Environmental Toxicology and Chemistry, Vol. 33, No. 2, pp. 341–349, 2014 # 2013 SETAC Printed in the USA

UNCOATED AND COATED ZNO NANOPARTICLE LIFE CYCLE IN SYNTHETIC SEAWATER ALEXANDRE GELABERT,*y YANN SIVRY,y ROSELYNE FERRARI,y ASSIA AKROUT,y LAURE CORDIER,y SOPHIE NOWAK,z NICOLAS MENGUY,x and MARC F. BENEDETTIy yInstitut de Physique du Globe de Paris, UMR 7154, Paris Diderot University, Sorbonne Paris Cite, Paris, France zITODYS, UMR 7086, Paris Diderot University, Sorbonne Paris Cite, Paris, France xInstitut de Mineralogie et de Physique des Milieux Condenses, UMR 7590, Pierre and Marie Curie University, Paris, France (Submitted 29 May 2013; Returned for Revision 27 July 2013; Accepted 28 October 2013) Abstract: The increasing production of nanoparticles has raised strong concerns regarding their environmental release. In life cycle scenarios of nanoparticles, marine systems constitute one of the main final compartments, and the fate of nanoparticles in marine environments needs to be assessed. The dissolution kinetics of commercial uncoated and organic-coated ZnO nanoparticles in synthetic seawater were investigated using the Donnan membrane technique and 1000-Da pore size ultrafiltration. Uncoated nanoparticles reach a maximum dissolution within the first hour, approximately 24% of total ZnO at pH 8.2, and 4% at pH 7.7, followed by secondary carbonated phase precipitation (hydrozincite) until the system reaches a steady state after 30 d of interaction. Assuming a pseudo first-order kinetics for hydrozincite precipitation allowed calculation of kinetics constant values k0 p of 208  104 mol L1 h1  15  104 mol L1 h1 (standard deviation) at pH 7.7, and 57  104 mol L1 h1  11  104 mol L1 h1 at pH 8.2. The presence of an organic coating drastically modifies the life cycle of nanoparticles, with a maximum dissolution reached after 7 d of interaction, followed by a stationary phase lasting from 1 wk to 3 wk, and a subsequent Zn carbonate precipitation until a steady state is reached after 1.5 mo. Monitoring changes in the physicochemical parameters of nanoparticles after exposure to synthetic seawater constitutes an important step in predicting their fate in environmental systems, with major implications for ecotoxicological studies in which metallic speciation is required for toxicity evaluation. Environ Toxicol Chem 2014;33:341–349. # 2013 SETAC Keywords: Zinc oxide

Nanoparticles

Solubility

Organic Coating

Seawater

Fate

description of the physicochemical properties of nanoparticles, resulting in only a partial understanding of their potential impact. An accurate evaluation of the impact of ENPs on ecosystems simulated by toxicity studies thus requires a precise knowledge of their speciation, aggregation state, dissolution rate, and surface physicochemistry as well as a clear statement of the evolution of these parameters over time. To date, only partial information is available regarding the general fate of ENPs in environmental systems [7,15–18]. This is partly because of the complexity inherent in their detection in natural settings. Thus, alternative modeling approaches to estimate their life cycle have been proposed [5]. Such studies have highlighted the role of surface waters as the principal ENP dispersion agent, demonstrating the importance of predicting ENP behavior and fate in aquatic systems (rivers, lakes, marine environments). Marine settings play a major role because, after their environmental release in freshwaters, ENPs may be stable enough to enter marine or estuarine systems [19], which can potentially constitute one of the ultimate ENP life cycle compartments. Despite a postulated rapid dissolution [20], ZnO ENPs can still be found in systems exposed to natural freshwaters after several weeks of interaction, corresponding to sufficient transfer time for ZnO ENPs to reach marine environments [21,22]. In addition, another important contamination point for marine systems derives from the use of sunscreens incorporating ZnO ENPs [23] that, once applied onto the skin, can be released in seawater by immersion or sand abrasion [24] . Once into marine environments, ZnO ENPs can exhibit high reactive oxygen species generation; such toxicity is associated with the particulate form [25], especially when the aggregation of ENPs in saline waters is considered, resulting in their

INTRODUCTION

Strategies for nanoparticle synthesis and manipulation have undergone outstanding progress during the last 20 yr. Because of their high surface area-to-volume ratio, resulting in an increased physicochemical reactivity compared with the corresponding bulk material [1,2], engineered nanoparticles (ENPs) are today incorporated into numerous industrial and technological fields. As a result, ENPs are produced in large quantities for a widespread range of applications [3], and the predicted annual increase in manufactured nanoparticle use is as high as 58 000 tons/yr between 2011 and 2020 [4]. Given the high fluxes associated with ENPs, many studies postulated a considerable release of this material into environmental settings [5–7]. Among the different structures specifically developed at the nanoscale, ZnO has long been a very promising material, and numerous applications have followed the development of ZnO nanowire, nanobelt, and nanoparticle [8] synthesis strategies. Today, among other applications, ZnO nanoparticles are present in various electronic devices, and are incorporated into textiles for antibacterial and ultraviolet (UV)-blocking properties [9]. Along with TiO2, they are one of the most important inorganic UV filters used in sunscreens because of their high photostability and low photoallergic potential [10]. Because of their specific properties, ENP behavior is difficult to predict from the corresponding bulk material reactivity [1,11]. As stated in review papers [12–14], the main weakness of most of the toxicity studies originates in the lack of a complete All Supplemental Data may be found in the online version of this article. * Address correspondence to [email protected]. Published online 31 October 2013 in Wiley Online Library (wileyonlinelibrary.com). DOI: 10.1002/etc.2447 341

