The impact of size on the fate and toxicity of nanoparticulate silver in aquatic systems

Chemosphere 93 (2013) 359–365

Contents lists available at SciVerse ScienceDirect

Chemosphere journal homepage: www.elsevier.com/locate/chemosphere

The impact of size on the fate and toxicity of nanoparticulate silver in aquatic systems Brad M. Angel ⇑, Graeme E. Batley, Chad V. Jarolimek, Nicola J. Rogers Centre for Environmental Contaminants Research, CSIRO Land and Water, Locked Bag 2007, Kirrawee, NSW 2232, Australia

h i g h l i g h t s  Citrate-coated AgNPs aggregated unless humic acid present.  PVP-coated AgNPs did not aggregate significantly in any medium.  Humic acid reduced toxicity but did not have a strong effect on dissolution.  Greater AgNP dissolution for higher chloride concentrations.  Toxicity was in the order micron silver  nanoparticulate silver < ionic silver.

a r t i c l e

i n f o

Article history: Received 4 March 2013 Received in revised form 18 April 2013 Accepted 26 April 2013 Available online 31 May 2013 Keywords: Nanoparticle Silver Dissolution Aggregation Toxicity

a b s t r a c t The increased use of silver nanomaterials presents a risk to aquatic systems due to the high toxicity of silver. The stability, dissolution rates and toxicity of citrate- and polyvinylpyrrolidone-coated silver nanoparticles (AgNPs) were investigated in synthetic freshwater and natural seawater media, with the effects of natural organic matter investigated in freshwater. When sterically stabilised by the large PVP molecules, AgNPs were more stable than when charge-stabilised using citrate, and were even relatively stable in seawater. In freshwater and seawater, citrate-coated AgNPs (Ag–Cit) had a faster rate of dissolution than PVP-coated AgNPs (Ag–PVP), while micron-sized silver exhibited the slowest dissolution rate. However, similar dissolved silver was measured for both AgNPs after 72 h in freshwater (500–600 lg L1) and seawater (1300–1500 lg L1), with higher concentrations in seawater attributed to chloride complexation. When determined on a mass basis, the 72-h IC50 (inhibitory concentration giving 50% reduction in algal growth rate) for Pseudokirchneriella subcapitata and Phaeodactylum tricornutum and the 48-h LC50 for Ceriodaphnia dubia exposure to Ag+ (1.1, 400 and 0.11 lg L1, respectively), Ag–Cit (3.0, 2380 and 0.15 lg L1, respectively) and Ag–PVP (19.5, 3690 and 2.0 lg L1, respectively) varied widely, with toxicity in the order Ag+ > Ag–Cit > Ag–PVP. Micron-sized silver treatments elicited much lower toxicity than ionic Ag+ or AgNP to P. subcapitata. However, when related to the dissolved silver released from the nanoparticles the toxicities were similar to ionic silver treatments. The presence of natural organic matter stabilised the particles and reduced toxicity in freshwater. These results indicate that dissolved silver was responsible for the toxicity and highlight the need to account for matrix components such as chloride and organic matter in natural waters that influence AgNP fate and mitigate toxicity. Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved.

1. Introduction Silver is one of the most widely used nanomaterials and probably one of the oldest. With the rapidly expanding use of nanomaterials in industrial applications and consumer products, the potential to enter aquatic systems and reach concentrations that can adversely affect biota will increase. Predicted environmental concentrations of silver nanomaterials are in the range <0.03– 0.32 lg L1 (Batley et al., 2013). As part of the regulatory manage⇑ Corresponding author. Tel.: +61 2 9710 6851; fax: +61 2 9710 6800. E-mail address: [email protected] (B.M. Angel).

ment of manufactured nanomaterials, the key question is whether nanosize imparts a different fate and toxicity from that of the equivalent macroforms. Understanding the critical factors that contribute to differences in behaviour of nanomaterials are important for regulation, including aggregation, solubility, dissolution, and the effect of coatings used in specialised formulations. Silver nanoparticles (AgNPs) may be uncoated or coated with layers such as the citrate anion, the polymer, polyvinylpyrrolidone (PVP), or surfactants (e.g. Tween-80), which may enhance electrostatic or steric repulsion and increase stability in suspension. The biocidal activity of AgNPs in aquatic systems is often attributable to ionic silver, produced by the oxidation of surface Ag0 to ionic

0045-6535/$ - see front matter Crown Copyright Ó 2013 Published by Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.chemosphere.2013.04.096

