Multi-screening analysis of single nanoparticles by the first multi-quadrupole ICPMS/MS

The recent expansion in the use of nanomaterials in various fields has led to a growing concern over their impacts on environmental systems. Accurate detection, quantification, characterization, and tracking of the nanoparticles are essential for assessing the ecological risks and their implications for environmental health. Inductively coupled plasma mass spectrometry (ICP-MS) operated in single-particle mode is an attractive analytical tool for quantification and sizing very small nanoparticles in aqueous suspensions. The ICP-MultiQuad-MS/MS can rapidly detect (including highly interfered elements) and characterize a large number of particles, determine size and size distributions, the particle number concentration in a sample, and the elemental mass concentration of both dissolved and particulate species. This paper provides the first systematic review of the ICP-MultiQuad-MS/MS to perform NPs analysis applied to different structures and compositions while simultaneously comparing them with the current state of the art of ICP-(CRC)-MS available. This study is done on composite nanomaterials with different shapes for elements with high sensitivity and no interference, like Au, Ag, and Pt, and elements with high interferences and lower sensitivity, like Fe, Si, and Ti.


Introduction
Nanotechnology is a fast-growing field of science that employs materials and their properties at a nanometer (10 −9 m) scale.The determination and characterisation of nanoparticles (NPs) are crucial for environmental and toxicological studies and are an increasing concern to regulatory agencies and researchers.A wide variety of engineered NPs is used in consumer products such as cosmetics (TiO 2 and ZnO), textiles (Ag), polishing slurries (SiO 2 , Al 2 O 3 , CeO 2 ), and electronic manufacturing [1][2][3][4].Typically, NPs morphology and size are determined by microscopy techniques, SEM and TEM.These methods provide high spatial resolution, but accuracy depends on the number of particles used for mean size calculation [5].Also, artifacts can be generated during sample preparation and at low NP concentrations.In addition, very few robust analytical approaches detect NPs in the environment where the NPs are predicted to occur in very low concentrations, in the range of ng.l −1µg.l−1 [1,6].In the past decade, NPs characterisation using single particle inductively coupled plasma mass spectrometry (SP-ICP-MS) has been preferred [7].The technique uses a standard ICP-CRC-MS setup and timeresolved detection to probe NPs introduced into diluted suspensions.While NPs analysis by SP-ICP-MS/MS has reached reliable standards, the determination of size, composition, and concentration of materials comprising interfering elements remains challenging.
With the introduction of four quadrupoles, the ICP-MultiQuad-MS/MS NexION 5000 has the potential to improve the reliability and performance of the reaction cell methods.It can rapidly detect (including highly interfered elements) and characterize many NPs, determine size and size distributions, the particle number concentration in a sample, and the elemental mass concentration of both dissolved and particulate species.To detect different elements, each characterized by a specific massto-charge (m/z) ratio, different settling and dwell times are crucial parameters for an accurate analysis of the NPs.The main advantage of the MultiQuad ICP-MS/MS is its ability to work with MultiModes (reaction DRC, collision KED, or standard (no gas), Extraction or Focused Ion, Hot, Cold, Wet, Dried Plasma mode), depending on the analyte sensitivity.It, therefore, provides a high level of interference removal [8,9].
In this paper, studies are done on composite nanomaterials, including spherical, core-shell particles with a non-uniform distribution of elemental components and different shapes like cube, rods, etc. Elements with high sensitivity and not with many interferences, like Au, Ag, and Pt, and materials with high interferences and low sensitivity, like Fe, S, and Ti, were analyzed.This paper provides the first systematic and critical review of the ICP-MultiQuad-MS/MS to perform NPs analysis (or characterization) applied to different structures and compositions while simultaneously comparing them with the current state of the art of ICP(CRC)MS available.

Instruments and materials
The instruments used in this study are the ICPMulti-QuadMS/MS (NexION 5000, Perkin Elmer, France) and the ICP-CRC-MS (7900 series, Agilent Technologies, France) in the nano/single particle mode.The nebulizer is an inert concentric PFA with an ultra-inert cyclonic chamber (SilQ, Perkin Elmer).The sampling cone and skimmer are Platinum cones, and the Hyperskimmer is Gold covered.The sonicator used is from VWR (USC 300 THD/HF).For particle size and concentration analysis, the dwell time is set at 100 µs and the settling time at 100 s for both instruments for most elements except Fe, Si, and Ti.For these three elements, the dwell time was set at 50 µs with the settling time at 100 s due to laboratory constraints.For dwell time analysis comparison, only ICPMultiQuadMS/MS is set at 10 µs, 25 µs, 50 µs, and 100 µs as Agilent 7900 does not allow a dwell time below 100 µs.The NP standards were ordered from Sigma Aldrich (France), nanoComposix (USA), and Merck (France).The ultra-pure water for dilution and NPs solution are Type I from Advantage (Merck-Millipore).