342

Environ Toxicol Chem 33, 2014

A. Gelabert et al.

precipitation and the subsequent accumulation of these particles in sediments [26]. However, this effect can be counterbalanced by rapid dissolution [27], which complicates evaluation of their environmental impact. Because of the coexistence of the particle and ionic forms, toxicity interpretations for ZnO ENPs are difficult, and the relative contribution to toxicity of each form remains unclear, even in controlled culture media [28–36]. Moreover, Zn can be a limiting factor for marine organisms [37], and input from ZnO ENPs could modify the equilibrium of existing ecosystems. Hence, it is critical to have reliable estimations of Zn speciation and ZnO ENP dissolution in seawater at concentrations relevant for toxicity studies. From this perspective, numerous studies were devoted to defining the dissolution characteristics of ZnO nanoparticles [20,21,27,30,32,34,38–42]. It was noted that, in these reports, most separation processes were sometimes inaccurate or inappropriate because nanoparticles are strongly subjected to aggregation and sedimentation [43]. However, David et al. [20] employed electrochemical techniques to monitor dissolution kinetics and thermodynamics. They showed that a dissolution size effect could be measured for nanoparticles below 20 nm. In addition to these valuable datasets, experiments integrating the complexity of natural matrices are urgently needed. A few studies have focused on the dissolution of ZnO ENPs in marine environments, highlighting a rapid dissolution similar to that of other electrolytes [27,38]. However, no studies have been performed on the precipitation of secondary phases, although the formation of these precipitates could result in a surface passivation and a subsequent decrease in nanoparticles reactivity. In addition, the impact of organic coatings on ZnO dissolution has not been systematically investigated, whereas these coatings are frequently associated with commercial nanoparticles. Because of their organic nature, these coatings could potentially control the dynamics of ZnO nanoparticles in marine and estuarine environments, in a similar way to natural organic matter [26]. In the present study, we give kinetic data relative to the dissolution rate of ZnO nanoparticles in synthetic seawater, and try to link the dissolution processes in this natural analog to a thermodynamic description of the system. Moreover, we focus on bare and coated ENPs to highlight the role of coatings in the fate of ENPs. The dissolution kinetics were monitored as a function of pH for both types of nanoparticles. High ZnO nanoparticle concentrations were considered in the present study for 2 reasons: to accurately assess the kinetic parameters in our systems, and also because most ecotoxicological studies, for which an accurate description of the physicochemical properties of nanoparticles is required, have used typically high nanoparticle concentrations in closed systems [28,31,33,34,39,40]. Many analytical separation methods and strategies were tested in the present study, and were complemented with transmission electron microscopy (TEM) observations as well as X-ray photoelectron spectrometry (XPS) and X-ray diffraction (XRD) measurements.

MATERIALS AND METHODS

Physicochemical characterization of nanoparticles

The ENP synthesis protocol provides most of the physicochemical properties of nanoparticles, such as size, shape, crystallinity, or presence of an amorphous layer [27,38]. We focused on commercial ZnO nanoparticles (kindly provided by Kobo Products), which are likely to be released into environmental systems, instead of using those synthesized in the laboratory [44]. Two types of ZnO nanoparticles were used: coated (termed c-NPs in the present study) and uncoated (termed nc-NPs), as both forms are used for industrial applications and are expected to behave differently because of the organic coating [26,38]. An extensive description of the 2 forms has already been given by Sivry et al. in a previous report [22]. In brief, both nanoparticles display a wurtzite crystal structure, with an average size of 21.0 nm  (standard deviation) 3.3 nm and 17.5 nm  2.5 nm for nc-NPs and c-NPs, respectively. Moreover, the nc-NP surface is composed of a 1.6-nm-thick Zn(OH)2 layer. According to the manufacturer (Supplemental Data, Table S1), the c-NP organic coating is composed of a mixture of C12-15 alkyl benzoate, polyhydroxystearic acid, and triethoxycaprylylsilane, and is approximately 2 nm thick. Both nanoparticles exhibit an anisotropic shape (spheroid-like), but the particle population is globally homogeneous, with an associated specific surface area (Brunauer–Emmett–Teller [BET] analysis) of 37.5 m2 g1. Hydrodynamic size was measured with dynamic light scattering using a Malvern Zetasizer HS 3000 at a concentration of 2.5 mmol L1 of Zn introduced as ZnO. Seawater characterization

We used a synthetic seawater sample following the protocol given by Dickson [45] to obtain a good control for the exposure media parameters. The chemical composition of the seawater is given in Table 1. No natural organic matter was added because natural organic matter concentration in marine water is usually rather low (40 mmolc L1) compared with terrestrial aquatic ecosystems. Moreover, the absence of natural organic matter in our synthetic seawater minimizes interference by probing only the effect of the organic coating on the ZnO ENP dissolution kinetics. To test the pH dependence for ZnO dissolution, 2 different pH conditions corresponding to the pH range of natural seawater were used for each nanoparticle type (7.7  0.3 and 8.2  0.3 for nc-NPs, and 7.8  0.5 and 8.2  0.2 for c-NPs; Supplemental Data, Figure S1). Nanoparticle solubility experiments

For the solubility studies, ZnO c-NPs or nc-NPs corresponding to a total zinc concentration of 2.5  103 mol L1 were suspended in 250 mL of synthetic seawater in polypropylene bottles and agitated with a magnetic stirrer. No buffer solutions were used in these experiments. The solubility was monitored at different time steps (from 1 h to 3 mo) using the ultrafiltration technique (UFT) with 1000-Da nominal pore size membranes

Table 1. Chemical composition of the synthetic seawater used in the present studya Naþ (mmol L1) 486 a

Mg2þ (mmol L1)

Ca2þ (mmol L1)

Kþ (mmol L1)

Cl (mmol L1)

HCO3 (mmol L1)

SO42 (mmol L1)

H3BO4 (mmol L1)

pH

54.7

10.8

10.6

569

0.699

29.27

0.066

7.6–8.3

Data from Dickson [45].

ZnO nanoparticle life cycle in synthetic seawater

Residual and newly formed phase characterization

The mineral phases present in our experimental systems were identified using XRD with a Panalytical X’pert Pro diffractometer equipped with a multichannel X’celerator detector, and using a Co Ka radiation (l ¼ 1.7889 A) in the 2u range of 20 8 to 100 8. In addition, TEM (Jeol 2100 FEG operating at 200 kV) was used to characterize the evolutions in size and shape of the particles during our experiments. To do so, drops of dilute suspensions of the experimental solutions were deposited on copper grid and left to evaporate for a few minutes. To perform chemical analysis, this microscope was coupled with electrondispersive X-ray spectroscopy using a JEOL detector with an ultrathin window allowing detection of low atomic mass elements. X-ray photoelectron spectrometry was performed using an ESCALAB 250 (ThermoElectron) on samples settled on a carbon adhesive tape, and previously degassed overnight to remove atmospheric carbon and water.

