Real-time size discrimination and elemental analysis of gold nanoparticles using ES-DMA coupled to ICP-MS
We report the development of a hyphenated instrument with the capacity to quantitatively characterize aqueous suspended gold nanoparticles (AuNPs) based on a combination of gas-phase size separation, particle counting, and elemental analysis. A customized electrospray-differential mobility analyzer (ES-DMA) was used to achieve real-time upstream size discrimination. A condensation particle counter and inductively coupled plasma mass spectrometer (ICP-MS) were employed as downstream detectors, providing information on number density and elemental composition, respectively, of aerosolized AuNPs versus the upstream size selected by ES-DMA. A gas-exchange device was designed and optimized to improve the conversion of air flow (from the electrospray) to argon flow required to sustain the ICP-MS plasma, the key compatibility issue for instrumental hyphenation. Our work provides the proof of concept and a working prototype for utilizing this construct to successfully measure (1) number- and mass-based distributions; (2) elemental compositions of nanoparticles classified by size, where the size classification and elemental analysis are performed within a single experiment; (3) particle concentrations in both solution (before size discrimination) and aerosol (after size discrimination) phases; and (4) the number of atoms per nanoparticle or the nanoparticle density.
KeywordsDMA Electrospray Gold Nanoparticle ICP-MS Quantitative analysis
Nanoparticles (NPs), especially in the colloidal form (i.e., dispersed in a fluid medium), are increasingly used in the production of consumer products, biomedical and diagnostic devices, drug delivery systems, and advanced materials for catalysis, coatings, and food packaging, among other uses [1, 2, 3, 4, 5, 6, 7, 8, 9, 10]. For these applications, physical size and elemental composition are among the most important properties that influence the performance of nanobased products [11, 12, 13, 14, 15]. For instance, nanomedicines that use NPs as drug delivery vehicles are under development [11, 13, 14, 16, 17, 18]. Studies have shown that therapeutic performance is related to the size and chemical composition of NPs used in the formulations [12, 15, 19]. Traceable, quantitative analytical methods for determining size and chemical composition are necessary to accurately and precisely correlate these properties to efficacy and safety and are critical for the advancement of NP-based therapeutics. For particle size characterization, techniques based on physical measurements, especially classification prior to detection (e.g., chromatography, electrophoresis, and field flow fractionation), are widely used due to their capacity to enable size-resolved measurements [20, 21, 22]. To characterize elemental composition with high sensitivity, techniques such as mass spectrometry, optical emission spectrometry, X-ray photoelectron spectroscopy, and energy dispersive X-ray spectroscopy are used extensively [12, 23, 24, 25, 26, 27, 28]. By combining the information obtained by multiple characterization methods, the material properties of NPs can be presented more completely and accurately.
In addition to applying complementary measurements independently [11, 16, 23, 25, 29, 30, 31, 32], a recent advance is the development of real-time, quasi-simultaneous characterization methods through the hyphenation of complementary instrumentation: e.g., upstream size separation coupled with downstream elemental detection [20, 33, 34]. Hyphenation not only combines the advantages of using these instruments individually, but also offers a major advantage by providing real-time analysis of the material properties of NPs for a more synchronized data comparison. For example, a complex mixture of NPs can first be classified and subsequently analyzed, in order to differentiate among coexisting populations, with minimal perturbation to the native sample.
