DRAFT OECD TEST GUIDELINE FOR THE TESTING OF CHEMICALS

Transcription

1 DRAFT OECD TEST GUIDELINE FOR THE TESTING OF CHEMICALS Agglomeration Behaviour of Nanomaterials in Different Aquatic Media 15 1

2 INTRODUCTION 1. The enhanced production and wide usage of manufactured nanomaterials suggest the increased probability of finding them in natural systems, especially aqueous ones. This fact has risen up a discussion, whether the existing regulatory protocols are sufficient and adequate to assess the fate of manufactured nanomaterials and their impact on the systems of interest. Agglomeration behaviour was identified as important parameter affecting the environmental behaviour of nanomaterials. This parameter depends on the physicochemical characteristics of the nanomaterial itself, the physicochemical characteristics of the suspension medium, suspension preparation, concentration of the nanomaterial and concentration of other substances and particles in the suspension. The agglomeration behaviour is also highly dynamic in many cases, because it is controlled by kinetics (energy barriers) rather than thermodynamic equilibrium. Therefore information on nanomaterials agglomeration behaviour is, beside e.g. dissolution rate, one prerequisite for a proper further testing of nanomaterials. 2. This test guideline will provide reliable and reproducible results on the agglomeration behaviour of different types of nanomaterials in aquatic media, which will support decisionmaking on further ecotoxicology and environmental fate testing. To represent natural environmental conditions the test guideline prescribes testing under different conditions with varying parameters like ph, ionic strength, and presence of dissolved organic matter (DOM). The test guideline will be applied by experimenters conducting tests and by regulators evaluating tests. The data is important for hazard and risk assessment activities. 3. This test guideline will provide a simple and effective tool that will allow analysing the agglomeration behaviour in aqueous media under conditions representative of natural surface water chemistry. Such characterization shall be performed with the purpose of investigating the fate of nanomaterials in natural waters, potential behaviour in test media and their impact on the environment based on the criteria listed above. DEFINITIONS AND UNITS 4. Definitions and Units are set out in Annex 1. INITIAL CONSIDERATIONS 5. When a nanomaterial is insoluble or sparingly soluble in water and considered within the aqueous phase it s appropriate to consider it as a particle dispersion. Dispersions of solid particles are inherently unstable and the particles eventually agglomerate. An energy barrier, resulting from the interplay of various particle-particle interaction forces can prevent the attachment of particles to each other and therefore prevent the process of agglomeration. These force are among others controlled by the particle surface chemistry and the hydrochemistry. Among the various forces existing between particles, there are three major forces of different nature acting within the dispersion, controlling the particle interactions. These are the forces of 2

3 electrostatic interactions (attractive or repulsive), steric forces emerging from surface bound macromolecules (repulsive or attractive), and forces of Van der Waals attraction. In special cases magnetic forces might play a role in-between particles which reveal magnetic properties. Due to the forces acting in-between the colloidal particles they can form (meta-) stable dispersions or undergo the processes of agglomeration. Due to always attractive Van der Waals forces, lyophobic dispersion colloids, in which category most of the currently discussed nanomaterials fall, agglomerate as long as no stabilizing, repulsing forces between particles are present. A stable dispersion therefore always represents a non-equilibrium situation which is kinetically hindered to reach equilibrium in the agglomerated state. Formed agglomerates (working definition from ISO TS ) are described as collection of weakly bound particles where the resulting surface area is similar to the sum of surface areas of initial individual particles. Thus, structures formed by attachment of particles to each other from a formerly dispersed stage would likely be termed agglomerates. Formation of agglomerates is one of the main factors influencing the dispersion stability, since particle agglomeration increases particle size, which in turn enhances their sedimentation. Since the main purpose of this guideline is to monitor the fate of nanomaterials under environmentally relevant conditions, the choice of the media for the preparation of the dispersion is the task of paramount importance, as it shall represent natural surface waters. The choice of parameters and parameter ranges must be based on the effectiveness of the parameter to influence colloidal stability and the likelihood that the parameter has a magnitude in natural waters which results in an important impact on colloidal stability. For example, Sodium as a monovalent cation typically influences the colloidal stability of dispersions only at a concentration not often encountered in natural freshwaters. At the same time it is likely that Calcium, which has an order of magnitude higher effect on colloidal stability because of its divalent character, is found often at concentrations where it influences colloidal stability. Thus the key parameters of media, such as ph, concentration of Calcium ions and of NOM are chosen as most relevant parameters for dispersion stability under natural conditions and are applied at concentrations which represent 90-95% of natural surface waters (according to FOREGS Database (1)). Ca(NO 3 ) 2 electrolyte is chosen because of both the dominant effect of multivalent cations (Ca 2+ ) on particle aggregation and their presence in relevant concentrations (compared to less effective monovalent ions and less abundant trivalent ions). At the same time the system shall be possibly simple to be conducted in standard laboratory conditions. A concept that allows doing so was proposed by F. von der Kammer (2) and further developed and checked for environmental relevance by Ottofuelling et al. (3). It is based on the construction a multidimensional test matrix including environmentally relevant test parameters predominantly influencing nanoparticle agglomeration. These parameters include presence of various cations and anions, organic matter and ph. Such approach is utilized in the present guideline, in a more universal and less complex procedure. The parameters that influence particle agglomeration behaviour and dispersion stability are described below: 6. Ionic strength Presence of electrolyte in the aquatic media is essential for the formation of the Electric Double Layer (EDL), the ionic structure in the medium that appears close to the 3

