Characterization of Silver Nanoparticle Products Using Asymmetric

Feb 2, 2012 - EDCE Civil and Environmental Engineering, EPFL − Ecole Polytechnique Federale ..... KONA Powder and Particle Journal 2016 33 (0), 48-6...
1 downloads 0 Views 2MB Size
Article pubs.acs.org/ac

Characterization of Silver Nanoparticle Products Using Asymmetric Flow Field Flow Fractionation with a Multidetector Approach − a Comparison to Transmission Electron Microscopy and Batch Dynamic Light Scattering H. Hagendorfer,*,†,§ R. Kaegi,∥ M. Parlinska,‡ B. Sinnet,∥ C. Ludwig,§,⊥ and A. Ulrich† †

Laboratory of Analytical Chemistry, ‡Electron Microscopy Center, EMPA − Swiss Federal Laboratories for Materials Testing and Research, Ueberlandstrasse 129, CH-8600 Duebendorf, Switzerland § EDCE Civil and Environmental Engineering, EPFL − Ecole Polytechnique Federale de Lausanne, CH-1015 Lausanne, Switzerland ∥ Department of Process Engineering, EAWAG - Swiss Federal Institute of Aquatic Science and Technology, Ueberlandstrasse 133, CH-8600 Duebendorf, Switzerland ⊥ Chemical Processes and Materials Research Group, PSI − Paul Scherrer Institute, CH-5232 PSI Villigen, Switzerland S Supporting Information *

ABSTRACT: Due to the already prevalent and increasing use of silvernanoparticle (Ag-NP) products and the raised concerns in particular for the aquatic environment, analytical techniques for the characterization of such products are of need. However, because Ag-NP products are of different compositions and polydispersities, analysis especially of the size distribution is challenging. In this work, an asymmetric flow field flow fractionation (A4F) multidetector system (UV/vis, light scattering, inductively coupled plasma mass spectrometry - ICPMS), in combination with a method to distinguish and quantify the particle and dissolved Ag fractions (ICPMS after ultracentrifugation), for the characterization of AgNP products with different degrees of polydispersities is presented. For validation and to outline benefits and limitations, results obtained from batch dynamic light scattering (batch-DLS) and transmission electron microscopy (TEM) were compared. With the developed method a comprehensive understanding in terms of dissolved Ag and Ag-NP concentration as well as an element selective, mass- and number particle size distribution (PSD) was obtained. In relation to batch-DLS, the reliability of the data was improved significantly. In comparison to TEM, faster measurement times and the ability to determine the samples directly in dispersions are clearly advantageous. The proposed setup shows potential for a rapid- and reliable characterization method of virtually any polydisperse metallic NP dispersion, many of them available on the market already.

Silver-nanoparticles (Ag-NP) are identified as one of the fastest growing products in the nanotechnology industry.1,2 They show a remarkable antimicrobial behavior and thus are frequently used for all kind of applications already for over 100 years.3 Also in the field of consumer products silver is stated as the most common nanomaterial on the market.4 Again, the antimicrobial properties in combination with low cost manufacturing are the main reasons for their increased use in several products like cosmetics, wall paints, biocide sprays, textiles, laundry detergents, and others.5 Though, besides the beneficial use of Ag-NP, also concerns about adverse effects regarding release,6−8 fate,9,10 and toxicity11−13 to the aquatic environment have been raised.14,15 For example, a release of 30% of the total Ag concentration after the first year of use, for Ag-NP incorporated in facade paints, was reported.7 In terms of toxicity, next to physicochemical parameters such as particle number concentration and PSD,16 also the availability of dissolved Ag,5,17,18 possibly resulting from the oxidation of the Ag-NP,19 seem to have an important influence. However, a recent review20 stressed that results are still contradictory and © 2012 American Chemical Society

that in many of these studies stock suspensions of commercial available Ag-NP were employed, sometimes with limited characterization only. As a result of the increased use of Ag-NP products as well as to draw reliable conclusions from environmental and toxicological studies, appropriate analytical techniques are of need. Indeed, numerous methods for the characterization of metallic engineered nanoparticles (ENPs) such as Ag-NP have been developed already.21−23 Common used techniques for size characterization are dynamic light scattering (DLS)- and scanning/transmission electron microscopy (SEM, TEM, and STEM). However, the biased response of DLS 24 for polydisperse samples toward larger particle size and the timeand cost intensive investigations using electron microscopy25 limits the applicability of these methods, calling for alternative techniques. A promising method already implemented for size Received: October 17, 2011 Accepted: February 2, 2012 Published: February 2, 2012 2678

dx.doi.org/10.1021/ac202641d | Anal. Chem. 2012, 84, 2678−2685

Analytical Chemistry

2679

22 (14−45) 2 (150) 90 (3−>150) n.n. −30 ± 1 n.n. a

PSD: particle size distribution.

