Tc

Da

Sb

h

•••••

a

ARRAA

KNPNPA

1

etCwpimwc

(

h0

Journal of Hazardous Materials 276 (2014) 271–277

Contents lists available at ScienceDirect

Journal of Hazardous Materials

j o ur nal ho me pa ge: www.elsev ier .com/ locate / jhazmat

race elements and nutrients adsorption onto nano-maghemite in aontaminated-soil solution: A geochemical/statistical approach

omingo Martínez-Fernándeza,∗, Deniz Bingölb, Michael Komáreka

Czech University of Life Sciences Prague, Faculty of Environmental Sciences, Department of Environmental Geosciences, Kamycká 1176, 165 21 Prague 6uchdol, Czech RepublicKocaeli University, Faculty of Science and Arts, Department of Chemistry, 41380 Kocaeli, Turkey

i g h l i g h t s

PO43− reduced the As and Cd adsorption capacity of NM in the soil solution.

High P level in the soil solution improved the immobilization of Al by NM.CCD was useful to explain the effects of K+ and NH4

+ on the adsorption of TEs by NM.High KNO3 and NH4NO3 concentrations reduced TEs levels in soil solution with NM.K+ and NH4

+ could improve the effectiveness of NM during phytoremediation tasks.

r t i c l e i n f o

rticle history:eceived 18 February 2014eceived in revised form 1 May 2014ccepted 14 May 2014vailable online 23 May 2014

eywords:ano-oxidesotassium

a b s t r a c t

Two experiments were carried out to study the competition for adsorption between trace elements(TEs) and nutrients following the application of nano-maghemite (NM) (iron nano-oxide; Fe2O3) to a soilsolution (the 0.01 mol L−1 CaCl2 extract of a TEs-contaminated soil). In the first, the nutrients K, N, and Pwere added to create a set of combinations: potential availability of TEs during their interaction with NMand nutrients were studied. In the second, response surface methodology was used to develop predictivemodels by central composite design (CCD) for competition between TEs and the nutrients K and N foradsorption onto NM. The addition of NM to the soil solution reduced specifically the concentrations ofavailable As and Cd, but the TE-adsorption capacity of NM decreased as the P concentration increased.The CCD provided more concise and valuable information, appropriate to estimate the behavior of NM

itrogen

hosphorusvailability

sequestering TEs: according to the suggested models, K+ and NH4+ were important factors for Ca, Fe,

Mg, Mn, Na, and Zn adsorption (R2adj

= 95%, except for Zn with R2adj

= 87%). The obtained information andmodels can be used to predict the effectiveness of NM for the stabilization of TEs, crucial during thephytoremediation of contaminated soils.

. Introduction

The bioremediation of soils contaminated by trace elements (TEs) through thestablishment of a vegetation cover (phytostabilisation) is a viable alternative forheir recovery and the conservation of the surrounding areas and groundwater.onditioning the soil is a key factor for the survival and growth of the plants,ith nanoparticles being promising materials for the stabilization of inorganicollutants [1,2], especially for multi-element-contaminated areas. Contaminant-

mmobilizing amendments decrease TEs bioavailability by inducing adsorption toineral surfaces, surface precipitation, and ion exchange [3]. Adsorption technologyith no chemical degradation is attractive due to its merits of effectiveness, effi-

iency and economy [4]. Adsorbents like activated carbons, zeolites, clays, industrial

∗ Corresponding author. Tel.: +420 224382663E-mail addresses: domingo.marfer@gmail.com, martinez-fernandez@fzp.czu.cz

D. Martínez-Fernández).

ttp://dx.doi.org/10.1016/j.jhazmat.2014.05.043304-3894/© 2014 Elsevier B.V. All rights reserved.

© 2014 Elsevier B.V. All rights reserved.

by-products, agricultural wastes, biomass or polymeric materials suffer from lowadsorption capacities and separation inconvenience [5]. Alternatively, the synthe-sis and utilization of iron oxide nanomaterials with novel properties and functionshas received much attention due to their surface modifiability, excellent superpara-magnetic properties, and great biocompatibility [6,7], and especially due to theirreactivity and relatively-large specific surface area (tens to hundreds of m2 g−1). Forthese reasons, iron nano-oxides are important scavengers of contaminants [8], beinga good choice for contaminated soils because of their sorption properties [9]. Named“eco-nanomagnet” by Girginova et al. [10], nano-maghemite (NM) has been investi-gated due to its abundance and effectiveness at removing the most-toxic form of As(arsenite, As3+) [11] and other elements, like Cu2+, Zn2+, and Pb2+ [12], from contam-inated water, and as an in situ remediation material [13]. Moreover the potentialavailability of nutrients and toxic TEs to living organisms will depend on the nature

and the strength of the interactions with the nano oxide surfaces, as well as theconditions for the interaction.

In complex systems, such as soil solution, there are ions other than toxic metalions that may affect the adsorption of metals. Therefore, there is a need to investigatewhether the effectiveness of an adsorbent is affected by the nutrient supply in the

272 D. Martínez-Fernández et al. / Journal of Haz

Table 1Characterization of the contaminated soil from Mokrsko (Czech Republic), with thetotal element concentrations, and its soil solution (extracted by 10 mmol L−1 CaCl2)used for the experiments.