4 particle surface as response to the surface charge of particles. When particles with the opposite surface charge are considered, low electrolyte concentration leads to faster particles agglomeration and sedimentation. In this case increased thickness of EDL enables particles to approach each other and interact faster through forces of electrostatic attraction. Oppositely, when particles with the same polarity of surface charge are under investigation, high electrolyte concentration promotes their agglomeration and sedimentation. Compact, thin EDL (at high electrolyte concentration) allows particles to approach closer to each other and interact more effectively based on short-range Van der Waals attractive forces and diminished forces of electrostatic repulsion. 7. ph ph has a crucial effect on particle agglomeration by influencing the surface charge of particles carrying variable charges at the surface. Dispersions of particles are most unstable at the ph of Point of Zero Charge (PZC). At this ph the net charge at the particle surface that forms the EDL is minimal. As a result, the inter-surface electrostatic repulsion forces are minimal, that allows particles to agglomerate and sediment. 8. Concentration of Dissolved Organic Matter (DOM) DOM influences the stability of particle dispersions through adsorption on the particle surfaces. A positive surface charge can be reversed to negative charge due to the adsorption of DOM which has in general an aionic character, uncharged or negatively charged surfaces can experience an (increasing) negative potential, thereby providing enhanced stability. Particle dispersions generally become more stable due to such interactions. However, when the adsorbed DOM reduces or only balances a positive charge on the particles surfaces, DOM driven agglomeration could be observed. DOM molecules in the liquid phase also act as a buffering agent, keeping the ph of particle dispersion at a constant level. 9. Particle concentration For dispersions of particles, particle concentration is typically given in mass concentration of particles per volume of dispersion liquid. However, the number concentration of particles is key for the number of particle collisions in a dispersion per unit time and therefore for the rate of the agglomeration process and hence for the dispersion stability. At the same ratio of successful to unsuccessful collisions (attachment efficiency α) larger numbers of particles in the dispersion increase the probability of inter-particle collisions. Larger number of collisions per time lead to faster agglomeration and subsequently the sedimentation of particles. Particle number concentrations are related to mass concentrations via particle density and size (and geometry of the particles). 10. Particle nature and surface chemistry The surfaces of various particles are terminated with surface functional groups which determine the surface chemistry of the particles in the dispersion. The amount and type of surface groups determine the extent in which particles are able to develop surface charge and build EDL. These two features play crucial role in particle dispersion stability as described above, were permanent charge and ph-dependent variable charge are the two major factors to consider Additionally surface-bound molecules may 4

5 introduce a non-electrostatic stability factor, the steric stabilization, where the molecules prevent attachment of the particles by creating a steric barrier. If sterically stabilizing molecules are only weakly bound to the surface, dilution of the particles in dispersion can lead to desorption of the molecules, which can further lead to de-stabilization of the dispersion through particle agglomeration. 11. Particle size Large particles sediment faster. Thereby, also agglomeration directly leads to faster particle settling. Diffusion counters sedimentation, but much less for larger particles. 12. Particle density Particles with high density (e.g. Gold-Nanoparticles) show considerable sedimentation rates even in the non-aggregated form (e.g. 100 nm Gold-NPs settle by 4.3 mm/d at 20 C). 13. Presence/absence of CO 2 in the atmosphere Presence of CO 2 in the atmosphere above the particle dispersion can influence the stability of dispersion through different processes. One is the absorption of CO 2 by the water phase at high ph values (above ph=5.6) thereby decreasing the ph of dispersio Dissolved carbonate ions likely take part in adsorption at the mineral surface in a form of counter ion, changing the surface charge, the point of net zero proton charge (not to be confused with the isoelectric point measured by zeta potential measurements) and thereby the properties of EDL, consequently influencing the dispersion stability. PRINCIPLE OF THE TEST 14. Performed tests are designed in the form of multi-dimensional matrix testing. Experimental matrix includes environmentally relevant test parameters predominantly influencing nanoparticle agglomeration, such as ph, type and concentration of electrolyte and organic matter, absence of CO 2 in the atmosphere (for materials prone to interaction with carbonate absence and presence of CO 2 in the atmosphere above the dispersion). To check the dispersion stability in the prepared samples, the agglomeration and resulting sedimentation of the particles is probed by determining the remaining particle concentration in the top cm volume of dispersion every hour (in triplicates) during a time-period of 6 hours (0-6 hour measurements). Before such determination at hour 6, samples undergo centrifugation imitating enhanced particle sedimentation. For a fully stable dispersion, the concentration of particles in the top cm volume of the dispersion is not supposed to change over time, while agglomeration and consequently settling shall gradually reduce the concentration of particles in this volume of the sample. The losses of material during experimental procedures (section 20) (%) should be found through the calculation of the difference between the expected concentration value based on the mass of the dispersed material and the concentration of material detected during 0-hour measurement. Thereby losses during the preparation of the dispersion and the hydrochemical conditions in the test vial are neglected. However, the difference between expected concentration (according to the weighted amount) and the start of the agglomeration 5

