Analytical Capability of High-Time Resolution-Multiple Collector-Inductively Coupled Plasma-Mass Spectrometry for the Elemental and Isotopic Analysis of Metal Nanoparticles

We measured the Re/Os (185Re/188Os) and 187Os/188Os ratios from nanoparticles (NPs) using a multiple collector-inductively coupled plasma-mass spectrometer equipped with high-time resolution ion counters (HTR-MC-ICP-MS). Using the HTR-MC-ICP-MS system developed in this study, the simultaneous data acquisition of four isotopes was possible with a time resolution of up to 10 μs. This permits the quantitative analysis of four isotopes to be carried out from transient signals (e.g., <0.6 ms) emanating from the NPs. Iridium–Osmium NPs were produced from a naturally occurring Ir–Os alloy (ruthenosmiridium from Hokkaido, Japan; osmiridium from British Columbia, Canada; iridosmine from the Urals region of Russia) through a laser ablation technique, and the resulting nanoparticles were collected by bubbling water through a suspension. The 187Os/188Os ratios for individual NPs varied significantly, mainly due to the counting statistics of the 187Os and 188Os signals. Despite the large variation in the measured ratios, the resulting 187Os/188Os ratios for three Ir–Os bearing minerals, were 0.121±0.013 for Hokkaido, 0.110±0.012 for British Columbia, and 0.122±0.020 for the Urals, and these values were in agreement with the ratios obtained by the conventional laser ablation-MC-ICP-MS technique. The data obtained here provides a clear demonstration that the HTR-MC-ICP-MS technique can become a powerful tool for monitoring elemental and isotope ratios from NPs of multiple components.


INTRODUCTION
Mass spectrometry utilising an inductively coupled plasma as an ion source (ICP-MS) was used to measure both the sizes and number concentrations of nanoparticles (NPs) in aqueous solutions. [1][2][3] e typical settling time for mass switching (switching delay time), used in the conventional quadrupole-based MS, is about 1 ms, which is comparably longer than the time duration of a signal event obtained from a single NP (i.e., 0.2 to 0.6 ms). 4) is indicates that only a single element (isotope) can be monitored using the conventional ICP-MS system (e.g., single particle-ICP-MS: spICP-MS). [5][6][7][8][9] Although fast size distribution analysis of NPs was achieved, analytical restriction still remains, since the system is not capable of monitoring the elemental composition or the isotope ratios of the constituent elements.
Elemental analysis for multicomponent NPs (e.g., metal-lic alloy or non-metallic compounds) or structures (e.g., core-shell or polarized structures) are increasingly needed. An ICP-based mass spectrometer using a time of ight type mass spectrometer (ICP-TOF-MS) was reported to be able to monitor multiple elements (isotopes) of NPs. [10][11][12] With the ICP-TOF-MS system, multiple elements or isotopes can be monitored from the transient signals obtained from NPs. 13) Despite its ability to obtain elemental ratio data from single NPs, it should be noted that the time e ciency of signal integration (i.e., duty cycle of the data acquisition) achieved by the ICP-TOF-MS was signi cantly lower than that for a spICP-MS system. is suggests that the uncertainties in the size distribution analysis obtained with the ICP-TOF-MS system could be greater than the corresponding values for a conventional spICP-MS system. Another e ective method for detecting multiple elements or isotopes from single NPs is to use magnetic sector based ICP-MS instruments equipped with a multiple collector (MC) system. e simultaneous detection of multiple elements or isotopes can be made without lowering the time e ciencies for the signal integrations. To obtain quantitative signal intensity data from small ion sizes (e.g., 10-1,000 ions) and transient signals (e.g., time duration of 0.2-0.6 ms), high-gain ion detectors, such as electron multipliers 14) or Daly ion detectors 15) must be used. Using the multiple collector system with high-gain ion detectors, the signal intensities of multiple elements (isotopes) can be monitored with time integration e ciency better than that achieved by the ICP-TOF-MS system.
