Journal of Molecular Liquids 299 (2020) 112144

Contents lists available at ScienceDirect

Journal of Molecular Liquids

journal homepage: www.elsevier.com/locate/molliq

Investigating the design parameters for a permeable reactive barrier

consisting of nanoscale zero-valent iron and bimetallic iron/copper for
phosphate removal
Osama Eljamal a,⁎, Ian P. Thompson b, Ibrahim Maamoun a, Tamer Shubair a, Kareman Eljamal c,
Khemmathin Lueangwattanapong b, Yuji Sugihara a
a
Environmental Fluid Science, Department of Earth System Science and Technology, Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, 6-1 Kasuga-Koen Kasuga, Fu-
kuoka 816-8580, Japan
b
Department of Engineering Science, University of Oxford, Parks Road, Oxford OX1 3PJ, UK
c
Department of Clinical Pharmacology, Graduate School of Medical Sciences, Kyushu University, Fukuoka, Japan

a r t i c l e i n f o a b s t r a c t

Article history: There is a growing interest in deploying nanoscale zero valent iron (NZVI) in permeable reactive barriers (PRBs)
Received 24 September 2019 for groundwater remediation. In the present study a series of packed-column experiments were conducted in
Received in revised form 11 November 2019 order to investigate the effectiveness of phosphorus removal from groundwater using NZVI and bimetallic
Accepted 15 November 2019
NZVI/Cu as reactive materials within PRBs. Seven sets of packed-column experiments were conducted in order
Available online 19 November 2019
to study the effect of different design parameters for PRB; including delivery approach of NZVI into porous
Keywords:
media, PRB's configuration, coexisting groundwater ions and change in flowrate. Results implied that doping
Nanoscale zero valent iron (NZVI) NZVI surface with copper had an anti-aggregation effect and enhanced its performance in terms of phosphorus
Bimetallic NZVI (NZVI/cu) removal 2.2 times higher than bare NZVI. Moreover, the lower flowrate (10 ml/min) demonstrated improved
Phosphorus removal phosphorus removal by 22% compared with higher flowrate (60 ml/min). Additionally, groundwater ions barely
Packed-column interfered phosphorus removal process with only ±6%. Overall, geochemical properties and characteristics of the
Groundwater remediation supporting materials were key parameters in the removal process of phosphorus by NZVI/Cu.
© 2019 Elsevier B.V. All rights reserved.

1. Introduction P-saturation is reached in the soil, high levels of phosphate could

leach to deeper layers of the soil and groundwater as it has been re-
Phosphorus (P) is a critical chemical element for the growth of mi- ported before in some parts of Northwestern Europe (0.10–3.3 mg/l)
croorganisms and plants as long. The main source of P is geological [10–12]. Moreover, that may represent an additional contribution to
phosphate which is expected to become exhausted within a century the nutrient leaching pathways to surface water bodies increasing the
[1,2]. However, its excessive and at times indiscriminate use in agricul- cause of the eutrophication. To counter this, several methods are avail-
tural fertilizers, presence in many industrial wastes, and landfill waste able for removing P from water including biological, adsorption, precip-
disposal are some of the causes for the presence of high concentrations itation, co-precipitation and crystallization [13,14]. Recently, as one of
of phosphorus in water systems [3,4]. Recovery of P from water has be- the most commonly used adsorbents for soluble contaminants, zero
come an essential task to solve P depletion problem and to reduce its valent iron (ZVI) has been employed in P removal from aqueous solu-
negative effects on the ecosystem. A high concentration of P in the tions [15,16]. At nano-scale ZVI represents a novel technology for sev-
water causes eutrophication which resulting in the depletion of oxygen eral environmental remediation applications such as the removal of
concentration that can adversely affect aquatic life [5,6]. Three forms of different pollutants from aqueous solutions [17]. Its unique properties,
dissolved P are detected in water; organic, ortho and poly phosphates such as the high reduction potential and the large specific surface area
[7–9]. Correspondingly, phosphate may enters groundwater in some provide it with the adequate sorption characteristics towards different
vulnerable areas, where subsoil is shallow (depth ≤ 90 cm), in a signifi- contaminants in water [18,19]. Moreover, NZVI has the additional ad-
cant amounts. Normally, soils have a definite capacity for the sorption of vantage that it is non-toxic and its core-shell structure feature, gives it
such presented phosphate in shallow groundwater. However, once the the ability to remove phosphorus from water employing different re-
moval mechanisms such as adsorption, precipitation and co-precipita-
tion [20]. Indeed its adsorption capacity for phosphorus especially
⁎ Corresponding author. with elevated concentrations compares very well to goethite [3,21], ka-
E-mail address: [email protected] (O. Eljamal). olinite [6], and activated carbon [22,23].

https://doi.org/10.1016/j.molliq.2019.112144
0167-7322/© 2019 Elsevier B.V. All rights reserved.
2 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144