Total digestion (mg kg−1) Soil solution (mg L−1)

pH 6.15 4.66Al 22,383 –As 389 0.11Ca – 423Cd b.d. 0.03Cu 21 0.04Fe 23,782 0.05K – 7.76Mg – 12.68Mn 499 1.10Na – 5.86

sttnunTFHaa

aaagmgt

2

<twncic(oiubattttt

2

iC4acw

pendent variables and they were factor levels coded as–1 (low),0 (middle), and +1 (high) (Table 3). Using MINITAB software ver-sion 16.1.1. [28], a CCD of 13 runs (in triplicate) was designed forKNO3 and NH4NO3 (from 0 to 6 mmol L−1). The same conditions

Table 2Concentration of each salt used to create a set of combinations of nutrients in theinteraction media used in the first experiment. The name of the combination denotesthe final concentration of each element in the soil solution (0, 3 or 6 mmol L−1). Theinteraction forms were K+ for K, NH4

+ and NO3− for N, and PO4

3− for P.

Name ofcombination

KNO3 (mmol L−1) NH4NO3 (mmol L−1) KH2PO4 (mmol L−1)

K0N0P0 0 0 0K0N3P0 0 1.5 0K0N6P0 0 3 0K3N0P3 0 0 3K3N3P0 3 0 0K3N3P3 0 1.5 3K3N6P0 3 1.5 0K3N6P3 0 3 3K6N0P6 0 0 6

Pb 26 –Zn 93 0.12

oil. On the other hand, low nutrient content is one of the most-limiting problems forhe growth of plants in contaminated soils [14] therefore, it is crucial to evaluate howhe application of nano-oxides might reduce the bioavailability of nutrients, and howutrients can alter their effectiveness. The related literature summarizes mainly these of nanoparticles for the cleanup of drinking-water and their competition withutrients: for example, As–P competition during Fe–Al nano-oxides adsorption [15],Es removal by zero-valent iron as affected by PO4

3− and NO3− [16], and the use of

e3O4 and Fe2O3 for the removal of either As (III) or As (V) from drinking water [17].owever, there is very little information about these processes in contaminated soilsnd soil solutions, the consequent changes in element concentrations following NMpplication, or the implications for plant nutrition and growth [18].

In this work, we study the effect of NM on the potential availability of nutrientsnd TEs in a contaminated soil solution, using a coupled geochemical/statisticalpproach: the results should have implications for phytoremediation technologiesnd for studies using NM in hydroponic culture. The nutrients potassium (K), nitro-en (N), and phosphorus (P) were selected because they are the most-importantacronutrients, and they are usually supplied to contaminated soil to improve plant

rowth during phytoremediation tasks. Metals and metalloids were analyzed, withhe focus on those with high total concentrations in the soil.

. Material and methods

The adsorption of TEs by nano-maghemite (NM; �-Fe2O3;50 nm nanopowder; Sigma–Aldrich) was tested in a soil solu-ion with/without NM and increasing concentrations of nutrients,ith the emphasis on the competition between contaminants andutrients for the sorption sites. Two experimental designs wereompared. The first used increasing concentrations of nutrientsn a soil solution, with NM interaction, and subsequent statisti-al comparison. The second followed a central composite designCCD) approach, which has several advantages over the traditionalne-variable-at-a-time approach. It was used to predict the behav-or of each element, since CCD is the most-popular class of designssed for fitting a second-order model [19]. The CCD approach cane used to determine the statistically-significant factors affecting

response, and to predict mathematically how a response relateso the values of various factors [20]. The mathematical models canhen be utilized to find the optimum response (through the statis-ical significance of various parameters) and to find the conditionshat result in a maximum or minimum value, as appropriate, withinhe experimental study.

.1. Soil solution

The Mokrsko–West mesothermal gold deposit is a former min-ng area affected by high TEs concentrations, located in centralzech Republic [21,22]. Samples from the top soil (0–25 cm;

◦ ′ ′′ ◦ ′ ′′

9 44 18 N 14 19 49 E) were collected, air-dried, sieved through 2-mm sieve, and homogenized to determine selected initial soilharacteristics (Table 1). The pH and electrical conductivity (EC)ere measured in saturated soil/water pastes (1:3, w/w). Cations

ardous Materials 276 (2014) 271–277

were measured by inductively-coupled plasma-optical emissionspectrometry (ICP-OES) (Varian, VistaPro, Australia). A soil solutionwas obtained by mixing 10 g of soil with 100 mL of 10 mmol L−1

CaCl2, to extract the potentially-available TEs in the soil fraction[23]. In order to maintain the pH following the addition of the dif-ferent nutrients, MES buffer (C6H13NO4S) was added to the soilsolution at a concentration of 1.5 mmol L−1.

