Impact of Fe0 amendment on methylene blue discoloration by sand columns 1 
2 
3 
4 
5 
6 
7 
8 
9 
10 
11 
12 
13 
14 
15 
16 
17 
18 
19 
20 
21 
22 
23 
24 
25 
26 
27 
Miyajima K.a, Noubactep C.a,b,* 
aAngewandte Geologie, Universität Göttingen, Goldschmidtstraße 3, D-37077, Göttingen, Germany. 
bKultur und Nachhaltige Entwicklung CDD e.V., Postfach 1502, D-37005 Göttingen, Germany 
* corresponding author: e-mail: cnoubac@gwdg.de; Tel. +49 551 39 3191, Fax: +49 551 399379. 
 
Abstract 
The influence of metallic iron (Fe0) amendment on the efficiency of sand to discolor a 2.0 mg 
L-1 methylene blue (MB) solution was investigated in column studies. MB was used as an 
indicator to identify the optimum Fe0/sand ratio for efficient filtration systems. Columns 
contained 0, 100 or 200 g of a Fe0 material. The volumetric proportion of Fe0 in the reactive 
layer of the columns with 100 g of material varied from 10 to 100 %. Results showed that, Fe0 
amendment significantly impaired MB discoloration by sand for experiments lasting for up to 
132 days. Early MB breakthrough in Fe0/sand columns delineated the paramount importance 
of particle cementation, which has caused preferential flow with a negative impact on 
discoloration efficiency. The most efficient Fe0/sand mixtures were the ones with 30 to 50 % 
Fe0 (v/v). These volumetric ratios correspond 33 to 41 % weight ratios showing that the 
commonly used 1:1 weight ratio (50 %) may not be optimal. Further research with 
compounds exhibiting different affinities to both Fe0 and sand is needed before this 
observation can be generalized. 
Keywords: Fe0/sand filters, Particle cementation, Permeability loss, Water treatment, Zero-
valent iron. 
 