343

Modeling

At the end of the experiment, the system is potentially at steady state, allowing thermodynamic modeling with the Visual MINTEQ v. 3.0 code [47]. Input concentrations of synthetic seawater elements are displayed in Table 1; the Visual MINTEQ thermodynamic database was used for these calculations (with the following solubility products: hydrozincite Zn5(OH)6(CO3)2 log Ks ¼ 14.9; smithsonite ZnCO3 log Ks ¼ 10.0; zincite ZnO log Ks ¼ 11.23; zinc hydroxyde Zn(OH)2 log Ks ¼ 12.47), assuming activity correction using the Davies equation for high ionic strength. RESULTS AND DISCUSSION

Validation of the 1000-Da ultrafiltration separation process in seawater

Dynamic light scattering measurements performed on ncZnO revealed 3 distinct size populations below 1 mm (20–50 nm, 150–200 nm, and 400–500 nm), much larger than the ultrafiltration filter pore size. Dissolved Zn concentrations measured in each ultrafiltrate as a function of time are displayed in Figure 1 for nc-NPs. The 2 DMT experimental concentrations are shown as gray circles after 10 d and 16 d of exposure (Figure 1). At pH 8.2, the free Zn2þ concentration was 34 mmol L1 (at 10 d) and 28 mmol L1 (at 16 d). These values are very close to the calculated 21 mmol L1 for [Zn2þ] at pH 8.2 using Visual MINTEQ code, and postulating that the system has reached pseudo-equilibrium after 10 d. This calculation uses the total dissolved concentration given by UFT (38.6  1.0 mmol L1), and the complexing and hydrolysis properties of the synthetic seawater. Because the difference between the measured and calculated free Zn is minor, we postulate that in this system involving nc-NPs, most of the dissolved Zn is free Zn2þ with negligible participation of partially dissolved ZnO nanoparticles of 2 nm in size or less. The good agreement between measured (DMT) and predicted (UFT/model) free Zn concentrations thus allows validation of the separation method based on 1 kDa UFT. 700 600 500

[Zn] dissolved ( M)

(2 nm), which allows isolation of small inorganic complexes and free metal ions from the rest of the solution. Ultrafiltration is filtration promoted by the acceleration provided by centrifugation. The ultrafiltration device consists of 2 tubes separated by a porous membrane of 1000-Da nominal pore size. The sample is initially introduced into 1 tube; after centrifugation, the second tube contains material whose size is smaller than 1000 Da. For clarity, the Zn measured using UFT will be referred to as dissolved Zn. For each time step, 2 aliquots of 3.5 mL were picked up: the first was ultrafiltered using UFT cells (Microsep Centrifugal Devices, 1 kDa) at 2060 g (Eppendorf Centrifuge 5810R), and the second was used to determine total Zn concentrations to correct for ZnO sorption onto the polypropylene bottle walls, which can lead to considerable experimental artifacts given the dispersion difficulties of ENPs. For each set of experiments, triplicate samples were taken at 3 different exposure times to assess experimental reproducibility. In addition, the Donnan membrane technique (DMT) was used to determine the free Zn2þ concentration in our experiments. The Donnan membrane device consists of a cation exchange membrane separating the donor (synthetic seawater supplemented with ZnO nanoparticles) and acceptor solutions (NaCl 0.6 mol L1). Free ions diffuse through the membrane, and a chemical concentration equilibrium is reached within 2 d to 3 d for divalent cations like Zn2þ [46] in both donor and acceptor compartments (i.e., the same free metal ion concentration in both compartments). The different parts of the system (Donnan membrane parts, Teflon vessels, and connecting devices) were washed, and the cationic membrane was preconditioned as described by Jouvin et al. [46]. Both donor and acceptor sides were sampled after 10 d and 16 d of interaction between the nanoparticles and the seawater. Three runs were performed for each experiment (2 replicates and 1 blank) to assess the external reproducibility of the method. To evaluate the efficiency of the UFT/DMT-coupled approach for ZnO nanoparticle solubility and Zn speciation in seawater, the experimental results were compared with a thermodynamic simulation of the solutions at 10 d and 16 d. All ultrafiltrates, filtration blanks, Donnan membrane acceptors, and donor samples were weight-diluted in ultrapure HNO3 2% prior to analysis. The elemental dissolved concentrations in these samples were measured using inductively coupled plasma optical emission spectrometry and high-resolution inductively coupled plasma mass spectrometry (ThermoScientific iCAP 6200 and Element II). At the end of the experiments, the solid phases containing both nanoparticles and secondary minerals were separated by centrifugation (3200 g, 20 min), and freeze-dried before microphysical characterizations.

Environ Toxicol Chem 33, 2014

400 300 200 pH = 7.7

100

pH = 8.2

DMT

0 0

500

1000 Time (h)

1500

2000

Figure 1. Uncoated-nanoparticle ZnO dissolution kinetics. The Zn concentrations in ultrafiltrates reported as a function of time at pH 7.7 (squares) and pH 8.2 (diamonds). The red and blue lines represent the theoretical results of first-order dissolution kinetics corrected for the steady-state Zn concentration, with k’ ¼ –208  104 mol L1 h1  15  104 mol L1 h1 (red) and k’ ¼ –57  104 mol L1 h1  11  104 mol L1 h1 (blue). The gray circles represent the free Zn2þ concentration measured by the Donnan membrane technique. Incertitude (s) takes into account the analytical standard deviation and the triplicate measurement standard deviation.

344

Environ Toxicol Chem 33, 2014

However, given the relatively high equilibration time for Donnan membrane devices (3 d), UFT was preferred for dissolution measurements in the present study. Uncoated nanoparticle solubility kinetics