SP-ICPMS/MS measurements and data analysis
The number of particles entering the plasma depends on the particle concentration in the sample, the sample introduction system used (i.e., the transport efficiency (TE) to the plasma), the sample uptake rate, and the sample matrix [10].The TE was calculated every day before analysing the samples using 1 ppb Au as dissolved standard and Au NPs at the particle concentration of 98,900 parts/ml (around 125ppt Au equivalent for 50 nm).For ICP-MultiQuadMS/MS, TE calculated for particle size was 11-14%.The solutions were sonicated for 4 min before and after the dilutions with an ice pack within the water volume to keep the temperature below 20 °C and to have a homogenous NP solution/ to prevent aggregation.For each element containing NPs, the relative dissolved chemical element was used for instrument calibration on both instruments.The concentration was set at 1 ppb for Au, Ag, and Pt, 5 ppb for Fe and Ti, and 10 ppb for Si.All elements in this study were analyzed only in extraction mode (focused mode is mainly beneficial for Sulfur containing NPs).The gas modes used to analyze each element are presented in Table 1 below.The parameters for instrument analysis are mentioned in Table 2. Additionally, formulas used to convert particle size to mass concentration are mentioned in Table 3.  , where the core is silver, and the shell is silica, Agilent gives a value (81.6nm) approx.3nm less than ICPMulti-QuadMS/MS (85nm).For the shell elements, it is visible that both instruments give us the diameter of the whole nanoparticle rather than just the shell, as it does not know the mass fraction of the element in the sample.The software in both instruments assumes that the whole nanoparticle is the same element rather than just a shell.For sample 6, where the shell is Ag (14nm), and the core is Au (50nm), the whole diameter of the nanoparticle is 80nm.We get a "good" value for core Au from both instruments, but NexION reports 73nm as the shell value, and Agilent reports 71.2 nm, both of which are close to the total diameter of the nanoparticle.For sample 5, only NexION can give the value for the Au shell (136nm) as the core is Si.However, it also gives a value close to the total size of the nanoparticle and not just the gold shell.For Si, the value is close to the actual core value given by TEM, but the number of peaks is too low for the data to be statistically relevant.Agilent 7900, however, cannot give a value from the sample even (for Au shell and Si) when 100x more concentration is used compared to ICP-MultiQuad-MS/MS.

Table 3 Formulas to convert particle size to mass concentration for ICPMultiQuadMS/MS by assuming compact spherical particles
Where p is the mean (n = 3) particle/ml in the solution, 1.00E + 09 is the conversion factor from g/cm 3 to ppb, ρ is the element's density in g/cm 3 , and DF is the dilution factor for the sample

Radius to mass (m) of 1 NP (g/part) Mass of the element in the solution analyzed (ppb)
Mass of the element in the original solution (ppb) Formulas used to convert spherical nanoparticle diameter of nanorods to length and width using aspect ratio

Characterization value Formula
Aspect ratio (AR) l/w Width of the rod (w)

Non-spherical NPs
Rods Both instruments give a value of 22nm for the gold NPs.For the "approximate" calculation of the length (l) and width(w) of the rod, the following formulas [11] (Table 4) are used with the assumption that ICPMS measures the spherical diameter of the NPs.For the given nanorod, the aspect ratio (AR) (characteristic of the material) is used from the values provided by the supplier.Also, the length and width formulas of the rod are based on the volume formula for hemispherical capped cylindrical nanorods as, according to the TEM images (by the supplier), the rods are this shape.
The calculated value given by ICPMultiQuadMS/MS for "l" is 42.5nm, and "w" is 10.2nm.For Agilent 7900, the value for "l" is 40.8nm, and "w" is 10 nm.Again, ICPMul-tiQuadMS/MS demonstrates a slightly better accuracy than ICPMS; however, both are in the statistical error range given by the TEM.
Cube Sample 8 is a silver cube of 100nm in length.
Using SP-ICP-MS, the particle size (as a spherical equivalent diameter) was 92.5 nm by ICPMultiQuadMS/MS, which is closer to the value given by TEM and in the given error range.However, the value by Agilent 7900 is 4nm lower than ICPMultiQuadMS/MS and falls 3nm outside the given error range.

