Journal of Hazardous Materials 271 (2014) 16–23

Contents lists available at ScienceDirect

Journal of Hazardous Materials

journal homepage: [Link]/locate/jhazmat

Immobilization of arsenate in a sandy loam soil using

starch-stabilized magnetite nanoparticles
Qiqi Liang a , Dongye Zhao a,b,∗
a
Environmental Engineering Program, Department of Civil Engineering, Auburn University, Auburn, AL 36849, USA
b
Institute of Environmental Science, Taiyuan University of Science and Technology, Shanxi, China

h i g h l i g h t s

• First study on starched magnetite nanoparticles for in situ immobilization of As(V).

• A small dose of the nanoparticles in soil greatly enhances soil sorption of As(V).
• The stabilized nanoparticles are deliverable in soil under moderate pressure.
• The nanoparticles are effective for in situ immobilization of As(V) in a model soil.
• Delivered nanoparticles will not spread under natural groundwater flow conditions.

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

Article history: This study investigated effectiveness of starch-stabilized magnetite nanoparticles for in situ enhanced
Received 25 September 2013 sorption and immobilization of arsenate, As(V), in a model sandy loam soil. Batch tests showed that the
Received in revised form 21 January 2014 nanoparticles offered an As(V) distribution coefficient of 10,000 L/g, which is >3 orders of magnitude
Accepted 31 January 2014
greater than that for the soil. Batch and column experimental results revealed that the nanoparticle
Available online 7 February 2014
treatment greatly reduced water-leachable As(V) and the leachability of As(V) remaining in the soil per
TCLP (Toxicity Characteristic Leaching Procedure) analysis. Column tests showed that water-leachable
Keywords:
As(V) from the As(V)-laden soil containing 31.45 mg/kg was reduced by ∼93% and the TCLP leachability
Metal immobilization
Magnetite
by >83% when the soil was treated with 34 pore volumes of a 0.1 g-Fe/L of the nanoparticle suspension.
Arsenic While the nanoparticles are deliverable in the soil, the effective travel distance of the nanoparticles can be
Nanoparticle manipulated by controlling the injection flow rate. Under natural groundwater flow conditions (velocity
Groundwater remediation ≤2.4 × 10−4 cm/s), the delivered nanoparticles are confined within a limited distance (<6.1 cm).
Starch stabilization © 2014 Elsevier B.V. All rights reserved.
Soil remediation

1. Introduction The predominant As species in groundwater are inorganic As(III)

and As(V) [6,7]. Although both species can be adsorbed on soil,
Arsenic (As) has been widely detected in soil and groundwa- they also undergo slow desorption/dissolution in the subsurface,
ter worldwide, particularly in the Bengal Delta and western United posing long-term threat to groundwater quality. Depending on the
States [1]. Based on analysis of 30,000 groundwater samples in biogeochemical conditions, three principal mechanisms have been
the U.S., ∼10% exceeded 10 ␮g/L [2]. Excessive levels of As were proposed for As mobilization: desorption in alkaline conditions,
detected at over 1/3 of the National Priorities List sites [3]. Expo- competitive sorption of co-ions, and reductive release [3]. Reduc-
sure to As in drinking water has been associated with development ing As(V) to As(III) increases As in groundwater [8], and dissolution
of cancers and non-carcinogenic effects [4,5]. To mitigate human of iron oxyhydroxide minerals causes release of As(V) [9].
exposure, U.S. EPA lowered the maximum contaminant level (MCL) Iron oxides have been known to offer high adsorption affinity
of As in drinking water from 50 to 10 ␮g/L effective in 2006. for both As(V) and As(III) [10–14]. Investigations using extended
X-ray absorption fine structure (EXAFS) [15,16], energy disper-
sive analysis of X-rays (EDAX) [17], and Fourier Transform Infrared
Spectroscopy (FTIR) [18] showed that As(V) forms strong inner-
sphere Fe–O–As complexes at the mineral surface.
∗ Corresponding author at: Environmental Engineering Program, Department of
Over the past decade, nanoscale adsorbents have gained strong
Civil Engineering, Auburn University, Auburn, AL 36849, USA. Tel.: +1 334 844 6277;
fax: +1 334 844 6290. momentum in water and soil remediation engineering [19–23].
E-mail address: zhaodon@[Link] (D. Zhao). Compared to the bulk counterparts, nanoparticles offer much

0304-3894/$ – see front matter © 2014 Elsevier B.V. All rights reserved.
[Link]
 Q. Liang, D. Zhao / Journal of Hazardous Materials 271 (2014) 16–23 17