360

B.M. Angel et al. / Chemosphere 93 (2013) 359–365

silver (Ag+) and subsequent dissolution via desorption (Miao et al., 2009; Sotiriou and Pratsinis, 2010). However, ionic silver is not always solely responsible for toxicity, as shown by the direct uptake of waterborne AgNP by snails (Croteau et al., 2011). Soluble silver is typically separated from nanosilver by either dialysis or ultrafiltration (Navarro et al., 2008; Liu and Hurt, 2010). Few studies have measured the fate and toxicity of AgNPs concurrently, particularly for marine organisms, which makes it difficult to determine the mechanism of toxicity. The current study presents a comparative investigation of the fate and toxicity of ionic, nano- and micron-sized silver in aquatic systems. Toxicity to the freshwater alga, Pseudokirchneriella subcapitata), the marine alga, Phaeodactylum tricornutum and the freshwater daphnid, Ceriodaphnia dubia were measured. The effects of the aquatic medium, the duration in solution, and the particle concentration on the dissolution of AgNPs was investigated to provide information on the mechanism of silver toxicity. 2. Materials and methods 2.1. General analytical All new plasticware was acid-washed with 10% v/v nitric acid (Tracepur, Merck, Darmstadt, Germany) by soaking for 24 h followed by rinsing with at least five rinses with deionised water (18 MX cm, Milli-Q, Millipore). All plastic- and glassware exposed to nano- and micron-sized particles in the various experiments was soaked in a 4% v/v HNO3/12% v/v HCl (Tracepur, Merck, Darmstadt, Germany) solution for at least 72 h to remove contamination and rinsed with Milli-Q water before use. Measurements of pH used a Wissenschaftlich-Technische Werkstattan (WTW, Weilheim, Germany) meter equipped with a pH probe (Orion sure-flow combination pH 9165BN). 2.2. Reagents The physico-chemical characterisations of nanoparticles were studied in Milli-Q water and/or toxicity test media. The freshwater toxicity tests used filtered (0.22 lm) synthetic fresh water (SFW) prepared using a modified USEPA medium (USEPA, 1994) having a low chloride content (50 lm) (Lee et al., 2005) so that negligible AgCl would form. The SFW was buffered with 2 mM piperazineN,N0 -bisethanesulfonic acid (PIPES: Sigma–Aldrich) and had a hardness of 19 mg CaCO3 L1. The marine toxicity tests used filtered (0.45 lm) natural seawater (SW) (Cronulla, Sydney, Australia) (Angel et al., 2010). The silver particles tested comprised citrate-coated AgNPs (Ag– Cit) obtained from ABC Nanotech (Daejeon City, South Korea) with a nominal particle size of 14 nm, PVP-coated AgNPs (Ag–PVP) obtained from Mercator Gmbh (Magdeburg, Germany) with a nominal particle size of 15 nm, and micron-sized metallic silver with a nominal particle size between 2000 and 3500 nm, obtained from Sigma Aldrich (St. Louis, MO, USA). 2.3. Nanoparticle stability – size and zeta potential measurements Measurements of z-average hydrodynamic diameter (z-ave HD) and zeta potential of the AgNP dispersions were performed by dynamic light scattering (DLS) and electrophoretic light scattering using a Malvern Zetasizer Nano ZS (4 mW He–Ne laser at 633 nm, with MPT-2 autotitrator) (Malvern Instruments, UK) using operating conditions described by Rogers et al. (2010). Suspensions were diluted to 10 mg L1 in MQ, SFW, SFW + Suwannee River humic acid (SRHA, IHSS, St. Paul, MN, USA), or SW. Particle size was also measured immediately after addition to the media using dif-

ferential centrifugal sedimentation (DCS 24 000UHR, CPS Instruments, USA) to produce an absorption intensity curve as a function of time, and conversion of time to an equivalent spherical (Stokes) diameter (Coleman et al., 2011). The surface plasmon resonance (SPR) of the AgNP suspensions was investigated by measuring the kmax using UV–visible spectrometry (Lambda 45, Perkin Elmer) with a 1-cm path length quartz cell. A decrease in the absorbance at kmax is indicative of aggregation and/or sedimentation. The spectra were collected daily after addition of AgNPs to Milli-Q water, SFW, SFW + SRHA and SW. 2.4. Measurements of nanoparticle dissolution Equilibrium dialysis was used to measure the dissolution of 40 mg L1 of AgNPs in each suspension, with ultrafiltration used for AgNP concentrations 6 1 mg L1. Preliminary tests indicated good agreement between the concentrations of dissolved silver measured by the two techniques. The dissolution tests were performed in SFW, SFW + SRHA, SW, and various concentrations of NaCl. The equilibrium dialysis method as described by Franklin et al. (2007) used membranes with a molecular weight cut-off (MWCO) of 1 kDa (Cole Parmer Spectra/Por 7). The membranes were washed with Milli-Q water, filled with 20 mL of Milli-Q water, sealed with plastic dialysis clips and placed into triplicate 40 mg L1 AgNP solutions to initiate tests. The test solutions were stirred with PTFE magnetic stirrer bars under continuous light and temperature (24 °C) conditions for 72 h. At each sampling time a cell from each of three replicates was removed and sub-sampled for the dialysed metal concentration, and a filtered (0.1 lm) sample was taken from the bulk solution. Ultrafiltration used the method described by Cornelis et al. (2010), with Macrosep ultrafiltration devices (Pall-Gellman, Port Washington, USA) with a 1 kDa MWCO. The ultrafiltration devices were pre-conditioned with 0.1 M copper nitrate to minimise ionic silver adsorption onto membrane surfaces, and rinsed with Milli-Q before use. The solutions were centrifuged at 1400g for 20 min followed by sub-sampling of the filtrate. 2.5. Algal bioassays The toxicity of ionic, nanoparticulate and micron-sized silver was assessed using the freshwater alga, P. subcapitata (ATCC 22662, Maryland, USA) and the marine alga, P. tricornutum (Bohlin, Hobart, Australia). P. subcapitata was cultured axenically in SFW under continuous light (Philips TL 40 W cool white fluorescent lighting, Danvers, MA, USA, 70 lmol photons m2 s1) at 24 °C. P. tricornutum was cultured in f2 growth medium (half-strength f medium, according to the method described by Levy et al. (2008). The tests were performed in 30-mL borosilicate glass mini-vials using 6 mL of either SFW (P. subcapitata), or SW (P. tricornutum). Exponentially-growing cells were harvested by centrifugation (700g, 7 min), rinsed three times with the test medium, and inoculated into test vials to achieve an initial cell density of 3  104 cells mL1. Algal cell densities were counted daily using a 4-colour BD-FACSCalibur™ (Becton Dickinson BioSciences, San Jose, CA, USA) flow cytometer (Levy et al., 2008). The IC50 value (the inhibitory concentration giving 50% reduction in algal growth rate after 72 h compared to controls) was calculated for each exposure using linear interpolation (ToxCalc Version 5.0.23C, Tidepool Software, McKinleyville, CA, USA). 2.6. Water flea bioassay Cultures of the water flea, C. dubia were originally isolated from Lake Parramatta, Sydney, Australia. Mass cultures were maintained