Differential mobility analysis (DMA), also called scanning ion mobility spectrometry, is a high-resolution size classification technique relevant to the analysis of discrete (dispersed) nanoscale species such as NPs, viruses, proteins, and DNA [35, 36]. For sampling from an aqueous solution, an aerosol generator, such as an electrospray (ES) device, can be combined with DMA, allowing the DMA size discrimination to be applied to particles as small as a few nanometers in diameter [11, 16, 30, 31, 35, 36, 37, 38, 39, 40]. The major advantage of using electrospray-differential mobility analyzer (ES-DMA) is that the ES can generate well-controlled, monodisperse, sub-micrometer droplets containing a single particle per droplet [35, 36, 37, 41, 42]; thus, it is ideal for characterizing NPs at appropriate particle number concentrations (1010 to 1013/cm3) with resolution down to about 0.3 nm. In the classical application of ES-DMA, a condensation particle counter (CPC) is attached to detect aerosolized particles of a specific size, selected by controlling the applied voltage in the DMA classifier. In this configuration the technique is limited to providing aerodynamic size-resolved concentration, albeit with a very high degree of size resolution.
For quantitative elemental analysis, inductively coupled plasma mass spectrometry (ICP-MS) is the technique of choice, due to sensitivity at sub-nanograms per kilogram (parts per trillion) to micrograms per kilogram (parts per billion) levels [22, 43, 44]. By combining DMA with ICP-MS, it should be feasible to directly obtain size-resolved composition and mass-weighted size distributions of nanoscale species with high sensitivity. Previous studies have also shown the feasibility to measure the composition of aerosolized NPs directly by integrating ICP-MS with different types of aerosol generators (e.g., atomizer/nebulizer and laser ablation) [34, 43, 45]. Elemental composition, on a basis of single particle analysis, could be achievable for the NPs having sufficiently large size and at low number concentration in solution.[43, 46] We are aware of only two previously published studies in which ES-DMA has been directly hyphenated to ICP-MS: Kapellios et al.  and Carazzone et al. partially addressed the problem of incompatibility of air with the ICP-MS plasma (air is typically used in ES and as the sheath flow in DMA). As we will show, gas incompatibility is the principal coupling issue for hyphenation of ES-DMA with ICP-MS.
The ES device requires air or other compatible conductive gases and cannot be operated using Ar due to the low electrical breakdown voltage of the latter. On the other hand, the ICP-MS operates with Ar, where air content must be minimized to sustain a stable plasma. Moreover, the high oxygen content present in air interferes with the detection and quantification of several key elements, such as S, so it is advantageous to minimize air entering the ICP. Hence, the air in the gas flow carrying the analyte must be exchanged with argon before introduction into the ICP-MS system. One partial solution  is to utilize the DMA itself as a gas converter (i.e., use a sufficient flow of Ar as the sheath gas in the DMA) to achieve sufficiently low air content to allow stable on-line ICP-MS measurements. Although these previous studies show it is possible to overcome the air/plasma stability issue by using the DMA as the sole gas converter , the necessity of using a higher sheath flow rate significantly restricts the operational size range of the DMA and decreases the concentration of aerosolized NPs delivered to the downstream ICP-MS detector. As a result, the gas-exchange efficiency is coupled to the size resolution and dynamic range of the DMA. Additionally, the residual air flow is still high enough to present an interference issue in the MS detection of specific elements.
Our primary objective is to develop and validate a high-resolution, hyphenated measurement system capable of providing quantitative information on the size-resolved elemental composition of engineered NPs and their functional coatings (where applicable). Through a systematic study of ES-DMA coupled with ICP-MS (denoted as ES-DMA/ICP-MS), we optimized the operational window (e.g., size range, sensitivity of elemental analysis, etc.) for this hyphenated system, and extended the measurement capability to obtain other important material properties (e.g., particle concentration and particle density). We designed and constructed a gas-exchange device (GED) installed just prior to the ICP, which affords the capacity to operate the DMA at the desired size range while achieving the gas-exchange efficiency necessary for high sensitivity elemental analysis. The GED concept is based on the membrane gas exchange apparatus described previously by Nishiguchi et al.  for direct monitoring of airborne particulates by ICP-MS. Gold nanoparticles (AuNPs) were chosen as the test material for the present studies, because of their widespread use in biomedical applications and because of our substantial prior work on this material [11, 16, 17, 18, 23, 29, 30, 47]. The GED was systematically studied to optimize efficiencies of gas exchange and particle transport. Additionally, the advantages of this technique are discussed for applications related to NPs and nanobased products.
