Single Particle Inductively Coupled Plasma Time-of-Flight Mass Spectrometry—A Powerful Tool for the Analysis of Nanoparticles in the Environment

: Single-particle inductively coupled plasma-mass spectrometry (SP-ICP-MS) has emerged as an important tool for the characterization of inorganic nanoparticles (NPs) in the environment. Although most SP-ICP-MS applications rely on the quadrupole ICP-MS (ICP-QMS), it is limited by the slow scanning speed of the quadrupole. Recent advancements in instrumentation have led to the development of inductively coupled plasma time-of-ﬂight mass spectrometry (ICP-TOF-MS) which offers a viable solution. In this review, we discuss the recent advances in instrumentation and methodology of ICP-TOF-MS, followed by a detailed discussion of the applications of SP-ICP-TOFMS in analyzing NPs in the environment. SP-ICP-TOFMS has the potential to identify and quantify both anthropogenic and natural NPs in the environment, providing valuable insights into their occurrence, fate, behavior, and potential environmental risks.


Introduction
The widespread use of nanoparticles (NPs) in industry continues to pose threats to the environment and increase health risks to organisms [1,2]. The release of engineered NPs from industrial products (such as metal nanoparticles and carbon nano-tubes) inevitably results in human exposure to NPs and is increasing with the rapidly expanding production of engineered NPs [3]. In this context, controlling NP discharge, evaluating NP health risks, and developing new regulations on NPs depend on improving the current knowledge about the occurrence, fate, behavior, and potential risks of NPs in the environment [4]. Therefore, there is an urgent need for the development of innovative and reliable methods of NP analysis [5], which require increasingly sophisticated nanometrology capable of providing accurate and robust quantitation and characterization of NPs.
NPs are generally heterogeneous in size, composition, crystallinity, and surfaces, and these characteristics significantly impact the industrial performance and environmental fate of NPs. Although instrumentation and standardized methods have been developed for decades to examine nano-scale features [6,7], the laborious sample preparation and insufficient detection limits (e.g., much higher than the actual concentration of µg·L −1 in environmental samples) hinder reliable, accurate, and high-throughput analysis. Moreover, many of the available methods are not suitable for analyzing NPs in a real environmental sample with complex matrixes and interferences. Furthermore, there is even less characterization of individual nanoparticles, and analysis of NPs at a single particle level To improve the accuracy of particle analysis, the signal can be elongated to several milliseconds using collision cell technology to achieve the accurate and precise analysis of more than one isotope in a single NP [40]. Multi-collector ICP-MS(MC-ICP-MS) is an alternative technique to the multi-element analysis of a single NP due to its capability to simultaneously monitor multiple isotopes [41]. Equipped with a fast detector (e.g., a multichannel ion counter), MC-ICP-MS achieves the simultaneous acquisition of several isotopes with a very short dwell time (e.g., 30 µs [42]), making it a powerful tool for the multi-element analysis of a single NP. However, when analyzing NPs with more complex elements, only a limited number of isotopes within a restricted mass range can be detected. If the sample is unknown, the target elements must be screened first, followed by targeted analysis, which may reduce sample utilization rates.
In contrast to the above mass analyzers, time-of-flight (TOF) mass analyzers provide a quasi-simultaneous detection of all elements and have great advantages for multi-element and high throughput analysis of single particles. Commercial TOF mass spectrometry was introduced in the 1950s and its outstanding features have been confirmed. The principle of TOFMS involves generating ions, followed by ion acceleration, and measuring the flight time of ions in a drift tube. During the process, all ions acquire the same kinetic energy in the acceleration region, with differences in velocity arising from variations in mass-tocharge. If the flight distance is known, it is possible to determine the mass-to-charge ratio of the ions by measuring the flight time of the ions, as demonstrated in Equations (1) and (2): ·L (2) where v is the speed of an ion; z is the charge of the ion; e is the electron charge; V is the acceleration voltage; m is the mass of the ion; t is the flight time; and L is the flight distance of the ion. In a commercial ICP-TOFMS, the heaviest ions reach the detector in the tens of microseconds, which means that a few full mass scans can be completed for a transit signal from a single NP [43]. A new generation of ICP-TOFMS instruments is currently on the market, including the icpTOF from TOFWERK, Vitesse from Nu Instruments, and CyTOF from Standard BioTools (formerly known as Fluidigm). These instruments use the same orthogonal design and single-pass reflectron TOF design. With a fast acquisition speed (i.e., 30 µs for one TOF full mass spectrum extractions [34]), ICP-TOFMS can determine multi-elements in a single NP and become a promising technique for single cell analysis and single particle analysis [6,44,45].