361

B.M. Angel et al. / Chemosphere 93 (2013) 359–365

in 500 mL of 5-fold diluted mineral (PerrierÒ) water containing 0.12 mM Cl and supplemented with 100 mg L1 of vitamin B12 and 5 lg L1 of Na2SeO3, at 25 ± 18 °C under ambient laboratory light levels (Binet et al., 2010). The cultures were fed 400,000 cells mL1 of P. subcapitata and 6 mg L1 fish food supplement (Algro-Natural, Cognis, Tullamarine, Victoria, Australia) three times a week. Neonates (<24-h old) were transferred into 30-mL borosilicate glass mini-vials containing 20 mL of FSW to initiate tests (Binet et al., 2010). The mobile organisms were counted after 24 and 48 h following vial agitation. The EC50 values were calculated for each exposure using linear interpolation (ToxCalc Version 5.0.23C, Tidepool Software, McKinleyville, CA). 2.7. Metal analyses The concentrations of silver in SFW samples were measured using inductively coupled plasma-mass spectrometry (ICPMS) (Agilent, 7500CE). The seawater and high chloride samples were analysed for silver using inductively coupled atomic emission spectrometry (ICPAES) (Varian730 ES). Instrument calibrations were performed using matrix-matched (Plasma Chem. Corp., Eaglewood, FL, USA) standards and analytical drift was monitored periodically throughout each analysis according to manufacturers standards. 3. Results and discussion 3.1. Particle size and zeta potential Data for the characterisation of silver particles immediately after addition to Milli-Q water, SFW, and SW are shown in Supplementary Table S1. The differential centrifugal sedimentation (DCS) measurements of the Ag–Cit and Ag-PVP nanoparticles immediately after addition to Milli-Q water and SFW were similar to the nominal size reported by the suppliers for these suspensions. The DLS measurements of z-average hydrodynamic diameter (z-ave HD) for Ag–Cit were similar to those measured by DCS, whereas the DLS measurements of Ag–PVP were smaller than those measured by DCS. This difference may have been related to the density of Ag–PVP being significantly less than Ag0 due to the relatively large PVP coating affecting the DCS weight calculations. These measurements indicated that neither of the AgNPs underwent immediate aggregation upon addition to MQ or SFW, and PVPcoated silver was even relatively stable in SW. The Ag–Cit was not stable in SW and immediately aggregated to form large particles in the millimetre size range that could not be measured due to rapid sedimentation. The Ag–Cit and Ag–PVP z-ave HDs measured by DLS after 72 h in Milli-Q water, SFW, SFW + SRHA and SW are shown in Supplementary Fig. S1A and B. The Ag–Cit z-ave HDs (mean ± S.D.) in Milli-Q water was initially 14.4 ± 0.1 nm and did not increase significantly until after 24 h. When dispersed in SFW, the z-ave HD of the Ag–Cit was initially 13.4 ± 1.2 nm, but exhibited a steady increase in size over 72 h (Huynh and Chen, 2011). However, when dispersed in SFW containing 4 mg L1 of SRHA the z-ave HD remained steady in the range 10–20 nm over 72 h, indicating the humic acid prevented aggregation. The z-ave HD of the Ag-PVP measured by DLS immediately after dispersion in MQ, SFW, and SFW containing 4 mg L1 SRHA was between 4 and 10 nm, and remained steady over 72 h in each respective solution (Supplementary Fig. S1B). The z-ave HD of the Ag–PVP in SW was initially 9.9 ± 1.9 nm and increased to 20 ± 10 nm after 72 h. The high stability of the Ag–PVP in all solutions can be attributed to the steric hindrance of the large noncharged polymer, which has been shown to be more effective than