Rent-A-Scientist AgPure (RM OECD NM 300K) Colloidal Ag 50 ppm 2

3

NanoSys NanoArgentum 1

50 mg/kg

48.5 ± 0.5 mg/kg

23

14 (9−44) 8 (2−65) 8 (5−90) 35 (5−90) 99.5 99 ± 9 g/kg

99

7 (3−44) 6 (6−160) 55 (6−160) 26 −42.5 ± 0.5 n.n.

product name

1000 mg/kg 10% w/w

940 ± 7 mg/kg

60

TEM number PSDa mode (range) A4F number PSDa mode (range) A4F mass PSDa mode (range) specified mean size [nm] zeta potential [mV] (n = 3) measured particle fraction [%] specified particle fraction [%] measured total conc. (n = 3) specfied total conc.

EXPERIMENTAL SECTION Chemicals. Nitric acid (65% w/w) and 2-propanol (both suprapur) are purchased from Merck GmbH (Germany). Sodium dodecyl sulfate (SDS) and dodecyltrimethylammonium bromide (DTAB) in pro analysis grade were obtained from Sigma Aldrich GmbH (Germany). A Milli-Q A-10 water unit from Millipore AG (Switzerland) was applied for the production of 18 MOhm cm deionized water (DI-water). ICPMS standards were prepared from single element standards in ICP quality from Merck GmbH (Germany). Nanoparticle Samples. All samples were available on the Swiss market and were labeled as nano- or colloidal Ag dispersions. Details concerning producer and application are given in the Supporting Information. Information about specified concentrations and size- are given in Table 1. All samples were stored in light protected glass or PE vessels at room temperature, and DI water was used for dilution of the samples. Analysis was always performed immediately after opening and diluting, and the containers were purged with N2 before restoring. A4F Multidetector Setup. The A4F system (Eclipse3, Wyatt Technology Europe, Germany) was connected to a

product no.



determined size [nm]

Table 1. Producer Specified and Determined Concentration- and Size Values (Particle Mass and Particle Number) for the Investigated Samples

batch-DLS number PSDamode (range)

separation of natural- and engineered nanoparticles, providing a wide dynamic range and analysis times comparable to conventional HPLC, is the so-called Asymmetric Flow Field Flow Fractionation (A4F) technique.26 Since A4F separates over size (hydrodynamic radius), particle dimensions can be derived directly from the theory and retention time calibration with particle standards.27 However, the employment of particle standards or field flow fractionation theory does not account for the influence of particle-membrane interactions on retention time. In consequence, this can lead to erroneous size information.28,29 Thus, especially for unknown samples, the implementation of a size specific detector after A4F separation is recommended. For mass-, element-, and isotopic specific information, combination of A4F with inductively coupled plasma mass spectrometry (ICPMS) can be realized. Several publications describe the use of ICPMS to reveal important information about trace metal behavior in natural colloidal systems.30 Characterization of metallic ENP with A4F in combination with ICPMS was reported already for quantum dots 31,32 and Au nanoparticles (Au-NP).29,33 Also for monodisperse or very well-defined Ag-NP standards and dispersions34−37 the use of A4F coupled to ICPMS is reported. However, size was determined calibrating with particle standards rather than implementing a size specific detector, possibly problematic for unknown and polydisperse samples. The aim of this work was to provide a rapid-, universal-, and in situ technique to characterize polydisperse, commercial available Ag-NP products (number- and mass size distribution and the ratio between the dissolved and particle Ag). For this, we present the application of an A4F multidetector approach coupled to UV/vis, online DLS, and ICPMS. With the setup, the determination of mass and number PSD simultaneously, without the need of either size or mass calibration, is possible in less than 30 min. Additionally, to quantify the toxicological relevant nanoparticle- and dissolved Ag fraction with minimum sample alteration, ICPMS after ultracentrifugation was applied. To verify the obtained results with the proposed setup, as well as to reveal its benefits and limitations, the figures of merit were compared to common used techniques such as TEM and batchDLS.