2.2. K, N, and P during TEs adsorption by NM

In the first experiment, KNO3, NH4NO3, and KH2PO4 were usedto add K, N, and P, respectively, to the soil solution before additionof NM. These compounds are typically used for nutritive hydro-ponic solutions at concentrations of 2–4 mmol L−1 [24–26]. Stocksolutions of 0.1 mol L−1 were prepared for each salt. In order tohave three concentrations of each nutrient (0, 3, and 6 mmol L−1),different combinations of volumes of the stocks were added tothe soil solution (diluted), in 10-ml tubes, in triplicate. For this,since K and N are common to more than one compound, a set ofcombinations was designed to study the individual effects of eachelement (Table 2). The software Visual MINTEQ [27] showed thatin these solutions more than 98.8% of the K, N, and P were availablein free forms, as K+, NO3

−, NH4+, and PO4

3−. After the nutrientswere mixed with the soil solution, 1 ml of 10% w/v NM in purewater was added to each tube to obtain 1% w/v NM in the soil-nutrients-solution, in the final volume of 10 ml. These dose wasselected because usually, 1–2 wt.% of nano-amendment is suffi-cient for metal stabilization [9]. The resultant pH of the solutionwas 4.66. Controls without NM (1 mL of pure water added to eachtube) were used to determine the effect of NM on the equilibrium.After agitation for 24 h, in darkness, the pH was determined and thesamples were centrifuged and filtered using 0.45-�m nylon filters,which were efficient due to the aggregation of the nano particles(not detected in the filtrates). Samples were diluted using HNO3and analyzed using ICP-OES (Varian, VistaPro, Australia).

2.3. Central composite design (CCD) for K and N

In this experiment, the response surfaces methodology (RSM)was employed to investigate the influence of K+ and NH4

+ (addedas KNO3 and NH4NO3) on the bioavailability of nutrients and TEsafter NM application to the soil solution. The nutrients were inde-

K6N3P6 0 1.5 6K6N6P0 6 0 0K6N6P3 3 1.5 3K6N6P6 0 3 6

D. Martínez-Fernández et al. / Journal of Hazardous Materials 276 (2014) 271–277 273

Table 3Design matrix and experimental results based on the central composite design (CCD).

Nutrient Symbol Coded variable levels

Lowest Low Middle High Highest–� (−

√2) –1 0 +1 +� (+

√2)

KNO3 (mmol L−1) X1 0 0.9 3 5.1 6NH4NO3 (mmol L−1) X2 0 0.9 3 5.1 6

Run Coded level of variables Actual level of variables Response (as mean of triplicate)

X1 X2 (mmol L−1) (mg L−1) (�g L−1)

KNO3 NH4NO3 Ca Mg Na Mn Fe Zn

1 −1 −1 0.9 0.9 45.51 1.42 0.95 0.114 4.7 33.52 +1 −1 5.1 0.9 41.90 1.23 0.87 0.106 4.4 21.03 –1 +1 0.9 5.1 43.56 1.29 0.80 0.110 6.7 34.94 +1 +1 5.1 5.1 39.72 1.17 0.84 0.101 3.5 20.25 −√

2 0 0.0 3.0 45.55 1.37 0.74 0.115 4.1 20.36 +

√2 0 6.0 3.0 39.76 1.17 0.86 0.101 3.4 18.9

7 0 −√2 3.0 0.0 43.77 1.29 0.84 0.110 4.7 18.5

8 0 +√

2 3.0 6.0 40.68 1.20 0.80 0.103 3.3 29.59 0 0 3.0 3.0 42.19 1.24 0.83 0.106 3.8 18.010 0 0 3.0 3.0 42.39 1.25 0.83 0.107 4.0 20.8

42.242.342.2

aitwtdat

y

wa(idtcf

x

wr

2

ctrS1wIco

11 0 0 3.0 3.0

12 0 0 3.0 3.0

13 0 0 3.0 3.0

s in the first experiment were used (10-ml tubes, 24 h agitationn darkness, centrifugation, filtration, pH measurement, acidifica-ion, ICP-OES analyses). Accordingly, adsorption surface isothermsere obtained for TEs in the presence of NM and nutrients, simul-

aneously, to graphically illustrate the relationship between theifferent experimental variables and the responses. To determinen optimum, it was necessary that the polynomial function (y) con-ained quadratic terms:

= ˇ0 +k∑

i=1

ˇixi +k∑

i=1

ˇiix2i +

k∑

1≤i≤j

ˇijxixj + ε (1)

here x1, x2, . . ., xk are the independent variables, which haven influence on the response y; ˇ0, ˇi (i = 1, 2, . . ., k), ˇij, and ˇiji = 1, 2, . . ., k; j = 1, 2, . . ., k) are the intercept, linear, quadratic, andnteraction constant coefficients, respectively, and ε is either a ran-om error or allows for the description or uncertainties betweenhe predicted and measured value. For statistical calculations, thehosen independent variables Xi were coded as xi according to theollowing relationship:

i = Xi − X0

�x(2)

here X0 is the uncoded value of Xi at the center point and �xepresents the step change [19,29].

.4. Statistical analysis

Analysis of variance (ANOVA) and Student’s t test (95% ofonfidence interval) were used to determine the significance ofreatment effects for the comparison of three and two means,espectively, with post hoc analysis by Tukey’s test (IBM-SPSStatistics 19 software). When necessary, values were transformed/x to satisfy normality and variance homogeneity tests. ANOVA

as used to evaluate the fit of the mathematical models in the RSM.

n order to compare significance of each element on the same scale,oded coefficients were used in the RSM to make sure that the rangef each of the factors was the same [20].