6 test should be recorded. The experimental design as well as typical experimental results for TiO 2 (NM 105) and Ag (NM300K) nanomaterials, presented for comparison, can be found in annex 2. INFORMATION ON THE TEST CHEMICAL 15. For a sound test performance and interpretation of test results information on nanomaterials composition, density, particle size, specific surface area and preferable also solubility is needed before conduction the test and should be reported together with the results. Density and particle size is needed to calculate the required sample mass in the test to obtain the desired particle number concentration. PROPOSED REFERENCE SUBSTANCES 16. Three nanomaterials were used to underpin the applicability of the proposed experimental routines as being positive and negative controls plus a borderline case. These nanomaterials are Ag (NM300K) (4), Carbon nanotubes CNTs (NM400) (5) and TiO 2 (NM105) (6), nanoparticles. Dispersions of Ag (NM300K) nanoparticles were completely stable over time, while dispersions of CNTs were not stable through entire test conditions. These materials were used as positive and negative controls. 17. Dispersions of TiO 2 (NM105) nanoparticles showed intermediate stability dependent on test conditions. Therefore this material was used as an appropriate nanomaterial for quality control in further testing of nanomaterials and inter-laboratory comparison. VALIDITY OF THE TEST 18. Present guidelines are applicable to the dispersions with the approximate particle concentration of particles/l (±1/2 order of magnitude). The number of particles in the dispersion can be calculated according to the formula: N = M P V where N is the number of particles in the dispersion, M is the mass of the powder measured to prepare the dispersion, P (kg/m 3 ) is the density of the particle material (known value), V (m 3 ) is the volume of particles calculated either from DLS size-measurements (use the formula for the volume of sphere and the first mode of the intensity weighted distribution, not the Z-average) or by other means (microscopy) (use specific formulas for volume calculations). A Microsoft Excel test preparation tool is available on OECD website for calculation of particle number concentration from mass concentrations based on the model of a homogeneous solid sphere with known density and particle diameter. 19. This guideline is applicable for particles having a density not smaller than 1 kg/m 3. Density of particles is the significant factor influencing their sedimentation rate, especially at the 6

7 stage of sample s centrifugation. Thus for the nanomaterials with the density smaller than 1 kg/m 3 (dendrimers, polymers, etc.) alternative methods other than based on sedimentation have to be applied, since these materials will preferably locate at the air/water interface in the test vessels. Such methods are however out of the focus of the current TG. 20. Performed tests can be considered valid when the losses of investigated nanomaterial during preparation of the dispersion in the test vessels do not exceed 50% of initially dispersed material. Such losses should be calculated based on the difference between the initial material concentration measured at 0 h and expected concentration of material in the sample, based on the amount of material weighed and dispersed during the preparation step (section 25). Large experimental losses can occur due to: (a) Incomplete dispersion of the analysed nanomaterial (b) Dissolution of nanomaterial during experiments (c) Fast sedimentation of large particles (d) Sample heterogeneity (e) Precipitation of the particles on the walls of experimental tubes 21. In case the material losses during experimental manipulations exceed 50% of initially dispersed material, the reasons of such behaviour of the samples have to be investigated by particle size analysis, analysis of material dissolution in the applied conditions or analysis of material precipitation on the walls of experimental vessels. The use of a different vessel material should be considered to minimize those losses. DESCRIPTION OF THE METHOD Apparatus and chemical reagents 22. Standard laboratory equipment, including but not limited to: (a) Suitable calibrated pipets for sample preparation (5 ml, 2 ml, ml volume) (b) Ultrasonic probe for homogenization of particle dispersion (c) ph-meter to measure the ph of the dispersion (d) DLS devise for measuring the particle size (e) 50 ml conical bottom polypropylene tubes (N) to perform the agglomeration experiments (f) 10 ml polypropylene tubes (N) to prepare the samples for ICP-OES/MS analysis (g) Standard lab centrifuge, capable of 4000 relative centrifugal force (rcf) (h) Inductively Coupled Plasma Mass Spectrometry Device or Inductively Coupled Plasma Optical Emission Spectrometry device for analysis of nanoparticle dispersions 23. Materials: (a) Water (H 2 O) ultrapure de-ionized water (18 Ohms resistivity) 7

