Laser ablation-inductively coupled plasma-mass spectrometry using a double-focusing sector field mass spectrometer of Mattauch–Herzog geometry and an array detector for the determination of platinum group metals and gold in NiS buttons obtained by fire assay of platiniferous ores

Abstract

This paper explores the potential of laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) for the determination of Au, Ir, Pd, Pt, Rh and Ru in NiS buttons obtained by fire assay of platiniferous ores using a new type of ICP-MS instrument equipped with an array detector Mattauch–Herzog spectrometer. The method evaluated comprises the NiS fire assay of a representative amount of the ore sample (40–75 g), grinding of the NiS buttons subsequently obtained, pelleting of the resulting powders using polyethylene wax as a binder, and LA-ICP-MS analysis of the sample using in-house matrix-matched standards for calibration. The use of this new ICP-MS device proved very beneficial in this context, offering a remarkable level of precision (2–3% RSD for the most abundant analytes, Pt, Pd, Rh and Ru, and 6–11% RSD for Au and Ir) owing to its simultaneous capabilities and its extended linear range, which enabled an improved performance of the internal standard (⁶¹Ni). Moreover, the low level of argide-based interferences and the detection power of the instrument provided low limits of detection (10 ng g⁻¹ level) even for those elements that could be potentially affected by ArNi⁺ and ArCu⁺ overlap in this particular matrix (Pd, Rh, Ru). The accuracy finally obtained (90% of results within 10% of the reference value, and 70% of results within 5%) is fit-for-purpose for this application. The resulting method can be thus considered very attractive as a faster and greener strategy for the control of these types of samples in the platinum group metal industry, circumventing the need for the cumbersome and environmentally unfriendly digestion procedures currently employed.

---

1. Introduction

Because of their resistance to oxidation, catalytic activity and physical properties, platinum group metals (PGMs) and Au are extensively used in many applications. PGMs and Au are found in platiniferous ores as sulfides, or are associated with base metal sulfides within a silicate or chromite matrix.¹ A high degree of accuracy and precision is required for analysis of these ores owing to the high value of these elements, but, unfortunately, this type of analysis is very challenging because of (i) the low concentration at which the PGMs occur (typically ≤ 10 μg PGMs g⁻¹); (ii) their heterogeneous distribution within the matrix; (iii) the difficulties in bringing these elements into solution in a quantitative way; and (iv) the presence of other elements in the ores at considerably higher concentration, which can give rise to significant problems for PGM determination (e.g., matrix effects and spectral overlaps).^2–4 Fire-assay procedures allow these drawbacks to be circumvented, as they permit the use of a substantial amount of ore as sample (tackling the problem of heterogeneity) and involve extraction of the target elements into a molten collector (most often Pb, NiS or Cu).^5–7

Of the different types of fire assay procedures that can be deployed, the NiS fire assay technique is recognized as the most reliable approach.⁵ In the NiS fire assay technique, a relatively large amount of sample (tens of grams) is mixed with a flux and fused. Most of the matrix components (gangue) react with flux reagents to form a slag, while the PGMs and Au are concentrated in the molten NiS matte phase. After cooling and solidification, the two phases can be mechanically separated from one another and the NiS button containing the analytes is further prepared for subsequent analysis. The traditional approach now aims at digestion of the NiS buttons, bringing the analytes into an aqueous solution, prior to analysis, usually by means of inductively coupled plasma-mass spectrometry (ICP-MS).⁵ This procedure is reliable, but quite time-consuming (4–6 hours) and costly.⁴

Furthermore, there is one important additional disadvantage associated with the dissolution of NiS buttons, when used for routine laboratory analysis at an industrial level. This procedure results in the generation of an enormous quantity of waste HCl solution containing high levels of Ni, which is very costly to dispose of. Even worse, it also produces a large amount of very toxic H₂S (g), which has to be removed by scrubbing with NaOH solution and oxidized using H₂O₂, which eventually leads to an enormous amount of Na₂SO₄ solution that it is also necessary to dispose of. This is a case where the implementation of a “greener” analytical approach will provide important benefits from both a financial and an environmental point of view. This aspect, the development of greener procedures, is still overlooked in atomic spectrometry, but that situation may well change in the future.⁸

It is probably safe to assume that carrying out the NiS fire assay is necessary, in order to guarantee the required level of accuracy. However, a much greener strategy could be to use an analytical technique that allows for the direct analysis of the resultant NiS buttons, thus circumventing the need for any wet chemical preparation. This approach will solve the environmental issues mentioned before resulting in practically no waste, since the solid NiS samples can easily be recycled in a smelter without any additional sample preparation. In this way, both the PGMs and Ni can be recovered.

Among the different techniques that could be used for this purpose, laser ablation (LA) coupled to ICP-MS is one of the most powerful ones.^9–12 This technique offers sufficient sensitivity for this type of application and requires only minimal sample preparation of the NiS buttons (grinding and pelletization).^2,13 It is however necessary to use suitable instrumentation and/or appropriate strategies to cope with spectral overlap.^3,14–16 In this regard, previous results reporting on PGM determination in geological materials by means of NiS fire assay followed by LA-ICP-MS analysis have not always been successful.^17–20 A previous work from our research group showed promising performance for this type of application based on: (i) the grinding of the NiS buttons to achieve a more homogeneous material; (ii) the evaluation of different binding agents, leading to the final use of polyethylene wax; (iii) the use of line profiling ablation for the sampling of a representative portion of the button; and (iv) the use of the dynamic reaction cell pressurized with NH₃ to overcome spectral overlaps that hamper the accurate determination of several PGMs of medium mass (namely Rh, Pd and Ru, due to the formation of ArNi⁺ and ArCu⁺). This approach provided good accuracy, precision values in the 5–10% RSD range and limits of detection around 10 ng g⁻¹.²¹

Despite this satisfactory performance, it would still be desirable to improve the precision of the method. Moreover, careful optimization of the reaction cell conditions (NH₃ flow and RPq setting) was required, because it was necessary to remove ArCu⁺ and ArNi⁺ ions, but also to prevent the formation of clusters such as Cu(NH₃)_0–3⁺ and Ni(NH₃)_0–4⁺ that would otherwise lead to new spectral interferences.