greater specific surface area and greater adsorption capacity. comparison, the nanoparticle suspension was also prepared at
For example, iron oxide nanoparticles were found to offer 10 0.50 g-Fe/L with 0.4 wt.% of starch.
times greater sorption capacity than the micro-scale counterparts
[10,18]. In addition, well dispersed/stabilized nanoparticles offer 2.3. Preparation of As(V)-spiked soil and soil analysis
another unique advantage that the nanoparticles may be directly
delivered into contaminated soil or deep aquifers to facilitate in situ A sandy loam soil was obtained from the E.V. Smith Research
immobilization of target contaminants. For example, Xiong et al. Center (EVSRC) in Shorter, AL, Before use, the soil was sieved
[24] developed a class of CMC (carboxymethyl cellulose) stabi- through a standard sieve of 2 mm opening, and then washed with
lized FeS nanoparticles and found that the nanoparticles were able tap water to remove suspended colloids and water soluble compo-
to immobilize Hg2+ in situ. Xu and Zhao [25] demonstrated that sitions. The washed soil can be completely separated from water
CMC-stabilized ZVI nanoparticles were effective for in situ reduc- through centrifugation at 400 g-force. The soil was then air-dried
tive immobilization of Cr(VI) in a sandy loam soil. Compared to and stored in a sealed glass bottle. Soil analyses were performed
conventional technologies (e.g., permeable reactive barrier, pump- by the Soil Testing Laboratory at Auburn University. Table SM-1
and-treat, and excavation), the in situ immobilization approach is (Supplementary Materials) provides pertinent physical and chem-
not only potentially much more cost-effective, but is likely the only ical properties of the soil and the corresponding analytical methods
way to reach contaminant plumes in deep aquifers and sites that [28]. The pH at the point of zero salt effect (PZSE) was determined to
are not accessible by conventional technologies. be 6.0 following the potentiometric titration method [29] (Details
However, nanoparticles without a stabilizer tend to aggregate, and the titration curves are provided in SM, Section 2).
thereby losing soil deliverability. For example, Shipley et al. [26] As(V) was loaded to the soil following the batch adsorption pro-
tested non-stabilized magnetite and hematite nanoparticles for cedure by Yang et al. [30] and is described in SM, Section 3. The
arsenic immobilization in soil. Yet, physical mixing was required arsenic loading on the soil was determined to be 31.45 mg/kg by
to deliver the particles into the soil, which greatly limited the measuring the difference between the initial and final As(V) con-
feasibility for treating deep aquifers. Kanel et al. [27] tested centrations in the aqueous phase. The adsorbed arsenic was further
surfactant-modified ZVI nanoparticles, which showed improved verified via acid digestion of the soil per USEPA Method 3050B [31]
transportability when tested in sand-packed columns. and the difference was <5%. The As loading falls in the typical range
In our prior work, we synthesized a class of starch-stabilized of <0.1–97 mg/kg in U.S. soil according to the ToxGuide for Arsenic
magnetite nanoparticles, and preliminarily tested the nanopar- by the Agency for Toxic Substances and Disease Registry. The As(V)
ticles for arsenate removal in simulated groundwater [23]. The spiked soil was then air-dried for subsequent experimental uses.
starch-stabilized nanoparticles offered a 2.2 times greater As(V)
uptake capacity than conventional non-stabilized magnetite parti-
2.4. Immobilization of As(V) in soil: batch kinetic tests
cles. However, the effectiveness of the stabilized nanoparticles for
treating soil-sorbed As and related particle transport issues remain
Batch kinetic tests of As(V) leaching from the As(V)-laden
unexplored.
soil were carried out in the presence and absence of the starch-
Given the great potential of the starch-stabilized magnetite
stabilized nanoparticles. The tests were conducted using a batch
nanoparticles for in situ immobilization of arsenic in soil and
of Corning plastic (polyethylene tetraphthalate or PET) centrifuge
groundwater, this study aimed to: (1) probe the effectiveness of
tubes (53 mL). Typically, 5.3 g or 21.2 g of the As(V)-laden soil was
the stabilized nanoparticles for in situ immobilization of soil-sorbed
added to the centrifuge tubes each containing 53 mL of the nanopar-
As(V) through batch and column experiments; and (2) test and
ticle suspension containing 0.5 g-Fe/L of the nanoparticles (with
model the soil deliverability and transport behavior of the stabi-
0.4 wt.% starch), resulting in a soil-to-liquid ratio (SLR) of 0.1 and 0.4
lized nanoparticles.
(g/mL), respectively. The tubes were then capped and equilibrated
on an end-to-end rotator operated at 50 rpm at room tempera-
2. Materials and methods ture (22.0 ± 0.1 ◦ C). The solution pH was kept at 6.8 ± 0.4 through
intermittent adjustment. At pre-determined times, the tubes were
2.1. Chemicals sacrificially sampled by centrifuging the mixtures at 400 g-force for
20 min. The supernatants were then passed through the 25 nm Mil-
A potato starch (hydrolyzed for electrophoresis) and fer- lipore membrane filters, and the filtrates acidified to pH < 2.0 with
rous sulfate heptahydrate (FeSO4 ·7H2 O) were obtained from 0.1 N HNO3 and then analyzed for total As. The membrane was able
Acros Organics (Pittsburgh, PA, USA). Sodium arsenate heptahy- to completely remove the nanoparticles, but did not remove soluble
drate (Na2 HAsO4 ·7H2 O) was purchased from Sigma–Aldrich (St. As. All tests were conducted in duplicate.
Louis, MO, USA). Ferric chloride (FeCl3 ) and sodium hydroxide To determine the As(V) sorption capacity of the soil, arsenate
(NaOH) were obtained from Fisher Scientific (Pittsburgh, PA, USA). sorption isotherms were constructed for the soil through batch
Hydrochloric acid and nitric acid were purchased from Mallinck- equilibrium experiments (Supplementary Materials, Section 4).
rodt Chemical (St. Louis, MO, USA). All chemicals are of the ACS
reagent grade. 2.5. Toxicity characteristic leaching procedure (TCLP) tests

2.2. Preparation of starch-stabilized magnetite nanoparticles TCLP tests were performed following U.S. EPA Method 1311 [32]
to determine the leachability of As(V) in the soil, and the test pro-
Starch-stabilized magnetite nanoparticles were prepared fol- cedures are provided in Supplementary Materials (Section 5).
lowing the co-precipitation approach [23], and is described
in Section 1 of the Supplementary Materials (SM). Typically, 2.6. Soil sorption and transport of starch-stabilized Fe3 O4
the nanoparticle suspension was prepared at 0.10 g-Fe/L, which nanoparticles
was stabilized with 0.04 wt.% of starch. Our previous study
indicated that the TEM-based particle size and DLS-based hydro- Interactions between the soil particles and starch-stabilized
dynamic diameter were 75 ± 17 nm and 129 nm, respectively, magnetite nanoparticles were investigated through batch sorption
for the nanoparticles stabilized with 0.1 wt.% of starch [23]. For kinetic experiments (Supplementary Materials, Section 6).
18 Q. Liang, D. Zhao / Journal of Hazardous Materials 271 (2014) 16–23

To test the particle transportability, column breakthrough (a) 1200

tests were carried out. The experimental setup includes a Har- 0.4% starch ctrl (SLR=0.1)

vard Apparatus PHD 2000 syringe pump (Plymouth Meeting, PA, 1000
USA), a Plexiglas column (ID = 10 mm, length = 100 mm) (Omnifit,
Cambridge, UK), and a fraction collector (Eldex Laboratories, Napa,
800

As(V) ( g/L)
CA, USA). To facilitate a uniform soil packing and avoid air
entrapment, a nitrogen-sparged background solution (0.84 mM
NaCl + 0.16 mM NaHCO3 , pH 7.5) was pre-introduced into the col- 600

umn in the down-flow mode, and then 12 g of the soil was added
from the top into the solution while the column was constantly 400
patted with a plastic rod, yielding a porosity of 0.36 and a bulk
bed volume of 7.23 mL. Then, a stabilized Fe3 O4 nanoparticle sus- 200
pension (Fe = 0.1 g/L, starch = 0.04 wt.%) was pumped through the 0.5g Fe/L 0.4% starch magnetite
column in the down-flow mode at an EBCT (Empty Bed Contact
0
Time) of 118 min and at a pore water velocity of 3.5 × 10−3 cm/s. 0 50 100 150 200
The effluent was collected by the fraction collector, and the sam-
Time( hrs)
ples were then acidified by concentrated (12 M) HCl (4 mL acid for
each mL of the sample) to dissolve the nanoparticles. The concen-
tration of the nanoparticles was then determined by measuring (b) 3000
the total iron content. For comparison, the breakthrough behav-
0.4% starch ctrl (SLR=0.4)
ior of a tracer (50 mg/L Br− ) was also tested in parallel. Separate 2500
elution tests indicated that leaching of iron native in the soil was
negligible. 2000