The maximum dissolved Zn concentration was reached within the first hour of interaction, up to 585 mmol L1 (24% of total Zn) and 99 mmol L1 (4% of total Zn), at pH 7.7 and 8.1, respectively (Figure 1). Then the dissolved Zn concentration diminished quickly within the first day for both pH values, finally reaching steady state with stable concentrations of 83 mmol L1  17 mmol L1 at pH 7.7, and 26 mmol L1  2 mmol L1 at pH 8.2 (3.3% and 1.0% of total Zn, respectively). A strong pH control over ZnO nanoparticle dissolution can thus be observed with a systematic higher dissolution amount (up to 5fold) under more neutral conditions. This trend confirms the pH dissolution dependence proposed in previous works related to ZnO dissolution [34]. Moreover, the rapid dissolution of ZnO nanoparticles observed is a common element for studies dealing with ZnO nanoparticle stability in seawater [27,35,38,40]. This trend seems to indicate a common dissolution control for most ZnO nanoparticles, regardless of their synthesis routes and physicochemical properties. To the best of our knowledge, the subsequent decrease in dissolved Zn and the later stabilization are reported here for the first time in the literature. Because a 1.6-nm-thick Zn(OH)2 layer is located on the ncZnO nanoparticle surfaces, the initial dissolution observed in synthetic seawater (Figure 1) is likely to be controlled by this amorphous phase instead of the bulk ZnO wurtzite structure alone, as shown for nc-ZnO nanoparticle dissolution in freshwater [22]. The Zn(OH)2 layer was previously calculated to be 22.3% of total Zn in nanoparticles [22]. If total dissolution of the hydroxide layer occurred in our experimental systems, this would result in a dissolved Zn concentration of 556 mmol L1 in the 250-mL solution. This calculated value is very close to the measured concentration at 1 h of 585 mmol L1  4 mmol L1 at pH 7.7, and approximately 5 times more than the measured Zn

A. Gelabert et al.

concentration of 99 mmol L1  3 mmol L1 at pH 8.2. For both pH values, this finding suggests a strong control from the Zn(OH)2 phase on the initial steps of nc-ZnO nanoparticle dissolution in seawater. Presence of secondary phases

The solid phases remaining in solution after 3 mo of nc-ZnO nanoparticle interaction with seawater were harvested after centrifugation and characterized by TEM (Supplemental Data, Figure S2). The TEM images reveal the presence of 2 distinct particle populations: nanosized structures directly comparable with native nc-NPs, and needle-shaped or elongated structures of a few hundred nanometers. The XRD measurements performed on the same samples (Figure 2) showed the presence of a wurtzite structure along with Zn carbonate phases such as hydrozincite, Zn5(OH)6(CO3)2, as well as other carbonated phases such as calcite and dolomite. Given the size and shape of the smallest particles and the presence of a wurtzite phase, this likely indicates that part of the nc-ZnO nanoparticles persisted as ENPs in our system after 3 mo of interaction. On the other hand, the carbonated phases may correspond to secondary minerals resulting from Zn2þ released during nanoparticle dissolution and its subsequent association with carbonated species present in the system. The XPS analysis performed on the solid phases sampled after 3 d and 3 mo of interaction with synthetic seawater at pH 8.2 are displayed in Table 2 and Figure 3. The ratio C(carbonate)/Zn did not significantly evolve during the first 3 d of interaction, from 0.06 (native nc ZnO nanoparticles) to 0.05 (3 d), while after 3 mo this ratio reached 0.27. Similarly, the ratio O/ C(carbonate) did not change significantly during the first 3 d, staying around 1.2, but reached a value of 2.4 after 3 mo of interaction. At 3 d, assuming a spherical shape for the nanoparticles, the XPS results allowed an estimation of a Zn(OH)2 thickness of 1.7 nm, and a Zn carbonate thickness of 0.3 nm. At 3 mo, the calculated atomic ratios for O/Zn and C(carbonate)/Zn were close to those given for hydrozincite (2.4 and 0.4, respectively) and indicate the presence of this phase in

Figure 2. Solid-phase identification using X-ray diffraction (XRD) after 3 mo of interaction. The XRD analyses were performed on native uncoated-nanoparticles ZnO (red), and on the corresponding solid phase after 3 mo of interaction (blue). After 3 mo, hydrozincite ( , the main diffraction peak positions) is 1 of the main phases present in the system, while a low proportion of ZnO wurtzite structures can still be detected.

ZnO nanoparticle life cycle in synthetic seawater

Environ Toxicol Chem 33, 2014

345

Table 2. Results of X-ray photoelectron spectrometry analysis for solid material collected after 3 d and after 3 mo

Peak BE (eV)

3 d (At. %) 3 mo (At. %)

C1s

C1s C

C1s B

C1s A

F1s

O1s

O1s A

Zn2p3

283.0  0.1

284.3  0.1

287.7  0.5

289.7  0.1

687.1  0.1

528.4

530.0  0.7

1021.5  0.9

C-C

C-O

C (carbonate)

C (Teflon)

F1s

O1s

OH-, carbonates

Zn2p3

17.87 17.63

1.44 6.39

1.51 4.54

1.45 5.11

3.27 6.49

17.69 -

23.40 38.89

33.39 17.02

BE ¼ binding energy; At. % ¼ atomic concentration; peak designation (atomic orbitals) are reported as column headings.

our system. However, XPS analysis did not allow us to detect ZnO structure after 3 mo of interaction. Because the presence of ZnO was confirmed by TEM and XRD experiments, and because XPS is a surface-specific analysis, this result may indicate that a significant amount of ZnO particles are coated with Zn carbonated phases, which could result in a possible passivation. After an initial dissolution of nc-ZnO nanoparticles, the dissolved Zn concentration in solution decreased, reaching steady state after 1 mo (Figure 1). The experimental conditions obtained for pH 8.2 and 7.7 after 1 h of interaction were modeled using the Visual MINTEQ code. Given the considerable dissolution occurring during the first hour, the system becomes oversaturated in free Zn, and the thermodynamic model predicts that the precipitation of hydrozincite would reach a Zn2þ concentration equal to 53.8 mmol L1 at pH 7.7, and 6.8 mmol L1 at pH 8.2, while the measured dissolved Zn values were higher, at 84 mmol

L1  3 mmol L1 and 24 mmol L1  3 mmol L1, respectively. The calculated dissolved Zn concentrations and the measured Zn concentrations have the same order of magnitude. In the long term, this modeling suggests a nc ZnO nanoparticle fate thermodynamically controlled by carbonate secondary-phase precipitation. Impact of organic coating on nanoparticle solubility

Dissolved Zn concentrations as a function of interaction time in seawater for c-NPs display a different pattern compared with nc-NPs (Figure 4). Indeed, for the 2 pH values investigated, the maximum dissolved Zn concentration is not reached immediately but only after 7 d of interaction. Then a plateau can be observed at approximately 546 mmol L1  16 mmol L1 and 135 mmol L1  6 mmol L1 (22% and 6%, respectively) for the pH 7.8 and pH 8.2 experiments, followed by a decrease in dissolved Zn concentration, finally reaching a second plateau at

Figure 3. Solid-phase identification using X-ray photoelectron spectrometry (XPS). The XPS analyses were performed on the solid phase collected after 3 d (a, b), and 3 mo (c, d) of interaction for uncoated-nanoparticle ZnO. The carbon atomic orbital C1s energy spectra are displayed in (a) and (c), and the oxygen atomic orbital O1s energy spectra are displayed in (b) and (d). The C1s peak corresponding to C in carbonate phases (287.7  0.5 eV) and the O1s peak for O in carbonates (530.0  0.7 eV) increase significantly with time.

