Characterization of Nanoparticles in Mixtures by Taylor Dispersion Analysis Hyphenated to Inductively Coupled Plasma Mass Spectrometry

A novel methodology for investigating the behavior of nanoparticles in their mixtures in aqueous high-ionic strength conditions is presented in this work. Our approach utilizes Taylor dispersion analysis in capillaries connected to inductively coupled plasma mass spectrometry (ICP-MS) to probe metal-derived nanoparticles. This methodology simultaneously distinguishes between different kinds of nanoparticles and accurately determines their essential parameters, such as hydrodynamic size, diffusion coefficient, and elemental composition. Moreover, the isotope-specific ICP-MS detection allows for unique targeting of the fate of isotopically enriched nanoparticles. The complexity of our methodology opens the way for studying barely explored areas of interparticle interactions or unequivocal characterization of one type of nanoparticle in complex mixtures without any need for calibration as well as labor-consuming sample preparation.


■ INTRODUCTION
The last few decades have witnessed tremendous progress in the design of metallic nanoparticles (NPs) and ultrasmall NPs for various applications, such as medicine, sensing, biotechnology, water and sewage treatment, construction, paints, cosmetics, electronics, and energy storage. 1,2Interestingly, our understanding of the self-assembly process of NPs and their interactions with other NPs is still quite limited.It is known that interparticle forces among building blocks play a vital role in constructing assembled superstructures; however, the evolution of this process has not been explored yet. 3Moreover, introducing micro-and nanosystems into complex real-life conditions, such as living bodies or the environment, could have other important effects on their overall assembled structure.It appears that NP−NP interactions play a crucial role in all the processes and applications described. 4,5However, there is a lack of analytical techniques that allow for the simultaneous characterization of freely moving NPs, and their mutual interactions under high ionic strength conditions mimic the real-life situation in living bodies or the environment.
Recently, only a few studies dealing with the behavior of NPs in the presence of other NPs have been published.Pyrgiotakis et al. 6 used an atomic force microscopy (AFM)-based platform to investigate the NP−NP interaction in biological media by employing NP-modified AFM tips.However, the Fe 2 O 3 , CeO 2 , and SiO 2 NPs were attached on the AFM tip, and they were not freely moving.Hence, the results only partially reflect the situation in living entities.Riedesel et al. 7 used titration under UV−vis, while Kaur et al. 8,9 implemented other "traditional" techniques such as transmission electron microscopy (TEM) or dynamic light scattering (DLS) to study the NP−NP interaction with surfactants.In both papers, NPs were freely moving in solution but could not be distinguished.Therefore, the results cannot be attributed to specific types of analyzed NPs.
Nowadays, Taylor dispersion analysis (TDA) is gaining more attention for the characterization of NPs.TDA is an absolute method for determining the diffusion coefficient of a given analyte from a very small sample volume (subnanoliters).Therefore, it does not require calibration or prior knowledge of the sample concentration.The hydrodynamic radius and size distribution of a given NP population can be calculated using the Stokes−Einstein formula. 10,11TDA can be considered the mathematical framework for the analysis of dispersion in solutions using a parabolic velocity profile of pressure-driven laminar flow in a cylindrical tube such as a fused silica capillary.When a sample plug is introduced into the flow, it spreads axially due to convection and diffusion.The longitudinal diffusion is negligible when conducted under well-controlled conditions; thus, the broadening of an analyte plug relates to the diffusion coefficient. 12Moreover, TDA measurements do not require any special instrumentation.They can be conducted, for example, on a standard capillary electrophoresis (CE) instrument.−15 Sensitive inductively coupled plasma mass spectrometry (ICP-MS) detection can transform TDA into a tool that is specifically designed for probing the size of NPs and even ultrasmall NPs in real samples, such as cells, tissues, air, water, and soil, where various NPs from different sources may be present simultaneously.So far, the TDA-ICP-MS approach has been employed to characterize only a specific type of NPs.Labied et al. 16 demonstrated that the TDA-ICP-MS approach can precisely determine the radius of ultrasmall Gd-containing NPs in various media.The data were also compared with UV detection, although only under specific conditions.Subsequently, the same group investigated Gd release from Gdchelated polysiloxane NPs. 17 Finally, Degasperi et al. examined the radius of core−shell gold/silica NPs interacting with proteins, namely, bovine serum albumin and fetuin. 18To the best of our knowledge, there has been no publication describing the characterization of NPs in their liquid mixtures.
