**3. Results and Discussion**

### *3.1. Hydrochemical Conditions to Investigate ENMs Dissolution in Natural Aquatic Media*

Studies on solubility and dissolution kinetics are usually performed using deionized water or simple background electrolyte solutions (e.g., NaCl, NaNO3 or Ca(NO3)2) for specific pH values [15,16]. However, the lack of complexity in the exposure media is not suited to mimic ENMs dissolution in natural waters, which depends on an interplay between intrinsic properties and extrinsic environmental parameters, unique to each environment. Few variations are observed within one river system, such as the Danube and Rhine rivers, and to a larger extent for the Elbe River. In contrast, the FOREGS database, issued from the Geochemical Atlas of Europe [48], presents a wide range of values (Figure 2) covering the diversity of various catchments on the continent. The FOREGS database is, to date, the most relevant database to define media composition, mimicking a large pool of natural surface waters, from alkaline rivers such as Danube and Rhine (Figure 2) to more acidic waters mostly present in base-poor buffering capacity regions and/or organic-rich acid buffering regions. However, it does not provide data on all chemical components, such as orthophosphates. For this study, orthophosphate concentrations that are representative of a pool of rivers were obtained from the European Environment Agency (EEA) database "Waterbase-Water Quality" [49].

As recommended by OECD GD318 for the testing of nanomaterials dispersion stability [32], exposure media harboring low concentrations of the major ions reported in surfaces waters, Ca2+ and Mg2+, would allow a better simulation of the ionic background of surface waters [48]. In addition to standard tests performed using 10 mM Na(NO3), we thus propose to use 2 mM of Ca(NO3)2 and 0.5 mM of MgSO4 to mimic a more realistic scenario of natural waters. Such values are based on Ca2+ and Mg2+ concentrations reported by the FOREGS database, with average concentration of 1.4 mM and 0.5 mM for Ca2+ and Mg2+, respectively [48]. To simulate alkaline surface waters (pH 7–8), 2 to 5 mM HCO3 − should be added to the exposure media.

**Figure 2.** Boxplots illustrating the range values of (**a**) pH, (**b**) conductivity, (**c**) calcium concentration, (**d**) DOC, (**e**) orthophosphate and (**f**) magnesium concentrations in specific and a pool of surface waters. Data from 2015 to 2017 were obtained from The International Commission for the Protection of the Danube River (ICPDR) [45], the River Basin Communities of Rhine [46] and the Elbe Data Information System (FIS) of the River Basin Community [47] for the Danube, Rhine, and Elbe rivers, respectively. Data from the Rhine River were obtained at the locations Bad Honnef and Karlsruhe. The FOREGS (Forum of European Geological Surveys) Geochemical database [48] and the European Environment Agency (EEA) database were used to represent European stream waters. \* Data obtained from the EEA database "Waterbase-Water Quality" [49].

For all databases, phosphate concentrations are significant, with values ranging between 0.001 to 0.190 mg.L−<sup>1</sup> (Figure 2). Regarding the expected concentrations of ENMs into aquatic environments, for example a maximum of 150 ng.L−<sup>1</sup> for ZnO NPs [5], phosphate can thus play an important role by initiating transformation or formation of lower soluble metal-phosphate coatings. Similarly, NOM in surface waters (Figure 2) might play an important role, enhancing or inhibiting dissolution of ENMs [21,22,50]. Thus, for a more realistic scenario, test medium should also include NOM and inorganic phosphate. We should, however, point out the importance of separating two pivotal processes in environmental aquatic media: dissolution and chemical transformation. Testing using a complex aquatic chemistry (i.e., sulfide, Cl<sup>−</sup>, PO4<sup>3</sup><sup>−</sup>, HCO3<sup>−</sup>, DOC) might trigger ENM transformations into less soluble phases [51–53] and affect the direct evaluation of the solubility and dissolution rate. Determination of the dissolution rates in a medium that represents natural conditions but does not induce other reactions is then necessary. In this regard, precipitation of secondary species must be preliminarily predicted using thermodynamic modeling.

