Fast Separation, Characterization, and Speciation of Gold and Silver

Article pubs.acs.org/ac

Fast Separation, Characterization, and Speciation of Gold and Silver Nanoparticles and Their Ionic Counterparts with Micellar Electrokinetic Chromatography Coupled to ICP-MS Bastian Franze† and Carsten Engelhard‡,* †

University of Münster, Institute of Inorganic and Analytical Chemistry, Corrensstrasse 28/30, 48149 Münster, Germany University of Siegen, Department of Chemistry & Biology, Adolf-Reichwein-Strasse 2, 57076 Siegen, Germany

‡

S Supporting Information *

ABSTRACT: In this study, a method for separation, size characterization, and speciation of gold and silver nanoparticles was developed through the use of micellar electrokinetic chromatography (MEKC) coupled to inductively coupled plasma-mass spectrometry (ICP-MS) for the first time. Figures of merit in this proof-of-principle study include peak area precision of 4−6%, stable migration times (1.4% with internal standard), and capillary recoveries on the order of 72−100% depending on species and nanoparticle size, respectively. Detection limits are currently in the sub-microgram per liter range. For example, a total of 1500 50-nm-sized gold nanoparticles were successfully detected. After careful optimization, MEKC-ICP-MS was used to separate engineered nanoparticles (ENPs) of different composition. Speciation analysis of ENPs and free metal ions in solution was feasible using a complexing agent (penicillamine). Gold speciation analysis of a dietary supplement, which contained approximately 6-nm-sized gold nanoparticles, was demonstrated.

E

hydrodynamic diameter of particles in suspension. This is a relatively simple and rapid method but results may be skewed when a sample is polydispersed and size overestimation may occur.11 In recent years NP size distribution was successfully performed using inductively coupled plasma mass spectrometry in the so-called single particle detection mode (SP-ICP-MS). Fundamental works were performed by Degueldre and colleagues.12 Recent studies in this field dealt with the influence of different sample introduction techniques on SP-ICP-MS performance13,14 and fundamental considerations with respect to detection efficiencies, detection limits and precision.13−15 The method itself is based on recording a time-resolved spectrum from a diluted nanoparticle suspension. Transient signals, each corresponding to the detection of a single nanoparticle, can be used to calculate mass and hence size of a nanoparticle (assuming solid spheres) with a current minimum detectable particle size in the range of 10−20 nm depending on the species of interest. Additionally, signal frequency offers information about the particle number concentration present in a sample. Recently, an interlaboratory study demonstrated the suitability of this methodology but concluded that method precision needs to be further improved.16 Currently, time resolution of common ICP-MS

ngineered nanoparticles (ENPs) exhibit interesting physical and chemical properties and are used increasingly in many commercial products.1 Prominent ENPs include metal nanoparticles, such as silver and gold nanoparticles (AgNPs and AuNPs, respectively), as well as metal oxide NPs such as TiO2, ZnO, and SiO2. Applications cover healthcare, electronics, and consumer goods, which exploit the unique properties of ENPs (e.g., the antimicrobial activity of AgNPs or the optical properties of TiO2 and ZnO NPs).2,3 Clearly, a general increase in use of ENPs will inevitably raise exposure to the environment and organisms. For example, exposure models predict concentrations in surface waters of 30 ng L−1 to 10 μg L−1, 3 ng L−1 to 10 μg L−1, and 1 ng L−1 to 60 ng L−1 for AgNPs, nanoparticulate TiO2, and ZnO, respectively.4−6 Unfortunately, quantifying the amounts of ENPs produced and incorporated in products has been proven to be difficult in the past.7 Sensitive and rapid methods for nanoparticle characterization in complex matrices are required for exposure, biodistribution, and toxicity studies. Ideally, such methods would also be able to distinguish between nanoparticulate and ionic forms of the respective element.8,9 Typically, size and morphology of nanoparticles are determined with microscopy-based techniques (e.g., scanning/transmission electron microscopy, SEM and TEM). These methods provide high spatial resolution, but accuracy is dependent on the number of particles used for mean size calculation. Also, artifacts can be generated during sample preparation and at low NP concentration, respectively.10,11 Dynamic light scattering (DLS) is applied to characterize the © 2014 American Chemical Society

Received: December 10, 2013 Accepted: May 22, 2014 Published: May 22, 2014 5713

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720

Analytical Chemistry

Article

(GE) coupled to ICP-MS for Au nanoparticle characterization.27 In a later study, the method was used for the determination of Au/S-ratios in gold nanoparticles covered by mercaptosuccinic acid.27,33 Baker et al. demonstrated that MEKC coupled to ICP-MS is suitable for elemental analysis of cobalamin species.34 In the current study, MEKC-ICP-MS is introduced for nanoparticle size characterization for the first time. On the basis of the studies conducted by Liu et al., a MEKC method was developed to separate polydispersed nanoparticle suspensions and to detect ionic species and nanoparticles in the same run. This method extends the family of ICP-MS-based nanoparticle characterization tools. Strengths include the relatively high resolution compared to published separation protocols, as well as the possibility to perform speciation analysis.