A new type of ICP-MS instrument has been introduced into the market recently by SPECTRO Analytical Instruments GmbH. This instrument is based on the use of a double-focusing sector field mass spectrometer of Mattauch–Herzog geometry that focuses ions of different masses onto the same focal plane but spatially separated.²² This mass spectrometer enables simultaneous monitoring of the entire elemental mass spectrum. While in the oldest examples of mass spectrometers this was accomplished via ion-sensitive emulsions, this instrument uses an array detector that covers the mass range from 5 to 240 u in 4800 channels (Faraday strips) for this purpose. Each of these channels is equipped with two circuits, providing two levels of gain, thus resulting in an expanded linear range. The obvious benefit of this instrumental configuration is that it permits monitoring of all the nuclides of interest in truly simultaneous fashion.

This instrumental concept is based on the pioneering work of Cromwell and Arrowsmith²³ and of Burgoyne et al.²⁴ Since these works, several improved prototypes have been developed and evaluated by the research groups of G. M. Hieftje, M. B. Denton and D. W. Koppenaal,²⁵ using 31-channel,²⁶ 128-channel,^27,28 512-channel²⁹ and 1696-channel array detectors,³⁰ and the benefits of such an instrument for the monitoring of transient signals such as those generated by laser ablation³¹ or electrothermal vaporization³² have been demonstrated.

It is the purpose of this work to evaluate the performance of laser ablation coupled to this new type of commercially available ICP-MS, based on an array detector, Mattauch–Herzog spectrometer, for the direct determination of Au, Ir, Pd, Pt, Rh and Ru in NiS buttons obtained by fire assay of platiniferous ores. The performance of such a device in terms of accuracy, precision, potential to deal with spectral overlaps from argide-based polyatomic molecules, spectral resolution and limits of detection will be critically evaluated in the context of the application mentioned above. Some indications that may permit the extrapolation of the performance of this instrument to other contexts will also be provided, whenever possible.

2. Experimental

2.1. Instrumentation

A frequency-quintupled Nd:YAG laser system operating at 213 nm (LSX 213 from CETAC, USA) coupled to a double-focusing sector field Mattauch–Herzog ICP-mass spectrometer equipped with a Faraday strip array (Ion 120)³³ with 4800 independent channels was used for analysis of the NiS buttons. This new type of instrument, named SPECTRO MS, has been made commercially available very recently (SPECTRO Analytical Instruments GmbH, Kleve, Germany).

2.2. Samples and standards

2.2.1. Samples and standards. For samples, eight reference materials (RMs), with certified values for all PGMs (except for Os) and Au, were selected. SARM 7B, SARM 65 and SARM 73 were commercially available from MINTEK (South Africa, SA). AMIS 07 and AMIS 10 were available from African Mineral Standards (SA). Finally, MR RF05/1, UG2 RF05/1 and MR FW05/1 were reference materials made commercially available by Ore Research & Exploration Pty Ltd. (Australia). They are platiniferous ores from the Merensky or UG2 reefs, except for MR FW05/1, which was obtained from the Merensky footwall (area below the main reef, still rich in PGM).

The PGM sponges of >99.99% purity were obtained from the Precious Metal Refiners (SA), while Au grain was obtained from the Rand Refinery (SA). The binder for blending with the pulverized NiS buttons, Ceridust 130 (polyethylene wax), was provided by Clariant (SA). Ni and Cu powder from the OM Group was provided by Protea Industrial Chemicals (SA). Other flux reagents such as sodium tetraborate, sodium carbonate, silica and sulfur were all of industrial grade and were supplied by Protea Industrial Chemicals.

2.3. Procedure for the analysis of the NiS buttons by means of LA-ICP-MS

2.3.1. Preparation of NiS buttons by fire assay of the samples. The NiS buttons were prepared by weighing appropriate masses (based on grade) of the reference material (typically 40–75 g) and adding 330 g of a flux containing Ni, Cu, Fe, S, SiO₂, Na₂B₄O₇, Na₂CO₃, KNO₃ and CaF₂. Details of this procedure can be found elsewhere.²¹ The final NiS buttons were typically 20–23 g in mass, containing 64% Ni, 32% S and 3% Cu (added with the flux).

2.3.2. Preparation of NiS pellets. Once the NiS buttons were obtained, they were subsequently crushed and ground. Afterwards, an aliquot of 9 g of the NiS powder was mixed together with 1 g of binder (polyethylene wax) and the resulting mixture was pelletized. Details of this procedure can be found elsewhere.²¹

2.3.3. Preparation of the standards used for calibration. As a result of the relatively simple matrix composition of the samples analyzed in this work, matrix-matching was feasible. The preparation of the standards was carried out as described elsewhere.²¹ The resulting NiS standards contain practically the same amount of Ni, S and Cu as the samples, and a known amount of PGMs and Au, covering the range of concentrations in which the analytes are expected to be present after the fire assay procedure.

