Strategies to enhance figures of merit in ICP-ToF-MS

Abstract


Introduction
Inductively coupled plasma mass spectrometry (ICP-MS) has become the method of choice for the characterisation of trace elements.New applications and technology have been developed and expanded the capabilities of ICP-MS vastly since its inception.[1,2] Following its commercial introduction in 1983 [3], solid sampling [4], bioimaging [5] and speciation analysis [6] were established by hyphenating ICP-MS with laser ablation (LA) and chromatography [7,8].For the former, a pulsed laser is used to raster over a solid specimen and the ablated material is transported as dry aerosol for elemental analysis via ICP-MS.In conjunction with adequate soft-and hardware, LA-ICP-MS is capable of reconstructing the two-dimensional distribution of trace elements [7].ICP-MS can further be employed to characterise particle dispersions using an approach pioneered by Degueldre in 2003 [9] known as single particle (SP) ICP-MS.Here, nanoparticles (NPs) are introduced individually into the plasma and following the generation of elemental cations, a spatially separated ion cloud can be extracted for each particle for mass spectrometry.The operation of rapid mass analysers enables the probing of each ion cloud with several data points and to count and characterise individual NP signal pulses.This provides the means to construct models on particle number concentrations (PNCs) and mass/size distributions [10].
Technological advances such as the introduction of the collision reaction cell (CRC) [11], sector field analysers [12] and tandem mass spectrometry (ICP-MS/MS) [13,14] redefined the figures of merit for ICP-MS and enabled new applications.Despite early efforts, the development of time-of-flight (ToF)based mass spectrometry was hampered due to the noisy ICP ion source, the high abundance of Ar + ions, low duty cycles and mass bias effects.[15,16] However, improvements in instrumentation made within the last twenty years have enabled modern ICP-ToF-MS instruments to set new benchmarks in multi-elemental imaging and the analysis of individual particles.Although sensitivity is lower when compared against quadrupole-based ICP-MS (ICP-QMS), ICP-ToF-MS has the crucial advantage to acquire full elemental mass spectra rapidly (at rates exceeding 35 kHz) and save binned mass spectra continuously at rates exceeding 10 kHz.[17,18] This capability promotes both the multi-elemental characterisation of individual particles via SP analysis as well as the rapid mapping of elemental distributions in micro-scaled structures.For the latter, fast washout systems are typically installed to promote rapid aerosol introduction and enable the transient resolution of individual laser pulses.[19] While this is the basis for fast imaging and increases signal to noise ratios (SNRs), it limits the maximum number of analysable isotopes when applying sequentially operating mass analysers such as quadrupoles.[20,21] The extraction of ablated aerosol plumes typically occurs in the lower millisecond range, which means that only a handful of elements can be monitored with ICP-QMS performing m/z scanning operations on a similar time scale.The case is different when using LA-ICP-ToF-MS, which records the full elemental mass spectra and therefore enables the simultaneous acquisition of elements in every pixel.[17] The rapid acquisition of full mass spectra is also a critical advantage for the characterisation of individual NPs in single particle mode.Here, quadrupoles are too slow to analyse individual ion clouds for different isotopes/elements and therefore, only one element per particle is https://doi.org/10.26434/chemrxiv-2023-326fjORCID: https://orcid.org/0000-0003-3880-9385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 detectable using SP ICP-QMS.This limits the non-target analysis of particles in unknown samples as well as the inquiry of particle compositions severely.[22,23] Both SP and LA-ICP-ToF-MS set a new paradigm for the characterisation of elemental distributions and compositions on the micro-and nanoscale, respectively.More comprehensive investigations into the interplay and associations between different elements in discrete structures are now possible.However, when compared against ICP-QMS, ICP-ToF-MS lacks sensitivity, preventing some applications such as the characterisation of small and/or heterogeneous NPs or the analysis of trace entities at high resolutions via LA-ICP-MS.This study demonstrates two strategies to optimise ICP-ToF-MS duty cycles and to increase sensitivity.These strategies were applied to characterise heterogenous nanoparticles and to map essential low level trace elements in rat brain tissue.