In our previous study of phosphorus removal from aqueous solution (FeCl3.6H2O, 99.0%, Junsei Chemical Co., Japan) for NZVI synthesis. An-
by NZVI, we demonstrated that the high efficiency of phosphorus re- hydrous copper chloride (CuCl2, 99.9%, Aldrich Inc., USA) was used to
moval by NZVI (30 mg/g) and even higher efficiency (50 mg/g) in the synthesize NZVI/Cu. For phosphate solution preparation, potassium
presence of copper [24]. Almeelbi et al. also conducted a batch test to as- dihydrogen phosphate (KH2PO4 99.5%, Kanto Chemical Co., Japan) was
sist the NZVI ability to remove P from the aqueous solution, showing used. Moreover, sodium sulphate (Na2SO4, 99%, Wako Co., Japan), mag-
that the efficacy of the NZVI was 13.9 times higher than ZVI with P ad- nesium chloride hexahydrate (MgCl2.6H2O, 97%, Wako Co., Japan, so-
sorption capacity of 24.4 mg/g at pH 4.0 [13]. In another study, Tu et dium bicarbonate (NaHCO3, 99%, Wako Co., Japan), and calcium
al. investigated nano-scale bimetal ferrites for P removal from water, chloride dihydrate (CaCl2.2H2O, 70–79%, Junsei Chemical Co., Japan)
where the results indicated that the P adsorption capacity (13.5 mg/g) were used for simulated groundwater. Sodium hydroxide (NaOH, 97%,
at pH 2.6 [25]. Another study reported that using ZVI in packed columns Wako Co., Japan), hydrochloric acid (HCl, 35–37%, Wako Co., Japan),
on phosphate removal from aqueous solution demonstrated a homoge- and pH buffer solution (Sansyo Co., Japan) were used for pH adjustment.
neous distribution of P along the column with adsorption capacity All solutions were prepared using deionized water along with 20 min
132 mg/g [26]. deoxygenating by nitrogen gas. Furthermore, standard sand (ASONE
Despite its exceptional properties, NZVI still has some drawbacks Corp., Japan) and river sand (Kobaya, Miyazaki, Japan) were used as po-
when it comes to the real environmental remediation including the rous medium for column experiments.
poor mobility, particles agglomeration as well as the limited suspension
stability [27]. The aggregation of NZVI is mainly associated with the oxida- 2.2. Synthesis of NZVI
tion of its metal core and the deposition of corrosion products on its outer
surface. Such phenomenon is basically originated from magnetic attrac- NZVI was synthesized following the chemical reduction equation of
tion force between the particles as well as the Van Der Waals interactions ferric chloride by sodium borohydride [24,41]:
which result in larger size of NZVI, less reactive sites on the surface and
deteriorated remediation performance [28,29]. Nevertheless, such phe- 2FeCl3 þ 6NaBH4 þ 18H2 O→2Fe0 þ 21H2 þ 6BðOHÞ3 þ 6NaCl ð1Þ
nomenon could be useful in the partial recovery of the reacted NZVI
from the treated solution which can be achieved using magnetic field In order to synthesize 10 g of NZVI the following steps were under-
[30]. Hence, several modifications of the material have been studied in taken. Firstly, reductant solution was prepared by dissolving amount of
order to overcome such drawbacks. Hybridizing NZVI with the addition 35 g sodium borohydride in 1250 ml of deoxygenated deionized water
of other metals on its surface is known to improve the redox potential (DDIW). Secondly, 50 g of ferric chloride hexahydrate was dissolved in
of the composite and reduced aggregation potential. Several metals 1250 ml of DDIW to prepare the precursor solution which was kept in
have been used in the bimetallic NZVI composites such as palladium 5000 ml four-neck glass flask at constant temperature 25 ± 0.5 °C
(Pd) [31–33], Nickel (Ni) [34], and silver (Ag) [35]. Meanwhile, previous using water bath. Afterward, the reductant solution was pumped
studies have reported the usage of copper (Cu) with NZVI serving as a cat- drop-wisely into the synthesis flask at a rate of 1 L/h along with vigorous
alyst resulting in enhanced reactivity of the particles [36,37]. mechanical mixing at 250 rpm. The mixture was left for a 20 min aging
Several researchers have investigated the efficiency of NZVI and bi- time to ensure the reaction was complete. Furthermore, nitrogen gas
metallic NZVI/Cu in phosphorus removal. However, bench-scale tests was purged continuously into the mixture during the whole synthesis
are necessary to get some understanding of their performance in field to provide the anaerobic atmosphere. Finally, the resultant black precip-
conditions. Therefore, column tests using porous material are an infor- itates were separated using a vacuum filtration system consisted of a
mative intermediate step towards full-scale applications [38]. Most of vacuum pump for solution siphoning from the synthesis suspension.
the previous studies using ZVI/NZVI packed columns have focused on The NZVI particles were collected on a double-layer membrane filters
the removal of heavy metals from water [39,40]. Furthermore, there (Cellulose mixed ester, AS ONE Co., Japan) with pore diameter of 0.22
have been very few reports of the removal of P using ZVI packed-col- μm. The aggregated size of the synthesized NZVI allowed the particles
umn as technique to inform on the design of the PRB, and even less in- to be retained on the surface of the formerly mentioned membrane
formation on phosphorus removal using NZVI or NZVI/Cu and how the filter with the minimum solid losses. Nitrogen purging was continu-
effectiveness of these materials could be influenced by field condition, ously provided through the filtration process to prevent the possible
local porous materials and flow rates. oxidation of the collected particles. Thenceforth, the collected particles
In the present study, the main objective is to investigate different fac- were washed three times with DDIW before using directly in the
tors that affect the removal of phosphorus from groundwater by NZVI. experiments.
Therefore, seven sets of twelve column experiments were conducted con-
sidering several design conditions and their effects on phosphorus re- 2.3. Synthesis of NZVI/Cu
moval, including; the effect of modifying NZVI by cupper addition
(NZVI/Cu), the effect of changing the configuration of PRB layers (one/ The bimetallic nano iron particles were synthesized following the
double-layer), the effect of the delivery approach of NZVI into porous same formerly mentioned synthesis procedure of NZVI except for the
media (injection/packing), the interfering effect of groundwater ions, addition of CuCl2 to the ferric chloride solution prior the injection of re-
the effect of flowrate variance, and the reactive longevity effect. Samples ductant solution. The addition of CuCl2 was at rate of 0.5 g per 10 g of
were characterized using transmission electron microscopy (TEM) and NZVI (mass ratio), which was considered as the optimum addition
X-ray diffraction (XRD) for morphological and chemical composition rate for promoted efficiency [42]. The redox effect of the addition of
properties. Moreover, geochemical measurements such as pH, oxidation Cu2+ to NZVI was based on the following reaction:
reduction potential (ORP), and dissolved oxygen (DO) were recorded to
study their influence on phosphorus removal performance. Fe0 þ Cu2þ →Fe2þ þ Cu0 ð2Þ

2. Materials & methods

2.4. Characterization
2.1. Materials
The physical characteristics of both NZVI and NZVI/Cu were investi-
The following chemicals and materials were used in this study as gated. X-ray diffraction (XRD) analysis was performed in order to inves-
received without any pre-modification. Sodium borohydride (NaBH4, tigate the crystallinity and the exact composition of the synthesized
98.0%, Sigma–Aldrich Inc., USA), and ferric chloride hexahydrate particles. Samples were analyzed in X-ray diffractometer (XRD, TTR,
 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144 3

2.6. Column experiments

Column experiments were conducted in Plexiglas columns with di-

mensions of 65 cm length and 10 cm inner diameter (Fig. 1). Two
types of silica sand, river sand and standard sand, were used as a filling
porous material at the top and the bottom of the columns and also as a
supporting material for NZVI and NZVI/Cu in the reactive mixture at the
middle of the columns with (Fe/sand) mixing mass ratio of 0.01. The
mixing ratio of Fe/Sand was decided considering three crucial points;
a) unifying the NZVI or NZVI/Cu dosages in each column's set that stud-
ies one factor (for fair comparison), b) optimizing the Fe/Sand ratio
through set of batch experiments corresponding to higher removal effi-
ciency of phosphorus, and c) reasonable permeability of the PRB layer
that should not cause clogging problems during column experiments.
River sand was selected as a supporting material because of its high
drainage characteristics as well as the low cost. Moreover, unlike stan-
dard sand, river sand represented the real state of the aquifer material
Fig. 1. Schematic diagram of the column system. in field without any treatment of the impurities on the sand surface.
The configuration of the reactive layers as well as the operating condi-
tions was modified in order to investigate several factors that might af-
fect the performance of the presented PRB's design in phosphorus
Rigaku, Tokyo, Japan) using Cu Kα radiation (k = 1.5418 Å) operating at removal (Table 1). Therefore, the flowchart displayed in (Fig. 2) shows
40 kV and 200 mA with scanning rate of 3° min−1 and scanning angle the framework of the multiple conducted experiments and the consid-
ranged between 3° and 90° [24]. Furthermore, transmission electron ered factors for phosphorus removal through the columns. Column 1
microscopy (TEM, JEM-ARM 200F, JEOL Co., Japan) was used to investi- (C1) was integrated as a control column containing only river sand.
gate the particles morphology. Additionally, scanning electron micros- Key factors of the current PRB design were investigated as follows:
copy (JSM-6060LA/VI, JEOL Co., Japan) connected with energy
dispersive X-ray spectroscopy (EDS) was used to determine the ele- 1- The efficiency of modifying the surface of NZVI with the addition of
mental composition of the particles. In order to measure the Copper was examined where flowrate, supporting material, and
Brunauer–Emmett–Teller specific surface area (BET SSA) of the synthe- state of reactive material remained without change (C2, C4) and
sized particles, surface area analyzer (Micromeritics 3Flex, USA) was (C3, C5).
used based on N2 adsorption methodology at 77 K [36]. Moreover, par- 2- Secondly, the effect of multi-layer design of PRB on phosphorus re-
ticle size analyzer (SALD-2300, Shimadzu Co., Japan) was used to esti- moval was investigated (C2, C3) and (C4, C5).
mate the size of synthesized particles after being ultrasonicated in 3- Then, the influence of changing the reactive material state inside the
ethanol for 30 min. columns whether supported on sand, pure, or injected with a con-
trolled rate (C2, C6, and C7).
4- Based on the high performance of (C4) in the former experiments,
2.5. Analysis the efficiency of the two supporting types of sand in removal perfor-
mance was examined (C4, C8).
Phosphorus concentration in solution samples was colorimetri- 5- Moreover, operating with variable flowrate (Table 2) was performed
cally measured via UV–Vis spectrophotometer (DR 3900, Hach Co., in (C4, C9).
USA) using ascorbic acid method (USEPA PhosVer 3) at 880 nm 6- The interference of groundwater ions (Table 3) was simulated in (C4,
[43]. Additionally; pH, and oxidation reduction potential (ORP) mea- C10).
surements were obtained for samples and solutions using pH/ORP 7- Finally, the long-term experiments were conducted (C11, C12).
digital meter (D-72, LAQUAact, Horiba Co., USA). Dissolved oxygen
(DO) was also recorded using portable digital-meter (HQ30d flexi, Aqueous solution with initial phosphorus concentration of 25 mg/l
Hach Co., USA). was pumped downward into the columns using a peristaltic pump.