As(V) ( g/L)
To determine the maximum transport distance of the nanopar-
ticles as a function of injection pore velocity, column breakthrough
1500
studies were carried out at four pore velocities (3.5 × 10−3 ,
1.8 × 10−3 , 3.5 × 10−4 , and 2.2 × 10−4 cm/s, respectively) to simu-
late different groundwater flow conditions. 1000

500
0.5g Fe/L 0.4% starch magnetite
2.7. In situ immobilization of As(V) in soil: column tests
0
0 50 100 150 200
The effectiveness of the stabilized nanoparticles for in situ
immobilization of As(V) in the soil was tested with the same col- Time( hrs)
umn setup. Briefly, 12 g of the As-laden soil was wet-packed in the
column with a porosity of 0.36. Then, an influent of 0.1 g-Fe/L of Fig. 1. Desorption kinetics of As(V) from an As-laden sandy loam in the absence
and presence of starch-stabilized magnetite nanoparticles (0.5 g-Fe/L with 0.4 wt.%
the nanoparticle suspension was passed through the soil bed at starch). The solid-to-liquid ratio was 0.1 in case (a) and 0.4 for case (b). Control
an EBCT of 118 min. Then, the effluent As(V) concentration was tests were conducted with 0.4 wt.% starch solution only. Suspension pH was kept
followed in two ways. First, the samples, which included both dis- at 6.8 ± 0.4. For all figures, data plotted as mean of duplicates, and errors refer to
solved and nanoparticle-associated As, were acidified at pH < 2.0 deviation from the mean.
with 1 N HNO3 and analyzed for total As; second, the effluent sam-
ples were first filtered through the 25 nm membrane to remove the 3. Results and discussion
nanoparticles and As sorbed thereon, and then, the filtrates were
acidified and analyzed for As, which gave the soluble As. For com- 3.1. As(V) immobilization: batch tests
parison, As elution history was also determined by subjecting the
identical soil bed to a simulated groundwater (SGW) with a chem- The effectiveness of stabilized Fe3 O4 nanoparticles for immobi-
ical composition of 7 mM NaCl + 0.86 mM CaSO4 (pH 6.5) under the lizing As(V) in the soil was tested in batch desorption experiments.
same hydrodynamic conditions. Fig. 1 compares arsenate desorption kinetics from the As-laden
To determine the spatial distribution of the retained nanopar- soil in the absence and presence of the nanoparticles at a SLR of
ticles along the soil bed following the nanoparticle treatment, the 0.1 and 0.4. When the soil was mixed with the 0.04 wt.% starch
soil bed was evenly sectioned into five layers. Each segment was solution without the nanoparticles, the equilibrium concentration
then air-dried and subjected to the EPA 3050B method for total Fe of As in the aqueous phase reached ∼1000 ␮g/L at SLR = 0.1 and
content. The nanoparticle concentration in each section was then ∼2200 ␮g/L at SLR = 0.4, respectively. In contrast, in the presence of
quantified by subtracting the background Fe content (1.11 ␮g-Fe/g- the nanoparticles, the water-soluble As concentration was reduced
soil) in the soil from the measured total Fe. by ∼98% in both cases. The distribution coefficient (Kd ) of arsenic
between the combined solid (soil + Fe3 O4 ) phase and water is cal-
culated to be 50 and 121 L/g at SLR = 0.1 and SLR = 0.4, respectively.
Separate sorption tests with the soil (no nanoparticles) revealed
2.8. Chemical analyses a nearly linear sorption isotherm (Fig. SM-2) with a mean Kd(soil)
of 2.6 L/kg over an equilibrium concentration range of 0–40 ␮g/L
Total As was analyzed using a PerkinElmer Graphite Atomic of As, which is in the typical Kd range of 0.3–4.3 L/kg [33]. Then,
Absorption Spectrometer 3110 with a detection limit of ∼5 ␮g/L. the distribution coefficient of As with respect to the nanoparticles,
Br− was analyzed using a Dionex Ion Chromatography (DX-120) Kd(Fe3 O4 ) , can be obtained via:
equipped with an AS14 column (detection limit = 0.5 mg/L). Total
iron was analyzed with a VARIAN 220FS Flame Atomic Absorption Ws
Kd(Fe3 O4 ) = (1)
Spectrometer (detection limit = 0.1 mg/L). WFe3 O4 (Kd − Kd(Soil) )
 Q. Liang, D. Zhao / Journal of Hazardous Materials 271 (2014) 16–23 19

