Bulk analysis of columbite ores by LA-ICP-MS: development of reference materials and investigation into matrix effects

Abstract

Determination of metal elements in columbite ores is of great importance in understanding the potential economic significance and the origin of deposits. LA-ICP-MS is recognized as a green and efficient method for the bulk analysis of trace elements in natural samples. However, a major issue is the lack of matrix-matched reference materials for columbite ores. In this study, we developed an advanced method to produce pressed ultrafine-powder pellets with appropriate concentrations by applying a wet-mill method. The optimized scheme achieves a typical grain size of d₉₀ = 1.74 μm, forming pressed powder pellets with great cohesion and homogeneity, suitable for LA-ICP-MS. The relative standard deviation (RSD) values obtained from repeated measurements are <10% for more than 50 elements, comparable to those of homogeneous reference glasses. A systematic investigation of glass reference materials, commercial iron ore pellets, and our synthetic columbite ore pellets revealed significant matrix effects in various materials when using nanosecond laser ablation. Although the use of a femtosecond laser ablation system can partially suppress the matrix effect that occurs during the laser ablation process, the matrix effect in the ICP caused by differences in chemical composition remains challenging to resolve. The use of a “wet” plasma mode enhanced the matrix effect. Our results highlight the importance of matrix-matched reference materials in quantitative analysis of columbite ores and other ore samples. Utilizing the matrix-matched calibration method, we successfully determined trace elements in pressed powder pellets of iron ore (MAKR-NP) and columbite ore (LSC), with discrepancies of less than 10% for most elements.

---

1. Introduction

Determination of metal elements in ores is always an important and exciting research area, which can help to understand the potential economic significance of ores and provide essential information for understanding the origin of deposits.^1–3 Columbite-group minerals are important accessory minerals in granitic pegmatites, and constitute the principal Nb–Ta ore minerals, including columbite–Mn, columbite–Fe and several others. Columbite-group minerals are primary sources of niobium (Nb) and tantalum (Ta), which are critical elements with extensive applications in the aerospace, electronics, and steel industries due to their ability to enhance strength and corrosion resistance of alloys.^4,5 In addition, there are numerous metallic elements in columbite ores, due to their similar geochemical behaviors, such as Zr, Hf, and rare earth elements (REEs). Accurate quantification of these elements in columbite ore not only aids in evaluating ore reserves but also provides insights into parent rocks and mineralization processes. This information can be used as a guide for exploration and subsequent extraction processes.

The trace element determination in columbite ore is a significant challenge. Inductively coupled plasma mass spectrometry (ICP-MS) is frequently used for the analysis of trace elements in natural ores. However, complete digestion of columbite ore using a mixture of acids is often hampered due to the presence of resistant minerals. Furthermore, Nb and Ta have a strong tendency for polymerisation and hydrolysis, resulting in insoluble compounds in solution, while they can be stabilized by adding hydrofluoric acid (HF), forming soluble fluoro-complexes.^6–8 Consequently, a large amount of hydrofluoric acid is required to keep Nb and Ta in solution before nebulization in ICP-MS, especially for columbite ore. Alternatives to acid dissolution include fusion and sintering methods with alkali salts, which have been frequently applied to samples containing resistant minerals.^9–11 In addition, the samples could be diluted using high mass proportions of flux agents, increasing the difficulty of trace element analysis.¹²

Concerns have been raised about the efficiency and health and safety risks in geological sample preparation. Laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) has been developed as a green technique for the determination of major and trace elements in complex geological samples.^13–18 LA-ICP-MS is more economical and environmentally friendly. LA-ICP-MS directly samples solid materials, which avoids the need to completely digest materials into a stable solution, and reduces the risks of precipitation and hydrolysis. The sample preparation processes for LA-ICP-MS are simple and rapid, including the pressed powder pellet with or without adhesive^13,18 and high-temperature melting with or without flux.^19–21 The LA-ICP-MS method also provides adequate sensitivity, precision, and accuracy over a wide range of concentrations. As a result, LA-ICP-MS has been employed for the bulk analysis of elements in soil, sediment, rocks and ores (peridotite, basalt, shale, and granite).^{13,14,19,20,22–24}

Although significant contributions and improvements have been reported in LA-ICP-MS, element quantification remains a limitation and requires further development. Elemental fractionation is an inherent problem in LA-ICP-MS, and significantly affects the accuracy of quantitative results.^25–30 It is considered a non-stoichiometric conversion of elements during laser ablation, aerosol transport, and ICP excitation/ionization, which depends on the physical and chemical properties of the analyte (matrix effect) and the operational conditions of LA-ICP-MS.^25,31–34 The elemental fractionation is not yet fully understood due to its complex process. Therefore, external calibration using matrix-matched standards, based either on available certified reference materials or laboratory prepared “home-made” standards, is still a commonly accepted method for LA-ICP-MS calibration. Unfortunately, commercially available reference materials do not cover either the type of matrix or the range of concentrations of elements in the sample. This limitation is particularly relevant for the mineral resources and mining industry, where the type of ore samples is more complex. Many studies have shown that using existing reference materials of silicate glasses to correct natural minerals could obtain erroneous results.^33–38 A major concern in this study is that columbite ore contains a large amount of Nb and Ta, which have high melting and boiling points (melting point = 2477 °C and boiling point = 4744 °C for Nb, melting point = 3007 °C and boiling point = 5458 °C for Ta). The massive amount of Nb and Ta entering the ICP may affect the ionization efficiency, resulting in significant matrix effects, especially when using a large spot size for the bulk analysis. However, the matrix effect of columbite ore relative to silicate glass reference materials is still unknown.

Recently, the pressed powder pellet made with ultrafine powder has emerged as a technique for the rapid production of matrix-matched reference materials. This technique has been used to synthesize reference materials for element quantitative analysis and isotope ratio analysis. For example, Garbe-Schönberg and Müller (2014)¹⁸ developed an ultrafine powder pellet method by applying a wet milling protocol with a high-power planetary ball mill. Jochum et al. (2019) synthesized nano-pellets of MACS-3NP, JCp-1NP and JCt-1NP for trace element and Sr isotope analyses in carbonate samples.³⁹ Feng et al. (2022) used ultrafine powder to synthesize pyrite and chalcopyrite reference materials for in situ Fe and S isotope analyses.⁴⁰ In addition, “Nano-Pellets” as calibration standards also have been commercially available, produced and sold by the company “myStandards GmbH” for LA-ICP-MS analysis. However, no reference material has been developed for trace element analysis in columbite ore so far.