Sample Introduction Systems for Single Particle Analysis
The sample introduction system is regarded as the Achilles' heel of ICP-MS. A standard sample introduction system contains a chamber and a nebulizer with a typical transport efficiency of less than 5%, which decreases with an increase in the sample uptake rate [46]. However, by using a modified sample introduction system with a single-pass chamber and a low-consumption nebulizer, transportation efficiency can be improved [47]. For example, Tharaud et al. achieved~100% transport efficiency by using a direct injection high-efficiency nebulizer [48].
Standard sample introduction systems generate polydisperse aerosols that lead to an inaccurate analysis with SP-ICP-MS due to different ionization processes and sampling biases that NPs undergo. Moreover, standard sample introduction systems often suffer from severe matrix effects. To address these issues, monodisperse droplets generated by either a commercial piezoelectric dispenser [49] or a microfluidics-based droplet dispenser [50]  the microdroplets generated by an online introduction system can be used as an accurate and matrix-independent calibration for single particle analysis with SP-ICP-TOFMS [51].
In addition to solution analysis, in situ solid sampling can be achieved using laser ablation (LA) as a sample introduction system. In LA-ICP-MS analysis, solid samples are ablated by high-power laser shots, and the resulting aerosols are transported and analyzed by ICP-MS. When the laser fluence is attenuated at a suitable level, LA-ICP-MS could be used as a sensitive tool for analyzing and imaging NPs as the intact NPs are transported into the ion source by a carrier gas. This method provides in situ information on particle size and number. Metarapi et al. used LA-SP-ICP-MS to image and discriminate silver NPs (AgNPs) and silver ions in sunflower roots [52]. Additionally, Wang et al. imaged AgNPs and released Ag ions in the organs of mice exposed to AgNPs using LA-SP-ICP-MS, providing a valuable tool to study the uptake, translocation, and degradation characteristics of NPs in organisms [53]. The sample introduction system is crucial for successful SP-ICP-MS analysis, and different sample introduction methods are gradually overcoming the challenges faced by standard methods [48,51]. Together with recent developments in TOF instrumentation, the sample introduction systems make it more feasible to analyze single particles by ICP-TOFMS.

Identification of NPs from Backgrounds
The complexity of signals in SP-ICP-MS surpasses that of traditional ICP-MS solution analysis. In SP-ICP-MS, a single particle is statistically introduced into the plasma at a time, producing an ion cloud that represents the elemental composition of the particle. The ion cloud is then passed through the ion optics and mass analyzer, resulting in a transient signal with typical durations between 300 and 1000 µs [6,54,55]. Separating the transient signal from the steady-state background is a critical aspect of SP-ICP-MS. An accurate measurement of the size and concentration of NPs is only possible if the NP signal can be distinguished from the background signal.
The most common strategy for detecting NPs in SP-ICP-MS is to treat NP signals as outliers from background signals. An iterative algorithm is used to discriminate a NP from the dissolved background when the NP signal exceeds the critical value (L C ), as shown in Equation (3) [56][57][58].
where L C is the critical value; σ b is the standard deviation of the background; n is the abscissa of the standardized normal distribution defined by false-positive errors (α). There is no agreement on the value of n, and different values are found in the literature, typically ranging from 3σ [59,60] to 5σ [59,61,62]. It should be noted that the n-σ threshold method assumes that the background signals in SP-ICP-MS follow a Gaussian distribution.