Unknown shape NPs
The analysis for the following NPs was done at 50us dwell time due to some laboratory constraints.Only ICPMul-tiQuadMS/MS can give the value for these elements as they are highly interfering.

SiO 2
The NPs are made of SiO 2 with a 200nm size range.ICPMultiQuadMS/MS gave the value of 114nm for the NPs, which is not in the error range of the value given by the manufacturer.Multiple reasons can explain the difference in size: -1) The software assumption that NPs are spherical.The error can be expected with Si being highly interfering and the shape being unknown. 2) The assumption while analyses that the mass fraction of the sample is 100% Si is not valid.So, if these conditions are optimized, the particle size value can be improved.

Fe 3 O 4
The NPs are made of Fe (II, III) oxide with a 30nm size.ICPMultiQuadMS/MS gave a value of 36nm for the NPs, which is close to the value given by TEM.This can be seen as the number of particles decrease with increasing particle size, indicating aggregation.It should be noted that the experimental conditions of Fe NPs should be monitored.They oxidize and aggregate quickly (less than an hour).Therefore, the dissolved standard and the final NPs dilutions are made right before the experiment, and even then, sonication is required by the time the third run for the NPs is done.
The NPs are made of TiO 2 with a 100-200nm size range.ICPMultiQuadMS/MS gave a value of 98nm for the NPs, which is close to the value given by the manufacturer.

Conclusion
In Only ICP-MultiQuad-MS/MS can give values because the reaction cell with high interference removal uses multiple gases for highly interfering elements like Fe, Si, and Ti.The particle size for Fe and Ti were in the range the manufacturer gave using NH 3 (MS/MS Mode) and O 2 (Mass Shift Mode) in the reaction cell, respectively.However, the particle size for Si was smaller than expected.If the ionization of this element and the method are improved further with upcoming enhancement, we believe a "better" value for Si NPs can also be achieved.

Determination of particle number and concentration
TE for this study was done in particle size mode, so the values for mass concentration by ICPMultiQuadMS/ MS are calculated by particle size (to keep the data connected to the previous section).The software can also calculate the concentration if the TE is done for particle concentration.However, there is a bias in the values provided for particle size if TE is done for particle concentration and vice versa.As mentioned before, for this study, TE is calculated as 1ppb Au dissolved and around 125ppt Au 50nm NPs.When TE is calculated by particle size, the size value for the 50 nm Au sample is 50nm, and the particle concentration of 7,474,100 parts/ml.On the other hand, for the same solution, when TE is calculated by particle concentration, the size value is 46nm, and the particle concentration of 9,449,800 parts/ml.So, the TE should be chosen concerning the experiment's goal (particle size or concentration), keeping in mind the errors that might occur.However, the values provided by Agilent 7900, the concentration values were given by the software as calculation of TE for both particle size and concentration is compulsory before starting the experiment.Additionally, for the NPs with multiple elements, correction with the mass fraction is done manually for each element's mass for both instruments.

Spherical NPs
Spherical mono-element NPs Table 6 shows the NPs concentration determined by both instruments and compared to the value provided by the supplier.For samples 1-3 which are spherical mono-element NPs, the value for both particle concentration and size is close to the value provided by the supplier for both instruments.On average, this is 80-98% of the particle concentration provided by the supplier.
Core/shell particles Samples 4,5,6, and 9 are all made from two elements.For core elements like Au and Ag (pt. 4 and 6), both ICPMS give a value close to the manufacturer's value for mass concentration and the number of particles.For sample 5, only ICPMultiQuadMS/MS can give a value for mass concentration as the core is Si.Even for the Au shell, Agilent 7900 had insignificant peaks during analysis.For sample 9, as the shell is Si, but the core is Ag, the mass concentration from both instruments is similar but less than the value provided by the manufacturer.The calculation of TE can explain the error in particle size mode.

Non-spherical NPs
Rods and cube Samples 7 and 8 are Au nanorod and Ag cube, respectively.Both ICPMS have a lower value of mass concentration and the number of particles compared to the value provided by the manufacturer.This loss in particle mass could be a combination of the assumptions of spherical particles by the software while also keeping in account the error that comes with the calculation of particle mass with TE done with particle size.