346

Environ Toxicol Chem 33, 2014

A. Gelabert et al.

700

[Zn] dissolved ( M)

600 500 400 pH = 7.8

300 200 Zn-L

100

Zn2+

0 0

pH = 8.2 DMT

500

1000 1500 Time (h)

2000

2500

Figure 4. Coated-nanoparticle ZnO dissolution kinetics. Ultrafiltration technique (UFT) Zn concentrations reported as a function of time at pH 7.8 (squares) and pH 8.2 (diamonds). The gray circles are the measured free Zn2þ concentration (by DMT). The 2 rectangles represent the calculated Zn speciation in the ultrafiltrate at 336 h based on synthetic seawater composition.

the end of the experiment, at 272 mmol L1  18 mmol L1 and 19 mmol L1  1 mmol L1 for pH 7.8 and pH 8.2, respectively. As in the nc-NP experiments, the observed decrease in dissolved Zn concentrations was related to the precipitation of carbonated secondary phases. Again, a strong control of pH can be seen, with total dissolved Zn values from 4 times to 15 times higher at pH 7.8 than at pH 8.2. The DMT method was used after 10 d and 16 d of interaction, and only 35 mmol L1 and 37 mmol L1 of free Zn2þ, respectively, were measured in seawater at pH 8.2. Zinc speciation modeling based on synthetic seawater composition and Zn concentrations measured by UFT at 14 d predicts a free Zn2þ concentration of 63 mmol L1 at this pH (represented as rectangles in Figure 4). This difference between DMT and UFT modeled concentrations indicates that a nonnegligible fraction of total dissolved Zn measured by UFT is actually complexed by

the organic molecule moieties released during the c-NP ZnO coating dissolution, which is not taken into account during the modeling. These organic–Zn complexes are calculated to account for approximately 20% of the total dissolved Zn. Moreover, the presence of these organic ligands in solution results in an increased UFT dissolved Zn concentration at steady state compared with nc-NP ZnO. The organic coating may also influence the dissolution kinetics of the nanoparticles, and may be responsible for the late dissolution maximum reached after only 7 d. This may be the result of the protective effect of the organic shell around the ZnO/Zn(OH)2 core. As a result, the c-NP dissolution is a 2-step process, with the ZnO/ Zn(OH)2 core dissolution being kinetically limited by the organic coating solubilization. Furthermore, after 14 d of interaction at pH 8.2, TEM observations revealed the presence of ZnO nanoparticles included in macrostructures composed of amorphous carbon, thus lowering the contact between nanoparticles and bulk solution (Figure 5), and limiting the diffusion of Zn2þ from the mineral surfaces to the bulk solution. As a result, it is likely that the presence of organic coatings lowers the nanoparticle dissolution kinetics. However, Zn was detected by electron-dispersive X-ray spectroscopy not only on the nanoparticle spots but almost everywhere. Hence, one can assume that these organic structures contain both complexed Zn and ZnO nanoparticles and that the amount of dissolved Zn measured in the ultrafiltrates is probably underestimated. The final decrease in UFT-dissolved Zn concentrations observed after 30 d was similar to the profile obtained for nc-NPs. It may be the result of a late degradation of organic ligands inducing a release of dissolved Zn that then precipitates into carbonated secondary phases, or, more likely, different reaction kinetics between the 2 reactions in competition: Zn complexation by organic ligands and formation of Zn carbonates. Then, the late precipitation of a highly insoluble phase could also induce, by a competition effect, the release of ions formerly complexed by the organic ligands. Dissolution kinetics

In our system, characterization of the collected solid phase demonstrated the formation of hydrozincite, Zn5(OH)6(CO3)2,

Figure 5. Transmission electron microscopy (TEM) pictures (b, c) and corresponding electron-dispersive X-ray spectroscopy (EDXS) maps (a) for solid samples collected after 3 mo of interaction between coated-nanoparticle ZnO and seawater. The TEM image demonstrates the presence of nanosized particles (20 nm) inside an organic matrix (b), and the red circle represents the zoomed area corresponding to image (c). The shape and size of particles found in image (c) correspond to native ZnO. The EDXS maps (a) display the location of O, C, and Zn in image (b), showing that Zn is largely distributed inside the organic matrix instead of being concentrated at the particle locations.

ZnO nanoparticle life cycle in synthetic seawater

Environ Toxicol Chem 33, 2014

with time according to the precipitation reaction 5Zn2þ þ 6HO þ 2CO2 3 , Zn5 ðOHÞ6 ðCO3 Þ2

ð1Þ

Thus, considering a control over the solution chemistry by Zn carbonate precipitation, the net experimental precipitation rate (R ¼ D[Zn2þ]/Dt) can be expressed as the difference between the precipitation rate (Rp) and the dissolution rate (Rd), and can be written as follows R ¼ ðRp  Rd Þ ¼ kp ðaZn2þ Þn1 ðaOH Þn3  kd ðaZn5 ðOHÞ6 ðCO3 Þ2 Þn4 ð2Þ Where kp and kd are the forward and backward reaction constants, and ai and ni are the activity and the partial reaction order of component i, respectively. Given the high oversaturation (V) for the hydrozincite phase (log [V] ranged from 1.5 to 5.0 during the first 2 wk of the experiment), this reaction is largely controlled by the precipitation rate, and no significant effect arises from dissolution (i.e., Rd ! 0) [48]. The hydroxide and carbonate activities can be considered constant during the first days of the experiment and Equation 2 can be expressed as  n1 d ð½Zn2þ Þ ¼ kp 0 Zn2þ dt

