Journal of Chromatography A, 1372 (2014) 204–211

Contents lists available at ScienceDirect

Journal of Chromatography A journal homepage: www.elsevier.com/locate/chroma

Asymmetrical flow-field-flow fractionation coupled with inductively coupled plasma mass spectrometry for the analysis of gold nanoparticles in the presence of natural nanoparticles Boris Meisterjahn a,b , Elisabeth Neubauer a , Frank Von der Kammer a,∗ , Dieter Hennecke b , Thilo Hofmann a,∗ a b

Department of Environmental Geosciences, University of Vienna, Althanstr. 14 UZA II, 1090 Vienna, Austria Fraunhofer Institute for Molecular Biology and Applied Ecology IME, Auf dem Aberg 1, 57392 Schmallenberg, Germany

a r t i c l e

i n f o

Article history: Received 11 June 2014 Received in revised form 6 October 2014 Accepted 25 October 2014 Available online 3 November 2014 Keywords: Hyphenated field flow fractionation Engineered nanoparticles Natural nanoparticles Natural organic matter Heteroaggregation

a b s t r a c t Flow-Field-Flow Fractionation (Flow-FFF), coupled with online detection systems, is one of the most promising tools available for the analysis and characterization of engineered nanoparticles (ENPs) in complex matrices. In order to demonstrate the applicability of Flow-FFF for the detection, quantification, and characterization of engineered gold nanoparticles (AuNPs), model dispersions were prepared containing AuNPs with diameters of 30 or 100 nm, natural nanoparticles (NNPs) extracted from a soil sample, and different concentrations of natural organic matter (NOM), which were then used to investigate interactions between the AuNPs and the NNPs. It could be shown that light scattering detection can be used to evaluate the fractionation performance of the pure NNPs, but not the fractionation performance of the mixed samples that also contained AuNPs because of specific interactions between the AuNPs and the laser light. A combination of detectors (i.e. light absorbance and inductively coupled plasma mass spectrometry (ICP-MS)) was found to be useful for differentiating between heteroaggregation and homoaggregation of the nanoparticles (NPs). The addition of NOM to samples containing 30 nm AuNPs stabilized the AuNPs without affecting the NP size distribution. However, fractograms for samples with no added NOM showed a change in the size distribution, suggesting interactions between the AuNPs and NNPs. This interpretation was supported by unchanged light absorption wavelengths for the AuNPs. In contrast, results for samples containing 100 nm AuNPs were inconclusive with respect to recovery and size distributions because of problems with the separation system that probably related to the size and high density of these nanoparticles, highlighting the need for extensive method optimization strategies, even for nanoparticles of the same material but different sizes. © 2014 Elsevier B.V. All rights reserved.

1. Introduction Engineered nanoparticles (ENPs) are being increasingly used in consumer products. Metal and metal oxide nanoparticles, consisting for example of Ag, Au, ZnO, and TiO2 and carbon-based nanomaterials such as fullerenes and carbon nanotubes, are currently the most commonly used materials [1]. Gold nanoparticles (AuNPs) are mainly used for biomedical applications [2–4] or in sensing devices [e.g. 5–8] and are available as analytical standard materials [9]. ENPs can be released from products in which they are used into the environment [10–13]. ENPs released into aquatic

∗ Corresponding authors. Tel.: +43 1 4277 533 20; fax: +43 1 4277 9533. E-mail addresses: [email protected] (F. Von der Kammer), [email protected] (T. Hofmann). http://dx.doi.org/10.1016/j.chroma.2014.10.093 0021-9673/© 2014 Elsevier B.V. All rights reserved.

and terrestrial environments can aggregate, undergo transformations [14–17], or interact with a complex background of natural nanoparticles (NNPs) that may include clay or oxide particles and natural organic matter (NOM) [15,18–21]. The NNPs are present in much higher (mass) concentrations than the ENPs and have the potential to dominate the fate and behavior of the ENPs [16]. It has for example been shown that natural organic matter (NOM) can replace citrate coatings on AuNPs, enhancing their dispersion stability [22,23], and recent systematic investigations into the fate and behavior of ENPs in aquatic systems have shown that the behavior of such particles is mainly influenced by their surface functionalization [23,24]. However, methods for the detection, quantification, and characterization of ENPs within such complex environmental matrices are either generally absent or still under development [25,26]. Dispersed nanoparticles (NP) systems are metastable systems and low