1 Introduction 
Metallic iron (Fe0) has been demonstrated in numerous studies to represent one of the best 
available materials for subsurface permeable reactive barriers [1-6]. Fe0 is also a very efficient 
material for above-ground wastewater treatment and safe dinking water provision [7-9]. New 
 1
applications of Fe0 for water treatment usually involve extensive pilot-scale studies although 
several models have been developed for predicting the performance of Fe
28 
29 
30 
31 
32 
33 
34 
35 
36 
37 
38 
39 
40 
41 
42 
43 
44 
45 
46 
47 
48 
49 
50 
51 
52 
53 
0 material [10-13]. 
Moreover, despite 20 years of intensive research, the question as to whether iron should be 
used alone or mixed to a cost efficient material is yet to be properly addressed [14-24]. 
The suitability of Fe0 for water treatment arises from its aqueous instability (Eq. 1 - 4). 
Immersed Fe0 is oxidized by water according to Eq. 1: 
Fe0 + 2 H+ ⇒ Fe2+ + H2↑     (1) 
Fe0 + ½ O2 + H2O ⇒ Fe2+ + 2 HO-    (2) 
2 Fe2+ + ½ O2 + H2O ⇒ 2 Fe3+ + 2 HO-   (3) 
Fen+ + n HO- ⇒ Fe(OH)n ⇒ FeOOH/FexOy  (4) 
In the presence of dissolved O2 the more favourable redox reaction is given by Eq. 2, but 
reaction following Eq. 1 still significantly occurs due to the abundance of water [25]. 
Moreover, it has been traceably shown that even under external oxic conditions, Fe0 is 
oxidized by water (Eq. 1) and FeII by O2 (Eq. 3) [26]. In other words, accelerated Fe0 
oxidation under oxic conditions results from FeII consumption by O2 (Le Chatelier’s 
principle) and not from any direct interactions between Fe0 and O2 [26,27]. 
The sustainability of Fe0 filtration systems primarily depends on the intrinsic reactivity of Fe0 
which oxidative dissolution induces both contaminant removal [28] and porosity loss due to 
the volumetric expansive nature of iron corrosion [29]. Aqueous iron oxidation yields 
hydroxides (e.g. Fe(OH)2) and oxides (e.g. Fe3O4) (Eq. 4) which volume varies from 2.1 to 
6.4 times the volume of Fe0 in the metal [30]. The transformation of Fe0 to (hydr)oxides (Eqs. 
1 through 4) goes through colloidal intermediates with cementing capacities. Here, 
cementation means that the inter-granular space is filled by FeII/FeIII precipitates and the solid 
particles are “cemented” to each other. This cementation results in material consolidation and 
in permeability loss in column experiments [14,16,31]. The ease with which cementation is 
observed primarily depends also on the reaction environments (oxic vs. anoxic). However, the 
 2
first tool to restrict cementation is to decrease the proportion of cement producers, that is to 
mix Fe
54 
55 
56 
57 
58 
59 
60 
61 
62 
63 
64 
65 
66 
67 
68 
69 
70 
71 
72 
73 
74 
75 
76 
77 
78 
79 
0 with a non expansive material (e.g. activated carbon, anthracite, clay, gravel, organic 
substances, pumice, sand) [32].  
The literature contains many contradictory findings regarding the question as to whether 
admixing inert materials to micro-scale Fe0 is beneficial for its performance or not 
[17,20,24,33]. Inert materials have been routinely mixed with Fe0 to (i) avoid/delay/minimize 
Fe0 cementation [14-16], (ii) increase the hydraulic conductivity [1,15,19] and (iii) decrease 
the cost of the barrier [18]. Recent theoretical works have demonstrated that admixing non 
expansive materials (e.g. gravel, MnO2, pumice, sand, TiO2) to Fe0 is a pre-requisite for 
sustainable Fe0 filtration systems [34].  
The present work is the first attempt to experimentally test the validity of the concept that 
admixing non expansive materials to Fe0 is a pre-requisite for sustainable (e.g. long-term 
efficient) Fe0 filtration systems. For this purpose, the discoloration of aqueous methylene blue 
(MB) is investigated in ten columns containing 0 g (1 column), 100 g (8 columns) or 200 g (1 
column) of a commercial Fe0. The tested Fe0/sand volumetric ratios were 0/100, 10/90, 20/80, 
30/70, 40/60, 50/50, 70/30, 80/20 and 100/0. Individual systems are characterized by the time-
dependent evolution of (i) the extent of iron release, (ii) the MB breakthrough and (iii) the 
hydraulic conductivity. 
2 Background of the experimental methodology 
The choice of methylene blue (MB) as model pollutant to ascertain the optimal Fe0/sand ratio 
for sustainable efficient Fe0-based filters arises from three major factors: (i) the ease of MB 
determination by cost-efficiency colorimetric methods, (ii) the low affinity of MB to iron 
oxides (corrosion products) [35], and the high affinity of MB to sand [36,37]. The low affinity 
of MB for iron corrosion products suggests that the retention time of MB in a Fe0-based 
system will be minimal (rapid breakthrough). The breakthrough time is further lowered by 
working under atmospheric conditions (oxic conditions) where voluminous corrosion 
 3
products are generated [30,38]. Despite low affinity to iron oxides, MB is removed in Fe0-
based filters mainly by adsorptive size-exclusion. However, adsorption and co-precipitation 
still occur [39], but their extent depends on the flow velocity or the residence time of the 
solution within the filter. 
80 
81 
82 
83 
84 
85 
86 
87 
88 
89 
90 
91 
92 
93 
94 
95 
96 
97 
98 
99 
100 
101 
102 
103 
104 
105 
The suitability of Fe0-amended sand filters to investigate the impact of the corrosion process 
on the efficiency (MB discoloration, permeability loss) of a sand filter (reference system) 
arises from the historical observation by Mitchell et al. [35], that sand is a better adsorbing 
agent for MB than iron oxide coated sand. In other words, the discoloration performance of a 
Fe0-amended sand filter is worsen by the progressive coating of sand by in-situ generated iron 
oxides (assumption 1). On the other hand, the accumulation of corrosion products in a system 
will reduce the porosity and lower its hydraulic conductivity (assumption 2). 
The used methodology for the investigation of the impact of the Fe0/sand ratio on the 
efficiency investigated system comprises testing the validity of assumption 1 and assumption 
2 by following the extent of (i) MB discoloration and (ii) permeability loss in Fe0-ammended 
sand columns. 
3 Materials and methods 
3.1 Solutions 
3.1.1 Methylene blue (MB ) 
MB is widely used as model contaminant to characterize the suitability of various materials 
for water treatment [35-37, 40-42]. The used methylene blue (MB) was of analytical grade. 
The working solution has a concentration of 2.0 mgL-1 and was weekly prepared by diluting a 
1000 mg L-1 stock solution using the tap water of the city of Göttingen. Its average 
composition (in mg L-1) was: Cl–: 12.9; NO3–: 7.5; SO42-: 35.5; Na+: 9.7; K+: 0.9; Mg2+: 8.2; 
Ca2+: 37.3. The pH value of the initial solution was 8.2. The used concentration 2.0 mg L-1 or 
6.3 μM was selected to approach the concentration range of natural waters (MB as model 
micro-pollutant). 
 4
3.1.2 Iron 106 
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 
A standard iron solution (1000 mg L-1) from Baker JT® was used to calibrate the 
spectrophotometer used for analysis. All other chemicals used were of analytical grade. In 
preparation for spectrophotometric analysis, ascorbic acid was used to reduce FeIII in solution 
to FeII. 1,10 orthophenanthroline (ACROS Organics) was used as reagent for FeII 
complexation. Other chemicals used in this study included L(+)-ascorbic acid and L-ascorbic 
acid sodium salt. 
3.2 Solid materials 
3.2.1 Metallic iron (Fe0) 
The used Fe0 material was purchased from iPutech (Rheinfelden, Germany). The material is 
available as fillings with a particle size between 0.3 and 2.0 mm. Its elemental composition as 
specified by the supplier was: C: 3.52%; Si: 2.12%; Mn: 0.93%; Cr: 0.66%. The material was 
used without any further pre-treatment. Fe0 was proven a powerful discoloration agent for MB 
with the particularity, that discoloration agents are progressively generated in-situ [39]. 
Therefore, the discoloration capacity of used Fe0 can not be exhausted within the 
experimental duration (here 132 days). 
3.2.2 Sand 
The used sand was a commercial material for aviculture (“Papagaiensand” from RUT – 
Lehrte/Germany). Papagaiensand was used as received without any further pre-treatment or 
characterization. The particle size was between 0.5 and 2.0 mm. Sand was used as an MB 
adsorbent [35] because of its worldwide availability and its use as admixing agent in Fe0/H2O 
systems [1,24,34]. The adsorption capacity of sand for MB can be exhausted within the 
experimental duration. 
3.3 MB discoloration 
As-received Fe0 from iPutec GmbH (Rheinfelden, Germany) was used. The materials were 
packed into columns in a dual manner. Sand (Hsand,1 and Hsand,2 – Tab. 1) and pure Fe0 layers 
 5
(columns 9 and 10 – Tab. 1) were wet packed. For all other reactive zones (< 100 % Fe0), dry 
homogenized 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 
157 
0/sand mixtures were introduced into the column in small lofts, which were 
wetted and compacted with manual tapping. To warrant optimal compaction, columns were 
gently tapped with a 100 mL PET flacon containing water. Tested volumetric ratios of Fe0 in 
the reactive zone (Tab. 1) were built while using the volume occupied by 100 g of Fe0 (32 mL 
– apparent volume) as unity. The resulting sand masses are documented in Tab. 1. For 
example, the system with 20 % (v/v) Fe0, was made up of one volume of Fe0 and four 
volumes of sand. The corresponding mass of sand was 77 g yielding a weight ratio of 43.5 %. 
The 2.0 mg L-1 MB solution was pumped upwards from PE bottles using a peristaltic pump 
(Ismatec, ICP 24). Tygon tubes were used to connect inlet reservoir, pump, column and outlet. 
Ten glass columns (40 cm long, 2.6 cm inner diameter) having a section of 5.31 cm2 were 
used. The columns were mostly packed with sand (Tab. 1, Fig. 1). The extent of MB 
discoloration by packed materials and the extent of the permeability loss by individual column 
set-ups (Tab. 1) were the sole targets. The experiments were performed at room temperature 
(22 ± 3 °C). A stable flow rate of 0.1 mL min-1 was maintained throughout the experiment. 
The whole effluent was collected. The volume recorded as function of the elapsed time served 
for the assessment of flow velocity or hydraulic conductivity. Each collected fraction was 
analysed for MB and dissolved Fe.  
3.4 Analytical methods 
MB and aqueous iron concentrations were determined by a Cary 50 UV-Vis 
spectrophotometer (Varian) at a wavelength of 664.5 nm and 510 nm respectively. Cuvettes 
with 1 cm light path were used. The iron determination followed the 1,10 orthophenanthroline 
method [43]. The spectrophotometer was calibrated for MB concentrations ≤ 2.5 mg L-1 and 
iron concentrations ≤ 10.0 mg L-1. The pH value was measured by combined glass electrodes 
(WTW Co., Germany). 
3.5 Presentation of experimental results 
 6
The MB breakthrough curves are expressed in terms of normalized concentration defined as 
the ratio of effluent dye concentration to inlet dye concentration (C/C
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 
183 
0) as a function of time 
or volume of effluent for a given bed height. For each column set-up, the extent of MB 
discoloration (efficiency, E in %) at each time was calculated according to the following 
equation (Eq. 5): 
E = [(ΣVi*C0 - ΣVi*Ci)/ΣVi*C0] * 100%    (5) 
where C0 is the initial aqueous MB concentration (2.0 mg L-1), while Ci gives the MB 