ð3Þ

Where k’p ¼ kp  (gZn2þ)n1  (aCO32)n2  (aOH)n3, and gZn2þ is the activity coefficient. Assuming a pseudo first-order precipitation kinetics in synthetic seawater, the linearization of the data corresponding to the initial decrease in Zn2þ concentration (the first 50 h of experiment) allows us to determine kinetics constant values k’p of 208  104 mol L1 h1  15  104 mol L1 h1 at pH 7.7 and 57  104 mol L1 h1  11  104 mol L1 h1 at pH 8.2. Based on these k’p values and considering a pseudo first order for hydrozincite precipitation in our system, the theoretical Zn2þ concentrations in solution were simulated beyond 50 h as a function of time for pH 7.7 and pH 8.2 (Figure 1, red and blue lines, respectively). To be able to visually compare experimental and theoretical data, the theoretical kinetics equation was corrected from the steady-state Zn2þ concentration values by adding a 100 mM constant at pH 7.7, and a 20 mM constant at pH 8.2. For both pH values, the experimental data plotted correctly to the corresponding theoretical line, suggesting a global pseudo first-order precipitation for hydrozincite in our system. Seawater compared with freshwater systems

When compared with the nc-NP dataset collected in natural freshwater [22], a different slope for the decrease in free Zn concentration was observed. Actually, after 3 mo of interaction the freshwater system did not reach the predicted thermodynamic steady state, while this steady state was attained only after 1 mo (700 h) in seawater for both pH values. For pH values in the range of 8.0 to 8.2, the initial precipitation kinetics in freshwater (k’p ¼ 182  104 mol L1 h1) was high compared with the seawater system (57  104 mol L1 h1  11  104 mol L1 h1). However, at longer times, this trend was reversed, and a considerable decrease in the precipitation constant occurred for the freshwater system (k’p ¼ 8  104 mol L1 h1), in contrast to seawater. It has been postulated that this decrease in precipitation kinetics for the freshwater system could be the result of the presence of nanoparticle aggregates [20], or the

347

embedding of the dissolving nanoparticles in a carbonated matrix [22]. The typical point of 0 charge reported for ZnO nanoparticles is more than 9.1 [49], whereas the reported isoelectric point for synthetic Zn(OH)2 is approximately 9.8 [50]. Given the small differences in pH for both systems studied, and the considerable pH differences between our experimental settings and the reported Zn(OH)2 isoelectric point value, the electrostatic repulsion and aggregation states may not be significantly different in seawater and freshwater systems. In addition, 2 major differences between freshwater and synthetic seawater are the presence of natural organic matter, which is able to inhibit aggregation, in the freshwater experiment (2.5 mg L1) [42,49], and the higher ionic strength in synthetic seawater, which should favor aggregation by compaction of the electric double layer. In this case, aggregation should occur preferentially in seawater, and result in lower dissolution kinetics. This trend was observed initially, with lower dissolution kinetics for seawater compared with freshwater experiments: k’p of 57  104 mol L1 h1 and 182  104 mol L1 h1, respectively. Moreover, the 2 systems are oversaturated relative to the carbonate phases, and passivation of the ZnO surfaces by carbonate precipitation has been observed in both systems. Because these carbonate coatings are also present in the seawater settings, the observed 2nd step lower kinetics for the freshwater system cannot only be the result of a carbonated matrix formation. However, the presence of organic matter can inhibit the formation of carbonates, as shown for calcite with sorption of humic substances onto the calcite surface nucleation sites [51–53]. Because the dissolved organic carbon content of Seine River water is 2.5 mg/L and the maximum dissolved Zn concentration is only 20 mM in this system, a possible hypothesis is that the presence of fulvic substances is responsible for the decrease in hydrozincite precipitation kinetics in the freshwater system. CONCLUSIONS

In summary, we have developed a robust protocol to measure ZnO nanoparticle dissolution in natural waters, based on the combined use of DMT and 1000-Da pore size ultrafiltration. Application of this analytical strategy allowed us to compare c-NP and nc-NP life cycles in synthetic seawater, highlighting the strong control imposed on nanoparticle dissolution by an organic coating. Given the high occurrence of these coatings on ENPs, their presence constitutes a key parameter to be considered when toxicity experiments are conducted in natural or biological media, or when the fate of nanoparticles in environmental settings is being predicted. Thus, further investigations focusing on the effects of different commercial organic coatings need to be conducted. In addition, we have shown the importance of the inorganic Zn(OH)2 layer during the initial dissolution steps, as well as the control imposed by the precipitation of hydrozincite in our systems. Assuming a pseudo first-order kinetics for hydrozincite precipitation in nc-NP experiments allowed us to calculate kinetics constants of k’p ¼ 208  104 mol L1 h1  15  104 mol L1 h1 at pH 7.7 and 57  104 mol L1 h1  11  104 mol L1 h1 at pH 8.2. Importantly, the precipitation of such carbonated species may potentially result in nanoparticle surface passivation, as supported by XPS measurements. Finally, the electrolyte pH, ionic strength, and organic matter content have been shown to constitute major parameters in ZnO nanoparticle life cycles by partly controlling their dissolution. It follows from the results of the present study that the persistence of ZnO nanoparticles mainly depends on the relative kinetics difference between 2 competitive processes: Zn–

348

Environ Toxicol Chem 33, 2014

carbonaceous species precipitation and nanoparticle dissolution. Thus, although it is rapid, ZnO nanoparticle total dissolution is far from being instantaneous, and the free Zn(II) concentration evolves over at least 1 mo before the system reaches steady state. Thus, if ecotoxicological studies show a rapid ZnO dissolution in the short term [28,34,35], persistence of nanoparticles and free Zn(II) concentration evolution should also be considered, especially in carbonated waters. Even over the long term, the coexistence in toxicological experiments of these different forms of Zn(II), which are able to generate reactive oxygen species or free metal stress, has to be carefully taken into account for an accurate understanding of the toxicity results [28–36]. Furthermore, the present study was conducted at high nanoparticle concentrations, which are appropriate for laboratory conditions in ecotoxicological research. However, lower concentrations are expected in environmental settings, in which some reactivity modifications will certainly occur. For instance, aggregation state is favored at a high nanoparticle concentration [26], thus lowering the dissolution kinetics by minimizing the particle surface exposed to the solution. In addition, given the rapid initial dissolution of ZnO nanoparticles observed in our systems, and the continuous renewal of the solution occurring in open natural waters, faster ZnO nanoparticle dissolution can be expected in environmental settings. Consequently, the ultimate fate of nanoparticles in natural waters remains linked to the relative kinetic equilibrium between dissolution and surface stabilization by carbonaceous species precipitation. However, such kinetics can be strongly modified by external parameters (for instance, pH, presence of natural organic matter, nanoparticle concentration, etc.), and further investigation is required to determine their impact on the fate of ZnO ENPs in natural settings. SUPPLEMENTAL DATA

Table S1. Figures S1–S2. (1.5 MB DOCX). Acknowledgment—The present study was supported by grant APR Environnement Sante Travail-07-28 of the Agence Française de Securite Sanitaire de l’Environnement et du Travail, and the French National Program EC2CO (Ecosphère continentale et côtière) NANOSOL from the Institut National des Sciences de l’Univers. We are grateful to Kobo Products, who kindly provided the nanoparticles. Three anonymous reviewers as well as the associate editor are thanked for their critical comments.