Isotopic data of the constituent elements can also provide key information concerning the origin and formation sequence of the NPs. Changes in isotopic abundance, due to the radioactive decay of parent nuclides, have been widely used as geochronology and isotope tracers. Among the isotope systematics, a 187 Re-187 Os decay system based on the radioactive decay of 187 Re with a half-life of about 40 Byrs, can be used to identify the timing of crust/mantle separation, or metal/silicate segregation of planets or the parent body of a meteorite. [16][17][18][19] Osmium is enriched in metallic nuggets, or certain Os-bearing minerals, such as iridosmine, osmiridium, rutheosmiridium, and these minerals contain virtually no Re, suggesting that the contribution of isobaric interference on 187 Re on 187 Os would be negligible. [16][17][18] More importantly, recent studies revealed that the 187 Os/ 188 Os ratios re ect the nucleosynthetic processes of heavy elements. [20][21][22][23][24] Based on the elemental and isotopic signatures derived from individual NPs, information such as the origin of the elements (Fe to U), the transport mechanism of NPs, and the formation processes of our Solar system, can be obtained. Several pioneering research studies revealed that the Os is present as metallic alloy with Ir, and some other platinum group elements (PGE), whrease Re was present in the form of separated grains associated with other chalcophile elements such as S, or Mo, obviating the larger contribution of mass spectrometric interferences on 187 Os by 187 Re, or vice versa. 25,26) However, because of the limited sizes of the these metallic grains, which are µm-sized or nm-sized metallic nuggets or particles, the elemental and isotopic analysis of the NPs has not progressed, mainly due to analytical di culties. Because of the analytical di culty in obtaining the Os isotope data from individual micro-grains, precise isotope ratio measurements of Os from bulk chondrites have been used as acceptable. [27][28][29] However, the Os isotope data obtained from bulk chondrites are averaged values obtained from the analysis of large numbers of small grains. To overcome this, we developed a new analytical technique for measuring 187 Os/ 188 Os ratios from Os-Ir bearing NPs of various sizes.
In this study, we interfaced a four-channel high-time resolution ampli er system to an MC-ICP-MS equipped with a multiple ion counting system. With high-time resolution ion counters, the signal intensities of up to four elements (isotopes) can be quantitatively monitored with a time resolution of up to 10 µs, and therefore, quantitative signal intensity data can be obtained from single signal events with short time durations, thus enabling us to monitor the elemental or isotope ratios of the NPs. Nevertheless, the isobaric interference on 187 Os by 187 Re, and 190 Os by 190 Pt, and 192 Os by 192 Pt, must be taken into account. Since Re (chalcophile element) has di erent geochemical features from Os and Ir (siderophile elements), the Re/Os abundance ratios for Ir-Os bearing minerals would be lower than 10 −4 , hence the contribution of the isobaric interferences by Re would be negligibly small. [16][17][18][19]25,26) In contrast, the geochemical characteristics of Pt are similar to Os, and isobaric interferences by Pt can cause systematic errors in the measured Os isotope ratios. Due to both the limited number of isotopes (i.e., four isotopes) and mass dispersion of the mass spectrometer, we were not able to monitor Pt isotopes using the present system setup. Hence, correction of the mass discrimination e ect was corrected by normalising the 189 Os/ 188 Os ratio. e application of the present HTR-MC-ICP-MS technique can be further extended to the precise isotopic ratio analyses of trace-elements. e measured count rates cannot re ect the true count rate of the ion signals.
is is especially true in the case of transient signals. Data acquisition with a poorer time resolution (i.e., a longer dwell time) can lower the measured count rate, resulting in under estimation of the counting loss due to dead time. Both the high-time resolution and the signal-output linearity of the ion counters are still a key issue in terms of improving data quality in the isotope ratio analysis.
EXPERIMENTAL Instrumentation e ICP-MS instrument used in this study was a multiple collector ICP-MS (Nu Plasma 2, Nu Instruments, Wrexham, UK).
e system is capable of measuring ion currents of a total of 22 isotopes using 16 Faraday detectors and six high-gain ion detectors (three electron multipliers and three Daly ion detectors). 28,29) e Faraday detectors were used for the isotope analysis of Os using the conventional laser ablation sampling technique. Hence, the detector arrangement is Faraday L3 for 185 Re, L2 for 186 (W+Os), L1 for 187 (Re+Os), Axial for 188 Os, H1 for 189 Os, H2 for 190 (Os+Pt), H3 for 191 Ir, H4 for 192 (Os+Pt), H5 for 193 Ir, and H6 for 194 Pt ( Fig. 1).