quadrupole mass analyzers is, besides the minimum detectable particle size, another limitation preventing the analysis of heterogeneous particles samples. This shortcoming could be overcome with the use of simultaneous multicollector or timeof-flight instrumentation.17,18 In addition, it was demonstrated by several groups that SP-ICP-MS could be used, in principle, for speciation analysis of NP and free metal ions in solution. Currently, this is feasible for selected NP size and concentration ranges, respectively. A limitation does exist if particles are present in the sample with sizes smaller than the method particle size detection limit. Those particles might easily be misinterpreted as ionic species (or even background). This shortcoming could be overcome with the use of separation techniques. Separation techniques coupled to ICP-MS provide added dimensionality, which also enables smaller particles to be detected at concentrations in the nanogram per liter range. For the analysis of polydispersed suspensions, field-flow fractionation (FFF), size exclusion or hydrodynamic chromatography (SEC or HDC, respectively) were recently applied. For example, AgNPs in biological media, AuNPs in tissue, as well as nanoparticulate TiO2 in sunscreen were successfully characterized using FFF-ICP-MS.19−21 With FFF, however, recovery rates were reported between 50% and 95%. In addition, simultaneous determination of ENPs and ionic counterparts is currently not possible. Total analysis time is typically more than 25 min.22 Alternative separation approaches were studied as well. For example, Tiede et al.23 demonstrated the possibility to analyze environmental samples and sewage sludge with HDC-ICP-MS, but with limited separation power. In a proof-of-principle study published in 2012, Pergantis et al.24 demonstrated the benefits of transient detection in HDCSP-ICP-MS and obtained 2D contour plots with particle number concentration for AuNPs. Later, Rakcheev et al.25 used HDC-SP-ICP-MS to separate primary AuNPs from agglomerated particles. Size exclusion chromatography was also applied to separate particles and ions in solution. Here, Wei et al.26 demonstrated that sorption of particles on the packing material can be minimized by adding a surfactant (SDS) to the mobile phase. This approach was also used by Helfrich et al.27 to separate AuNPs mixtures (5−20 nm), but with limited separation power. In the same study, gel electrophoresis (GE) was used as an alternative and improved peak separation was obtained. Several groups reported the fundamental possibility to differentiate metal ions in solution and NPs by means of single particle detection or chromatography. For example, SotoAlvaredo et al. developed a silver speciation method using reversed-phase LC-ICP-MS, which was then applied to determine free Ag(I) and AgNPs in extracts from sport socks.26,28 More recently, Fabricius et al. compared different sample preparation procedures for nanoparticle characterization by ICP-MS analysis. They identified centrifugal ultrafiltration as a relatively easy tool for the separation of the dissolved fraction from the particle fraction in suspension.29 Electrophoretic methods exhibit great promise for nanoparticle characterization. Already in 1991, McCormick et al.30 achieved good resolving power for differently sized silica particles in the range of 5−500 nm. Liu et al.31,32 studied different buffers and revealed the potential of MEKC for the analysis of ENPs. Because CE with UV/vis detection was used, sample concentrations were in the milligram per liter range. As mentioned above, Helfrich et al. applied gel electrophoresis

■

EXPERIMENTAL SECTION Sample Preparation. A series of spherical silver and gold nanoparticles with nominal diameters of 7.2 (±1.2), 30.7 (±3.6) (AgNPs), and 4.8 (±0.6), 9.1 (±0.8), 20 (±1.8), 48.1 (±4.2) nm (AuNPs, first standard deviation in brackets), respectively, were obtained from nanoComposix (San Diego, CA, USA). In addition, a certified AuNP reference material (RM 8012, NIST, Gaithersburg, USA) with nominal diameter of 27.6 (±2.1) nm was used. Nanoparticle suspensions were stabilized in tannic acid (AuNPs) and citrate (AgNPs) at concentrations of 0.05 and 0.02 mg mL−1, respectively. All samples were stored in darkness at 2 °C prior to analysis. Before diluting to the desired concentration, suspensions were shaken for 30 s. All chemicals were obtained from SigmaAldrich (Steinheim, Germany) unless otherwise stated. Internal standardization was performed by adding 10 mg L−1 sodium diatrizoate to each sample. Buffer solution for CE contained sodium dodecyl sulfate (SDS) and N-cyclohexyl-3-aminopropanesulfonic acid (CAPS). Additionally, 10 μg L−1 Cs was added to monitor electroosmotic flow (EOF), which is a measure for stable analyte transport from the CE capillary to the ICP-MS. A solution of 60 mM SDS and 10 mM CAPS at pH of 10 provided best separation conditions with respect to resolution, peak shape, and migration time. Solution pH was adjusted with a pH-meter (CALIMATIC 761, Knick GmbH & Co. KG, Berlin, Germany) and 1 M NaOH. In some experiments (cf., speciation analysis), 50 μM penicillamine was added to the sample prior dilution with buffer. Sheath liquid contained 10 μg L−1 In and 2% HNO3 (CertiPUR, Merck KGaA, Darmstadt, Germany). All buffers were filtered through a 0.2 μm cellulose-acetate syringe filter (VWR International GmbH, Darmstadt, Germany). Aqueous Au+, In+, and tuning solutions were prepared by diluting standard stock solutions of 1000 mg L−1 (In, Cs, Au) in 2−3% HNO3 with double-distilled water (AQUATRON A4000D, Bibby Scientific, Stone, UK) to the desired final concentration. Instrumentation. CE-ICP-MS. A capillary electrophoresis system (HP G1600AX, Agilent Technologies, Germany) with a polyimide-coated fused-silica capillary (70 cm long, 75-μm i.d., and 375-μm o.d.; Polymicro Technologies, Phoenix, AZ, USA) was coupled to a model 7500ce ICP-MS (Agilent Technologies, Santa Clara, CA, USA). A standard torch with a 1.5 mm wide injector was used. The liquid-introduction interface was based on a model DS-5 nebulizer (CETAC Technologies, Omaha, NE, USA) equipped with a low-volume spray chamber and a Tpiece to merge sheath liquid and capillary flow. Electrical circuit 5714