concentration in each collected sample Vi. 
In order to characterize the effect of the tested column set-ups on MB discoloration at the end 
of the experiment, the discoloration efficiency (E) and the specific discoloration (Es) were 
calculated using Eq. 6 and Eq. 7.  
E = mdiscol/min*100      (6) 
Es = mdiscol/mFe*100     (7) 
where min is the mass of MB flowed into the column, mdiscol is the MB mass discoloured 
within the reactive zone, and mFe the mass of Fe0 present in the column. It is operationally 
assumed that Hsand,2 (Tab. 1) does not significantly impact MB discoloration if significant 
flow disturbance happens in the reactive zone (assumption 3). The extent of MB discoloration 
(mMB in mg) within the reactive zone of individual columns was calculated from Hsand,1 using 
the rule of proportion. In this effort, column 1 was used as reference (44 cm of sand for 48.78 
mg MB) (Eq. 8). 48.78 mg is the value of mdiscol (mdiscol =ΣVi*C0) at MB breakthrough in the 
pure sand column. 
mMB (mg) = (Hsand,1/44) * 48.78   (8) 
4 Results and Discussion 
4.1 Visual analysis of the columns at day 132 
Figure 1 reveals 4 main colorations: blue, brown, dark-green and ‘white’. White is the 
operational colour of sand as seen in the upper part of columns 9 and 10, blue is the colour of 
 7
adsorbed methylene blue (column 1), brown is the product of Fe0 oxidation by dissolved O2, 
and dark-green is the colour of Fe
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 
II species (including green rust). The formation of rust 
(brown coloration) is the most tangible evidence of corrosion by O2. That is either the local 
iron oxidation by O2 or the migration of FeIII species (e.g. Fe(OH)3, Fe2O3•H2O). In other 
words, a brown coloration ‘far’ from the entrance zone of the reactive layer is an evidence of 
O2 breakthrough. 
It is important to notice that MB breakthrough is observed in all systems although an 
homogeneous blue coloration is only see in system 1 (0 % Fe0) and in the lower part of all 
other columns (Hsand,1 – Tab. 1). This observation indicates that flow disturbance was caused 
within the Fe0-containing layer. In other words, uniform flow may be limited to pure sand 
layers (column 1, Hsand1). 
In general, from the inlet to the outlet, Fe0-containing columns exhibited the following 
coloration sequence: (i) blue (Hsand,1), (ii) brown (entrance of the reactive layer), (iii) grey or 
dark (reactive layer), (iv) dark-green (entrance of Hsand,2) and (v) white (Hsand,2). Columns 7 
and 8 exhibited a different behaviour as the entrance zone of the Hsand,2 layer was brown 
coloured (not green-dark). This coloration indicates that dissolved O2 could be quantitatively 
transported across the reactive layer in these 2 systems. Additionally, a blue coloration is seen 
in the upper part of column 7. 
The presence of a brown coloration in the connecting tube at the column outlet for all Fe0-
containing systems shows that there is an axial breakthrough of O2 as a rule. In other words, 
the O2 breakthrough is the first hint for preferential flow due to particle cementation. The 
second hint for preferential flow is the blue coloration in upper sand layer of column 7. The 
absence of the brown coloration in the upper sand layer of systems with less than 70 % Fe0 is 
concordant with the use of Fe0/sand mixtures as ‘O2 scavengers’ [2,3,14]. 
4.2 Evidence for an optimal system with less than 70 % Fe0 
 8
Table 2 summarizes the results from the 10 columns. It shows that about 38.0 L of the MB 
solution flow through each column during the 132 days. This corresponds to about 69 mg of 
MB (effective initial concentration 1.83 mg L
209 
210 
211 
212 
213 
214 
215 
216 
217 
218 
219 
220 
221 
222 
223 
224 
225 
226 
227 
228 
229 
230 
231 
232 
233 
234 
-1) from which 42.0 to 63.3 % was discoloured 
within the reactive layer in Fe0-containing systems (columns 2 to 10). On the other hand, 
dissolved iron release from the same columns (4.0 to 47.1 mg) represented only less than 0.05 
% of the initial Fe0 amount (100 or 200 g). 
An important result from Tab. 2 is that the reference system (0 % Fe0 – column 1) was the 
most efficient system with 72.3 % MB discoloration efficiency at the day 132 (E value - Fig. 
2). Therefore, assumption 1 is validated. However, at this date the discoloration capacity of 
this system is almost exhausted because this is a pure adsorption system [44,45]. In column 1 
(97 % breakthrough), 48.79 mg MB is adsorbed onto 360 g of sand, yielding an adsorption 
capacity of 0.138 mg g-1. For Fe0-containing system, not such adsorption capacity can be 
defined because MB is not adsorbed onto a defined surface, rather MB is adsorbed or 
enmeshed in the matrix of iron corrosion products [46]. Additionally, the extent of Fe0 
depletion is difficult to access. 
Figure 2 summarizes the results of the evolution of the specific MB discoloration (Es in μg g-
1) as a function of the volumetric proportion of Fe0 in the reactive zone. It is seen that Es 
evolutes through a maximum at 30 % Fe0. Es then decreases to a minimum at 80 % Fe0. The 
overall trend that, for long enough experimental durations, contaminant removal should be 
minimal (i) in the sand column (pure adsorbent with limited capacity) and (ii) in pure Fe0 
column (no room for expansive iron corrosion) is found here. The fact that the pure Fe0 
system (100 % Fe0) is more efficient than the 80 % Fe0 system is due to the simplistic nature 
of assumption 3 (no MB discoloration in Hsand,2) as attested by the blue coloration in the 
Hsand,2 layer in column 7 (Fig. 1). MB discoloration is normalized to 360 g of sand in the 
reference system and to (only) 100 g Fe0 in all other systems. This is the reason for the 
apparent discrepancy between E and Es values in Fig. 2. 
 9
The most important output from Fig. 2 is that for the same mass of Fe0, various Es values (143 
to 239 μg g
235 
236 
237 
238 
239 
240 
241 
242 
243 
244 
245 
246 
247 
248 
249 
250 
251 
252 
253 
254 
255 
256 
257 
258 
259 
260 
-1) are obtained after 132 days. These results clearly question the 
suitability/validity of Es as indicator of Fe0 efficiency when experimental conditions are not 
identical/similar. Fe0 is definitively not an adsorbent but a generator of ‘MB scavengers’. 
Accordingly, the discoloration efficiency/capacity of Fe0 for MB can not be universally 
defined. Therefore, results from Fig. 2 radically refute the generalized use of ‘removal 
capacity’ as indicator of the efficiency of Fe0 materials. 
4.3 Iron breakthrough 
Figure 3 summarizes the results of the evolution of dissolved iron concentration in the 
effluent (also see Tab. 2). It is seen that Fe release and migration occurs in three stages: (i) 
dissolution from Fe0, (ii) short distance migration due to the low solubility at pH > 5, and (iii) 
adsorption and precipitation on sand particle (in situ coating). This in situ coating is 
responsible for the accelerated MB breakthrough because of porosity loss and lower 
adsorptive affinity [35]. Fig. 3 shows that the largest iron release (up to 8.0 mg L-1) was 
observed in the system with the lowest Fe0 volumetric proportion of Fe0 (10 %). This column 
corresponds to the system with Hsand,2 = 0. The next system (30 % Fe0) exhibiting noticeable 
iron breakthrough (> 2 mg L-1) hast the lowest Hsand,2-value (11 cm). This observation 
confirms the fact, that the long-distance transport of dissolved iron is not favoured at pH > 5 
[47]. In all other systems, even more iron could have been dissolved but it is retained within 
the reactive layer and the under laying sand (Hsand,2 > 11 cm) (dark-green coloration – section 
3.1). The retention mechanisms are (i) adsorption onto available iron oxides or onto sand, or 
(ii) precipitation as iron (hydr)oxides. It is very important to notice that the extent of iron 
release depends primarily on the intrinsic reactivity of used Fe0. Although data on iron release 
from column experiments are available in the literature [15] it is quite impossible to make a 
quantitative comparison. In fact, a parameter (or an index) to characterize the intrinsic 
reactivity of Fe0 is still lacking [48]. 
 10
The fact that in situ dissolved iron mostly remains in the system is a hint, that MB should be 
more or less quantitatively removed by adsorption, co-precipitation and adsorptive size 
exclusion (straining) [46]. As recalled in section 3.2, all Fe
261 
262 
263 
264 
265 
266 
267 
268 
269 
270 
271 
272 
273 
274 
275 
276 
277 
278 
279 
280 
281 
282 
283 
284 
285 
286 
0/sand systems should be more 
efficient that pure sand system (assumption 4). Testing the validity of assumption 4 will 
sustain the further presentation. 
4.4 Behaviour of the columns 
Fig. 4 (a and b) and Tab. 3 summarize the results of the 10 columns (see also Tab. 2). Figure 4 
depicts the time-dependant evolution of the MB mass in the effluent. It can be seen that 
breakthrough occurs after at least 18 mg of MB was retained in all systems. This corresponds 
to complete discoloration of 9.0 L of the initial solution.  
Fig. 4b shows that the reference system (0 % Fe0) is the most efficient at discoloring MB for 
the tested experimental duration (132 days). This observation seemingly disproves assumption 
4. However, it should be recalled that the adsorption capacity of sand is limited. Therefore, it 
can be argued that ‘early breakthrough’ in Fe0-amended systems is due to the low reactivity of 
used Fe0 as it could not produce enough ‘scavengers’ for MB removal (assumption 5). 
The system with 80 % Fe0 was the less efficient (Tab. 2). The remaining systems could be 
ordered as follows: 70 % < 100 % < 10 % < 20 % < 50 % < 40 % < 30 %. Since sand is a 
good adsorbent for MB, the classification has to take the thickness of sand preceding the 
reactive layer into account (Hsand,1 - see sections 3.2 and 3.5). However, the absence of a 
monotone trend in the evolution of the systems disproves assumption 5. Therefore, another 
process, yet to be identified is responsible for the ‘early breakthrough’ of MB in Fe0-amended 
systems. 
Fig. 4b compares the efficiency of the reference system (0 % Fe0) with that of pure Fe0 
systems (column 9 and 10 – Tab. 1). The sand-column remains the most efficient but the 
column with 200 g Fe0 was more efficient than the column with 100 g Fe0. This is a further 
evidence that Fe0 disturbs the efficiency of sand for MB discoloration but not in a monotone 
 11
linear way. While further disproving assumption 5 (but validating assumption 1), this 
observation suggest that a stochastic process is responsible for the ‘early breakthrough’ of MB 
in systems with 100 g Fe
287 
288 
289 
290 
291 
292 
293 
294 
295 
296 
297 
298 
299 
300 
301 
302 
303 
304 
305 
306 
307 
308 
309 
310 
311 
312 
0 (assumption 6). 
Fig. 4 also shows that after breakthrough the mass MB (mg) in the effluent linearly increases. 
The regression parameters (a and b values) of the lines [MB]effluent = a*[MB]influent + b were 
determined with Origin 6.0 and summarized in Tab. 3. The slope of this line can be regarded 
as the rate of MB increase in the effluent after breakthrough. For a pure absorbent the ‘a 
value’ ideally approaches zero corresponding to a S-shape breakthrough. In other words, the 
smaller the ‘a value’, the more the system is close to an adsorbing system. The classification 