Table 1
Configuration of column experiments.

Column No of Layer depth Reactive Dosage State Supporting Flowrate Feeding Duration
no. layers (cm) material (g) material (ml/min) solution (days)
C1
– – – – – 15 14
(control)
C2 1 20 NZVI 20 Mixture 15 14
C3 2 10 NZVI 20 Mixture 15 14
C4 1 20 NZVI/Cu 10 Mixture River sand 15 14
C5 2 10 NZVI/Cu 10 Mixture 15 P-solution1 14
C6 1 – NZVI 10 Pure 15 14
10
C7 – – NZVI Injection 15 14
(2g/day)
C8 1 20 NZVI/Cu 10 Mixture Standard sand 15 14
C9 1 20 NZVI/Cu 10 Mixture 60–10 14
Simulated
C10 1 20 NZVI/Cu 10 Mixture 15 14
River sand GW2
C11 2 10 NZVI 10 Mixture 5 90
P-solution
C12 2 10 NZVI/Cu 10 Mixture 5 90
1
P-solution: phosphorus solution 2GW: Groundwater.
4 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144

Table 3
Chemical compositions of simulated groundwater used in C10.

Chemical compound Ca2+ Cl− Na+ SO2−

4 HCO−
3 Mg2+ P

Concentration (mg/l) 31.36 79.0 131.0 203.0 89.58 8.07 25.0

(pH = 7, ORP = −55 mV and DO = 10 mg/l)

Additionally, TEM investigation of synthesized NZVI showed a

chain-like aggregated structure and nearly spherical shape of the parti-
cles as displayed in Fig. 4a [47]. Meanwhile, due to copper addition, a
relatively separate distribution was observed for the spherical NZVI/Cu
Fig. 2. Framework for sets of column experiments and factors investigated.
particles in the exhibited microscopic morphology in Fig. 4b. Moreover,
the areas with brighter color in the images represented the thin iron
The chosen phosphorus concentration was decided based on a previous oxide/hydroxide outer shell on both of NZVI and NZVI/Cu surfaces.
optimization of NZVI-P initial conditions (data is not presented), as well Surface morphology of NZVI and NZVI/Cu particles was further ana-
as the commonly reported concentration in previous studies [26,44,45]. lyzed by SEM images. The irregular microscale of sand particles was
The initial pH, ORP, and DO values of the feeding P-solution were 7, clearly observed in the reactive sand mixture with either iron or bime-
−25 mV, and 8 mg/l respectively. The top sand layer (10 cm length) tallic iron (see Fig. S2a & b in Supplementary Materials). The EDS anal-
provided a uniform flow of the phosphate solution inside the column. ysis for spent samples of both NZVI and bimetallic NZVI sand mixtures
Moreover, mesh filters were used at the inlet and outlet of the columns was performed in order to determine the elemental composition. Si,
in order to prevent the inflow and the outflow of soil particles. Effluent Fe and O were the main elemental components in the EDS spectra of
treated water samples as well as samples along the columns were col- the spent NZVI-sand sample (Fig. 5a). Whereas, the same former ele-
lected periodically for phosphorus and concentration analysis along ments were detected as well as a small amount of copper (Cu) in the
with the daily measurements of pH, ORP, and DO of outlet samples. All spent NZVI/Cu-sand sample (Fig. 5b). Concurrently, silicon (Si) peaks
the column results are presented as average data of duplicate
experiments.

2.7. Statistical analysis

One-way ANOVA test was performed by GraphPad Software for

Windows (San Diego California USA) to evaluate the significant differ-
ence between the results obtained from columns, using 95% confidence
interval, i.e. p b 0.05 indicating statistical significance.

3. Results & discussion

3.1. Characterization of synthesized nanoparticles

The two peaks at 2θ = 44.8° and 82.6° in the XRD analysis of fresh
mixture samples (NZVI-sand and NZVI/Cu-sand) confirmed the pres-
ence of zero-valent iron (α-Fe0). Meanwhile, the weak peaks of hema-
tite (Fe2O3) in Fig. 3a indicated the presence of a relatively small
amount of iron oxides which contributed significantly in phosphorus
adsorption. Furthermore, the apparent peaks of copper ferrite
(CuFe2O4) in the spent samples of NZVI/Cu-sand mixture (Fig. 3b) en-
dorsed the co-precipitation of cupper ion on the surface of the bimetallic
iron particles following this reaction [46]:

CuO þ Fe2 O3 →CuFe2 O4 ð3Þ

The presence of iron oxides in the fresh mixture of either NZVI-sand

or NZVI/Cu could be mainly attributed to the possible rapid oxidation of
the NZVI particles during the mixing process or some oxide traces from
the sand material itself. However, in case of the fresh state of the NZVI
and NZVI/Cu there was no indication of oxidation as was observed in
XRD pattern in Fig. S1.

Table 2
Flowrate variation in C9.