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720

Analytical Chemistry

Article

of the CE was completed via a grounded platinum wire in the sheath liquid container. A detailed schematic of the interface is provided in the Supporting Information, Figure S1. Experimental parameters of the CE-ICP-MS setup are summarized in Table 1. Table 1. Instrumental Parameters of MEKC-ICP-MS ICP-MS system instrument RF power (W) cooling gas Ar (L min−1) auxiliary gas Ar (L min−1) sample carrier gas Ar (L min−1) sampling depth (mm) CE system voltage (kV) current (μA) hydrodynamic injection electrokinetic injection

Agilent 7500ce 1500 15 1 0.8 7 ∼29 45 50 mbar for 3 s 20 kV for 8 s

Figure 1. Representative electropherograms for mixtures of ∼5, ∼20, and ∼50-nm-sized AuNPs obtained with CE/MEKC-ICP-MS (monitoring m/z 197) and increasing SDS concentrations in the buffer. (A) no SDS, (B) 20 mM SDS, (C) 40 mM SDS, and (D) 60 mM SDS. See Table 1 and Experimental Section for CE/MEKC-ICPMS conditions.

Before use, the capillary was rinsed for 10 min each with 1 M NaOH, double-distilled water, and running buffer. Between runs the capillary was rinsed with running buffer for 1 min. Further, the position of the capillary was optimized to achieve a low dead-volume. Here, 115In, present in sheath liquid, and 133 Cs, added to the separation buffer, were monitored while flushing the system. After optimization, a steep rising edge of 133 Cs signal and a sharp falling edge for the 115In signal (both on the order of 5−10 s until equilibrium) were obtained. Sample injection was carried out hydrodynamically with 50 mbar for 3 s (corresponds to approximately 16.5 nL injection volume). Analytical figures of merit were compared with those obtained by electrokinetic injection (20 kV, 8 s). Maximum applied current was 45 μA, which resulted in a voltage of approximately 29 kV. Electropherograms were recorded by ChemStation software (B.03.04, Agilent Technologies) and data evaluation was performed with Origin 8.5 (OriginLab Corporation, Northampton, MA, U.S.A.).

AuNPs seems not to be high enough for a resolved electropherogram. Adding a surfactant to the mobile phase changes the surface chemistry of the particles and, in turn, the separation mechanism to that of micellar electrokinetic chromatography (MEKC). Here, SDS was used as surfactant, which features an anionic character due to the sulfate headgroup. Monomers and SDS micelles show an electrophoretic mobility that is counter to the direction of a strong EOF (basic conditions) with a net movement toward the capillary outlet (cathode). In Figure 1b−d, electropherograms of 5, 20, and 50 nm AuNPs are presented with increasing SDS concentration, which clearly improves the separation capabilities. For example, 5 and 20-nmsized particles could not be separated at 20 mM SDS. In contrast, almost baseline separation of 5, 20, and 50 nm AuNPs is successfully achieved using 60 mM SDS. Total separation time is only 10 min. Consequently, the experiments described below were all performed with 60 mM SDS unless otherwise stated. It was found that higher SDS concentrations enhanced the electrophoretic mobility under the conditions used here. Charge-to-size ratio of the AuNPs is proportional to the number of SDS molecules on the surface and a higher electrophoretic mobility of large AuNPs is the result. In contrast, smaller particles elute earlier due to the smaller surface area and lower charge-to-size ratio. Thus, elution order is reversed compared to SDS-free conditions (cf., Figure 1a). Liu et al.32 reported similar elution orders for 5 and 19-nm-sized AuNPs and suggested that the surfactant in the buffer associates with AuNPs to form a protective layer. Further, it was suggested that the total charge of a nanoparticle is linked and limited to the number of SDS molecules, which are available for adsorption onto the particle surface. Thus, at a sufficiently high SDS concentration, SDS molecules saturate the surface of larger particles and result in a higher NP charge-to-size ratio and