8 (b) Sodium Hydroxide solution (NaOH) 0.1M solution in Ultrapure DI water, for ph establishing (c) Hydrochloric acid (HCl) 0.1M solution in Ultrapure DI water, for ph establishing (d) NOM solution- solution of Natural Organic Matter (1g/L of DOM) (e) Calcium Nitrate (Ca(NO 3 ) 2 ) 0.1M solution to establish the needed concentration of electrolyte (f) Analysed material in the form of dry powder or in a form of dispersion General conditions 24. All described experimental conditions are chosen to serve two main purposes. They should resemble natural conditions which are controlling dispersion stability and be easy to reproduce in the laboratory environment. All experiments should be performed in triplicates at conditions close to standard ( K (25 C, 77 F), kpa (14.7 psi, 1.00 atm, and bar)). For preparation of dispersions, it is recommended to use 50 ml plastic tubes (polypropylene, 50 ml conical bottom centrifuge tubes, 3 cm in diameter, 11.5 cm tall). In case of observed attachment of particles to vessel walls alternative vessel materials should be tested for better suitability. Preparation of nanoparticle stock dispersion 25. In case the tested material is provided in the form of dry powder, it should be pre-wetted in deionized water and left in the form of wet-paste for 24 h to insure the proper interaction of material surface with water. After 24 h of pre-wetting the resulted wet paste is dispersed into known volume of MQ water, thus providing a stock solution with determined material concentration. The concentration of stock solution shall be sufficient to allow the further dilution to the required nanomaterial number concentration (0.5 to 5 x particles/l) within the analysed samples. At the same time the concentration of the stock solution should not be too high to avoid facilitation of nanomaterial agglomeration reaction in the stock before dilution in the test vessels. Thus the recommended concentration of nanomaterial within stock solution shall not exceed the concentration of nanomaterial within analysed samples more than 20 times. Resulting dispersion of stock solution is sonicated for 10 minutes at the power level of 40 W to ensure proper dispersion and homogenization. Description of calibration process of sonication probe can be found in annex 2. Usage of sonication probe is recommended compared to the usage of cup-horn sonicator and bath sonicator. Sonication probe is more effective in dispersing the material since it interacts directly with the dispersion, while other mentioned devices interact with the dispersion through the walls of the experimental vessels. In addition usage of sonication probe allows calibrating the energy input to the dispersion. A Microsoft Excel test preparation tool supporting the calculation of delivered sonication energy is available on the OECD website. In case the investigated nanoparticles are delivered in a form of stable stock dispersion, the stock dispersion is directly diluted to the concentration, convenient for further sample preparation. No additional sonication is required in this case. 8

9 Agglomeration behaviour in presence of electrolyte in the dispersions 26. To study the agglomeration behaviour of particles in dispersion in presence of electrolyte, the usage of Ca(NO 3 ) 2 electrolyte is recommended for particles having a negative charge under the tested ph or test dispersion contains NOM. In case that the tested nanoparticles are inherently positively charged in the tested ph range, it is suggested to use a salt with a divalent anion, as Na 2 SO 4 or CaSO 4. The possibility of specific interaction of the Sulphate ion with the particle surface and risk of bridging flocculation needs to be considered then. The aliquots of sonicated stock dispersion (section 25) are taken into 50 ml polypropylene tubes and diluted with deionized water to the volume of 20 ml. Dilution of dispersion of tested material is made to minimize the contact of particles in dispersion with concentrated solution of electrolyte, added in the next step. Stock solution of electrolyte is added to the previously diluted dispersion of tested material to achieve the final concentrations of 0, 1, and 10 mm respectively in the final volume of the sample. The samples are filled with MQ water to achieve the final volume (40 ml) thus providing the nanomaterial number concentration of (0.5 to 5 x particles/l) within the sample. The aliquots of samples are taken and diluted for the further instrumental analysis (sections 32-35) Agglomeration behaviour in presence of DOM in the dispersions Preparation of NOM stock solution 27. Prior to the observation of particle agglomeration behaviour in presence of DOM, stock solution of corresponding NOM shall be prepared and characterized. Any NOM can be utilised for these purposes if it reveals the levels of purity regarding present electrolyte ions. The recommended concentrations of electrolyte ions in the NOM material should be comparable to such concentrations as of 2R101N Suwannee River NOM (SRNOM) (7) (Tables 5 and 6, annex 2). The purification of NOM material can be performed through the dialysis or purification with ion-exchange columns. If it is not possible to perform such purification, SRNOM is recommended for preparation of NOM solution as standardized and purified material. The stock solution of NOM is prepared by addition of required amount of NOM powder to the de-ionized water. After adjustment of ph to ph=8.5-9, NOM solution is left under vigorous stirring for 24 hours. After that the ph is measured and adjusted to the same values (ph=8.5-9) if needed. The solution is filtrated using a vacuum filtration set-up, consisting of vacuum pump and bottle-top 0.2 µm PES filter. Thereafter the solution of NOM is analysed for the DOC content. Resulted solution is kept at 4 C in brown-glass bottles avoiding exposure to the light. Due to the aging of DOM followed by precipitation, it is recommended to store the NOM stock solution no longer than 4 weeks. Addition of NOM stock solution to the sample 28. Addition of NOM solution to the samples can serve two purposes. Primary purpose is achievement of greater dispersion stability through the adsorption of NOM molecules at the particle surfaces, what simulates the interaction of the nanomaterial with DOM present in natural 9