0 1.24 0.82 0.107 4.4 14.99 1.25 0.83 0.107 4.0 20.80 1.24 0.82 0.107 44 14.9

3. Results

3.1. Effects of K, N, and P on adsorption by NM

The addition of NM to the contaminated soil solution reducedsignificantly the concentrations of available As and Cd in the equi-librium (Table 4A), by 28.5% and 6.2%, respectively, denoting thespecificity of NM for As adsorption. The concentration of K+ inthe medium affected the Ca, Mg, Mn, and Na adsorption by NM(Table 4B), so that the concentrations of Ca, Mg, and Mn were lowerat 3 mmol K L−1 and even lower at 6 mmol K L−1. In contrast, the Naconcentration was higher as the K concentration in the equilibriumincreased. The effect of N concentration (as NH4

+ and NO3−) on the

adsorption of TEs by NM was studied at three different concentra-tions of K and P (0, 3, and 6 mmol L−1), but, since no differences weredetected among the K and P concentrations, only data without K andP are shown (Table 4C). Only the highest concentration of N in themedium affected the Ca, Mg, and Mn in the equilibrium (reductionsof 2.9, 2.7, and 2.4%, respectively). For the increasing concentrationof P in the medium (Table 4D), the highest level reduced the poten-tial availability of Al, Ca, Mg, and Mn, and increased the As, Fe, andNa. The concentration of Na was proportional to the P concentra-tion. No changes on the Al, As, or Fe were detected when the K andP concentrations increased simultaneously (Table 4E). The concen-trations of Co, Cr, Ni, Se and Pb were below the ICP-OES detectionlimits.

3.2. Effects of K and N on adsorption by NM, using CCD

The responses, as concentrations in the equilibrium, are given asthe average of three results for Ca, Fe, Mg, Mn, Na, and Zn (Table 3),since no significant differences were detected for pH, Al, As, Cd, orCu (R2

adj< 0.8), and their quadratic models were not created. The

regression models for TEs removal, based on RSM from the data ofTable 3, are given in the following equations (Eqs. (3)–(8)). Theseexplain the relationship of the two variables, KNO3 and NH4NO3concentration, to each element concentration remaining in the

solution:

Ca (mg L−1) = 42.301 − 1.955 · X1 − 1.062 · X2 + 0.054 · X21

(R2Adj = 99.94%) (3)

274 D. Martínez-Fernández et al. / Journal of Hazardous Materials 276 (2014) 271–277

Table 4Values of pH and concentrations of elements in the equilibrium with the NM treatment and the three different nutrients (0, 3, or 6 mmol L−1). The table is divided accordingthe comparisons carried out: (A) with and without NM; (B) increasing concentration of K; (C) increasing concentration of N; (D) increasing concentration of P; (E) increasingconcentration of K and P. Mean values denoted by the same letter in a column do not differ significantly according to Tukey’s test (p > 0.05); ns not significant. The Zn datawere transformed with 1/x to satisfy normality.

Treatment pH �g L−1 mg L−1

Al As Cd Cu Fe Zn Ca Mg Mn Na

(A)Control 4.66 15.77 25.60 a 2.73 a 2.93 1.03 58.57 47.67 1.47 0.125 0.75NM 4.66 16.23 18.30 b 2.56 b 2.77 19.10 20.87 47.54 1.46 0.124 0.79t Student ns ns ** ** ns ns ns ns ns ns ns

(B) K treatmentNM+ K0N6P0 4.65 16.23 18.03 2.80 3.27 10.87 31.80 46.16 a 1.42 a 0.121 a 0.78 cNM+ K3N6P0 4.68 17.20 17.16 2.63 3.03 – 22.30 43.16 b 1.30 b 0.112 b 0.88 b

NM+ K6N6P0 4.67 18.17 16.46 2.60 3.33 – 18.43 41.77 c 1.25 c 0.108 c 0.94 aANOVA ns ns ns ns ns ns ns *** *** *** ***

(C) N treatmentNM+ K0N0P0 4.66 16.23 18.30 2.57 2.77 19.10 20.87 47.54 a 1.46 a 0.124 a 0.79NM+ K0N3P0 4.63 17.50 20.20 2.67 2.57 – 32.00 46.86 ab 1.44 ab 0.123 ab 0.76NM+ K0N6P0 4.65 16.23 18.03 2.80 3.27 10.87 31.80 46.16 b 1.42 b 0.121 b 0.78ANOVA ns ns ns ns ns ns ns ** ** * ns

(D) P treatmentNM+ K6N6P0 4.67 18.16 a 16.46 b 2.60 b 3.33 0.16 b 18.43 41.77 a 1.25 a 0.108 a 0.93 cNM+ K6N6P3 4.64 16.40 a 20.46 a 2.56 b 3.23 2.30 b 17.37 41.35 a 1.25 a 0.107 a 1.84 bNM+ K6N6P6 4.61 13.50 b 21.40 a 2.86 a 3.63 6.10 a 19.53 40.10 b 1.22 b 0.105 b 2.73 aANOVA ns *** * ** ns ** ns *** *** *** ***