In this study, we developed a novel approach involving ultrafine milling in the presence of ethanol to prepare pressed pellets of columbite ores with appropriate concentrations. We then investigated in detail the elemental fractionation and matrix effect between the pellets of columbite ore and glass reference materials (NIST 610 and GSE-2G). Additionally, two “Nano-Pellets” standards of magnetite (MAKR-NP) and hematite (HMIE-NP) produced by “myStandards GmbH” were studied together. Significant matrix effects were observed among columbite ores, silicate glasses, magnetite and hematite, when a nanosecond laser ablation system (ns-LA) was used. In order to suppress the matrix effect, a femtosecond laser ablation system (fs-LA) was employed instead of ns-LA, and a “wet” plasma mode was investigated. The final results indicated that the matrix effects cannot be entirely eliminated, emphasizing the necessity of matrix-matched reference materials for the quantitative analysis of trace elements in columbite ores or iron ores using LA-ICP-MS.

2. Experimental

2.1. Reagents and standards

Ultrapure water, obtained from a Milli-Q water purification system (Millipore, Bedford, MA, USA), with a resistivity of 18.2 MΩ cm, was used to ensure the requisite purity for experimental protocols. Nitric acid (68%, GR grade) and hydrofluoric acid (40%, GR grade) were prepared through a sub-boiling distillation process to achieve optimal purity. Certified multielement and single-element standard solutions, including NCS148479, NCS148724, NCS141749, NSC148674, NCS141562, GSB-G-62064-90, and GSB-G-62073-90 (1000 mg L⁻¹), were procured from the National Centre for Analysis and Testing of Steel Material, China. These standard solutions include most of the elements required for the analysis. Analytical grade ethanol (99.7%), sourced from Sinopharm Group Chemical Reagent Co., Ltd, served as the dispersant in wet milling procedures.

For LA-ICP-MS analysis, certified reference materials of magnetite (MAKR-NP) and hematite (HMIE-NP) were procured from the myStandards GmbH company. Glass reference materials NIST 610 (National Institute of Standards and Technology) and GSE-2G (United States Geological Survey) were used to study the elemental fractionation and matrix effect.

2.2. Sample preparation

The traditional preparation method of preparing ultrafine powder involves grinding the natural sample to a fine particle size.¹⁸ However, the concentration of some elements in natural samples is low, and not suitable as external standards. The reference materials HMIE-NP and MAKR-N also have the same issue.

In this study, we doped trace elements into the ultrafine powders by adding a multielement-mixed standard solution. This solution was produced using the gravimetric method from seven certified standard solutions. The standard solution of NCS148479 contains a high concentration of HF (2%, wt) in order to stabilize elements that are susceptible to hydrolysis, such as Zr, Nb, Hf, and Ta. In order to dilute HF as much as possible, the multielement-mixed standard solution was heated at 120 °C until nearly dry (approximately 150 minutes) and redissolved in 2% HNO₃ to reduce the concentration of HF. This step can prevent the potential acid corrosion between agate balls and the multielement-mixed standard solution.

The preparation experiment was performed at the Geological Processes and Mineral Resources Laboratory in Wuhan, China (GPMR). The initial sample was a natural niobium ore (260.85 g). It was reduced to a granular size of less than 5 mm with a jaw crusher. Then the fragments underwent a two-stage milling process. Firstly, there was a brief milling for 20 seconds in a vibratory disc mill (Retsch RS 200) equipped with a tungsten carbide disc. Secondly, a more refined grinding process was applied using an agate disc mill for 10 minutes. The preliminary powder of columbite ore, with particle sizes in the micron range, was stored in a plastic bag within a dry cabinet to maintain purity and prevent contamination.

Next, the preliminary powder of columbite ore underwent wet milling using a high-energy planetary ball mill (Chishun Tech, Nanjing, China) equipped with a pair of 45 ml agate jars. Agate balls weighing 20 g with a diameter of 3 mm were placed in each jar. Afterward, 2 g of preliminary powder of columbite ore were added to the agate balls in each jar. This was followed by the addition of 9.25 g of absolute ethanol, sufficient to submerge the mixture for optimal milling. The combined volume of the agate balls, sample ore, and ethanol occupied approximately 50% of the agate jar's total height. Then a small amount of mixed standard solution was added. The milling procedure was conducted over a duration of 20 hours at a velocity of 650 rpm, consisting of 3-minute milling periods followed by 5-minute intervals for cooling. After wet milling completion, the jar lids were carefully opened, and the suspension was collected into a glass beaker and dried on a hotplate at 60 °C for 3–4 hours. To enhance the milling process, a series of parameters were investigated, including the sizes and quantities of agate milling balls, sample volumes, dispersion ethanol content, and milling speed. The produced powders were measured for grain size using a laser particle size distribution analyzer (BT-9300ST, Bettersize Instruments).

An approximate weight of 0.15 g of the ultrafine powder was carefully placed into a round mold (6 mm in diameter and 2 mm in depth). A hand-operated hydraulic press (TP40, Herzog, Germany) was then employed to apply a finely calibrated pressure of 10 kN for 5 minutes. No binder was added to the pellet. The produced pellets were further measured by LA-ICP-MS to verify the homogeneity of trace elements.

2.3. LA-ICP-MS analysis

An Agilent 7900 quadrupole ICP-MS (Agilent Technology, Tokyo, Japan) in combination with a GeoLas HD laser ablation system (Coherent Inc, Gottingen, Germany) was employed at GPMR. The GeoLas HD system employs a standard ablation cell with a closed design, comprising a cylindrical volume of around 40 cm³, featuring an inlet nozzle with an internal diameter of less than 0.5 mm and a broad outlet of 1.5 mm internal diameter. A signal smoothing device was used to produce a smooth signal. In the experimental setup, helium gas served as the carrier gas. Prior to the introduction of the aerosol into the plasma, helium was combined with argon via a T-connector, which acted as the makeup gas. The best combination of carrier gas and makeup gas was optimized by ablation of NIST 610 silicate glass to obtain a maximum signal intensity for ²³⁸U⁺ while keeping the ²³⁸U⁺/²³²Th⁺ ratio close to 1 and ²³²Th¹⁶O⁺/²³²Th⁺ < 0.3%. The operating conditions of the instruments and the measurement parameters for both the ICP-MS and the laser ablation system are concisely presented in Table 1.