Some researchers have also developed critical value approaches based on Poissondistributed background signals [5]. As defined by Currie [63] and adopted by IUPAC [64], there are two detection criteria: the critical value (L c , the minimum detectable signal) and the detection limit (L D , the minimum signal level that results in reliably detected signals), which are defined by false-positive errors (α) and false-negative errors (β) [65]. Poisson distributions show a significant asymmetry for the low values of its mean. The typical value usually used for α and β is 0.05 as shown in Equations (4) and (5) [63].
where λ b is the average count rate of the background signal. However, the shape of mass spectrometric signals obtained by an analog-to-digital conversion (ADC) with high-speed digitizers in modern ICP-TOFMS instruments often does not follow a Gaussian distribution, especially for low-count signals [65,66]. This is because the use of such high-speed digitizers increases the variance from Poisson noise and causes the output signals of an electron multiplier detector to have a distribution (i.e., the pulse-height distribution, PHD) [65]. In this case, the shape of ICP-TOFMS signals can be described by a compound Poisson distribution of the measured PHD of the detector and a Poisson distributed ion arrival at the detector [65,66]. Gundlach-Graham et al. developed a Monte Carlo simulation of the TOF signals of single particle analysis by ICP-TOFMS. They proposed a new method to calculate Lc and L D , which is used as the threshold for single particle analysis by ICP-TOFMS. The new α and β values are 0.001 and 0.05, respectively [65]. As shown in Equations (6) and (7) where λ b is the average count rate of the background signal. This new method can effectively separate the overlapping background from NP distributions, resulting in a more accurate detection threshold and size measurement of NPs [65]. Moreover, it can be applied to other mass spectrometers that are equipped with electron multiplier detectors and fast digitizers.

Quantification for SP-ICP-TOFMS
Calibration is an essential step in SP-ICP-TOFMS, as it enables the accurate and quantitative analysis of NPs in solution. In SP-ICP-TOFMS, the intensity of NP signals is proportional to the mass of the NPs, and the number of events detected is proportional to the number of NPs in the sample solution. Currently, many methods are used for quantitative analysis by SP-ICP-MS [5,45].
The first quantitative method involves utilizing NP standard materials to establish a functional relationship between the particle size and signal response. However, the lack of standard NP materials of similar composition, shape, and size as those of NP samples limits its applicability.
The second quantitative method, widely used by researchers, relies on the calibration curve of a standard solution and the measurement of transport efficiency. This method assumes that the ionization efficiency difference between the standard solution and NPs can be disregarded, which is generally true. To achieve accurate quantification, measuring the transport efficiency (η neb ) is crucial. Three methods have been proposed for measuring transport efficiency, including the waste collection, particle frequency, and particle size methods [57].
The waste collection method indirectly determines transport efficiency by collecting the waste solution out of the spray chamber and calculating the actual amount of analyte entering the ICP-MS. However, this method may overestimate transport efficiency due to the presence of water vapor and residual liquid in the spray chamber [57]. The particle frequency method calculates transport efficiency by dividing the number of detected events by the total number of NPs sampled during the data acquisition time. However, determining the accurate concentration of NPs is challenging due to the lack of NP standard materials and potential NP aggregation. The particle size method involves comparing the sensitivity of an element in NPs with that obtained from the standard solution of the same element. The transport efficiency can be calculated by dividing the two sensitivities. Many studies show that the particle size method provides superior accuracy [45]. However, if there is a difference in the ionization efficiency of the NPs and standard solutions, it may introduce additional errors [67].
The two quantitative methods mentioned above do not fully utilize the benefits of the full-element analytical capability available with SP-ICP-TOFMS. The third quantitative method is the use of monodisperse microdroplets as quantitative standards of NPs [68,69]. This method, illustrated in Figure 1 [70], uses a two-sample introduction system where a microdroplet containing a known element concentration is merged into an aerosol produced Processes 2023, 11, 1237 6 of 18 by a pneumatic nebulizer and then introduced into the ICP-TOFMS. The online microdroplet calibration technique offers an automatic matrix-matching calibration of signals from individual NPs [51,71]. Mehrabi et al. determined the size and concentration of NPs by the online microdroplet calibration method while accounting for matrix effects in the single particle analysis in a single step [70]. Harycki et al. conducted a study to evaluate the effectiveness of online microdroplet calibration for quantifying nanoparticles in three organic matrices-ethanol, acetone, and acetonitrile. Despite these matrices causing signal attenuation up to 35 times and having a nebulizer transport efficiency 4 times higher than pure water matrices, the NP sizes and particle number concentration (PNC) in the organic matrices were determined with 98% accuracy [72].