B. Meisterjahn et al. / J. Chromatogr. A 1372 (2014) 204–211

perturbing methods are therefore required to address these tasks. The use of Flow-Field-Flow Fractionation (Flow-FFF) coupled to a detection system such as inductively coupled plasma mass spectrometry (ICP-MS) is a promising technique for the detection and quantification of ENPs in complex matrices because of its particle size related separation power, its versatility, and its elemental specificity [27,28]. In brief, Flow-FFF is a technique for the continuous separation of particles and macromolecules with diameters between 1 and 1000 nm, which is based on hydrodynamic principles and avoids the use of a stationary phase. Details concerning its principles of operation and applications have been published elsewhere [27,29–32]. Flow-FFF has been widely applied to analyze NNPs, e.g. when investigating interactions between trace elements and NNPs within complex environmental matrices [30,32–34]. The use of Flow-FFF for ENPs such as AuNPs or silver NPs (AgNPs) has, to date, mainly focused on aquatic systems and stability studies [35–38]. Few studies have focused on the analysis of metallic ENPs extracted from biological material, foodstuffs, or organic tissue [26,39–41]. Very few reports have been published on the use of Flow-FFF methods for analyzing ENPs in complex environmental matrices such as soil and sediment, or for investigating interactions between metallic ENPs and NNPs [42]. For Flow-FFF analysis of ENPs in complex samples (such as soils) the NPs first need to be extracted from the matrix and transferred into a stable aqueous dispersion. The resulting dispersion will contain not only ENPs but also NNPs such as clay particles, oxides, and varying amounts of NOM, which will affect the recovery and aggregation state of both ENPs and NNPs [23,43–45]. The objective of this study is to demonstrate the applicability of Flow-FFF multidetection for the detection, quantification, and characterization of metallic ENPs in the presence of NNPs. For this purpose we used mixtures of different sized AuNPs, with sizes at the lower and upper end of the nano-size-range (1–100 nm), with NNPs extracted from a soil sample [34,46] as model dispersions. The effect of different NOM concentrations on the stability and interactions of the NPs was investigated by analyzing dispersions with added NOM, and dispersions without additional any NOM. 2. Materials and methods 2.1. Chemicals The Milli-Q water used in this study was prepared using a Millipore Advantage A10 system (Millipore, Billerica, USA) equipped with a Bio-PakTM Ultrafilter (5000 g mol−1 molecular weight cutoff) for final clean-up. Sodium pyrophosphate (analytical grade) was purchased from Merck (Darmstadt, Germany), as were the HNO3 (68% suprapure) and HCl (30% suprapure) used for aqua regia digestion. Aqueous suspensions of citrate-stabilized AuNPs with nominal diameters of 30 nm (30 nm AuNPs) and 100 nm (100 nm AuNPs) were purchased from BBI International (Cardiff, UK). The Au concentrations in the 30 nm AuNP and 100 nm AuNP suspensions were determined by inductively coupled plasma optical emission spectroscopy (ICP-OES) analysis (Perkin Elmer Optima 5300 DV, PerkinElmer, Beaconsfield, UK) following digestion with concentrated aqua regia solution at 80 ◦ C, and were found to be 63 mg L−1

205

and 50 mg L−1 , respectively. Suwannee River natural organic matter (SRNOM) was purchased from the International Humic Substance Society (St. Paul, MN, USA). NanosphereTM latex beads of 40, 80, 200 and 400 nm diameter were purchased from Duke Scientific (Palo Alto, CA, USA) for calibration of the channel. A sample of NNPs from a colloidal soil extract, with a particle mass concentration of 200 mg L−1 , was used as a model system for a complex NNP matrix [46]. The wet extraction procedure used to prepare the NNP extract has been previously described in [18]. The extract contains particles with sizes ranging between 10 and 400 nm [46] and is routinely used as an in-house reference sample to validate and compare the performance of Flow-FFF analytical methods [34,47]. 2.2. Samples Two series of aqueous samples containing AuNPs (at a calculated particle mass concentration of 2 mg L−1 ) and NNPs (at a calculated particle mass concentration of 143 mg L−1 ) were investigated in this study. The samples were prepared by mixing the AuNP stock suspensions with the NNP suspension and diluting the mixture with Milli-Q water. The samples were then stored cool (4 ◦ C) and dark place for two months prior to taking any measurements, in order to ensure complete interaction. Samples were prepared both with and without the addition of SRNOM (final calculated concentration 28 mg L−1 ) in order to investigate the influence of NOM on the stability of the AuNP-NNP system. Total Au concentrations were determined following aqua regia digestion; the nominal Au and NNP concentrations of the samples are shown in Table 1. 2.3. Asymmetric Flow-FFF-UV/VIS-MALLS-ICP-MS A Wyatt Eclipse 3+ Flow-FFF system (Wyatt Technology Europe, Dernbach, Germany) was used for the fractionation of the samples. The channel had a length of 29.6 cm and was equipped with a 10 kDa MWCO regenerated cellulose membrane (NADIR, Wiesbaden, Germany). The carrier solution was delivered by an Agilent 1200 series quaternary HPLC pump equipped with a micro vacuum degasser. Carrier solutions were filtered with Anodisc 0.02 ␮m membrane filters (Whatman, Maidstone, UK) prior to use. The outflow was analyzed using a UV/Vis diode array detector (Agilent Technologies 1200 series DAD; absorption wavelengths selected: 260, 280, 540 nm), multi angle laser light scattering MALLS (DAWN HELEOS II, Wyatt Technologies, Dernbach, Germany), and ICP-MS (Agilent Technologies 7700x, Waldbronn, Germany). The channel recovery was determined by injecting samples without applying any cross flow. The details of the Flow-FFF and ICP-MS run conditions and parameters are summarized in Table 2. The ICP-MS was calibrated by manual injection of metal standard solutions via a Rheodyne 9725i switch valve (Rheodyne, CA, USA) between the channel outlet and the ICP-MS interface. In addition to Au, Fe and Al were also chosen as representative elements for the NNP sample, which mainly contains oxides (e.g. iron oxides) or clay minerals. 2.4. Data treatment The Flow-FFF separation of the samples was evaluated using MALLS analysis and by calculating radius of gyration values. The

Table 1 Nominal concentrations of AuNPs and NOM, particle mass concentrations of NNPs in mixed samples. Sample

AuNP diameter (nm)

[Au] nominal (mg L−1 )

[NNP] (mg L−1 )

Au30 + NNP Au30 + NNP + NOM Au100 + NNP Au100 + NNP + NOM

30 30 100 100

2 2 2 2

143 143 143 143

[NOM] (mg L−1 ) 28 28

206

B. Meisterjahn et al. / J. Chromatogr. A 1372 (2014) 204–211

Table 2 Flow-FFF and ICP-MS operational parameters. Channel length (tip to tip) (cm) Spacer height (␮m) Focus flow rate (mL min−1 ) Focus time (min) Cross flow rate (mL min−1 ) Channel flow rate (mL min−1 ) Carrier composition RF- power (W) Carrier gas flow (He) (L min−1 ) Collision cell gas flow (mL min−1 ) Torch material Spray chamber Nebulizer Sample & skimmer cone