Instrumentation
A Vitesse ICP-ToF-MS system by Nu Instruments (Wrexham, UK) was operated at varying acquisition frequencies ranging from 35 to 180 kHz.The plasma was operated at 1.4 kW and the segmented reaction cell was operated with He and H2 flow rates between 5 and 12 mL/min.The typical nebuliser flow rate was approximately 1.35 L/min.For SP experiments, binned spectra were written to disc at frequencies between 10 and 12.5 kHz, and for LA experiments at 1 kHz.The LA system was an Analyte G2 excimer system (193 nm, Teledyne Photon Machines, Omaha, US) equipped with an aerosol rapid introduction system (ARIS, Teledyne Photon Machines, US) achieving wash out times of 15 ms.The laser was operated at 1 J/cm 2 and rastered at 250 Hz using a 35 µm square and a scan speed of 2,187.5 µm/s.
The LA-ICP-ToF-MS set-up was tuned analysing a 612 NIST "Trace Elements in Glass" standard.For SP analysis, the ICP-ToF-MS system was equipped with a concentric nebulizer (Glass expansions) and a cyclonic spray chamber.Transport efficiency was 1.7% and was determined analysing Au NP (nanoComposix, US) and ionic standards with known mass and concentration, respectively.

Software and data processing
Data acquisition was performed using Nu Codaq software (Nu Instruments).For elemental mapping, Nu Quant (Nu Instruments) was used to export mass spectra which were further processed using Pew 2 software by Lockwood et al. [24].For Mo, the isotopes 95 Mo, 97 Mo, 98 Mo and 100 Mo were accumulated and their sum was abbreviated with ΣMo.For Se, the isotopes 78 Se, 80 Se and 82 Se were accumulated to form ΣSe. For the calculation of SNRs, noise was determined as standard deviation of the gas blank and used as quotient for the signal in each pixel.For comparisons, ΣSe and ΣMo were overlaid with maps of 56 Fe and 64 Zn to visualize neuroanatomy.SP ICP-ToF-MS raw data was analysed using SPCal developed by Lockwood et al. [25] and adapted for ICP-ToF-MS data structure.Decision limits were determined using compound Poisson sampling of a lognormal approximation of the signal ion distribution [26].For each value k of the probability mass function of the expected background the sum of k lognormal distributions was computed.The cumulative https://doi.org/10.26434/chemrxiv-2023-326fjORCID: https://orcid.org/0000-0003-3880-9385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 density function of each lognormal was then calculated and weighted by the Poisson probability of k and summed.The decision limit was then determined as the first value above the desired zero-truncated quantile.The determined decision limit (α=10 -6 ) was used to differentiate background and noise from SP signals.Contiguous regions above the mean signal were accumulated for mass and size calibration.
Signals from different elements detected coincidently were considered for modelling single particle composition.Three sums of isotopes were compared against the analysis of individual isotopes: ΣYb accumulating all Yb isotopes, ΣGd accumulating all Gd isotopes, and ΣYb + ΣGd accumulating all Gd and all Yb isotopes.

Consumables and samples
Elemental standards at 1000 µg/L for ICP-MS (Single-Element ICP-Standard-Solution Roti®Star) were diluted to working conditions using ultra-pure water (18.2MΩ cm, Merck Millipore, Bedford, USA).A 80 nm Au NP standard was obtained from nanoComposix (San Diego, US).Dilutions were performed in tubes made of polypropylene.
Tissues for LA-ICP-ToF-MS were obtained as frozen specimens crytocut to a thickness of 30 μm.
UCNPs samples were manufactured following previous reports by Liu et al. [27,28]. 1 mmol (in total) lanthanide chlorides ( GdCl3, YbCl3 and ErCl3) were dissolved in methanol base at a molar ratio of 50:49:1 and then mixed with 6 ml oleic acid and 15 ml octadecene.The methanol was evaporated completely during the heating process and the mixture was kept at 150 °C for 0.5 h.The solution was then cooled to room temperature (RT) and spiked with 10 mL methanol solution containing 4 mmol NaOH and 2.5 mmol NH4F was added and stirred for 0.5 hrs at RT.The mixture was heated up again to 90 °C and kept for 0.5 h and then further heated to 150 °C and kept for 10 minutes to evaporate the methanol completely.
Finally, the solution was heated and kept at 300 °C for 1.5 hrs.The size of the UCNPs was tailored by controlling the heating rate.The UCNPs were washed three times using a mixed solution of oleic acid, cyclohexane, methanol and ethanol after the reaction solution was cooled to RT.The morphology of the UCNPs were characterised using Transmission Electron Microscopy (TEM, FEI Tecnai T20 Transmission Electron Microscope at 200 kV accelerating voltage) as shown in Figure 1.UCNPs had a hexagonal shape and mean sizes were 54.6 nm.For SP ICP-ToF-MS, UCNPs were dispersed and diluted in water and directly analysed.