Table 1 Summary of the operating parameters for LA-ICP-MS measurements

ICP-MS (Agilent 7900)
RF power	1350 W
Plasma gas (Ar)	15.0 L min^−1
Auxiliary gas (Ar)	1.0 L min^−1
Dwell time	10 ms
Acquisition mode	Time-resolved analysis
Detector mode	Dual
Isotopes	^7Li, ^9Be, ^45Sc, ^51V, ^52Cr, ^59Co, ^60Ni, ^63Cu, ^66Zn, ^75As, ^85Rb, ^88Sr, ^89Y, ^91Zr, ^93Nb, ^97Mo, ^118Sn, ^121Sb, ^133Cs, ^137Ba, ^139La, ^140Ce, ^141Pr, ^146Nd, ^147Sm, ^153Eu, ^158Gd, ^159Tb, ^163Dy, ^165Ho, ^166Er, ^169Tm, ^172Yb, ^175Lu, ^178Hf, ^181Ta, ^205Tl, ^208Pb, ^232Th, ^209Bi, ^238U
Ns-laser ablation system (Geolas HD)
Wavelength	193 nm
Energy density	5 J cm^−2
Spot sizes	44 μm
Repetition rate	5 Hz
Pulse count	250
Sampling strategy	Single spot
Carrier gas (He)	0.60 L min^−1
Makeup gas (ar)	0.88 L min^−1
Fs-laser ablation system (NWR FemtoUC)
Wavelength	257 nm
Energy density	∼2.5 J cm^−2
Spot sizes	44 μm
Repetition rate	5 Hz
Pulse count	250
Sampling strategy	Single spot
Carrier gas (He)	0.60 L min^−1
Makeup gas (Ar)	0.85 L min^−1

---

Each analysis incorporated approximately 20 s background acquisition followed by 50 s of data acquisition from the sample and 20 s of purging ready for the next analysis. Signals from the first 5 s and last 5 s time intervals of the laser ablation were not integrated with the sample analysis. The software ICPMSDataCal, which is based on Excel, was used to perform offline selection and integration of background and analyzed signals, time-drift correction, and quantitative calibration for major and trace element analysis. In the following sections, the details of the quantitative strategy are discussed.

In this study, a significant matrix effect was observed between the ore samples and the silicate glasses. To reduce the matrix effect, a femtosecond laser system (fs) and “wet” plasma mode were employed. An NWR FemtoUC femtosecond system (New Wave Research, Fremont, CA, USA), which consisted of a 300 fs Yb:KGW femtosecond laser amplifier (PHAROS, Light Conversion Ltd, Vilnius, Lithuania) with a wavelength of 257 nm, was used for laser ablation. The system was equipped with a two-volume cell. The carrier gas, make-up gas and the signal smoothing device used were the same as those in the GeoLas HD system. In the “wet” plasma mode, ultrapure water was introduced into the ICP through a glass spray chamber and nebulizer (100 μL min⁻¹). The mixing scheme of water and laser-ablated aerosol was referred to Zhang et al. (2022).⁴¹ They established a new method of fs-LA-MC-ICP-MS and “wet” plasma for the determination of Zr isotope ratios in zircons.

2.4. Trace element determination using solution-ICP-MS

Two ultrafine powders of columbite ore with different concentrations were produced and designated as “High Sample Concentration” (HSC) and “Low Sample Concentration” (LSC). Both samples were dissolved by the conventional high-pressure bomb digestion method. Specifically, 25 mg of columbite ore powder were weighed into a 10 ml PTFE-lined stainless-steel bomb (consisting of a 10 ml PTFE inner vessel with a lid, which was fitted tightly into an outer stainless steel pressure jacket). 1.5 ml HF and 0.5 ml HNO₃ were added slowly. The stainless-steel bomb was then tightly screwed and heated to 190 °C in an electric oven (Shanghai Jing Hong Laboratory Instrument Co., Ltd, Shanghai, China) for 48 hours. After cooling down, the clear solution obtained was transferred to a polyethylene bottle and diluted to 100 g by the addition of 2% HNO₃. In this process, HF was not removed through repeated evaporation, because a large number of fluorine ions are needed for stabilizing Nb and Ta.

Due to the high concentration of F ions and Nb–Ta elements, the working curve of ICP-MS established using the conventional standard solutions (2% or 5% HNO₃) could result in the systematic deviation of quantitative results. We hence chose the standard addition method to avoid the matrix effect. The standard addition method is described in detail in the ESI.† The trace elements were determined using a high-sensitivity sector field ICP-MS Element XR (Thermo Fisher Scientific Inc. Waltham, MA, USA) at GPMR. The Element XR in this study is equipped with a quartz spray chamber. The oxide formation rates during measurement were less than 3% (CeO⁺/Ce⁺). All elements were measured in low mass resolution mode. Signal drift was corrected by repeatedly analysing a quality control solution (QC) every ten samples as a drift monitor. The operating conditions of the instruments are listed in Table S1 in the ESI.†

3. Results and discussion

3.1. Optimal milling conditions

The experimental protocols for the ultrafine powder preparation have been studied in previous studies. An obvious difference in this study is that a mixed standard solution was added to the samples. Therefore, the new sample milling conditions were optimized to ensure the homogeneous distribution of trace elements from the mixed standard solution. A systematic investigation was carried out, including the agate ball size, the amount of ore sample, the amount of ethanol dispersion, and the rotation speed of the planetary ball mill. Columbite ore samples were milled for 20 hours under each set of conditions. The produced powders were then pressed as pellets and measured by LA-ICP-MS. The laser ablation parameters were fixed at a spot size of 44 μm, a frequency of 5 Hz, and a laser fluence of 5 J cm⁻². Each pellet was measured repeatedly (n = 20), and the relative standard deviation (RSD, %) was used to evaluate the homogeneity of the trace elements.

The agate ball size is an important parameter in the preparation of ultrafine powders. Garbe-Schönberg and Müller (2014) and Peters and Pettke (2017) recommended agate balls of 7 mm.^17,18 Wu et al. (2018) utilized agate balls of 5 mm to mill samples.¹⁵ In this study, we experimented with smaller agate balls, including sizes of 5 mm, 3 mm, 1 mm and the mixtures (5 mm + 3 mm and 3 mm + 1 mm). The result shows that high values of RSD (>5%) are observed when agate balls of 5 mm were used (Fig. 1a). However, the values of RSD are lower than 5% when using 3 mm, 1 mm or a mixture of 3 mm + 1 mm, indicating that good homogeneity can be achieved using the smaller balls. The agate balls of 1 mm may be damaged in a high-speed ball mill, leading to contamination of Si, so we chose agate balls of 3 mm in this study.

Fig. 1 A systematic investigation of new sample milling conditions: (a) agate ball size; (b) amount of ore sample; (c) amount of ethanol dispersion; (d) rotation speed of the planetary ball mill.

Other parameters do not show the difference in the distribution of trace elements (Fig. 1b–d). For example, the values of RSD are from 0.3% to 6.4% when the sample amounts are added from 1 g to 4 g. The use of ethanol as a suspension medium showed that increasing its amount from 6.25 g to 9.25 g does not significantly impact the values of RSD, which ranged from 0.3% to 6.9%. Similarly, varying the rotation speed of the planetary ball mill from 350 rpm to 650 rpm results in RSD values between 0.3% and 5.9%.