All of the used running buffers (ionic strength of 40 mM) were prepared by dissolving the corresponding amount of the abovementioned acids in ultrapure water, and then 50% (w/w) NaOH was added until the desired pH value was reached.The amounts of acids needed for buffer preparation (to match the ionic strength of 40 mM) were calculated using the Peakmaster software. 23,24ials for TDA measurements were filled with buffer solutions with either 10% (v/v) of a suspension of the gold NPs (0.057 mg/mL) or ultrapure water (when blank analyses were performed).The sample consisted of concentrated Fe 3 O 4 @ COOH NPs directly pipetted into the sample vial.In the case of size investigation on Au and Ag NPs, these were injected independently or as a mixture (50:50, v/v).Here, the running buffer was not modified (neither H 2 O nor any NPs were added).
Taylor Dispersion Analysis.All the TDA experiments were conducted on the CE instrument CE7100 by Agilent Technologies (Waldbronn, Germany) equipped with a diodearray detector (DAD).Analyses were carried out in fused silica capillaries (50 μm I.D., effective length of 51.5 cm in the case of TDA-DAD and 70 cm in the case of TDA-ICP-MS) purchased from Molex (Lisle, IL, USA).They were conditioned by flushing with 0.1 M NaOH for 20 min and subsequently rinsed with H 2 O for another 20 min before first use.Between analyses, capillaries were flushed with 0.1 M NaOH for 2 min, deionized water for 2 min, and BGE for 3 min.Flushing was done at a pressure drop of 935 mbar, and samples were introduced into the capillary at 50 mbar for 5 s.Then, the sample zone was pushed through the capillary toward the detector at a pressure drop of 50 mbar at a temperature of 25 °C.The signal of Fe-based NPs was collected by using a DAD under 200 nm.All analyses were performed in three replicates; thus, the results are expressed as the arithmetic mean ± standard deviation.−28 TDA-ICP-MS Method.−21 Since no electricity is needed during TDA analyses, deionized water with the addition of Sc, Tb, and Bi (10 ng/mL) was used as a sheath liquid to monitor the stability of an aerosol supplied to an ICP-MS instrument.The RSD of the signal intensity for 45 Sc, 159 Tb, and 209 Bi was less than 6%, indicating the stability of the sheath liquid supply.The ICP-MS instrument was tuned daily to achieve the highest S/N ratio for the analyzed elements.Optimized operating conditions for ICP-MS were as follows: an RF power of 1550 W; a plasma gas flow rate of 15.0 L/min; an auxiliary gas flow rate of 0.9 L/min; a nebulizer gas flow rate of 0.75−0.8L/min; a makeup gas flow rate of 0.3−0.4L/min; a collision gas He flow rate of 3.0 mL/min (for Fe NPs determination); and a dwell time of 150 ms for 54 Fe and 56 Fe; 100 ms for 107 Ag, 109 Ag, 159 Tb, 197 Au, 12 C, and 34 S; and 80 ms for 45 Sc, 159 Tb, and 209 Bi.The signal for the most abundant Fe isotope, 56 Fe, and signals of 107 Ag, 109 Ag, and 197 Au were used for the nanoparticle size calculations, whereas 54 Fe was used to prove TDA results for only Fe-containing NPs.On top of that, the monitoring of 34 S and 12 C was adopted as tracers of DMSO representing a nonreacting, well-defined compound.The limit of detection (LOD) was calculated as the signal intensity that equaled three times the S/N ratio.All analyses were performed in three replicates; thus, the results are expressed as an arithmetic mean ± standard deviation.