2.3.4. Analysis by means of LA-ICP-MS. The parameters used for analysis are summarized in Table 1. Prior to the analysis, the surface of the samples was cleaned with ethanol. The pellets were analyzed using the line profiling mode (1200 firings per sample), resulting in a sample throughput of approximately 3 minutes. Five replicate measurements were considered per sample (10 seconds of ablation each) and their average was taken as a representative value. The signal intensity for ⁶¹Ni was used as an internal standard in all cases.

Table 1 Optimum conditions selected for LA-ICP-MS measurements

Cetac LSX 213 nm laser ablation system
Output energy	∼2.0 mJ on sample surface
Repetition rate	20 Hz
Beam diameter	200 μm
Translation speed	0.1 mm s^−1
Distance ablated per replicate	6 mm
Cell volume	∼60 cm^3
SPECTRO MS instrument
RF power	1410 W
Ar plasma/auxiliary gas flow rates	12/2.2 L min^−1
Carrier gas flow rate	0.89 (Ar) + 0.30 (He) L min^−1
Data acquisition parameters
Acquisition mode	Transient
Sampling rate	10 Hz
Running time per sample	180 s

---

3. Results and discussion

3.1. Sample preparation

The goal of this work is to propose a straightforward procedure that also enables accurate and precise results to be obtained. In principle, when carrying out a NiS fire assay, a large (and, thus, likely representative) amount of the target material (40–75 g) is sampled. However, previous works have shown that the resulting NiS buttons are not very homogeneous,^34,35 such that poor precision would be obtained when analysing them directly. In a previous work it was demonstrated that grinding the NiS buttons, mixing them with a known amount of binder (polyethylene wax that contains 10% of PTFE, added in a 1 : 9 binder/sample ratio) and pelletizing is almost essential for achieving reliable results.²¹ This aspect does not reduce sample throughput very significantly, as it takes only 10 minutes to do so. Moreover, it has to be mentioned that grinding the NiS buttons is also required when digestion of the samples is opted for. The resulting pellets show good physical properties for LA-ICP-MS analysis (good cohesion and very flat surfaces).

The average level of Ni (64%) in both samples and matrix-matched standards was observed to be very constant (at the mm level) by means of XRF monitoring (0.2% RSD). This aspect is noteworthy, since it permits the Ni signal to be selected as the internal standard during LA-ICP-MS analysis, without the need to carry out an independent determination of the Ni content for each sample.

3.2. LA-ICP-MS monitoring of the samples

Once NiS-based pellets originating from platiniferous ores were obtained according to the procedure described before, the ablation and plasma conditions were optimized aiming at the best signal-to-blank ratios. Optimum results were achieved with the maximum beam diameter (200 μm) and repetition rate (20 Hz), 0.3 L min⁻¹He (only gas going through the ablation cell) plus 0.89 L min⁻¹ Ar (admixed only after the ablation cell) and laser energy values of 1.5 mJ or higher. The line profiling ablation mode was preferred as discussed before. The conditions finally adopted are displayed in Table 1.

3.2.1. Characteristics of the LA-ICP-MS spectrum. As discussed in the Introduction, one of the most attractive features of the double-focusing sector field Mattauch–Herzog mass spectrometer used in this work is that it permits the truly simultaneous monitoring of the entire elemental mass spectrum, from 5 to 240 u. Fig. 1a shows a typical spectrum obtained when a NiS pellet is ablated. As can be seen, the spectrum is very rich in lines in the low mass region and the baseline shows some structure in this region. The main reason for this is that the detector is not a pulse counting device but it is based on an operational amplifier designed for charges delivered to the surface of the detector. Depending on the working parameters of this amplifier, the baseline can be positive or negative (“DC offset”). This offset can also be influenced by very large signals at the same or adjacent pixels, as verified by the baseline shift in the low mass regions adjacent to the ¹⁶O or ⁴⁰Ar signals, very similar to oversaturation effects seen in optical semiconductor multichannel detector devices, e.g., CCDs or CMOS. However, since this baseline is measured truly simultaneously with the analytical signal, the selection of an appropriate setting of integration ranges and baseline correction points should eliminate the baseline variation from the measured signal. In fact, the software allows the operator to zoom into any part of the spectrum, determine which elements are present based on the relative abundance of their different isotopes and properly baseline-correct the signal for every mass of interest using various types of algorithms for signal deconvolution.

Fig. 1 (A) Complete mass spectrum obtained upon truly simultaneous LA-ICP-MS monitoring of a ground and pelletized NiS button standard, prepared as described in Section 2.3.3, using the high gain setting of the detector. (B) Complete mass spectrum obtained upon LA-ICP-MS monitoring of NIST SRM 610 trace elements in glass, using the same conditions as in (A). (C) Mass spectrum corresponding to the exact same measurement shown in (A), but now displaying the data obtained with the low gain detector setting.

The spectrum shows the presence of Au and the PGMs of interest in the pellet. Also, the presence of Pb is evident, which is not surprising because the NiS pellets are prepared in the same laboratory where Pb fire assay is also carried out. Very high signals for Ni and Cu are also observed, as could be anticipated considering the expected level of these elements (see Section 2.3.1). Fig. 1b shows a spectrum obtained upon LA-ICP-MS monitoring of a NIST 610 glass under the same conditions. The corresponding result may serve as a reference as this sample has already been monitored with many types of LA-ICP-MS set-ups. As can be seen, in this case, the spectrum is very rich in lines since this glass is spiked with tens of elements at the 500 μg g⁻¹ level, for which signal intensities in the 1 000 000 charges per second range (the term counts per second is not accurate for this type of detector) are obtained for the most abundant isotopes.