Duty cycle and isotope accumulation
The ToF analyser samples only a fraction of ions extracted from the plasma.Ions in a push out region are accelerated orthogonally into the reflectron to equal kinetic energies and as such, different m/z travel at different velocities, which is the fundament for m/z calibration.The duty cycle of a ToF analyser is defined by the period it takes to complete the acquisition of a mass spectrum.When acquiring the complete mass range, considerable time is required before the next push out event can be triggered, which results in a relatively low fraction of ions analysed.While full mass ranges (5-250 amu) could be acquired at approximately 35 kHz, this speed was too fast for the processing units and spectra were therefore binned before writing data to disc at a maximum rate of approximately 12.5 kHz.The analysed mass range could be controlled using a Bradbury-Nielsen gate, further referred to as "ion blanker".The blanker was situated between the push-out region and the reflectron and consisted of a set of wires, which generated an electric field on demand when applying a current.The timely and repeated activation of the ion blanker restricted the acquired mass range by scattering ions at specific time slots.The removal of m/z ranges on both sides of the mass spectrum enabled acquisitions of restricted mass regions at higher rates.Acquisition frequency was increased from 35 kHz (full mass spectra) to up to 177 kHz for SP ICP-ToF-MS applications and to up to 180 kHz for LA-ICP-ToF-MS applications.The reduction of the acquired mass spectrum allowed to accelerate ion push out, which translated into increasing duty cycles and improved sensitivity.
Although the principle of ToF-MS is the sequential registration of different m/z at the detector, respective ions were extracted from the ion beam at the same point of time and therefore, analyses can be considered simultaneous.This provided opportunities to increase the figures of merit of polyisotopic elements.Sequentially operating mass analysers like quadrupoles can only analyse a fraction of a polyisotopic element at a time and recent studies focussed on improving SNRs by increasing the mass bandpass beyond 1 amu to transmit several isotopes simultaneously.These 'increased bandpass' methods have been successfully applied to LA-ICP-MS [29], SP ICP-MS [30] and LC-ICP-MS [31], allowing the analysis of certain trace elements.The simultaneous acquisition of various isotopes is an intrinsic feature of ICP-ToF-MS, and a similar strategy can be implemented using a facile post-analysis processing step, in which several isotopes are accumulated to analyse a higher fraction of a polyisotopic element.The possibility to select or omit certain isotopes provides control to avoid spectral interferences and using the higher resolution of the ToF analyser as well as tuning the collision/reaction cell facilitates the bypassing of polyatomic interferences.
The following sections describe both described approaches in a practical setting.Duty cycles were increased using an ion blanker to acquire restricted spectra at up to 180 kHz and isotopes of the polyisotopic elements Se, Mo, Gd and Yb were accumulated in a post-analysis processing step to further enhance SNRs.