### *3.2. Dissolution of Ag NPs in Simple Background Electrolyte: Validation of the Continuous Flow System for Low Ag NPs Loadings*

For all experiments performed at pH 5 with 10 mM NaNO3 as the eluent, higher dissolved Ag concentrations were measured at the beginning of the exposure experiments (Figure 3). During the first minutes of the exposure, the dissolution behavior of the Ag NPs might be governed by the positioning of the particles in the filter holder. Since the particles are not immediately fixed onto the membrane, they may remain dispersed in the filter holder's dead volume above the membrane, increasing the contact time with the eluent and consequently the Ag concentration. Similarly, the increase in Ag concen-

trations after 60 min exposure for the experiment Ag\_0.2mL/min\_2.2μg\_NaNO3 and Ag\_0.2mL/min\_8.2μg\_NaNO3 correspond to an increase in contact time with the eluant due to the decrease in the flow rate from 1 mL.min−<sup>1</sup> to 0.2 mL.min−1. The presence of dissolved Ag+ in the injected Ag suspension and sorbed Ag+ at the surface of the NPs could also explain higher Ag concentrations at the beginning of the experiments.

**Figure 3.** Ag concentrations measured at the outlet of the flow-through setup for (**a**) 500 μL of a 4.3 mg.L−<sup>1</sup> Ag NPs suspension injected with a flow rate of 1 mL.min−<sup>1</sup> for 1 h and set to 0.2 mL.min−<sup>1</sup> for the rest of the experiment, Ag\_0.2mL/min\_2.2μg\_NaNO3 experiment (**b**) for 500 μL of a 16.4 mg.L−<sup>1</sup> Ag NPs suspension injected with a flow rate of 1 mL.min−<sup>1</sup> for 1 h and set to 0.2 mL.min−<sup>1</sup> for the rest of the experiment, Ag\_0.2mL/min\_8.2μg\_NaNO3 experiment and (**c**) for 500 μL of a 16.4 mg.L−<sup>1</sup> Ag NPs suspension injected with a flow rate set to 0.5 mL.min−<sup>1</sup> during all the time of the experiment, Ag\_0.5mL/min\_8.2μg\_NaNO3 experiment. (**d**) Conceptual illustration of the behavior of Ag NPs in the filter holder during the flow-through experiment.

Once the particles are deposited onto the membrane and are exposed to a continuous and unchanged flow rate, outflow Ag concentration becomes stable (Figure 3). The apparent dissolution rates calculated from the Ag concentrations (Figure 4) reached a steady state after 225 min for Ag\_0.2mL/min\_2.2μg\_NaNO3, 400 min for Ag\_0.2mL/min\_8.2μg\_NaNO3, and 345 min for Ag\_0.5mL/min\_8.2μg\_NaNO3. The average Ag concentrations and averaged dissolution rates were calculated on the steady state range. Experiments performed with 8.2 μg of Ag NPs loaded onto the membrane at two different flow rates show similar averaged dissolution rates (k = 0.10 ± 0.002 μg.m<sup>−</sup>2.s−<sup>1</sup> and k = 0.093 ± 0.009 μg.m<sup>−</sup>2.s−<sup>1</sup> for Ag\_0.2mL/min\_8.2μg\_NaNO3, and Ag\_0.5mL/min\_8.2μg\_NaNO3, respectively). Nevertheless, for the experiment performed at a flow rate of 0.2 mL.min−1, the averaged Ag concentration at the outlet was 2.7 times higher, due to a longer contact time of the eluent with Ag NPs (Figures 3 and 4d). Theoretically, lower flow rates would allow more contact time between the eluent and the surface of the NPs, leading to higher Ag concentrations in the reaction zone and result in a larger diffusion boundary layer, both limiting

the dissolution process. The study of Keller et al. [54] illustrates well the effect of the flow-rate on the dissolution kinetics of nanomaterials such as BaSO4 NPs, by performing long term experiments with higher NPs loadings and flow rate ramping between 0.1 and 3.0 mL.h−1. Dissolution rates are not much different between the experiments Ag\_0.2mL/min\_8.2μg\_NaNO3, and Ag\_0.5mL/min\_8.2μg\_NaNO3. However, it is likely that at higher flow rates a more pronounced difference will be observed. However, for highly soluble ENMs such as CuO and ZnO NPs, the influence of the flow rate on dissolution might be reduced [54].