Fig. 1a and b show the spectra as observed using the high gain setting (maximum sensitivity) but, for every pixel of the detector, it is also possible to select a low gain setting (post-acquisition), thereby reducing the sensitivity by roughly a factor of 800. Fig. 1c shows the results of the same LA-ICP-MS process as Fig. 1a, but now using low gain mode. Now, the spectrum becomes very clean, with an almost perfect baseline, and only those elements present at the % level (Ni, Cu, S) and the ones found in the gases (O, N, C or Ar) give rise to a significant signal. Please note that this device permits the monitoring of all the nuclides in the 5 to 240 u range, including ⁴⁰Ar⁺. This low gain setting is recommended for monitoring those elements found at high concentrations, for which the corresponding detector channels become saturated in the high gain mode. Thus, for ⁶¹Ni⁺, which was used as the internal standard in the current work, the low gain signal was used throughout this work, while for Au and PGM, the high gain signals were always chosen. It has to be indicated that both high and low gain signals are always recorded, so both of them are always available for the user, and the decision as to which one is more suitable does not have to be made a priori. In the same way, it is not necessary to decide a priori which nuclides should be selected for calculating the concentration of each element. The entire elemental mass spectrum with all its information is recorded every time.

Concerning the monitoring of the analytes of interest, Fig. 2 shows two sections of the mass spectrometer obtained for a NiS blank and for one of the samples investigated. In Fig. 2 the actual measurement data of every individual detector channel are indicated with squares. Various conclusions can be extracted from that figure. First of all, the resolution of the instrument is not constant over the entire mass spectrum. A resolution of 700–800 (R = m/Δm, with Δm the full width at half maximum—FWHM) is obtained for spectral peaks located in the low and medium mass range. However, for the high mass range (around 200 u), a value of only 270 was found. Only for the heaviest masses (e.g., Pb), no baseline-separated spectral peaks were obtained, indicating a mass resolution somewhat lower than that in quadrupole-based instrumentation in this range. In any case, this value is still acceptable for quantitative analysis. To avoid the possible influence of the tailing of an adjacent peak, only the central 35% of the peak was used for quantitative purposes. This is a setting that can be freely chosen by the analyst. Reprocessing of the signals with any other setting is always feasible since all the data are recorded.

Fig. 2 Mass spectrum obtained upon LA-ICP-MS monitoring of a ground and pelletized button obtained after NiS fire assay of a platiniferous ore (SARM 7B—blue line) and of a blank button (same NiS and binder matrix, but unspiked with any analyte—red line). Only the spectral areas in which the nuclides of interest are located are shown. The squares represent the actual signal intensity values of every detector channel in high gain mode.

A critical aspect when aiming at the determination of trace amounts of PGMs using ICP-MS is the possible occurrence of spectral overlap. A list of possible interferences is discussed in the literature.^14,36–39 When using LA for sample introduction, a dry plasma is obtained, such that the signal intensities of all O- and OH-based molecular ions are considerably reduced in comparison to those obtained with a wet plasma. Moreover, in this particular case, the method of sample preparation (pre-concentration and separation of PGMs and Au from the matrix by means of NiS fire assay) will have an important effect as the risk of spectral overlap for most nuclides should be significantly reduced, but, at the same time, it can be enhanced for those target nuclides affected by interferences from Ni- and Cu-based polyatomic ions (mainly Rh owing to ⁴⁰Ar⁶³Cu, Ru owing to ArNi species, and some Pd nuclides, e.g., ¹⁰⁵Pd owing to ⁴⁰Ar⁶⁵Cu).

In this particular case, it is possible to use an almost perfect blank containing the same amount of Ni and Cu as found in the samples and standards (after NiS fire assay). Thus, the risk of spectral overlap can be easily assessed. As seen in Fig. 2a and b, the influence of this possible overlap seems to be minimal, as the signals observed for a typical sample can be clearly differentiated from the blank signal for the most important isotopes. The expected ArNi⁺ or ArCu⁺ signals potentially overlapping with those of Rh, with the main isotopes of Ru (¹⁰¹Ru and ¹⁰²Ru) or with Pd main nuclides, are very low and can be easily blank corrected. Still, some ArNi is formed, as can be appreciated at masses 98 (⁴⁰Ar⁵⁸Ni) and 100 (⁴⁰Ar⁶⁰Ni). This could be expected taking into account the high amount of Ni and the abundance of ⁵⁸Ni (68.8%) and ⁶⁰Ni (26.22%). The use of nuclides ⁹⁸Ru and ¹⁰⁰Ru is therefore not recommended for Ru quantification, which in any case is not a relevant issue since ⁹⁹Ru,¹⁰¹Ru and ¹⁰²Ru are more abundant.

This aspect is worth discussing because in a previous work carried out with a quadrupole-based instrument, the influence of the ArNi⁺ and ArCu⁺ signals was proven to be significant, making the use of NH₃ in the reaction cell compulsory to minimize the spectral interferences they otherwise give rise to.²¹ In order to provide some quantitative data on this aspect, a ⁴⁰Ar⁵⁸Ni/⁵⁸Ni ratio of 3 × 10⁻⁶ was calculated in the current work, while in the previous work with the Perkin Elmer Elan DRC Plus a ratio of 5 × 10⁻⁴ was obtained. It has already been reported that such significant variation is feasible among different types of quadrupole-based ICPMS instruments, greatly depending on the design of the interface (sampler, skimmer and extraction lenses).⁴⁰ A relatively high pressure between the skimmer and the extraction lenses is key to a low argide level,⁴⁰ since those conditions favour the collisional dissociation of argides in this area.⁴¹ Interestingly, the Elan DRC instrument has been reported to provide a very high argide ratio,⁴⁰ while the SPECTRO MS evaluated in this work seems to perform slightly better than the best quadrupole evaluated in the paper mentioned above (Agilent 7700×).⁴⁰ Other values for the SPECTRO MS are a ⁴⁰Ar⁴⁰Ar/⁴⁰Ar ratio of 8 × 10⁻⁴ and a ¹⁷²Yb¹⁶O/¹⁷²Yb (it is not possible to monitor UO or ThO, since the limit is 240 u) ratio of 9 × 10⁻⁴.