In summary, most of the variables have few effects on the element distribution, except for the size of the agate ball, when the sample was ground for 20 hours. The optimized parameters in this study are an agate ball size of 3 mm, ore samples of 2 g, ethanol of 9.25 g and a rotation speed of 650 rpm. The grain size of the ultrafine powders obtained under the optimal conditions was determined using a laser particle size analysis system (BT-9300ST). The d₉₀ value is 1.74 μm, which is finer than that reported by Garbe-Schönberg and Müller (2014) (d₉₀ = 7.9 μm),¹⁸ due to the longer milling time and smaller agate ball size used in this study (Fig. 2). The small grain size provides good cohesion to form pressed powder pellets without any binders.

Fig. 2 The particle size distribution of milled columbite ore with the optimized parameters.

3.2. The homogeneity examination for the pressed ultrafine-powder pellets

The produced ultrafine powders were pressed into pellets and analysed using LA-ICP-MS. We prepared 20 pellets for each HSC and LSC (Fig. S2†). Ten pellets from each batch of HSC and LSC were randomly selected. Then each pellet was analysed 10 times using the single spot mode of ns-LA. The signal profiles of some elements during laser ablation are shown in Fig. 3. The quality of element signals in the pressed ultrafine-powder pellets resembles that of glass and is very stable, with no signal spikes observed.

Fig. 3 The ablation signal profiles for columbite ore pellets of HSC (a) and LSC (b) using LA-MC-ICP-MS in single spot mode.

The values of RSD for each pellet, calculated from 10 analyses, are shown in Fig. 4. Most elements in both HSC and LSC exhibit RSD of less than 10%, indicating the homogeneous distribution of elements in each pellet. However, the data for B, Cu, Ag, and Cd are exceptions. The values of RSD for B are from 9.2% to 23.5% in LSC, but only 2.7% to 6.9% in HSC. This contradiction is considered to be caused by contamination during the preparation of the LSC powder. For Cu, Ag and Cd, the high RSD values are observed in both HSC and LSC. Contamination is unlikely the main reason for the heterogeneous distribution of these elements. Instead, Cu, Ag, and Cd in the mixed standard solution may have reacted with the sample during milling, resulting in the local enrichment of elements. Therefore, adding solid powder might be a more suitable method for doping Cu, Ag, and Cd into the initial material.

Fig. 4 Homogeneity of elements in HSC (a) and LSC (b) based on repeated LA-ICP-MS analyses expressed as relative standard deviation (RSD%). 10 pellets of HSC and 10 pellets of LSC were analysed. Each pellet was analysed 10 times in single spot mode.

The homogeneity between pellets is evaluated through the RSD values that are calculated from the average concentration in each pellet. Fig. 5 shows the relationship between the concentration values and RSD for each element in HSC and LSC. The values of RSD for most elements are less than 10%, indicating a homogeneous distribution of elements between pellets. We should note that RSD values for B and Ag range from 3.9% to 7.4%, indicating that they are homogeneous between pellets. Cu and Cd also comply with the above element distribution pattern. However, Cu in both HSC and LSC and Cd in HSC still show inhomogeneity.

Fig. 5 The relationship between average concentrations and RSD values for selected elements in HSC and LSC. The RSD values were calculated from the average value in each pellet and represented the homogeneity between pellets.

The results demonstrate that a homogeneous distribution of most trace elements can be achieved on the scale of several ten microns in the pressed ultrafine-powder pellets. The homogeneity in the produced columbite ore pellets is comparable to that of silicate glasses. In addition, the concentration of elements can be adjusted to a suitable range to meet the demand of data calibration in LA-ICP-MS, which is a significant advantage over the traditional pressed powder pellet method.^16–18

3.3. Elemental fractionation and matrix effect

3.3.1 Elemental fractionation. To investigate the elemental fractionation in various matrices during LA-ICP-MS, the elemental fractionation index (FI) was measured and calculated for NIST 610, GSE-2G, HMIE-NP, MAKR-NP, HSC and LSC using ns-LA. The laser ablation conditions were fixed, including a spot size of 44 μm, a frequency of 5 Hz, and a laser fluence of 5 J cm⁻². FI was calculated as the ratio of the total counts determined in the second half of the measurements to the counts of the first half, and normalized to Si or Fe. Most FIs in NIST 610, GSE-2G, HSC and LSC range from 0.90 to 1.10, and show minimal fractionation in both normalized elements (Fig. 6). However, some elements in HMIE-NP show high FIs, such as Ho, Yb, and Hf (from 1.22 to 1.28), which could be caused by the low concentration (0.26 μg g⁻¹ for Ho, 0.73 μg g⁻¹ for Yb and 0.3 μg g⁻¹ for Hf). For MAKR-NP, FIs of some elements exceed 1.20, such as Y, REEs, Th and U, when Si is used as the normalized element. However, when Fe is used as the normalized element, these FIs reduce below 1.20. The results indicate that the behavior of Fe is more consistent with some elements during the analysis of MAKR-NP.

Fig. 6 Fractionation indexes (FIs) normalized to Si (a) and Fe (b) obtained from the analyses of elements in different samples using ns-LA-ICP-MS.

The same experiment as above was conducted again using fs-LA. The ablation conditions were a spot size of 44 μm, a frequency of 5 Hz, and a laser influence of 2.5 J cm⁻². Fig. 7 shows that most FIs in the six samples ranged from 0.90 to 1.10, except for some elements with low concentrations. Comparing the results of the two experiments, FIs from ns-LA are more dispersed, suggesting that the process of ns-LA could produce more significant elemental fractionation. In addition, the FIs in HSC and LSC obtained using both laser systems are close to unity, similar to those in silicate glasses. The result indicates that the synthetic samples of HSC and LSC have good cohesion due to the fine particle size, and obtain controlled ablation like silicate glasses.

Fig. 7 Fractionation indexes (FIs) normalized to Si (a) and Fe (b) obtained from the analyses of elements in different samples using fs-LA-ICP-MS.

3.3.2 Matrix effect. The values of FIs help us understand the relationships between elements during LA-ICP-MS analysis, but cannot identify the matrix effect. Therefore, a new indicator for the matrix effect is necessary. The internal standard correction strategy is one of the most commonly used quantitative methods for LA-ICP-MS. Longrich et al. (1996) have given a complete quantitative formula.⁴²where C and I represent concentration and signal intensity, respectively; Sam and Cal represent the measured sample and calibration standard, respectively; An and Is are the analyte element and internal standard element, respectively. S is the normalized sensitivity. Then the formula (1) and (2) are merged into a new formula (3).