charge stabilisation by citrate (Huynh and Chen, 2011). The surface plasmon resonance measurements correlated well with the DLS and DCS measurements as shown by the decrease over time in the kmax for Ag–Cit in Milli-Q water, SFW and SW (Supplementary Figs. S1C and D, S2 and S3). The zeta potentials of the Ag–Cit in Milli-Q water and SFW were in the range 46 to 26 mV, while those of Ag–PVP were in the range 18 to 3 mV. A zeta potential between 30 mV and +30 mV typically indicates that particles are unstable and will tend to aggregate (Rogers et al., 2010). However, zeta potential was a poor predictor of aggregation in the current study as Ag–Cit aggregated while Ag–PVP did not. The stability of Ag–PVP indicated that zeta potential may not be appropriate for predicting the aggregation of nanoparticles with large polymer coatings. No isoelectric point was observed in the SFW medium (pH 4.0–10.0) for the Ag–Cit and Ag–PVP. 3.2. Nanoparticle solubility Solubility data for 40 mg L1 AgNP samples after 72 h in fresh and marine waters are shown in Table 1 and kinetic data is shown in Fig. 1. The rate of release of dialysed silver into SFW was faster for Ag–PVP than for Ag–Cit, with the majority of the dialysed silver released by Ag–PVP within 8–21 h, while the dialysed silver from Ag–Cit had not plateaued after 72 h. Similar dialysed silver concentrations were released from 40 mg L1 of Ag–Cit and Ag–PVP in SFW after 72 h suggesting the particle coating affected the initial rate of silver release, but had little effect on longer-term solubility. The rates of silver release from Ag–Cit into SFW and SFW + SRHA were similar for approximately 32 h, after which the dialysed concentration exhibited a plateau in SFW + SRHA, but continued to increase in the absence of SRHA. This indicated that the humic acid decreased the rate of oxidative dissolution of Ag and/or the solubility of Ag–Cit in SFW after 32 h, possibly by adsorbing to oxidation sites on the surface of the Ag0 (Dubas and Pimpan, 2008). Humic acids have also been shown to consume H2O2 intermediates generated during Ag0 oxidation in a sacrificial reaction (Wang et al., 2001), and may lead to AgNP formation by reducing Ag+ to Ag0 (Akaighe et al., 2011). Humic acid had a negligible effect on the concentration of dialysed silver released from Ag–PVP in SFW. Similar concentrations of dissolved silver were released from Ag–Cit and Ag–PVP in SW, which was higher than in SFW or SFW + SRHA, as a result of chloride complexation (Liu et al., 2010). The dissolution of micron-sized silver (Ag-Mic) in SFW and SW was slower and dissolved silver concentrations were lower than released by the AgNPs in each respective medium. The dialysed silver concentration was approximately 20 times lower than that measured for either of the AgNPs in SFW after 72 h and approximately 10 times lower than that released by either of the AgNPs in SW after 72 h due to the relatively low surface area of Ag-Mic. 3.3. The effect of particle concentration on dissolution The percentage of silver released into the dissolved phase after 72 h was low (0.7–3.7%) for 40 mg L1 of Ag–Cit and Ag–PVP in

Table 1 The 72-h solubility of 40 mg L1 of silver particles in fresh and marine water and percentage dissolution. Nanoparticle

Ag–Cit Ag–PVP Ag–Mic

72-h Solubility (lg Ag L1) SFW

SFW + 4 mg L1 SRHA

SW

550 ± 40 (1.4%) 560 ± 40 (1.6%) 27 ± 1 (0.7%)

260 ± 20 (0.7%) 550 ± 40 (1.4%) –

1470 ± 200 (3.7%) 1320 ± 230 (3.3%) 110 ± 18 (0.3%)

362

B.M. Angel et al. / Chemosphere 93 (2013) 359–365

800

A Dissolved Ag (µg/L)

Dissolved Ag (µg/L)

2000

1500

1000

500

0 0

24

48

600

400

200

0 0.1

72

Time (h)

1

10

100

1000

10000

100000

AgNP (µg/L)

B

Fig. 2. The effect of particle concentration on the dissolution of Ag–Cit () and Ag– PVP (s) after 72 h in SFW.

1500 400

500

0 0

24

48

72

Time (h)

Dissolved Ag (µg/L)

150

Dissolved Ag (µg/L)

1000

C

120

A

300

200

100

0 1

10

100

1000

10000

100000 1000000

10000

100000 1000000

Cl- (µM) 90 400

B

60 30 0 0

24

48

72

Time (h) Fig. 1. The dissolution of citrate- (A) and PVP-coated (B) nanoparticulate silver and micron-sized silver (C) in synthetic soft water (), synthetic soft water containing 4 mg L1 humic acid (N) and natural seawater (d).

Dissolved Ag (µg/L)

Dissolved Ag (µg/L)

2000

300

200

100

0 1

SFW, SFW + SRHA and SW (Table 1). This nanoparticle concentration was relatively high so the effect of nanoparticle concentration on dissolution was investigated (Fig. 2). For total Ag–Cit concentrations between 1 and 100 lg L1, a linear increase in dissolved silver occurred, which was equivalent to 20–27% dissolution at each nanoparticle concentration. At nanoparticle concentrations of 1000 lg L1 and above, the percentage of dissolution decreased and was 8.4% and 1.0% for 1000 and 40 000 lg Ag–Cit L1, respectively. For Ag–PVP, the percentage dissolution decreased steadily from 52% to 1.6% as the concentration of nanoparticles increased. These results suggest that a significant percentage of silver may be released into the dissolved phase from the relatively low concentrations of AgNPs in the environment, particularly considering nanoparticles may reside in the environment for longer durations. 3.4. Effect of chloride on the dissolution Silver dissolution from Ag–Cit and Ag–PVP was measured as a function of chloride concentration, using 0.5 mg L1 Ag–Cit and Ag–PVP suspended in 0.01–564 (seawater) mM chloride (Fig. 3). For the Ag–Cit, the dissolved silver remained steady between

10

100

1000

Cl- (µM) Fig. 3. The effect of chloride concentration on the dissolution of 0.5 mg L1 Ag–Cit (A) and Ag–PVP (B) after 0.5 h () and 72 h (N).