29.6 350 0.4 6 0.4 1.0 MQ-water + 2.5% v/v 12.5 mM Na4 P2 O7 pH: 8 1550 1.04 4.00 Quartz Quartz Micromist Ni

radius of gyration (rg ) values were calculated with ASTRA 5.4 software (Wyatt Technology, Dernbach, Germany) using a first order ZIMM fit [46] of the Rayleigh ratios obtained at different angles. This method has been shown to be viable for determining size distributions and for evaluation of the quality of the fractionation of NNPs using Flow-FFF [46,48]. In order to derive a size distribution from the measured fractograms the retention times were converted to hydrodynamic sizes using the calibration function determined in fractionations of the 40 nm–400 nm diameter NanosphereTM standards, after subtracting the time of the void peak (the peak from unretained compounds). The standard sizes (independent of the manufacturer’s specifications) were determined by fitting the light scattering patterns of the peak maxima with a spherical particle model provided in the ASTRA 5.4 software (Fig. SI1). Calibration functions for converting the ICP-MS signals to concentrations were set up by plotting the averaged intensities of standard solutions against the standard concentrations, after subtracting the background signal. The signal intensities in the fractograms were then converted into concentration values. To determine the channel recovery percentage for AuNPs, the peak areas in fractograms obtained using different detectors were determined by integration using the OriginPro 8.5 software (OriginLab Corporation, Northampton, USA). The recoveries were then calculated as the ratio of the peak areas in the fractograms to those in the recovery runs without any cross flow. Total AuNP recoveries were calculated as the ratio of the Au concentrations derived from the peak areas in the Au Flow-FFF-ICP-MS-fractograms to the total Au concentrations determined by ICP-OES analysis following aqua regia digestion. Values of the mean cloud thickness in the Flow-FFF channel for the analyzed particles were obtained from the ISIS FFF-simulation software (Wyatt Technology Europe, Dernbach, Germany). 3. Results and discussion 3.1. NNP size distributions from Flow-FFF-MALLS We first evaluated the fractionation of the NNPs under the chosen conditions. Fig. 1 shows the Flow-FFF-MALLS fractograms for the NNP dispersions, with and without additional NOM. The evolution of the radius of gyration over time shows the expected linear increase indicating ideal fractionation in the normal mode of Flow-FFF without visible steric inversion as it is known for this sample from previous studies [34,47]. Small interferences near the void peak are caused by larger, unretained or improperly focused particles. The samples are well within the usable size range of the method, as shown by the channel calibration with size standards.

Fig. 2(a and b) shows the Flow-FFF-MALLS fractograms and the calculated radius of gyration values for mixtures of NNPs with 30 nm AuNPs and with 100 nm AuNPs. The samples containing 30 nm AuNPs show a small shoulder at the beginning of the NNP distribution, but no deviation of the calculated radii from NNP sample with no added AuNPs could be observed. In the samples containing 100 nm AuNPs the scattering signal of the AuNPs dominates the signal trace and calculation of the radius of gyration is disturbed by the presence of the AuNPs. MALLS detectors are known not to be suitable for metallic ENPs because of their unusual scattering behavior due to the plasmon resonance effect; atypical refractive indices and depolarization of the scattered light are not taken into account by commonly used light scattering instruments and data processing software [41,49]. In addition, the Flow-FFF separation has insufficient resolution to provide the MALLS detector with monodisperse particle fractions for each acquisition interval. NNP size distribution analysis, and probably even the fractionation monitoring, may therefore not be possible using light scattering techniques when metallic nanoparticles such as AuNPs are present in the sample, even if they are present in much lower concentrations than the NNPs. 3.2. Channel recoveries and total recoveries for evaluation of Flow-FFF performance The channel recoveries of pure NNP samples, determined using MALLS and ICP-MS detection, are summarized in Table 3. Recoveries determined using the MALLS detector were >90% for the NNP samples without additional NOM (Fig. 1, black curve) and 56% for the NNP samples containing additional NOM (Fig. 1, gray curve; Table 3). This is contrary to the higher recoveries that might be theoretically expected for the NOM-containing samples due to colloidal stabilization of NNPs by the NOM. Recovery calculations based on 56 Fe and 27 Al ICP-MS signals confirmed the results obtained using MALLS detection, also recording lower channel recoveries for those samples containing additional NOM (Table 3). A possible explanation for these different recoveries might be that NNPs coated with NOM are probably more likely to become attached to the membrane surface, but this is contradicted by the observation of no differences in Flow-FFF-ICPMS fractograms of mixed AuNP-NNP samples for the NNPs (represented by Fe and Al) between samples with and without NOM (Figs. 3 and 4), In addition it has to be stated that the determination of channel recoveries is probably not accurate using the ratio between peak areas of runs with and without cross flow respectively. The peak areas of Fe and Al in runs without cross flow were observed to be lower for samples without NOM compared to the samples with NOM (Table SI1). Therefore the recoveries that are determined this way are not reliable because the reference runs are probably not representing the total amount of particles in the samples, which is a prerequisite for this method for recovery determination. Similar effects not only for the NNPs but also the AuNPs were observed for other samples and will be discussed in more detail below. The channel recoveries of Au for the mixture of NNPs with 30 nm AuNPs were 99% without additional NOM and 152% with additional NOM, indicating that there was no loss of particles during the fractionation process. However, for the AuNPs the peak area in the reference runs with no applied cross flow, which should represent the total injected amount of Au, was reduced to 68% of that in the samples without additional NOM (Table SI1). This explains the channel recovery >100% for the sample with additional NOM. The total Au concentration determined for the samples following acid digestion indicated a significant loss (∼30–40%) in both samples. This reduction in the total recovery relative to the total Au concentration is also observed in pure 30 nm AuNP dispersions and therefore an influence of the NNPs or the NOM could be excluded.