Commercially available AuNP suspensions (nominally 30 nm and 10 nm in diameter) were purchased from Ted Pella (Redding, CA).1 A 20-nm polystyrene (PS) nanosphere size standard was purchased from Thermo Scientific (Madison, WI). Ammonium acetate (Sigma-Aldrich, St. Louis, MO) was used to adjust the ionic strength of samples for ES operation. Biological-grade (18 MΩ × cm) high-purity filtered deionized water (Aqua Solutions, Jasper, GA) was used for all sample preparation and dilutions.
Citrate-stabilized AuNPs (1 mL) were centrifuged (MiniSpin Plus, Eppendorf North America, Hauppauge, NY) for one cycle (additional cycles prevented resuspension) under the following conditions: 30 nm AuNPs for 15 min at 9,660×g; 10 nm AuNPs for 45 min at 14,100×g. After centrifugation, the supernatant was removed, ammonium acetate solution (5 mmol L−1) was added for final Au concentrations of approximately 5–250 μg g−1, and samples were sonicated using a bath sonicator (model 1510, Branson, Danbury, CT) to aid resuspension of the particles.
Electrospray-differential mobility analyzer
Settings and parameters for the gas exchange device (GED)
Length of PGM (Lmem)
Pore size of PGM
Inner diameter (ID) of inner tube (PGM)
Outer diameter of inner tube (PGM)
ID of GED outer chamber
Length of GED outer chamber (L0)
Sweep flow rates (Qsweep)
0–6 L min−1
Sample flow rates (Qsamp)
0.25–1 L min−1
The Qsamp was varied from 0.25 to 1 L min−1, and the Ar sweep flow (Qsweep)was varied from 0 to 6 L min−1. The flow rate at the GED outlet was measured by a DryCal flow meter (ML-800-24, Bios International Corp., Butler, NJ) set for Ar. The relative standard uncertainty (precision) of the O2 measurements is typically less than or equal to 0.03 %.
Inductively coupled plasma mass spectrometer
Instrumental settings for ICP-MS operation
Radio frequency (RF) power (W)
Coolant gas (L min−1)
Auxiliary gas (L min−1)
Qsamp (L min−1)
Mass monitored (m/z)
Integration time (s)
List of other symbols not listed in Table 1
Particle mobility diameter (nm)
Particle number concentration in gas phase after DMA (cm−3)
Sheath flow rate (L min−1)
Particle loss through the GED
Percent O2 at the GED outlet
Percent O2 in the sample flow (before GED)
Scanning time (s)
Diffusion rates of Ar [mol/(m2s)]
Diffusion rates of air [mol/(m2s)]
Retention time of sample flow through the PGM (s)
Radius of PGM (nm)
Np measured before GED (cm−3)
Np measured after GED (cm−3)
Gas exchange efficiency
Concentration of particles in solution (cm−3)
As-received concentration of particles in solution (cm−3)
Np, l, 0
Number concentration of particles in gas phase before DMA (cm−3)
Number concentrations of PS in gas phase before DMA (cm−3)
Np, g, PS
Number concentrations of PS in solution phase (cm−3)
Np, l, PS
Average density of a AuNP (g/cm3)
Number of 197Au atoms per particle
Total 197Au atoms present in solution
Volume of a single AuNP (cm3)
Hybrid instrument operation (ES-DMA/ICP-MS)
Results and discussion
Study of GED
Here, JAr and Jair are the diffusion rates of Ar and air through the membrane, respectively, and tmem is the retention time of sample flow through the PGM. According to Eq. (3), JAr, Jair, and tmem are proportional to EffGED. Increasing Qsweep should increase the concentration gradient between the sweeping flow and the sample flow, resulting in an increase in both JAr and Jair. Decreasing Qsamp increases tmem (i.e., allowing more time for gas diffusion). Hence, decreasing Qsamp, or increasing Qsweep, will increase EffGED. Our experimental results (Fig. 3A) support this mechanism.