**Figure 4.** Dissolution rates of 80 nm Ag NPs in 10 mM NaNO3 at pH 5, for different particles loadings and flow rates. (**a**) Dissolution rates obtained for the Ag\_0.2mL/min\_2.2μg\_NaNO3 experiment corresponding to 2.2 μg Ag NPs exposed at a flow rate of 1 mL.min−<sup>1</sup> for 1 h and 0.2 mL.min−<sup>1</sup> for the rest of the experiment. (**b**) Dissolution rates obtained for Ag\_0.2mL/min\_8.2μg\_NaNO3 experiment corresponding to 8.2 μg Ag NPs exposed at a flow rate of 1 mL.min−<sup>1</sup> for 1 h and 0.2 mL.min−<sup>1</sup> for the rest of the experiment. (**c**) Dissolution rates obtained for Ag\_0.5 mL/min\_8.2μg\_NaNO3 experiment corresponding to 8.2 μg Ag NPs exposed at a flow rate of 0.5 mL.min−1. (**d**) Table showing the average [Ag] concentrations and the averaged dissolution rates calculated on the steady state ranges.

Ag NPs loading appear to influence significantly the dissolution rate of Ag NPs when applying the same 0.2 mL.min−<sup>1</sup> flow (experiments Ag\_0.2mL/min\_2.2μg\_NaNO3 and Ag\_0.2mL/min\_8.2μg\_NaNO3, Figure 4d). The dissolution rate was more than two times higher for the experiment performed with 2.2 μg of Ag NPs (Ag\_0.2mL/min\_2.2μg\_NaNO3, k = 0.28 ± 0.07 μg.m<sup>−</sup>2.s−1) compared with the experiment performed with 8.2 μg of Ag NPs (Ag\_0.2mL/min\_8.2μg\_NaNO3, k = 0.10 ± 0.05 μg.m<sup>−</sup>2.s−1). A high particle loading increases the dissolved ion concentration in the vicinity of the particle surfaces, limiting their dissolution. To avoid local saturation around the nanoparticles, a higher flow rate might be used. Nevertheless, the dissolution rates determined at a flow rate of 0.5 and

0.2 mL.min−<sup>1</sup> are similar (k = 0.09 ± 0.03 compared with k = 0.10 ± 0.05 μg.m<sup>−</sup>2.s−1, respectively, both at 8.2 μg loading and with NaNO3). This indicates that the dissolution rate of Ag NPs depends on the initial Ag NPs concentration. Concentration-dependent dissolution of Ag NPs was already reported by several studies, showing higher dissolution rates for lower Ag NPs concentrations [55,56]. Keller et al. [54] also highlighted the influence of the initial NPs loading on the dissolution rate for BaSO4, CuO, ZnO and TiO2 NPs. Such findings emphasize the importance to investigate ENMs dissolution at environmentally relevant ENMs concentrations. Indeed, performing flow through dissolution experiments using high initial NPs loading may result in wrong assessments of the dissolution rate of an ENM in natural aquatic systems. We may also hypothesize that a larger amount of Ag NPs injected does not result in the deposition of Ag NPs as a monolayer.

