BACKGROUND Molybdenum (Mo) is widely used in aerospace, nuclear industry, integrated circuits, flat display and photovoltaic solar energy and other fields^2-4. The purity or impurity content of high purity molybdenum has an important effect on material properties, so the accurate measurement of trace impurities is the key to quality control. In the China national standard “Molybdenum Powder” (GB/T 3461—2016), the limits of impurities such as Pb and Bi which are commonly in the range of 5 to 50mg/kg, are specified. Rapid and accurate analysis of trace or ultra-trace impurities is necessary with the development of high purity Mo requirements of high throughput analysis.　　The ICP-MS method has become the most powerful method for trace impurities analysis in high purity Mo, but the tedious and time-consuming matrix effect calibration strategies makes it a big challenge to achieve rapid and accurate analysis. ICP-MS has the advantages of high sensitivity, good accuracy and the possibility of simultaneous measurement of multiple elements, compared with AAS, ICP-OES, and GD-MS methods. However, the matrix effect is heavy in the analysis of high purity Mo, which seriously affects the accuracy of the measurement results. Matrix matching method, standard addition method and isotope dilution mass spectrometry (IDMS) based on isotope abundance ratio measurement can effectively compensate for the complex matrix effect, and accurate analytical results can be obtained, but the analysis processes are complicated, the efficiency is low, and the cost is high, making it difficult to meet the measurement requirements of high throughput.　　To meet the requirements of rapid analysis, on-line analysis based on flow injection²⁸ and tandem calibration²⁹ was used. The online matrix separation and pre-concentration process for flow injection not only eliminates the matrix effect and reduces the detection limits, but also improves the automation level, and achieves high throughput measurement^31-33. Wang et al.³² measured a variety of trace metal elements in seawater samples within 28.5min based on online ion exchange. An online calibration process through a new analytical methodology called tandem calibration could be achieved based on a dual sample introduction system with two nebulizers working in parallel²⁹ or a multiple-channel nebulizer³⁴, for which the special design of the nebulizer is necessary.

OBJECTIVES To develop a rapid and accurate method for the analysis of trace impurities in high purity molybdenum to meet the requirements of high throughput sample tests.

METHODS A dual-channel simultaneous sampling ICP-MS measurement method based on online-standard-addition was established by calibrating the difference in sample-standard injection flow rate. The in-house online platform was set up based on an auto-sampler and two peristatic pumps. The pure Mo reference material candidates containing homogeneous impurities with mass fraction of 0.1-200μg/g were used as samples. About 0.1g sample was digested based on the China national standard GB/T 4325.26—2013 and diluted about 1000-fold for analysis. The digested sample solution was introduced from a tube (designated as A) continuously by using a peristaltic to a T-joint in which the sample solution and a series of standard solutions from another tube (B) using an auto-sampler were mixed thoroughly. The mixed solutions were introduced through a nebulizer and rotary spray chamber, ionized in the plasma, and then introduced to mass spectrometry for analysis. For more accurate results, the difference flow rate between the sample and standard tube (the internal diameter of the sampling tube was 0.25mm and 0.38mm for the sample and standard, respectively) was calibrated by using the 20ng/mL standard solution. The ratio (R₁) of intensity using A tube injection to the intensity using B tube injection was 1.54±0.06 for each element, indicating the difference of injection flow for A and B tube. The ratio of intensities for simultaneous sample injection to the sum intensities of A and B injection respectively was about 1.

RESULTS Methodological parameters for ICP-MS method based on online-standard-addition were evaluated in terms of method detection limit (MDL), linearly dependent coefficient, precision and accuracy. The limits of detection of this method based on the 3-σ criterion ranged from 0.004μg/g to 0.90μg/g for 29 elements, which meets the requirements of trace impurities analysis. The linear correlation coefficient r of the standard curve was mostly more than 0.999 at the concentration range of 0-500ng/g. The precision was evaluated by introducing the Mo digested solutions from A tube and 5ng/mL standard solution from B tube simultaneously, and the relative standard deviations (RSDs) of 6 repeated measurements were low, ranging from 0.8%-3.9% for 28 elements, in spite of the relative high RSD for Nb (6.1%). The multi-elements standard solution of GBW(E)082429, GBW(E)082430, GBW(E)082431were used for accuracy evaluation. After flow rate calibration, the relative deviation (RDs) between measured values and certified values ranged from −6.8% to 5.1% for 27 elements. For Ca and Zn, the RDs were relatively large (32% for Ca and 13% for Zn), which was probably due to the contamination. To further evaluate the method reliability, the analytical results of K and Cr were compared with those by using the IDMS method. Relative deviation of less than 1% between the two methods was found.　　Calibration of the difference for flow rate between the sample and standard injection. When the sample and standard is introduced from different tubes or nebulizers, the efficiency should be calibrated. The mathematical correction used was first introduced by Salin et al.³⁶ A dual sample introduction system based on two nebulizers working in parallel was set up by Canals et al.²⁹ The transport efficiencies for online standard addition were calibrated and the deviations of less than 3% for Na, K, Ba and Rb elements analysis in different matrixes were obtained. Different from the reported systems which needed to be specially designed^29,34, the online platform in this work was simple and easy to apply, because it only used a T-joint for the mixing of the sample and standard. This similar setup was found elsewhere^37-38. In this setup, the same nebulizer was used for the sample and standard solution, and thus the same nebulization efficiency was obtained. The difference of signals between using A tube injection and B tube injection only resulted from the flow rates. In the flow process, fractionation could not occur. Therefore, the very consistent R₁ factors were found for different elements, which indicated that it was not needed for the measurement of R₁ factors for all elements.　　The merits of online-standard-addition ICP-MS method.Satisfactory limits of detection were obtained for 29 elements ranging from 0.004μg/g to 0.90μg/g, which was better than the MDL derived from the China national standard GB/T 4325.26—2013(1μg/g) and was comparable with the reported method based on ICP-MS/MS⁶. In fact, compared with the traditional standard addition method, the online method would attenuate the MDL due to the dilution of sample solution, although Schwartz et al.³⁷ deemed that there was no obvious difference between online and traditional methods in terms of SNR (Signal to Noise Ratio) and precision. The good precision 0.8%-3.9% for 28 elements was better than literature^37-38. Apart from Ca and Zn, the satisfactory accuracy for 27 elements (relative deviation ranged from −6.8% to 5.1%) was obtained, which was better than the online internal standard method³⁹ because of the calibration of matrix effect for the standard-addition method. The comparative results of Cr and K by using the online-standard-addition method (this work) and the IDMS method showed good agreement of ±1%, which further validates the reliability of this work. It can be concluded that the accuracy and precision of this established method were comparable with the IDMS method. However, the analysis efficiency and cost showed obvious advantages, because the measurement of 29 elements could be completed in 15min, and thus the analysis efficiency was hundreds of times higher than that of the IDMS method.

CONCLUSIONS The method combined the properties of high accuracy and high throughput, and thus can achieve rapid and accurate trace impurities analysis in pure Mo.