B. Meisterjahn et al. / J. Chromatogr. A 1372 (2014) 204–211

207

Fig. 1. Flow-FFF-MALLS fractogram for NNP samples with no AuNPs. Lines: MALLS signals at 90◦ ; dots: radius of gyration (1st order ZIMM fit). Black signals: sample without additional NOM; light gray signals: sample with additional NOM; left axis: MALLS signal intensity right axis: radius of gyration.

The total concentration of Au in both samples is similar and near to the calculated value of 2 mg L−1 from dilution of the stock dispersions. However, the similar AuNP concentrations for both samples indicate that the peak areas in the reference runs without applied cross flow should show not the observed variation (Table SI1). These results lead to the conclusion that there are losses of AuNPs in other steps of the experiment, possibly through the sedimentation of NPs in the sample vial, or through losses in the analytical system due, for example, to adsorption to the walls of tubing (Table 3), even in runs without any cross flow. These losses affect the precondition for the determination of channel recoveries that the total amount of particles should be detected in runs without cross flow. Therefore only the total recoveries can give reliable information about the performance of the system. The total recovery of AuNPs was higher for the samples containing additional NOM which indicates a better stabilization of the AuNPs in these samples and is in agreement with the results from the Flow-FFF-ICPMS fractograms discussed in Section 3.3. For the NNPs channel recoveries of nearly 100% were determined in both 30 nm AuNP-NNP-samples on the basis of 56 Fe and 27 Al signals. The presence of additional NOM had no effect on the Fe and Al recoveries. The MALLS intensities were also reduced in those samples containing additional NOM (Fig. 2a), as well as in the pure NNP samples, which can be explained by absorption of the laser light by the additional NOM in these samples, and probably also by the AuNPs.

In the samples containing 100 nm AuNPs and NNPs the NNP channel recovery, represented by the recoveries of Fe and Al, was close to 100%, while the channel recovery of 100 nm AuNPs was low compared to the recovery of AuNPs in samples containing 30 nm AuNPs and NNPs. The peak areas of the AuNPs and the NNPs, represented by Fe and Al, in the reference runs with no applied cross flow showed a variation between the different samples and therefore the derived values are probably not reliable and for evaluation of the AuNP separation the total recoveries should be used. In contrast to the 30 nm AuNPs, the 100 nm AuNPs exhibited lower total recoveries in the NOM-containing samples. This reduced recovery indicates a loss of 100 nm AuNPs following the addition of NOM in the analytical system or in the sample itself, since both samples contained the same amount of Au, as determined by ICP-MS analysis following aqua regia digestion. The reason for this apparent selective loss of 100 nm AuNPs following the addition of NOM remains unclear. Effects such as bridging between adsorbed NOM layers by divalent cations [50] seem to be an unlikely explanation because all samples should be hydrochemically identical and no similar particle losses were observed for 30 nm AuNPs in the presence of NNPs and additional NOM. The 100 nm AuNPs show even lower total recoveries than the 30 nm AuNPs (Table 3) in case of the pure dispersion. The presence of NNPs thus seems to have a stabilizing effect on the 100 nm AuNPs. Additionally, for an explanation of the low

Table 3 Channel recoveries for the analyzed samples calculated from MALLS fractograms (Figs. 1 and 2) and ICP-MS analysis (Figs. 3 and 4). Sample

Channelrecovery (%) (MALLS 90◦ )

Channel-recovery (%) (ICP-MS) 56 Fe/27 Al

NNPs NNPs + NOM AuNP 30 nm (pure dispersion) AuNP 100 nm (pure dispersion) AuNP 30 nm + NNP AuNP 30 nm + NNP + NOM AuNP 100 nm + NNP AuNP 100 nm + NNP + NOM

91 56

136/101 74/46

a b

-a -a -a -a

88/96 87/95 146/131 96/97

Channelrecovery (%) (ICP-MS) 197 Au

[Au] (mg L−1 ) (from digestion)

[Au] (mg L−1 ) (from peak area fractograms)

Total recovery Aub (%)

114 2 99 152 100 76

63 51 2.01 1.85 1.89 1.98

34.8 0.029 1.25 1.31 0.77 0.30

55 0.6 62 71 41 15

Not determined because of overlap of AuNP and NNP scattering and AuNP light absorption. Ratio of Au concentrations from peak area analysis and digestion results.

208

B. Meisterjahn et al. / J. Chromatogr. A 1372 (2014) 204–211

Fig. 2. Flow-FFF-MALLS fractograms for samples containing AuNPs and NNPs: (A) samples with 30 nm AuNPs + NNPs; lines: MALLS signals at 90◦ ; dots: radius of gyration values; black: samples without additional NOM; light gray: samples with added NOM; (B) samples with 100 nm AuNPs + NNPs; left axis: MALLS signal intensity; right axis: radius of gyration.

recoveries of the 100 nm AuNPs it has to be considered that due to the high density particles do sediment a distance of a few micrometers during the time scale of a fractionation run. This distance is comparable to, or even exceeds, the dimension of the mean cloud thickness of the diffusional cloud in the Flow-FFF channel, which is, according to FFF-theory, approximately 3.5 ␮m for the 30 nm AuNPs and 1.7 ␮m for the 100 nm AuNPs. With the dimensions and density of the used AuNPs the sedimentation velocities can be calculated. According to these considerations the 30 nm AuNPs travel the distance of their respective mean cloud thickness in 7 min and the 100 nm AuNP in 0.2 min, demonstrating a huge difference due to the particle dimensions and the high density of the AuNPs. The reduced recovery (as well as other irregularities

observed in the fractionation) for the samples containing 100 nm AuNPs may be partially attributable to this effect. Lower recoveries could also be explained by sedimentation in the sample vial prior to the analysis, as these vials were not shaken immediately prior to the fractionation run. However, since samples both with and without NOM were fractionated under identical conditions in the Flow-FFF channel, the influence of sedimentation on the recovery should be identical for all of the samples containing 100 nm AuNPs. Nevertheless, this effect needs to be taken into account in future optimization of Flow-FFF methods for similar ENPs and NNPs. The results from the analyzes of samples containing 100 nm AuNPs are therefore not fully quantitative and highlight the need for proper method development and optimization that takes into account these sorts