the full-element analytical capability available with SP-ICP-TOFMS. The third qua tive method is the use of monodisperse microdroplets as quantitative standards of [68,69]. This method, illustrated in Figure 1 [70], uses a two-sample introduction sy where a microdroplet containing a known element concentration is merged into an sol produced by a pneumatic nebulizer and then introduced into the ICP-TOFMS online microdroplet calibration technique offers an automatic matrix-matching calibr of signals from individual NPs [51,71]. Mehrabi et al. determined the size and conce tion of NPs by the online microdroplet calibration method while accounting for m effects in the single particle analysis in a single step [70]. Harycki et al. conducted a s to evaluate the effectiveness of online microdroplet calibration for quantifying nanop cles in three organic matrices-ethanol, acetone, and acetonitrile. Despite these ma causing signal attenuation up to 35 times and having a nebulizer transport efficien times higher than pure water matrices, the NP sizes and particle number concentr (PNC) in the organic matrices were determined with 98% accuracy [72]. for the calibration of NP mass. The droplet signals are measured in the "Microd Burst Regions" of the TOF time trace. At the same time, NP-containing samples are introduce the ICP via conventional pneumatic nebulization. NP signals are analyzed from the "sp-Regio the TOF time trace, which typically lasts a few minutes in duration. Through the addition known amount of plasma uptake standard to both nebulized samples and microdroplet stand the plasma uptake rate is determined in each analysis, which is then used to calibrate the PNC plasma uptake standard is usually Cs (Reprint with permission from Kamyar M. Environ Nano. 2019, 6, 3349-3358. Copyright 2019, Royal Society of Chemistry [70]).
Compared to other single-particle calibration methods, the microdroplet calibr method offers the advantage of the elimination of matrix effects. In addition, mass q tification is not reliant on measuring the sample transfer efficiency. Furthermore, the is calibrated for each particle measurement, compensating for the instrument drif other possible negative effects during a long run. Microdroplets composed of multi-element solutions are introduced into the ICP to provide absolute sensitivities (counts· g −1 ) for the calibration of NP mass. The droplet signals are measured in the "Microdroplet Burst Regions" of the TOF time trace. At the same time, NP-containing samples are introduced into the ICP via conventional pneumatic nebulization. NP signals are analyzed from the "sp-Region" of the TOF time trace, which typically lasts a few minutes in duration. Through the addition of a known amount of plasma uptake standard to both nebulized samples and microdroplet standards, the plasma uptake rate is determined in each analysis, which is then used to calibrate the PNC. The plasma uptake standard is usually Cs (Reprint with permission from Kamyar M. Environ. Sci.: Nano. 2019, 6, 3349-3358. Copyright 2019, Royal Society of Chemistry [70]).
Compared to other single-particle calibration methods, the microdroplet calibration method offers the advantage of the elimination of matrix effects. In addition, mass quantification is not reliant on measuring the sample transfer efficiency. Furthermore, the mass is calibrated for each particle measurement, compensating for the instrument drift and other possible negative effects during a long run.

Simultaneous Quantification of Multiple Elements in a Single Particle
SP-ICP-TOFMS is a promising approach that enables the multiplexed detection and quantification of diverse metal and metal-oxide NPs [35]. ICP-TOFMS is non-targeted multi-element measurement that allows the quantification of individual particles, enabling the accurate measurement of high-throughput and in situ multi-elements. For diverse environmental samples, SP-ICP-MS still has the potential to measure real environmental samples at a level of 10 2 -10 6 NPs·mL −1 [73]. This method is practical for quantifying natural and anthropogenic nanoparticles in complex or unclear environmental matrices, which is critical for the ecotoxicological risk assessment of NPs, including engineered nanoparticles and natural nanoparticles [74][75][76].