DLS and Zetametry Measurements.The mean hydrodynamic diameters and zeta potential values were determined by DLS measurements using the Malvern Zetasizer Nano instrument (Malvern Instruments, Worcestershire, UK).NPs were carefully dispersed in the electrolytes and immediately measured.All analyses were performed in three replicates; thus, the results are expressed as an arithmetic mean ± standard deviation.■ RESULTS AND DISCUSSION TDA-ICP-MS of NPs in NP-Based Media.First, a suspension of 20 nm Au NPs was added to various background media (10%, v/v) with the same ionic strength of 40 mM but varying pH values: 2.5, 4.5, 7.5, and 9.5.Then, a sample of 4.5 mg/mL Fe 3 O 4 @COOH NPs (TEM 25 nm) was subjected to conventional TDA using a DAD and ICP-MS in specific buffers.Data acquired at 200 nm were compared with blank measurements, where 10% ultrapure water was added instead of Au NPs.The resulting taylorgrams are shown in Figure 1.The modified data processing procedure was adopted to obtain a smoothed Gauss function. 29The characteristics for the nonaffected NP peak eluting in a void volume were used for the calculations of diffusion coefficients and, subsequently, NPs' hydrodynamic diameters (see Table 1 and Supporting Information for data processing).
Both analyzed NPs increased their zeta potential with increasing pH: Fe 3 O 4 @COOH NPs from 3.6 mV at pH 2.5 to −10.6 mV at pH 9.5 and Au NPs from −1.6 mV at pH 2.5 to −22.3 mV at pH 9.5 (see the Supporting Information for more details).As a result, electrostatic repulsion between NPs can be expected only in an alkaline environment.We believe that this could be the reason for the higher fluctuation of the 197 Au signal at pH 2.5 and 4.5, where NP−NP aggregation might occur.
Furthermore, a significant decrease in the 197 Au signal was observed at these pH levels, which would be barely possible to discern in a common TDA-DAD.This might be attributed to the adsorption of Au NPs onto the capillary wall.This observation aligns with DLS measurements, where the results for Au NPs at pH 2.5 and 4.5 were not reproducible.The possible aggregation can also lead to an overestimation of the diameter of NPs in the mixture, especially in the case of DAD (as DAD represents a universal detector).
Interestingly, most of the DAD data show partially sharp profiles, resulting in unrealistic NPs' diameters ranging from 0.04 to 1.7 nm (D h1 ).This has been explained by the release of NPs' stabilizing ligands, which have a higher absorption coefficient than that of the NPs themselves. 30The observed sharp peak corresponds to much smaller molecules and not to NPs with a TEM size of 25 nm.On the other hand, the non-Gaussian peaks may be attributed to the partial adsorption of NPs on the capillary wall, NPs' polydispersity, or open-tube hydrodynamic chromatography. 10,31,32In this context, only the use of a specific detection, such as ICP-MS, led to accurate results.
In comparison with traditional DLS, the DLS results cannot be considered reliable since DLS is not element-specific.It can determine only the average size of the given mixture rather than each type of NP independently.The inconsistency with the values declared by TEM (25 nm) is because both DLS and TDA take into account the hydrodynamic diameter in a specific environment, whereas TEM can determine only the core of the studied NPs.