400
1.0 KBr
350
0.040 wt.%

300 0.035 wt.%

0.8
As(V) conc, g/L

0.070 wt.%
250

Fe, C/C0
200 0.6

150 0.10 wt.%

0.4
100

50 0.2

0
ent ol
tm ntr 0.0
a Co
Tre 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
Pore Volume
Fig. 2. Arsenic concentrations in TCLP fluid #1 for an As-laden sandy loam with or
without the nanoparticle treatment. Fig. 3. Breakthrough curves of a tracer (Br− ) and Fe3 O4 nanoparticles prepared at
various concentrations of starch through a sandy loam bed. Experimental condi-
tions: Influent Br− = 50 mg/L, Fe3 O4 = 0.1 g-Fe/L, EBCT = 103 min, v = 0.001273 cm/s.
Symbols: Experimental data; lines: CXTFIT model simulations. Red squares corre-
where Ws and WFe3 O4 are the weights of soil and the nanoparti- spond to the 0.10 wt.% line.
cles, respectively. The resultant Kd(Fe3 O4 ) is 10,000 L/g (SLR = 0.1)
and 96,798 L/g (SLR = 0.4), which is more than three orders of mag-
nitude greater than Kd(soil) for the soil. This observation indicates C/C0 reached ∼0.68 before 13 BVs, but then, plunged to a steady
that adding a small fraction of the nanoparticles to the soil greatly level of ∼0.3 after 15 BVs.
enhances immobilization of As by transferring water leachable Based on the classical filtration theory [36], filtration removal
As from the soil onto the nanoparticles. Based on prior spectro- of fine particles in porous media involves two consecutive steps:
scopic studies [18,34], the enhanced As immobilization by the (i) mass transfer of particles to the matrix surfaces by Brownian
nanoparticles is attributed to the formation of the strong complexes diffusion, interception, and/or gravitational sedimentation, and (ii)
(Fe–O–As) between As(V) and the nanoparticles. deposition of particles to the matrix surface. Particle size, liquid
The TCLP approach has been commonly used to quantify the viscosity, media pore size and surface potential can affect mass
leachability of As in solid wastes under simulated landfill condi- transfer and particle-collector interactions. For examples, smaller
tions [35]. Fig. 2 shows that the nanoparticle amendment was able particles tend to have faster diffusion, but slower sedimentation
to reduce soluble As from 240 ␮g/L for the untreated soil to 46 ␮g/L, rate; Suspensions of higher viscosity tend to slow down the mass
a reduction of ∼81%. The results suggest that the nanoparticle treat- transfer process and may cause clogging of the soil pores. Our pre-
ment may also be applied to As-laden solid and hazardous waste vious study indicated that higher starch concentrations resulted
to convert a hazardous waste to a non-hazardous waste. in smaller nanoparticles with larger specific surface area [23]. At
1 h of particle aging, the hydrodynamic diameter was measured to
be 206, 172, 153, and 129 nm, respectively, for particles stabilized
3.2. Transport of starch-stabilized magnetite nanoparticles in soil with 0.035, 0.04, 0.07, and 0.10 wt.% of starch, and accordingly, the
suspension viscosity was measured to be 1.057, 1.063, 1.075 and
For in situ immobilization of arsenic in soil and groundwater, 1.086 cP. Although the particles prepared with 0.1 wt.% of starch
the nanoparticles must meet some fundamental requirements: (1) resulted in the smallest size, the gaining of faster mass transfer
they must be deliverable into the contaminated zones, and (2) once rate was diminished by elevated viscosity and partial clogging of
the external injection pressure is removed, the delivered nanopar- soil pores. It was also determined that the critical stabilization con-
ticles should remain within a confined domain (i.e., under natural centration (CSC, defined as the minimum concentration of starch
groundwater conditions, the delivered nanoparticles will serve as required for complete stabilization of nanoparticles) for 0.1 g/L as
an immobile sink for capturing soluble As). Fe of the nanoparticles was 0.04 wt.%. Because 0.035 wt.% of starch
Fig. 3 shows breakthrough curves of a tracer (bromide) and the was insufficient to fully stabilize the nanoparticles, more of the
starch-stabilized nanoparticles through the soil bed. The hydro- particles were retained by the soil. The increased removal of the
dynamic dispersion coefficient (D) was obtained by fitting the particles prepared at elevated starch concentrations (>0.07 wt.%)
classic 1-D convection-dispersion equation to the experimental are attributed to: (1) partial clogging of soil pores, resulting in more
breakthrough data of the tracer using the CXTFIT code (STAN- straining of the particles; and (2) increased collisions of the smaller
MOD software, USDA). The resultant D value was then applied to particles with collector surfaces, especially those not accessible by
modeling the breakthrough curves of the nanoparticles at various larger particles, favoring deposition of more particles [37].
starch concentrations. In all cases, the nanoparticles in the efflu- In addition, the surface potential also affects transport of the
ent reached a plateau after ∼15 PVs (pore volumes), whereas the nanoparticles. For bare magnetite particles, the pH at point of zero
full breakthrough occurred at ∼3 PVs for the tracer. Nanoparticles charge (PZC) was measured to be ∼6.1 [23]. The starch coating
prepared at various starch concentrations displayed different levels greatly suppressed the sensitivity of the  potential to the potential-
of the breakthrough plateaus (C/C0 ), with the nanoparticles stabi- determining ions (H+ /OH− ). For instance, between pH 5 and 8,
lized at 0.04 wt.% starch appearing most transportable (C/C0 = 0.91). the  potential varied from +20 to −40 mV for the bare particles,
At a starch level of 0.035 wt.%, more nanoparticles were removed whereas the range was narrowed down to −0.69 to −15.7 mV for
by the soil bed (C/C0 = 0.80). On the other hand, increasing the sta- 0.04 wt.% starch-stabilized nanoparticles. Noting that the pHZSE of
bilizer concentration to 0.07 wt.%, the level of C/C0 was lowered the soil also 6.1 (Fig. SM-1), the surfaces for both the collectors
to 0.73. Further increasing the starch concentration to 0.10 wt.% and nanoparticles are negatively charged at the experimental pH
resulted in a distorted breakthrough profile: the full breakthrough of 6.8. Consequently, nanoparticles stabilized with lower starch,
20 Q. Liang, D. Zhao / Journal of Hazardous Materials 271 (2014) 16–23

9 (a) 16000
Soil type: sandy soil
8 Arsenate content = 31.45 mg/kg
14000
EBCT = 117.8 min
Column height (cm)

7
12000
6

As(V), µg/L
5 10000
Total As w/ DI
4 8000 Total As w/SGW
Total As w/starch-Fe3O4
3
6000 Soluble As w/starch-Fe3O4
2
4000
1

0 2000
0 10 20 30 40 50
0
Fe distribution (mg/kg) 0 5 10 15 20 25 30 35 40 45

Fig. 4. Spatial distribution of starch-stabilized magnetite nanoparticles (measured Pore Volume

as Fe) that are retained in the soil bed following the As immobilization column run
of Fig. 5a. (b) 400

350
which are more negatively charge, undergo greater electrostatic
300
repulsive forces.