Unknown shape NPs
For the following NPs, only ICP-MultiQuad-MS/MS calculated values are mentioned as the elements are highly interfering, and Agilent 7900 cannot analyse them.
SiO 2 and TiO 2 are samples 10 and 12, respectively.Both have lower mass concentrations than the value manufacturer.An error is expected as they are highly interfered elements with unknown shapes.Also, the calculation of TE for the analysis was done in particle size mode.

Fe 3 O 4
The calculated mass concentration value for Fe 3 O 4 is very close to the value provided by the manufacturer.

Conclusion
This section compared mass concentration and number of particles via ICPMultiQuadMS/MS (NexION 5000) and ICP-CRC-MS (Agilent 7900).For non-interfering spherical NPs, both instruments had similar values.However, with samples including highly interfering elements, only ICPMS/MS gives values very close to those given by the manufacturer.For some other cases, a decrease in the mass fraction is observed, resulting in the fact that TE calculation is done in particle size mode, and the mass is calculated via particle size for ICPMultiQuadMS/MS.Additionally, the fraction of tiny particles that signal fall below LOD contributes to underestimating particle mass and number concentration.Furthermore, interlaboratory exercises with SP-ICPMS have shown inaccuracies of up to 100% for particle number and mass concentrations.[7,12]

Limits of detection for particle size
In SP-ICPMS, the "d LOD " (limit of detection for size) depends on the fraction of analyte in the nanoparticle, polyatomic and isobaric interferences, and chemical and electronic noise [7].D LOD is usually calculated using a 3-sigma approach.However, in this study for ICP-Multi-Quad-MS/MS d LOD values, the values provided are taken from the Syngistix software for 100 µs dwell time analysis (Table 7).This section compares the d LOD for multiple instruments, and the references are mentioned next to the values.It is to note that the method of calculation of d LOD might be different.LOD for ICPMultiQuadMS/MS is comparable to Agilent 8900 (both MS/MS) for noninterfering elements but significantly lower once we consider the interfering elements.

Dwell time comparisons 3.4.1 Effect of dwell time on particle integration
Capturing an entire peak event is essential in calculating particle concentration and mass accurately and precisely.It has been previously reported that the full width of a particle event is less than 0.5ms [10,20]; hence the choice of dwell time for the analysis is known to affect the maximum number of particles that can be introduced in the plasma within a given amount of time without causing the overlap between signals from individual particles.Working with dwell times in the millisecond range (3-10ms) increases the risk of measuring two or more particles within one dwell time, resulting in systematic error when counting particle events.It has been reported that a dwell time of <100µs significantly decreases the errors even for low-concentration samples (as expected in environmental samples).This section compares the dwell time data for multiple dwell times for ICPMulti-QuadMS/MS instrument.
Figure 1 shows NP events using dwell times 100us, 50us, 25us, and 10us.Reviewing the number of points per peak, the importance of shorter dwell time is emphasized in the resulting single event profile.More points for a single particle event for shorter dwell times suggest higher precision and size resolution, and the peak area can be related to the mass of the element within the NP.Hineman et al. reported that NexION 350D ICPMS could also use a dwell time of 10µs while using 60nm Au gold Nps as the sample [10].They reported that for dwell time 10µs, 30µs, and 50µs, the number of peaks per NP event was 31, 13, and 8, respectively, which is significantly lower than the ICPMultiQuadMS/MS reported for the 50nm Au NPs sample in this study (Table 8).It is to note that the number of points for a NP is size dependent, so 60nm Au NPs for NexION 350D, in theory, should have more points than 50nm Au for a similar dwell time.[10] However, the trend for the current study that decreasing dwell gives better resolution at defining a single NP event aligns with the literature [10,21].