10 waters. This stabilization originates predominantly from electrostatic effects (but steric stabilization might play an additional role) of the deprotonated, negatively charged, carboxylic groups in the DOM through the increase of the negative charge on the particles surfaces (stability from -/- repulsion). In the case of positively charged particle surfaces (stability from +/+ repulsion) the addition of DOM first reduces the positive charge on the particles, thereby reducing particle stability. If a too small amount of DOM is added the charge reversal (+/+ -/- ) is not reached or not sufficient to re-establish stability and particles will show dispersion instability from the DOM addition. This can be avoided using a minimum concentration, which is (simplified) depending on the specific surface area of the particles. Secondary purpose of DOM addition is the ph buffering of the dispersion at elevated ph against the effect of absorbing CO 2 without any further buffering additives which influence stability of the dispersion. 29. To check the dispersion stability of the particles in presence of DOM one shall strictly follow the procedure described in section 26. Prior to bringing the samples volumes to 40 ml, necessary amount of NOM stock solution (section 27) is added to the samples prepared as described in section 26 to achieve the desired concentration of DOC. In general, the concentration of DOC shall be sufficient for the DOM molecules uniformly adsorb (this TG assumes a minimum of 1 DOM molecule per nm 2 ) on the entire surface area of analysed material. While a concentration of 10 mg/l DOC was found sufficient for the TiO 2 (NM105) material and is recommended as a standard condition in the described testing procedure, the concentration per surface area is a more robust condition, especially when materials with high specific surface area are tested. In general, the minimal concentration of needed DOC shall be determined by the surface area of nanomaterial in the sample (SA) that is calculated as follows: SA=C s V s SSA where C s is the concentration of the sample (g/l), V s is the volume of the sample (L) and SSA is the BET specific surface area of the nanomaterial (m 2 /g). The recommended minimal concentration of DOC is 0.004g per SA=1 m 2 of the material in the sample. Microsoft Excel test preparation tool is available on the OECD website to aid in calculation of the required amount of DOC in samples. After addition of DOM the volume of the samples is adjusted to 40 ml. The aliquots of samples are taken and diluted for the further instrumental analysis (sections 32-35). Agglomeration behaviour in presence/ absence of CO The influence of CO 2 on the dispersion stability can be checked both for electrolyte- and electrolyte/dom- containing systems. To check whether the dispersion is stable in presence or absence of CO 2 the general procedure described in section 26 or 29 shall be followed. However, to eliminate the possibilities of undesired surface reactions between particle surfaces and dissolved CO 2 /carbonate one has to degas all solutions used in experimental procedures. Thus the dispersion of nanomaterials, solutions of electrolyte and NOM undergo degassing that is 10

11 performed by vigorous bubbling the solutions by N 2 gas for at least 2 hours. Prior and during the experimental procedures all solutions and samples should be kept in closed vessels and tubes sealed with Parafilm when not used. Absence of CO 2 in the sample media or atmosphere around the sample is a crucial factor that can strongly influence the agglomeration behaviour of material in certain cases. Therefore testing the agglomeration behaviour of nanomaterials in absence of CO 2 is strongly recommended. Agglomeration behaviour in dispersions with various ph values 31. The ph values shall cover the ph range of natural waters. For these purposes ph values of 4, 7 and 9 are recommended (8). For working in an open, unbuffered system the presence of 5mM NaHCO 3 as a ph-buffer in the samples is recommended. Experimental data, presented in annex 2, section 15, revealed that smaller concentrations of this agent (1 mm) cannot provide significant buffering capacity, while the larger concentrations (10 mm) may enhance agglomeration and particle sedimentation. ph of the samples should be established using HCl and NaOH in relatively high concentration (0.1 M), so that the total volume of the samples could remain fairly constant. To check the dispersion stability of the particles at these ph values one shall follow the procedure described in section 26. Prior to bringing the sample volume to final volume of 40 ml, 100 mm NaHCO 3 solution is added to the experimental tubes in the volume of 2 ml to reach the concentration of 5 mm in the final volume of 40 ml. After volume adjustment the ph is established at ph= 4, 7 and 9 by adding 0.1 M solution of HCl and NaOH, respectively, and samples are remained intact for equilibration purpose for the time period of 24 hours. Afterwards each sample undergoes 30 second sonication at the power level of 40 W to ensure the particles in the dispersion are as well as possible dispersed under the selected condition. When handling the samples with established ph, the sample tubes should be kept closed in-between the sampling procedures to avoid the influence of atmospheric carbonate to the ionic strength of the samples as much as possible. The ph of dispersion is checked and adjusted if needed. Aliquots of samples are taken and diluted for further instrumental analysis as described in Sample analysis 32. The aliquots of samples prepared as described in sections are diluted for further analysis by addition of deionised water. Resulted samples are vigorously shaken by hand after preparation and right before the analysis, to ensure their proper homogenization. Samples are analysed by ICP-OES or ICP-MS methods. For particles constituted by elements not detectable by those instruments (e.g. CNTs or fullerenes) other analytical methods have to be used. Possible complications during sample preparation and analysis 33. Test dispersions as prepared in sections have to be handled with care to avoid remixing them before sampling the supernatant. Even very little shaking can result in undesired sample mixing especially in the top cm layer of the sample. 11