of the systems after the order of increasing ‘a values’ is: 0 % < 30 % < 40 % < 50 % < 20 % < 
10 % < 75 % = 100 % < 80 %. It is interesting that both columns with 100 % Fe0 (100 and 
200 g) exhibited the same a value (0.65). This fact is an important hint that despite differences 
in the thickness the same ‘stochastic processes’ are responsible for MB discoloration. The 
difference between both 100 % Fe0-columns is the ‘b value’: -14.6 for 100 g and -23.0 for 200 
g. Here again, the absence of a simple proportion between the two ‘b values’ is a hint of the 
stochastic nature of processes occurring in Fe0/H2O systems and leading to ‘early 
breakthrough’ relative to the pure sand system (0 % Fe0). 
4.5 Effect of the thickness Hsand,1
Fig. 1 showed that the reference system (first column from left) was entirely blue colored 
while in all other systems, the uniform blue color is limited to the sand before the Fe0 layer or 
reactive zone (Hsand,1 - Tab. 1). This observation suggests that, in Fe0-containing systems, 
granular particles (Fe0 and sand) are partly cemented and preferential flow is generated in the 
Fe0 layer. 
If it is mentally assumed that the Fe0 layer depicted no interactions with MB, the solution 
would have ‘traversed’ the Fe0 layer non-disturbed and the uniform blue coloration would 
have continued after the pure Fe0 (columns 9 and 10) or the Fe0-amended layer (columns 2 to 
 12
8). However, as reported by Imamura et al. [49] MB adsorbs strongly onto Fe0 materials. 
Therefore, the absence of a uniform blue coloration can be attributed to the flow disturbance 
in the Fe
313 
314 
315 
316 
317 
318 
319 
320 
321 
322 
323 
324 
325 
326 
327 
328 
329 
330 
331 
332 
333 
334 
335 
336 
337 
0 layer (assumption 2 is validated). Additionally, Mitchell et al. [35] have traceably 
demonstrated that iron oxide-coated sand is a poorer adsorbent for MB than pure sand. It is 
the objective of this study to characterize the processes occurring within the reactive layer. 
The initial Fe0 dissolution creates a local super-saturation of aqueous Fe which is short-term 
diffusive transported and precipitated as cement, in particular the observed brown coloration 
at the entrance of the reactive layer. Particle cementation is certainly a stochastic process, 
suggesting that assumption 6 may be valid. 
Fig. 5 summarized the results of a mathematical modelling of the time to breakthrough in the 
Fe0-amended columns. The time necessary for 0.5 breakthrough in the pure sand column (day 
97) is used as reference. The model used Eq. 8 while replacing the mass by the time where 
C/C0 = 0.5 in column 1 (96.83 days - Eq. 9). In the case “sand1 & 2”, “Hsand,1” is replaced by 
“Hsand,1 + Hsand,2” (Tab. 1). 
tBT (days) = (Hsand,1/44) * 96.83   (9) 
As a rule, the longer the time to breakthrough, the more efficient a system. Fig. 5 shows that 
the most efficient systems were the systems with 30, 40 and 50 vol % Fe0. It is seen that the 
observed efficiency is always higher than the efficiency of Hsand1, while for Fe0 > 40 % the 
efficiency of the system was lesser than that of sand alone (‘sand 1 & 2’). This observation 
corroborates the complexity of processes in the Fe0-ammended and pure Fe0 layers. It should 
be kept in mind that the data presented in Fig. 5 (‘observed’ in particular) are static snap-shots 
of dynamic processes (time for 0.5 MB breakthrough in the sand column). However, they 
clearly show that systems with Fe0 volumetric ratios ≥ 70 % are less efficient than systems 
containing 30 to 50 vol % Fe0. 
4.6 Hydraulic conductivity 
 13
338 
339 
340 
341 
342 
343 
344 
345 
346 
347 
348 
349 
350 
351 
352 
353 
354 
355 
356 
357 
358 
359 
360 
361 
362 
363 
The results presented in Fig. 6 clearly demonstrate that the hydraulic conductivity almost 
remained constant during the whole experiment. This unusual observation could be 
misinterpreted as the absence of porosity loss due to expansive iron corrosion. However, the 
used material is known for its reactivity and the observed brown coloration (section 3.1) 
attests that oxidative dissolution and subsequent precipitation of iron hydroxides has occurred. 
One should keep in mind, that permeability loss is observed whenever the sum of forces 
generated in the system (filter resistance) are superior to the pressure supplied by the 
peristaltic pump (initial driving force) (Eq. 10). 
Flow rate = Driving force / Filter resistance    (10) 
In the present work, it could be concluded that the used pump flow rate (0.1 mL min-1) was 
sufficient to transport enough in situ generated iron oxides out of the reactive zone and so to 
avoid significant clogging (for 132 days). Fig. 1b shows that, in all Fe0-based systems, 
colloidal Fe species have escaped from the columns and precipitated in the connection tubes. 
Figure 7 schematically shows the time-dependent evolution of the porosity in a Fe0/sand 
cylindrical column while Fe0 experiences uniform corrosion. It is seem that porosity loss 
should increase with the experimental duration (assumption 2). This trend was not observed in 
the current study (Fig 6). This observation is rationalized by the fact that the filter resistance 
generated by 100 or 200 g of Fe0 for up to 132 days was not sufficient to significantly induce 
a decrease of the flow rate (hydraulic conductivity). However, material compaction was 
observed and its thickness was smaller in systems with higher Fe0 proportions (Tab. 2). 
Accordingly, although no significant permeability loss is noticed, particle cementation has 
impacted the uniform flow in various extent in individual columns. Assumption 2 is thus 
validated. 
The driving force (hydraulic pressure) from Eq. 10 is ideally uniformly distributed in the 
whole column. That is on all Fe0 and sand particle within the 44 cm effective length. 
However, due to the cementation process described above, there are local ‘irregularities’, as 
 14
cemented particles are non accessible and cemented regions non/less permeable. For example, 
Eq. 10 suggests that if a 50 % porosity loss occurs (t
364 
365 
366 
367 
368 
369 
370 
371 
372 
373 
374 
375 
376 
377 
378 
379 
380 
381 
382 
383 
384 
385 
386 
387 
388 
389 
1/2 in Fig. 7), the hydraulic pressure 
should be doubled to maintain the observed constant flow. Actually, a local increase of the 
hydraulic pressure has occurred and has caused preferential flow at a certain distance from the 
‘clogged zone’. This accelerated flow is illustrated the best by the brown colour in the Hsand,2 
layer as observed in columns 7 and 8 (Fig. 1, section 3.1). 
The fact that column 7 and 8 depicted the lowest Es efficiency (Fig. 2) is a further 
confirmation of assumption 4 (“all Fe0/sand systems should be more efficient that pure sand 
system”) as the observed anomaly results from the particular difficulty to uniformly mix Fe0 
and sand when Fe0 is in volumetric abundance. It should be recalled that sound theoretical 
works [50] have proven the needlessness of testing these volumetric proportions. Therefore, it 
is not worth to continue the discussion on these systems. Accordingly, further research should 
focus on Fe0 ratios < 60 % while eventually test the pure Fe0 as negative reference. 
4.7 MB breakthrough 
Figure 8 summarizes the results of MB breakthrough. A S-shaped breakthrough curve was 
obtained for all systems but only the reference system experienced a complete breakthrough 
(C/C0 = 0.97 at day 132).  
The breakthrough time (time for [MB] ≠ 0) varied largely and was primarily dependent on the 
thickness of Hsand.1. As shown in Fig. 8a, individual columns depicted slightly different 
behaviour but the discoloration efficiency was levelled at about C/C0 = 0.70. This levelling at 
C/C0 = 0.70 suggests that the ‘residual’ (steady state) corrosion rate was sufficient to induce 
the discoloration of about 30 % of a 2.0 mg L-1 MB. In other words, an effluent solution 
containing about 0.60 mg L-1 MB (30 % of 2 mg L-1) could be efficiently treated (no MB 
breakthrough) for the 132 days of the experiment. As a rule, natural waters contain much 
lower levels of contaminants termed ‘micro-pollutants’ [51]. Remember that the experiments 
were designed to characterize the efficiency of Fe0 at various Fe0/sand ratios using limited 
 15
amounts of Fe0. Accordingly, both MB breakthrough and the characterization of the system 
thereafter were intended. 
390 
391 
392 
393 
394 
395 
396 
397 
398 
399 
400 
401 
402 
403 
404 
405 
406 
407 
408 
409 
410 
411 
Fig. 8b summarized the MB breakthrough behaviour in the 3 pure material systems (100 % 
sand or Fe0). Its shows that the sand systems is the most efficient for 132 days but that the 
breakthrough is quantitative (100 %) suggested that the adsorption capacity is exhausted. 
From this moment on, no significant MB discoloration is expected. On the contrary, in the 
pure Fe0 systems the initial S-shape breakthrough starting after 25 and 50 days respectively is 
levelled at about 70 % (C/C0 = 0.70). Fig. 8b reveals that after a short levelling at 70 % 
breakthrough, ‘additional removal’ occurred in the system with 200 g Fe0 followed by a phase 
of limited removal efficiency. This ‘fluctuation’ is also be evident from Fig. 8a, in particular 
for the 30 % Fe0 system. These ‘fluctuations’ document the uncertainty in the long-term 
reactivity of Fe0 materials. This observation supports the quest to accumulate data on long-
term reactivity of various Fe0 materials at laboratory scale [48]. 
The breakthrough results suggest the efficient use of Fe0/sand systems in beds for water 
treatment in deep filtration modus. However, design efforts are urgently needed. It is observed 
from Fig. 8a that the time to achieve breakthrough is the highest for systems containing 30 to 
50 % Fe0 (v/v). 
4.8 Discussion 
4.8.1 Validating the used experimental methodology 
The presentation herein is centred on testing the validity of two fundamental (1,2) and four 
operational (3-6) assumptions. Assumptions 1 through 6 are summarized and commented in 
this section. 
 16
Assumption 1: the (initial) MB discoloration performance of a Fe0-amended sand filter is 
worsen by the progressive coating of sand by in-situ generated iron oxides. Assumption 1 is 
verified by experimental data for up to 132 days, clearly showing that the sand system 
(reference system) is the most efficient for MB discoloration. 
412 
413 
414 
415 
416 
417 
418 
419 
420 
421 
422 
423 
424 
425 
426 
427 
428 
429 
430 
431 
432 
433 
434 
435 
436 
437 
Assumption 2: the accumulation of corrosion products in a system reduces its porosity and 
lowers the hydraulic conductivity (permeability loss). Assumption 2 is indirectly verified as 
the used pump flow rate (0.1 mL min-1) was sufficient to avoid accumulation of corrosion 
products in columns containing only 100 g of Fe0 during 132 days. It is recommended that 
lower pump flow rates are tested in future works. Alternatively, higher Fe0 amounts (> 100 g) 
may be tested using the pump flow rate of 0.1 mL min-1. 
Assumption 3: the sand layer after the Fe0 layer (Hsand,2 - Tab. 1) does not significantly 
impact MB discoloration if significant flow disturbance happens in the reactive zone. This 
assumption was not verified showing the complexity of processes coupled to Fe0 amendment 
of sand filters. 
Assumption 4: All Fe0/sand systems are more efficient than the pure sand system. This 