The complexity of their composition and structure makes it difficult to determine the properties of NPs. At present, the research on composite nanoparticles mainly focuses on the core/shell structure of spherical nanoparticles with an uneven distribution of element components. Au in core and Ag in shell structure has usually been used for the evaluation of multi-element accuracy [66]. Generally, the measurement sensitivity of these elements is relatively high without much interference [29]. However, their behavior cannot be generalized and extrapolated to other composite nanoparticles, such as nano-steel (a Fe, Cr, Ni, Mo alloy used in composites) and bismuth vanadate particles (BiVO 4 ) [77,78]. Naasz et al. [31] provided a systematic and critical evaluation of the performance of ICP-TOFMS and ICP-QMS instruments for the analysis of nanoparticles used in a variety of industrial applications with complex structures and compositions. They found that only SP-ICP-TOFMS can accurately assess the elemental composition of nano-steel particles. Erhardt et al. achieved full element quantification of ice core samples in the environment by a combination of SP-ICP-TOFMS with continuous flow analysis [79]. This setup allows for accurately measuring target element concentrations over the entire mass range without losing sensitivity as the number of analytes increases.
Different elemental compositions on a single particle can often indicate the source of the particle. Multi-element analysis by SP-ICP-TOFMS has been applied to more complex samples such as air samples (e.g., road dust [80], samples from the International Space Station [81], and biomass-burning aerosol and ash [82]), water samples (e.g., wastewater [83] and rainfall [84]), and geological samples (e.g., soil [30] and minerals [32]). These applications provide guidelines for exploring how trace elements are transported into the environment. In addition, SP-ICP-TOFMS is expected to be used in medical research. Nanoparticles are increasingly used in medical products and devices, and their properties are critical for such applications. Recently, Mehrabi et al. detected magnetic iron nanoparticles by SP-ICP-TOFMS and applied it to a case study of magnetic filtering medical devices. Magnetic filtration was shown to reduce the mass concentration of detectable C/Fe 3 C NPs by 99.99 ± 0.01% in water [85].
For many analytical techniques, it is difficult to assign the particle type in a sample that contains mixtures of NPs with similar major and minor element compositions. The elemental composition of a single particle can provide much information, especially in the environment. For example, the origin of Ce-NPs is related to the presence of other rare earth elements. Based on this characteristic, Szakas et al. reported a new class of anthropogenic accidental Ce-NPs, which cannot be distinguished from natural Ce-NPs by the previous binary classification approach [86].
Another major challenge is to distinguish and quantify anthropogenic particles from naturally occurring particles [1]. The unknown multi-element NPs constitute the bulk of accidental particles as the sources are often composed of many complex elements (e.g., brake and tire wear [87]). Particles from different regions have different elemental compositions that form specific clusters, which is called "elemental fingerprints". Elemental fingerprinting can be used for tracing and migration clustering of particulate matter [82,83]. Due to elemental complexity, the need for the analysis and integration of data generated by SP-ICP-TOFMS is gradually increasing. There is no complete inventory of commercial or industrial-engineered NPs, and few data are available on the abundance of natural NPs. Establishing inventories of engineered and natural NPs depends on the development of high-throughput analytical methods. SP-ICP-TOFMS provides a direct way to build such databases.
For the huge data obtained by SP-ICP-TOFMS, some studies have developed new datum processing methods. Baalousha et al. characterized soil NPs at the single particle level in order to determine the purity, composition, association, and ratio of the elements in the NPs [88]. To identify unique metallic fingerprints in natural NPs, cluster analysis was performed using MATLAB to identify clusters/groups of natural nanoparticles with similar elemental composition and to determine their average elemental composition. This method has also been applied to the element cluster analysis of mineral dust aerosols [89]. In addition, Mehrabi et al. proposed a method that employed automated single-nanoparticle quantification and classification for an unsupervised clustering analysis of multi-metal NPs to identify unique classes of NPs based on their element compositions [83,90]. Furthermore, Holbrook et al. built a machine-learning model using Pearson correlations and unsupervised t-distributed stochastic neighbor embedding (t-SNE) to find patterns of co-occurring elements and attribute possible particle sources based on the values reported in the literature [91]. As shown in Figure 2, Pearson correlations were used to find trends in element correlations, and t-SNE projects the high-dimensional dataset (a 25-element feature set with a 1-4 element dimension target) into a lower-dimensional space of 2 dimensions [91]. The factor that often has the strongest impact is the perplexity argument in t-SNE analysis, which was tested using values of 5, 30, and 50 by the author. The performance of the data result is the space between apparent clusters and data points (a value of 30 was chosen for these samples, as shown in Figure 2B). The information obtained from the correlation and t-SNE analysis was combined with the reported document element tags to create an efficient data processing pipeline using the LightGBM multi-class classifier.