As seen from formula (3), the fundamental principle of internal standard correction is that there is a correlation of K coefficient between the signal ratio of the analyte element and internal standard element and their concentration ratio. For accurate quantification, the K coefficient needs to be consistent between the measured sample and the calibration standard. Matrix effects can then be evaluated by comparing K coefficients between different matrices.

Here, the values of K coefficient of NIST 610, GSE-2G, HMIE-NP, MAKR-NP, HSC and LSC are calculated with Si or Fe as an internal standard element, when using ns-LA. The selection of internal standard elements is very important for quantitative analysis. Si and Fe are the best internal standards for the determination of elements in this study due to their percent level abundance in the ore samples. However, Si and Fe generally have different fractionation behaviors during the laser ablation process and ICP ionization process. Therefore, we want to evaluate the difference in calibration results between these two elements as internal standard elements.

In order to more clearly express these values, the K coefficients of GSE-2G, HMIE-NP, MAKR-NP, HSC and LSC are normalized to NIST 610 when Si is used as the internal standard. Then a relative deviation of K coefficient (RD-K, %) can be obtained based on the following formula RD-K = (K_sample/K_NIST610 − 1) × 100. When Fe is used as the internal standard, HMIE-NP, MAKR-NP, HSC and LSC are normalized to GSE-2G, due to the low concentration of Fe in NIST 610.

Most RD-K values in GSE-2G range from 5.5% to 15.1% in ns-LA with Si as the internal standard, indicating a system deviation between NIST 610 and GSE-2G (Fig. 8). This system deviation agrees with the previous study.^33,43 For example, measured results in basalt/komatiite samples deviated from the reference values by about 10–30% using NIST 610 as the calibration standard.⁴³ Kimura and Chang (2012) also found that the measured values of BCR-2G obtained from ns-LA were systematically higher than reference values when NIST 612 was used as the external standard.³³ The difference in major element composition between NIST 610 and GSE-2G could be the main reason for this system deviation. In addition, higher RD-K values are observed in HMIE-NP (18.9% to 80.8%) and MAKR-NP (17.8% to 87.4%), confirming a more significant matrix effect between glass of NIST 610 and the pellets of iron ores (Fig. 8a). When GSE-2G is used as the normalized standard and Fe is used as the internal standard, the RD-K values of HMIE-NP and MAKR-NP are slightly reduced, but still range from 11.3% to 69.9% and from 11.8% to 67.3%, respectively (Fig. 8b). The matrix effect is also observed in HSC and LSC. The RD-K values of REEs are from −17% to −40%, independent of the internal standard elements (Fig. 8).

Fig. 8 A relative deviation of K coefficient (RD-K) obtained from the analyses of elements in different samples using ns-LA-ICP-MS. (a) Si was used as the reference element and NIST 610 was used as the normalization standard. (b) Fe was used as the reference element and GSE-2G was used as the normalization standard.

The thermal effect arising from the laser ablation process can lead to elemental fractionation and varied distribution of aerosol particles. The latter could influence aerosol transport and ICP ionization. We carefully investigated the morphology of the ablation craters in NIST 610, GSE-2G, HMIE-NP, MAKR-NP, HSC and LSC using a scanning electron microscope (SEM) and white-light interferometer (WLI). Smooth craters are observed in NIST 610 and GSE-2G under a SEM (Fig. 9). However, the craters in HMIE-NP, MAKR-NP, HSC and LSC display significant melting structures at the bottom. The strong melting phenomenon could contribute to the matrix effect among glasses, iron ores and columbite ores. Please note that we can observe the rougher surface in the non-ablated area in the pellets of HMIE-N and MAKR-NP compared to that in HSC and LSC (Fig. 9). It is confirmed that the grain size of HSC and LSC is finer than that of the commercial pellets of HMIE-N and MAKR-NP.

Fig. 9 SEM images of laser ablation craters on various samples using ns-LA-ICP-MS, including GES-2G (a), NIST 610 (b), HMIE (c), MAKR (d), HSC-100 (e), LSC-50 (f).

A white-light interferometer (WLI) was employed to measure the depth and 3D morphology of the craters (Fig. 10). Smooth and cylindrical craters are observed in NIST 610 and GSE-2G. However, the 3D morphologies of craters in HMIE-NP, MAKR-NP, HSC and LSC exhibit rough walls and irregular bottoms with a lot of peaks. We speculate that these peaks, observed in WLI images (and SEM images), are caused by extruding melt through the pressure of the plasma plume. The crater depths in NIST 610 and GSE-2G are from 3.9 μm to 4.5 μm, while those in HMIE-NP, MAKR-NP, HSC and LSC are shallower, ranging from 2.5 μm to 2.9 μm under the same laser ablation conditions. Empirically, pressed powder pellets tend to have deeper craters than glass due to their looser structure, but our study shows the opposite result. Differences in chemical composition make it difficult to explain the depth phenomenon. This is because iron ores (HMIE-NP and MAKR-NP) contain more metal ions, and should theoretically be able to absorb laser energy more efficiently, leading to greater ablation depth. Further research is needed to elucidate this peculiar finding. But we suggest a possible point. The plasma plume generated above the laser radiation region can absorb the laser energy during ns-LA, which can be called “plasma-shielding”.⁴⁴ There are stronger “plasma-shielding” effects in the pressed powder pellets of both iron ores and columbite ores, resulting in excessive laser energy loss and reducing the rate of ablation. Therefore, the mass produced by ns-LA differs significantly between the silicate glasses and pellets.

Fig. 10 White-light interferometer (WLI) images of laser ablation craters on various samples using ns-LA-ICP-MS, including GES-2G (a), NIST 610 (b), HMIE (c), MAKR (d), HSC-100 (e), LSC-50 (f).

In summary, when using ns-LA to ablate silicate glasses, iron ore pellets and columbite ore pellets, the morphology and depth of craters are different, contributing to inconsistencies in the mass introduced into the ICP. These factors could collectively lead to the observed matrix effect.