7 (4−14)

Article

dx.doi.org/10.1021/ac202641d | Anal. Chem. 2012, 84, 2678−2685

Analytical Chemistry

Article

sizes even larger than 150 nanometres were observed. Applying a constant cross-flow, the small nanoparticle fractions in the tens of nanometers were resolved well, but nanoparticles with a size larger than 100 nm did not completely elute even after 30 min. Furthermore, due to the late retention time a significant peak broadening could be observed. With gradient elution a lower resolution power for the smaller particle fractions but separation of a larger size range in shorter time was achieved. The optimized A4F flow parameters for each sample are given in the Supporting Information (Table S-2). To test for reproducibility and reasonable recovery from the A4F channel the influence of Ag-NP interaction with seven different membrane types was investigated (for more specific information about the membrane material see Supporting Information section 5). The dispersion of product 1 was used, since monodisperse Ag-NP standards were not available. Recoveries were calculated by comparing the peak area from a flow injection to the area obtained after A4F separation. Elugrams were recorded with an UV/vis detector, as metallic Ag-NP show a distinctive absorption band (plasmon resonance) in the visible region. Thus detection of possible ionic Ag species in the dispersions, which would falsify recoveries, is prevented. The experiments revealed that several parameters have an influence on the particle recovery such as the numbers of samples already injected on the system. On new PVDF and regenerated cellulose (RC) membranes Ag-NP tend to be immobilized on the surface membrane resulting in a distinct yellow spot at the focus region of the A4F channel. The saturation of the membrane surface led to better recoveries, from 57% for new membranes up to 98% after 6 injections (see Figure S-1b). After saturation of the membrane, 20 to 30 injections could be performed (sample load of 10 to 20 mg/L) without a change in the separation behavior or any observable sample carry over. Membrane fouling could be observed after that by the appearance of ghost peaks in the elugram. For best performance of the system the membrane was changed usually after 25 injections. To prevent particle adsorption in the first place, the addition of surfactants (SDS or DTAB, 200 mg/L) was tested but resulted in a sample loss of over 90% with the employed PVDF and RC channel membranes. In consequence DI water was once more applied as carrier to avoid high salt loads in the ICPMS, possible formation of micelles from the surfactants during A4F separation (which would be detected in the light scattering apparatus), and a change of the charge stabilized samples due to altering the ionic strength. To test for size independent recovery the UV/vis chromatograms of six injections (new membrane up to membrane conditions with 98% recovery) for product 1 were compared (see Figure S-2). The elugrams feature the same characteristics and ratios of peak areas indicating a size independent loss, conserving the PSD. In terms of channel recovery, PVDF and RC in combination with DI water as carrier gave the best results (up to 100%, see Figure S-1a). The negatively charged Ag-NPs behave similar as charged stabilized Au-NP, whose particle-membrane interaction was investigated in more detail in a previous study already.28 The recovery as well as the separation behavior strongly depends on the hydrophobicity of the membrane material. The PVDF as well as the RC membrane (having a hydrophobic layer on top) provide best recoveries for product 1. With decreasing hydrophobicity, also the recovery decrease with full