There is obviously Ag contamination in the pellets (see the blank at masses 107 and 109 in Fig. 2a), and also the occurrence of traces of Hg can be discerned (see masses 198, 199 and 200 in Fig. 2b). However, the presence of these elements will not affect the measurement of Au and the PGMs targeted in the work.

Overall, it seems feasible to monitor the isotopes of main interest without significant overlaps under the conditions used in this work.

3.2.2. Temporal behaviour of the LA-ICP-MS signals and influence of the internal standard. It is generally accepted that the type of application for which the use of an ICP-MS instrument with simultaneous detection potential would be most beneficial is the monitoring of transient signals, such as those generated when a discontinuous sampling technique (e.g., LA or electrothermal vaporization—ETV—or a chromatographic technique) is used. Since with scanning-type ICP-MS instruments the signals need to be monitored sequentially to produce a mass spectrum, the precision with which it is possible to define the transient signal profiles will ultimately be limited, particularly when a large number of nuclides need to be monitored.⁴²

This effect will be the more detrimental the shorter the sampling event is. That means that, in principle, for an ETV signal that lasts a couple of seconds only, the situation seems a priori more critical than for signals that may be tens of seconds long.^43,44 In this particular case, the size and the flat surface of the pellet permit ablation of lines over a long period of time, if desired, with the only obvious drawback of decreasing the sample throughput. Since the idea of this work is to develop a fast method suitable for routine control, 60 seconds of ablation time per sample was selected.

However, despite this relatively long duration of each measurement, it has to be stressed that this is not a very homogeneous type of sample, even after grinding. Besides, mixing with a binder will always result in some further variations. A typical signal profile is shown in Fig. 3 and the fluctuations in the signal during the ablation process can be observed clearly. In this case, simultaneous monitoring is beneficial, because by selecting an internal standard that varies in a similar fashion as the analytes, it would be possible to minimize the effect on inhomogeneity as well as any other variations occurring during the ablation of the sample (variations in the ablation efficiency, in the transport, in the plasma conditions, etc.). As long as there is a good degree of correlation between the signal of the internal standard and those of the analytes, all these sources of noise should cancel out.⁴⁵

Fig. 3 Signal profiles (signal intensity as a function of time, baseline corrected) observed upon LA-ICP-MS analysis of a ground and pelletized button obtained after NiS fire assay of a platiniferous ore (MR RF05/1 QC). The ratio of ¹⁹⁵Pt to ⁶¹Ni (internal standard) is also shown for comparison of signal stability (in red). The portion of data actually used for deriving quantitative data (50 seconds, divided in 5 replicates of 10 seconds each) is also indicated.

That was the situation found in this work. Despite the fact that the minimum readout of the detector is 20 ms, a higher setting (100 ms) was used since obtaining more points was not assessed as providing further advantage in terms of precision (and reduces the data processing speed). The critical aspect is how well the signals of the different nuclides correlate with time. As shown in Fig. 3, there are obvious variations in the Ni content across the surface ablated, but it seems that those areas enriched in Ni are also enriched in Cu, PGMs and Au, and vice versa. There is therefore a very good degree of correspondence between the signals of the different nuclides monitored (the correlation coefficient between the signal of the analytes and that of ⁶¹Ni typically varied between 0.85 and 0.97) and normalizing the signal of the analytes to that of Ni permits a stable signal ratio to be obtained (see the case of Pt/Ni shown in Fig. 3), which should translate into a better precision. In this regard, for processing of the signal, the central portion of the signal that corresponds with the ablation (from 50 s to 100 s) was divided into 5 different replicates of 10 seconds each. Typically, the variation between these replicates improves when using the internal standard, decreasing from 4 to 6% RSD for the most abundant analytes (Pt, Pd, Rh and Ru) to only 2–3% RSD, and from 8 to 15% RSD for the less concentrated ones (Au and Ir) to 6–11% RSD. For samples in which Ir and Au are below 100 ng g⁻¹ (content in the original sample), the low signal intensity likely becomes a significant factor in contributing to the imprecision, such that it is not feasible to improve precision so much when using an internal standard.

Finally, it should also be mentioned that the plasma behaviour during sample ablation seems to be robust, as there are no indications of disturbances affecting the ⁸⁰Ar₂⁺ signal.

3.3. Analysis of the samples

Once all of the relevant settings were optimized, analysis of the samples was undertaken using the conditions shown in Table 1 and following the procedure described in Section 2.3.

In principle, one of the key benefits of this approach, NiS fire assay followed by direct solid sampling analysis of the buttons, is that it is feasible to prepare matrix-matched standards for calibration since the composition of the buttons is relatively simple, as predominantly only PGMs and Au are extracted into the NiS buttons.

Very satisfactory calibration curves could be constructed following the procedure described in Section 2.3.3. Moreover, the benefits of using an internal standard simultaneously monitored also applied here. Fig. 4a shows a typical calibration curve for the most abundant nuclides of Pt, Pd and Rh. While the calibration is relatively acceptable, there are indications of sensitivity drifts occurring during the measurements: the second standard in terms of analyte concentrations shows a signal that is a bit too high for all nuclides, while the third one shows too low a signal. It has to be stated that owing to the dimensions of the NiS buttons only one sample could be placed inside the ablation cell at the time, which made it necessary to open the cell after every analysis, thus possibly causing variations in the measurement conditions. This problem could be alleviated by using a cell designed to hold a higher number of samples, while restricting the effective volume for the expansion of the laser-generated aerosol, as recently described elsewhere.^39,46 In any case, these variations are clearly compensated for with the use of the ⁶¹Ni signal intensity as the internal standard, and an excellent linearity (r² > 0.999) was found for all the analytes.