Time (hr) Flowrate (ml/min)

0–12 60
12–24 40
24–36 30
36–48 15
Fig. 3. XRD patterns for fresh and spent reactive mixture samples of: (a) NZVI-sand, and
48-end of experiment 10
(b) NZVI/Cu-sand.
 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144 5

Fig. 4. TEM images of (a) NZVI and (b) NZVI/Cu.

were comparatively stronger than iron (Fe) peaks in the two EDS spec- 17.32 and 32.4 m2/g for NZVI and NZVI/Cu respectively, which was
trums which could be attributed to the Fe/sand mixing mass ratio of close to previously reported values [47–50]. The BET curves of N2-ad-
0.01. Moreover, weak phosphorus (P) peak was found in the EDS spec- sorption/desorption are presented in Fig. S3. The pore volume values
tra of the reacted samples confirming the adsorption process. Some of NZVI and NZVI/Cu based on BET measurements data were 0.0461
neglected traces of Al, Mg and Na was found in the samples. The SEM- and 0.0862 cm3/g respectively. Whereas, the measured values of aver-
EDS analysis for the fresh samples of NZVI and the bimetallic mixtures age pore size were 1.191 and 1.970 nm for NZVI and NZVI/Cu, respec-
is introduced in Fig. S1c & d respectively in the Supplementary Mate- tively, which is not so far from previously reported values [49].
rials. The BET measurements implied specific surface area values of Meanwhile, particle size analysis demonstrated a size distribution

Fig. 5. SEM and EDX spectra of: (a) NZVI-sand (spent), and (b) NZVI/Cu-sand (spent).
6 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144

within the range of about 13–70 and 15–117 nm for iron and bimetallic
iron particles respectively. Whereas, the average particle size was 44 nm Cu2 O þ H2 O→2CuO þ 2Hþ ð8Þ
for NZVI and 24 nm for NZVI/Cu.
The variation of effluent oxidation reduction potential (ORP) and
3.2. Column experiments dissolved oxygen (DO) values are exhibited in Figs. S5 and S6 respec-
tively. The results show that the effluent ORP values in NZVI/Cu columns
In order to investigate the effect of different column configurations (either one or double-layer design) were higher than that of NZVI col-
and operating conditions on the PRB's design in terms of phosphorus re- umns. In terms of the DO values, there were no significant (P =
moval efficiency was assessed by establishing different column configu- 0.1931 for C2 vs. C4) difference between the NZVI and NZVI/Cu columns
rations and operating was determined by laboratory microcosms (C1– except for the continuous fluctuation along the experiment time. How-
C12). The removal performance of phosphorus was assessed through ever, comparing to the control column (C1), the presence of either NZVI
several comparisons in terms of the investigated factors. Phosphorus or NZVI/Cu particles inside the other columns (C2–C5) had the main
adsorption capacity and the removal efficiency were determined for contribution in reducing the DO values due to the rapid aqueous corro-
comparing experiments performance. Phosphorus adsorption capacity sion and the formation of iron oxides (Fe2+/Fe3+) according to (Eqs. (9)
(qt) was calculated following this formula: and (10)). Iron measurements at the effluent samples in the column ex-
periments revealed that the 10 g of NZVI exhausted accumulative dis-
V ðC o −C t Þ solved ferrous and total iron with an average range of 0.9–1.1% and
qt ¼ ð4Þ
m 21–23%, respectively. Such released iron oxides range should not repre-
sent a deterioration in the drinking water quality standards. It is worth
where Co and Ct (mg/l) are initial and effluent phosphorus concentra- mentioning that the high initial DO value was adapted considering the
tion, m (g) is reactive material dosage, and V (L) is accumulated inflow state of the shallow aquifers for the application of PRBs, such alluvial
volume. While removal efficiency (R) was calculated based on (Eq. (4)): aquifers are considered to be a relatively oxic environment for sampling
which could result in rising DO concentration [51].
C o −C t
Rð%Þ ¼  100% ð5Þ
Co 2Fe0 þ 4Hþ þ O2 →2Fe2þ þ 2H2 O ð9Þ

Moreover, C1 was considered in all comparisons as the reference in

2Fe2þ þ 2H2 O→2Fe3þ þ 2H2 þ 2OH− ð10Þ
terms of performance for all column configurations.

3.2.1. NZVI modification (NZVI vs. NZVI/Cu) Phosphorus concentration profiles along the columns after 14 days
Phosphorus adsorption capacity of NZVI and NZVI/Cu was investi- (see Fig. S7a, b in Supplementary Materials) confirmed the positive ef-
gated in experiment of 14-day duration in one-layer 20 cm in depth fect of using the bimetallic NZVI on phosphorus removal the C4 and
contains 20 g of NZVI mixed with river sand (C2), 20 cm in depth con- C5 treatments comparing to the NZVI columns (C2, C3) and C1.
tains 10 g of NZVI/Cu mixed with river sand (C4) and double-layer
10 cm in depth for each contains 10 g of NZVI mixed with river sand
which are separated by 10 cm sand river layer (C3), 10 cm in depth
for each layer contains 5 g of NZVI/Cu mixed with river sand which
are separated by 10 cm sand river layer (C5) as shown in Fig. 6a and b
respectively. Results indicated that both NZVI and NZVI/Cu were effec-
tive in phosphorus removal from aqueous solution in porous media.
The plotted data depicted that the four former columns containing reac-
tive mixtures showed significantly higher phosphorus removal capacity
over the whole experiment time compared to C1 (P b 0.0001 for C2, C3,
C4 vs. C1) which reached saturation by the 10th day. Meanwhile, the bi-
metallic nano‑iron particles had a vital effect on phosphorus removal in
comparison with NZVI (P b 0.0001 for C2 vs. C4 and C3 vs. C5). For the
NZVI/Cu columns, maximum phosphorus adsorption capacity reached
about 380 and 345 (mg-P/g-Fe0) in C4 and C5, while the values were
lower than that in the NZVI columns (C2 and C3) with 173 and 223
mg-P/g-Fe0 respectively. Generally, the results confirmed that the addi-
tion of copper to NZVI resulted in doubling the phosphorus removal
performance. The general trend in regards all columns (C1-C5) was an
increase in the pH value (Fig. S4a, b). Concurrently, pH values in NZVI
columns (C2, C3) were slightly higher than that in C1 ranged between
8 and 8.5. Whereas, NZVI/Cu columns (C4, C5) had lower values com-
paring with C1. The comparative improvement in phosphorus adsorp-
tion capacity and the reduction in pH values in NZVI/Cu columns
could be interpreted by the reduction of the added copper ions (Cu2+)
to copper (Eq. (6)), following by oxidation and precipitation reactions
(Eqs. (7) and (8)) which had a notable effect on reducing solution pH
value due to the released H+ protons [36,42,43,46].