0.01 and 10 mM chloride. At chloride concentrations higher than 10 mM, the dissolved silver increased significantly for each incremental increase in chloride. Silver speciation modelling (Visual Minteq 3.0) indicated the increase in dissolved silver was accompanied by a corresponding increase in the concentrations of soluble y 3 AgClx complexes, with AgCl4 representing 82% of dissolved silver when marine chloride concentrations of 564 mM were present (Supplementary Fig. S4). After 72 h, 290 ± 30 lg L1 of dissolved silver was released in the 564 mM chloride treatment. This was 58% of the total silver present at the start of the test. Similar behaviour was seen for Ag-PVP as chloride concentrations increased. After 72 h, there was 170 ± 4 lg L1 of dissolved silver released in the 564 mM chloride treatment, which was 34% of the total silver present at the start of the test. For Ag–Cit, the dissolved silver only increased significantly at chloride equivalent to salinities of approximately 3‰ and above. For Ag–PVP, the dissolved silver only increased significantly at

B.M. Angel et al. / Chemosphere 93 (2013) 359–365 Table 2 The toxicity of ionic, nano- and micron-sized silver to P. subcapitata. Material

Ag+ Ag–Cit Ag–Cit + 4 mg L1 SRHA Ag–Cit + 8 mg L1 SRHA Ag–PVP Ag–PVP + 4 mg L1 SRHA Ag–PVP + 8 mg L1 SRHA Ag–Mic

DLS HDa (nm)

72-h IC50 (lg L1) Nominal concentration

Dissolved silver

nd 13.4 ± 1.2 14.7 ± 0.6

1.1 ± 0.5b 3.0 ± 0.7b 5.2 (4.6–5.8)c

0.5 (0.3–0.7)c 0.7 (0.5–1.1)c nd

nd

5.6 (5.5–5.7)c

nd

5.9 ± 4.4 4.2 ± 0.4

19.5 ± 6.1b 36.7 (33.8–38.8)c

3.0 (2.4–3.5)c nd

nd

48.9d

nd

nd

966 (900–1030)c

1.1 (0.9–1.4)c

nd = Experimental parameter not determined. a z-ave HD Determined immediately after addition to solution. b Values represent the mean ± S.D. of three separate bioassays. c 95% Confidence interval. d Insufficient concentration range tested for calculation of confidence intervals.

chloride equivalent to salinities of approximately 0.6‰ and above. These results suggest that organisms that reside in the low salinity zone of estuaries receiving inputs of AgNPs are likely to be exposed to pulses of dissolved silver as the salinity increases during flood tides (Angel et al., 2010). However, at salinities greater than 0.02‰, the concentration of Ag+ decreased and silver speciation y was dominated by AgCl and AgClx complexes that are much less toxic than free silver due to their lower bioavailability (Lee et al., 2005). Therefore, higher salinity causes increased dissolved silver that is likely to be less toxic, with increased aggregation and particle sedimentation is also likely to lead to less organism exposure to nanoparticles.

363

When calculated using the dissolved silver concentration, similar 72-h IC50 values were determined for Ag+ (0.5 lg L1), Ag– Cit (0.7 lg L1) and Ag–Mic (1.1 lg L1), while the IC50 for Ag– PVP (3.0 lg L1) was 3–6 times higher. These results indicate that dissolved Ag was the toxic agent for Ag–Cit and Ag–Mic and was a good predictor of the toxicity. The higher IC50 of Ag–PVP suggests that some of the dissolved silver is in non-bioavailable forms. The dispersant of the Ag–PVP stock contained several organic agents that may have complexed dissolved silver and decreased its bioavailability. 3.6. Toxicity of AgNPs to Phaeodactylum tricornutum in seawater The 72-h IC50 values (mean ± S.D. of three replicate tests) calculated using the total initial silver concentration for P. tricornutum exposure to ionic silver, Ag–Cit and Ag–PVP were 400 ± 110, 2380 ± 1880, 3690 ± 2380 lg L1, respectively, again indicating that ionic silver was more toxic than nanoparticulate silver. There was significant variability in the IC50 values between replicate tests, probably because large variations in the total silver only result in small changes in toxic free silver due to chloride complexation (Lee et al., 2005). When calculated using the dissolved silver concentration the 72-h IC50 of ionic silver (400 ± 110 lg L1) was similar to the value (95% confidence interval) for Ag–PVP (442 (333–500) lg L1) and only marginally more toxic than Ag–Cit (766 (692–816) lg L1, which indicated that dissolved silver was a better predictor of toxicity than the total silver concentration. Dissolved silver was also reported to be responsible for the toxic response elicited in the marine diatom, Thalassiosira weissflogii exposed to PVP-coated AgNPs in a study by Miao et al. (2009). The 72-h IC50 values for P. tricornutum exposed to ionic and nanoparticulate silver were more than two orders of magnitude higher than those for P. subcapitata. 3.7. Toxicity of AgNPs to Ceriodaphnia dubia