Fig. 4 LA-ICP-MS calibration curves constructed on the basis of the intensities obtained for ground and pelletized NiS buttons spiked with known amounts of the analytes, prepared as described in Section 2.3.3, (A) when not using any internal standard and (B) when using the signal intensity for ⁶¹Ni as internal standard.

The LODs achieved are shown in Table 2 and the results for the analysis of the samples (certified reference materials of platiniferous ores, as described in Section 2.2.1) are displayed in Table 3. Whenever possible (e.g., Pt, Ru, Pd, Ir) the signal of at least two isotopes was used in order to further confirm the absence of spectral overlap. No significant differences in the results obtained using different isotopes of the same target element were observed for any of the samples. For the monitoring of ¹⁰²Ru, mathematical correction was required in order to compensate for the isobaric interference of ¹⁰²Pd. This correction accounted for only 3 to 7% of the total signal intensity for ¹⁰²Ru⁺.

Table 2 Limits of detection obtained upon LA-ICP-MS analysis of ground and pelletized NiS buttons obtained by fire assay of platiniferous ores. The values are referred to the content of the analytes in the ore sample. The pre-concentration factor of the NiS fire assay is approximately 3. The results are compared with those previously obtained by means of a quadrupole-based ICP-MS equipped with a dynamic reaction cell.²¹ In both cases, LODs were calculated as 3 times the standard deviation of ten replicates of the blank standard divided by the sensitivity (slope of calibration curve)

Technique	^101Ru	^102Ru	^103Rh	^105Pd	^106Pd	^191Ir	^193Ir	^194Pt	^195Pt	^197Au
LA-Q-ICP-MS/ng g^−1	29	15	7	19	11	23	13	11	9	10
This work/ng g^−1	23	8	8	17	12	22	12	14	11	10

---

Table 3 Results achieved in the direct determination of Au, Ir, Pd, Pt, Rh and Ru in ground and pelletized NiS buttons obtained by fire assay of platiniferous ores using LA-ICP-MS. For every sample, a NiS button (∼64% Ni) was prepared and analyzed as described in Section 2.3. Uncertainties are provided as 95% confidence intervals (n = 5). For those elements for which several major isotopes were available (Ir, Pd, Pt, and Ru), the final value was calculated as the average of the two most abundant nuclides

Sample	Au	Ir	Pd	Pt	Rh	Ru
SARM 65
Certified value/μg g^−1	0.034 ± 0.005	0.183 ± 0.011	1.28 ± 0.05	2.64 ± 0.05	0.522 ± 0.018	0.853 ± 0.038
This work/μg g^−1	0.031 ± 0.004	0.208 ± 0.024	1.20 ± 0.06	2.63 ± 0.14	0.534 ± 0.015	0.899 ± 0.041
Ratio	0.91	1.14	0.93	1.00	1.02	1.05
SARM 7B
Certified value/μg g^−1	0.270 ± 0.015	0.090 ± 0.012	1.54 ± 0.03	3.74 ± 0.05	0.240 ± 0.013	0.460 ± 0.026
This work/μg g^−1	0.257 ± 0.037	0.087 ± 0.022	1.40 ± 0.04	3.87 ± 0.18	0.251 ± 0.010	0.435 ± 0.022
Ratio	0.95	0.97	0.91	1.03	1.04	0.95
SARM 73
Certified value/μg g^−1	0.19 ± 0.02	0.11 ± 0.04	1.56 ± 0.05	2.45 ± 0.06	0.26 ± 0.03	0.51 ± 0.06
This work/μg g^−1	0.19 ± 0.03	0.10 ± 0.02	1.65 ± 0.07	2.49 ± 0.14	0.26 ± 0.01	0.48 ± 0.02
Ratio	1.02	0.91	1.06	1.02	1.01	0.94
AMIS 07
Certified value/μg g^−1	0.13 ± 0.02	0.093 ± 0.026	1.50 ± 0.20	2.48 ± 0.28	0.25 ± 0.04	0.45 ± 0.06
This work/μg g^−1	0.12 ± 0.03	0.104 ± 0.024	1.35 ± 0.05	2.38 ± 0.12	0.25 ± 0.01	0.46 ± 0.03
Ratio	0.94	1.12	0.90	0.96	0.99	1.03
AMIS 10
Certified value/μg g^−1	0.026 (IND)	0.170 ± 0.036	1.33 ± 0.10	2.05 ± 0.29	0.41 ± 0.08	0.80 ± 0.14
This work/μg g^−1	0.028 ± 0.002	0.169 ± 0.022	1.32 ± 0.07	2.15 ± 0.10	0.44 ± 0.01	0.79 ± 0.02
Ratio	1.08	0.99	0.99	1.05	1.07	0.99
MR FW05/1 QC
Certified value/μg g^−1	0.244 ± 0.017	0.051 ± 0.004	1.34 ± 0.02	2.20 ± 0.08	0.143 ± 0.008	0.295 ± 0.011
This work/μg g^−1	0.225 ± 0.020	0.049 ± 0.011	1.34 ± 0.04	2.28 ± 0.14	0.133 ± 0.005	0.310 ± 0.011
Ratio	0.92	0.97	1.00	1.04	0.93	1.05
MR RF05/1 QC
Certified value/μg g^−1	0.420 ± 0.020	0.204 ± 0.014	3.09 ± 0.07	7.17 ± 0.15	0.522 ± 0.014	1.103 ± 0.032
This work/μg g^−1	0.400 ± 0.038	0.197 ± 0.020	3.19 ± 0.10	7.54 ± 0.28	0.522 ± 0.015	1.119 ± 0.065
Ratio	0.95	0.96	1.03	1.05	1.00	1.01
UG2 RF05 /1 QC
Certified value/μg g^−1	0.033 ± 0.007	0.241 ± 0.015	1.76 ± 0.06	2.93 ± 0.09	0.602 ± 0.017	1.060 ± 0.039
This work/μg g^−1	0.039 ± 0.005	0.238 ± 0.035	1.76 ± 0.08	2.91 ± 0.16	0.581 ± 0.011	1.046 ± 0.059
Ratio	1.19	0.99	1.00	0.99	0.97	0.99
Average ratio	1.00	1.01	0.98	1.02	1.00	1.00