(E) KP treatmentNM+ K0N6P0 4.65 16.23 18.03 2.80 3.27 10.87 31.80 46.16 a 1.42 a 0.121 a 0.77 cNM+ K3N6P3 4.61 15.80 17.36 2.93 3.33 0.23 16.53 42.57 b 1.29 b 0.111 b 1.71 b

NM+ K6N6P6 4.61 13.50 21.40 2.87 3.63 6.10 19.53 40.10 c 1.22 c 0.105 c 2.73 aANOVA ns ns ns ns ns ns ns *** *** *** ***

F

M

M

N

Z

evTtR

AtT

changes in the element concentrations since no significant changein pH was detected in either experiment, thanks to the MES bufferaddition. The adsorption capacity of NM is highly pH-dependent[30], and Tuutijärvi et al. [31] found that it was maximal at pH 3,

Table 5Optimum conditions found by the CCD models for the minimum and maximumresults.

TEs/nutrients Ca Mg Mn Na Fe Zn

KNO3 mmol L−1 6 6 6 0 6 6NH4NO3 6 6 6 6 6 2.5Min.response

mg L−1 38.52 1.17 0.10 0.71�g L−1 2.0 10.0

* p < 0.05.** p < 0.01.

*** p < 0.001.

e (�g L−1) = 4.278 − 0.564 · X1 − 0.097 · X2 − 0.715 · X1 · X2

(R2Adj = 95.59%) (4)

g (mg L−1) = 1.253 − 0.075 · X1 − 0.038 · X2 + 0.017 · X21

(R2Adj = 97.85%) (5)

n (mg L−1) = 0.1071 − 0.0046 · X1 − 0.0023 · X2 + 0.0007 · X21

(R2Adj = 99.42%) (6)

a (mg L−1) = 0.839 + 0.017 · X1 − 0.029 · X2 + 0.030 · X1 · X2

(R2Adj = 95.70%) (7)

n (�g L−1) = 19.428 − 3.644 · X1 + 2.033 · X2 + 4.197 · X22

(R2Adj = 86.55%) (8)

The adjusted determination coefficient (R2Adj

) values of the mod-ls (calculated by ANOVA), which do not always increase whenariables are added, are a measure of how much variability in eachE can be explained by KNO3 (X1), NH4NO3 (X2), and their interac-ions (Xl·X2, X2

1 and X22 ). The determination coefficients (acceptable

2Adj

≥ 0.8, for biological samples) of the models calculated byNOVA show that the quadratic models were a very good fit to

he data, with respect to describing the experimental region [29].he results show that only a negligible part of the total variation

was not explained by the models, thus confirming that the modelswere highly significant.

The mathematical prediction models derived from the statisticalanalysis were used to generate the response surface plots shown inFig. 1. These plots only provide an approximation within the experi-mental region. The optimum conditions for minimal and maximumremoval of elements from the soil solution by NM are given as theremaining concentrations of the elements in solution, at the 0.05%probability level (Table 5).

4. Discussion

Although the total concentrations of TEs in the soil (22,383, 389,21, 23,782, 499, 26, and 93 mg kg−1 for Al, As, Cu, Fe, Mn, Pb, andZn, respectively) were much higher than in the soil solution, thelatter were high enough to be able to see changes caused by theNM interaction. The pH could not have been responsible for the

KNO3 mmol L−1 0 0 0 0 0 0NH4NO3 0 0 0 0 6 6Max.response

mg L−1 47.05 1.49 0.12 0.93�g L−1 6.4 40.0

D. Martínez-Fernández et al. / Journal of Hazardous Materials 276 (2014) 271–277 275

proce

sr

4s

twodia

Fig. 1. Response surfaces plots for optimization of

o the obtained pH value of 4.66 in our experiments could haveeduced the effectiveness of NM.

.1. PO43− reduced the adsorption capacity of NM in the soil

olution

The concentrations of Al, As, and Cd were only affected byhe P concentration, so it was expected that no-significant effectould be detected for these elements in the CCD experiment, since

nly K and N were evaluated. The Al–P interference supposed aecrease of Al in the equilibrium at 6 mmol P L−1, so a high P level

n the soil could help to improve the adsorption of Al after NMpplication, since the Al concentration did not change when the

ss variables by CCD, in the experiment for K and N.

NM was used without P in the solution. The explanation may bethat Al3+ formed soluble complexes with phosphate [32] or thatphosphate facilitated sorption by the formation of intermediatecompounds with Al, like AlPO4

3+ and mainly AlHPO44+, this latter

compound being the most-abundant form of Al during the addi-tional speciation tests carried out with Visual MINTEQ. Phosphateinterferes with the binding of arseniate (AsO4)3− to different mate-rials, due to their structural molecular similarity [33]. In the caseof As and Cd, increases in their concentrations were detected in

the presence of P in the equilibrium; hence, P reduced the adsorp-tion capacity of NM for As and Cd. According to Chowdhury andYanful [34], the competition between AsO4