12 The aliquots of the supernatant of the samples prepared as described in sections are taken from the top cm volume of dispersion. In some cases particles will stick to the air/water interface and caution has to be taken not to get those into the sample aliquot. The recommended volume of taken aliquots is 0.5 ml. The aliquots are taken from the prepared samples every hour, starting hour 0 and finishing hour 6. Such frequency is found to be an optimal time to investigate the dispersibility and dispersion stability in the same time. Before the sampling at hour 6, the samples undergo centrifugation to enhance the particle sedimentation. A Microsoft Excel test preparation tool is available on the OECD website to aid calculation of required centrifugation time based on the particle size cut-off (recommended value 1000 nm) and their density as well as parameters of centrifuge, such as speed of rotation and diameter 35. Samples prepared as described in section 32 should be analysed within 24 hours from the time of preparation. Lacking to do so can result in unreliable results being obtained. Obtained results can be incorrect due to particle sedimentation and loss on the tube walls. Samples of acidsoluble particles might be stabilized by acid addition such as HNO 3 in the form of the dissolved ions. If a direct measurement of the particles shall be performed (either out of practical reasons or to avoid a tedious sample digestion) common practices of slurry analysis shall be applied. Particles then should be stable as dispersion or measures to provide stability should be applied. Samples then should be diluted with MQ water sufficiently to ensure additional dispersion stability or suitable stabilizing agents need to be added (e.g. surfactants, sodium diphosphate as appropriate). If particles cannot be stabilized against losses due to sedimentation or attachment to vessel walls, a complete digestion of the sample is required. Providing knowledge about the point of zero charge (PZC) of the investigated nanomaterials is available, one shall avoid crossing PZC when varying ph. Such variation is a direct reason for particles agglomeration and sedimentation during the analysis. Limit test - test on dispersibility 36. The described testing routine of agglomeration behaviour is performed through the testing of dispersion stability. During the dispersion stability test a number of experimental functions representing the concentration of remaining stable particle dispersion in the supernatant over experimental time are obtained. These functions describe the time-dependent agglomeration and settling behaviour of the test dispersion under the chosen hydro-chemical conditions. The shape of these functions contains valuable information about the character of the material but the interpretation requires fundamental knowledge about colloidal stability. The experimental approach for dispersion stability is cost and time demanding due to the requirement of sub-sampling and analysis in 1h time intervals. Although important information might be retrieved from the shape of the experimental stability functions, the interpretation if the test material is able to form a stable dispersion is taken from the last time-point in the series. Therefore, the described testing routine of agglomeration behaviour (7-point testing) can potentially be shortened and simplified in order to check only sample s dispersibility. In case of such dispersibility check, only 2 samplings/ measurements have to be done. One has to be done 12

13 directly after sample preparation (0h reaction time) and another after 6h of reaction time, and centrifugation of sample. Such experimental approach would provide the data about particle state in the beginning and in the end of experiment, but would not report the dynamics of particle agglomeration and sedimentation, provided by the testing of dispersion stability. Dispersibility testing appears nevertheless convenient, since it provides simple Yes or No answer to the question of particle dispersion stability and can readily be used for the initial assessment of particle agglomeration behaviour. The pitfall of the reduced test comes from numerous difficulties in understanding the origin of obtained data. For example, there would be no chance to determine whether the particles sediment due to a consecutively progressing agglomeration process or due to material heterogeneity. Thus, dispersibility and dispersion stability experimental approaches are strongly bounded to each other. The choice of any of these approaches depends on the aims of the test, and requires the understanding of advantages and disadvantages of each approach, as described above. It is therefore defined that if the simple dispersibility test does not produce an unambiguous decision of a stable or non-stable material (defined as more than 90% and less than 10% of the test material is found in the supernatant after the test) the extended 7-point testing under a broader set of hydrochemical conditions needs to be applied. INVESTIGATION OF AGGLOMERATION BEHAVIOUR BY MEANS OF UV-VIS AND ITS POSSIBLE COMPLICATIONS Investigation of agglomeration behaviour by means of UV-VIS 37. Continuous monitoring of the settling process can be achieved by using UV-VIS spectrometry. Samples for analysis of agglomeration behaviour by UV-VIS are prepared in the way described in sections in regards to sample composition. Samples of interest shall be analysed against blank samples in a double-beam photometer. Blank samples are prepared as the samples of interest, containing all solutes, but in absence of the investigated nanomaterials. As UV-VIS allows analysis of only one sample at a time, the analysis of only one experimental condition at a time can be performed. Thereby, the preparation of only one sample and one corresponding blank sample at a time is required. All samples have to be prepared directly before UV-VIS measurements to avoid particle agglomeration followed by enhanced sedimentation. 38. Prior to the measurements the baseline of UV-VIS spectrometer is set to zero by running the blank samples against each other. After this, the sample of interest is inserted to the devise and analysed against the blank sample for recommended time period of 24h, the spectrometer softare is configured so that a single measurement is being performed each 30 min. Depending on the manufacturer specific layout of the used spectrometer the analysis of dispersion is performed ~1.5 cm from the bottom of the UV-VIS cuvette, thereby lengthening the analysis time from 6h (supernatant analysis) to 24h in the UV-VIS spectrometer. The analysis of the samples is recommended at the wavelength of maximum light adsorption or scattering taking into account background signal from the blank. Too high background signal (> 0.1 AU) should 13