assumption was disproved by experimental data for up to 132 days. However, the fact that the 
MB removal capacity of the reference system (sand only) was almost exhausted at day 132 
suggests that this assumption could be valid for longer experimental durations if the systems 
are not clogged. Exactly this was the goal of the present study: identifying the optimal 
Fe0/sand ratio concealing increased permeability (more sand) and decreased efficiency (more 
sand or less Fe0). It is essential to point out, that MB presents all the characteristics of a tracer 
[52,53]. In particular, its very low affinity to the Fe0/H2O system makes it an ideal compound 
to characterize the impact of iron corrosion on the hydraulic properties of a Fe0 amended 
filter. 
Assumption 5: the ‘early’ MB breakthrough in Fe0-amended systems is due to the low 
reactivity of used Fe0 as it could not produce enough ‘scavengers’ for MB removal. This 
 17
438 
439 
440 
441 
442 
443 
444 
445 
446 
447 
448 
449 
450 
451 
452 
453 
454 
455 
456 
457 
458 
459 
460 
461 
462 
463 
assumption was disproved an corroborates the ‘tracer nature’ of MB for characterizing the 
process of iron corrosion in packed-bed filtration systems. In essence, produced iron oxides 
either accelerated MB breakthrough.  
Assumption 6: a stochastic (non linear) process is responsible for the ‘early’ MB 
breakthrough in Fe0-amended systems. Assumption 6 was validated by experimental data. 
This validation seriously questions the current approach of correlating Fe0 reactivity with 
monitored parameters at the system outlet including the extent of H2 evolution and the extent 
of contaminant removal [22-24]. 
The validation of the two fundamental assumptions (1 and 2) is regarded as the validation of 
the theory of Fe0-based filtration systems presented in ref. [34]. This makes the assertion that 
pure Fe0 systems (100 % Fe0) are not sustainable universal. This was even already 
experimentally observed as Fe0 systems used for As removal in South East Asia (Bangladesh, 
Nepal) were very efficient but not sustainable [7]. With respect of the longed optimal Fe0/sand 
ratio, the present work could identify the domain in which it should be sought: 30 to 50 % Fe0 
(v/v). 
4.8.2 Significance of achieved results  
The present study was designed to characterize the behaviour of Fe0 columns at discolouring 
MB when the 100 g Fe0 represents a volumetric proportion varying from 10 to 100 %. Using 
identical columns, the same initial flow rate (multi-channel peristaltic pump) and the same 
sand as an additive material, initial differences include the thickness of the reactive layer (Hrz) 
and the thickness of sand after the reactive layer (Hsand,2). To better discuss the impact of the 
thickness of sand after the reactive layer on the extend of Fe release from the column, the 
thickness of sand before the reactive layer (Hsand,1) was also varied (Tab. 1). 
Figure 9 summarizes the results of the cumulative MB discoloration (Eq. 5) as function of 
Hsand,1, Hrz and Hsand,2. It is evident that from each system, the E value decreases with 
increasing experimental duration. For example, the 0 % Fe0 system (Hsand,1 = 44 cm; Hrz = 
 18
464 
465 
466 
467 
468 
469 
470 
471 
472 
473 
474 
475 
476 
477 
478 
479 
480 
481 
482 
483 
484 
485 
486 
487 
488 
489 
Hsand,2 = 0 cm) exhibits E values of 100.0, 89.5 and 72.3 % for 55, 77 and 132 days 
respectively.  
Fig. 9 shows a rough trend that (i) E values increase with increasing Hsand,1 value; (ii) there is 
an optimum Hrz value of around 10 - 15 cm, (ii) no trend is observed for the Hsand,2 values. 
The optimum in the Hrz (11 to 15 cm) curve coincides with the results discussed in 3.5 
showing that 30 to 50 % Fe0 is the optimal range for efficient systems. This coincidence 
suggests that, despite variation in the Hsand,1 values, the proportion of Fe0 in the reactive layer 
is of paramount significance for system sustainability. 
Used originally as a model pollutant, methylene blue has exhibited the characteristic of an 
‘operational tracer’ or a true indicator. This study has demonstrated that there is much more 
uncertainty about data collection and their interpretation for the design of Fe0/H2O systems 
than is currently acknowledged. A significant uncertainty arises from the difficulty to assess 
the time-dependant extent of Fe0 consumption. An accurate evaluation of the extent of Fe0 
consumption would facilitate the prediction of long-term efficiency of Fe0/H2O systems. 
Accordingly, characterization efforts need to be coordinated. Based on the results of the 
present work, it is suggested that 100 g Fe0 making about 40 % (v/v) of reactive layers is 
routinely used in laboratory columns (corresponding Hrz about 13 cm for 2.6 cm ID). Ideally 
each experiment testing porous additives should be accompanied by a reference system where 
tested Fe0 is admixed to quartz. 
4.8.2 Promoting Fe0 filtration systems 
The results of this work lead to an avenue for reducing uncertainties in designing Fe0-based 
filtration systems. Volumetric Fe0 ratios ensuring long-term efficiency are lower than 50 %. 
This conclusion determines a 20 years lasting discussion on the suitability and the efficiency 
of admixing Fe0 with other materials (e.g. anthracite, gravel, pumice, sand).  
These results further demonstrated that contaminants should be characterized and tested 
according to their physical and chemical properties and particularly, their affinity to the 
 19
Fe0/H2O system or to iron corrosion products. The current approach of testing contaminants 
by their origin (e.g. agricultural wastes (fertilizers and pesticides), heavy metals, human 
wastes, industrial wastes, organic dyes, pharmaceuticals or radionuclides) is confusing [51]. 
For example, dyes are not an homogeneous class of substances as some of them contain the 
hydroxyl group (e.g. erichrome black T, bromocresol green, bromophenol blue, fluorescein) 
and some other (methyl orange, methyl red, and methylene blue) not [54]. With regard to 
interactions in Fe
490 
491 
492 
493 
494 
495 
496 
497 
498 
499 
500 
501 
502 
503 
504 
505 
506 
507 
508 
509 
510 
511 
512 
513 
514 
515 
0/H2O systems, dyes with hydroxyl groups exhibit a greater affinity to iron 
corrosion products. Accordingly, duplicating the experiments reported herein with fluorescein 
or bromophenol blue as model pollutant  will result in longer experimental duration.  
The results reported herein will enable/facilitate a science-based design of Fe0-based filtration 
systems. Based on the observation herein and the theory of the system established before [34], 
many reported discrepancies can be elucidated. In particular, it is unambiguously established 
that pure Fe0 systems are not sustainable. Accordingly, the recent article by Ruhl et al. [24] 
questioning the theory of the Fe0-based filters, validated herein, should be regarded as a 
classical case, illustrating the inefficiency of the data-based approach of developing a theory. 
This long history of mainstream science teaches that this approach can not be successful [55]. 
Ruhl et al. [24] have recently depreciated the theory of Fe0-based filters to an assumption and 
presented data seemingly challenging this theory. They tested anthracite (porous), gravel 
(compact), pumice (porous) and sand (compact) in dual granular mixtures with a commercial 
Fe0 (100 g) in long term column experiments (> 200 days) for the treatment of a 
trichloroethylene (TCE) contaminated groundwater. Their results confirmed differences in 
porosity for the tested materials but no difference in ‘reactivity’ as derived from H2 evolution. 
Due to increased cis-dichloroethylene (a TCE metabolite) concentrations in the effluent from 
all columns (TCE was completely removed), they concluded that "the mixed reactive filters 
are therefore not applicable for treatment of the here tested groundwater with its indigenous 
microorganisms". However, this conclusion disproved the results of O’Hannesin and Gillham 
 20
[1] who have demonstrated the suitability of Fe0 for contaminant removal at a site 
contaminated with TCE and PCE (tetrachloroethylene) using a Fe
516 
517 
518 
519 
520 
521 
522 
523 
524 
525 
526 
527 
528 
529 
530 
531 
532 
533 
534 
535 
536 
537 
538 
539 
540 
0:sand weight mixture of 
22:78.  
The work presented in ref. [24] shows clearly how misconceptions could be propagated based 
on wrong interpretation of good experimental observations. To accurately interpret their 
results, the authors would have additionally tested reference systems (e.g. pure anthracite, Fe0, 
gravel, pumice and sand). The results presented herein clearly demonstrated that for short 
experiments, an inert material could be more efficient than a reactive one. This result 
confirms that there are no generally efficient or non efficient materials but selected materials 
should be used to meet the site-specific requirements. In other words, there are appropriate 
and non appropriate materials. The present work hopes to have opened new avenues to 
optimised the appropriateness of Fe0 materials in packed filtration beds, whether they are used 
for groundwater remediation, wastewater treatment or save drinking water provision. 
5 Concluding remarks 
This study clearly demonstrates that a pure Fe0 (100 % Fe0) filtration system for water 
treatment is not sustainable. Moreover, even the commonly used 50 % weight ratio is 
demonstrated not to be the most efficient one. Accordingly, Fe0 filtration systems must 
contain less than 50 % (w/w) Fe0 to be sustainable. The presented work has also validated the 
suitability of using limited amount of Fe0 (here 100 g) to investigate relevant processes within 
reasonable experimental duration (here 4 months) [50]. The rationale for the selection of the 
admixing materials (e.g. activated carbon, biomaterials, charcoal, gravel, MnO2, pumice, 
wood gravel, sand) is yet to be realized. Future works under similar conditions, should use 
lower pumping rates to increase the probability to experimentally observe permeability loss. 
In fact, the tested pumping rate (0.1 mL min-1) has solely identified a region (30 to 50 vol % 
Fe0) of the optimal Fe0/sand ratio. An interesting case will consist to test gravity flow which is 
 21
541 
542 
543 
544 
545 
546 
547 
548 
549 
550 
551 
552 
553 
554 
555 
556 
557 
558 
559 
560 
561 
562 
563 
564 
565 
relevant for household water filters and for small water treatment plants in remote, low-
income communities.  
This study has further delineated the unsuitability of the expression of “removal capacity” for 
Fe0 materials. This expression could only be acceptable if materials are used under similar 
conditions, have similar intrinsic reactivity and used in the same mass loading.  
The last important feature from this study concerns the long-term reactivity of Fe0/sand 
systems. This study has shown that upon an initial increased reactivity Fe0 oxidation 
experienced a steady state which was able to produce enough scavengers for the quantitative 
removal of about 0.6 mg/L methylene blue (30 % of 2.0 mg L-1). Designing a Fe0/H2O system 
could be regarded as finding out such a steady state or creating the conditions for such a state 
with respect to the actual contaminant flux and the nature of the contaminant of interest. For 
future works more care should be taken while collecting and interpreting experimental data. 
 