The disparity between the dissolution rates calculated for each experiment performed using the same Ag NPs and the same exposure medium demonstrated that the dissolution kinetic of Ag NPs is dependent on the initial NPs concentration. The flow rate is also an important parameter to adjust for the success of the test and an appropriate representation of the system to mimic. A low flow rate allows for longer interaction between the medium and the ENMs, which would lead to a more reliable measurement of the dissolved fraction at the outlet. Suitable for ENMs with low solubility, it may also result in local saturation and underestimation of the dissolution rate. Experimental parameters (i.e., flow rate, exposure time and ENMs loading) need to be adjusted to the environmental conditions we want to mimic and to the solubility property of the studied ENM, keeping the feasibility of the test. Based on the results obtained through these tests, we recommend performing continuous flow dissolution experiments at a flow rate between 0.5 and 1 mL.min−1, sufficient to reach a steady-state after 5–6 h of exposure, and allowing robust determination of the outlet dissolved concentrations while reducing potential saturation effects.

### *3.3. Dissolution Rate of Ag NPs in Artificial Surface Waters*

Continuous flow experiments were also performed using artificial waters intended to mimic acidic streams (2 mM Ca(NO3)2 and 0.5 mM MgSO4 solution at pH 5) and nearneutral/alkaline surface waters (2 mM Ca(NO3)2, 0.5 mM MgSO4 and 5mM NaHCO3 − solution at pH 7.5). In both experiments, the steady-state plateau was reached after 400 min (Figure 5a,b). The averaged dissolution rate of Ag NPs calculated for the near-neutral artificial water exposure experiment was relatively low (Ag\_0.5mL/min\_8.2μg\_NSW, k = 0.08 μg.m<sup>−</sup>2.s−1). For this latter experiment, the averaged concentration of Ag measured in the eluant ([Ag] = 0.56 ± 0.01 μg.L−1) was higher than the LOD (0.05 μg.L−1) and LOQ (0.12 μg.L−1) values determined for the ICP-MS. However, for the lower Ag NP dissolution rate, lower Ag concentrations at the outlet might impact the relevance and significance of the test, for example, if dissolved Ag concentrations are too low to be accurately measured by ICP-MS (Figure 5c). In such case, lower dilution factors and/or a lower flow rate might thus be suitable.

At a lower pH, the averaged Ag dissolution rate was higher (Ag\_0.5mL/min\_8.2μg\_ASW, k = 0.13 μg.m<sup>−</sup>2.s−1) than at pH 7.5 (Ag\_0.5mL/min\_8.2μg\_NSW, k = 0.08 μg.m<sup>−</sup>2.s−1). This trend is consistent with previous work showing an increase in Ag NPs dissolution in acidic media [23]. Using the classic batch experiment, Mitrano et al. [57] investigated the dissolution rate of 100 nm and 60 nm spherical citrate-coated Ag NPs in artificial and natural media. They reported dissolution rate values as logr = −11.72 and −12.23 mol.cm−2.s−<sup>1</sup> in deionized water (pH = 6.7), logr = −12.14 and −12.71 mol.cm−2.s−<sup>1</sup> in a natural surface water (pH = 7.3) and logr = −12.46 mol.cm−2.s−<sup>1</sup> in deionized water containing 2 mg.L−<sup>1</sup> of DOC (pH = 4.8). Such values are close to the values obtained in this study, with log *k* ranging between −12.58 and −13.13 mol.cm−2.s−<sup>1</sup> for Ag\_0.2mL/min\_2.2μg\_NaNO3 and Ag\_0.5mL/min\_8.2μg\_NSW, respectively.

**Figure 5.** (**<sup>a</sup>**,**b**) Dissolution rates of 80 nm Ag NPs in environmental aqueous media. (**c**) Table showing the average [Ag] concentrations and the averaged dissolution rates calculated on the steady state ranges. Ag\_0.5mL/min\_8.2μg\_ASW was performed at pH 5 with 2 mM Ca(NO3)2 and 0.5 mM MgSO4 as exposure medium. Ag\_0.5mL/min\_8.2μg\_NSW was performed at pH 7.5 with 2 mM Ca(NO3)2, 0.5 mM MgSO4 and 5 mM of NaHCO3− as exposure medium. Error bars correspond to the incertitude of Ag concentrations measured by ICP-MS.
