Wiley

Chúng tôi mô tả các quy trình phân tích cho việc xác định các nguyên tố vi lượng được phát triển tại CNRS Service d'Analyse des Roches et des Minéraux (SARM) và báo cáo kết quả thu được cho năm vật liệu tham chiếu địa hóa: bazan BR, điôrit DR‐N, serpentinit UB‐N, anorthosit AN‐G và granit GH. Kết quả cho các nguyên tố đất hiếm, U và Th cũng được báo cáo cho các vật liệu tham chiếu khác bao gồm dunit DTS‐1, peridotit PCC‐1 và bazan BIR‐1. Tất cả các mẫu đá đã được phân hủy bằng cách nóng chảy kiềm. Các phân tích được thực hiện bằng cách tiêm dòng ICP‐MS và bằng sắc ký lỏng áp suất thấp online (LC)‐ICP‐MS cho các mẫu có nồng độ REE, U và Th rất thấp. Phương pháp sau mang lại giới hạn phát hiện thấp hơn nhiều so với dữ liệu bằng cách giới thiệu trực tiếp và loại bỏ khả năng can thiệp đồng vị trên các nguyên tố này. Mặc dù các kết quả đồng ý với hầu hết các giá trị làm việc, khi có sẵn, nhưng kết quả cho một số nguyên tố khác nhau nhẹ so với nồng độ khuyến nghị. Trong những trường hợp này, chúng tôi đề xuất các giá trị mới cho Co, Y và Zn trong bazan BR, Zr trong điôrit DR‐N, Sr và U trong granit GH, và Ga và Y trong anorthosit AN‐G. Hơn nữa, mặc dù nồng độ Sb đo được trong AN‐G rất gần với giới hạn phát hiện của chúng tôi, giá trị của chúng tôi (0.3 ± 0.1 μg g⁻¹) thấp hơn nhiều so với giá trị làm việc được báo cáo là 1.4 ± 0.2 μg g⁻¹. Các giá trị mới này cần được xác nhận bởi một chương trình liên phòng thí nghiệm mới để phân loại thêm các vật liệu tham chiếu này.

Kết quả thu được cho nồng độ REE, Th và U sử dụng LC‐ICP‐MS áp suất thấp online cung cấp giới hạn phát hiện tốt (ng g⁻¹ đến sub‐ng g⁻¹ cho đá và ng l⁻¹ đến sub‐ng l⁻¹ cho nước tự nhiên) và kết quả chính xác. Hiệu quả tách nền cho phép đo chính xác Eu mà không cần chỉnh sửa can thiệp đồng vị BaO cho các mẫu có tỷ lệ Ba/Eu cao tới 27700. Đối với nồng độ REE trong PCC‐1 và DTS‐1, sự khác biệt với các giá trị được báo cáo trong tài liệu được giải thích là do khả năng không đồng nhất của các vật liệu tham chiếu. Các giá trị Thorium và U được đề xuất cho hai mẫu này, cũng như cho AN‐G và UB‐N.

Selenium has been determined in sixty five geological reference materials of different origins by graphite furnace atomic absorption spectrometry. Samples were decomposed with a mixture of nitric and hydrofluoric acids. Selenium was reduced to Se^IV with hydrochloric acid, and then fixed and separated from the matrix on thiol cotton. After digestion of the thiol cotton in hot nitric acid, the Se concentration was measured using palladium and magnesium nitrates as a matrix modifier. The limit of determination was 0.02 μg g⁻¹, the precision of the results (relative standard deviation of 3 to 8 replicates) varied from 2.6 to 17.7% with an average of 7.9% in the range 0.02‐42.7 μg g⁻¹ and was similar to the value obtained for synthetic samples. Our results are in good agreement with available literature values.

Elemental composition data on eight older (AGV‐1, BCR‐1, DTS‐1, G‐1, G‐2, GSP‐1, PCC‐1 and W‐1) and three newer (BIR‐1, DNC‐1 and W‐2) USGS rock standards have been collected from institutional reports and journal articles from 1972–1981. This collection was combined with data from previous compilations and “consensus values” for up to 79 elements determined by comparing overall means, medians, and individual means based on analytical techniques.

Forty two major (Na, Mg, Ti and Mn) and trace elements covering the mass range from Li to U in three USGS basalt glass reference materials BCR‐2G, BHVO‐2G and BIR‐1G were determined using laser ablation‐inductively coupled plasma‐mass spectrometry. Calibration was performed using NIST SRM 610 in conjunction with internal standardisation using Ca. Determinations were also made on NIST SRM 612 and 614 as well as NIST SRM 610 as unknown samples, and included forty five major (Al and Na) and trace elements. Relative standard deviation (RSD) of determinations was below 10% for most elements in all the glasses under investigation. Consistent exceptions were Sn and Sb in BCR‐2G, BHVO‐2G and BIR‐1G. For BCR‐2G, BHVO‐2G and BIR‐1G, clear negative correlations on a logarithmic scale exist between RSD and concentration for elements lower than 1500 μg g^‐1 with logarithmic correlation coefficients between ‐0.75 and ‐0.86. There is also a clear trend of increasing RSD with decreasing concentration from NIST SRM 610 through SRM 612 to SRM 614. These suggest that the difference in the scatter of apparent element concentrations is not due to chemical heterogeneity but reflects analytical uncertainty. It is concluded that all these glasses are, overall, homogeneous on a scale of 60 μm.