As(V) conc, µg/L

The software STANMOD has been widely employed to simu- 250
late solute transport in soil and groundwater [38]. The software
200
package includes a modified version of the CXTFIT code for estimat-
ing solute transport parameters following a nonlinear least-squares 150
parameter optimization method. The one-dimensional equilibrium
100
advection-dispersion equation is given by:
50
∂C ∂2 C ∂C
R =D 2 −v (2)
∂t ∂x ∂x 0
s h W ted
where C is the solute concentration in solution, x is distance, t is Fre SG ea
e tr
netit
time, D is the dispersion coefficient, v is the average pore water M a g
velocity, and R is the retardation factor, defined as:
Fig. 5. (a) Arsenic elution histories from an As(V)-laden sandy loam when sub-
K
R=1+ (3) jected to nanoparticle suspension, SGW or DI water. Nanoparticle suspension:
 Fe3 O4 = 0.1 g-Fe/L, starch = 0.04 wt.%; SGW: NaCl = 7 mM, CaSO4 = 0.86 mM. Pore vol-
where  is the soil bulk density, K is the distribution constant,  is ume = ratio of liquid volume through the column to void volume of the soil bed. (b)
As concentrations in TCLP fluid for an As-laden soil that was untreated, eluted with
the volumetric water content. SGW, or subjected to the nanoparticle amendment in (a).
The computer code was used to solve the governing transport
equation and to simulate the breakthrough curves of starch-
stabilized Fe3 O4 nanoparticles at various starch concentrations. immediate peaking during the first 5 PVs followed by a gradual
Fig. 3 shows that the model was able to adequately simulate the tailing. Based on mass balance, SGW and DI-water eluted 19% and
breakthrough curves except for the case of 0.10 wt.% starch. 11% of soil-sorbed arsenic, respectively, while the nanoparticle
Fig. 4 shows the spatial distribution of the nanoparticles retained suspension leached 6% during the first 5 PVs of elution. Further
in the soil bed following an As immobilization column run. The As examination of the effluent nanoparticle suspension revealed that
distribution profile mimics that of colloids retention in a typical 93% of the eluted As was associated with the nanoparticles, while
filtration process [39,40]. A concentration peak of the nanopar- all As eluted by SGW or DI-water was soluble. After ∼5 PVs of the
ticles was observed in the top layer of the soil column, while nanoparticle treatment, the soluble As(V) concentration reached a
nanoparticle retention was evident throughout the column. Similar steady level of <5 ␮g/L, while As in the SGW and DI-water efflu-
nanoparticle deposition phenomenon was also reported by other ents remained at ∼450 and >2000 ␮g/L, respectively. Compared to
researchers [41,42]. In this case, the full breakthrough C/C0 was the SGW elution, the nanoparticle treatment reduced the peak sol-
∼0.91. The results suggest that the nanoparticles can be well deliv- uble As by 90% from 13.7 mg/L to 1.32 mg/L, and the peak total
ered and distributed in the soil bed, and the particle retention and (soluble + nanoparticle-associated) As by 61% from 13.7 mg/L to
spatial distribution may be manipulated by tuning the injection 5.3 mg/L.
pressure/velocity and/or particle size as predicted by the filtration Hydrous oxides of transition metals such as Fe, Al and Cu in soil
theory. are known to bind with arsenate through specific interactions (e.g.,
Lewis acid-base interaction) [43]. Based on the soil analysis (Table
3.3. Immobilization of As(V) in soil: column tests SM-1), the soil contains 22 ppm of Al and 5 ppm of Fe, while other
transition metals are ≤ 2 ppm. The relatively low contents of the
The potential of starch-stabilized nanoparticles was tested metals in the soil may account for the fairly high water leachability
for in-situ immobilization of As(V) in the soil through column of As from the soil. Wenzel et al. [43] referred the most labile arsenic
experiments. Fig. 5a shows arsenate elution histories during two sorbed in a soil to as the exchangeable fraction (i.e., As sorbed
parallel column runs: one with SGW and the other with the through non-specific Columbic forces or outer sphere complexa-
starch-stabilized nanoparticle suspension under otherwise iden- tion). Noting that the SGW contains 0.86 mM of CaSO4 and that
tical conditions. In both cases, the elution curves displayed an SO4 2− offers greater affinity than non-specifically sorbed H2 AsO4
 Q. Liang, D. Zhao / Journal of Hazardous Materials 271 (2014) 16–23 21

Table 1
Column experimental conditions and calculated model parametersa for simulating the breakthrough curves of starch-stabilized magnetite nanoparticles.

Porous media Flow rate (mL/min) Pore velocity (cm/s) C/C0 D I × 10−5 G × 10−5 0 A ˛
−4
Sandy soil 0.0038 2.24 × 10 0 0.025 12.5 52.1 0.0259 6.46 6.46
0.006 3.54 × 10−4 0.51 0.018 11.9 31.4 0.0187 0.63 0.63
0.03 1.77 × 10−3 0.8 0.0058 9.69 5.26 0.0059 0.05 0.05
0.06 3.54 × 10−3 0.91 0.0035 8.89 2.44 0.0036 0.09 0.09

The parameters are summarized as follows:

Parameter Definition Physical interpretation

2(1− 5 )
As  = (1 − f)1/3 , f = porosity, porosity-dependent parameter
2−3+3 5 −2 6
dp
NR dc
Aspect ratio
Udc
NPe D∞
Peclet number characterizing ratio of convective transport to diffusive transport
NvdW A/kT van der Waals number characterizing ratio of van der Waals interaction energy to the particle’s thermal energy
a2 ( −f )g
p p
2
NG 9 U
dc
C Gravity number: ratio of Stokes particle settling velocity to approach velocity of the fluid
A − 23 (1−f )L0
ln C0
Attachment efficiency for given physicochemical conditions

Other parameters are defined as follows: dp is the particle diameter (171 nm), dc is the collector diameter (0.3 mm), U is the flow velocity, A is the Hamaker constant, k is
the Boltzmann constant (1.3805 × 10−23 J/K), T is fluid absolute temperature (298 K), ap is particle radius (85.5 nm), D∞ is the bulk diffusion coefficient (described by the
kT
Stokes–Einstein equation (D∞ = 6 a = 2.22 × 10−23 ), p is the particle density (5180 g/m3 ), f is the fluid density (997 g/m3 ),  is the absolute fluid viscosity (0.0011475 Pa s),
P
L is the filter medium packed length (L = 9.19 cm), and g is the gravitational acceleration (9.8 m/s2 ).
a
Note: The calculations are based on Tufenkji and Elimelech [39].
D = 2.4As NR−0.081 Npe
−0.715 0
1/3
NvdW · 052 (a)
I = 0.55As NR1 · 5sNP−0.12s NvdW
0
· 125 (b)
e
G = 0.22NR−0.24 NG1 · 11NvdW
0
· 053 (c).