Effect of dwell time on particle size
Calibration was performed using a 50nm Au standard and 1ppb solution of dissolved gold to evaluate the effects of dwell time on particle size calculation.The calibration was conducted at 100s scan time for all dwell times, except for 10µs, which had a scan time of 60s, as the Syngistix software and Microsoft, Suite limitations do not allow a higher value than 10,000,000 events per single acquisition.The sample flow rate was kept between 0.300 and 0.350mL.min - .Figure 2 shows different element and size NPs with the error bars from the data by the supplier.The particles' size stays within the error range (provided by TEM) for all the dwell times and elements.Figure 3 shows the number of peaks for the same NPs with different dwell times.The no. of peaks for NP events decreases drastically for 10us but stays in a similar range for 100, 50, and 25us.It may be because if the "peak width" of the signal from one individual particle is a significant fraction (more than the dwell time), then it is likely that part of the signal (whole signal) will not occur within the dwell time, considering that the particles arrive in a random pattern [21].This will reduce the number of particles being detected, and/or the result might be a signal that is a fraction of the whole signal due to one particle that is indistinguishable from the entire signal produced by a smaller particle.It will, therefore, affect the precision of the mass analyte measurement from particle to particle.So, choosing the short dwell time available is tempting because of the higher number of points per peak; it is not advisable as the dwell time may be shorter than the signal peak width.Finding a good balance between the resolution, the number of events and the abundance significantly about the dissolved background should be the goal while choosing dwell time for the analysis.

Concluding remarks
In this paper, we compared the performance of two "ICPMS" instrument geometries, MultiQuad-MS and Single MS.Particle size and concentration, number of particles per ml, d LOD , and dwell time comparison data were shown.Both instruments' particle size results were comparable when non-interfering elements like Au, Ag and Pt were analyzed.In our conditions, single MS is not able to analyse as accurately as MS/MS the interfering elements like Fe, Si, and Ti.ICPMS is an ideal and fast probe system for a laboratory seeking possible detection limits and the highest level of productivity available in NPs characterization.However, MS/MS was able to give a comparable particle size value for these elements.In general, from the    noise.ICP-MultiQuad-MS/MS can be used to identify NPs (including the highly interfered element containing) in various real samples (and matrices) to get accurate information (concentration and number of particles) of the elements present, whether in the dissolved or nanoparticulate state.Despite the dilutions made to reduce matrix effects (for example, as low as 2 ppt for Au), it can still give accurate results regarding the concentration and size of the particles.

Fig. 1
Fig. 1 Raw data of a 50 nm gold nanoparticle acquired at different dwell times comparisons made with the literature, ICP-MultiQuad-MS/MS showed smaller particle size detection limits (d LOD ) than the other ICPMS.It can be concluded that with a shorter analysis time and a universal cell that can switch between five gases (H 2 , He as collision gases and CH 4 , NH 3 , and O 2 as reaction gases), ICP-MultiQuad-MS/MS, can be the tool of choice for highly interfered containing element nanoparticle analysis with its high interference removal efficiency and with its ultra-high sensitivity (as expected for environmental samples) of analytes.It is also possible to work at shorter dwell times, till 10µs, where 25 & 50µs seem to be the best compromise, providing better resolution at defining a single NP event and, in parallel, keeping the sensitivity.Also, using shorter dwell times improves the d LOD by enhancing the background signal and decreasing the

Fig. 2 Fig. 3
Fig. 2 NP's median size data were acquired at different dwell times.The error bars are the standard deviation acquired from the data of the manufacturer

Table 1
The gas modes in the reaction/collision cell used for the analysis of each element for NexION 5000

Table 2
Operating Parameters for NexION 5000 and Agilent 7900

Table 5
Median particle sizes (mean) ± average standard deviation determined by SP-ICPMS/MS (calculated equivalent to spherical diameters).The error represents the average standard deviation of the median particle size in 3 independent runs this section, multiple NPs sizes have been compared between ICP-MultiQuad-MS/MS and ICP-CRC-MS.For non-interfering spherical NPs, both instruments had similar values.In some cases, ICP-MultiQuad-MS/ MS had slightly better accuracy than single MS (Agilent 7900), but the differences are not statistically significant.It could be attributed to the fact that the number of peaks/NPs measured by ICPMultiQuad-MS/MS is higher than by ICP-CRC-MS.For instance, for 30nm Au NPs, ICPMultiQuadMS/MS and Agilent 7900 measured 3604 and 2283 particles, respectively, for the same dilution of the sample.For Au nanorods and Ag cube, NexION has values for particle size closer to the value provided by the manufacturer than Agilent 7900.ICPMultiQuadMS/ MS could measure signals for non-interfering elements slightly more accurately than the ICP-CRC-MS.However, this is expected as Agilent 7900 is one generation older than ICPMultiQuadMS/MS.

Table 6
Particle mass calculated for NexION 500 and Agilent 7900 compared to the value given by the supplier

Table 7
Limit of detection (nm) for various elements on three different ICPMS

Table 8
Summary of points per peak for different dwell times of single nanoparticle (Au-50 nm) events for NexION5000