High‐Precision Fe and Mg Isotope Ratios of Silicate Reference Glasses Determined In Situ by Femtosecond LA‐MC‐ICP‐MS and by Solution Nebulisation MC‐ICP‐MS

In this study, a technique for high precision in situ Fe and Mg isotope determinations by femtosecond‐laser ablation‐multi collector‐ICP‐MS (fs‐LA‐MC‐ICP‐MS) was developed. This technique was employed to determine reference values for a series of common reference glasses that may be used for external standardisation of in situ Fe and Mg isotope determinations in silicates. The analysed glasses are part of the MPI‐DING and United States Geological Survey (USGS) reference glass series, consisting of basaltic (BIR‐1G, BCR‐2G, BHVO‐2G, KL2‐G, ML3B‐G) and komatiitic (GOR128‐G and GOR132‐G) compositions. Their Fe and Mg isotope compositions were determined by in situ fs‐LA‐MC‐ICP‐MS and by conventional solution nebulisation multi‐collector ICP‐MS. We determined δ⁵⁶Fe values for these glasses ranging between ‐0.04‰ and 0.10‰ (relative to IRMM‐014) and δ²⁶Mg values ranging between ‐0.40‰ and ‐0.15‰ (relative to DSM‐3). Our fs‐LA‐MC‐ICP‐MS results for both Fe and Mg isotope compositions agreed with solution nebulisation analyses within analytical uncertainties. Furthermore, the results of three USGS reference glasses (BIR‐1G, BHVO‐2G and BCR‐2G) agreed with previous results for powdered and dissolved aliquots of the same reference materials. Measurement reproducibilities of the in situ determinations of δ⁵⁶Fe and δ²⁶Mg values were usually better than 0.12‰ and 0.13‰ (2s), respectively. We further demonstrate that our technique is a suitable tool to resolve isotopic zoning in chemically‐zoned olivine crystals. It may be used for a variety of different applications on isotopically‐zoned minerals, e.g., in magmatic or metamorphic rocks or meteorites, to unravel their formation or cooling rates.     

Determination of Boron Concentration in Geochemical Reference Materials Extracted by Pyrohydrolysis and Measured by ICP‐MS

This article presents new boron concentrations for nine geochemical reference materials (GS‐N, FK‐N, GL‐O, BX‐N, DT‐N, AN‐G, GH, Mica‐Fe, Mica‐Mg). After extraction by a modified pyrohydrolysis technique, boron concentrations were measured by ICP‐MS. The blank levels for the whole procedure were 0.091 ± 0.020 ng ml^?1 or 14 ± 5 ng of boron in total. The method was first validated by measuring nine reference materials with known boron concentrations. The determined boron concentrations are all within the range of recommended or published values, which means that the yields were 100%, and show precisions below 10% for samples containing over 2 μg g^?1 of boron.     

High‐Precision Iron Isotope Analysis of Geological Reference Materials by High‐Resolution MC‐ICP‐MS

We report high‐precision iron isotopic data for twenty‐two commercially available geological reference materials, including silicates, carbonatite, shale, carbonate and clay. Accuracy was checked by analyses of synthetic solutions with known Fe isotopic compositions but different matrices ranging from felsic to ultramafic igneous rocks, high Ca and low Fe limestone, to samples enriched in transition group elements (e.g., Cu, Co and Ni). Analyses over a 2‐year period of these synthetic samples and pure Fe solutions that were processed through the whole chemistry procedure yielded an average δ⁵⁶Fe value of ?0.001 ± 0.025‰ (2s, n = 74), identical to the expected true value of 0. This demonstrates a long‐term reproducibility and accuracy of < 0.03‰ for determination of ⁵⁶Fe/⁵⁴Fe ratios. Reproducibility and accuracy were further confirmed by replicate measurements of the twenty‐two RMs, which yielded results that perfectly match the mean values of published data within quoted uncertainties. New recommended values and associated uncertainties are presented for interlaboratory calibration in the future.     