3.5. Toxicity of AgNPs to Pseudokirchneriella subcapitata The concentrations of nanoparticulate, micron-sized, and ionic silver that caused 50% growth rate inhibition (IC50) to P. subcapitata are shown in Table 2. When the 72-h IC50 was calculated using the total silver concentration spiked into solution, the IC50 values for ionic silver, Ag–Cit, Ag–PVP and Ag–Mic were all significantly (p < 0.05) different. The IC50 values were in the order Ag+ > Ag–Cit > Ag–PVP  Ag–Mic. Ionic silver (from AgNO3) and the AgNPs were approximately three and two orders of magnitude, respectively, more toxic than the micron-sized silver. There have been few studies of the effect of organic matter on the aquatic toxicology of AgNPs (Fabrega et al., 2009; McLaughlin and Bonzongo, 2012). The presence of SRHA at concentrations representative of local freshwater systems increased the 72-h IC50 values for Ag–Cit and Ag–PVP by factors up to 1.9 and 2.5, respectively, which is likely the result of complexation reducing the bioavailability of dissolved silver. Organic matter may also mitigate the reactivity of the nanoparticle surface (Dubas and Pimpan, 2008; Akaighe et al., 2011) or scavenge free radicals (Wang et al., 2001; Liu and Hurt, 2010). Ultrafiltration was utilised to measure the dissolved silver in toxicity test solutions to provide data on the dose metrics of dissolved silver released from the AgNPs. This was technically difficult because the relatively low concentrations of silver lead to relatively high percentage adsorptive losses of dissolved silver onto the test vials over the duration of the 72-h bioassays. Up to a 50% decrease in dissolved silver was measured for ionic silver treatments. A linear decrease with time was assumed for the purpose of determining the dose of dissolved silver in bioassays.

The 48-h LC50 values for ionic Ag, Ag–Cit and Ag–PVP were 0.11 ± 0.02, 0.15 ± 0.04 and 2.0 ± 0.8 lg L1 total silver, respectively, indicating that ionic silver was marginally more toxic than Ag–Cit, which was 18-fold more toxic than Ag–PVP. The 48-h LC50 values for C. dubia exposure to ionic and nanoparticulate silver were marginally lower than those obtained for each toxicant exposed to P. subcapitata, indicating that C. dubia is more sensitive to silver. The high sensitivity of C. dubia to ionic silver was also reported by Bielmyer et al. (2002), who measured a 48-h LC50 of 0.32 lg L1. The order of toxicity was Ag+ > Ag–Cit > Ag–PVP and was the same as measured for P. subcapitata. When tested in SFW containing 4 mg L1 SRHA, the 48-h LC50 values for Ag–Cit and Ag–PVP were 0.70 (0.54–0.97) and 5.7 lg L1 respectively, which was between 2.9 and 4.7 times less toxic than without SRHA. 3.8. General discussion The toxicity test data indicated significant differences between the toxicity of ionic, nanoparticulate and micron-sized silver to algae and cladocerans. Ionic silver was the most toxic form of silver and its release from nanoparticulate and micron-sized silver explained their toxic mechanism. Micron-sized Ag was the least toxic due to the slow rate of dissolution from its low surface area. The freshwater toxicity tests with the alga, P. subcapitata and the water flea, C. dubia showed that citrate-coated AgNPs were significantly more toxic than PVP-coated AgNPs, despite releasing similar concentrations of dissolved silver into solution, possibly due to complexation by organic dispersants in the Ag–PVP stock solution.