---

The LODs obtained can be considered as satisfactory for the application intended, which was further proved by analysis of all the samples of interest. It has to be mentioned that in this case, the pre-concentration factor of the NiS procedure was relatively moderate (3 times). Achieving higher factors is in principle possible, if analysis at lower levels is required. These LODs are very similar to those obtained in a previous work using quadrupole-based instrumentation under optimized dynamic reaction cell conditions for resolving spectral overlap. It is necessary to be cautious when establishing this comparison because the laser ablation device was not the same in both works (for the current system the mass ablated was approx. 1 μg per second, while in the previous work it was approx. 0.4 μg per second).

As shown in Table 3, an excellent agreement with the reference values was obtained. The corresponding confidence interval overlaps in all of the cases but one (Pd in SARM 7B). The difference between the certified value and the value obtained was lower than 10% in practically 90% of the cases, which is similar to what was already obtained with the quadrupole-based ICP-MS method.²¹ However, in the current work, the difference was lower than 5% in 70% of the cases, clearly improving the results for quadrupole-based ICP-MS (50% of cases only). If we consider the most abundant elements (Pd, Pt, Ru and Rh), these values further increase up to 97% of results within 10% of the reference value, and 81% within 5%. This level of accuracy is the highest reported so far for this approach (NiS followed by LA-ICP-MS),^17–21 and to some extent seems to be a consequence of the excellent performance deriving from the truly simultaneous monitoring of the signal for the internal standard and the analytes. This aspect can be regarded as very important considering the economic value of these analytes, which makes any improvement in precision and accuracy to be looked upon with high interest.

Conclusions

The use of a new type of ICP-MS device, equipped with an array detector, Mattauch–Herzog spectrometer, proved to be beneficial for the LA-ICP-MS analysis of buttons obtained by NiS fire assay of platiniferous ores. The truly simultaneous capabilities of this instrument permitted obtaining an improved level of precision and, subsequently, of accuracy, in comparison with prior works^17–21 for analysis of eight different reference materials. Moreover, this instrument provided sufficient detection power for sub μg g⁻¹ analysis, and a low level of argide-based interferences, which made it possible to monitor those elements typically affected by ArNi⁺ and ArCu⁺ overlap in these samples (Pd, Rh, Ru) in a very straightforward way.

Acknowledgements

This work has been funded by the Spanish Ministry of Science and Innovation (Project CTQ2009-08606) and the Aragon Government (Departamento de Ciencia, Tecnología y Universidad del Gobierno de Aragón y Fondo Social Europeo). The authors are very grateful to SPECTRO Analytical Instruments GmbH and, in particular, to Willi Barger, Ralf Geerling and Dirk Ardelt for permitting the access to all the instrumentation, for all the support while measuring, as well as for the fruitful discussions on the characteristics of this new type of ICP-MS device and on the results.