Our TDA-ICP-MS approach can accurately determine the diameter of Fe 3 O 4 @COOH NPs both alone and in the presence of Au NPs, as shown in Table 1.The values observed are consistent at the same pH levels.Variations between electrolytes can be attributed to the different surface chemistry of NPs,  1 stand for D h2 (see Figure 1).c At 56 Fe. d Irreproducible, possible aggregation.including zeta potentials and their hydrodynamic behavior.Furthermore, as indicated by the ICP-MS profiles in Figure 1, multiple peaks were observed, which may be indicative of NPs' adsorption on the capillary wall or interactions between Fe 3 O 4 @ COOH NPs and Au NPs.This complexity also cannot be resolved using conventional TDA-DAD or other measurement techniques, such as DLS or TEM.TDA-ICP-MS of NPs' Mixtures.The multielemental and isotope-specific character of ICP-MS detection offers another unique capability: the simultaneous TDA sizing of NP mixtures in complex media.In our approach, we injected a mixture of Au and Ag NPs (1:1, v/v) with sizes of 20 and 10 nm (as determined by TEM), respectively, into the capillary and conducted TDA (Figure 2).It is evident that the peak representing Au NPs is broader than the one corresponding to Ag NPs, indicating that Au NPs have a greater diffusion coefficient and thus hydrodynamic diameter.This finding was confirmed by the DLS data obtained from the diluted individual NP samples (Table 2) and the reported sizes of the analyzed NPs.
The ability to simultaneously determine the NP's size in complex mixtures is one of the key advantages of our isotopeand element-specific TDA-ICP-MS approach because this task remains challenging for other techniques.Such NPs are often difficult to detect using DAD due to their low absorbance and high scattering.Therefore, conventional TDA-DAD and even DLS provide only an average profile, and other characterization techniques like high-resolution TEM or single-particle ICP-MS are constrained by the environment.
Our TDA-ICP-MS approach is versatile and readily applicable to aqueous and high-ionic-strength environments, e.g., citrate buffer at pH 7.0.Furthermore, the reliability of the calculated hydrodynamic diameters can be confirmed by the analysis of multiple isotopes (not applicable for monoisotopic elements).For example, the results for Ag NPs were confirmed for both isotopes, thus accounting for possible interferences in the setup (Table 2).
Another remarkable feature of the TDA-ICP-MS approach is its ability to determine isotope ratios (see the Supporting Information for procedures 33,34 ).The signals obtained for 107 Ag and 109 Ag isotopes are shown in Figure 2B.In this case, the 107 Ag/ 109 Ag ratio is 1.0770 ± 0.0046, while the natural 107 Ag/ 109 Ag ratio is 1.0764. 35This suggests that commercially available Ag NPs are manufactured from natural Ag sources.This feature of our TDA-ICP-MS approach can be highly beneficial for process control when isotopically enriched NPs are produced or when investigating NPs composed of different isotopes and their fate in the environment or cellular uptake. 36inally, the LOD values were determined to be 37.8 μg/mL for Fe, 0.26 μg/mL for Ag, and 0.24 μg/mL for Au, showcasing the limits of our TDA-ICP-MS approach.These values align with the superior LODs and wide linear dynamic range associated with ICP-MS detection.

■ CONCLUSIONS
In conclusion, our work has introduced a novel multielemental and isotope-specific TDA-ICP-MS methodology for studying the behavior of NPs in the presence of other NPs with aqueous high-ionic strength conditions that mimic real-life situations in cells and the environment.We conducted experiments with carboxylated magnetite (Fe 3 O 4 @COOH) and Ag NPs in the presence of Au NPs under various conditions, demonstrating the extraordinary potential of our method.We believe that our methodology has wide applicability for addressing numerous questions in the fields of nanomaterial chemistry, nanotoxicology, and medicine, for example, studying the selfassembly processes, effects of different NPs, NPs' functional agglomerates, and metal-based nanorobots freely moving in living organisms or the environment.

Figure 1 .
Figure 1.Comparison of the TDA-DAD (left) and TDA-ICP-MS taylorgrams for Fe 3 O 4 @NPs (4.5 mg/mL) in the presence of the Au NPs (TEM 20 nm, 6 μg/mL) at various environments differing in pH.

Table 1 .
Comparison of Results Provided by TDA and DLS a At 200 nm.b Values in Table