The machine learning model ultimately automates the dataset labeling and classification work, providing a fast and efficient method for inter/intra sample comparison in terms of multi-element NP elemental correlations. The pipeline can be further developed in the future to fully automate the analysis process for large particle datasets. In addition, a binomial logistic regression (LR) written by Bland et al. used the Python Sci-Kit learning module to compile a binomial LR combined with the SP-ICP-MS dataset to discriminate engineered titanium dioxide nanomaterials from natural titanium nanomaterials in soil [26]. Table 1 shows the selected applications of SP-ICP-TOFMS. The multiple elements in each particle were quantified and tracked. It was possible to develop the basis of the field of particle-by-particle geology.
[ The source of burned biomass was discussed. The source of burned biomass Zn and other crustal elements after biomass burning were more likely to be present in ash than in the biomass burning aerosol.
[82] Machine learning was developed to label and classify particle samples, providing a fast and effective method for inter-and intra-sample comparisons based on multi-element particle correlations. [91] icpTOF R After cells were stained with Ru red, it was found that the Ru content was directly related to cell volume, and cell size could be calculated by combining it with known cell shapes, leading to the calculation of the concentration of the target element in individual cells. [96] Vitesse Global surface waters and precipitation Sonication and filtration Aridus II desolator Ti, Ce, and Ag The concentrations of Ti-, Ce-, and Ag-containing NPs were presented for both surface waters and precipitation. The origin was determined from the size and composition of the nanoparticles. [97] icpTOF 2R Pt NPs Leached with diluted nitric acid and dilution Microdroplet generator introduction, control, and autosampler system Pt isotope and W isotope Using the online isotope dilution analysis method, particles were characterized with a 194 Pt/ 195 Pt ratio while monitoring 182 W/ 183 W for mass bias correction, allowing an accurate quantification at a high matrix concentration. [98] icpTOF R Soil spiked TiO 2 <500 nm particle extraction from soil and sludge, enrichment with cloud point extraction, and dilution for analysis Concentric borosilicate glass nebulizer and baffled cyclonic, high-purity quartz spray chamber Ti, Ce, Ba, Rb, Fe, Mg, Mn, Nb, Pb, and other earth-abundant elements Machine learning models of elemental fingerprinting and mass distribution were used to identify TiO 2 ENPs and NNPs in soil; this method effectively reduced the effect of a high matrix. [26] icpTOF R Soil Same as the last one Concentric borosilicate glass nebulizer and baffled cyclonic, high-purity quartz spray chamber

Ti isotope
This study is to evaluate the traceability of isotopically enriched ENPs at the individual particle level in soil and provides guidance on the isotope enrichment requirements for the quantification of ENPs from earth-abundant elements in soils. [99] icpTOF 2R Gunshot residues Settling to remove large particles and collecting the suspension's surface PFA MicroFlow pneumatic nebulizer and quartz cyclonic spray chamber

Mg-U (65 species)
The GSR particles were classified and their particle size was determined. In addition, the composition of the GSR particles was analyzed. [100] icpTOF S2 Nano-scale mineral dust aerosols (MDAs) in snow Sonication and filtration Micro FAST MC autosampler, PFA pneumatic nebulizer, and cyclonic spray chamber Al, Ti, Mn, Fe, Cu, Zn, Y, Zr, Nb, La, Ce, Nd, Pb, Th, and U The particle size and composition of MDAs in wet deposits could be effectively analyzed by SP-ICP-TOFMS, but the quantification of the particle number has a greater uncertainty. The characterization of nanoscale MDAs can be used to better understand particle dynamics in the atmosphere. [89] icpTOF S2 TiO 2 in organic matrices Sonication and dilution in ultrapure water PFA MicroFlow pneumatic nebulizer and piezoelectric droplet generator cyclonic spray chamber Cs and Ti TiO 2 NPs in the organic matrix were accurately quantified by using the online microdroplet calibration method. [72] icpTOF S2 Microplastic containing metals Aqueous dispersions Cyclonic spray chamber, quartz nebulizer with nanoparticle measurement, and pneumatic nebulizer with an autosampler for microplastic measurement and microdroplet introduction C, Ag, Au, Ce, Eu, Ho, and Lu Low m/z detection capabilities were explored by analyzing carbon and metals in both microdroplets and uniform polystyrene (PS) beads. [101] icpTOF R Anisotropic copper crystals Dilution Microdroplet introduction Cu, Au, Ag, and Pd Bimetallic physical mixtures (CuAg + CuPd) could be distinguished from multi-metallic NPs. Nanoscale structures relevant to bulk phenomena could be easily quantified and characterized with ensemble-representative reliability. [102] icpTOF 2R C/Fe 3 C NPs in whole blood 10 6 × dilution Pneumatic nebulizer and cyclonic spray chamber Cr, Fe, and Ce By analyzing the NP mass distributions, the study showed the effect of NP surface modification on the aggregation of C/Fe 3 C NPs in whole blood. Magnetic filtration was able to significantly reduce detectable particles in water.