Determination of Cr,Cu, Ni,Sn, Sr and Zn Mass Fractions in Geochemical Reference Materials by Isotope Dilution ICP‐MS

Isotope dilution (ID) mass spectrometry is a primary method of analysis suited for the accurate and precise measurement of several trace elements in geological matrices. Here we present mass fractions and respective uncertainties for Cr, Cu, Ni, Sn, Sr and Zn in 10 silicate rock reference materials (BCR‐2, BRP‐1, BIR‐1, OU‐6, GSP‐2, GSR‐1, AGV‐1, RGM‐1, RGM‐2 and G‐3) obtained by the double ID technique and measuring the isotope ratios with an inductively coupled plasma‐mass spectrometer equipped with collision cell. Test portions of the samples were dissolved by validated procedures, and no further matrix separation was applied. Addition of spikes was designed to achieve isotope ratios close to unity to minimise error magnification factors, according to the ID theory. Radiogenic ingrowth of ⁸⁷Sr from the decay of ⁸⁷Rb was considered in the calculation of Sr mass fractions. The mean values of our results mostly agree with reference values, considering both uncertainties at the 95% confidence level, and also with ID data published for AGV‐1. Considering all results, the means of the combined uncertainties were < 1% for Sr, approximately 2% for Sn and Cu, 4% for Cr and Ni and almost 6% for Zn.     

Reference Values Following ISO Guidelines for Frequently Requested Rock Reference Materials

We present new reference values for nineteen USGS, GSJ and GIT‐IWG rock reference materials that belong to the most accessed samples of the GeoReM database. The determination of the reference values and their uncertainties at the 95% confidence level follows as closely as possible ISO guidelines and the Certification Protocol of the International Association of Geoanalysts. We used analytical data obtained by the state‐of‐the‐art techniques published mainly in the last 20 years and available in GeoReM. The data are grouped into four categories of different levels of metrological confidence, starting with isotope dilution mass spectrometry as a primary method. Data quality was checked by careful investigation of analytical procedures and by the application of the Horwitz function. As a result, we assign a new and more reliable set of reference values and respective uncertainties for major, minor and a large group of trace elements of the nineteen investigated rock reference materials.     

Precise Determination of Silicon Isotopes in Silicate Rock Reference Materials by MC‐ICP‐MS

A HF‐free sample preparation method was used to purify silicon in twelve geological RMs. Silicon isotope compositions were determined using a Neptune instrument multi‐collector‐ICP‐MS in high‐resolution mode, which allowed separation of the silicon isotope plateaus from their interferences. A 1 μg g^‐1 Mg spike was added to each sample and standard solution for online mass bias drift correction. δ³⁰Si and δ²⁹Si values are expressed in per mil (‰), relative to the NIST SRM 8546 (NBS‐28) international isotopic RM. The total variation of δ³⁰Si in the geological reference samples analysed in this study ranged from ‐0.13‰ to ‐0.29‰. Comparison with δ²⁹Si values shows that these isotopic fractionations were mass dependent. IRMM‐17 yielded a δ³⁰Si value of ‐1.41 ± 0.07‰ (2s, n = 12) in agreement with previous data. The long‐term reproducibility for natural samples obtained on BHVO‐2 yielded δ³⁰Si = ‐0.27 ± 0.08‰ (2s, n = 42) on a 12 month time scale. An in‐house Si reference sample was produced to check for the long‐term reproducibility of a mono‐elemental sample solution; this yielded a comparable uncertainty of ± 0.07‰ (2s, n = 24) over 5 months.     

Determination of Trace Elements in Geological Reference Materials G‐3, GSP‐2 and SGD‐1a by Low‐Dilution Glass Bead Digestion and ICP‐MS

We present a revised alkali fusion method for the determination of trace elements in geological samples. Our procedure is based on simple acid digestion of powdered low‐dilution (flux : sample ≈ 2 : 1) glass beads where large sample dilution demanded by high total dissolved solids, a main drawback of conventional alkali fusion, could be circumvented. Three geological reference materials (G‐3 granite, GSP‐2 granodiorite and SGD‐1a gabbro) decomposed by this technique and routine tabletop acid digestion were analysed for thirty trace elements using a quadrupole ICP‐MS. Results by conventional acid digestion distinctly showed poor recoveries of Zr, Hf and rare earth elements due to incomplete dissolution of resistant minerals. On the other hand, results obtained by our method were in reasonable agreement with reference data for most analytes, indicating that refractory minerals were efficiently dissolved and volatile loss was insignificant.     