364

B.M. Angel et al. / Chemosphere 93 (2013) 359–365

Supplementary Table S2 summarises the reported toxicity data and demonstrates the variability of AgNP toxicity. Toxicity may be related to the surface coating, the surface charge, the aggregation state, and the release of dissolved silver, and certain characteristics may be interconnected, e.g. coating, aggregation and dissolution. Kennedy et al. (2010) reported that citrate-coated AgNPs were more toxic than PVP-coated AgNPs to Daphnia magna, Pimephales promelas and P. subcapitata in agreement with the results in the current study, but attributed this to differences in dissolution. However, Yang et al. (2012) reported the opposite results for exposure to Caenorhabditis elegans, with greater aggregation of citratecoated AgNP resulting in less dissolved silver release than PVPcoated silver and subsequently less toxicity. Zhao and Wang (2012) also reported different LC50s for D. magna exposure to lactate- and PVP-coated AgNPs due to differences in the release of dissolved silver. El Badawy et al. (2011) reported that PVP–Ag elicited greater toxicity than citrate-coated AgNP to bacillus species, but attributed this to a more positive surface charge resulting in greater interaction with the membrane of the gram-positive bacteria rather than to release of dissolved silver. In the current study, the coatings affected the degree of aggregation, but dissolution was not affected. Considering that the current study showed that citrate-coated AgNPs underwent aggregation while PVP-coated particles did not, the results indicated that the small size of the PVP-coated particles did not elicit greater toxicity. The dissolution of AgNPs was the main driver of toxicity and it was not affected by aggregation, as shown by the similar concentrations of dissolved silver for the aggregated Ag–Cit and non-aggregated Ag–PVP. Kennedy et al. (2010) reported similar LC50s for D. magna exposed to 10, 30 and 50 nm uncoated AgNPs, but a higher LC50 for 80 nm AgNPs, and attributed the toxicity to dissolution. Yang et al. (2012) also reported no effect of particle size for toxicity in C. elegans, with no difference in toxicity between PVP-coated AgNPs of different sizes and different toxicities for similar sized PVP- citrate- and gum arabic-coated AgNPs. These and current study results indicate that dissolution may be the toxic mechanism, but its rate may be driven by the size and thus surface area of particles. The freshwater toxicity tests also indicated that C. dubia was more sensitive to each form of silver than P. subcapitata. Kennedy et al. (2010) and Bielmyer et al. (2002) reported similarly low LC50 values in the sub-lg L1 range for D. magna and C. dubia, respectively, with D. magna exhibiting greater sensitivity than P. subcapitata. The EC50 values determined for P. subcapitata and C. dubia exposure to AgNPs in the current study are lower than those reported for AgNPs in previous studies (Griffitt et al., 2008; Kennedy et al., 2010; McLaughlin and Bonzongo, 2012). The lower toxicity in other studies can be attributed to water quality parameters such as higher concentrations of chloride and dissolved organic matter (DOM) (Kennedy et al., 2010; McLaughlin and Bonzongo, 2012) decreasing the free silver concentration, and/or less dissolution of AgNPs (Griffitt et al., 2008). The study by McLaughlin and Bonzongo (2012) also suggested that the type of DOM as well as its concentration is important, as toxicity was lowered by between 2 and 3 orders of magnitude in P. subcapitata and C. dubia tested in freshwater marsh wetland water containing 65 mg L1 DOM compared to a synthetic soft water. The results from the current study have confirmed that complexation by chloride and humic acid can have a major effect on the stability and dissolution of AgNPs, as well as on their toxicity. Anions such as citrate, sulfide, and thiosulfate may also affect aggregation, dissolution and toxicity. Liu et al. (2010) showed that dissolved sulfide inhibited the release of Ag+ by the formation of insoluble Ag2S on the surface of the nanoparticles preventing oxidation of the Ag core. Dissolved sulfide may also bind to Ag+ released from the particles to form insoluble Ag2S. Therefore, the

effects of AgNPs in natural waters are likely to be mitigated by a number of anions and DOM decreasing the free silver available for uptake by organisms either by complexing dissolved silver or causing precipitation. 4. Conclusions Solubility, particle size, and the extent of aggregation are important controls on nanoparticle toxicity, however, characterisation at environmental concentrations remains an issue. Dynamic light scattering, disk centrifugation and surface plasmon resonance techniques can provide complementary information on particle size but have problems with low mg L1 nanoparticle concentrations. Solubility was usefully measured using a combination of dialysis separation and ultrafiltration and provided discrimination between truly dissolved, nano- and micron-sized particles. Silver nanoparticles were less toxic than ionic silver, but much more toxic than micron-sized silver. Toxicity was related to the release of dissolved silver, which was released more slowly from micron-sized silver due to the low surface area. The citrate-coated AgNPs were more toxic than the PVP-coated AgNPs to the freshwater organisms despite undergoing much greater aggregation, indicating the type of coating affected toxicity whilst aggregation to a larger size did not. The presence of humic acid or chloride mitigated toxicity. Greater percentage dissolution occurred for lower concentrations of AgNPs. The results indicated that significant dissolution may occur in natural waters, but the toxicity of the dissolved silver is likely to be mitigated by complexation of free silver by anions and dissolved organic matter or through precipitation reactions such as with sulfide or chloride. Acknowledgements The authors would like Darren Koppel, Steven Leahy, Josh King and Rob Jung from CSIRO for assistance with chemical and toxicological analyses, and Victoria Coleman from the National Measurement Institute of Australia for assistance with the differential centrifugal sedimentation measurements. This work was funded by a Grant from the New South Wales Environmental Trust, Australia. Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chemosphere. 2013.04.096. References Akaighe, N., MacCuspie, R.I., Navarro, D.A., Aga, D.S., Banerjee, S., Sohn, M., Sharma, V.K., 2011. Humic acid-induced silver nanoparticle formation under environmentally relevant conditions. Environ. Sci. Technol. 45, 3895–3901. Angel, B.M., Simpson, S.L., Jolley, D.F., 2010. Toxicity to Melita plumulosa from intermittent and continuous exposures to dissolved copper. Environ. Toxicol. Chem. 29, 2823–2830. Batley, G.E., Kirby, J.K., McLaughlin, M.J., 2013. Fate and risks of nanomaterials in aquatic and terrestrial environments. Acc. Chem. Res. 46, 854–862. Bielmyer, G.K., Bell, R.A., Klaine, S.J., 2002. Effects of ligand-bound silver on Ceriodaphnia dubia. Environ. Toxicol. Chem. 21, 2204–2208. Binet, M.T., Stauber, J.L., Adams, M.S., Rhodes, S., Wech, J., 2010. Toxicity of brominated volatile organics to freshwater biota. Environ. Toxicol. Chem. 29, 1984–1993. Coleman, V.A., Jamting, A.K., Catchpoole, H.J., Roy, M., Herrmann, J., 2011. Nanoparticles and metrology: a comparison of methods for the determination of particle size distributions. In: Postek, M.T., Coleman, V.A. (Eds.), Instrumentation, Metrology, and Standards for Nanomanufacturing, Optics, and Semiconductors, Proceedings of SPIE, vol. 8105, San Diego, CA, USA. Cornelis, G., Kirby, J.K., Beak, D., Chittleborough, D., McLaughlin, M.J., 2010. A method for determination of retention of silver and cerium oxide manufactured nanoparticles in soils. Environ. Chem. 7, 298–308.