■

RESULTS AND DISCUSSION Optimization of CE/MEKC Separation. Nanoparticle charge can be described by the ζ-potential, which is a measure of the effective charge on the surface of the nanoparticle. AuNPs are usually coated, e.g., with tannic acid, to achieve high colloidal stability by electrostatic and steric stabilization. This coating results in a negative ζ−potential and prevents agglomeration.35 Thus, electrophoretic separation of NPs should be feasible. In the current study, initial CE experiments were carried out but without the addition of surfactants or complexing agent to the running buffer (10 mM CAPS, 10 μg L−1 Cs, pH = 10). Under these conditions, a separation of 5 and 20-nm-sized AuNPs could be achieved (cf. Figure 1a). However, resolution was too poor to adequately resolve 20 nm particles from those that were 50 nm in size. As the electrophoretic mobility is proportional to the charge-to-size ratio of the analyte, mobility of 5-nm-sized AuNPs is significantly larger than that of the 20 and 50-nm-sized AuNPs. Smaller particles migrate faster counter the EOF direction and, therefore, elute at a later time compared to larger particles. Under these CE conditions the difference in charge-to-size ratio of 20 and 50-nm-sized 5715

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720

Analytical Chemistry

Article

chemistry (see below). Further studies are planned to investigate whether this approach can be transferred to other surface chemistries and environmental samples. In Figure 3a, the influence of nanoparticle size (5−50 nm AuNP) on migration time is depicted. In this experiment, the

electrophoretic mobility. An increase in SDS concentration will also increase the ionic strength, which, in turn, will reduce EOF velocity. Presumably, this is why the continuous delay in migration time is observed (cf., Figure 1b and d). Once a MEKC-ICP-MS method is optimized, retention time matching can be used to assign NP sizes but it is critical to provide stable separation conditions. Preliminary data (not shown here) suggest that a prerequisite for maintaining resolution (and to prevent misalignment of NP sizes) is to guarantee constant separation conditions (e.g., buffer composition, temperature, applied voltage, and capillary length). Analytical Figures of Merit of MEKC-ICP-MS. Exemplarily, Figure 2 shows an electropherogram obtained under

Figure 2. MEKC-ICP-MS electropherogram of ∼5, ∼20, and ∼50-nmsized AuNP (black, 197Au) obtained after method optimization. 133Cs signal (red) indicates stable current and spray conditions. Iodinecontaining diatrizoate (green, 127I) was used as internal standard to improve migration time precision.

optimal conditions for the separation of a mixture of 5, 20, and 50 nm AuNPs. Nanoparticles from the sample mixture elute at 415 (5 nm), 450 (20 nm), and 530 s (50 nm). Elution order was confirmed by individual measurements of the differently sized AuNPs (data not shown here). Stability of the CE flow was monitored using the 133Cs signal, which increases at the beginning of the experiment and levels off to a relatively constant value throughout the experiment. Relative standard deviations (RSDs) of 197Au peak areas of 4−6% for hydrodynamic injection and 7−10% for electrokinetic injection, respectively, were obtained (n = 3). To improve migration time precision, an internal standard (IS) correction approach was used. Here, iodine-containing sodium diatrizoate, which did not affect AuNP migration time, was successfully applied. In Figure 2, the first transient signal appears at 250 s and stems from the IS monitored at 127I. Diatrizoate elutes earlier from the capillary compared to AuNP because incorporation into the micellar phase is less likely and, thus, migration counter the EOF is reduced. After IS correction, RSD values for migration times ranged from 0.4% to 1.4%. Results indicate that an acceptable reproducibility of AuNP migration is achievable in MEKC. Clearly, successful external size calibration with nanoparticle standard suspensions is dependent on the ability of SDS molecules to replace the surface chemistry already present in the sample suspension. With tannic acid and citrate, which both can be considered easily replaceable capping agents, migration time was found to be independent from the NP surface

Figure 3. Influence of nanoparticle size and concentration on MEKCICP-MS response (197Au).(A) Linear weighted migration time calibration curve obtained from ∼5, ∼10, ∼20, and ∼50-nm-sized AuNP (black). Note that a NIST AuNP standard (red) was characterized using this calibration curve (see Results and Discussion section for details). Y-error bars indicate RSDs of migration time from six consecutive measurements and x-error bars indicate the AuNP size distribution (provided by the manufacturer). Ratio of NP migration time (MTNP) versus internal standard (IS) migration time (MTIS) was used to give an IS-corrected migration time. IS-corrected migration time is linearly correlated with nanoparticle size (R2 > 0.999, y = 0.0114x + 1.6532). (B) Peak area as a function of injected concentration and mass for ∼5 (black), ∼20 (red), and ∼50 nm (green) AuNP.