B. Meisterjahn et al. / J. Chromatogr. A 1372 (2014) 204–211

209

Fig. 3. Flow-FFF-ICPMS fractograms for pure dispersion of 30 nm AuNPs and mixed samples containing 30 nm AuNPs and NNPs (solid lines: samples without NOM; dashed lines: samples with added NOM). Left axis: Au-concentration, right axis: Fe and Al concentration.

of effects. The reason for the variation and reduced peak areas of both AuNPs and NNPs in the reference runs without any cross flow remains unclear. For the AuNPs, the peak areas were reduced in samples with additional NOM, while the amount of sample injected for the NNPs, represented by the Fe and Al peak areas in runs without applied cross flow, increased in those samples with additional NOM (Table SI1). This second observation can probably be explained by enhanced colloidal stabilization caused by the NOM as it is well known for NNPs and ENPs [23,50,51], or by signal enhancement due to the presence of organic carbon [52], buts this should have increased the AuNP signals as well. One difference compared to the fractionation runs with applied field is the parallel elution of

the NOM and AuNPs under no-field conditions, whereas they are transferred separately to the ICP-MS interface in runs with applied field. A loss of sample due to adsorption of AuNPs to the connection tubings between the Flow-FFF and ICPMS-interface under these conditions might be an explanation for the observations described above. In general our results have shown that the commonly used approach to calculate recoveries in Flow-FFF applications might lead to erroneous results. It seems that the precondition that there are no losses in runs without cross flow is not always fulfilled. In general, this method has the problem, that dissolved species are detected also in such runs and the calculated recoveries are

Fig. 4. Flow-FFF-ICPMS fractograms for pure dispersions of 100 nm AuNPs and for mixed samples containing both 100 nm AuNPs and NNPs (solid lines: samples with no added NOM; dashed lines: samples with added NOM).

210

B. Meisterjahn et al. / J. Chromatogr. A 1372 (2014) 204–211

underestimated, especially when using a detector that cannot differentiate between particles and dissolved compounds such as conventional ICPMS. For certain NPs like Ag or AuNPs this can be accounted for by using particle selective detection such as e.g. the specific UV Vis absorption of these NPs. However, these detection methods are not always sensitive enough to detect these NPs in the low ␮g L−1 range. However, if particle losses are also observed in these reference runs the determination of the recovery is not reliable any more. In this case, the total recovery should be used for the evaluation of the analytical system. This method has also drawbacks as it can lead to underestimated recoveries for the same reason of being blind to differences between particles and dissolved compounds. However, the total recovery can give reliable information about the recovery for NPs such as the AuNPs, which are unlikely to dissolve.

3.3. Elemental size distributions from Flow-FFF-ICP-MS The elemental size distributions for mixtures of 30 nm and 100 nm AuNPs with NNPs are shown in Figs. 3 and 4. In all samples the Fe and Al signals show a continuous distribution over the diameter range from 1 to 400 nm, with the Fe-containing NNPs showing a tendency toward smaller particle sizes (peak maximum at approximately 60 nm) and the Al-containing NNPs showing a tendency toward larger particle sizes (peak maximum at 120 nm). Al can be taken as a representative metal for the clay particles in the sample which are larger platelike particles, while the Fe represents mainly Fe-oxide particles which are present in form of smaller, more compact particles and more enriched in the nano-size range [34,46]. The AuNP distributions for the samples containing NNPs and NOM differed from those for AuNP-NNP-samples with no added NOM. The particle size distributions for the 30 nm AuNP samples containing NOM were the same as those for a pure dispersion of 30 nm AuNPs (Fig. 3). In contrast, in AuNP-NNP-samples without additional NOM the AuNP peak showed an asymmetric and broadened shape. We assume that the change of the size distribution was caused by heteroaggregation of the AuNPs with NNPs, resulting in some of the AuNPs being eluted according to the size of the (larger) NNPs, on which the AuNPs may have been transported through the channel. This assumption is supported by the fact that there is no visible red-shift in the absorption wavelength in the UV–Vis detection (Fig. SI3a), as has been reported for homoaggregation between AuNPs [22]. The surfaces of NNPs are also known to interact with NOM [44,45,50,53,54], and the citrate coating of AuNPs can be replaced by NOM [22,23,55]. This leads to an additional steric and (depending on the mineral surface) electrostatic stabilization and prevents the association of AuNPs with NNPs. Because fractograms of pure 30 nm AuNP dispersions, with and without added NOM, show the same distribution (Fig. SI 2), the observed change in peak shape in the AuNP-NNP and AuNP-NNP-NOM samples cannot be caused by interactions between the 30 nm AuNPs and parts of the separation system (e.g. the membrane in the Flow-FFF channel). Although the distribution of the 30 nm AuNPs changed in the presence of NNPs with no added NOM due to possible interactions with the NNPs, the channel recovery of 30 nm AuNPs remained quantitative and no 30 nm AuNPs were removed from the sample by the processes that caused the broadening of the peak. Elemental size distributions for samples containing 100 nm AuNPs and NNPs are shown in Fig. 4, revealing no difference in peak shape between samples with added NOM and those without. Hence, associations between the 100 nm AuNPs and the NNPs could not be revealed by changes of the size distribution. Due to the low recoveries the signal intensities in the UV-DAD 3D spectra were greatly reduced and no additional information