Characterisation of Apatites as Potential Uranium Reference Materials for Fission‐track Dating by LA‐ICP‐MS

We report homogeneity tests on large natural apatite crystals to evaluate their potential as U reference materials for apatite fission‐track (AFT) thermochronology by laser ablation‐inductively coupled plasma‐mass spectrometry (LA‐ICP‐MS). The homogeneity tests include the measurements of major element concentrations by electron probe microanalysis (EPMA), whereas for U concentration, isotope dilution (ID) ICP‐MS and laser ablation (LA) ICP‐MS were employed. Two apatite crystals are potential reference materials for LA‐ICP‐MS analysis: a 1 cm³ fraction of a Durango crystal (7.5 μg g^?1 U) and a 1 cm³ Mud Tank crystal (6.9 μg g^?1 U). The relative standard deviation (1 RSD) of the U concentration determined by ID‐ICP‐MS of both apatite crystals was ≤ 1.5%, whereas 1 RSD for the LA‐ICP‐MS results was better than 4%, providing sufficient homogeneity for fission‐track dating. The results on the U homogeneity for two different apatite samples are an important step towards establishing in situ dating routines for AFT analysis by LA‐ICP‐MS.     

Abundances of Sulfur,Selenium, Tellurium,Rhenium and Platinum‐Group Elements in Eighteen Reference Materials by Isotope Dilution Sector‐Field ICP‐MS and Negative TIMS

Geological reference materials (RMs) with variable compositions and NIST SRM 612 were analysed by isotope dilution mass spectrometry for bulk rock concentrations of chalcogen elements (sulfur, selenium and tellurium), rhenium and platinum‐group elements (PGEs: Ru, Pd, Os, Ir and Pt), including the isotope amount ratios of ¹⁸⁷Os/¹⁸⁸Os. All concentrations were obtained from the same aliquot after HCl‐HNO₃ digestion in a high pressure asher at 320 °C. Concentrations were determined after chemical separation by negative TIMS, ICP‐MS and hydride generation ICP‐MS (Se, Te). As in previous studies, concentrations of the PGEs in most RMs were found to be highly variable, which may be ascribed to sample heterogeneity at the < 1 g level. In contrast, S, Se and Te displayed good precision (RSD < 5%) in most RMs, suggesting that part of the PGE budget is controlled by different phases, compared with the chalcogen budget. The method may minimise losses of volatile chalcogens during the closed‐system digestion and indicates the different extent of heterogeneity of chalcogens, Re and PGEs in the same sample aliquot. OKUM, SCo‐1, MRG‐1, DR‐N and MAG‐1 are useful RMs for the chalcogens. NIST SRM 612 displays homogenous distribution of S, Se, Te, Pt and Pd in 30 mg aliquots, in contrast with micro‐scale heterogeneity of Se, Pd and Pt.     

High‐Precision Measurement of Molybdenum Isotopic Compositions of Selected Geochemical Reference Materials

Here we describe high‐precision molybdenum isotopic composition measurements of geological reference materials, performed using multi‐collector inductively coupled plasma‐mass spectrometry (MC‐ICP‐MS). Purification of Mo for isotopic measurements was achieved by ion exchange chromatography using Bio‐Rad AG^® 1‐X8 anion exchange resin. Instrumental mass bias was corrected using ¹⁰⁰Mo‐⁹⁷Mo double spiking techniques. The precision under intermediate measurement conditions (eighteen measurement sessions over 20 months) in terms of δ^98/95Mo was 0.10‰ (2s). The measurement output was approximately four times more efficient than previous techniques, with no compromise in precision. The Mo isotopic compositions of seven geochemical reference materials, seawater (IAPSO), manganese nodules (NOD‐P‐1 and NOD‐A‐1), copper‐molybdenum ore (HV‐2), basalt (BCR‐2) and shale (SGR‐1b and SCo‐1), were measured. δ^98/95Mo values were obtained for IAPSO (2.25 ± 0.09‰), NOD‐P‐1 (?0.66 ± 0.05‰), NOD‐A‐1 (?0.48 ± 0.05‰), HV‐2 (?0.23 ± 0.10‰), BCR‐2 (0.21 ± 0.07‰), SCo‐1 (?0.24 ± 0.06‰) and SGR‐1b (0.63 ± 0.02‰) by calculating δ^98/95Mo relative to NIST SRM 3134 (0.25‰, 2s). The molybdenum isotopic compositions of IAPSO, NOD‐A‐1 and NOD‐P‐1 obtained in this study are within error of the compositions reported previously. Molybdenum isotopic compositions for BCR‐2, SCo‐1 and SGR‐1b are reported for the first time.     