REFERENCES 1. Auffan M, Rose J, Bottero J-Y, Lowry GV, Jolivet J-P, Wiesner MR. 2009. Towards a definition of inorganic nanoparticles from an environmental, health and safety perspective. Nat Nanosci 4:634– 641. 2. Hochella MF, Lower SK, Maurice PA, Penn RL, Sahai N, Sparks DL, Twining BS. 2008. Nanominerals, mineral nanoparticles, and earth systems. Science 319:1631–1635. 3. Wiesner MR, Lowry GV, Jones KL, Hochella JMF, Di Giulio RT, Casman E, Bernhardt ES. 2009. Decreasing uncertainties in assessing environmental exposure, risk, and ecological implications of nanomaterials. Environ Sci Technol 43:6458–6462. 4. United Nations Environment Programme. 2007. Emerging challenges— Nanotechnology and the environment. Global Environment Outlook Year Book. Nairobi, Kenya. 5. Gottschalk F, Sonderer T, Scholz RW, Nowack B. 2009. Modeled environmental concentrations of engineered nanomaterials (TiO2, ZnO, Ag, CNT, fullerenes) for different regions. Environ Sci Technol 43:9216–9222. 6. Kaegi R, Ulrich A, Sinnet B, Vonbank R, Wichser A, Zuleeg S, Simmler H, Brunner S, Vonmont H, Burkhardt M, Boller M. 2008. Synthetic TiO2 nanoparticle emission from exterior facades into the aquatic environment. Environ Pollut 156:233–239.

A. Gelabert et al. 7. Nowack B, Bucheli TD. 2007. Occurrence, behavior and effects of nanoparticles in the environment. Environ Pollut 150:5–22. 8. Djurisic AB, Chen X, Leung YH, Man Ching Ng A. 2012. ZnO nanostructures: Growth, properties and applications. J Mater Chem 22:6526–6535. 9. Becheri A, Dürr M, Lo Nostro P, Baglioni P. 2008. Synthesis and characterization of zinc oxide nanoparticles: Application to textiles as UV-absorbers. J Nanoparticle Res 10:679–689. 10. Burnett ME, Wang SQ. 2011. Current sunscreen controversies: A critical review. Photodermatol Photoimmunol Photomed 27:58–67. 11. Waychunas GA, Zhang H. 2008. Structure, chemistry, and properties of mineral nanoparticles. Elements 4:381–387. 12. Mahendra S, Zhu H, Colvin VL, Alvarez PJ. 2008. Quantum dot weathering results in microbial toxicity. Environ Sci Technol 42:9424– 9430. 13. Hardman R. 2006. A toxicologic review of quantum dots: Toxicity depends on physicochemical and environmental factors. Environ Health Perspect 114:165–172. 14. Foss Hansen S, Larsen BH, Olsen SI, Baun A. 2007. Categorization framework to aid hazard identification of nanomaterials. Nanotoxicology 1:243–250. 15. Moore MN. 2006. Do nanoparticles present ecotoxicological risks for the health of the aquatic environment? Environ Int 32:967–976. 16. Wiesner MR, Lowry GV, Alvarez P, Dionysiou D, Biswas P. 2006. Assessing the risks of manufactured nanomaterials. Environ Sci Technol 40:4336–4345. 17. Neal A. 2008. What can be inferred from bacterium–nanoparticle interactions about the potential consequences of environmental exposure to nanoparticles? Ecotoxicology 17:362–371. 18. Handy R, Owen R, Valsami-Jones E. 2008. The ecotoxicology of nanoparticles and nanomaterials: Current status, knowledge gaps, challenges, and future needs. Ecotoxicology 17:315–325. 19. Chinnapongse SL, MacCuspie RI, Hackley VA. 2011. Persistence of singly dispersed silver nanoparticles in natural freshwaters, synthetic seawater, and simulated estuarine waters. Sci Total Environ 409:2443– 2450. 20. David CA, Galceran J, Rey-Castro C, Puy J, Companys E, Salvador J, Monn J, Wallace R, Vakourov A. 2012. Dissolution kinetics and solubility of ZnO nanoparticles followed by AGNES. J Phys Chem C 116:11758–11767. 21. Reed RB, Ladner DA, Higgins CP, Westerhoff P, Ranville JF. 2012. Solubility of nano-zinc oxide in environmentally and biologically important matrices. Environ Toxicol Chem 31:93–99. 22. Sivry Y, Gelabert A, Ferrari R, Juillot F, Menguy N, Benedetti M. 2013. Impact of organic coating on the persistence of manufactured ZnO nanoparticles in freshwater. Chemosphere, in press DOI: 10.1016/j. chemosphere.2013.09.110. 23. Langford KH, Thomas KV. 2008. Inputs of chemicals from recreational activities into the Norwegian coastal zone. J Environ Monit 10:894–898. 24. Stokes, Diffey. 1999. The water resistance of sunscreen and day-care products. Br J Dermatol 140:259–263. 25. Bennett SW, Keller AA. 2011. Comparative photoactivity of CeO2, gFe2O3, TiO2 and ZnO in various aqueous systems. Appl Catalysis B Environ 102:600–607. 26. Keller AA, Wang H, Zhou D, Lenihan HS, Cherr G, Cardinale BJ, Miller R, Ji Z. 2010. Stability and aggregation of metal oxide nanoparticles in natural aqueous matrices. Environ Sci Technol 44:1962–1967. 27. Miller RJ, Lenihan HS, Muller EB, Tseng N, Hanna SK, Keller AA. 2010. Impacts of metal oxide nanoparticles on marine phytoplankton. Environ Sci Technol 44:7329–7334. 28. Brayner R, Ferrari-Iliou R, Brivois N, Djediat S, Benedetti MF, Fivet F. 2006. Toxicological impact studies based on Escherichia coli bacteria in ultrafine ZnO nanoparticles colloidal medium. Nano Lett 6:866–870. 29. Franklin NM, Rogers NJ, Apte SC, Batley GE, Gadd GE, Casey PS. 2007. Comparative toxicity of nanoparticulate ZnO, Bulk ZnO, and ZnCl2 to a freshwater microalga (Pseudokirchneriella subcapitata): The importance of particle solubility. Environ Sci Technol 41:8484–8490. 30. Song W, Zhang J, Guo J, Zhang J, Ding F, Li L, Sun Z. 2010. Role of the dissolved zinc ion and reactive oxygen species in cytotoxicity of ZnO nanoparticles. Toxicol Lett 199:389–397. 31. Shaw BJ, Handy RD. 2011. Physiological effects of nanoparticles on fish: A comparison of nanometals versus metal ions. Environ Int 37: 1083–1097. 32. Quik JTK, Vonk JA, Hansen SF, Baun A, Van De Meent D. 2011. How to assess exposure of aquatic organisms to manufactured nanoparticles? Environ Int 37:1068–1077. 33. Poynton HC, Lazorchak JM, Impellitteri CA, Smith ME, Rogers K, Patra M, Hammer KA, Allen HJ, Vulpe CD. 2010. Differential gene