[85] The SP-ICP-TOFMS method was developed to extract Pt NPs from LDSK, and its multi-element analysis was used to analyze the symbiotic elements of Pt in LDSK. [103] icpTOF 2R Ag NPs and algal cells exposed to Ag NPs Centrifugation and dilution PFA MicroFlow pneumatic nebulizer and quartz cyclonic spray chamber Ag isotope The ability to monitor AgNPs and intracellular silver isotope ratios was investigated. [

SP-ICP-TOFMS Isotope Analysis
Specialized ICP-MS instruments such as the Multi-Collector ICP-MS (MC-ICP-MS) and ICP-TOFMS are used to measure accurate isotopic ratios [34,109]. MC-ICP-MS can provide high-precision isotopic ratios. Also, MC-ICP-MS has been shown to be capable of detecting multiple isotopes in single particles with an excellent accuracy [41,42,110].
Although MC-ICP-MS can measure a number of isotopes simultaneously, the m/z range and number of the isotopes are limited, which is depending on the number of the detectors installed on the instrument, making it difficult to apply to extensive elemental analysis. In addition, Faraday detectors on MC-ICP-MS instruments have a slow response time [110]. Consequently, the detector may not be able to detect the transient signals generated from NPs.
An advantage of SP-ICP-TOFMS over ICP-QMS is the ability to measure isotopic ratios in single particles. Tian et al. used several types of ICP-MS (ICP-QMS, ICP-TOFMS, and MC-ICP-MS) to simultaneously detect 107 Ag and 109 Ag in single AgNPs and single cyanobacterial cells exposed to AgNPs [104]. The results showed that ICP-QMS has a poor performance in isotope ratio analysis, but accurate silver ratios can be obtained by ICP-TOFMS and MC-ICP-MS. Compared to MC-ICP-MS, ICP-TOFMS can detect almost 100% paired events of single particles [104]. Bland et al. determined 47 Ti-enriched TiO 2 NPs in soil using SP-ICP-TOFMS and evaluated the tracking ability of isotope-enriched engineered NPs at the single particle level in soil [102]. The selected applications of SP-ICP-TOFMS in isotope ratio analysis of single particles are shown in Table 1.
Processes 2023, 11, x FOR PEER REVIEW 9 of Figure 2. A machine learning model based on Pearson correlation and unsupervised T-distribute random neighbor embedding. A. Data analysis scheme for the multi-element particle analysis usin dimensionality reduction (t-SNE), Pearson correlation analysis, and the creation of an automate LGBM classifier from particle mass data in combination with reported particle fingerprint marker B. t-SNE plot of a sedimentation basin sample; the axis shows the dimension of the reduced datase The color and shape indicate specific particle types (i.e., purple circle: SrLaCe containing particle C. Starburst plots of the particle counts of CeLa-containing particles and their associated elemen Class 1: CeLa containing particles (purple). Class 2: Ce containing particles with a Ce/La ratio great than 3 (red). The machine learning model ultimately automates the dataset labeling and classif cation work, providing a fast and efficient method for inter/intra sample comparison Figure 2. A machine learning model based on Pearson correlation and unsupervised T-distributed random neighbor embedding. A. Data analysis scheme for the multi-element particle analysis using dimensionality reduction (t-SNE), Pearson correlation analysis, and the creation of an automated LGBM classifier from particle mass data in combination with reported particle fingerprint markers. B. t-SNE plot of a sedimentation basin sample; the axis shows the dimension of the reduced dataset. The color and shape indicate specific particle types (i.e., purple circle: SrLaCe containing particles). C. Starburst plots of the particle counts of CeLa-containing particles and their associated elements. Class 1: CeLa containing particles (purple). Class 2: Ce containing particles with a Ce/La ratio greater than 3 (red).