Cu2þ þ 2e− →Cu ð6Þ

Fig. 6. Effect of modifying reactive material on phosphorus adsorption in: (a) one-layer,
2Cu þ H2 O→Cu2 O þ 2Hþ ð7Þ and (b) double-layer columns.
 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144 7

3.2.2. Column configuration (one-layer vs. multi-layer)

In order to investigate the effect of one/multi-layer configuration of
PRB on phosphorus removal, adsorption capacity results for set no. 2 of
experiments was considered. Plotted data in Fig. 7a and b compares the
phosphorus adsorption capacity between the NZVI and NZVI/Cu respec-
tively, either one or double-layer column designs. For C2 and C3 (P =
0.2314 for C2 vs. C3), using NZVI as the reactive material in the mixture
inside the two columns had no detectable difference with the first three
days of the experiments. Afterward, the double-layer column design
(C3) showed a slightly better performance than (C2) with one-layer
configuration till the end of the experiments. In case of NZVI/Cu col-
umns, there was an unexpected increase in phosphorus adsorption ca-
pacity of one-layer column (C4) comparing with the double-layer
column (C5) within the first 9 days of the experiments. Later, the differ-
ence fluctuation for the two curves was insignificant for the rest of the
experiments (P = 0.8021 for C4 vs. C5). The slight improvement in Fig. 8. Effect of reactive material state on phosphorus adsorption; mixture of NZVI/river
phosphorus removal which accompanied the double-layer design sand layer (C2), pure NZVI layer (C6) and direct injection of pure NZVI (C7).
could be attributed to several possible explanations. Firstly, the pres-
ence of the separating sand layer between the two reactive layers en- ORP and DO values of column effluent samples are displayed in Figs.
hanced phosphorus co-precipitation with the released iron oxides (Fe2 S9 and S10 respectively. The results suggest that there was no detect-
+
/Fe3+) from the first reactive layer [44]. Moreover, the clogging effect able effect of the reactive layer design on either ORP or DO values except
due to the occupation of the surface with the adsorbed compounds as for the previously mentioned material effect. Column profiles of phos-
well as the precipitates was decreased through sand layer and the con- phorus concentration are presented in Fig. S11a, b (see the Supplemen-
tact time of phosphorus and reactive iron particles increased conse- tary Materials).
quently. Generally, the influence of multi-layer design of PRB was not
greatly observed on phosphorus removal performance. 3.2.3. Reactive material state (mixture vs. pure vs. injection)
Fig. S8a displays pH effluent results with respect to time for the two The reactive material state was investigated by the comparison be-
NZVI columns. As shown in the figure, the pH values for the two col- tween three column experiments; one-layer 20 cm in depth contained
umns (C2 and C3) fluctuated too much to detect a trend in terms of de- 20 g of NZVI mixed with river sand fixed between two layers of river
sign to the configuration. In contrast, pH values of C4 increased over the sand (C2), one-layer of 2 cm in depth containing 10 g of pure NZVI con-
first three days compared with the two-layer column (C5) as shown in fined between two layers of river sand (C6) and the direct injection of
Fig. S8b. The pH values tended to decrease in NZVI/Cu columns compar- 10 g pure NZVI at the top of column with a daily dose 2 g/day (C7).
ing to NZVI columns for the same reasons explained in the previous The effect of the NZVI state in the porous columns on phosphorus ad-
section. sorption capacity is shown in Fig. 8. Up to the second day of the exper-
iment, C6 which contained one pure layer of NZVI which was slightly
more effective in phosphorus removal. The direct injection of pure
NZVI at the top of column (C7) revealed a higher adsorption capacity
compared to the mixture column C2 (P b 0.0001 for C2 vs. C7) whilst
the direct injection (C7) demonstrated a slightly similar capacity to
the pure state column C6 (P = 0.1088 for C6 vs. C7). In general, the
change in reactive material status had an influence on phosphorus re-
moval performance compared to column with pure and injection col-
umns. Meanwhile, there were no observable significant differences
observed in the pH and ORP results presented in Fig. S12a, b except
for C6 in which pure NZVI showed lower values of pH and ORP com-
pared to the other columns. Furthermore, the variation of DO values
followed the same pattern (Fig. S12c). In contrast, Fig. S13 suggest
that phosphorus concentration along the injection column (C7) were
lower than that inside the other columns, which could be attributed to
the fresh state of the injected (as-synthesized) NZVI slurry in addition
to the possible surface oxidation of the particles during the packing pro-
cess in C2 and C6. Overall the three different states of NZVI were effec-
tive at removing phosphorus but the best performance is dictated by
the specific condition of the site. Nevertheless, it is suggested that the
reactive mixture (NZVI/sand) with supporting material would be pref-
erable for the long-term applications in terms of the cost and the high
porosity comparing to the pure NZVI and the injection.

3.2.4. Supporting material (river sand vs. standard sand)

In addition, the potential bimetallic NZVI was investigated in
supporting material (river sand/standard sand) in terms of its effective-
ness in removal of phosphorus. Two Columns packed with reactive layer
20 cm in depth containing 10 g of NZVI/Cu was mixed with river sand
(C4) or standard sand (C8). Fig. 9 shows phosphorus removal rate
within the two columns (C4 and C8) in comparison with C1. Results in-
Fig. 7. Multi-layer effect on phosphorus adsorption in: (a) NZVI, and (b) NZVI/Cu columns. dicated that phosphorus removal efficiency was 98% in all columns for
8 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144

Fig. 9. Effect of supporting material on phosphorus removal rate. Fig. 10. Effect of flowrate variation inside the columns on phosphorus removal rate.

almost the first two days. Afterward, the removal rate declined in the had an obvious positive effect on phosphorus removal rate owing to the
river sand column (C4), whereas the breakthrough time in the column increased contact time between the reactive mixture and the phospho-
with standard sand (C8) as well as C1 was only 1 day. In subsequent rus. Nevertheless, there was no significant influence of flowrate varia-
days, tC4 filled with river sand showed better performance with final tion on the effluent pH, ORP, and DO values in Fig. S16a, b, and c
phosphorus removal rate of about 50%, while it was around 14% in respectively. Generally, the residence time of the reactive material in-
case of column (C8) filled with standard sand. This suggests that the side the porous media seemed to be a critical factor in terms of the re-
river sand had nearly three-times the effectiveness in terms of phospho- moval efficiency of phosphorus.
rus removal rate than standard sand over the 14-day study (P = 0.1088
for C4 vs. C8).
The pH values revealed a steep increase at the beginning of the ex- 3.2.6. Feeding solution (simulated GW vs. P-solution)
periment reaching 8.6 in (C8) with standard sand (Fig. S14a) and then The sixth set of experiments was conducted in order to investigate
stabilized to a neutral value. Whilst, the column with river sand (C4) the degree of interference of groundwater ions on phosphorus removal
showed lower pH values ranging between 5.6 and 7. It is worth men- performance. The results of phosphorus adsorption capacity in Fig. 11
tioning that C1 had almost the same tendency of C4 except with a de- show that there was no detectable difference up to the first two days be-
crease in pH in the presence of NZVI/Cu. In the case of ORP results, tween C4 (with phosphate solution influent) and C10 (with simulated
standard sand showed a sharp decrease in ORP value in C8 to reach groundwater influent) (P = 0.1051 for C4 vs. C10). Afterward, the dif-
−120 mV within the first 12 h, thereafter the value increased steeply ference fluctuated to the end of the experiment whilst both showed a
fluctuating around −40 mV to end of the experiment (Fig. S14b). gradual increase. The interfering effect of groundwater ions was insig-
Whereas, the reduction rate of ORP values in C1 and C4 (containing nificant on phosphorus adsorption performance (±6%) which is corre-
river sand) was much lower than that of C8 (containing standard lated with other reported studies of unsupported NZVI [13]. Effluent
sand). Furthermore, DO results showed an expected decreasing in the pH results showed a slightly higher values in C10 than that of C4 (Fig.
two columns with NZVI/Cu (C4 and C8) compared to C1 (Fig. S14c). Ad- S17a), whereas an inverse effect was observed for ORP values (Fig.
ditionally, the vertical phosphorus profiles confirmed that standard S17b). This change may be attributed to the possible effect of ground-
sand had a lower performance in terms of phosphorus decline than water ions in response to the high pH and low ORP conditions. In case
river sand with respect to phosphorus removal (Fig. S15). Overall it of DO results, the reduction rate in the dissolved oxygen values with re-
could be concluded that standard sand as a supporting material in the spect to the influent values in C4 and C10 were similar (Fig. S17c). Phos-
reactive mixture was less effective at phosphorus removal and this phorus concentration profiles of this experiment are provided in the
was attributed to the increase of pH values which affected removal effi- supplementary materials (Fig. S18).
ciency. Moreover, the natural composition of the river sand which
contained traces of aluminum (Al), sodium (Na), and magnesium
(Mg), contributed acidic conditions which enhanced the phosphorus re-
moval process. Overall, these results imply that supporting material in
the reactive mixture is a crucial factor that is needed to be considered
in the PRBs design.