concerning the aggregation process could be derived from this detector (Figs. SI 3d and e). The peaks for the 100 nm AuNPs show a pronounced tailing, even in the fractograms of pure dispersions containing neither NNPs nor NOM (dashed black line in Fig. 4). In general, large nanoparticles such as the 100 nm AuNPs are fractionated closer to the accumulation wall than small nanoparticles, where they can experience more frequent or stronger attractive interactions with this surface. It is therefore likely that retention in the channel is partially also occurring due to interactions with the accumulation wall (membrane). The peak tailing that was visible even in fractionations of pure 100 nm AuNP samples (dashed black line in Fig. 4) can be explained by (i) additional chromatography-like processes in the channel, where particles experience reversible attractive interactions with the membrane surface, and to a certain extent, also by (ii) homoaggregation. The surface chemistry of the NNPs differs from that of AuNPs and these particles would therefore require different carrier compositions in FFF to ensure optimal separation, good channel recovery, and accurate sizing. Hence, the run conditions in the channel, and also the composition of the carrier liquid (additives, ionic strength, pH) represented the best compromise for both types of NPs. This proved to work well for the 30 nm AuNPs but less well for the larger ones. Further experiments and additional optimization of the separation system (specifically for the 100 nm particles) would be necessary to explain the nature of the interactions and differences in behavior between the 100 nm AuNPs and the 30 nm AuNPs. 4. Conclusion The results presented in this paper demonstrate the applicability (and limitations) of Flow-FFF-UV/Vis-MALLS-ICP-MS for the analysis of metallic ENPs in complex environmental matrices. The use of static light scattering to check separation performance can be of limited value when metallic NPs (such as gold or silver NPs) are present in the sample, as these are incompatible with currently available light scattering detectors. The combination of different detection systems allows in principle the investigation of the interactions between ENPs and NNPs which has been demonstrated for the 30 nm AuNPs although this was not possible for the 100 nm AuNPs. The comparison of samples with different NOM concentrations mitigate the common assumption that heteroaggregation will always dominate the fate of ENPs in environmental matrices, resulting in the removal of ENPs from the aqueous phase [16,21]. Our results have highlighted the need for a proper optimization of Flow-FFF methods in order to exclude or minimize any possible influences of the separation system itself. In addition, if real environmental samples are to be analyzed using Flow-FFF the ENPs first need to be extracted from the matrix into a stable liquid suspension. The resulting colloidal extract would be similar to the model systems presented herein, so that such models will be a useful tool for the optimization and evaluation of Flow-FFF methods prior to the application to real samples. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.chroma. 2014.10.093. References [1] Project on Emerging Nanotechnologies, http://www.nanotechproject.org/ inventories/consumer, accessed October 2012. [2] E. Katz, I. Willner, Integrated nanoparticle–biomolecule hybrid systems: synthesis, properties, and applications, Angew. Chem. Int. Ed. 43 (2004) 6042–6108.