ratio of NP migration time (MTNP) versus IS migration time (MTIS) was used to give an IS-corrected migration time. It was found that the IS-corrected migration time is linearly correlated with nanoparticle size (R2 > 0.999, y = 0.0114x + 1.6532). Because a linear relationship exists, electrophoretic mobility of AuNP can be considered to be proportional to AuNP radius, which was already theoretically described in a simplified way as follows. Details can be found elsewhere.32 Briefly, the number of associated SDS molecules is proportional to the surface area of the nanoparticle, ANP, and can be expressed as no. SDS molecules ∝ ANP = 4πr 2 5716

(1)

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720

Analytical Chemistry

Article

where r is spherical particle radius. Electrophoretic mobility for spherical particles, μ, is the sum of physical factors that affect particle migration during electrophoresis and given by μ=

q 6πηr

Table 2. Capillary Recoveries and Detection Limits

(2) a

where q is net charge of the particle and η is viscosity of the surrounding medium. The net charge of a particle is proportional to the number of SDS molecules attached to the NP surface. Because the number of SDS molecules is considered to be proportional to surface area (see eq 1), the net charge can be expressed as a function of surface area. Thus, electrophoretic mobility and spherical particle radius are proportional given by μ∝

4πr 2 ∝r 6πηr

species

detection limit (μg L−1)a

Au(I) 5 nm AuNPs 20 nm AuNPs 50 nm AuNPs

0.27 0.18 0.22 0.55

capillary recovery (%)b 102 75.5 80 72.3

± ± ± ±

8.7 14.1 5.9 12.2

3σ criterion. bN = 3.

principle study are already quite attractive compared to other methods (800 fg−500 pg) cited above. Separation techniques are often accompanied by analyte mass loss. For example, membrane sorption and permeation is of great concern in FFF and can lead to sample recoveries between 50% and 100% (see the introductory section). In the current study, mass loss and sample recoveries in MEKC were characterized via offline collection of specific species fractions and ICP-MS analysis. Species-specific external calibration revealed recoveries of approximately 72−80% for nanoparticulate species and approximately 102% for their ionic counterparts (cf., Table 2 for details). Note that the figures of merit for ionic species were determined after the addition of a complexing agent to the sample suspension, which is discussed in detail below. Speciation of Nanoparticles and Corresponding Ionic Species. As discussed above, the developed method worked well for the separation of AuNPs mixtures. Interestingly, this changed dramatically with a loss in resolution and the occurrence of major peak broadening when an ionic gold solution (250 μg L−1) was spiked to a sample. In this case, proper separation of ions and nanoparticles was not feasible (cf., Figure 4a). Presumably, gold ions associate with the sulfate group of SDS as well, and neutralize/reduce the negative charge on the surface of the differently sized NP, which, in turn, decreases electrophoretic mobility and prevents a successful separation. To overcome this limitation, penicillamine was used to complex ionic gold in solution. The complexing agent (50 μM) was added to the sample before analysis. Penicillamine is an αamino acid and was used in the past as complexing agent for gold and other heavy metal ions.39 According to Pearson’s acid base concept, the thiol group binds to the soft gold ion to form a stable complex. First results indicate that this approach is promising in MEKC-ICP-MS. Separation and speciation of AuNPs and gold ions from a mixture was successfully achieved (cf., Figure 4b). After addition of a gold standard solution to the AuNP suspension, one additional peak with a migration time of 360 s appeared, which was verified to be the penicillamine−gold complex. Migration of the complex is influenced by differential partitioning between micelles and buffer. A relatively short migration time indicates that the complex is in a rather charged state because neutral species would elute later. In general, penicillamine could also interact with the AuNP surface. In the current experiment, only small changes in NP migration times were observed compared to penicillamine-free conditions but could be an indication for this effect. The possibility to detect free ions and ENPs simultaneously is considered a very attractive feature of MEKC-ICP-MS. Both elemental speciation and NP size analysis can be achieved in a quasi-simultaneous measurement. Consequently, this method could help further determine the fate of ENPs, especially for

(3)