New ICP‐MS Results for Trace Elements in Five Iron‐Formation Reference Materials

Iron formations (IFs) typically contain low mass fractions of most trace elements, including the rare earth elements (REE), and few publications describe analytical methods dedicated to this matrix. In this study, we used bomb and table‐top acid dissolution procedures and ICP‐MS to determine the mass fractions of trace elements in IF reference materials FER‐1, FER‐2, FER‐3, FER‐4 and IF‐G. The full digestion of the IF samples with the bomb procedure required the addition of a small amount of water together with the acids. The results obtained by this method mostly agreed statistically with published values. The most remarkable exception was the higher values obtained for the heavy REE in FER‐3. The recoveries of the REE obtained with the table‐top procedure were slightly higher than those of the bomb digestion, except for the values of the heavy REE in FER‐3 and FER‐4, which were up to 30% lower than published values. Sintering of the samples with sodium peroxide was performed to determine the REE, but the results tended to be lower than those derived following acid digestion. On the whole, the recoveries showed dependence on the conditions of digestion, but subtle differences in trace mineral composition between samples also exerted influence on the analytical results for trace elements.     

Nickel Isotope Variations in Terrestrial Silicate Rocks and Geological Reference Materials Measured by MC‐ICP‐MS

Although initial studies have demonstrated the applicability of Ni isotopes for cosmochemistry and as a potential biosignature, the Ni isotope composition of terrestrial igneous and sedimentary rocks, and ore deposits remains poorly known. Our contribution is fourfold: (a) to detail an analytical procedure for Ni isotope determination, (b) to determine the Ni isotope composition of various geological reference materials, (c) to assess the isotope composition of the Bulk Silicate Earth relative to the Ni isotope reference material NIST SRM 986 and (d) to report the range of mass‐dependent Ni isotope fractionations in magmatic rocks and ore deposits. After purification through a two‐stage chromatography procedure, Ni isotope ratios were measured by MC‐ICP‐MS and were corrected for instrumental mass bias using a double‐spike correction method. Measurement precision (two standard error of the mean) was between 0.02 and 0.04‰, and intermediate measurement precision for NIST SRM 986 was 0.05‰ (2s). Igneous‐ and mantle‐derived rocks displayed a restricted range of δ^60/58Ni values between ?0.13 and +0.16‰, suggesting an average BSE composition of +0.05‰. Manganese nodules (Nod A1; P1), shale (SDO‐1), coal (CLB‐1) and a metal‐contaminated soil (NIST SRM 2711) showed positive values ranging between +0.14 and +1.06‰, whereas komatiite‐hosted Ni‐rich sulfides varied from ?0.10 to ?1.03‰.     

Non‐Matrix‐Matched Calibration for the Multi‐Element Analysis of Geological and Environmental Samples Using 200 nm Femtosecond LA‐ICP‐MS: A Comparison with Nanosecond Lasers

LA‐ICP‐MS is one of the most promising techniques for in situ analysis of geological and environmental samples. However, there are some limitations with respect to measurement accuracy, in particular for volatile and siderophile/chalcophile elements, when using non‐matrix‐matched calibration. We therefore investigated matrix‐related effects with a new 200 nm femtosecond (fs) laser ablation system (NWRFemto200) using reference materials with different matrices and spot sizes from 10 to 55 μm. We also performed similar experiments with two nanosecond (ns) lasers, a 193 nm excimer (ESI NWR 193) and a 213 nm Nd:YAG (NWR UP‐213) laser. The ion intensity of the 200 nm fs laser ablation was much lower than that of the 213 nm Nd:YAG laser, because the ablation rate was a factor of about 30 lower. Our experiments did not show significant matrix dependency with the 200 nm fs laser. Therefore, a non‐matrix‐matched calibration for the multi‐element analysis of quite different matrices could be performed. This is demonstrated with analytical results from twenty‐two international synthetic silicate glass, geological glass, mineral, phosphate and carbonate reference materials. Calibration was performed with the certified NIST SRM 610 glass, exclusively. Within overall analytical uncertainties, the 200 nm fs LA‐ICP‐MS data agreed with available reference values.     