− or HAsO4 2− [44] (which is evident from the 2.6 times difference distance of the nanoparticles)? And (2) Will the delivered nanopar-
in the As peak height in Fig. 5a), the high elution peak of As by ticles be mobile after the external pressure is released, and if so,
SGW can be attributed to sulfate exchange for the loosely-bonded how far can the nanoparticles (and associated arsenic) transport
arsenate species. under natural groundwater flow conditions?
During the 34 PVs of column run, ∼9% of the nanoparticles were To address these issues, transport modeling based on the filtra-
retained in the soil bed due to the filtration effect. The added Fe tion theory and more detailed experimental breakthrough studies
content to the soil increases the sorption capacity and affinity for of the starch-stabilized nanoparticles were carried out. Fig. 6 shows
arsenate, which can be easily revealed by the much greater Kd breakthrough histories of the nanoparticles at pore flow velocities
value for the nanoparticles than for the soil as illustrated earlier. from 3.5 × 10−3 cm/s to 2.2 × 10−4 cm/s. The pore velocities were
While the exchangeable fraction of As in the soil was leached out chosen to simulate the flow conditions during injection/delivery of
in both cases, the physical forms of leached As differed. While SGW the nanoparticles and after the injection pressure is released (i.e.,
leached As is fully soluble, the nanoparticle-leach As is firmly asso- under natural groundwater flow). Fig. 6 indicates that pore velocity
ciated with the effluent nanoparticles, which greatly reduces the strongly impacts the mobility, or soil retention, of the nanoparticles
bioavailability and mobility of As. in the soil. At the injection velocity of 3.5 × 10−3 cm/s, the nanopar-
TCLP tests were performed to probe the leachability of the resid- ticles were transportable with a full breakthrough C/C0 of ∼0.91.
ual As in the soil beds following the column elution tests in Fig. 5a. Halving the velocity to 1.8 × 10−3 cm/s decreased the C/C0 to 0.80.
For comparison, untreated As-laden soil samples were also tested. Further decreasing the velocity to 3.5 × 10−4 cm/s lowered the C/C0
Fig. 5b compares the leached As concentration in the TCLP solutions to 0.51. When the flow rate was reduced to a hypothetical ground-
for the soil samples. To facilitate a fair comparison, the weight of water flow velocity (2.2 × 10−4 cm/s), the nanoparticles became
the soil samples was adjusted to ensure that the As mass in the
TCLP tests was the same in all cases while maintaining the solid-
1.0
to-liquid ratio of 1 g to 20 mL as prescribed by the test protocol. The 3.5 10-3 cm/s
results indicate that the nanoparticle treatment reduced the TCLP 0.9
leachability of As in the soil by >83% compared to the SGW eluted 0.8
soil, and >90% to the untreated soil. Fig. 5a and b indicates that the
0.7 1.8 10-3 cm/s
nanoparticle amendment not only reduces the soluble As concen-
Total Fe, C/C0

tration in the downstream groundwater, but also diminishes the 0.6

leachability of the remaining As(V) in the soil phase.
0.5
3.5 10-4 cm/s
0.4
3.4. Maximum transport distance of starch-stabilized Fe3 O4
nanoparticles 0.3
0.2
The nanoparticle breakthrough curves (Fig. 3) and column
soil treatment data (Fig. 5a) indicate that the starch-stabilized 0.1 2.2 10-4 cm/s
nanoparticles are deliverable to the arsenic contaminated soil 0.0
under external pressure (i.e., at elevated pore flow velocities), and 0 5 10 15 20 25 30 35 40
most of As eluted was associated with the nanoparticles. From
Pore volume
the standpoint of As immobilization, two critical questions must
be addressed: (1) How far will the nanoparticles travel under the Fig. 6. Breakthrough histories of starch-stabilized magnetite nanoparticles through
external injection hydraulic gradient (i.e., what is the effective a soil bed at various pore velocities. Influent: Fe3 O4 = 0.1 g-Fe/L, starch = 0.04 wt.%.
22 Q. Liang, D. Zhao / Journal of Hazardous Materials 271 (2014) 16–23

5 4. Conclusions

To our knowledge, this has been the first study on the effec-
4 tiveness of starch-stabilized magnetite nanoparticles for in situ
Transport distance, m

immobilization of arsenic in soil and groundwater. The key findings

are summarized as follows:
3
(1) Starch-stabilized Fe3 O4 nanoparticles offer much greater affin-
ity and capacity for As(V) than soil. The As distribution
2 coefficient for the nanoparticles is >3 orders of magnitude
higher than that for the sandy loam soil.
(2) Batch tests indicated that addition of a small fraction of the
1
nanoparticles to soil greatly enhances sorption of As(V), thereby
reducing soluble arsenic concentration in groundwater. When
the As-laden soil was treated with 0.1 g-Fe/L of the nanoparti-
0
cles at a SLR of 0.1, water-leachable As was reduced by ∼98%
0 5e-4 1e-3 2e-3 2e-3 3e-3 3e-3 4e-3 4e-3
and the TCLP-leachable As by 80%
Pore velosity, cm/s (3) Column breakthrough studies confirmed that the stabilized
nanoparticles are deliverable in the model soil under moder-
Fig. 7. Maximum transport distance of starch-stabilized magnetite nanoparticles in
ate pressure. The transportability was found dependent on the
a sandy loam as a function of pore velocity.
stabilizer concentration, and the particles stabilized at a starch
concentration equal to the CSC value offered the highest mobil-
ity.
virtually immobile (C/C0 ≈ 0). Similar velocity effects have been
(4) Column tests revealed that the stabilized nanoparticles were
observed with CMC-stabilized ZVI nanoparticles [37] and fullerene
promising for in situ immobilization of As(V) in the model soil.
and oxide nanoparticles [45] in porous media. The results suggest
When an As-laden soil bed was treated with 34 PVs of the 0.1 g-
that the particle retention and transport may be manipulated by
Fe/L nanoparticle suspension, >93% of leachable As is converted
controlling the injection velocity, and the delivered nanoparticles
to nanoparticle-sorbed As, which is retained by soil once exter-
will remain in a confined domain under natural groundwater flow
nal pressure is released.
conditions.
(5) The effective travel distance of the stabilized Fe3 O4 nanoparti-
The deposition and transport of the nanoparticles are modeled
cles can be manipulated by controlling the injection pressure
in accord with the classical filtration theory. In this work, the mod-
or flow rate. Once delivered, the nanoparticles will remain
eling protocol by [37] was adopted to calculate the maximum travel
within a confined distance under typical groundwater condi-
distance of the nanoparticles as a function of the pore flow velocity:
tions, serving as a fixed sink for immobilization of arsenic in
2 dc groundwater.
Lmax = − ln(0.01) (4) (6) This work demonstrated great potential of the stabilized
3 (1 − f )˛0
nanoparticles, and the proposed in situ remediation technology
where Lmax is defined as the maximum travel distance over which may provide a powerful alternative for mitigating the health
99% removal of the nanoparticles occurs, dc is the collector grain impacts of arsenic in soil and groundwater, especially at sites
diameter, f is the bed porosity, ˛ is the attachment efficiency rep- where conventional technologies are inaccessible.
resenting the fraction of collisions between particles and collectors
that result in attachment, 0 is the overall single collector removal Acknowledgements
efficiency. ˛ is calculated via
The research was partially funded by the USDA-AAES Hatch and
 2 dc
C
Multistate program and National Natural Science Foundation of
˛= =− ln (5)
0 3 (1 − f )L0 C0 China (Grant 41230638).