References

1. N. G. Randolph, Pure Appl. Chem., 1993, 65, 2411–2416.
2. R. R. Barefoot, Anal. Chim. Acta, 2004, 509, 119–125.
3. L. Bencs, K. Ravindra and R. van Grieken, Spectrochim. Acta, Part B, 2003, 58, 1723–1755.
4. R. R. Barefoot, J. Anal. At. Spectrom., 1998, 13, 1077–1084.
5. C. R. M. Rao and G. S. Reddi, TrAC, Trends Anal. Chem., 2000, 19, 565–586.
6. R. R. Barefoot and J. C. Van Loon, Talanta, 1999, 49, 1–14.
7. M. Resano, E. García-Ruiz, K. S. McIntosh, J. Hinrichs, I. Deconinck and F. Vanhaecke, J. Anal. At. Spectrom., 2006, 21, 899–909.
8. M. De la Guardia, TrAC, Trends Anal. Chem., 2010, 29, 577.
9. D. Günther and B. Hattendorf, TrAC, Trends Anal. Chem., 2005, 24, 255–265.
10. C. C. Garcia, H. Lindner and K. Niemax, J. Anal. At. Spectrom., 2009, 24, 14–26.
11. M. Resano, E. García-Ruiz and F. Vanhaecke, Mass Spectrom. Rev., 2010, 29, 55–78.
12. J. Koch and D. Günther, Appl. Spectrosc., 2011, 65, 155A–162A.
13. S. F. Boulyga and K. G. Heumann, Anal. Bioanal. Chem., 2005, 383, 442–447.
14. M. Resano, E. García-Ruiz, M. A. Belarra, F. Vanhaecke and K. S. McIntosh, TrAC, Trends Anal. Chem., 2007, 26, 385–395.
15. F. Vanhaecke, M. Resano, E. Garcia-Ruiz, L. Balcaen, K. R. Koch and K. McIntosh, J. Anal. At. Spectrom., 2004, 19, 632–638.
16. P. R. D. Mason and W. J. Kraan, J. Anal. At. Spectrom., 2002, 17, 858–867.
17. K. E. Jarvis, J. G. Williams, S. J. Parry and E. Bertalan, Chem. Geol., 1995, 124, 37–46.
18. E. K. Shibuya, J. E. S. Sarkis, J. Enzweiler, A. P. S. Jorge and A. M. G. Figueiredo, J. Anal. At. Spectrom., 1998, 13, 941–944.
19. A. P. S. Jorge, J. Enzweiler, E. K. Shibuya, J. E. S. Sarkis and A. M. G. Figueiredo, Geostand. Newsl., 1998, 22, 47–55.
20. A. M. G. Figueiredo, J. Enzweiler, J. E. S. Sarkis, A. P. S. Jorge and E. K. Shibuya, J. Radioanal. Nucl. Chem., 2000, 244, 623–625.
21. M. Resano, E. García-Ruiz, K. S. McIntosh and F. Vanhaecke, J. Anal. At. Spectrom., 2008, 23, 1599–1609.
22. N. Jakubowski, T. Prohaska, L. Rottmann and F. Vanhaecke, J. Anal. At. Spectrom., 2011, 26, 693–726.
23. E. F. Cromwell and P. Arrowsmith, J. Am. Soc. Mass Spectrom., 1996, 7, 458–466.
24. T. W. Burgoyne, G. M. Hieftje and R. A. Hites, J. Am. Soc. Mass Spectrom., 1997, 8, 307–318.
25. C. Engelhard, Anal. Bioanal. Chem., 2011, 399, 213–219.
26. J. H. Barnes, IV, G. D. Schilling, R. Sperline, M. Bonner Denton, E. T. Young, C. J. Barinaga, D. W. Koppenaal and G. M. Hieftje, Anal. Chem., 2004, 76, 2531–2536.
27. G. D. Schilling, F. J. Andrade, J. H. Barnes, IV, R. P. Sperline, M. Bonner Denton, C. J. Barinaga, D. W. Koppenaal and G. M. Hieftje, Anal. Chem., 2006, 78, 4319–4325.
28. G. D. Schilling, S. J. Ray, R. P. Sperline, M. Bonner Denton, C. J. Barinaga, D. W. Koppenaal and G. M. Hieftje, J. Anal. At. Spectrom., 2010, 25, 322–327.
29. G. D. Schilling, S. J. Ray, A. A. Rubinshtein, J. A. Felton, R. P. Sperline, M. Bonner Denton, C. J. Barinaga, D. W. Koppenaal and G. M. Hieftje, Anal. Chem., 2009, 81, 5467–5473.
30. J. A. Felton, G. D. Schilling, S. J. Ray, R. P. Sperline, M. Bonner Denton, C. J. Barinaga, D. W. Koppenaal and G. M. Hieftje, J. Anal. At. Spectrom., 2011, 26, 300–304.
31. G. D. Schilling, F. J. Andrade, J. H. Barnes, IV, R. P. Sperline, M. Bonner Denton, C. J. Barinaga, D. W. Koppenaal and G. M. Hieftje, Anal. Chem., 2007, 79, 7662–7668.
32. B. U. Peschel, F. Andrade, W. C. Wetzel, G. D. Schilling, G. M. Hieftje, J. A. C. Broekaert, R. Sperline, M. Bonner Denton, C. J. Barinaga and D. W. Koppenaal, Spectrochim. Acta, Part B, 2006, 61, 42–49.
33. mailto:http://www.spectro.com/pages/e/p010402tab_overview.htm, last accessed 24 August, 2011.
34. H. Urban, F. Zereini, B. Skerstupp and M. Tarkian, Fresenius' J. Anal. Chem., 1995, 352, 537–543.
35. T. Paukert and I. Rubeska, Anal. Chim. Acta, 1993, 278, 125–136.
36. M. Motelica-Heino, S. Rauch, G. M. Morrison and O. F. X. Donard, Anal. Chim. Acta, 2001, 436, 233–244.
37. M. Resano, E. García-Ruiz, C. Crespo, F. Vanhaecke and M. A. Belarra, J. Anal. At. Spectrom., 2003, 18, 1477–1484.
38. F. Vanhaecke, M. Resano, M. Pruneda-López and L. Moens, Anal. Chem., 2002, 74, 6040–6048.
39. F. Vanhaecke, M. Resano, J. Koch, K. McIntosh and D. Gunther, J. Anal. At. Spectrom., 2010, 25, 1259–1267.
40. M. Guillong, L. Danyushevsky, M. Walle and M. Raveggi, J. Anal. At. Spectrom., 2011, 26, 1401–1407.
41. R. S. Houk and N. Praphairaksit, Spectrochim. Acta, Part B, 2001, 56, 1069–1096.
42. M. Tanner and D. Günther, Anal. Chim. Acta, 2009, 633, 19–28.
43. M. Resano, M. Verstraete, F. Vanhaecke and L. Moens, J. Anal. At. Spectrom., 2001, 16, 1018–1027.
44. M. Resano, M. Aramendía and F. Vanhaecke, J. Anal. At. Spectrom., 2009, 24, 484–493.
45. M. Resano, M. P. Marzo, R. Alloza, C. Saénz, F. Vanhaecke, L. Yang, S. Willie and R. E. Sturgeon, Anal. Chim. Acta, 2010, 677, 55–63.
46. M. B. Fricker, D. Kutscher, B. Aeschlimann, J. Frommer, R. Dietiker, J. Bettmer and D. Günther, Int. J. Mass Spectrom., 2011, 307, 39–45.

---