3.4. Suppression matrix effect: fs laser system and “wet” plasma

3.4.1 fs laser system. At present, the use of an fs laser system is considered as an effective method to suppress the thermal effect associated with the laser ablation process.^29,30,45,46 In this study, a fs laser system with a wavelength of 257 nm was used to measure NIST 610, GSE-2G, HMIE-NP, MAKR-NP, HSC and LSC. The experimental procedures were the same as those for ns-LA described above. The results for the RD-K are shown in Fig. 11. The values of RD-K in GSE-2G are within 10% with Si as the internal standard element (Fig. 11a), indicating that the matrix effect between NIST 610 and GSE-2G has been eliminated. This means that the systematic deviation between NIST 610 and GSE-2G observed in ns-LA is attributable to different ablation behaviors. However, the patterns of RD-K in HSC and LSC obtained using the fs laser system are similar to those observed with the ns laser system (Fig. 11). In particular, the RD-K values for REEs are from −22% to −51%. The replacement of ns-LA with fs-LA does not completely suppress the matrix effect between glasses and columbite ores. The analytical results for HMIE-NP and MAKR-NP show a more complex phenomenon. The values of RD-K in 60% of elements are from −9.23% to 15.60% in HMIE-NP and MAKR-NP, which are significantly close to the expected value (±15%). However, some elements still show abnormal values, including Sn, Ba, W, Tb, Zr, La, Y, Dy, As, Hf, Zn, and Mn in HMIE-NP, and Sn, As, La, Zr, W, Mn, Nb, Cr, Y, Zn, and Hf in MAKR-NP. Consequently, fs-LA does suppress the matrix effect between glasses and iron ores to a certain extent, rather than completely resolving it.

Fig. 11 A relative deviation of K coefficient (RD-K) obtained from the analyses of elements in different samples using fs-LA-ICP-MS. (a) Si was used as the reference element and NIST 610 was used as the normalization standard. (b) Fe was used as the reference element and GSE-2G was used as the normalization standard.

We also characterized the ablation craters obtained from fs-LA using SEM and WLI. The SEM results show the same surface texture with minimal melting phenomena in the six samples (Fig. 12). The data of WLI show that the number of melt-forming peaks at the bottom of craters is significantly reduced, confirming the reduced thermal effect in fs-LA (Fig. 13). The craters produced by fs-LA are irregular due to the Gaussian distribution of laser energy, which makes it difficult to measure the depth of the craters accurately. However, it can be roughly estimated that the ablation depths for silicate glasses range from 6.1 μm to 7.5 μm, which can be compared to the depths of iron ores and columbite ores from 6.0 μm to 7.9 μm. Fs-LA demonstrates consistent ablation efficiency across various samples. The similar ablation efficiency is beneficial to keep the same mass of aerosols entering into the ICP.

Fig. 12 SEM images of laser ablation craters on various samples using fs-LA-ICP-MS, including GES-2G (a), NIST 610 (b), HMIE (c), MAKR (d), HSC-100 (e), LSC-50 (f).

Fig. 13 White-light interferometer (WLI) images of laser ablation craters on various samples using fs-LA-ICP-MS, including GES-2G (a), NIST 610 (b), HMIE (c), MAKR (d), HSC-100 (e), LSC-50 (f).

The above results of systematic experiments show that fs-LA effectively reduces the thermal effect and provides a consistent ablation across all samples. The matrix effect between NIST 610 and GSE-2G is completely resolved, and it is partially resolved between glasses and iron ores. However, there is no significant improvement between glasses and columbite ores. These residual matrix effects may therefore be attributed to the ICP process.

3.4.2 “Wet” plasma. In the ICP process, the decomposition and ionization of aerosols are influenced by the properties of the plasma, the chemical composition of the aerosol, and the distribution of aerosol sizes.^47–49 Therefore, the ICP process is considered an important source of the matrix effect. Recently, it has been reported that “wet” plasma is helpful in solving the matrix effect in the process of ICP. These studies have primarily focused on non-matrix-matched analyses for U–Pb dating,⁵⁰ and metal stable isotope measurements.^41,51,52 However, there are few studies on using “wet” plasma for multi-element quantitative analysis. In this study, a combination of fs laser ablation and “wet” plasma was used to measure trace elements in NIST 610, GSE-2G, HMIE-NP, MAKR-NP, HSC and LSC.

The experimental procedures were the same as those for fs-LA analysis described above, and the results of RD-K are shown in Fig. 14. The values of RD-K in GSE-2G are mostly within the range of ±15%. But their average value is 9.44%, which is slightly higher systematically (Fig. 14a). The patterns of RD-K in HSC and LSC remain the same as before (Fig. 14). The values of RD-K for REEs in HSC and LSC are still from −36% to −59%, indicating that the “wet” plasma does not overcome the matrix effect between glasses and columbite ores.

Fig. 14 A relative deviation of K coefficient (RD-K) obtained from the analyses of elements in different samples using fs-LA-ICP-MS combined with “wet” plasma. (a) Si was used as the reference element and NIST 610 was used as the normalization standard. (b) Fe was used as the reference element and GSE-2G was used as the normalization standard.

The values of RD-K in HMIE-NP range from −39.0% to 29.6% and from −68.5% to 63.7% when using Si and Fe as internal standards, respectively (Fig. 14). The data for MAKR-NP show the greatest RD-K deviations, ranging from 19.0% to 177.1% and from −71.9% to 161.0% with Si and Fe as internal standards, respectively (Fig. 14). Firstly, the large fluctuation of RD-K caused by the selection of the internal standard element only occurs in the pellets of iron ores, not in the other samples. Secondly, the pellets of iron ores show the largest matrix effect in the “wet” plasma mode. We think that the commercial pellets of iron ores could contain large particles, which have been evidenced by SEM imaging (Fig. 9). This could result in a non-uniform distribution of aerosols generated by laser ablation. When these non-uniform aerosols enter the ICP, their ionization efficiency becomes sensitive to either the plasma state or the aerosol size. Kimura and Chang et al. (2012) pointed out that the aerosol size was an important factor affecting ionization in the ICP.³³ The finer median aerosol size led to less elemental fractionation in the ICP ion source. In this study, the temperature of the center channel in ICP is reduced with the addition of water, resulting in changes in the ionization process of aerosols. Consequently, the pellets of iron ores suffered the strongest effects in the “wet” plasma mode due to the non-uniform distribution of aerosols.