metal-free HPLC system (DGU-20A3 degasser, LC-10Ai pump, SIL-20AC autosampler, CBM-20A control unit) from Shimadzu Europe GmbH (Germany). A 30 kD polyvinylidendifluoride (PVDF, GE Osmonics Germany, Sepa CF-PVDFUF-JW) channel membrane and prefiltered DI water as carrier was applied. The A4F system was coupled inline to an UV/vis diode array detector (SPD-M20A, Shimadzu GmbH, Germany), followed by an multiangle laser-light scattering (MALLS) detector (DAWN Heleos2, Wyatt Technology Corporation, USA) and finally to a magnetic sector field plasma mass spectrometer (Element2, ThermoScientific, Germany). The MALLS, operating at a laser wavelength of 658 nm, was modified at an angle of 108°, where a dynamic light scattering (DLS) apparatus (NanoStar, Wyatt Technology Corporation, USA) was connected via a glass fiber cord. To record the UV/vis absorption of Ag nanoparticle the scan range was adjusted from 200 to 800 nm with a step size of 1.2 nm and a scan rate of 12.5 Hz. For online introduction of the samples to the ICPMS, the flow passing the UV/vis and MALLS/DLS detector was reduced by a factor 200 using a split interface with an self-aspirating nebulizer, described in detail in a previous work.29 To correct for nonspectral interferences an internal standard (10 μg/L Rh) in 10% nitric acid was added to the remaining flow. Further details concerning ICPMS and A4F conditions are given in the Supporting Information (Tables S-1 and S-2). Ultracentrifugation and ICPMS. Total, dissolved, and particle fractions were quantified as follows: To determine the total Ag concentration, the samples were dissolved in nitric acid (65% w/w) and further diluted with 1% nitric acid (v/v), appropriate for measurements with ICPMS (≈10 μg/L). Clear and noncolored solutions indicated full mineralization of the samples. To determine the dissolved fraction, the Ag-NP dispersions were ultracentrifugated (Centrikon T-2000 with fixed angle rotor TFT 70.38, KONTRON Instruments, Switzerland). After centrifugation for 6 h at 120.000g a calculated cutoff of Ag-NP > 0.5 nm was obtained. After acidification with nitric acid (1% v/v) the supernatant was also measured with ICPMS, and the determined Ag concentration was equated to the dissolved Ag fraction (ionic Ag or subnano Ag atom clusters). The particle fraction corresponds to the difference between the dissolved and the total fraction. Further details about sample handling and calculations of centrifugation time are given in the Supporting Information. TEM and Batch-DLS. For validation of the A4F method, data were compared to TEM analysis and batch-DLS measurements. A more detailed description of sample preparation, size calibration, and applied instruments are given in the Supporting Information.



RESULTS AND DISCUSSION Method Optimization for Ag-NP. For measurement of the Ag-NP products a previously described A4F multidetector setup29 was adapted with adjustments of several separation parameters. At first A4F parameters such as cross-, focus-, and injection flow were optimized for good separation efficiency in less than 30 min. For the cross-flow, defining the resolution power of the A4F separation, two different settings were investigated − a constant cross-flow and a cross-flow gradient. Constant cross flows worked well for monodisperse nanoparticle standards such as Au or polystyrene. For the investigated polydisperse Ag-NP products, next to particles in the size range in the tens of nanometres, also particles with 2680

dx.doi.org/10.1021/ac202641d | Anal. Chem. 2012, 84, 2678−2685

Analytical Chemistry

Article

Figure 1. Multidetector elugrams for a.) product 1, c.) product 2, and d.) product 3. Gray points in the light scattering elugram indicate the obtained hydrodynamic diameter with the online DLS detector. b.) Extracted UV/vis spectra for product 1 at retention time of 6.6 min (black) and 10.1 min (gray).

absorption of Ag-NP on the surface for the nanofiltration (NF) membranes (a yellow spot at the A4F focus zone was already visible after 1 to 2 injections). Additionally, the zeta-potential (ZP) of charge stabilized Ag-NP plays an important role. The same membrane type (PVDF) resulted in a recovery (n = 3) of 102 ± 2% for product 1, having a ZP of −42.5 ± 0.5 mV. Recovery of product 3 showing a lower ZP of −30 ± 1 mV decreased to 72 ± 2%. However, product 2 which features an even lower ZP of −5 ± 1 mV gave comparable recoveries of 75 ± 4%. The low ZP indicates that the particles are stabilized via steric hindering, as also specified by the producer.38 For such samples, particle−membrane interactions cannot be accounted to the ZP alone. In account of the obtained results, particle-membrane interactions are a major problem when analyzing unknown samples with A4F. Optimization of separation conditions for optimal recoveries for one specific sample does not imply that these are useful for all samples. This reveals the limits of the method and clearly demonstrates that direct quantification with the A4F system is challenging, when unknown samples are investigated. In turn determination of the PSD with size specific detectors is possible also without complete recovery. PSD from the A4F Multidetector Setup. In Figure 1a the elugrams for product 1, simultaneously determined with UV/ vis, MALLS at 90°, DLS, and ICPMS are presented. The online DLS was not able to detect particle fractions eluting right at the beginning because of too low sensitivity. The measurable hydrodynamic diameters were in a range between 6 to 160 nm (Ag-NP concentration of 11 mg/L; 50 μL injection volume).