where  is the single collector removal efficiency, and L is the length Appendix A. Supplementary data
of the soil bed (m). ␩0 can be broken into three key components
[39]: Supplementary material related to this article can be found,
in the online version, at [Link]
0 = D + I + G (6) 01.055.

where D , I , and G are single-collector contact efficiencies due References

to diffusion, interception, and gravitational sedimentation, respec-
tively. Table 1 summarizes the model calculations and resultant [1] M. Berg, H.C. Tran, T.C. Nguyen, H.V. Pham, R. Schertenleib, W. Giger, Arsenic
model parameters. Based on the simulation, Fig. 7 plots the max- contamination of groundwater and drinking water in Vietnam: a human health
threat, Environ. Sci. Technol. 35 (2001) 2621–2626.
imum travel distance as a function of pore velocity. Considering
[2] A.H. Welch, D.B. Westjohn, D.R. Helsel, R.B. Wanty, Arsenic in ground water
a high groundwater flow velocity of 10 ft./d (v = 3.5 × 10−3 cm/s) of the United States: occurrence and geochemistry, Ground Water 38 (2000)
(e.g., in sandy or gravelly aquifers) and a low velocity of 0.62 ft./d 589–604.
(v = 2.4 × 10−4 cm/s) [46], the estimated Lmax ranges from 4.5 m to [3] J.L. deLemos, B.C. Bostick, C.E. Renshaw, S. Sturup, X.H. Feng, Landfill-stimulated
iron reduction and arsenic release at the Coakley Superfund Site (NH), Environ.
6.1 cm. Fig. 7 indicates that for in situ remediation uses, the effective Sci. Technol. 40 (2006) 67–73.
area or particle travel distance may be manipulated by controlling [4] S. Kapaj, H. Peterson, K. Liber, P. Bhattacharya, Human health effects from
the injection flow rate/pressure. Once delivered, the nanoparticles chronic arsenic poisoning – a review, J. Environ. Sci. Heal. A 41 (2006)
2399–2428.
will remain within a confined soil domain under typical ground- [5] C.O. Abernathy, Y.P. Liu, D. Longfellow, H.V. Aposhian, B. Beck, B. Fowler, R.
water flow conditions. Goyer, R. Menzer, T. Rossman, C. Thompson, M. Waalkes, Arsenic: health effects,
 Q. Liang, D. Zhao / Journal of Hazardous Materials 271 (2014) 16–23 23