High-gain ion detectors were used for the elemental and isotopic analysis of the NPs. Because the time response of the Faraday detector (i.e., current monitoring) is not fast enough to keep up with the rapid changes in the signal intensities emanating from the NPs. 30) e quantitative signal intensity data from transient signals were detected by highgain ion detectors combined with high-time resolution data acquisition systems (we refer to this system as high-time resolution-MC-ICP-MS: HTR-MC-ICP-MS). Four hightime resolution ampli ers were applied to the MC-ICP-MS system, and a wide range of the dwell times from 1 s down to 10 µs can be selected. In this study, dwell times of 30 µs (i.e., sampling rate of ca. 33 kHz) were used for the four ion counters in all of the measurements. e resulting time resolution is su ciently rapid to permit the rapid changes in the signal intensities for the transient signals obtained from the NPs to be followed. 31) e dead time of the ion counters were measured separately by monitoring the isotope ratios obtained by a combination of high-gain ion detectors and Faraday detectors. Details of procedures for the calibration of dead time have been reported in our previous paper. 14)

Corrections of detector gain
Among the six high-gain ion detectors, four ion detectors of various combinations were applied for the simultaneous detection of ion signals of W, Re, and Os isotopes from the NPs (Fig. 1). e gains of the high-gain ion detectors were dependent upon the high voltages applied to the multipliers and discrimination settings for individual ion counters. e relative di erence in the detector gain was calibrated using 182 W, 183 W, and 184 W isotopes. e gain calibration consists of two-sequential isotope ratio measurements using di erent combinations of two high-gain detectors: (1) 183 W (Daly 1: D1)/ 182 W (Daly 2: D2), and 184 W (Electron Multiplier: EM0)/ 182 W(D2) ratio measurement, and followed by (2) 184 W (D2)/ 182 W (EM3) ratio measurements (Fig. 1). e relative gains of the four high-gain ion detectors were calculated by normalizing the measured 183 W/ 182 W, and 184 W/ 182 W ratios being 0.5428 and 1.172, respectively, using a simple linear set up. 32) Details of the instrumentation and operational settings were listed in Table 1.

Data reduction
With the present system setup, signal intensity data for 185 Re, 187 (Re+Os), 188 Os, and 189 Os isotopes were simultaneously measured every 30 µs (dwell time) for 30 s. e resulting signal intensity pro les were then used to calculate the signal intensities (total counts of each isotope) of signal events using our in-house so ware "NanoQuant." is so ware is used to calculate the total ion counts of the individual signal events. e signal events originating from the NPs were de ned as continuous signal counts (>1 counts) for three time slices (i.e., time duration of 90 µs). e total signal counts of the signal events were calculated by summing up the signal counts obtained for each time slice, hence the end of the signal events were de ned by continuous signal counts of zero for three time slices (i.e., a time duration of 90 µs). To avoid the duplicated integration of two or more signal events (mainly due to the overlap of the two or more signal events reaching the detector concurrently), the signal intensity pro le of all events (signal intensity plotted against time) were visualized automatically by the NanoQuant so ware. Data showing potential overlaps of multiple signal events were not used in further calculations.
Corrections for counting loss due to dead time were made on individual signal counts based on a conventional non-extended protocol. For the isotope ratios, the signal intensity of the isotopes was calculated by integrating the total ion counts for individual signal event using the NanoQuant so ware. e resulting signal intensity data for each isotope were then used for the calculation of  Background correction e background count rate for 185 Re, 187 (Re+Os), and 188 Os were all <20 cps. e signal count was less than 1 count per single data acquisition sequence when the dwell time was set to 30 µs. is suggests that the contribution of Os memory on the instrument and blank Os in the analysis solution will be almost identical to the electrical noise.

Samples and sample preparation
For the Re/Os and Os isotope ratio measurements, three naturally occurring Ir-Os samples containing minerals from Japan, Canada, and Russia were used. Osmiridium and iridosmine is a natural alloy of osmium and iridium, with traces of other platinum-group metals, and ruthenosmiridium is a natural alloy of ruthenium, osmium, and iridium.