3− and PO43− for the

sorption sites in contaminated water reduced the effectiveness

2 of Haz

oraowcigFHdetooooiicte(abns

4

hadtuttamtoftqa2bffoKH

aiCilecoivNe

76 D. Martínez-Fernández et al. / Journal

f maghemite nanoparticles, in agreement with our results. Aseviewed by Komárek et al. [9], based on spectroscopic, kineticnd titration measurements, As(V) predominantly adsorbs to Fexides as inner-sphere surface complexes through ligand exchangeith −OH groups at the mineral surfaces resulting from bidentate

orner-sharing of AsO4 and FeO6 polyhedra. In other words, themmobilization of As occurs by replacement of the surface hydroxylroups by the As ions, as well as by the formation of amorphouse(III) arsenates and/or insoluble secondary oxidation minerals [3].owever, in other interference studies with Fe2O3, Luther et al. [17]id not detect any effect of PO4

3− or SO42−, but there was a specific

ffect of CO32− on As(V). Cadmium may be fixed by complexation

o oxygen atoms in the oxyhydroxy groups at the surface of the ironxide nanoparticles [34]. According to Turner et al. [35], adsorptionf Cd by Fe oxides decreased with increasing salinity. The increasef the available Cd in the soil solution by increasing concentrationsf P could be caused since they can form electrostatic and/or chem-cal interactions between Cd(II) and orthophosphate ions at theron oxide surfaces [36]. Then, although maghemite nanoparticlesan be regarded as a cheap and promising candidate for exploita-ion in the field, for the removal of undesired pollutants [37], theirffectiveness in the treatment of As and/or Cd-contaminated soilsand in hydroponic solutions) may be reduced by the simultaneouspplication of P. Arsenic and Cd adsorption by NH was not affectedy the K and N concentrations in the first experiment, so theseutrients seem not to compete with As or Cd for the adsorptionites on NM, as was confirmed in the CCD experiment.

.2. Strengths of CCD for K+ and NH4+ adsorption effects

Conducting the two experiments with different methodologiesad advantages over the individual use of each separately. The CCDllowed obtaining statistically-significant quadratic models to pre-ict the concentrations of TEs remaining in solution with/withouthe nutrients during NM application. Furthermore, the CCD wasseful to study the relationship between the two variables, ando explain with higher resolution the effects of K+ and NH4

+ onhe adsorption by NM. The concentrations of Cu and Zn were notffected by NM or the presence of nutrients in the first experi-ent, but the CCD made it possible to predict the dependence of

he Zn concentration on the K and N concentrations in the sec-nd experiment. The response surface showed that the available Znollowing NM treatment depended inversely on the K+ concentra-ion and proportionally on the NH4

+ concentration (Fig. 1F), with auadratic interaction between the Zn and NH4NO3 concentrationsnd a higher adsorption capacity of NM at 6 mmol L−1 KNO3 and.5 mmol L−1 NH4NO3, respectively (Table 5). These results coulde explained by the competition between the metal ions and NH4

+

rom the soil solution for adsorption onto the iron nano-oxide sur-ace. So, the presence of K+ could have facilitated the adsorptionf Zn on the NM until reaching a limit, from which both NH4

+ and+ equally decreased Zn adsorption, in agreement with Wang andarrell [38].

In both experiments the Ca, Mg, and Mn concentrations showed tendency to be lower in the presence of NM and nutrients, reach-ng their lowest values at the highest K–N–P concentrations. TheCD was more useful with regard to identifying the quadratic

nteraction of the Ca, Mn, and Mg concentrations with KNO3. Theatter exhibited a positive effect upon the concentrations of theselements in the equilibrium, meaning that K+ reduced the con-entrations that remained in solution. Then, as the concentrationsf KNO3 and NH4NO3 increased, the removal of Ca, Mn, and Mg

ncreased (maximum response at the minimum concentration andice versa; Table 5). So, during TEs removal from the soil solution byM, as the concentration of KNO3 or NH4NO3 increased in the pres-nce of the other, the result was a reduction in the Ca, Fe, Mg, and

ardous Materials 276 (2014) 271–277

Mn concentrations because of the optimization of their adsorptiononto the surface of the NM.

For Na adsorption, KNO3 was not a statistically-significant factorat the 95% confidence level (p-value > 0.05), since it was unchangedin the presence or absence of KNO3. The presence of K+ and NH4

+

had a relatively-minor effect upon Cu, Cd, and As and, in fact, wasnot statistically significant during the competition between TEs andnutrients in the soil solution.

As shown in Eqs. (3)–(8), higher KNO3 and NH4NO3 concentra-tions favored lower levels of the analyzed elements remained insoil solution (always negative relation), except for Na and Zn. Abetter measure of significance comes from using the coefficientsof the coded factors, in which the concentration of KNO3 appearsto have a much larger influence; regardless mark the concentra-tion of NH4NO3 is seen to be quite small for Ca, Fe, Mn, Mg and Zn.In contrast, it appears that NH4NO3 is much more influential thanKNO3 for Na. So, based on this analysis, the use of high values of K+

and NH4+ could improve the effectiveness during phytoremedia-

tion technologies since higher KNO3 and NH4NO3 concentrationsfavored lower TEs concentrations.