As shown in Fig. 3B, increasing Ω pushes EffGED toward 100 %, suggesting a longer retention time, or a higher concentration gradient, is beneficial for achieving higher Eff,GED. The result also indicates that Lmem and rmem could be increased to maintain higher Ω, if an increased value of Qsamp is required for measurement resolution or sensitivity.
In addition to investigating EffGED, we also evaluated the potential influence of Qsamp and Qsweep on the loss of particles during transport. However, even without using the GED, the particle concentration from the ES process was reduced significantly by reducing Qsamp from 1 to 0.25 L min−1. We attribute this loss to the deposition of highly charged AuNPs generated by the ES onto the wall of the orifice plate, thus failing to reach the neutralization chamber. This effect is apparent at low carrier flow rates. ICP-MS requires Qsamp of ≈1 L min−1 to ensure sample penetration through the plasma, thus Qsamp at the GED inlet must be at least 1 L min−1 for this application. We therefore set Qsamp equal to 1 L min−1 for the measurement of AuNP transport.
Particle loss due to diffusion and/or deposition onto the surface of the GED was quantified based on the measured particle counts in sample flow using the ES-DMA-CPC configuration (Fig. 3C). Number concentration of size-classified AuNPs was measured under various Qsweep (0 to 6 L min−1) at a Qsamp of 1 L min−1. Particle loss through the GED, PL[(1 − (Np, GED/Np, samp), where Np, samp and Np, GED are Np measured before and after the GED, respectively], was determined using AuNPs of two nominal diameters: 30 and 10 nm (denoted as 30AuNP and 10AuNP, respectively). Unlike gas molecules, the size of the particles used in our experiment are within one order of magnitude of the pore size of the PGM. The steric hindrance of the pores may therefore restrict NP diffusion toward the outer chamber of the GED—a beneficial effect of size discrimination (i.e., gas molecules can diffuse through the membrane, but generally not NPs).
For 30AuNPs, particle loss due to the GED was negligible (≈1.5 % when Qsweep = 0 L min−1). Interestingly, Ar sweep flow through the GED increased the particle number concentration beyond the baseline value (measured without the GED) by ≈4 % at Qsweep = 3 L min−1 and 8 % at Qsweep = 6 L min−1. The reason for the increase in particle concentration is not entirely clear. However, it could be due to a slight increase in the Ar content of the carrier flow post-GED (i.e., less diffusion loss due to the denser Ar carrier gas), or possibly due to the uncertainty in the measurement of particle concentration (typically ≈3 % but can be as high as 20 % under some conditions) [31, 35]. Details of the calculation of particle loss are described in Section 2 of the ESM.
For 10AuNPs, particle loss due to the GED was >23 %. This relatively high loss could be the result of increased diffusional deposition of the smaller particles and/or an increase in penetration through the pores of the GED membrane itself. Based on calculations of diffusion loss , diffusion accounts for ≈4 % of the particle loss for 10AuNPs and 1 % for 30AuNPs (details in ESM). Thus, particle loss for the 30AuNPs is attributed principally to diffusion, but the most of the particle loss for the 10AuNPs can be attributed to penetration through the membrane pores. We also observed that increasing Qsweep increased the particle number concentration relative to the baseline by ≈4 % at Qsweep = 3 L min−1 and >7 % at Qsweep = 6 L min−1, similar to the increase for the 30AuNPs at the same value of Qsweep. Hence, increasing Qsweep (or the relative pressure outside the membrane) may suppress the diffusion loss of particles (through the membrane or in transport).