14 be avoided. Spectral properties of the nanomaterial and the test solutions, especially in the presence of NOM, need to be considered in the selection of wavelength. Complications during UV-VIS measurements 39. Typical UV-VIS spectrometer allows analysis of only one sample in 24 hours and therefore the measurements by this technique might be time consuming. Thus to study the particle agglomeration behaviour in all samples mentioned in 25-30, without performing any duplicate measurements, 30 full measurement days would be needed. 40. Particle concentration can be monitored by UV-VIS spectrometer based on light scattering or light adsorption of the particles. The signal to mass ratio for light scattering is typically much lower than for true absorption. Care should be taken to avoid influence of background absorption originating from the NO - 3 ion or added NOM. Therefore, the final results are recommended for presentation in the form of relative to the first measurement adsorption, (A/A 0 ). 41. One shall take into account the possible photocatalytic properties of the investigated nanomaterials. If the particles under investigation reveal photocatalytic activity, the analysis of NOM containing samples shall be performed at the wavelength 380 nm or at least outside of the absorption band of the photocatalyst. Otherwise, the interactions of analyzed particles with UV irradiation may cause the (partial) destruction of NOM molecules via radical formation. As mentioned in section 29 the concentration of NOM shall be adjusted in each special case, based on the specific surface area of the investigated nanomaterial. 42. Special attention should be paid to the choice of cuvettes for UV-VIS measurements. Plastic cuvettes are recommended when UV irradiation is undesirable. 43. Finally, the cleanness of UV-VIS cuvettes should be addressed. Since particles can likely attach to the cuvette walls, cuvettes have to be prewashed with 5N HNO 3 prior to each measurement. A mechanical cleaning of surfaces might be necessary to avoid carry-over. Deposition of particles at the surfaces of the cuvettes can interfere with the stability measurement by suggesting a higher concentration of the NM in the dispersion as in fact is existent. DATA AND REPORTING Data Treatment 44. Results obtained from analysis of the samples prepared as described in sections and analysed as described in are presented in the form of plots, where X-axis stays for time of sampling (0 to 6 hours, 0 to 24h in the UV-VIS measurements) and Y-axis stays for the percentage (%) of the concentration of analysed material relative to the initial concentration (0h measurement) of the material in the samples of interest. The data of the ph and ionic strength 14

15 dependency (each 3 experiments) are plotted together in one graph to visualize the effect of both parameters. Standard deviations should be calculated based on the performed triplicate measurements of each time point and could be displayed in the graphs. The first measurement at 0h might differ from the calculated expected concentration from the dilution of the stock due to losses in the dispersion routine and during sample preparation (setting of ph and ionic strength). The experiment is normalized to the 0h time point (set to 100%) and the deviation from the expected value at 0h is recorded. 45. In case of investigations of particle agglomeration behaviour by UV-VIS the results are presented in the form of plots, where X-axis stays for time (0 to 24h), and Y-axis represents the absorption measured at a certain timepoint relative to the initial value of absorption at 0h (A/A 0 ) Test report 46. The test report should include the following information: (a) Information about the tested material as appropriate (b) Density of the tested material (c) Mass and particle number concentration of material in the investigated samples based on the information of average particle diameter (from DLS analysis) and particle density (from material characterization or manufacturer information. A spherical particle is assumed as particle geometry. (d) Loss of sample at 0h time point referring to the expected value from dilution of the stock (e) Concentration of electrolyte in the investigated samples (f) Concentration of DOM (as DOC) in the investigated samples (g) ph of investigated samples in the moment of preparation, after equilibration time (if applicable) and after the completion of experiment. (h) Plots of relative concentration (C measured / C initial ) 100 (%) versus time (0-6 hours) in case of supernatant analysis (i) Values of final concentration (C measured / C initial ) 100 (%), (6 hours value) in case of supernatant analysis (j) Sample centrifugation parameters (k) Plots of relative adsorption (A measured/ A measured initial) versus time (0-24 hours) in case of UV-VIS measurements (l) Interpretation of the agglomeration behaviour of the investigated material based on the relative scale presented in scheme (m) Discussion on agglomeration reasons FOLLOWING THE TIERED AGGLOMERATION BEHVAIOUR SCHEME AND INTERPRETATION OF OBTAINED RESULTS 47. The developed Test Guideline answers the question if the analysed materials should be tested specifically under consideration of a nanomaterial character in further environmental tests 15