B. Meisterjahn et al. / J. Chromatogr. A 1372 (2014) 204–211 [3] D.C. Hone, P.I. Walker, R. Evans-Gowing, S. FitzGerald, A. Beeby, I. Chambrier, M.J. Cook, D.A. Russell, Generation of cytotoxic singlet oxygen via phthalocyanine-stabilized gold nanoparticles: a potential delivery vehicle for photodynamic therapy, Langmuir 18 (2002) 2985–2987. [4] D. Pissuwan, S.M. Valenzuela, M.C. Killingsworth, X. Xu, M.B. Cortie, Targeted destruction of murine macrophage cells with bioconjugated gold nanorods, J. Nanopart. Res. 9 (2007) 1109–1124. [5] R. Elghanian, Selective colorimetric detection of polynucleotides based on the distance-dependent optical properties of gold nanoparticles, Science 277 (1997) 1078–1081. [6] J. Kneipp, H. Kneipp, W.L. Rice, K. Kneipp, Optical probes for biological applications based on surface-enhanced Raman scattering from indocyanine green on gold nanoparticles, Anal. Chem. 77 (2005) 2381–2385. [7] L. Polavarapu, Q.-H. Xu, Water-soluble conjugated polymer-induced selfassembly of gold nanoparticles and its application to SERS, Langmuir 24 (2008) 10608–10611. [8] S. Eustis, M.A. El-Sayed, Why gold nanoparticles are more precious than pretty gold: noble metal surface plasmon resonance and its enhancement of the radiative and nonradiative properties of nanocrystals of different shapes, Chem. Soc. Rev. 35 (2006) 209. [9] NIST RM 8012 30 nm Gold nanoparticles, www-s.nist.gov/srmors/view detail.cfm?srm=8012 [10] T.M. Benn, P. Westerhoff, Nanoparticle silver released into water from commercially available sock fabrics, Environ. Sci. Technol. 42 (2008) 4133–4139. [11] J. Farkas, H. Peter, P. Christian, J.A. Gallego Urrea, M. Hassellöv, J. Tuoriniemi, S. Gustafsson, E. Olsson, K. Hylland, K.V. Thomas, Characterization of the effluent from a nanosilver producing washing machine, Environ. Int. 37 (2011) 1057–1062. [12] R. Kaegi, B. Sinnet, S. Zuleeg, H. Hagendorfer, E. Mueller, R. Vonbank, M. Boller, M. Burkhardt, Release of silver nanoparticles from outdoor facades, Environ. Pollut. 158 (2010) 2900–2905. [13] B. Nowack, J.F. Ranville, S. Diamond, J.A. Gallego-Urrea, C. Metcalfe, J. Rose, N. Horne, A.A. Koelmans, S.J. Klaine, Potential scenarios for nanomaterial release and subsequent alteration in the environment, Environ. Toxicol. Chem. 31 (2012) 50–59. [14] G.E. Batley, J.K. Kirby, M.J. McLaughlin, Fate and risks of nanomaterials in aquatic and terrestrial environments, Acc. Chem. Res. (2012), 120703143823005. [15] M. Delay, F. Frimmel, Nanoparticles in aquatic systems, Anal. Bioanal. Chem. 402 (2011) 583–592. [16] S.J. Klaine, P. Alvarez, G. Batley, T. Fernandes, R. Handy, D. Lyon, S. Mahendra, M. McLaughlin, J. Lead, Nanomaterials in the environment: behavior, fate, bioavailability, and, effects, Environ. Toxicol. Chem. 27 (2008) 1825–1951. [17] R. Kaegi, A. Voegelin, C. Ort, B. Sinnet, B. Thalmann, J. Krismer, H. Hagendorfer, M. Elumelu, E. Mueller, Fate and transformation of silver nanoparticles in urban wastewater systems, Water Res. 47 (2013) 3866–3877. [18] M. Wiesner, G. Lowry, E. Casman, P. Bertsch, C. Matson, R. Di Giulio, J. Liu, M. Hochella, Meditations on the ubiquity and mutability of nano-sized materials in the environment, ACS Nano 11 (2011) 8466–8470. [19] Y. Ju-Nam, J. Lead, Manufactured nanoparticles: an overview of their chemistry, interactions and potential environmental implications, Sci. Total Environ. 400 (2008) 396–414. [20] N.S. Wigginton, K.L. Haus, M.F. Hochella Jr., Aquatic environmental nanoparticles, J. Environ. Monit. 9 (2007) 1306. [21] A. Praetorius, M. Scheringer, K. Hungerbühler, Development of environmental fate models for engineered nanoparticles—a case study of TiO2 nanoparticles in the Rhine River, Environ. Sci. Technol. 46 (2012) 6705–6713. [22] S. Diegoli, A.L. Manciulea, S. Begum, I.P. Jones, J.R. Lead, J.A. Preece, Interaction between manufactured gold nanoparticles and naturally occurring organic macromolecules, Sci. Total Environ. 402 (2008) 51–61. [23] J. Liu, S. Legros, G. Ma, J. Veinot, F. von der Kammer, T. Hofmann, Influence of surface functionalization and particle size on the aggregation kinetics of engineered nanoparticles, Chemosphere (2012). [24] A.R. Poda, A.J. Kennedy, M.F. Cuddy, A.J. Bednar, Investigations of UV photolysis of PVP-capped silver nanoparticles in the presence and absence of dissolved organic carbon, J. Nanopart. Res. 15 (2013). [25] F. von der Kammer, P.L. Ferguson, P.A. Holden, A. Masion, K.R. Rogers, S.J. Klaine, A.A. Koelmans, N. Horne, J.M. Unrine, Analysis of engineered nanomaterials in complex matrices (environment and biota): general considerations and conceptual case studies, Environ. Toxicol. Chem. 31 (2012) 32–49. [26] A.R. Poda, A.J. Bednar, A.J. Kennedy, A. Harmon, M. Hull, D.M. Mitrano, J.F. Ranville, J. Steevens, A. Poda, A. Bednar, A. Kennedy, D. Mitrano, J. Ranville, Characterization of silver nanoparticles using flow-field flow fractionation interfaced to inductively coupled plasma mass spectrometry, J. Chromatogr. A 1218 (2011) 4219–4225. [27] F. von der Kammer, S. Legros, E. Larsen, K. Loeschner, T. Hofmann, Separation and characterization of nanoparticles in complex food and environmental samples by field-flow fractionation, Trends Anal. Chem. 30 (2011) 425–436. [28] D.M. Mitrano, A. Barber, A. Bednar, P. Westerhoff, C.P. Higgins, J.F. Ranville, Silver nanoparticle characterization using single particle ICP-MS (SP-ICP-MS) and asymmetrical flow field flow fractionation ICP-MS (AF4-ICP-MS), J. Anal. At. Spectrom. 27 (2012) 1131–1142. [29] M. Schimpf, K. Caldwell, J. Giddings (Eds.), Field-Flow Fractionation Handbook, Wiley-Interscience, New York, 2000. [30] E. Neubauer, F. von der Kammer, T. Hofmann, F.v.d. Kammer, Influence of carrier solution ionic strength and injected sample load on retention and recovery of

[31]

[32]

[33] [34]

[35]

[36]

[37]

[38]

[39]

[40]

[41]

[42]

[43]

[44]

[45]

[46]

[47]

[48]

[49]

[50]

[51]

[52]

[53]

[54]

[55]