Acknowledgments 
The draft manuscript was improved by the insightful comments from Thomas Ptak 
(Geosciences Center - University of Göttingen). Gerhard Max Hundertmark and Mohammad 
Azizur Rahman from the Geosciences Center (University of Göttingen) are acknowledged for 
technical support. The manuscript was improved by the insightful comments of anonymous 
reviewers from Chemical Engineering Journal. 
 
References 
[1] S.F. O´Hannesin, R.W. Gillham, Long-term performance of an in situ "iron wall" for 
remediation of VOCs, Ground Water 36 (1998) 164–170. 
[2] G. Bartzas, K. Komnitsas, Solid phase studies and geochemical modelling of low-cost 
permeable reactive barriers, J. Hazard. Mater. 183 (2010) 301–308. 
 22
566 
567 
568 
569 
570 
571 
572 
573 
574 
[3] L. Li, C.H. Benson, Evaluation of five strategies to limit the impact of fouling in 
permeable reactive barriers, J. Hazard. Mater. 181 (2010) 170–180. 
[4] S. Comba, A. Di Molfetta, R. Sethi, A Comparison between field applications of nano-, 
micro-, and millimetric zero-valent iron for the remediation of contaminated aquifers, 
Water Air Soil Pollut. 215 (2011) 595–607. 
[5] M. Gheju, Hexavalent chromium reduction with zero-valent iron (ZVI) in aquatic systems, 
Water Air Soil Pollut. 222 (2011) 103–148. 
[6] ITRC (Interstate Technology & Regulatory Council) Permeable Reactive Barrier: 
Technology Update. PRB-5. Washington, D.C.: Interstate Technology & Regulatory 
Council, PRB: Technology Update Team. www.itrcweb.org (2011). 575 
576 
577 
578 
579 
580 
581 
582 
583 
584 
585 
586 
587 
588 
589 
590 
[7] A. Hussam, Contending with a development disaster: SONO filters remove arsenic from 
well water in Bangladesh, Innovations 4 (2009) 89–102. 
[8] C. Noubactep, Metallic iron for safe drinking water worldwide, Chem. Eng. J. 165 (2010) 
740–749. 
[9] D.E. Giles, M. Mohapatra, T.B. Issa, S. Anand, P. Singh, Iron and aluminium based 
adsorption strategies for removing arsenic from water, J. Environ. Manage. 92 (2011) 
3011–3022. 
[10] L. Li, C.H. Benson, E.M. Lawson, Modeling porosity reductions caused by mineral 
fouling in continuous-wall permeable reactive barriers, J. Contam. Hydrol. 83 (2006) 
89–121. 
[11] S. Roy, P. Bose, Modeling arsenite adsorption on rusting metallic iron, J. Environ. Eng. 
136 (2010) 405–411. 
[12] S.-W. Jeen, R.W. Gillham, A. Przepiora, Predictions of long-term performance of 
granular iron permeable reactive barriers: Field-scale evaluation, J. Contam. Hydrol. 123 
(2011) 50–64. 
 23
591 
592 
593 
594 
595 
596 
597 
598 
599 
600 
601 
602 
603 
604 
605 
606 
607 
608 
609 
610 
611 
612 
613 
614 
615 
616 
[13] S.-W. Jeen, R.T. Amos, D.W. Blowes, Modeling gas formation and mineral precipitation 
in a granular iron column, Environ. Sci. Technol. 46 (2012) 6742–6749. 
[14] P.D. Mackenzie, D.P. Horney, T.M. Sivavec, Mineral precipitation and porosity losses in 
granular iron columns, J. Hazard. Mater. 68 (1999) 1–17. 
[15] P. Westerhoff, J. James, Nitrate removal in zero-valent iron packed columns, Wat. Res. 
37 (2003) 1818–1830. 
[16] O.X. Leupin, S.J. Hug, Oxidation and removal of arsenic (III) from aerated groundwater 
by filtration through sand and zero-valent iron, Wat. Res. 39 (2005) 1729–740. 
[17] E. Bi, J.F. Devlin, B. Huang, Effects of mixing granular iron with sand on the kinetics of 
trichloroethylene reduction, Ground Water Monit. Remed. 29 (2009) 56–62.  
[18] 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. 
[19] N. Moraci, P.S. Calabrò, Heavy metals removal and hydraulic performance in zero-
valent iron/pumice permeable reactive barriers, J. Environ. Manag. 91 (2010) 2336–
2341. 
[20] S. Ulsamer, A model to characterize the kinetics of dechlorination of tetrachloroethylene 
and trichloroethylene by a zero valent iron permeable reactive barrier, Master thesis, 
Worcester Polytechnic Institute (2011) 73 pp.  
[21] A.D. Henderson, A.H. Demond, Impact of solids formation and gas production on the 
permeability of ZVI PRBs, J. Environ. Eng. 137 (2011) 689–696. 
[22] A.S. Ruhl, A. Weber, M. Jekel, Influence of dissolved inorganic carbon and calcium on 
gas formation and accumulation in iron permeable reactive barriers, J. Contam. Hydrol. 
142–143 (2012) 22–32.  
[23] A.S. Ruhl, C. Kotré, U. Gernert, M. Jekel, Identification, quantification and localization 
of secondary minerals in mixed Fe0 fixed bed reactors, Chem. Eng. J. 172 (2011) 811–
816. 
 24
617 
618 
619 
620 
621 
622 
623 
624 
625 
626 
627 
628 
629 
630 
631 
632 
633 
634 
635 
636 
637 
638 
639 
640 
641 
[24] A.S. Ruhl, N. Ünal, M. Jekel, Evaluation of two-component Fe(0) fixed bed filters with 
porous materials for reductive dechlorination, Chem. Eng. J. 209 (2012) 401–406. 
[25] R. Crane, C. Noubactep, Elemental metals for environmental remediation: learning from 
hydrometallurgy, Fresenius Environ. Bull. 21 (2012) 1192–1196. 
[26] M. Stratmann, J. Müller, The mechanism of the oxygen reduction on rust-covered metal 
substrates, Corros. Sci. 36 (1994) 327–359. 
[27] F. Togue-Kamga, K.B.D. Btatkeu, C. Noubactep, P. Woafo, Metallic iron for 
environmental remediation: Back to textbooks, Fresenius Environ. Bull. 21 (2012) 
1992–1997. 
[28] J. Farrell, J. Wang, P. O'Day, M. Conklin, Electrochemical and spectroscopic study of 
arsenate removal from water using zero-valent iron media, Environ. Sci. Technol. 35 
(2001) 2026–2032.
[29] N.B. Pilling, R.E. Bedworth, The oxidation of metals at high temperatures, J. Inst. Metals 
29 (1923) 529–591. 
[30] 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. 38 
(2008) 1079–1091. 
[31] M. Rossi, O. Vidal, B. Wunder, F. Renard, Influence of time, temperature, confining 
pressure and fluid content on the experimental compaction of spherical grains, 
Tectonophysics 441 (2007) 47–65. 
[32] H. Jia, C. Wang, Adsorption and dechlorination of 2,4-dichlorophenol (2,4-DCP) on a 
multi-functional organo-smectite templated zero-valent iron composite Original, Chem. 
Eng. J. 191 (2012) 202–209.
[33] K. Miyajima, C. Noubactep, Effects of mixing granular iron with sand on the efficiency 
of methylene blue discoloration, Chem. Eng. J. 200–202 (2012) 433–438. 
 25
642 
643 
644 
645 
646 
647 
648 
649 
650 
651 
652 
653 
654 
655 
656 
657 
658 
659 
660 
661 
662 
663 
664 
665 
666 
667 
[34] S. Caré, R. Crane, P.S. Calabro, A. Ghauch, E. Temgoua, C. Noubactep, Modelling the 
permeability loss of metallic iron water filtration systems, Clean – Soil, Air, Water, doi: 
10.1002/clen.201200167. 
[35] G. Mitchell, P. Poole, H.D. Segrove, Adsorption of methylene blue by high-silica sands. 
Nature 176 (1955) 1025–1026. 
[36] S.B. Bukallah, M.A. Rauf, S.S. AlAli, Removal of methylene blue from aqueous solution 
by adsorption on sand, Dyes and Pigments 74 (2007) 85–87. 
[37] C. Varlikli, V. Bekiari, M. Kus, N. Boduroglu, I. Oner, P. Lianos, G. Lyberatos, S. Icli, 
Adsorption of dyes on Sahara desert sand, J. Hazard. Mater. 170 (2009) 27–34. 
[38] M. Cohen, The formation and properties of passive films on iron, Can. J. Chem. 37 
(1959) 286–291. 
[39] C. Noubactep, Characterizing the discoloration of methylene blue in Fe0/H2O systems, J. 
Hazard. Mater. 166 (2009) 79–87. 
[40] N.W. Barker, H.G. Linge, Methylene blue dye adsorption on sulphide minerals - 
Relevance to surface area measurement, Hydrometallurgy 6 (1981) 311–326. 
[41] P.G. Tratnyek, T.E. Reilkoff, A. Lemon, M.M. Scherer, B.A. Balko, L.M. Feik, B. 
Henegar, Visualizing redox chemistry: Probing environmental oxidation-reduction 
reactions with indicator dyes, Chem. Educator 6 (2001) 172–179. 
[42] R.L. Frost, Y. Xi, H. He, Synthesis, characterization of palygorskite supported zero-
valent iron and its application for methylene blue adsorption, J. Colloid Interface Sci. 
341 (2010) 153–161. 
[43] W.B. Fortune, M.G. Mellon, Determination of iron with o-phenanthroline: a 
spectrophotometric study, Ind. Eng. Chem. Anal. Ed. 10 (1938) 60–64. 
[44] M.R. Khan, S.I. Mozumder, A. Islam, D.M. Reddy Prasad, M. Mohibul Alam, 
Methylene blue adsorption onto Water Hyacinth: Batch and column study, Water Air 
Soil Pollut. 223 (2012) 2943–2953. 
 26
668 
669 
670 
671 
672 
673 
674 
675 
676 
677 
678 
679 
680 
681 
682 
683 
684 
685 
686 
687 
688 
689 
690 
691 
692 
693 
[45] N.N. Nassar, A. Ringsred, Rapid adsorption of methylene blue from aqueous solutions 
by goethite nanoadsorbents, Environ. Eng. Sci. 29 (2012) 790–797. 
[46] C. Noubactep, The fundamental mechanism of aqueous contaminant removal by metallic 
iron, Water SA 36 (2010) 663–670. 
[47] Y.N. Vodyanitskii, Iron hydroxides in soils: A review of publications, Eurasian Soil Sci. 
43  (2010) 1244–1254.. 
[48] C. Noubactep, T. Licha, T.B. Scott, M. Fall, M. Sauter, Exploring the influence of 
operational parameters on the reactivity of elemental iron materials, J. Hazard. Mater. 
172 (2009) 943–951. 
[49] K. Imamura, E. Ikeda, T. Nagayasu, T. Sakiyama, K. Nakanishi, Adsorption behavior of 
methylene blue and its congeners on a stainless steel surface, J. Colloid Interf. Sci. 245 
(2002) 50–57. 
[50] C. Noubactep, S. Caré, Designing laboratory metallic iron columns for better result 
comparability, J. Hazard. Mater. 189 (2011) 809–813. 
[51] K. Kümmerer, Emerging contaminants versus micro-pollutants, Clean – Soil, Air, Water 
39 (2011) 889–890. 
[52] K. Nödler, T. Licha, K. Bester, M. Sauter, Development of a multi-residue analytical 
method, based on liquid chromatography–tandem mass spectrometry, for the 
simultaneous determination of 46 micro-contaminants in aqueous samples, J. Chromat. 
A 1217 (2010) 6511–6521. 
[53] O. Hillebrand, K. Nödler, T. Licha, M. Sauter, T. Geyer, Caffeine as an indicator for the 
quantification of untreated wastewater in karst systems Original. Water Res. 46 (2012) 
395–402. 
[54] B. Saha, S. Das, J. Saikia, G. Das, Preferential and enhanced adsorption of different dyes 
on iron oxide nanoparticles: A comparative study, J. Phys. Chem. C 115 (2011) 8024–
8033. 
 27
694 
695 
696 
[55] S. Brenner, Sequences and consequences, Phil. Trans. R. Soc. B 365 (2010) 207–212. 
 