(a) Hokkaido (ruthenosmiridium): Ruthenosmiridium was de ned as a new species of a natural alloy of Ru-Os-Ir. 35,36) In the chemical compositional diagram for minerals, the compositional eld for ruthenosmiridium or rutheniridosmine was restricted by the revised nomenclature for alloys in which Ir was the dominant element. 37) e ruthenosmiridium samples from the Sorachi River, Hokkaido, Japan, were found as placers, and were considered to be derived from ultrama c rocks in the Kamuikotan metamorphic belt. e size of these samples were 1-2 mm.
(b) British Columbia (osmiridium): Osmiridium crystallizes in the isometric system and exhibits an iridium crystal lattice structure. Osmiridium occurs as an irregular shaped material and various sizes ranging from µm to mm scale, sometimes as concretions of grains and crystals are common. 38,39) e osmiridium samples from British Columbia, Canada, are associated with young ultrama c rocks in the northern part of the Garlock fault, and source of the Ir-Os bearing minerals could be the Tulameen ma c-ultrama c complex. e typical size of the sample is 0.05-0.5 mm.
(c) Urals (iridosmine): e placers of the Urals are associated with dunites in the Hercynian ophiolite belt. Since the precise isotopic data for iridosmine from the Urals Nevyansk were reported by Allègre and Luck,19) and also by us, 40) this sample was used for the inter-laboratory comparison of the Os isotopic data. e typical size of the sample is about 0.5-0.8 mm.
Sample grains used in the Os isotopic analyses were mounted using a resin (UK-M, PRESI, Eybens, France), and the sample surface was polished successively using #8000 polishing paper (Scotch Bright Polishing Sheet, 3M, St. Paul, MN, USA) and diamond paste (#10,000) to produce a at surface. Photographic images of the sample are shown in Fig. 2.
NPs of the samples stated above were produced by laser ablation by using a 4HG (fourth harmonic generation) Yb:KGW femtosecond laser operating at a wavelength of 260 nm. A uence of 4-8 J cm −2 and a repetition rate of 1 kHz were used. Laser ablation was carried out for 5 min by rastering the laser beam with a speed of 10 µm/s. e sampling areas are shown as white squares in Fig. 2. e laser induced sample aerosols were collected by bubbling air through a sample in 5 mL of deionized water. e number concentrations of the resulting NPs were about 10 5 -10 6 particles/mL. Since higher number concentrations will induce overlaps of two or more signal events, the number concentrations of the NP solutions were adjusted to 10 5 particles/mL using deionized water prior to the HTR-MC-ICP-MS measurements. e solutions were then introduced into the ICP using a Micromist nebulizer (Glass Expansion, Port Melbourne, Australia).

Os isotope ratio analysis using conventional LA-ICP-MS
Prior to the isotopic ratio measurements on the NPs, reference Os isotope ratios ( 186 Os/ 188 Os, 187 Os/ 188 Os, 189 Os/ 188 Os, 190 Os/ 188 Os, and 192 Os/ 188 Os) for the three Ir-Os bearing minerals were separately measured using the conventional laser ablation-MC-ICP-MS technique (LA-MC-ICP-MS) using the same laser ablation system for NP generation. e ion currents for 185 Re, 186 Os, 187 (Re+Os), 188 Os, 189 Os, 190 (Os+Pt), 191 Ir, 192 (Os+Pt), and 193 Ir, isotopes were measured by Faraday detectors equipped with an ampli er using a negative feedback register of 10 11 ohm. e 190 Os and 192 Os isotopes might be interfered by the 190 Pt and 192 Pt isobars, respectively, and therefore, the mass discrimination e ect found in the MC-ICP-MS system was corrected by normalizing the 189 Os/ 188 Os to be 1.216 using the exponential law. 17,32) Hence, no correction of the isobaric interferences on 190 Os and 192 Os by 190 Pt and 192 Pt respectively were made in this study. e analytical session consisted of 30 s on-peak baseline measurements without laser ablation, followed by a 60 s data acquisition with laser ablation using a time-resolved analysis (TRA) mode. A dwell time of 0.2 s was adopted. A er the subtraction of the baseline signals, signal intensity data of 192 (Os+Pt) that was >1 V were used for calculating the isotope ratios. e resulting Re/Os and Os isotope ratios for three Ir-Os bearing minerals are summarised in Table 2. e measured 185 Re/ 188 Os ratios for Hokkaido, British Columbia, and Urals samples were <6.5×10 −5 , <2.5×10 −4 , and 10 −4 , respectively. e contribution of isobaric interferences on 187 Os by 187 Re was smaller than 10 −2 .