As discussed above, experimentally obtained nanoparticle mean sizes from MEKC-ICP-MS were found to be correlated with TEM data provided by the manufacturer (NanoComposix). In principle, this method should then be applicable to characterize unknown AuNP materials after external calibration using commercial standards. Consequently, the calibrated system was tested using a certified reference material (AuNP, NIST RM8012, cf., Figure 3a). Analysis with MEKC-ICP-MS resulted in a mean size of 25.7 (±2.3) nm. This is in good agreement with the particle mean size of 27.6 ± 2.1 nm given by the NIST certificate. In principle, it should be feasible to obtain information on the particle size distribution in a given NP ensemble based on peak broadening in MEKC. It is noteworthy, however, that other factors besides the size distribution in the original sample may contribute to peak broadening and are known from classical CE and MEKC. For example, Joule heating is a common shortcoming in CE separations and leads to diffusion inside the capillary. Future work is necessary to study the method-based effects on the NP size distribution in a quantitative fashion. In Figure 3b, calibration curves for different concentrations (1.4−41.6 pg total injected mass) of 5-, 20-, and 50-nm-sized AuNP are presented. A linear response was obtained with correlation coefficients better than 0.99 in all three cases. Interestingly, the sensitivity of the method is slightly different for 5 nm AuNPs compared to 20 and 50 nm AuNPs. Recently published studies on particles in the ICP suggest that sizedependent changes in sensitivity could be due to vaporization and diffusion effects, respectively, in the ICP. Critical upper size limits were reported to be 150 nm36 for AuNP (studied size range 50−200 nm), where incomplete vaporization of particles started to play an important role. Another study reports similar effects but was based on larger particles (>400 nm).37 In the current study, particles were smaller than those reported values, and no studies are reported on fundamentals in a coupled system such as MEKC-ICP-MS. Therefore, future work is required to address the sensitivity changes observed in Figure 3b. Detection limits and sample recoveries are listed in Table 2. Detection limits were in the submicrogram-per-liter range and comparable to FFF and SEC when coupled to ICP-MS.28,29,38 Because the total injected volume in MEKC was in the nanoliter range, this corresponds to low femtograms of total detected mass. Therefore, detection limits of this proof-of5717

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720

Analytical Chemistry

Article

Figure 5. Separation and detection of a mixture of AuNPs, AgNPs, and Ag+ in solution using MEKC-ICP-MS. An offset was added to the m/z trace 197 for illustrative purposes. Note the later migration time of 30 nm labeled AgNPs versus 30 nm AuNPs. This is explained by an increase in size (from ∼30 to ∼50−60 nm) because of agglomeration/ aging on the shelf.

AgNPs as well. Investigation of method adaption to other inorganic NPs typically used, such as Ag, Cd, Ti, and Zn, will be subject of future research. The results demonstrate that this method is suitable for both size characterization and quantification of samples with unknown NP size distribution. Additionally, it is possible to separate similar sized nanoparticles with differences in chemical composition. First results also indicate that easily replaceable surface coatings do not skew migration time, which is important when using external size calibration with nanoparticle standard suspensions. Further investigations with more complex surface chemistries are planned in the future. Analysis of Gold Species in a Nutritional Supplement. Colloidal gold preparations are marketed (mainly via the Internet) as “nutritional supplements” and “universal remedy,” respectively. For this study, a supplement (MesoGold) was purchased from an online distributor,40 which claims that it is “99.99% pure gold particles” suspended in deionized water without any free gold ions. The product is advertised, for example, “to improve motor skills for better sports performance,” and 5−15 mL are recommended to be taken four times daily. Gold speciation analysis on the sample was performed with MEKC-ICP-MS. Sample preparation was straightforward and consisted only of sample dilution with separation buffer (with 50 μM penicillamine) followed by CE injection. The obtained electropherogram produced a large peak at a migration time of 460 s, which corresponds to a mean particle size of approximately 6 nm (cf., Figure 6a). This result is in good agreement with the mean size stated by the manufacturer (3.2 nm, DLS analysis40). A closer look (cf., Figure 6b) reveals an additional small peak at 360 s. Because similar migration times were obtained previously for the goldpenicillamine complex (cf. Figure 4b), it can be concluded qualitatively that ionic gold is present in the sample. This is important because this “nutritional supplement” likely contains gold ions, which are known to cause toxic effects in living organisms.41 As the manufacturer claims that no other chemicals besides colloidal gold and deionized water are present, it is likely that particles undergo dissolution because no stabilizing agent is present. A quantitative study was beyond the

Figure 4. Separation of ∼10 and ∼30-nm-sized AuNPs and Au+ in solution (A) without and (B) with a complexing agent present in the buffer solution.

risk assessment and dissolution behavior of ENPs under environmental and biological conditions. Simultaneous Detection of Au and Ag. To test the suitability of the method for more complex samples, a mixture of gold and silver nanoparticles different in size and with different surface chemistries (10 and 30 nm AuNPs; 10 and 30 nm AgNPs, citrate-capped) was analyzed with MEKC-ICP-MS. Figure 5 depicts a successful separation of both species (107Ag and 197Au). For illustrative purposes an offset was added to the m/z trace 197. First, a complex of penicillamine and Ag+ elutes from the capillary. Second, comigration of both 10-nm-sized AuNPs and AgNPs was observed, which indicates a similar charge-to-size ratio for both species. Similarly, 30-nm-sized AuNPs and AgNPs were expected to comigrate, however, the 30 nm AgNPs eluted later from the capillary. The nominal size information from the manufacturer was therefore questioned with in-house TEM analysis. Indeed, a TEM experiment of the original sample suspension revealed that these 30 nm labeled AgNPs underwent agglomeration and now exhibited a mean size distribution of 50−60 nm. Any agglomeration behavior in the sample prior or during analysis can be excluded because repetitive analyses were performed for all NP sizes. If agglomeration would occur, a migration time shift for the specific size would have been observed. It is assumed that agglomeration was a result of aging on the shelf since these particles were handled for over 2 years. In general, it is noteworthy that even if species were to comigrate, the extracted ion electropherogram could be used to differentiate nanoparticles based on composition. Hence, the method is not limited to AuNPs, but can also be used to study 5718