B.M. Angel et al. / Chemosphere 93 (2013) 359–365 Croteau, M.N., Misra, S.K., Luoma, S.N., Valsami-Jones, E., 2011. Silver bioaccumulation dynamics in a freshwater invertebrate after aqueous and dietary exposures to nanosized and ionic Ag. Environ. Sci. Technol. 45, 6600– 6607. Dubas, S.T., Pimpan, V., 2008. Humic acid assisted synthesis of silver nanoparticles and its application to herbicide detection. Mater. Lett. 62, 2661–2663. El Badawy, A.M., Silva, R.G., Morris, B., Scheckel, K.G., Suidan, M.T., Tolaymat, T.M., 2011. Surface charge-dependent toxicity of silver nanoparticles. Environ. Sci. Technol. 45, 283–287. Fabrega, J., Fawcett, S.R., Renshaw, J.C., Lead, J.R., 2009. Silver nanoparticle impact on bacterial growth: effect of pH, concentration, and organic matter. Environ. Sci. Technol. 43, 7285–7290. Franklin, N.M., Rogers, N.J., Apte, S.C., Batley, G.E., Gadd, G.E., Casey, P.S., 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. Griffitt, R.J., Luo, J., Gao, J., Bonzongo, J.C., Barber, D.S., 2008. Effects of particle composition and species on toxicity of metallic nanomaterials in aquatic organisms. Environ. Toxicol. Chem. 27, 1972–1978. Huynh, K.A., Chen, K.L., 2011. Aggregation kinetics of citrate and polyvinylpyrrolidone coated silver nanoparticles in monovalent and divalent electrolyte solutions. Environ. Sci. Technol. 45, 5564–5571. Kennedy, A.J., Hull, M.S., Bednar, A.J., Goss, J.D., Gunter, J.C., Bouldin, J.L., Vikesland, P.J., Steevens, J.A., 2010. Fractionating nanosilver: importance for determining toxicity to aquatic test organisms. Environ. Sci. Technol. 44, 9571–9577. Lee, D.Y., Fortin, C., Campbell, P.G.C., 2005. Contrasting effects of chloride on the toxicity of silver to two green algae, Pseudokirchneriella subcapitata and Chlamydomonas reinhardtii. Aquat. Toxicol. 75, 127–135. Levy, J.L., Angel, B.M., Stauber, J.L., Poon, W.L., Simpson, S.L., Cheng, S.H., Jolley, D.F., 2008. Uptake and internalisation of copper by three marine microalgae: comparison of copper-sensitive and copper-tolerant species. Aquat. Toxicol. 89, 82–93.

365

Liu, J.Y., Hurt, R.H., 2010. Ion release kinetics and particle persistence in aqueous nano-silver colloids. Environ. Sci. Technol. 44, 2169–2175. Liu, J.Y., Sonshine, D.A., Shervani, S., Hurt, R.H., 2010. Controlled release of biologically active silver from nanosilver surfaces. ACS Nano 4, 6903– 6913. McLaughlin, J., Bonzongo, J.C.J., 2012. Effects of natural water chemistry on nanosilver behavior and toxicity to Ceriodaphnia dubia and Pseudokirchneriella subcapitata. Environ. Toxicol. Chem. 31, 168–175. Miao, A.J., Schwehr, K.A., Xu, C., Zhang, S.J., Luo, Z.P., Quigg, A., Santschi, P.H., 2009. The algal toxicity of silver engineered nanoparticles and detoxification by exopolymeric substances. Environ. Pollut. 157, 3034–3041. Navarro, E., Piccapietra, F., Wagner, B., Marconi, F., Kaegi, R., Odzak, N., Sigg, L., Behra, R., 2008. Toxicity of silver nanoparticles to Chlamydomonas reinhardtii. Environ. Sci. Technol. 42, 8959–8964. Rogers, N.J., Franklin, N.M., Apte, S.C., Batley, G.E., Angel, B.M., Lead, J.R., Baalousha, M., 2010. Physico-chemical behaviour and algal toxicity of nanoparticulate CeO2 in freshwater. Environ. Chem. 7, 50–60. Sotiriou, G.A., Pratsinis, S.E., 2010. Antibacterial activity of nanosilver ions and particles. Environ. Sci. Technol. 44, 5649–5654. USEPA,1994. Short-term methodology for estimating the chronic toxicity of effluents and receiving waters to freshwater organisms, third ed. United States Environmental Protection Agency Report 1994, EPA-600-4-91-002. Cincinnati, OH, USA. Wang, G.S., Liao, C.H., Wu, F.J., 2001. Photodegradation of humic acids in the presence of hydrogen peroxide. Chemosphere 42, 379–387. Yang, X.Y., Gondikas, A.P., Marinakos, S.M., Auffan, M., Liu, J., Hsu-Kim, H., Meyer, J.N., 2012. Mechanism of silver nanoparticle toxicity is dependent on dissolved silver and surface coating in Caenorhabditis elegans. Environ. Sci. Technol. 46, 1119–1127. Zhao, C.M., Wang, W.X., 2012. Importance of surface coatings and soluble silver in silver nanoparticles toxicity to Daphnia magna. Nanotoxicology 6, 361–370.