mechanisms of actions, and research issues, Environ. Health Persp. 107 (1999) [25] Y.H. Xu, D.Y. Zhao, Reductive immobilization of chromate in water and soil
593–597. using stabilized iron nanoparticles, Water Res. 41 (2007) 2101–2108.
[6] S. Dixit, J.G. Hering, Comparison of arsenic(V) and arsenic(III) sorption onto [26] H.J. Shipley, K.E. Engates, A.M. Guettner, Study of iron oxide nanoparticles in
iron oxide minerals: implications for arsenic mobility, Environ. Sci. Technol. 37 soil for remediation of arsenic, J. Nanopart. Res. 13 (2011) 2387–2397.
(2003) 4182–4189. [27] S.R. Kanel, D. Nepal, B. Manning, H. Choi, Transport of surface-modified iron
[7] P.L. Smedley, D.G. Kinniburgh, A review of the source, behaviour and distribu- nanoparticle in porous media and application to arsenic(III) remediation, J.
tion of arsenic in natural waters, Appl. Geochem. 17 (2002) 517–568. Nanopart. Res. 9 (2007) 725–735.
[8] P.H. Masscheleyn, R.D. Delaune, W.H. Patrick, Effect of reodx potential and [28] G.W. Gee, D. Or, Particle-size analysis: pipette method, J.H. Dane, G.C. Topp
pH on arsenic speciation and solubility in a contaminated soil, Environ. Sci. (Eds.), Methods of Soil Analysis (Part 4): Physical Methods, Soil Science Society
Technol. 25 (1991) 1414–1419. of America, Madison, Wisconsin, 2002, pp. 272–278.
[9] L.E. Deuel, A.R. Swoboda, Arsenic solubility in a reduced environment, Soil Sci. [29] E. Marcano-Martinez, M.B. McBride, Comparison of the titration and ion
Soc. Am. Proc. 36 (1972) 276–278. adsorption methods for surface charge measurement in oxisols, Soil Sci. Soc.
[10] S. Yean, L. Cong, C.T. Yavuz, J.T. Mayo, W.W. Yu, A.T. Kan, V.L. Colvin, M.B. Tom- Am. J. 53 (1989) 1040–1045.
son, Effect of magnetite particle size on adsorption and desorption of arsenite [30] J.K. Yang, M.O. Barnett, P.M. Jardine, N.T. Basta, S.W. Casteel, Adsorption,
and arsenate, J. Mater. Res. 20 (2005) 3255–3264. sequestration, and bioaccessibility of As(V) in soils, Environ. Sci. Technol. 36
[11] J. Gimenez, M. Martinez, J. de Pablo, M. Rovira, L. Duro, Arsenic sorption (2002) 4562–4569.
onto natural hematite, magnetite, and goethite, J. Hazard. Mater. 141 (2007) [31] USEPA, Acid Digestion of Sediments, Sludges, and Soils, Method 3050B, Wastes
575–580. – Hazardous Waste – Test Methods, U.S. Environmental Protection Agency,
[12] M. Jang, S.H. Min, T.H. Kim, J.K. Park, Removal of arsenite and arsenate using Washington, DC, 1996.
hydrous ferric oxide incorporated into naturally occurring porous diatomite, [32] USEPA, Toxicity Characteristic Leaching Procedure, Method 1311, Washington,
Environ. Sci. Technol. 40 (2006) 1636–1643. DC, 1992.
[13] K.P. Raven, A. Jain, R.H. Loeppert, Arsenite and arsenate adsorption on ferrihy- [33] J.D. Allison, T.L. Allison, Partition coefficients for metals in surface water, soil,
drite: kinetics, equilibrium, and adsorption envelopes, Environ. Sci. Technol. 32 and waste, EPA/600/R-05/074, U.S. Environmental Protection Agency, Wash-
(1998) 344–349. ington, DC, 2005.
[14] C.T. Yavuz, J.T. Mayo, W.W. Yu, A. Prakash, J.C. Falkner, S. Yean, L.L. Cong, [34] A. Jain, K.P. Raven, R.H. Loeppert, Arsenite and arsenate adsorption on ferri-
H.J. Shipley, A. Kan, M. Tomson, D. Natelson, V.L. Colvin, Low-field mag- hydrite: surface charge reduction and net OH− release stoichiometry, Environ.
netic separation of monodisperse Fe3 O4 nanocrystals, Science 314 (2006) Sci. Technol. 33 (1999) 1179–1184.
964–967. [35] A. Ghosh, M. Mukiibi, W. Ela, TCLP underestimates leaching of arsenic from solid
[15] X.J. Guo, Y.H. Du, F.H. Chen, H.S. Park, Y.N. Xie, Mechanism of removal of arsenic residuals under landfill conditions, Environ. Sci. Technol. 38 (2004) 4677–4682.
by bead cellulose loaded with iron oxyhydroxide (beta-FeOOH): EXAFS study, [36] R. Kretzschmar, M. Borkovec, D. Grolimund, M. Elimelech, Mobile subsur-
J. Colloid Interf. Sci. 314 (2007) 427–433. face colloids and their role in contaminant transport, Adv. Agron. 66 (1999)
[16] B.A. Manning, M.L. Hunt, C. Amrhein, J.A. Yarmoff, Arsenic(III) and arsenic(V) 121–193.
reactions with zerovalent iron corrosion products, Environ. Sci. Technol. 36 [37] F. He, M. Zhang, T.W. Qian, D.Y. Zhao, Transport of carboxymethyl cellulose sta-
(2002) 5455–5461. bilized iron nanoparticles in porous media: column experiments and modeling,
[17] S.R. Kanel, B. Manning, L. Charlet, H. Choi, Removal of arsenic(III) from J. Colloid Interf. Sci. 334 (2009) 96–102.
groundwater by nanoscale zero-valent iron, Environ. Sci. Technol. 39 (2005) [38] D.T. Feinstein, W.X. Guo, STANMOD: a suite of Windows-based programs for
1291–1298. evaluating solute transport, Ground Water 42 (2004) 482–487.
[18] B. An, Q.Q. Liang, D.Y. Zhao, Removal of arsenic(V) from spent ion exchange [39] N. Tufenkji, M. Elimelech, Deviation from the classical colloid filtration theory in
brine using a new class of starch-bridged magnetite nanoparticles, Water Res. the presence of repulsive DLVO interactions, Langmuir 20 (2004) 10818–10828.
45 (2011) 1961–1972. [40] J.Y. Chen, C.H. Ko, S. Bhattacharjee, M. Elimelech, Role of spatial distribution of
[19] C.M. Su, R.W. Puls, In situ remediation of arsenic in simulated groundwater porous medium surface charge heterogeneity in colloid transport, Colloid Surf.
using zerovalent iron: laboratory column tests on combined effects of phos- A 191 (2001) 3–15.
phate and silicate, Environ. Sci. Technol. 37 (2003) 2582–2587. [41] X.Y. Liu, M. Wazne, C. Christodoulatos, K.L. Jasinkiewicz, Aggregation and depo-
[20] F. He, D.Y. Zhao, J.C. Liu, C.B. Roberts, Stabilization of Fe-Pd nanoparticles with sition behavior of boron nanoparticles in porous media, J. Colloid Interface Sci.
sodium carboxymethyl cellulose for enhanced transport and dechlorination 330 (2009) 90–96.
of trichloroethylene in soil and groundwater, Ind. Eng. Chem. Res. 46 (2007) [42] J.J. Adler, Y.I. Rabinovich, B.M. Moudgil, Origins of the non-DLVO force between
29–34. glass surfaces in aqueous solution, J. Colloid Interface Sci. 237 (2001) 249–258.
[21] K. Hristovski, A. Baumgardner, P. Westerhoff, Selecting metal oxide nano- [43] W.W. Wenzel, N. Kirchbaumer, T. Prohaska, G. Stingeder, E. Lombi, D.C. Adri-
materials for arsenic removal in fixed bed columns: from nanopowders to ano, Arsenic fractionation in soils using an improved sequential extraction
aggregated nanoparticle media, J. Hazard. Mater. 147 (2007) 265–274. procedure, Anal. Chim. Acta 436 (2001) 309–323.
[22] H.J. Shipley, S. Yean, A.T. Kan, M.B. Tomson, Adsorption of arsenic to mag- [44] B. An, T.R. Steinwinder, D.Y. Zhao, Selective removal of arsenate from drinking
netite nanoparticles: effect of particle concentration, pH, ionic strength, and water using a polymeric ligand exchanger, Water Res. 39 (2005) 4993–5004.
temperature, Environ. Toxicol. Chem. 28 (2009) 509–515. [45] H.F. Lecoanet, M.R. Wiesner, Velocity effects on fullerene and oxide nanoparti-
[23] Q.Q. Liang, D.Y. Zhao, T.W. Qian, K. Freeland, Y.C. Feng, Effects of stabilizers and cle deposition in porous media, Environ. Sci. Technol. 38 (2004) 4377–4382.
water chemistry on arsenate sorption by polysaccharide-stabilized magnetite [46] W.M. Alley, T.E. Reilly, O.L. Franke, Sustainability of Ground-Water Resources
nanoparticles, Ind. Eng. Chem. Res. 51 (2012) 2407–2418. (U.S. Geological Survey Circular 1186), U.S. Geological Survey, Denver, Col-
[24] Z. Xiong, F. He, D.Y. Zhao, M.O. Barnett, Immobilization of mercury in sediment orado, 1999.
using stabilized iron sulfide nanoparticles, Water Res. 43 (2009) 5171–5179.