AN OBJECTIVE ASSESSMENT OF ANALYTICAL METHOD PRECISION - COMPARISON OF ICP-AES AND XRF FOR THE ANALYSIS OF SILICATE ROCKS

The precision of an analytical method has been evaluated objectively by applying the method of Thompson and Howarth (1976) to the analysis in duplicate of 55 igneous rocks covering a range of silicate matrix types and analyte concentrations. Results were analysed to characterise the change in precision (S-c) of the analytical method with concentration (c) according to the equation S-c=S-o+kc, where the k parameter represents the limiting high-level precision and S-o, the precision at zero concentration, which is related to the method detection limit (MDL), Test materials were analysed using four analytical methods based on two analytical techniques, inductively coupled plasma-atomic emission spectrometry (ICP-AES) and X-ray fluorescence spectrometry (XRF), as operated under routine working conditions in the two participating laboratories. The two XRF methods were major elements on fused glass discs and trace elements on powder pellets, and the two ICP-AES methods were major elements after a fusion decomposition technique and trace elements together with selected major elements, after an acid attack. Statistical evaluation of the data showed that significant changes in precision as a function of concentration (i.e. the k factor) were determined in 34 cases out of 78 analyte-method combinations. In cases where no significant change in precision could be detected, a grand mean precision, representative of the concentration range analysed was calculated. The S-o parameter was found to be significantly different from zero in 36 cases out of 72. To allow evaluation of the detection limit performance of all data, a maximum method detection limit (MMDL) was calculated, which was estimated to be on average 1.62 times greater than the MDL derived from significant values of S-o. In terms of the four methods studied, median high-level precision of the techniques used to determine major elements were found to be 0.23% relative (XRF/glass discs), 0.43% (ICP-AES/fusion decomposition) and 0.70% (ICP-AES/acid attack). Typical precision values in the determination of trace elements by both techniques was 1.5%, providing elemental concentrations extended over a significant range. MMDL's varied from element to element but for XRF/powder pellet data were found to be approximately equivalent to instrumental detection limits (IDL's) calculated from background count rates. However, for trace elements determined by ICP-AES/acid attack, MMDL were found to be on average three times larger than IDL's measured from repeated analysis of an aqueous blank. As a result of an evaluation of these data, it is proposed that appropriate figures of merit to describe the analytical performance of a technique are: (1) median precision in the determination of major elements; and (2) the number of trace elements that can be determined to MDL's of less than one-tenth the crustal abundance of the element. These factors should then be evaluated in conjunction with logistical factors including the rate at which samples can be analysed and the cost per determination. The influence of these factors on applications of the techniques studied in pure and applied geochemistry are discussed.