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720

Analytical Chemistry

Article

important in a follow-up study to check whether the nanoparticle surface chemistry influences migration times. Changing surface chemistries may induce migration time shifts. Ultimately, this would lead to miss-assigned nanoparticle sizes and is of concern in ENP analysis especially in environmental and unknown samples.

■

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. The material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +49 (0) 271 740 2041. Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript. The authors declare no competing financial interest. Notes

Figure 6. Analysis of a dietary supplement containing AuNP. (A) Electropherogram of the sample reveals a mean Au particle size of ∼6 nm. (B) Zoom-in presumably indicates ionic gold in the suspension.

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS Partial financial support from the German Research Foundation (DFG, grant EN 927/1-1) and the Fonds der Chemischen Industrie (FCI) is gratefully acknowledged. We thank Dr. Daniel Pröfrock (Helmholtz-Zentrum Geesthacht, Germany) for support with the CE interface, Jörg Roscher (Institute of Inorganic and Analytical Chemistry, IAAC, University of Muenster) for discussions concerning CE, and Michael Epping (Institute of Physics, University of Muenster) for performing TEM measurements. Technical support by the machine shop and the glassblowers at IAAC is greatly appreciated. Finally, we thank three anonymous reviewers for their comments and suggestions, which have helped to improve the quality of the manuscript.

scope of this work but the applicability of MEKC-ICP-MS to real world samples was demonstrated.

■

CONCLUSIONS In this study, we identify MEKC-ICP-MS as a promising analytical tool for nanoparticle research. Separation, characterization, and speciation analysis of polydisperse gold nanoparticles (5−50 nm) and ionic gold, respectively, was performed. The key to a successful separation is the addition of a surfactant (SDS) to the running buffer. Surfactant molecules attach to the surface of the NPs and determine its charge-to-size ratio. As a result, small particles elute prior to large particles with the migration time being correlated to nanoparticle size. Migration time precision was improved to 1.4% by use of an internal standard. Capillary recoveries were on the order of 72% to 100% (depending on the species) and detection limits were in the sub-microgram per liter range. In addition, capabilities of MEKC-ICP-MS for speciation of nanoparticles and ionic species were evaluated using a complexing agent (penicillamine). Gold speciation analysis was then performed on a real world sample (nutritional supplement) that contained AuNP as well as ionic gold. Analytical figures of merit in this proof-of-principle study are promising. Most importantly, a calibrated MEKC-ICP-MS method would allow the user to size and quantify nanoparticles (within the current size and detection limits) in a suspension in less than 10 min. Future applications could include NP agglomeration/dissolution studies, process control, and environmental analysis. Clearly, several current limitations exist that should be addressed in future studies. One example is the accessible NP size range. In principle it should be feasible to separate particles with sizes covering the whole nanoscale (1−100 nm). However, extensive peak broadening for larger particles with a broad size distribution is a common problem in particle separation techniques and would also deteriorate detection limits and analysis time after MEKC separation. Second, it will be

■

REFERENCES

(1) The Project on Emerging NanotechnologiesConsumer Products Inventory. http://www.nanotechproject.org/cpi/products/ (accessed Oct 22, 2013). (2) Chen, X.; Schluesener, H. J. Toxicol. Lett. 2008, 176, 1−12. (3) Nohynek, G.; Dufour, E. Arch. Toxicol. 2012, 86, 1063−1075. (4) Gottschalk, F.; Sonderer, T.; Scholz, R. W.; Nowack, B. Environ. Sci. Technol. 2009, 43, 9216−9222. (5) Gottschalk, F.; Sun, T.; Nowack, B. Environ. Pollut. 2013, 181, 287−300. (6) Maurer-Jones, M. A.; Gunsolus, I. L.; Murphy, C. J.; Haynes, C. L. Anal. Chem. 2013, 85, 3036−3049. (7) Piccinno, F.; Gottschalk, F.; Seeger, S.; Nowack, B. J. Nanopart. Res. 2012, 14, 1−11. (8) Review of Environmental Legislation for the Regulatory Control of Nanomaterials. http://ec.europa.eu/environment/chemicals/ nanotech/pdf/review_legislation.pdf (accessed Oct 14, 2013). (9) Howard, A. J. Environ. Monit. 2010, 12, 135−142. (10) Mavrocordatos, D.; Perret, D.; Leppard, G. G. In Environmental Colloids and Particles; John Wiley & Sons, Ltd.: Chichester, U.K., 2007; pp 345−404. (11) Hassellov, M.; Readman, J.; Ranville, J.; Tiede, K. Ecotoxicology 2008, 17, 344−361. (12) Degueldre, C.; Favarger, P. Colloids Surf., A 2003, 217, 137− 142.