SP ICP-ToF-MS of UCNPs
Engineered NPs are an innovative driving force in various scientific disciplines and new materials are manufactured at increasing rates.One example are upconversion NPs (UCNPs), which have applications in diverse fields due to their unique optical, electronic, and magnetic properties.They consist of a host crystal (e.g., NaYF4 or NaGdF4) that is doped with lanthanide ions to promote the consecutive absorption of low-energetic photons and the subsequent photon upconversion.Fields of applications include high-resolution microscopy, biosensing, solar energy harvesting, optical bioimaging, drug delivery vehicles and multi-modal diagnostics [32][33][34][35][36].The optical properties and nanophotonic behaviours of UCNPs highly depend on their size, structure, and stoichiometry and alas, a bottle neck for the development and application of UCNPs is the limited number of suitable analytical techniques to characterise these particles with sufficient detail.SP ICP-QMS was recently used to characterise UCNPs regarding their size and bulk composition.[30] However, ToF analysis provides additional detail by analysing elemental composition on a single particle basis.
This study focussed on UCNPs containing Gd and Yb, which suggested to centre the investigated mass region around 152 Gd and 176 Yb as lightest and heaviest isotopes, respectively, to enhance duty cycle.
However, generation of the electric field by the ion blanker created settling effects that required the expansion of the acquired mass range on the low mass end.The optimum size of the mass range (and therefore the optimum acquisition frequency) was investigated by evaluating background intensities for elements briefly after the blanking events.Background/noise levels were predicted for masses between 130 and 185 amu by recording full spectra without blanking event and factoring in the expected level of signal enhancement caused by higher duty cycles.Respective values were compared against an experimental value obtained when acquiring the isolated mass range at higher rates using the ion blanker as described in equation 1.Here, N signifies the level of background/noise and I were intensities recorded for a solution containing Yb and Gd at 35 and 177 kHz, respectively.
Figure 2 shows the noise level N over the acquired mass range, and it was evident that immediately after the blanking event (red), noise levels were significantly altered in a mass range from 130-147 amu, which corresponded to a time of approximately 1 μs (yellow).Only after this settling time did noise levels return to baseline values.Consequently, when analysing isotopes with masses between 152 and 176 amu the low mass cut off was lowered to 130 amu to ensure the absence of ion blanker induced signal artifacts in the mass regions of interest.
Given that Yb and Gd were always detected coincidently suggested the joint acquisition of both Gd and Yb isotopes resulting in additional improvements of up to 6.8 cts g/ng at 177 kHz, which corresponded to a 32-fold increase when compared to 158 Gd alone.Increasing acquisition frequency as well as accumulating signals from different isotopes were inherent with increasing noise levels.Therefore, improvements in sensitivity were not translated linearly into SNRs or limits of analysis.In SP ICP-ToF-MS, noise levels were considered using Compound Poisson statistics and the decision limit at which a signal was detected as particle increased with growing background signal.Here, an α value of 10 -6 was used to determine a decision limit, which means that from 10 6 detection intervals, one interval was false positively detected as particle.Figure 3B shows how this decision limit (red) increased with growing background levels.Here, higher fractions analysed through isotope accumulated are indicated with increasing data point sizes and 6 different acquisition frequencies were tested for each isotope and isotope accumulate.It was apparent that larger analysed fractions and faster acquisition rates increased background values and therefore the decision limit.However, resulting size detection limits for an UCNP (NaGd0.5Yb0.5F4)were determined (blue) and demonstrated how figures of merit improved despite increasing decision limits.Size detection limits decreased from 47.7 and 38.5 nm for 158 Gd and 174 Yd, https://doi.org/10.26434/chemrxiv-2023-326fjORCID: https://orcid.org/0000-0003-3880-9385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 respectively, and down to 25 and 21.3 nm when acquiring ΣYb and ΣGd.The joint acquisition of both Yb and Gd isotopes (ΣYb + ΣGd at 177 kHz) decreased the limit further to 15.2 nm.The increased limits of analysis enabled the characterisation of single UCNPs as demonstrated in Figure 4. Figure 4A shows the detection of one UCNP signal containing signals from Gd and Yb isotopes.It is evident that post-analysis isotope accumulation increased signal drastically.Figure 4B shows the effect of isotope accumulation for a small UCNP.Here, signals for 174 Yb and 158 Gd were below the decision limit, and the particle was only detected based the joint consideration of lanthanide isotopes.Figure 4C compares the fraction of particles which were detected based on individual isotopes (17.4%), on the sum of isotopes of one element (37.8%) and the sum of both Yb and Gd (44.8%).The analysis of individual isotopes would have resulted in a loss of 82.6% of particles and therefore, the combined acquisition of Gd and Yb enabled the detection a substantial number of particles, which would have been missed otherwise.The comparison of accumulated signal fractions in individual particles is plotted in Figure 4D and demonstrated a consistent stoichiometry across detected particles and a mean molar ratio of 1.86 Gd:Yb was determined on a single particle basis.Figure 4E

LA-ICP-ToF-MS for Bioimaging
The strategies to increase acquisition frequency and to accumulate isotopes via post-analysis processing were also applicable in LA-ICP-ToF-MS and promoted the mapping of elements by increasing SNRs.
The spatial resolution of LA-based mapping is primarily associated with SNRs and options to increase these are in high demand to promote both more sensitive trace analyses as well as higher resolution mapping.As a proof of principle, this study focused on the analysis of Se and Mo as two essential trace elements with low abundance in rat brain.For comparison, one brain hemisphere was analysed at 180 kHz (m/z 56-100) accumulating selected isotopes of Se and Mo, and the other hemisphere was analysed at 35.5 kHz (m/z 5-250) targeting the most abundant isotopes 80 Se and 98 Mo.ΣSe and ΣMo considered 78 Se, 80 Se and 82 Se as well as 95 Mo, 97 Mo, 98 Mo and 100 Mo and other isotopes were omitted from analysis due to impurities in the glass and interferences, which would have decreased contrasts.Figure 5A and    B show the SNRs for ΣSe and the most abundant isotope, 80 Se, at 180 kHz and 35.5 kHz, respectively.
Both increased acquisition frequency as well as isotope accumulation improved SNRs and enabled the mapping of Se in brain tissues with sufficient contrast.The same approach was applied for the mapping of Mo and is shown in Figure 5C and D. As before, significantly improved SNRs were obtained when accumulating Mo isotopes and increasing acquisition frequency.In summary, isotope accumulation and higher acquisition frequency improved SNRs and enabled the mapping of Se and Mo at 35 µm spatial resolution in the cerebral cortex and caudate putamen.The described strategies have a high utility to be https://doi.org/10.26434/chemrxiv-2023-326fjORCID: https://orcid.org/0000-0003-3880-9385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 applied in conjunction with the multi-elemental capabilities of ToF.Although increasing duty cycles are inherent with decreasing mass ranges, acquisition frequencies and mass windows can be tuned to include relevant entities.In this example, a mass range of 56 to 100 amu was chosen to enable the coincidental mapping of Mn, Fe, Cu, Zn, Co and other relevant elements in addition to Se and Mo