211

natural nanoparticles using flow field-flow fractionation, J. Chromatogr. A 1218 (2011) 6763–6773. M. Hassellöv, F. von der Kammer, R. Beckett, Characterisation of aquatic colloids and macromolecules by field-flow fractionation, in: K. Wilkinson, J. Lead (Eds.), Environmental Colloids and Particles: Behavior, Separation and Characterisation, J. Wiley & Sons, New York, 2007, pp. 224–276. E. Neubauer, F.v.d. Kammer, T. Hofmann, Using FLOWFFF and HPSEC to determine trace metal–colloid associations in wetland runoff, Water Res. 47 (2013) 2757–2769. M. Hassellöv, F. von der Kammer, Iron oxide as geochemical nanovectors for metal transport in soil–river systems, Elements 4 (2008) 399–404. K. Plathe, F. von der Kammer, M. Hassellöv, J. Moore, M. Murayama, T. Hofmann, M. Hochella, Using FlFFF and aTEM to determine trace metal–nanoparticle associations in riverbed sediment, Environ. Chem. 7 (2010) 82–93. S. Cumberland, J. Lead, S.A. Cumberland, J.R. Lead, Particle size distributions of silver nanoparticles at environmentally relevant concentrations, J. Chromatogr. A 1216 (2009) 9099–9105. W. Sermsri, P. Jarujamrus, J. Shiowatana, A. Siripinyanond, W. Sermsri, P. Jarujamrus, J. Shiowatana, A. Siripinyanond, Flow field-flow fractionation: a versatile approach for size characterization of ␣-tocopherol-induced enlargement of gold nanoparticles, Anal. Bioanal. Chem. 396 (2010) 3079–3085. T.J. Cho, V.A. Hackley, Fractionation and characterization of gold nanoparticles in aqueous solution: asymmetric-flow field flow fractionation with MALS, DLS, and UV–vis detection, Anal. Bioanal. Chem. 398 (2010) 2003–2018. L. Calzolai, D. Gilliland, C.P. Garcìa, F. Rossi, Separation and characterization of gold nanoparticle mixtures by Flow-Field-Flow Fractionation, J. Chromatogr. A 1218 (2011) 4234–4239. B. Schmidt, K. Loeschner, N. Hadrup, A. Mortensen, J.J. Sloth, C. Bender Koch, E.H. Larsen, Quantitative characterization of gold nanoparticles by field-flow fractionation coupled online with light scattering detection and inductively coupled plasma mass spectrometry, Anal. Chem. 83 (2011) 2461–2468. K. Loeschner, J. Navratilova, C. Købler, K. Mølhave, S. Wagner, F. Kammer, E.H. Larsen, Detection and characterization of silver nanoparticles in chicken meat by asymmetric flow field flow fractionation with detection by conventional or single particle ICP-MS, Anal. Bioanal. Chem. 405 (2013) 8185–8195. K. Loeschner, J. Navratilova, S. Legros, S. Wagner, R. Grombe, J. Snell, F. von der Kammer, E.H. Larsen, Optimization and evaluation of asymmetric flow field-flow fractionation of silver nanoparticles, J. Chromatogr. A 1272 (2012) 116–125. A.R. Whitley, C. Levard, E. Oostveen, P.M. Bertsch, C.J. Matocha, F. von der Kammer, J.M. Unrine, Behavior of Ag nanoparticles in soil: effects of particle surface coating, aging and sewage sludge amendment, Environ. Pollut. 182 (2013) 141–149. J. Buffle, K. Wilkinson, S. Stoll, M. Filella, J. Zhang, A generalized description of aquatic colloidal interactions: the three-colloidal component model, Environ. Sci. Technol. 32 (1998) 2887–2899. A. Franchi, C.R. O’Melia, Effects of natural organic matter and solution chemistry on the deposition and re-entrainment of colloids in porous media, Environ. Sci. Technol. 37 (2003) 1122–1129. P. Christian, F. Kammer, M. Baalousha, T. Hofmann, Nanoparticles: structure, properties, preparation and behaviour in environmental media, Ecotoxicology 17 (2008) 326–343. F. von der Kammer, M. Baborowski, K. Friese, Field-flow fractionation coupled to multi-angle laser light scattering detectors: applicability and analytical benefits for the analysis of environmental colloids, Anal. Chim. Acta 552 (2005) 166–174. S. Dubascoux, F. von der Kammer, I. Le Hécho, M.P. Gautier, G. Lespes, Optimisation of asymmetrical flow field flow fractionation for environmental nanoparticles separation, J. Chromatogr. A 1206 (2008) 160–165. M. Baalousha, F. von der Kammer, M. Motelica-Heino, H. Hilal, P. Le Coustumer, Size fractionation and characterization of natural colloids by flow-field flow fractionation coupled to multi-angle laser light scattering, J. Chromatogr. A 1104 (2006) 272–281. H. Hagendorfer, R. Kaegi, J. Traber, S.F. Mertens, R. Scherrers, C. Ludwig, A. Ulrich, Application of an asymmetric flow field flow fractionation multi-detector approach for metallic engineered nanoparticle characterization – prospects and limitations demonstrated on Au nanoparticles, Anal. Chim. Acta 706 (2011) 367–378. S. Sander, L.M. Mosley, K.A. Hunter, Investigation of interparticle forces in natural waters: effects of adsorbed humic acids on iron oxide and alumina surface properties, Environ. Sci. Technol. 38 (2004) 4791–4796. M. Delay, T. Dolt, A. Woellhaf, R. Sembritzki, F.H. Frimmel, Interactions and stability of silver nanoparticles in the aqueous phase: influence of natural organic matter (NOM) and ionic strength, J. Chromatogr. A 1218 (2011) 4206–4212. P. Allain, L. Jaunault, Y. Mauras, J.-M. Mermet, T. Delaporte, Signal enhancement of elements due to the presence of carbon-containing compounds in inductively coupled plasma mass spectrometry, Anal. Chem. 63 (1991) 1497–1498. M. Baalousha, A. Manciulea, S. Cumberland, K. Kendall, J.R. Lead, Aggregation and surface properties of iron oxide nanoparticles: influence of pH and natural organic matter, Environ. Toxicol. Chem. 27 (2008) 1875–1882. M. Baalousha, Aggregation and disaggregation of iron oxide nanoparticles: influence of particle concentration, pH and natural organic matter, Sci. Total Environ. 407 (2009) 2093–2101. D.P. Stankus, S.E. Lohse, J.E. Hutchison, J.A. Nason, Interactions between natural organic matter and gold nanoparticles stabilized with different organic capping agents, Environ. Sci. Technol. 45 (2011) 3238–3244.