Size distribution analysis of Ir-Os bearing NPs
Sizes of the NPs were calculated based on the signal intensities of single events of the samples and those of the standard size NPs. Generally, size calibration is carried out using standard NPs having the same chemical composition as the unknown samples. 3,9) However, there are no commercially available size standard NPs available for the Ir-Os alloy. Hence, the size analysis was based on a calibration using the commercially available Au NP (sizes 60±6 nm) from nanoComposix (San Diego, CA, USA) under the assumption that the elemental sensitivity of Au was identical to that for Ir and Os. e transmission e ciencies of Au and Pt (from sample to ion detector) were 0.058±0.006% and 0.052±0.006%, respectively, suggesting that there is a small di erence in the transmission e ciencies between Au and Pt when the same system setup and operational settings are used. e similarity in transmission e ciencies of Au and Pt can also be demonstrated when conducting size calibration of 50 nm Pt NPs using the 50 nm Au NPs (47.8±1.8 nm: nanoComposix). e resulting size of the 50 nm Pt NPs was 50.0±4.9 nm, in good agreement with the data sheet provided by the manufacturer. Another important assumption is that there were no variations in the Ir/Os abundance ratios among the produced NPs. Due to the limited mass dispersion of the highgain ion counters, it was not possible to simultaneously determine the 185 Re/ 189 Os, 187 (Re+Os)/ 189 Os, and 191 Ir/ 189 Os ratios. Our conventional LA-MC-ICP-MS analysis on the Ir-Os bearing minerals samples showed that Ir/Os abundance ratios varied from 1.5 to 1.7 with the analysis spot, suggesting that Ir and Os is distributed heterogeneously within a single grain. e e ect of the variations in the Ir/Os ratio on the measured sizes of the NPs was smaller than 5%.
is is much smaller than the variations in the size distribution of the resulting Ir-Os bearing NPs, so we assumed that Ir/Os ratio for the NPs would be 1.6. Figure 3 illustrates the measured size distribution of the laser induced NPs. e total number of particles, obtained from three Ir-Os bearing minerals, having di erent totals numbere of particles, were normalized to 100 in order to obtain a better comparison. e resulting size distribution for the Hokkaido and Urals samples showed that the frequency of the particles decrease with increasing size. A nearly linear correlation of the number with the diameter (Fig. 3: double logarithmic scale) suggests that the resulting particle size distribution re ects Gaudin-Schumann or Roshin-Rammler distributions. 41,42) e largest particles were <80 nm for both the Hokkaido and Urals samples. On the other hand, the British Columbia samples showed a unique distribution pattern, with the highest sizes at about 60 nm. In addition, the size distribution data for the British Columbia sample showed the presence of large-sized particles being >80 nm. is contrasts with the size distributions for the Hokkaido and Urals samples, and this di erence could be attributed to the crystallographical features of the sample. As shown in Fig. 2, ruthenosmiridium from Hokkaido and iridosmine from the Urals have well crystalline morphologies, whereas there are vacancies in the osmiridium sample collected from British Columbia, suggesting that the small grains may have undergone agglomeration in the sample. is implies that the laser induced shockwave causes the release of small pieces or fragments from the osmiridium samples during the laser ablation procedure.
Under the assumption that the overall transmission efciency of Os is identical to that of Au, the particle size detection limits for Os achieved by the present system setup is about 10 nm. is is twice as large as that for other elements, such as Au, or Ag. e poor detection limits for Os are mainly due to the low isotopic abundance of the moni- tored isotopes ( 188 Os: 13.3% 33) ). If a higher isotopic abundance of Os (e.g., 192 Os) was monitored, the resulting size detection limits could have been improved.