 28
Table 1: Parameters of the packed columns. Hsand,1 is the thickness of sand preceding the 
reactive layer; H
696 
697 
698 
699 
700 
701 
702 
rz is the thickness of the reactive layer and Hsand,2 is the thickness of 
sand after the reactive layer. The internal diameter of the column is 2.6 cm, its length 40 
cm as given by the manufacturer but the measured length was 44 cm. The effective 
volume of the column was 230 mL. The volume occupied by 100 g of Fe0 was used as 
unit for the mixture of Fe0 and sand. 
 
Column Fe0 Sand Fe0 Sand Hsand,1 Hrz Hsand,2
 (g) (g) (% v/v) (% w/w) (cm) (cm) (cm) 
1 0.0 - 0.0 100 44.0 0.0 0.0 
2 100 87 10.0 46.5 10.0 34.0 0.0 
3 100 77 20.0 43.5 11.0 20.0 13.0 
4 100 66 30.0 40.5 18.0 15.0 11.0 
5 100 58 40.0 36.6 19.0 13.0 12.0 
6 100 48 50.0 32.5 18.0 11.0 15.0 
7 100 29 70.0 22.4 15.0 7.5 21.5 
8 100 19 80.0 16.2 13.0 6.5 24.5 
9 100 0.0 100.0 0.0 13.0 6.0 25.0 
10 200 0.0 100.0 0.0 15.0 12.5 16.5 
 703 
704 
 29
Table 2: Summary of the results of the column experiments after 132 days. VT is the total 
volume of MB solution which has flowed through each column, Fe
704 
705 
706 
707 
708 
709 
710 
711 
712 
713 
effluent and 
MBeffluent are the cumulative mass of Fe and MB in the effluent. MBinfluent is the total 
mass of MB which has flowed through the columns. Feeffluent (%) is the ratio of 
dissolved Fe which has escaped from the column relative to the used mass of Fe0. E 
(%) and Es (mg/g) are the MB discoloration efficiencies (see text). Hcake is the 
thickness of the cake at the entrance of the Fe0-containing layer as measured at the 
end of the experiments. 
 