Determination of Gold in e‐Waste Dust Samples and Geological Matrices by ICP‐MS after Extraction by an HClO4‐HBr‐HI‐Aqua Regia Mixture

This study describes two methods (Procedures‐1 and ‐2) for the direct extraction of Au by an inorganic acid mixture (HClO₄‐HBr‐HI‐aqua regia) from complex sample matrices. Standard PTFE jars at 200 °C were used to decompose test portions of 0.5–1 g, with subsequent precise and accurate analysis by ICP‐MS without any other preconcentration or separation. Procedure‐1 decomposed samples effectively without the necessity of leaching with HF and was developed for dust samples from e‐waste (electronic waste) processing; however, testing on geological reference materials showed very good results. The analyses of replicate decompositions (n = 5) from both procedures yielded very good precision (< 5% RSD) for most of the reference materials. The accuracy achieved was better than ± 10%, with the exception of NIST SRM 2782 data from Procedure‐1. Two unknown samples of dust from e‐waste processing (P‐1 and VM‐1) exhibited elevated concentrations of Au (21.31–61.64 μg g^?1) with precision better than 10% (n = 5). The proposed techniques are simple, sensitive and sparing in the use of chemicals, and are designed for a variety of e‐waste dust samples. No significant influences were observed for the predicted spectral interferences on mass ¹⁹⁷Au.     

Determination of Precise and Accurate 51V/50V Isotope Ratios by Multi‐Collector ICP‐MS,Part 2: Isotopic Composition of Six Reference Materials plus the Allende Chondrite and Verification Tests

We present the first measurements of vanadium (V) stable isotopes for six reference materials – USGS PCC‐1, BHVO‐2, BCR‐2, BIR‐1a, GSP‐2 and AGV‐2 – plus the widely available carbonaceous chondrite Allende. We present standard addition and matrix spiking tests to assess the robustness and reproducibility of our data. Standard addition utilised an enriched ⁵⁰V solution designated VISSOX (Vanadium Isotope Standard Solution OXford). We further assessed the veracity of the method by spiking collected sample matrices with the same amount of a V standard solution, whose isotopic composition was defined as 0‰. Standard addition and matrix spiking tests recorded no appreciable artificial isotope fractionation. We estimate that the best currently attainable long‐term reproducibility of stable ⁵¹V/⁵⁰V isotope measurements in complex matrices is 0.15‰, which is in the same order as the reproducibility achievable with standard solutions. Finally, a large range of ～ 1.2‰ in stable V isotopic composition was documented, with ～ 0.5‰ of that variation in high temperature igneous materials alone. The range and resolving power of V stable isotopes, with respect to igneous material, compared favourably with the magnitude of fractionation reported for other non‐traditional stable isotope systems, which bodes well for the utility of this new system.     

Determination of Thallium in the USGS Glass Reference Materials BIR‐1G,BHVO‐2G and BCR‐2G and Application to Quantitative Tl Concentrations by LA‐ICP‐MS

Here, we present determinations of thallium (Tl) concentrations in the USGS reference materials BIR‐1G, BHVO‐2G and BCR‐2G measured by solution ICP‐MS. The Tl content in these three glasses spans a range of about 2–230 ng g^?1, which is similar to the values published for the respective powder materials. The determined range of Tl concentrations in these three glass reference materials makes them ideal for investigating Tl concentrations in basaltic and andesitic volcanic glasses. We also performed a series of laser ablation ICP‐MS measurements on the three samples, which show that this technique is able to determine Tl concentrations in glass samples with concentrations as low as 2 ng g^?1.     

Classical and New Procedures of Whole Rock Dissolution for Trace Element Determination by ICP‐MS

Sample digestion is a critical stage in the process of chemical analysis of geological materials by ICP‐MS. We present a new HF/HNO₃ procedure to dissolve silicate rock samples using a high pressure asher system. The formation of insoluble AlF₃ was the major obstacle in achieving full recoveries. This was overcome by setting an appropriate digestion temperature and adding Mg to the samples before digestion. Sodium peroxide sintering was also investigated and the inclusion of a heating step to the alkaline sinter solution improved the recoveries of thirteen elements other than the lanthanides. The results of these procedures were compared with data sets generated by common acid decomposition techniques. Forty‐one trace elements were determined using an ICP‐QMS equipped with a collision cell. Under optimum conditions of gas flow and kinetic energy discrimination, polyatomic interferences were eliminated or attenuated. The measurement bias obtained for eight reference materials (BCR‐2, BHVO‐2, BIR‐1, BRP‐1, OU‐6, GSP‐2, GSR‐1 and RGM‐1) and intermediate precision (RSD) were generally better than ± 5%. The expanded measurement uncertainties estimated for two certified reference materials were mostly between 7 and 15%. New data sets for the reference materials are provided, including constituents with previously unavailable values and also for the USGS candidate reference material G‐3.     