3.2.5. Flowrate (fixed vs. variable)

Flowrate in the columns was investigated as it has a significant effect
on the residence time of the reactive material inside the porous media.
Fig. 10 shows phosphorus removal efficiency under constant and vari-
able flowrates. The variation of the flowrate in C9 from 60 to
10 ml/min over the experiment resulted in lower phosphorus removal
rate compared to that of C4 with constant flowrate (15 ml/min) (P =
0.0402 for C4 vs. C9). The removal rate was significantly low corre-
sponding to the higher flowrate values in C9 within the first two days.
Afterward, there was a dramatic enhancement in phosphorus removal
after decreasing the flowrate to 10 ml/min. Over the second weeks of
the experiment, there was a slight improvement in C9 performance in
comparison with C4, indicating that the low flowrate inside the column Fig. 11. Interference effect of groundwater ions on phosphorus adsorption.
 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144 9

3.3. Removal mechanisms

Batch experiments were executed on the bare and bimetallic NZVI

for P removal from aqueous solutions (data are not presented). Hence,
from the conducted batch tests and the presented column experiments
the main possible interactions involved in P removal are as follows: a)
the corrosion of NZVI core by the release of iron oxides (Fe2+/Fe3+) fol-
lowing by the formation of the iron oxy/hydroxide layer which has a
great adsorption capacity for the aqueous phosphate species [13]. b)
the co-precipitation of the released Fe3+ with phosphate species (PO34
−
) on the surface of either NZVI or NZVI/Cu, as in Eq. (11) [3]. Corre-
spondingly, the presence of the FeP peaks in the XRD patterns of the
spent NZVI and NZVI/Cu, which are mainly associate to the formed
FePO4, confirmed the co-precipitation on the surface of the iron parti-
cles (Fig. 13). The currently used neutral pH conditions in this study,
Fig. 12. Phosphorus adsorption capacities in long-term column experiments.
which was lower than the point of zero charge pH (pHPZC) values for
both NZVI and NZVI/Cu, maintained the positive surface charge of the
particles that induced the electrostatic attraction of the anionic phos-
phate species to the particle's surface.

3.2.7. Long-term experiments (C11 vs. C12) Fe3þ þ PO4 3− →FePO4 ð11Þ
The long-term phosphorus removal performance of the proposed re-
active mixture (NZVI/sand and NZVI/Cu-sand) was investigated over
90 days' duration. The flowrate was adjusted to 5 mL/min in order to in- 4. Conclusions
crease the residence time of the reactive iron particles. Additionally, the
double-layer design reactive layer was included since previous results The main aim in this study was to propose a rational design of PRBs
suggested improvement in phosphorus removal when compared to for phosphorus removal in porous media by NZVI through a bench-scale
the single-layer design. Results in Fig. 12 indicate that there was no dif- column experiment. Several experimental factors were investigated
ference between the two columns in terms of phosphorus adsorption with respect to their influence on phosphorus removal. Employing ei-
capacity within the first 15 days of the experiment. Over the following ther bare NZVI or NZVI/Cu in the PRB's reactive mixtures resulted in a
month, C11 with NZVI/sand reactive mixture demonstrated greater great enhancement in phosphorus removal comparing with the control
phosphate removal than C12 with the bimetallic reactive mixture column. Surface modification of NZVI particles with copper resulted in a
reaching a maximum phosphorus adsorption capacity of 291 mg-P/g- smaller average size of the bimetallic particles (24 nm) comparing with
Fe0. Nevertheless, adsorption capacity in C11 then dropped significantly bare NZVI (44 nm). Additionally, TEM morphology investigation con-
during the last month and half while C12 showed a near stable removal firmed the anti-aggregation effect in NZVI/Cu by obvious discrete spher-
performance. The reactive mixtures inside the two columns needed ical particles comparing with the agglomerated clusters of bare NZVI.
about 3 months to lose their phosphorus removal capacity suggesting Results showed that maximum phosphorus adsorption capacity (mg-
the two columns were fully saturated once again with the initial phos- P/g-Fe0) increased by 54.47% and 35.36% in NZVI/Cu columns (C4 and
phorus concentration after 90 days of the experiment. Generally, the C5) respectively, comparing with bare NZVI columns (C2 and C3). A
long-term performance of both reactive mixtures was effective in phos- slight enhancement in phosphorus removal was observed in the dou-
phorus removal through porous media. The observed difference was not ble-layered PRBs comparing with the one-layer design which could be
enough to determine a clear preference of one to another as they both attributed to the effect of separating sand layer between the two reac-
showed a great potential on the long-term with an average phosphorus tive layers on either the co-precipitation removal mechanism or the
adsorption capacity over the 3 months of 143.34 and 128.76 mg-P/g-Fe0 clogging phenomena. Moreover, the comparison between different de-
for C11 and C12 respectively (P = 0.0845 for C11 vs. C12). livery methods of NZVI into PRBs demonstrated that injecting reactive

Fig. 13. Possible interaction in P removal by NZVI [5,45,46].