We have demonstrated the proof of concept for using the GED to discriminate the diffusion of gas molecules from AuNPs, and then to achieve sufficient gas-exchange efficiency without substantial loss of particles during transport. Based on the particle size involved, and the tolerance of ICP-MS to air (i.e., an air content below 10 % is necessary for the stability of the plasma), parameters such as the length and pore size of the membrane can also be adjusted for optimization. This is the key advantage of using a modular GED instead of relying on the DMA alone for reducing air content. For example, in order to avoid exceeding the threshold of electrical breakdown for the Ar sheath gas inside the electrostatic classifier, the DMA voltage has to be operated under ≈−3 kV (based on our experimental setup), which is not applicable for NPs having dp, m > 40 nm. With the assistance of the GED, we were able to perform measurements for dp, m > 40 nm while maintaining stable plasma operation (i.e., operation at lower Qsh while maintaining the air content below 10 % for the ICP). In the next section, we discuss the application of the GED to the hyphenation of ES-DMA and ICP-MS.
Hybrid instrument (ES-DMA-GED/ICP-MS)
Particle size distribution with hyphenated elemental analysis
In this section, we discuss characterization of AuNPs using our customized ES-DMA system coupled to the ICP-MS through the GED. The ES-DMA was operated at an aerosol flow of 1 L min−1 to ensure that particles from the ES were successfully delivered to the DMA. We chose Qsh = 10 L min−1 to ensure sufficiently low air content delivery to the ICP. Qsweep of 3 L min−1 was used for the hyphenated measurements. Although the ICP-MS can operate at Qsh = 10 L min−1 without using the GED, the addition of the GED can reduce the level of air to improve the plasma stability and potentially reduce interference from O species.
By combining the data obtained from the CPC and the ICP-MS, the elemental composition of particles can be identified in a size-dependent manner. After the CPC data had been acquired, the direction of particle flow was switched from the CPC toward the ICP-MS, and the 197Au signal intensity was acquired at each size step using the ES-DMA/ICP-MS with the ICP-MS set to TRA mode. In Fig. 4A, AuNPs are clearly distinguishable from the salt residue, based on the coincidence of the peak (enlarged in Fig. 4B) at ≈32 nm (CPC) and the 197Au peak (ICP-MS). However, there is no 197Au signal corresponding to the peak at 9.4 nm, confirming the peak consists only of salt residue (mainly sodium ≈20 ppm and potassium ≈30 ppm found in the 30AuNPs using the semi-quantitative mode of the ICP-MS). Thus, this hybrid instrument can distinguish the components in a mixed sample based on elemental composition of size-separated particles.
The hyphenated method described here provides a promising way to resolve the issue of salt residue that frequently occurs during ES-DMA analysis . It is well known that salt residue from the ES process interferes with ES-DMA measurements; therefore, it is necessary to remove non-volatile salts and add volatile ammonium acetate solution prior to ES-DMA measurements. If the ES-synthesized (artificial) salt particles have a size that is close to, or even larger than, the size of the NP analyte, the number-based PSD of the NPs will be obscured by the salt peak . The ICP-MS detector, on the other hand, can successfully distinguish the population of artificial salt particles from the AuNPs based on elemental analysis. Hence, we can still measure an accurate PSD of NPs while avoiding the deleterious effects of non-volatile salt removal (e.g., centrifugation-induced aggregation). Also, the capacity to obtain mass-based PSD for comparison to the number-based distribution provides the opportunity for a more sophisticated analysis.
Measurement of the concentration of AuNPs in solution
Measurement of the density of AuNPs
The methodology proposed here would be very useful for supporting other types of instrumentation requiring an accurate value of particle density, such as disc centrifugation and analytical centrifugation. For these centrifugation-based measurement methods, the principle to differentiate particle size is based on the mass of the NPs. However, the density of NPs is usually unknown and always assumed to be equal to the value of the corresponding bulk material. Although our results show that AuNP density is close to the value of bulk Au, this might not be true for other types of NPs, due, for instance, to differences in synthetic processes. Take amorphous silica NPs for example, the density can vary from 1.8 to 2.3 g cm−3  and is generally less than the bulk value (≈2.65 g cm−3).