In summary, there is a minor matrix effect between silicate glasses, but it can be overcome by fs-LA. A significant matrix effect is observed between silicate glasses and pellets of columbite ores. This matrix effect is most likely to occur during the ICP ionization process, and cannot be resolved or suppressed by fs-LA and “wet” plasma. Therefore, for the analysis of columbite ores, matrix-matched reference materials are essential. For commercial pellets of iron ores, the matrix effect is particularly complex, involving both the laser ablation process and the ICP process. Fs-LA can suppress part of the matrix effect caused by the thermal effect. However, other matrix effects in the ICP process remain or are even enhanced in the “wet” plasma mode. In addition, the matrix effect in columbite ores is stronger than in iron ores. This may be due to the presence of large amounts of Nb and Ta. The high melting and boiling points of Nb and Ta make them more susceptible to the plasma state. Steenstra et al. (2019) also reported that most refractory elements suffered from matrix effects easily during both the laser ablation and ICP processes.³⁴ The “wet” plasma cannot fully resolve the problem of the matrix effect in the ICP process. Therefore, the development of matrix-matched reference materials is still important for correcting elemental fractionation or evaluating data quality, especially for ore samples with complex matrices. The synthesized samples HSC and LSC in this study show a very stable state similar to the silicate glasses, proving the superiority of our new synthesis technique. Recently, a new non-matrix-matched calibration method in LA-ICP-MS has been proposed. Mervic et al. (2024) developed a new data correction strategy based on an ablation volume normalization. This allows for accurate and precise multi-element quantification for geological and biological materials. The authors proposed that it is crucial to use the most suitable fluence for each material, such as slightly above the ablation threshold and associated with the highest signal-to-noise ratios. It is possible that this approach can solve the matrix effect problem, but more research is needed.⁵³

3.5. Determination of trace elements using LA-ICP-MS

Based on the aforementioned research results, NIST 610 and GSE-2G are unsuitable as external standards for iron ores and columbite ores. There is also an obvious matrix effect between iron ores and columbite ores. Consequently, we opted for the matrix-matched calibration method to determine the elemental composition in ore samples. The primary advantage of fs-LA is its ability to eliminate the thermal effect during the laser ablation process. However, fs-LA is twice as expensive as ns-LA. Therefore, from the point of saving experimental costs, it is more economical to choose ns-LA for the trace element analysis of a large number of ore samples.

The trace elements in MAKR-NP and LSC were measured, with HMIE-NP and HSC used as external standards, respectively. Ns-LA was employed in the experiment. The measured data are presented in Fig. 15 and Table S2.† For MAKR-NP, the measured values of most trace elements are consistent with the reference values within a 10% range. The data obtained using Si or Fe elements as internal standards also show consistency. Similarly, the measured values of most trace elements in LSC agree with the solution values within a 10% range. The analytical data are not influenced by the selection of the internal standard.

Fig. 15 The relationship between reference values (solution values) and measured values for trace elements in MAKR-NP (a) and LSC (b) using ns-LA-ICP-MS. The dashed lines show the uncertainty range of ±10%.

4. Conclusions

LA-ICP-MS is a highly attractive technique for rapid and green quantitative analysis of ore samples. However, the complex matrices of ore samples, such as columbite ores in this study, hinder the accurate analysis of trace elements. Through systematic investigation, we have reached the following conclusions:

(1) There are significant matrix effects between the conventional silicate glass standards and the pressed pellets of ore samples. These matrix effects are primarily observed in both the laser ablation and ICP ionization processes.

(2) Fs-LA can suppress a part of matrix effects by reducing thermal effects and ablating various materials equivalently. However, the matrix effect in the ICP process is difficult to resolve. The “wet” plasma mode reinforces the matrix effect unexpectedly.

(3) The inevitable problem of the matrix effect reveals the importance of matrix-matched reference materials, particularly for ore samples. A new preparation method of reference materials for columbite ores is suggested in this study. The new method needs only 24 hours to complete the preparation of ultrafine powder. Elements of interest can be added to the ultrafine powder according to the requirements of data calibration.

(4) The HSC and LSC synthesized using the new method show smooth surfaces and stable signal profiles similar to silicate glasses. Moreover, they have a consistent matrix effect under different laser conditions, such as ns-LA, fs-LA and fs-LA combined with “wet” plasma. This property also resembles that of silicate glasses, but clearly differs from that of commercial pellets of iron ores. This is attributed to the finer particles in HSC and LSC (d₉₀ = 1.74 μm), resulting in a more homogeneous aerosol size during the laser ablation process.

The results of this study demonstrate the importance of matrix-matched reference materials in the quantitative analysis of columbite ores or other ore samples. The new preparation method for reference materials provides significant support for resolving the issue of lack of matrix-matched reference materials.

Data availability

The data supporting this article have been included as part of the ESI.†

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

This research was supported by the National Key R&D Program of China (2021YFC2903003) and the most special fund from the State Key Laboratory of Geological Processes and Mineral Resources, China University of Geosciences (MSFGPMR04 and MSFGPMR08).