10 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144

material into porous media could be the most effective technique corre- [15] D. O'Carroll, B. Sleep, M. Krol, H. Boparai, C. Kocur, Nanoscale zero valent iron and
bimetallic particles for contaminated site remediation, Adv. Water Resour. 51
sponding to the highest overall phosphorus adsorption. However, in (2013) 104–122.
terms of cost and high porosity effect, sand supported NZVI mixture [16] J.-Y. Jeong, B.-M. Ahn, Y.-J. Kim, J.-Y. Park, Continuous phosphorus removal from
might be preferable for long-term applications. Moreover, results sug- water by physicochemical method using zero valent iron packed column, Water
Sci. Technol. 70 (2014) 895–900.
gested that river sand as a supporting material for NZVI in (C4) had a tri- [17] G. Vilardi, M. Stoller, N. Verdone, L. Di Palma, Production of nano zero valent iron
ple phosphorus removal rate over the 14 days experiment to that of particles by means of a spinning disk reactor, Chem. Eng. Trans. 57 (2017)
standard sand in (C8), confirming the preferable effect of using river 751–756, https://doi.org/10.3303/CET1757126.
[18] G. Vilardi, L. Di Palma, N. Verdone, A physical-based interpretation of mechanism
sand in NZVI reactive mixture. Retention time investigation indicated and kinetics of Cr(VI) reduction in aqueous solution by zero-valent iron nanoparti-
that low constant flowrate inside the column resulted in more contact cles, Chemosphere. 220 (2019) 590–599, https://doi.org/10.1016/j.chemosphere.
time between the reactive mixture and the solute phosphorus which 2018.12.175.
[19] G. Vilardi, L. Di Palma, N. Verdone, Competitive reaction modelling in aqueous sys-
had a clear positive effect on phosphorus removal rate comparing
tems: the case of contemporary reduction of dichromates and nitrates by nZVI,
with high or varied flowrates. The interfering effect of groundwater Chem. Eng. Trans. 60 (2017) 175–180, https://doi.org/10.3303/CET1760030.
ions was barely observed on phosphorus adsorption performance (± [20] T. Tosco, M.P. Papini, C.C. Viggi, R. Sethi, Nanoscale zerovalent iron particles for
6%) corresponding to the synthesized phosphate solution. Furthermore, groundwater remediation: a review, J. Clean. Prod. 77 (2014) 10–21.
[21] R. Chitrakar, S. Tezuka, A. Sonoda, K. Sakane, K. Ooi, T. Hirotsu, Phosphate ad-
pH conditions proved to be a key factor during phosphorus removal sorption on synthetic goethite and akaganeite, J. Colloid Interface Sci. 298
processes in the presence of either NZVI or the bimetallic iron particles (2006) 602–608.
where acidic conditions were favorable for better phosphorus ad- [22] L.-J. Yu, K. Rengasamy, K.-Y. Lim, L.-S. Tan, M. Tarawneh, Z.B. Zulkoffli, E.N. Se
Yong, Comparison of activated carbon and zeolites' filtering efficiency in fresh-
sorption. Over a 90 days experiment, the observed difference be- water, J. Environ. Chem. Eng. 7 (2019), 103223. https://doi.org/10.1016/j.jece.
tween NZVI and NZVI/Cu was not enough to determine a clear 2019.103223.
preference of one to another as they both showed a great potential [23] S. Hussain, H.A. Aziz, M.H. Isa, A. Ahmad, J. Van Leeuwen, L. Zou, S. Beecham, M.
Umar, Orthophosphate removal from domestic wastewater using limestone and
on the long-term with an average phosphorus adsorption capacity granular activated carbon, Desalination. 271 (2011) 265–272.
of 143.34 and 128.76 mg-P/g-Fe0 for C11 (with NZVI) and C12 [24] O. Eljamal, A.M.E. Khalil, Y. Sugihara, N. Matsunaga, Phosphorus removal from aque-
(with NZVI/Cu) respectively. Generally, further studies should be ous solution by nanoscale zero valent iron in the presence of copper chloride, Chem.
Eng. J. 293 (2016) 225–231, https://doi.org/10.1016/j.cej.2016.02.052.
performed based on an optimization model in order to integrate [25] Y.-J. Tu, C.-F. You, Phosphorus adsorption onto green synthesized nano-bimetal fer-
such presented results in the determination of the optimum design rites: equilibrium, kinetic and thermodynamic investigation, Chem. Eng. J. 251
and conditions of PRBs towards better performance in the real in- (2014) 285–292.
[26] N. Sleiman, V. Deluchat, M. Wazne, M. Mallet, A. Courtin-Nomade, V. Kazpard, M.
situ remediation applications.
Baudu, Phosphate removal from aqueous solution using ZVI/sand bed reactor: be-
havior and mechanism, Water Res. 99 (2016) 56–65.
Appendix A. Supplementary data [27] X. Guan, Y. Sun, H. Qin, J. Li, I.M.C. Lo, D. He, H. Dong, The limitations of applying
zero-valent iron technology in contaminants sequestration and the corresponding
countermeasures: the development in zero-valent iron technology in the last two
Supplementary data to this article can be found online at https://doi. decades (1994–2014), Water Res. 75 (2015) 224–248.
org/10.1016/j.molliq.2019.112144. [28] G. Vilardi, M. Parisi, N. Verdone, Simultaneous aggregation and oxidation of nZVI in
Rushton equipped agitated vessel: experimental and modelling, Powder Technol.
353 (2019) 238–246, https://doi.org/10.1016/j.powtec.2019.05.033.
[29] G. Vilardi, J. Rodríguez-Rodríguez, J.M. Ochando-Pulido, N. Verdone, A. Martinez-
References Ferez, L. Di Palma, Large laboratory-plant application for the treatment of a tannery
wastewater by Fenton oxidation: Fe(II) and nZVI catalysts comparison and kinetic
[1] D. Cordell, J.-O. Drangert, S. White, The story of phosphorus: global food security and modelling, Process. Saf. Environ. Prot. 117 (2018) 629–638, https://doi.org/10.
food for thought, Glob. Environ. Chang. 19 (2009) 292–305. 1016/j.psep.2018.06.007.
[2] J. Wei, J. Ge, A.A. Rouff, X. Wen, X. Meng, Y. Song, Phosphorus recovery from waste- [30] G. Vilardi, J. Rodriguez-Rodriguez, J. Miguel Ochando-Pulido, L. Di Palma, N. Verdone,
water using light calcined magnesite, effects of alkalinity and organic acids, J. Envi- Fixed-bed reactor scale-up and modelling for Cr(VI) removal using nano iron-based
ron. Chem. Eng. 7 (2019), 103334. https://doi.org/10.1016/j.jece.2019.103334. coated biomass as packing material, Chem. Eng. J. 361 (2019) 990–998, https://doi.
[3] H. Liu, T. Chen, X. Zou, Q. Xie, C. Qing, D. Chen, R.L. Frost, Removal of phosphorus org/10.1016/j.cej.2018.12.166.
using NZVI derived from reducing natural goethite, Chem. Eng. J. 234 (2013) [31] H.-L. Lien, W. Zhang, Hydrodechlorination of chlorinated ethanes by nanoscale Pd/
80–87, https://doi.org/10.1016/j.cej.2013.08.061. Fe bimetallic particles, J. Environ. Eng. 131 (2005) 4–10.
[4] Z. Wen, Y. Zhang, C. Dai, Removal of phosphate from aqueous solution using nano- [32] H.-L. Lien, W.-X. Zhang, Nanoscale Pd/Fe bimetallic particles: catalytic effects of pal-
scale zerovalent iron (nZVI), Colloids Surfaces A Physicochem. Eng. Asp. 457 (2014) ladium on hydrodechlorination, Appl. Catal. B Environ. 77 (2007) 110–116.
433–440. [33] X. Wang, C. Chen, H. Liu, J. Ma, Preparation and characterization of PAA/PVDF mem-
[5] H. Hauduc, I. Takács, S. Smith, A. Szabo, S. Murthy, G.T. Daigger, M. Spérandio, A dy- brane-immobilized Pd/Fe nanoparticles for dechlorination of trichloroacetic acid,
namic physicochemical model for chemical phosphorus removal, Water Res. 73 Water Res. 42 (2008) 4656–4664.
(2015) 157–170. [34] B. Schrick, J.L. Blough, A.D. Jones, T.E. Mallouk, Hydrodechlorination of trichloroeth-
[6] C.J. Penn, J.G. Warren, Investigating phosphorus sorption onto kaolinite using iso- ylene to hydrocarbons using bimetallic nickel−iron nanoparticles, Chem. Mater. 14
thermal titration calorimetry, Soil Sci. Soc. Am. J. 73 (2009) 560–568. (2002) 5140–5147.
[7] J. Lu, J. Yang, K. Xu, J. Hao, Y.-Y. Li, Phosphorus release from coprecipitants formed [35] Y. Xu, W. Zhang, Subcolloidal Fe/Ag particles for reductive dehalogenation of chlori-
during orthophosphate removal with Fe(III) salt coagulation: effects of pH, Eh, tem- nated benzenes, Ind. Eng. Chem. Res. 39 (2000) 2238–2244.
perature and aging time, J. Environ. Chem. Eng. 4 (2016) 3322–3329, https://doi. [36] Y. Zeng, H. Walker, Q. Zhu, Reduction of nitrate by NaY zeolite supported Fe, Cu/Fe
org/10.1016/j.jece.2016.07.005. and Mn/Fe nanoparticles, J. Hazard. Mater. 324 (2017) 605–616.
[8] O. Eljamal, J. Okawauchi, K. Hiramatsu, Product rich in phosphorus produced from [37] T. Shubair, O. Eljamal, A.M.E. Khalil, N. Matsunaga, Multilayer system of nanoscale
phosphorus-contaminated water, Adv. Mater. Res., Trans Tech Publ 2014, zero valent iron and Nano-Fe/Cu particles for nitrate removal in porous media,
pp. 261–265. Sep. Purif. Technol. 193 (2018) 242–254.
[9] N.Y. Mezenner, A. Bensmaili, Kinetics and thermodynamic study of phosphate ad- [38] G. Vilardi, J.M. Ochando-Pulido, M. Stoller, N. Verdone, L. Di Palma, Fenton oxidation
sorption on iron hydroxide-eggshell waste, Chem. Eng. J. 147 (2009) 87–96. and chromium recovery from tannery wastewater by means of iron-based coated
[10] A.O. Fadiran, S.C. Dlamini, A. Mavuso, A comparative study of the phosphate levels in biomass as heterogeneous catalyst in fixed-bed columns, Chem. Eng. J. 351 (2018)
some surface and ground water bodies of Swaziland, Bull. Chem. Soc. Ethiop. 22 1–11, https://doi.org/10.1016/j.cej.2018.06.095.
(2008) 197–206, https://doi.org/10.4314/bcse.v22i2.61286. [39] M.S.H. Mak, P. Rao, I.M.C. Lo, Zero-valent iron and iron oxide-coated sand as a com-
[11] L. Mabilde, S. De Neve, S. Sleutel, Regional analysis of groundwater phosphate con- bination for removal of co-present chromate and arsenate from groundwater with
centrations under acidic sandy soils: edaphic factors and water table strongly medi- humic acid, Environ. Pollut. 159 (2011) 377–382.
ate the soil P-groundwater P relation, J. Environ. Manag. 203 (2017) 429–438, [40] S.M. Ponder, J.G. Darab, T.E. Mallouk, Remediation of Cr (VI) and Pb (II) aqueous so-
https://doi.org/10.1016/j.jenvman.2017.07.058. lutions using supported, nanoscale zero-valent iron, Environ. Sci. Technol. 34 (2000)
[12] E.P.A. Ireland, Ireland's Environment, Environmental, Protection Agency Ireland 2564–2569.
2008 (2008). https://www.epa.ie/pubs/reports/indicators/irlenv/. [41] Y.-H. Hwang, D.-G. Kim, H.-S. Shin, Effects of synthesis conditions on the character-
[13] T. Almeelbi, A. Bezbaruah, Aqueous phosphate removal using nanoscale zero-valent istics and reactivity of nano scale zero valent iron, Appl. Catal. B Environ. 105 (2011)
iron, Nanotechnol. Sustain. Dev, Springer 2012, pp. 197–210. 144–150.
[14] O. Eljamal, J. Okawauchi, K. Hiramatsu, Removal of phosphorus from water using [42] A.M.E. Khalil, O. Eljamal, S. Jribi, N. Matsunaga, Promoting nitrate reduction kinetics
marble dust as sorbent material, J. Environ. Prot. (Irvine,. Calif) 03 (2012) by nanoscale zero valent iron in water via copper salt addition, Chem. Eng. J. 287
709–714, https://doi.org/10.4236/jep.2012.38084. (2016) 367–380, https://doi.org/10.1016/j.cej.2015.11.038.
 O. Eljamal et al. / Journal of Molecular Liquids 299 (2020) 112144 11