Mass Fractions of S,Cu, Se,Mo, Ag,Cd, In,Te, Ba,Sm, W,Tl and Bi in Geological Reference Materials and Selected Carbonaceous Chondrites Determined by Isotope Dilution ICP‐MS

Mass fractions of S, Cu, Se, Mo, Ag, Cd, In, Te, Ba, Sm, W and Tl were determined by isotope dilution sector field ICP‐MS in the same sample aliquot of reference materials using HF‐HNO₃ digestion in PFA beakers in pressure bombs and glassy carbon vessels in a high‐pressure asher (HPA‐S) for comparison. Additionally, Bi was determined by internal standardisation relative to Tl. Because isobaric and oxide interferences pose problems for many of these elements, efficient chromatographic separation methods in combination with an Aridus desolvator were employed to minimise interference effects. Repeated digestion and measurement of geological reference materials (BHVO‐1, BHVO‐2, SCo‐1, MAG‐1, MRG‐1 and UB‐N) gave results with < 5% relative intermediate precision (1s) for most elements, except Bi. Replicates of NIST SRM 612 glass digested on a hot plate were analysed by the same methods, and the results agree with reference values mostly within 2% relative deviation. Data for the carbonaceous chondrites Allende, Murchison, Orgueil and Ivuna are also reported. Digestion in a HPA‐S was as efficient as in pressure bombs, but some elements displayed higher blank levels following HPA‐S treatment. Pressure bomb digestion yielded precise data for volatile S, Se and Te, but may result in high blanks for W.     

Precise and Accurate Doping of Nanoporous Silica Gel for the Synthesis of Trace Element Microanalytical Reference Materials

This study presents an experimental procedure to fabricate high‐purity silica glass containing a selected element at a specified mass fraction. The procedure was used to prepare glasses doped with trace‐level mass fractions of Ti with the goal of improving analytical confidence when measuring trace elements in quartz. Systematic tests were performed to determine the ideal conditions and procedures for doping nanoporous silica gel with the highest efficiency of dopant recovery. Silica gel was cleaned in concentrated HCl, immersed in a non‐polar doping medium at a controlled pH and doped with precise quantities of ICP‐MS standard solution. Using liquids composed of longer chain molecules as the doping medium diminishes recovery, suggesting that large molecules could obstruct nanopores to inhibit capillary uptake of the dopant. A control experiment using crystalline quartz reinforced the effectiveness of nanoporous silica gel for doping with trace‐level precision. Layered aggregates of silica gel doped with different Ti mass fractions were hot‐pressed to create multi‐layered reference materials that were analysed with multiple techniques at a variety of spatial scales. Analyses at the intra‐grain scale (cathodoluminescence scanning electron microscopy, electron probe microanalysis), at the single grain scale (SIMS), at the sample layer scale (EPMA, laser ablation‐ICP‐MS) and at the bulk scale (ICP‐OES) demonstrated acceptable homogeneity at sample volumes characteristic of most microanalysis techniques and show that nanoporous silica gel holds promise as a highly retentive doping substrate for preparing reference materials for laser‐, electron‐ and ion‐beam microanalysis.     

Accurate and High‐Precision Determination of Boron Isotopic Ratios at Low Concentration by MC‐ICP‐MS (Neptune)

We report an approach for the accurate and reproducible measurement of boron isotope ratios in natural waters using an MC‐ICP‐MS (Neptune) after wet chemistry sample purification. The sample matrix can induce a drastic shift in the isotopic ratio by changing the mass bias. It is shown that, if no purification is carried out, the direct measurement of a seawater diluted one hundred times will induce an offset of ?7‰ in the isotopic ratio, and that, for the same concentration, the greater the atomic mass of the matrix element, the greater the bias induced. Whatever the sample, it is thus necessary to remove the matrix. We propose a method adapted to water samples allowing purification of 100 ng of boron with a direct recovery of boron in 2 ml of 3% v/v HNO₃, which was our working solution. Boron from the International Atomic Energy Agency IAEA‐B1 seawater reference material and from the two groundwater reference materials IAEA‐B2 and IAEA‐B3, was chemically purified, as well as boron from the certified reference material NIST SRM 951 as a test. The reproducibility of the whole procedure (wet chemistry and MC‐ICP‐MS measurement) was ± 0.4‰ (2s). Accuracy was verified by comparison with positive‐TIMS values and with recommended values. Seawater, being homogeneous for boron isotope ratios, is presently the only natural water material that is commonly analysed for testing accuracy worldwide. We propose that the three IAEA natural waters could be used as reference samples for boron isotopes, allowing a better knowledge of their isotopic ratios, thus contributing to the certification of methods and improving the quality of the boron isotopic ratio measurements for all laboratories.