References

1. R. L. Linnen, M. Van Lichtervelde and P. Černý, Elements, 2012, 8(4), 275–280.
2. J. F. Slack, Econ. Geol., 2012, 107(6), 1089–1113.
3. E. A. Holley, N. Zaronikola, J. Trouba, K. Pfaff, J. Thompson, E. Spiller, C. Anderson and R. Eggert, Geology, 2023, 51(8), 773–778.
4. X. D. Deng, J. W. Li, X. F. Zhao, Z. C. Hu, H. Hu, D. Selby and Z. S. de Souza, Chem. Geol., 2013, 344, 1–11.
5. F. Melcher, T. Graupner, H.-E. Gäbler, M. Sitnikova, F. Henjes-Kunst, T. Oberthür, A. Gerdes and S. Dewaele, Ore Geol. Rev., 2015, 64, 667–719.
6. C. Münker, Chem. Geol., 1998, 144(1), 23–45.
7. S. Weyer, C. Münker, M. Rehkämper and K. Mezger, Chem. Geol., 2002, 187(3), 295–313.
8. R. Tanaka, A. Makishima, H. Kitagawa and E. Nakamura, J. Anal. At. Spectrom., 2003, 18(12), 1458–1463.
9. W. Zhang and Z. C. Hu, Spectrochim. Acta, Part B, 2019, 160, 105690.
10. S. N. H. Bokhari and T. C. Meisel, Geostand. Geoanal. Res., 2017, 41(2), 181–195.
11. G. Bayon, J. A. Barrat, J. Etoubleau, M. Benoit, C. Bollinger and S. Révillon, Geostand. Geoanal. Res., 2009, 33(1), 51–62.
12. S. Y. Zhang, H. L. Zhang, Z. H. Hou, D. A. Ionov and F. Huang, Geostand. Geoanal. Res., 2019, 43(1), 93–111.
13. L. Arroyo, T. Trejos, P. R. Gardinali and J. R. Almirall, Spectrochim. Acta, Part B, 2009, 64(1), 16–25.
14. G. O. Duodu, A. Goonetilleke, A. Charlotte and G. A. Ayoko, Anal. Chim. Acta, 2015, 898, 19–27.
15. S. T. Wu, V. Karius, B. C. Schmidt, K. Simon and G. Wörner, Geostand. Geoanal. Res., 2018, 42(4), 575–591.
16. P. K. Mukherjee, P. P. Khanna and N. K. Saini, Geostand. Geoanal. Res., 2014, 38(3), 363–379.
17. D. Peters and T. Pettke, Geostand. Geoanal. Res., 2017, 41(1), 5–28.
18. D. Garbe-schönberg and S. Müller, J. Anal. At. Spectrom., 2014, 29(6), 990–1000.
19. C. X. Zhang, Z. C. Hu, W. Zhang, Y. S. Liu, K. Q. Zong, M. Li, H. H. Chen and S. H. Hu, Anal. Chem., 2016, 88(20), 10088–10094.
20. D. A. Monsels, M. J. Van Bergen and P. R. D. Mason, Geostand. Geoanal. Res., 2018, 42(2), 239–251.
21. X. Y. Qiu, Z. C. Hu, T. He, T. Luo, W. Zhang, M. Li, K. Q. Zong, Z. C. Wang and Y. S. Liu, J. Anal. At. Spectrom., 2024, 39(2), 545–557.
22. M. E. Shaheen and B. J. Fryer, Spectrochim. Acta, Part B, 2011, 66(8), 627–636.
23. R. Hennekam, T. Jilbert, P. R. D. Mason, G. J. de Lange and G.-J. Reichart, Chem. Geol., 2015, 403, 42–51.
24. C. G. C. Patten, A. Beranoaguirre, S. Hector, D. Gudelius, J. Kolb and E. Eiche, Int. J. Mass Spectrom., 2023, 488, 117039–117048.
25. S. D. Zhang, M. H. He, Z. B. Yin, E. Y. Zhu, W. Hang and B. L. Huang, J. Anal. At. Spectrom., 2016, 31, 358–382.
26. H. P. Longerich, D. Günther and S. E. Jackson, Fresenius. J. Anal. Chem., 1996, 355(5), 538–542.
27. M. Guillong and D. Günther, J. Anal. At. Spectrom., 2002, 17(8), 831–837.
28. N. Miliszkiewicz, S. Walas and A. Tobiasz, J. Anal. At. Spectrom., 2015, 30, 327–338.
29. M. E. Shaheen, J. E. Gagnon and B. J. Fryer, Spectrochim. Acta, Part B, 2015, 107, 97–109.
30. T. Luo, Y. Wang, M. Li, W. Zhang, H. H. Chen and Z. C. Hu, At. Spectrosc., 2020, 40(1), 11–19.
31. D. Bleiner and A. Bogaerts, Spectrochim. Acta, Part B, 2006, 61(4), 421–432.
32. R. Hergenröder, J. Anal. At. Spectrom., 2006, 21(5), 505–516.
33. J. I. Kimura and Q. Chang, J. Anal. At. Spectrom., 2012, 27(9), 1549–1559.
34. E. S. Steenstra, J. Berndt, S. Klemme and W. van Westrenen, J. Anal. At. Spectrom., 2019, 34(1), 222–231.
35. K. P. Jochum, D. Scholz, B. Stoll, U. Weis, S. A. Wilson, Q. Yang, A. Schwalb, N. Börner, D. E. Jacob and M. O. Andreae, Chem. Geol., 2012, 318–319, 31–44.
36. D. S. Macholdt, K. P. Jochum, S. A. Wilson, L. M. Otter, B. Stoll, U. Weis and M. O. Andreae, Geostand. Geoanal. Res., 2016, 40(4), 493–504.
37. E. S. Steenstra, J. Berndt, S. Klemme, W. van Westrenen, E. S. Bullock and A. Shahar, J. Anal. At. Spectrom., 2020, 35(3), 498–509.
38. L. Danyushevsky, P. Robinson, S. E. Gilbert, M. Norman, R. Large, P. Mcgoldrick and M. Shelley, Geochem.:Explor., Environ., Anal., 2011, 11(1), 51–60.
39. K. P. Jochum, D. Garbe-Schönberg, M. Veter, B. Stoll, U. Weis, M. Weber, F. Lugli, A. Jentzen, R. Schiebel, J. A. Wassenburg, D. E. Jacob and G. H. Haug, Geostand. Geoanal. Res., 2019, 43(4), 595–609.
40. Y. T. Feng, W. Zhang, Z. C. Hu, T. Luo, M. Li, Y. S. Liu, H. Liu and Q. L. Li, J. Anal. At. Spectrom., 2022, 37, 551–562.
41. W. Zhang, Z. C. Hu, L. P. Feng, Z. C. Wang, Y. S. Liu, Y. T. Feng and H. Liu, J. Earth Sci., 2022, 33(1), 67–75.
42. H. P. Longerich, S. E. Jackson and D. Günther, J. Anal. At. Spectrom., 1996, 11(9), 899–904.
43. Y. S. Liu, Z. C. Hu, S. Gao, D. Gunther, J. Xu, C. G. Gao and H. H. Chen, Chem. Geol., 2008, 257(1–2), 34–43.
44. A. Bogaerts, Z. Chen, R. Gijbels and A. Vertes, Spectrochim. Acta, Part B, 2003, 58(11), 1867–1893.
45. K. P. Jochum, B. Stoll, U. Weis, D. E. Jacob, R. Mertz-Kraus and M. O. Andreae, Geostand. Geoanal. Res., 2014, 38(3), 265–292.
46. C. C. Wohlgemuth-Ueberwasser and K. P. Jochum, J. Anal. At. Spectrom., 2015, 30(12), 2469–2480.
47. M. Guillong, H. R. Kuhn and D. Günther, Spectrochim. Acta, Part B, 2003, 58(2), 211–220.
48. R. Glaus, R. Kaegi, F. Krumeich and D. Günther, Spectrochim. Acta, Part B, 2010, 65(9), 812–822.
49. T. Luo, Y. Wang, Z. C. Hu, D. Günther, Y. S. Liu, S. Gao, M. Li and S. H. Hu, J. Anal. At. Spectrom., 2015, 30(4), 941–949.
50. T. Luo, Z. C. Hu, W. Zhang, Y. S. Liu, K. Q. Zong, L. Zhou, J. F. Zhang and S. H. Hu, Anal. Chem., 2018, 90(15), 9016–9024.
51. X. Y. Zheng, B. L. Beard and C. M. Johnson, J. Anal. At. Spectrom., 2018, 33(1), 68–83.
52. J. Lin, Y. S. Liu, Z. C. Hu, W. Chen, C. X. Zhang, K. D. Zhao and X. Y. Jin, J. Anal. At. Spectrom., 2019, 34, 1145–1153.
53. K. Mervič, V. S. Šelih, M. Šala and J. T. van Elteren, Talanta, 2024, 271, 125712.

---

Footnote

† Electronic supplementary information (ESI) available: Photograph of the initial ore and synthetic pressed powder pellets; introduction of the “standard addition” method; the results of HSC, LSC, HMIE-NP and MAKR-NP. See DOI: https://doi.org/10.1039/d4ja00311j

---