Os/ 188 Os ratio analysis using HTR-MC-ICP-MS
e 187 Os/ 188 Os ratio measurements were carried out on three NPs prepared from Ir-Os bearing minerals (Hokkaido, British Columbia, and Urals) using the HTR-MC-ICP-MS system developed in this study. e total number of particles analysed in this study was 1300 for Hokkaido, 1200 for British Columbia, and 160 for the Urals. e number of particles for the Urals was signi cantly low, simply because of the smaller size of the Urals iridosmine sample (100×300 µm). e resulting 187 Os/ 188 Os ratios are listed in Table 2, together with the Os isotope ratio data obtained by the conventional LA-MC-ICP-MS technique described in the earlier section. In Fig. 3, the measured 187 Os/ 188 Os ratios for the individual grains are plotted against the size of the particles, showing that the variations in the measured 187 Os/ 188 Os ratios became smaller with increasing particle size. is suggests that the precision of the isotope ratio measurements can be controlled by the counting statistics on the analytes, which is consistent with our previous report. 41 e resulting precision of the 187 Os/ 188 Os ratio measurements were 10-20%, slightly poorer than those achieved in the 195 Pt/ 194 Pt ratio measurements from NPs of similar sizes. 43) In the case of the 195 Pt/ 194 Pt ratio measurements, a precision of better than 10% can be achieved from the Pt NPs with sizes being >50 nm. For the 187 Os/ 188 Os ratio measurements, the deterioration of the precision of the measurements can be attributed to a larger contribution of counting statistics originating from the smaller isotopic abundances for both 187 Os (1.6%) and 188 Os (13%). e possible variations in the 187 Os/ 188 Os ratios estimated, based on the counting statistics (±1σ and ±2σ) are also shown as grey-coloured curves in Fig. 4. Most of the data points for 187 Os/ 188 Os ratios were plotted within the range of ±2σ variations, suggesting that the precision of the 187 Os/ 188 Os ratio measurements was principally controlled by the counting statistics of the signals. If that is the case, improvements in the transmission e ciency of the mass spectrometer (i.e., sensitivity) is the most important factor in terms of obtaining better precision in the isotope ratio measurements from the NPs. Based on the results gathered in this study, a variation of 187 Os/ 188 Os was observed for minerals of di erent chemical compositions. Variations in the 187 Os/ 188 Os ratios in excess of 50% may possibly include information that re ects di erent nucleosynthetic processes (e.g., rapid, slow, or spallation processes). erefore, the HTR-MC-ICP-MS system developed in this study could be applied to understanding the origin of the heavy elements in small metal nuggets or NPs in meteorites. 21,25,26,44,45) CONCLUSION 185 Re/ 188 Os, 187 Os/ 188 Os, and 189 Os/ 188 Os isotope ratio measurements were made on a series of Ir-Os NPs using a high-time resolution-MC-ICP-MS (HTR-MC-ICP-MS) system. e Ir-Os bearing NPs were produced by laser ablation using a UV femtosecond laser (wavelength of 260 nm, and a pulse duration of 230 fs). e size of the resulting NPs was calibrated with Au NPs of known sizes based on two assumptions: (a) the transmission of Os was identical to Ir and Au, and (b) there were no variations in the Os/Ir ratio among the individual NPs. e resulting sizes of the laser induced NPs ranging from 20-160 nm was within the range of the present HTR-MC-ICP-MS technique. e resulting precisions of the 187 Os/ 188 Os ratio measurements were 10-20% (2σ m ), which were principally controlled by the counting statistics of the 187 Os and 188 Os signals. Despite the lower precision values obtained for the isotope ratio measurements, the measured 187 Os/ 188 Os ratios were in good agreement with those obtained by the conventional laser ablation-MC-ICP-MS. e precision of elemental or isotope ratio measurements can be further improved using an ICP-MS system with an enhanced sensitivity. e data reported herein demonstrate that the HTR-MC-ICP-MS system has the potential for being an effective tool for collecting elemental and isotopic ratio data from NPs. With the elemental data obtained from individual NPs, quality control of the industrial raw materials for cosmetics, catalysis, electronic devices, or semiconductor materials, can be made with a high analysis throughput. Moreover, the sensitive detection of NPs from biological tissue samples can provide key information concerning the toxicity of the metal NPs. More importantly, Pb isotope ratios for ne airborne particles, including PM0.1, can be used for both the fundamental studies and for understanding the transport mechanism of particulate matter.