The hybrid instrument developed in this study is applicable to a broad range of NP-based materials, including environmental, biological, medicinal, and energy-related materials. There is an interest in analytical characterization of metal-containing NP colloids, both engineered and naturally occurring, for environmental studies of particle transport, trace metal transport via NPs, and water quality effects [53, 54], in which accurate and precise characterization of the size of particles and their aggregates, as well as their chemical composition, is critical.
This hyphenated technique has limitations with respect to the size range, concentration, and elements that can be detected. For example, purely organic materials cannot be quantified using ICP-MS. Depending on the DMA column used, the ES-DMA can only classify particles in the size range from ≈3 to ≈1,000 nm, but this covers most of the nanoscale regime and many particles of technological interest. In this study, the particle number concentration could be calibrated up to a level equivalent to Au mass fraction of ≈10 μg g−1. Because the detection limit of the batch ICP-MS is much lower, with ES-DMA upstream classification, we can obtain the elemental composition of “size-defined” NPs (e.g., ±0.3 nm in diameter for 30AuNP). This will be especially useful for analyzing the elemental composition of NP samples having different mobility sizes (e.g., 10AuNPs mixed with 30 nm SiO2 NPs) or the AuNPs containing impurities that are size dependent.
We have developed a prototype hyphenated instrument to characterize liquid-dispersed NPs based on upstream size separation coupled with quantitative elemental analysis and particle counting. A modular GED can improve conversion of air flow from the ES to Ar flow for ICP-MS analysis and also expand the operational size range for the hyphenated technique. By combining the data obtained from the CPC and the ICP-MS using ES-DMA sample introduction, the hybrid instrument can provide real-time results on particle size distributions with quantitative elemental composition, population of NPs in solution, the number of atoms per particle, and the average density of individual NPs. This is a powerful measurement technique for the simultaneous characterization of particle size and elemental composition, with relevance for a wide range of nanotechnology applications.
Current work is focused on optimization of methodology for analysis of more complex NP systems, such as core/shell particles and particle–ligand conjugates, thus expanding the technique to a broader range of real-world applications. Specific issues requiring further research include reduction of the air (particularly O2) content, which interferes with the determination of S and other key elements, and validation of calibration methods for a broader range of nanomaterials.
The identification of any commercial product or trade name does not imply endorsement or recommendation by the National Institute of Standards and Technology.
This research was performed while S.E. held a National Research Council Research Associateship Award at NIST. The authors thank Yonggao Yan and Mindong Li at NIST for their help with developing the customized DMA program. The authors thank Robert Cook and Julien Gigault at NIST, and Prof. Michael Zachariah at the University of Maryland, for manuscript review and helpful discussions.
- 1.Henig RM (2007) Our silver-coated future. OnEarth. Fall 2007Google Scholar
- 2.Jin R (2012) Nanotechnology Review 1:31–56Google Scholar
- 3.Fakruddin M, Hossain Z, Afroz H (2012) Journal of Nanobiotechnology 10:31Google Scholar
- 4.Conde J, Rosa J, Lima JC, Baptista PV (2012) Int J Photoenergy. doi:10.1155/2012/619530
- 6.Riggio C, Pagni E, Raffa V, Cuschieri A (2011) J Nanomaterials. doi:10.1155/2011/164506
- 31.Pease LF, Tsai DH, Brorson KA, Guha S, Zachariah MR, Tarlov MJ (2012) Anal Chem 83(5):1753–1759Google Scholar
- 49.Hinds WC (1999) Aerosol technology: properties, behavior, and measurement of airborne particles, 2nd edn. Wiley, New YorkGoogle Scholar
- 51.National Institute of Standards and Technology Report of Investigation RM8011–RM8013 2008Google Scholar