Our first results on BHVO‐2G and BIR‐1G showed that they generally agreed with BHVO‐2/BHVO‐1 and BIR‐1 within 10% relative. Exceptions were Nb, Ta and Pb in BHVO‐2G, which were 14‐45% lower than reference values for BHVO‐2 and BHVO‐1. Be, Ni, Zn, Y, Zr, Nb, Sn, Sb, Gd, Tb, Er, Pb and U in BIR‐1G were also exceptions. However, of these elements, Be, Nb, Sn, Sb, Gd, Tb, Pb and U gave results that were consistent within an uncertainty of 2s between our data and BIR‐1 reference values. Results on NIST SRM 612 agreed well with published data, except for Mg and Sn. This was also true for elements with m/z 85 (Rb) in the case of NIST SRM 614.

The good agreement between measured and reference values for Na and Mg in BCR‐2G, BHVO‐2G and BIR‐1G, and for Al and Na in NIST SRM 610, 612 and 614 up to concentrations of at least several weight percent (which were possible to analyse due to the dynamic range of 10⁸) indicates the suitability of this technique for major, minor and trace element determinations.

Thirty‐seven trace elements, including rare‐earth elements, have been determined by ICP‐MS in twenty‐eight international rock standards using routine sample preparation techniques. Samples were decomposed by either pressurized HF‐HCIO₄‐aqua regia attack, or by lithium borate fusion. Generally, the ICP‐MS data for geological rock standards presented here agree well with certified values. However, the results for light rare earth elements appear to be systematically low in comparison with the published working values.

We found that the suppression of signals for ⁸⁸Sr, ¹⁴⁰Ce and ²³⁸U in rock solution caused by rock matrix in ICP‐MS (matrix effects) was reduced at high power operation (1.7 kW) of the ICP. To make the signal suppression by the matrix negligible, minimum dilution factors (DF) of the rock solution for Sr, Ce and U were 600, 400 and 113 at 1.1, 1.4 and 1.7 kW, respectively. Based on these findings, a rapid and precise determination method for Rb, Sr, Y, Cs, Ba, REE, Pb, Th and U using FI (flow injection)‐ICP‐MS was developed. The amount of the sample solution required for FI‐ICP‐MS was 0.2 ml, so that 1.8 mg sample was sufficient for analysis with a detection limit of several ng g^‐1. Using this method, we determined the trace element concentrations in the USGS rock reference materials, DTS‐1, PCC‐1, BCR‐1 and AGV‐1, and the GSJ rock reference materials, JP‐1, JB‐1, ‐2, ‐3, JA‐1, ‐2 and ‐3. The reproducibilities (RSD %) in replicate analyses (n=5) of BCR‐1, AGV‐1, JB‐1, ‐2, ‐3, JA‐1, ‐2, and ‐3 were < 6 %, and typically 2.5%. The difference between the average concentrations of this study for BCR‐1 and those of the reference values were < 2%. Therefore, it was concluded that the method can give reliable data for trace elements in silicate rocks.

The Istituto di Geoscienze e Georisorse (IGG), on behalf and with the support of the International Atomic Energy Agency (IAEA), prepared eight geological materials (three natural waters and five rocks and minerals), intended for a blind interlaboratory comparison of measurements of boron isotopic composition and concentration. The materials were distributed to twenty seven laboratories ‐ virtually all those performing geochemical boron isotope analyses in the world ‐which agreed to participate in the intercomparison exercise. Only fifteen laboratories, however, ultimately submitted the isotopic and/or concentration results they obtained on the intercomparison materials. The results demonstrate that interlaboratory reproducibility is not well reflected by the precision values reported by the individual laboratories and this observation holds true for both boron concentration and isotopic composition. The reasons for the discrepancies include fractionations due to the chemical matrix of materials, relative shift of the zero position on the δ¹¹B scale and a lack of well characterized materials for calibrating absolute boron content measurements. The intercomparison materials are now available at the IAEA (solid materials) and IGG (waters) for future distribution.

Microanalytical trace element techniques (such as ion probe or laser ablation ICP‐MS) are hampered by a lack of well characterized, homogeneous standards. Two silicate glass reference materials produced by National Institute of Standards and Technology (NIST), NIST SRM 610 and NIST SRM 612, have been shown to be homogeneous and are spiked with up to sixty one trace elements at nominal concentrations of 500 μg g‐1 and 50 μg g‐1 respectively. These samples (supplied as 3 mm wafers) are equivalent to NIST SRM 611 and NIST SRM 613 respectively (which are supplied as 1 mm wafers) and are becoming more widely used as potential microanalytical reference materials. NIST however, only certifies up to eight elements in these glasses. Here we have compiled concentration data from approximately sixty published works for both glasses, and have produced new analyses from our laboratories. Compilations are presented for the matrix composition of these glasses and for fifty eight trace elements. The trace element data includes all available new and published data, and summaries present the overall average and standard deviation, the range, median, geometric mean and a preferred average (which excludes all data outside ± one standard deviation of the overall average). For the elements which have been certified, there is a good agreement between the compiled averages and the NIST data. This compilation is designed to provide useful new working values for these reference materials.