ZnO nanoparticle life cycle in synthetic seawater

34. 35.

36.

37. 38. 39.

40. 41. 42.

expression in Daphnia magna suggests distinct modes of action and bioavailability for ZnO nanoparticles and Zn ions. Environ Sci Technol 45:762–768. Li M, Zhu L, Lin D. 2011. Toxicity of ZnO Nanoparticles to Escherichia coli: Mechanism and the influence of medium components. Environ Sci Technol 45:1977–1983. Fairbairn EA, Keller AA, Mädler L, Zhou D, Pokhrel S, Cherr GN. 2011. Metal oxide nanomaterials in seawater: Linking physicochemical characteristics with biological response in sea urchin development. J Hazard Mater 192:1565–1571. Xia T, Kovochich M, Liong M, Mädler L, Gilbert B, Shi H, Yeh JI, Zink JI, Nel AE. 2008. Comparison of the mechanism of toxicity of zinc oxide and cerium oxide nanoparticles based on dissolution and oxidative stress properties. ACS Nano 2:2121–2134. Price NM, Morel FMM. 1990. Cadmium and cobalt substitution for zinc in a marine diatom. Nature 344:658–660. Miao A-J, Zhang X-Y, Luo Z, Chen C-S, Chin W-C, Santschi PH, Quigg A. 2010. Zinc oxide–engineered nanoparticles: Dissolution and toxicity to marine phytoplankton. Environ Toxicol Chem 29:2814–2822. Yu L-p, Fang T, Xiong D-w, Zhu W-t, Sima X-f. 2011. Comparative toxicity of nano-ZnO and bulk ZnO suspensions to zebrafish and the effects of sedimentation, · OH production and particle dissolution in distilled water. J Environ Monit 13:1975–1982. Wong S, Leung P, Djurišić A, Leung K. 2010. Toxicities of nano zinc oxide to five marine organisms: Influences of aggregate size and ion solubility. Anal Bioanal Chem 396:609–618. Mudunkotuwa IA, Rupasinghe T, Wu C-M, Grassian VH. 2011. Dissolution of ZnO nanoparticles at circumneutral pH: A study of size effects in the presence and absence of citric acid. Langmuir 28:396–403. Bian S-W, Mudunkotuwa IA, Rupasinghe T, Grassian VH. 2011. Aggregation and dissolution of 4 nm ZnO nanoparticles in aqueous environments: Influence of pH, ionic strength, size, and adsorption of humic acid. Langmuir 27:6059–6068.

Environ Toxicol Chem 33, 2014

349

43. Petosa AR, Jaisi DP, Quevedo IR, Elimelech M, Tufenkji N. 2010. Aggregation and deposition of engineered nanomaterials in aquatic environments: Role of physicochemical interactions. Environ Sci Technol 44:6532–6549. 44. Botta C, Labille J, Auffan M, Borschneck D, Miche H, Cabie M, Masion A, Rose J, Bottero J-Y. 2011. TiO2-based nanoparticles released in water from commercialized sunscreens in a life-cycle perspective: Structures and quantities. Environ Pollut 159:1543–1550. 45. Dickson AG. 1990. Thermodynamics of the dissociation of boric acid in synthetic seawater from 273.15 to 318.15 K. Deep Sea Res A 37:755–766. 46. Jouvin D, Louvat P, Juillot F, Marechal CN, Benedetti MF. 2009. Zinc isotopic fractionation: Why organic matters. Environ Sci Technol 43:5747–5754. 47. Gustafsson JP. 2012. Visual MINTEQ 3.0 Program. http://www.lwr.kth. se/english/OurSoftWare/Vminteq/index.html. 48. Lopez O, Zuddas P, Faivre D. 2009. The influence of temperature and seawater composition on calcite crystal growth mechanisms and kinetics: Implications for Mg incorporation in calcite lattice. Geochim Cosmochim Acta 73:337–347. 49. Zhou D, Keller AA. 2010. Role of morphology in the aggregation kinetics of ZnO nanoparticles. Water Res 44:2948–2956. 50. Kosmulski M. 2009. Compilation of PZC and IEP of sparingly soluble metal oxides and hydroxides from literature. Adv Colloid Interface Sci 152:14–25. 51. Hoch AR, Reddy MM, Aiken GR. 2000. Calcite crystal growth inhibition by humic substances with emphasis on hydrophobic acids from the Florida Everglades. Geochim Cosmochim Acta 64:61–72. 52. Lebron I, Suarez DL. 1998. Kinetics and mechanisms of precipitation of calcite as affected by PCO2 and organic ligands at 258C. Geochim Cosmochim Acta 62:405–416. 53. Lin Y-P, Singer PC, Aiken GR. 2005. Inhibition of calcite precipitation by natural organic material: Kinetics, mechanism, and thermodynamics. Environ Sci Technol 39:6420–6428.