5. Conclusions

The effectiveness of NM for TEs adsorption is affected by theconcentration of nutrients in the soil solution. Among the nutrientsstudied, the presence of PO4

3− reduced the As and Cd adsorptioncapacity of NM in the soil solution, which might be crucial duringthe treatment of contaminated soils with NM and an amendmentrich in P. The central composite design was used to observe theinteractions between independent factors, to optimize them, andto develop a predictive model for competition between TEs andnutrients in the medium. Statistically-significant quadratic modelswere derived to obtain the concentrations of TEs remaining in solu-tion with/without the nutrients during NM application, providingan area of future research for the optimization of phytoremedia-tion processes. The determination coefficients of the models werefound to be R2

Adj≥ 0.8% for Ca, Fe, Mg, Mn, Na, and Zn adsorption,

showing that the quadratic models were satisfactory. The presenceof K+ and NH4

+ can improve the adsorption of Ca, Mg, Mn, andNa on the NM surface, reducing their concentrations in the soilsolution, but the magnitude of this effect was modest comparedto the dominant effect of the K+ concentration alone. This studyfurther showed another approach of studying TE interactions withstabilizing amendments using a combined geochemical-statisticalapproach.

Acknowledgments

The authors thank Zuzana Michálková and Hana Sillerová fortheir help with analyses. Domingo Martínez-Fernández is grate-ful for financial support from the European project Postdok CZU(ESF/MSMT CZ.1.07/2.3.00/30.0040). This study was supported bythe Czech Science Foundation (GACR 503/11/0840). The assistanceof Dr. David J. Walker for English revision is also acknowledged.

References

[1] N.C. Mueller, B. Nowack, Nanoparticles for remediation: solving big problemswith little particles, Elements 6 (2010) 395–400.

[2] K.R. Kim, B.T. Lee, K.W. Kim, Arsenic stabilization in mine tailings using nano-sized magnetite and zero valent iron with the enhancement of mobility bysurface coating, J. Geochem. Explor. 113 (2012) 124–129.

[3] J. Kumpiene, A. Lagerkvist, C. Maurice, Stabilization of As, Cr, Cu, Pb and Zn in

soil using amendments – a review, Waste Manag. 28 (2008) 215–225.

[4] G. Crini, Recent developments in polysaccharide-based materials used as adsor-bents in wastewater treatment, Prog. Polym. Sci. 30 (2005) 38–70.

[5] J.L. Gong, B. Wang, G.M. Zeng, C.P. Yang, C.G. Niu, Q.Y. Niu, W.J. Zhou, Y. Liang,Removal of cationic dyes from aqueous solution using magnetic multi-wall

of Haz

[

[

[

[

[

[

[

[

[[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

ticles toward arsenate removal from aqueous environment, Chemosphere 93

D. Martínez-Fernández et al. / Journal

carbon nanotube nanocomposite as adsorbent, J. Hazard Mater. 164 (2009)1517–1522.

[6] P. Xu, G.M. Zeng, D.L. Huang, C.L. Feng, S. Hu, M.H. Zhao, C. Lai, Z. Wei, C. Huang,G.X. Xie, Z.F. Liu, Use of iron oxide nanomaterials in wastewater treatment: areview, Sci. Total Environ. 424 (2012) 1–10.

[7] W. Tang, G. Zeng, J. Gong, Impact of humic/fulfic acid on the removal of heavymetals from aqueous solutions using nanomaterials: a review, Sci. Total Envi-ron. 468-469 (468) (2014) 1014–1020.

[8] G.A. Waychunas, C.S. Kim, J.F. Banfield, Nanoparticulate iron oxide minerals insoils and sediments: unique properties and contaminant scavenging mecha-nisms, J. Nanopart. Res. 7 (2005) 409–433.

[9] M. Komárek, A. Vanek, V. Ettler, Chemical stabilization of metals and arsenic incontaminated soils using oxides – a review, Environ. Pollut. 172 (2013) 9–22.

10] P.I. Girginova, A.L. Daniel-da-Silva, C.B. Lopes, P. Figueira, M. Otero, V.S. Amaral,E. Pereira, T. Trindade, J. Colloid Interface Sci. 345 (2010) 234–240.

11] B. Prasad, C. Ghosh, A. Chakraborty, N. Bandyopadhyay, R.K. Ray, Adsorptionof arsenite (As3+) on nano-sized Fe2O3 waste powder from the steel industry,Desalination 274 (2011) 105–112.

12] A. Roy, J. Bhattacharya, Removal of Cu (II), Zn (II) and Pb (II) from water usingmicrowave-assisted synthesized maghemite nanotubes, Chem. Eng. J. 211–212(2012) 493–500.

13] J. Chomoucka, J. Drbohlavova, D. Huska, V. Adam, R. Kizek, J. Hubalek, Magneticnanoparticles and targeted drug delivering, Pharmacol. Res. 62 (2012) 144–149.

14] D.J. Walker, M.P. Bernal, E. Correal, The influence of heavy metals and mineralnutrient supply on Bituminaria bituminosa, Water Air Soil Pollut. 184 (2007)335–345.

15] A.F. de Sousa, T.P. Braga, E.C.C. Gomes, A. Valentini, E. Longhinotti, Adsorptionof phosphate using mesoporous spheres containing iron and aluminum oxide,Chem. Eng. J. 210 (2012) 143–149.