 
Column VT Feeffluent MBeffluent MBinfluent E Feeffluent Es Hcake
 (L) (mg) (mg) (mg) (%) (%) (mg/g) (cm) 
1 37.6 0.00 19.1 68.89 72.3 - - - 
2 37.8 47.08 36.4 69.34 47.5 0.047 0.218 2.0 
3 37.6 4.01 35.5 68.88 48.4 0.004 0.211 3.0 
4 37.8 17.82 25.5 69.39 63.3 0.018 0.239 3.0 
5 36.5 6.05 26.6 66.96 60.3 0.006 0.192 2.3 
6 37.5 8.40 28.6 68.73 58.4 0.008 0.201 1.5 
7 37.6 6.96 36.4 68.97 47.2 0.007 0.159 1.3 
8 37.5 9.71 40.0 68.74 41.9 0.010 0.143 1.1 
9 38.1 7.82 35.8 69.77 48.7 0.008 0.195 1.0 
10 37.8 4.91 22.0 69.37 68.3 0.002 0.154 1.0 
714 
715 
716 
 
 
 30
Table 3: Correlation parameters of the straight lines [MB]effluent = a*[MB]inffluent + b for the 10 
columns. “a” is the rate at which MB concentration increases in the effluent after 
breakthrough. “b” is the point at which each straight line MB meats the [MB]
716 
717 
718 
719 
720 
721 
effluent-
axis. R is the correlation factor and N is the number of experimental points used for 
the plot. Fig. 4 suggests that the smaller the a values, the more efficient the systems. 
 
Fe0 N R a δa b δb 
(%,vol) (-) (-) (-) (-) (-) (-) 
0 4 0.9881 0.31 0.03 -26.3 3.3 
10 14 0.999 0.64 0.01 -14.4 0.4 
20 10 0.998 0.63 0.02 -15.3 0.6 
30 8 0.997 0.55 0.02 -17.1 0.8 
40 8 0.996 0.56 0.02 -16.8 0.9 
50 8 0.998 0.60 0.02 -17.8 0.7 
70 12 0.999 0.65 0.01 -13.4 0.3 
80 12 0.999 0.74 0.01 -14.4 0.3 
100 11 0.998 0.65 0.01 -14.6 0.5 
100 6 0.999 0.65 0.01 -23.0 0.5 
722 
723 
724 
 
 
 31
Figure 1 724 
725 
726 
727 
728 
729 
730 
 
 
 
 
 
 20 30 % Fe0 0 10
0 
70 50 40 80 100 100 
 32
Figure 2 730 
731  
0 20 40 60 80 100
140
160
180
200
220
240
 Es
 E
Fe0 / [%]
E s
 / 
[μg
/g
]
30
40
50
60
70
80
E / [%
]
 732 
733 
 33
Figure 3 733 
0 20 40 60 80 100
0
2
4
6
8
10
Fe0 ration (vol %)
 0
 10
 20
 30
 40
 50
 70
 90
 100
 100 (200 g)Fe
 / 
[m
g/
L]
elapsed time / [days]
 734 
735 
 34
Figure 4  735 
736  
0 25 50 75 100 125
0
8
16
24
32
40 (a)
% Fe0 (v/v)
 10
 20
 30
 40
 50
 70
 80
 100ΣM
B
 / 
[m
g]
elapsed time / [days]
 737 
0 25 50 75 100 125
0
8
16
24
32
40 (b)
% Fe0 (v/v)
 0
 100 (100 g)
 100 (200 g)
ΣM
B
 / 
[m
g]
elapsed time / [days]
 738 
739 
740 
741 
742 
743 
 
 
 
 
 35
Figure 5 743 
744  
0 20 40 60 80 100
25
50
75
100
 sand 1
 observed
 sand 1 & 2
t B
T /
 [d
ay
s]
Fe0 ratio / [vol %]
 745 
746 
747 
 
 36
 Figure 6 747 
748  
0 25 50 75 100 125
0
4
8
12
16
% Fe0 (v/v)
 10
 20
 30
 40
 50
 70
 80
 100
flo
w
 v
el
oc
ity
 / 
[m
L/
h]
elapsed time / [days]
 749 
750 
 37
Figure 7750 
751  
 752 
753 
754 
 
 38
Figure 8 754 
755   
0 25 50 75 100 125
0
20
40
60
80
100 (a)
% Fe0 (v/v)
 10
 20
 30
 40
 50
 70
 80
 100
[M
B
]/[
M
B
] 0 
/ [
%
]
elapsed time / [days]
 756 
0 25 50 75 100 125
0
20
40
60
80
100 (b)
% Fe0 (v/v)
 0
 100 (100 g)
 100 (200 g)
[M
B
]/[
M
B
] 0 
/ [
%
]
elapsed time / [days]
 757 
758 
 39
Figure 9 758 
10 20 30 40
40
60
80
100 (a)
 55 days
 77 days
 132 daysE
 v
al
ue
 / 
[%
]
Hsand,1 / [cm]
 759 
0 10 20 30
40
60
80
100 (b)  55 days
 77 days
 132 days
E 
va
lu
e 
/ [
%
]
H rz / [cm ]
 760 
0 1 0 2 0
4 0
6 0
8 0
1 0 0
(c )
 5 5  d a y s
 7 7  d a y s
 1 3 2  d a y s
E 
va
lu
e 
/ [
%
]
H s a n d ,2  / [c m ]
 761 
762 
 40
Figure captions 762 
763 
764 
765 
766 
767 
768 
769 
770 
771 
772 
773 
774 
775 
776 
777 
778 
779 
780 
781 
782 
783 
784 
785 
786 
 
Figure 1: Photograph of the experimental design, showing the 10 columns at the end of the 
experiment (day 132). Uniform migration of the MB front is observed before the Fe0 layer 
(Hsand,1 - Tab. 1). MB breakthrough is observed in all systems. The absence of an MB 
adsorption front beyond the Fe0 layer evidences the disturbance of the flow regime within this 
layer. 
 
Figure 2: Effect of Fe0 amendment on the discoloration (E and Es) of methylene blue after 
132 days. The lines are not fitting functions, they simply connect points to facilitate 
visualization. 
 
Figure 3: Time-dependant evolution of the iron concentration of column effluent. The lines 
are not fitting functions, they simply connect points to facilitate visualization. 
 
Figure 4: Effect of Fe0 amendment on the performance of sand column for methylene blue 
discoloration: (a) 100 g of Fe0 occupying a volumetric proportion of 10 to 100 % of the 
reactive zone; and (b) 100 and 200 g of Fe0 in a pure Fe0 reactive zone. The lines are not 
fitting functions, they simply connect points to facilitate visualization. 
 
Figure 5: Comparison of model predictions of MB breakthrough with experimental 
observations after 97 days. The model assumes uniform MB adsorption onto sand 1 and sand 
2 (see Table 1). The lines are not fitting functions, they simply connect points to facilitate 
visualization. 
 
 41
Figure 6: Time-dependant evolution of the flow velocity curves in the 8 tested systems. The 
lines are not fitting functions, they simply connect points to facilitate visualization. 
787 
788 
789 
790 
791 
792 
793 
794 
795 
796 
797 
798 
799 
800 
801 
802 
803 
 
Figure 7: Cross-sectional diagram of the time-dependent evolution of the porosity in a 
Fe0/sand cylindrical column in which Fe0 experiences uniform corrosion. t , t , t  and t  
(t ) are the times at which porosity loss is 0, 50, 75 and 100 % respectively. As a rule the 
‘f
0 1/2 3/4 ∞
inf
ilter resistance’ increases with increasing porosity loss. 
 
Figure 8: Breakthrough curves of MB discoloration in the 8 tested systems (a) and in the 3 
pure material systems (b). The lines are not fitting functions, they simply connect points to 
facilitate visualization. 
 
Figure 9: Correlation of the cumulative extent of MB discoloration at days 55, 77 and 132 as 
function of the high of (a) sand before the reactive zone (Hsand,1), (b) the reactive zone (Hrz) 
and (c) sand after the reactive zone (Hsand,2). The reference system (0 % Fe0) corresponds to 
Hsand,1 = 44 cm or Hrz = Hsand,2 = 0 cm. For the 10 % Fe0 system, Hsand,2 = 0. 
 
 42