16 and if so, provide the experimental routine to test the behaviour of such materials in aquatic media to estimate their possible fate in and impact on the environment. Presented tiered agglomeration behaviour scheme (see Figure 1) was designed to address the issue of nanomaterial agglomeration behaviour in such way that would require the minimal time- and cost- spending, but allow the complete understanding of reasons and factors influencing this process. 48. Depending on the results the investigated nanoparticles can classified according to the presented decision tree, where class (a) is assigned to non-dispersible materials, class (b) is assigned to materials with condition-dependent dispersion stability and class (c) is assigned to the case of fully dispersible materials. With materials of class (a) 10% of the initially detected (0 h measurement) material concentration remain in the form of a nanomaterial dispersion after 6 hours or experimental time in presence of factors, responsible for agglomeration (ph=4,7,9; [C] electrolyte =0,1,10 mm, Presence of DOM). Materials that remain 10-90% of material in the form of dispersion after 6 hours of experimental time and similar conditions are related to class (b). Finally, class (c) is prescribed to the materials that remained at 90% of initially detected material after similar experimental time and applied conditions. 49. Section one of the decision tree suggests to refer the analysed material to one of the mentioned classes (a), (b), or (c) by determination of material dispersibility through 2-points test (0, 6 h), The 2-point dispersibility testing is sufficient for testing of class (a) and class (c) materials. Contrary, (b) class material would require the full (7-point) dispersion stability test in all conditions, including and excluding presence of DOM. (Section 2 of decision tree). After 7- point dispersion stability test obtained results can be related to the scale of dispersion stability (Section 3 of decision tree). Obtained material concentration can stay within the range of 10-50% or 50-90% of initial measured concentration. Besides finding the remaining concentration of analysed materials in the dispersion, character of obtained dependencies shall be observed. Based on this information the agglomeration rate and agglomeration reasons (e.g. kinetics) shall be discussed. When working with the materials that reveal condition-dependent dispersion stability one shall keep special attention to the smallest variation of factors influencing agglomeration behaviour of nanomaterials

17 559 Figure 1: Tiered Agglomeration Behavior Scheme

18 LITERATURE (1) Salminen, R., et al. (2005). Geochemical Atlas of Europe. Part 1: Background Information, Methodology and Maps. Espoo: Geological Survey of Finland, p (2) von der Kammer, F., S. Ottofuelling and T. Hofmann (2010). Assessment of the Physico- Chemical Behavior of Titanium Dioxide Nanoparticles in Aquatic Environments Using Multi-dimensional Parameter Testing. Environmental Pollution, 158 (12): (3) Ottofuelling, S., F. von der Kammer and T. Hofmann (2011). Commercial titanium dioxide nanoparticles in both natural and synthetic water: comprehensive multidimensional testing and prediction of aggregation behaviour. Environmental Science & Technology, 45: (4) Klein, C. L., et al. (2011). NM-Series of Representative Manufactured Nanomaterials NM- 300 Silver Characterisation, Stability, Homogeneity. Luxembourg: Publications Office of the European Union. (5) Rasmussen, K., et al. (2014). Multi-walled Carbon Nanotubes, NM-400, NM-401, NM-402, NM-403: Characterisation and Physico-Chemical Properties. Luxembourg: Publications Office of the European Union. (6) Rasmussen, K., et al. (2014). Titanium Dioxide, NM-100, NM-101, NM-102, NM-103, NM- 104, NM-105: Characterisation and Physico-Chemical Properties. Luxembourg: Publications Office of the European Union. (7) Nelson, W.G, et al. (2015). Suwannee River Natural Organic Matter: Isolation of the 2R101N Reference Sample by Reverse Osmosis. Environmental Engineering Science, 32 (1): (8) OECD (2004). Guideline for the Testing of Chemicals, No. 111: Hydrolysis as a Function of ph, Organisation for Economic Cooperation and Development, Paris

19 ANNEX 1 DEFINITIONS AND UNITS 1. Agglomeration Reversible grouping of initial small particles in space, happening due to interparticle interactions through various attraction forces and resulting in formation of larger secondary particles (agglomerates). 2. NOM (DOM), SRNOM Natural Organic Matter (Dissolved Organic Matter), Suwannee River Natural Organic Matter 3. PZC Point of Zero Charge, such state at the surface of the material when there are no charged moieties (theoretical value). In contrast to PZC, IEP - Isoelectric Point is such a state on the material surface when the surface net charge equals to zero. 4. Experimental time time, required to test the dispersibility of particles or their dispersion stability is measured in hours (h). The recommended experimental time is 6 hours. 5. Experimental Endpoint Measurement that shows remaining concentration of particles in the top 1 cm of dispersion after 6 hours of experimental time, related to the initial concentration of particles, detected during first measurement, and expressed in percentage (%), such as: Experimental Endpoint = (C 6h /C 0 ) Dispersibility - Particles should be considered dispersible if after 6 hours of experimental time the concentration of particles in the top 1 cm of dispersion exceeds or equals 90% of initial particle concentration at the initial observation time (t 0 ). 7. Agglomeration behaviour (Dispersion Stability) Shows the change of particle concentration in the top 1 cm of dispersion over and during time period of 6 hours. Thus the term dispersion stability includes the term dispersibility and is recommended for further use. 8. Concentrations - Concentrations of the particles and DOM in the samples are given in g/l, mg/l (ppm) and μg/l (ppb). Concentrations of salts (Ca(NO 3 ) 2 and NaHCO 3 ) present in the samples are given in mm. 9. Size - Size of the particles is given in micrometers (μm) or nanometers (nm) and can be determined by using TEM imaging, or in (d.nm) using DLS size measurements, or provided by supplier information. Besides, unit measurement methods should be specified 19