16] K. Tyrovola, N.P. Nikolaidis, N. Veranis, N. Kallithrakas-Kontos, P.E. Koulouri-dakis, Arsenic removal from geothermal waters with zero-valent iron, effect oftemperature, phosphate and nitrate, Water Res. 40 (2006) 2375–2386.

17] S. Luther, N. Borgfeld, J. Kim, J.G. Parsons, Removal of arsenic from aqueoussolution: a study of the effects of pH and interfering ions using iron oxidenanomaterials, Microchem. J. 101 (2012) 30–36.

18] K.J. Dietz, S. Herth, Plant nanotoxicology, Trends Plant Sci. 16 (2011) 11.19] D.C. Montgomery, Design and Analysis of Experiments, 7th ed., John Wiley &

Sons, New York, 2008.20] R.G. Brereton, Applied Chemometrics for Scientists, John Wiley & Sons, England,

2007.21] M. Filippi, B. Dousová, V. Machovic, Mineralogical speciation of arsenic in soils

above the Mokrsko–West gold deposit, Czech Republic, Geoderma 139 (2007)154–170.

22] P. Drahota, J. Rohovec, M. Filippi, M. Mihaljevic, P. Rychlovsky, V. Cerveny,Z. Pertold, Mineralogical and geochemical controls of arsenic speciation andmobility under different redox conditions in soil, sediment and water at the

[

ardous Materials 276 (2014) 271–277 277

Mokrsko–West gold deposit, Czech Republic, Sci. Total Environ. 407 (2009)3372–3384.

23] Ph. Quevauviller, Operationally defined extraction procedures for soil and sed-iment analysis I. Standardization, Trends Anal. Chem. 17 (1998) 289–298.

24] D.R. Hoagland, D.I. Arnon, The water-culture method for growing plants with-out soil. UC Agric. Exp.Sta. Circ. No. 3347, 1938, pp. 1–39, Berkeley.

25] J. Coombs, Growing power—bioenergy for development and industry: A. S.Miller, I. M. Mintzer and S. H. Hoagland. World Resources Institute, 1986. StudyNo. 5, paperback viii + 100 pp. ISBN 0-915825-14-7, Biomass 11 (1986) 3, 240.

26] D. Martínez-Fernández, D.J. Walker, P. Romero-Espinar, P. Flores, J.A. del Río,Physiological responses of Bituminaria bituminosa to heavy metals, J. PlantPhysiol. 168 (2011) 2206–2211.

27] J.P. Gustafsson, Visual MINTEQ version 3.0, KTH (Royal Institute ofTechnology), Department of Land and Water Resources Engineering, Stock-holm, Sweden, 2011 http://www2.lwr.kth.se/English/OurSoftware/vminteq(accessed 20.07.13).

28] MINITAB version 16 is statistical software of Minitab Inc, USA,http://www.minitab.com

29] T. Lundstedt, E. Seifert, L. Abramo, B. Thelin, Å. Nyström, J. Pettersen, R. Bergman,Experimental design and optimization, Chemometr. Intell. Lab. 42 (1998) 3–40.

30] M. Hua, S. Zhang, B. Pan, W. Zhang, Lu. Lv, Q. Zhang, Heavy metal removalfrom water/wastewater by nanosized metal oxides: a review, J. Hazard. Mater.211–212 (2012) 317–331.

31] T. Tuutijärvi, J. Lu, M. Sillanpää, G. Chen, As (V) adsorption on maghemitenanoparticles, J. Hazard. Mater. 166 (2009) 1415–1420.

32] P.H. Hsu, Interaction between aluminum and phosphate in aqueous solution,in: R.A. Baker (Ed.), Trace Inorganics in Water, Advances in Chemistry Series,73, 1968, pp. 115–127.

33] D.C. Adriano, Trace Elements in Terrestrial Environments, Springer-Verlag, NewYork, 2001.

34] S.R. Chowdhury, E.K. Yanful, Arsenic and chromium removal by mixedmagnetite–maghemite nanoparticles and the effect of phosphate on removal,J. Environ. Manag. 91 (11) (2010) 2238–2247.

35] A. Turner, S.M. Le Roux, G.E. Millward, Adsorption of cadmium to iron andmanganese oxides during estuarine mixing, Mar. Chem. 108 (2008) 77–84.

36] E.J. Elzinga, R. Kretzschmar, In situ ATR-FTIR spectroscopic analysis of the co-adsorption of orthophosphate and Cd(II) onto hematite, Geochim. Cosmochim.Acta 117 (2013) 53–64.

37] M. Kilianová, R. Prucek, J. Filip, J. Kolarík, L. Kvítek, A. Panácek, J. Tucek, R. Zboril,Remarkable efficiency of ultrafine superparamagnetic iron (III) oxide nanopar-

(2013) 2690–2697.38] J.J. Wang, D.L. Harrell, Effect of ammonium, potassium, and sodium cations and

phosphate, nitrate, and chloride anions on zinc sorption and lability in selectedacid and calcareous soils, Soil Sci. Soc. Am. J. 69 (2005) 1036–1046.