Conclusions
ICP-ToF-MS offers conspicuous advantages for the mapping of elements as well as for the characterisation of individual particles.However, a lack of sensitivity may be a pitfall for certain applications when trace elements are acquired with high resolution in LA analysis or when https://doi.org/10.26434/chemrxiv-2023-326fjORCID: https://orcid.org/0000-0003-3880-9385Content not peer-reviewed by ChemRxiv.License: CC BY-NC-ND 4.0 small and heterogenous nanoparticles are to be characterised via SP analysis.Here we described two strategies that can be applied to increase sensitivity and to push limits of analysis.First, restricting mass spectra to critical mass ranges using an ion blanker increased the acquisition frequency, which enhanced duty cycles and sensitivity.Second, for polyisotopic elements, simultaneously acquired isotopes could be accumulated post-analysis to consider larger element fractions.
These strategies are simple to implement and were demonstrated for the analysis of individual UCNPs via SP ICP-ToF-MS and for the mapping of Se and Mo in rat brain via LA-ICP-ToF-MS.While increasing sensitivities are accompanied with growing levels of noise, overall SNRs increased drastically and reduced detection limits.The described strategies are universally applicable to enhance limits of analysis and can be considered for high resolution mapping applications as well as the analysis of small and heterogenous NPs.

Figure 1 :
Figure 1: TEM images of UCNPs with hexagonal shape and average size of 54.6 nm.

Figure 2 :
Figure 2: The noise level following a blanking event was investigated.Theoretical background values were determined by analysing a full mass spectrum at 35 kHz and determined intensities were multiplied by the expected signal amplification factor when acquiring at 177 kHz.Theoretical values were compared against experimental values according to equation 1. Blanking created signal artifacts for a period of approximately 1 µs.

Figure
Figure 3A considers the most abundant isotopes of Gd ( 158 Gd) and Yb ( 174 Yb) and demonstrates how sensitivities were increased when accelerating acquisition frequency.Sensitivities for 158 Gd and 174 Yb increased by a factor of approximately 3.2 from 0.215 and 0.407 cts g/ng to 0.69 and 1.25 cts g/ng, respectively, and size detection limits decreased from 38.6 to 29.5 nm.

Figure 3 :
Figure 3: A shows the effects of isotope accumulation and increasing acquisition frequencies on sensitivity.B compares the Compound Poisson decision limit (red), which increases when isotopes are accumulated at increasing acquisition rates.Despite increasing noise levels, enhanced size detection limits are achieved.The size of data points corresponds to the isotope fraction analysed in a particle due to post-analysis accumulation.

Figure 4 :
Figure 4: A and B show a transient signal acquisition of an UCNP.The two most abundant isotopes of Gd and Yb are indicated and compared against the signal resulting from isotope accumulation.C compares the fraction of detection events that were recorded based on individual isotopes and accumulated isotope signals.D shows the linear association between Yb and Gd on a single particle basis, which was used to determine the average molar composition of single particles.E shows the size distribution, which was determined using the previously determined stoichiometry.
isotopes.An example is shown Figure E and F which demonstrates the joint analysis of Se (E) and Mo (F) with Fe and Zn as two neurologically relevant entities.It was apparent that the simultaneous analysis of Se and Mo with Zn and Fe enabled a correlative analysis as well as the resolution of neurological anatomy and fine structures.

Figure 5 .
Figure 5. Isotope accumulation and increased acquisition frequencies were applied to element mapping via LA-ICP-ToF-MS.A shows the distribution of Se isotopes at 180 kHz in the left hemisphere of a rat brain and B shows the distribution of 80 Se in the right hemisphere of the same brain recorded at 35.5 kHz.C and D show the same set up for the analysis of ΣMo and 98 Mo.Despite reduced mass ranges, Fe and Zn could be mapped simultaneously and enabled correlative studies as shown in E and F.