Single Cell (SC)-ICP-TOFMS
In recent years, single-cell analysis has become a growing field and has widely applied in biomedical research. Single-cell analysis is essential to reveal population heterogeneity, identify minority subpopulations of interest, and discover the unique characteristics of individual cells. Although several methods are available to analyze single cells, they are usually time-consuming and unable to detect elements in single cells [111,112]. Single cell-inductively coupled plasma-mass spectrometry (SC-ICP-MS) can be used to quantify elements in single cells. When ICP-TOFMS is used for single cell analysis, full mass spectrum can be obtained and there is no need to compromise on the analytes measured. All intrinsic elements in single cells can be measured, providing more insights into single cells [113]. This type of analysis does not require labeling or staining, as cells are detected based on their "native" elemental fingerprints [114]. Cell species can be distinguished by measuring elemental micronutrients unique to a particular cell type. For example, algal cells are rich in Mg [115], a core component of the chlorophyll pigment that is essential for photosynthesis. Therefore, the elemental composition can be used as a unique fingerprint to clearly identify different cell species.
Mass cytometry (CyTOF) is a recently developed method that combines ICP-TOFMS with flow cytometry [116][117][118]. This technology uses metal isotopes instead of fluorophores for antibodies labelling. Compared to traditional flow cytometry, the number of analytical channels in CyTOF is over 100 and the interference between adjacent channels is as low as 0.1% [117], solving the problem of fluorescence crosstalk. CyTOF has the limitation on analytical throughput (~1000 events/s [119]) but provides more complex and multidimensional data than traditional flow cytometry. With the increasing demand for high throughput in bioanalytical research, it is critical to maximize the ability to produce information about multiple markers in a single run [120].
Currently, CyTOF allows for the simultaneous detection of up to 50 metal-isotope labels on a single cell [34]. Such highly multiparametric detection has provided new insights into the complexity of biology in applications ranging from the deep phenotyping of tumors to signaling pathways of the immune system [101,121]. Antibodies are mainly used as cell staining agents in CyTOF, which fails to detect most intrinsic elements (less than 75 Th) in single cells. Bendall et al. used labeled antibodies to bind to human bone marrow cells and simultaneously analyzed up to 34 different cell parameters by CyTOF [122]. Recently, Wen et al. explored the potential of ruthenium red as a stain for single-cell analysis [96], and ruthenium red allows the elemental content to be directly correlated with cell volume to accurately calculate the intracellular concentration of target elements in single cells. By measuring metal atoms at the cellular level, the fundamental biological processes regulated by metalloproteins and metalloenzymes can be better understood [123]. Table 1 also shows the selected application of CyTOF in single cell analysis.

Summary and Prospect
SP-ICP-TOFMS is a powerful analytical technique used for the characterization of NPs. Rapid advances in ICP-TOFMS instrumentation have made it possible to detect smaller NPs with greater efficiency and accuracy. The development of SP-ICP-TOFMS methodologies has also been successful in many research fields, allowing for the analysis of individual NPs with high sensitivity and specificity. SP-ICP-TOFMS is continuously evolving to overcome its current limitations and it is expected that new generations of ICP-TOFMS will further improve their ability to detect smaller NPs with better accuracy. Furthermore, the data processing for SP-ICP-TOFMS will become more automated with the development of advanced data processing programs. In addition, other analytical methods will also be coupled with SP-ICP-TOFMS to provide more comprehensive and useful information about NPs at the same time. As a result, it is expected that SP-ICP-TOFMS will continue to expand its applications and become a valuable tool in many fields, including nanotechnology, environmental science, and biomedicine.