ICPMS detection shows a bimodal particle mass distribution with two distinctive peaks corresponding to a hydrodynamic diameter of 9 and 50 nm. Also the UV/vis signal at 400 nm detected a distribution with the same maxima but of different intensities. Extracting the spectra at 6.6 and 10.1 min from the UV/vis signal (Figure 1b) reveals a shift of the absorption maximum from 420 to 480 nm with increasing size. In consequence, the recorded signal at 400 nm leads to an underestimation of the larger and an overestimation of the smaller nanoparticle fractions. The plasmon resonance of metallic nanoparticles, among other properties, strongly dependents on particle size and shape39 explaining this behavior. Thus, also if another wavelength is extracted, again one of the particle fractions is over- or underestimated due to the size dependent shift of absorption. UV/vis has to be used with care, because it can easily lead to biased and misleading results. Therefore, other and more specific detectors such as ICPMS are preferable for concentration dependent detection. In contrast to the UV/vis and ICPMS signals, the 90° MALLS signal features only one distinct broad peak with a different maximum. Light scattering intensity is not only dependent on particle concentration but also strongly biased by the particle size (scattering intensity is proportional d6). In consequence, the light scattering intensity show different characteristics when compared to the concentration depended UV/vis and ICPMS signal, respectively. Figure 1c show the results for product 2. The light scattering signal shows a broad peak, with the online-DLS able to obtain particle diameters between 5 and 90 nm. However, the signal 2681

dx.doi.org/10.1021/ac202641d | Anal. Chem. 2012, 84, 2678−2685

Analytical Chemistry

Article

Figure 2. Number size distribution obtained with A4F-DLS-ICPMS, TEM, and batch-DLS with corresponding TEM images for product 1 (a,b), product 2 (c,d), and product 3 (e,f).

reveals a constant particle size at an elution time between 12 and 14.5 min, which might be caused by strong particlemembrane interactions. The poor recovery of around 75% from the A4F channel supports this hypothesis. The UV/vis shows the same response as the ICPMS signal, both with a maximum corresponding to a hydrodynamic diameter of 30 nm.

The results for product 3 are presented in Figure 1d. Comparing the UV/vis spectrum to the ICPMS signal, the smaller particles are slightly underestimated and differences also for larger particles intensities can be observed. This fact is attributed again to the size dependent shift in the UV/vis absorption maximum. However, the peak maximum for both signals corresponds to the same hydrodynamic diameter of 90 2682

dx.doi.org/10.1021/ac202641d | Anal. Chem. 2012, 84, 2678−2685

Analytical Chemistry

Article

number distribution from batch-DLS is narrower with diameters between 4 and 14 nm. In Figure 2c the number distributions for the more polydisperse product 2 are presented. The corresponding TEM images (Figure 2d) reveal Ag-NP with a size of around 20 to 60 nm surrounded by numerous small particles in the range of 2 to 15 nm. Again, particles smaller than 6 nm, although observed in the TEM analysis, could not be determined with the online-DLS. However, the peak between 30 and 50 nm, also indicated in the number size distribution derived from TEM images, was clearly pronounced. The number size distribution derived from batch-DLS gave only one broad peak, biased to larger particle diameters. The mode of 15 nm diameter was not in agreement with either TEM or the A4F multidetector setup results, both revealing a mode of 8 nm. Number size distributions for product 3 (Figure 2e) obtained with the A4F multidetector approach are in good agreement with data derived from TEM images. For both a comparable mode of 2 nm (TEM) and 3 nm (A4F) were found. Smaller particles than 3 nm were not evaluable with the online light scattering system. Comparison with the data from the batchDLS reveals a strong bias toward larger dimensions, as a result of the high polydispersity of the investigated sample. The corresponding TEM images (Figure 2f) reveal a span of particle sizes from