[43] W.G. Walter, Standard methods for the examination of water and wastewater (11th [48] Y. Liu, S.A. Majetich, R.D. Tilton, D.S. Sholl, G.V. Lowry, TCE dechlorination rates,
ed.), Am. J. Public Heal. Nations Heal. 51 (1961) https://doi.org/10.2105/ajph.51.6. pathways, and efficiency of nanoscale iron particles with different properties, Envi-
940-a(940–940). ron. Sci. Technol. 39 (2005) 1338–1345.
[44] I. Maamoun, O. Eljamal, A.M.E. Khalil, Y. Sugihara, N. Matsunaga, Phosphate removal [49] R. Yuvakkumar, V. Elango, V. Rajendran, N. Kannan, Preparation and characteriza-
through nano-zero-valent iron permeable reactive barrier; column experiment and tion of zero valent iron nanoparticles, Dig. J. Nanomater. Biostructures. 6 (2011)
reactive solute transport modeling, Transp. Porous Media 125 (2018) 395–412, 1771–1776.
https://doi.org/10.1007/s11242-018-1124-0. [50] T. Shubair, O. Eljamal, A.M.E. Khalil, A. Tahara, N. Matsunaga, Novel application of
[45] S. Takami, O. Eljamal, A.M.E. Khalil, R. Eljamal, N. Matsunaga, Development of con- nanoscale zero valent iron and bimetallic nano-Fe/Cu particles for the treatment
tinuous system based on nanoscale zero valent iron particles for phosphorus re- of cesium contaminated water, J. Environ. Chem. Eng. 6 (2018) 4253–4264,
moval, J. Japan Soc. Civ. Eng. 7 (2019) 30–42, https://doi.org/10.2208/ https://doi.org/10.1016/j.jece.2018.06.015.
journalofjsce.7.1_30. [51] A. Ramesh Kumar, P. Riyazuddin, Seasonal variation of redox species and redox po-
[46] A.J. Bard, L.R. Faulkner, J. Leddy, C.G. Zoski, Electrochemical Methods: Fundamentals tentials in shallow groundwater: a comparison of measured and calculated redox
and Applications, Wiley, New York, 1980. potentials, J. Hydrol. 444–445 (2012) 187–198, https://doi.org/10.1016/j.jhydrol.
[47] Y. Zhang, Y. Li, J. Li, L. Hu, X. Zheng, Enhanced removal of nitrate by a novel compos- 2012.04.018.
ite: nanoscale zero valent iron supported on pillared clay, Chem. Eng. J. 171 (2011)
526–531.