5719

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720

Analytical Chemistry

Article

(13) Franze, B.; Strenge, I.; Engelhard, C. J. Anal. At. Spectrom. 2012, 27, 1074−1083. (14) Gschwind, S.; Flamigni, L.; Koch, J.; Borovinskaya, O.; Groh, S.; Niemax, K.; Gunther, D. J. Anal. At. Spectrom. 2011, 26, 1166−1174. (15) Laborda, F.; Jimenez-Lamana, J.; Bolea, E.; Castillo, J. R. J. Anal. At. Spectrom. 2011, 26, 1362−1371. (16) Linsinger, T. J.; Peters, R.; Weigel, S. Anal. Bioanal. Chem. 2013, 1−9. (17) Engelhard, C. Anal. Bioanal. Chem. 2011, 399, 213−219. (18) Borovinskaya, O.; Hattendorf, B.; Tanner, M.; Gschwind, S.; Gunther, D. J. Anal. At. Spectrom. 2013, 28, 226−233. (19) Römer, I.; White, T. A.; Baalousha, M.; Chipman, K.; Viant, M. R.; Lead, J. R. J. Chromatogr. A 2011, 1218, 4226−4233. (20) Poda, A. R.; Bednar, A. J.; Kennedy, A. J.; Harmon, A.; Hull, M.; Mitrano, D. M.; Ranville, J. F.; Steevens, J. J. Chromatogr. A 2011, 1218, 4219−4225. (21) Nischwitz, V.; Goenaga-Infante, H. J. Anal. At. Spectrom. 2012, 27, 1084−1092. (22) Schmidt, B.; Loeschner, K.; Hadrup, N.; Mortensen, A.; Sloth, J. J.; Bender Koch, C.; Larsen, E. H. Anal. Chem. 2011, 83, 2461−2468. (23) Tiede, K.; Boxall, A. B. A.; Tiede, D.; Tear, S. P.; David, H.; Lewis, J. J. Anal. At. Spectrom. 2009, 24, 964−972. (24) Pergantis, S. A.; Jones-Lepp, T. L.; Heithmar, E. M. Anal. Chem. 2012, 84, 6454−6462. (25) Rakcheev, D.; Philippe, A.; Schaumann, G. E. Anal. Chem. 2013, 85, 10643−10647. (26) Wei, G.-T.; Liu, F.-K. J. Chromatogr. A 1999, 836, 253−260. (27) Helfrich, A.; Bruchert, W.; Bettmer, J. J. Anal. At. Spectrom. 2006, 21, 431−434. (28) Soto-Alvaredo, J.; Montes-Bayón, M.; Bettmer, J. Anal. Chem. 2013, 85, 1316−1321. (29) Fabricius, A.-L.; Duester, L.; Meermann, B.; Ternes, T. Anal. Bioanal. Chem. 2014, 406, 467−479. (30) McCormick, R. M. J. Liq. Chromatogr. 1991, 14, 939−952. (31) Liu, F.-K.; Ko, F.-H.; Huang, P.-W.; Wu, C.-H.; Chu, T.-C. J. Chromatogr. A 2005, 1062, 139−145. (32) Liu, F.-K.; Wei, G.-T. Anal. Chim. Acta 2004, 510, 77−83. (33) Helfrich, A.; Bettmer, J. Int. J. Mass spectrom. 2011, 307, 92−98. (34) Baker, S. A.; Miller-Ihli, N. J. Spectrochim. Acta, Part B 2000, 55, 1823−1832. (35) nanoComposixStandard Capping Agents. http://www. nanocomposix.eu/sites/default/files/Standard%20Capping%20Agents. pdf (accessed Jan 24, 2014). (36) Ho, K.-S.; Lui, K.-O.; Lee, K.-H.; Chan, W.-T. Spectrochim. Acta, Part B 2013, 89, 30−39. (37) Olesik, J. W.; Gray, P. J. J. Anal. At. Spectrom. 2012, 27, 1143− 1155. (38) Mitrano, D. M.; Barber, A.; Bednar, A.; Westerhoff, P.; Higgins, C. P.; Ranville, J. F. J. Anal. At. Spectrom. 2012, 27, 1131−1142. (39) Eyring, E. J.; Engleman, E. P. Arthritis Rheum. 1963, 6, 216−223. (40) Nutritional Supplement MesoGoldDetails and Size Distribution Report. http://www.purestcolloids.com/mesogold.php (accessed Apr 1, 2014). (41) Macleod, J. G. Ann. Rheum. Dis. 1948, 7, 143−151.

5720

dx.doi.org/10.1021/ac403998e | Anal. Chem. 2014, 86, 5713−5720