Summary: | 碩士 === 長庚大學 === 企業管理研究所 === 92 === Uncertainty measurement has been recently mushroomed. With the trend of quality management on laboratory and the requirement of international standard such as ISO 17025, all laboratories have been devoted to establishing their uncertainty evaluation models.
Indeed, calibration curve is very critical in doing the analysis. Abundant of studies toward the uncertainty of calibration curve can be found in literature. Among them only the chemical analysis had improperly taken the independent variable as a constant. In fact, the uncertainty of calibration is formed with the uncertainties of both X and Y, and transmitted to the unknown determinants as well. As plenty of uncertain sources were added coincidently to the complex procedures of chemical analysis, the uncertainty of independent variables of calibration curve absolutely deserves our highly attention.
This study is in part to build a model to work out the problem that can’t be solved by the uncertainty of dependent variables of calibration curve itself, and in part to compute both the uncertainty of dependent and independent variables. That means, both of the variation of standard solution concentration and instrumental signal response have all been put into consideration in this model. By further following the procedures of international guide, this model is able to mistakably transmit uncertainty of all steps to the analysis results of unknown samples without distortion.
Diverse uncertainties have been appeared in the course of samples preparation, such as the balance on weighing samples, the volume of flask and the temperature on dilatation during dilution, the transfer volume of pipette being used, and even the adopted concentration of standard solution. By comparison, calibration curve has generated an uncertainty that is far reachable than the one obtained from the procedures of preparing samples. Evidently, it is too simplified to take the independent variable X as a constant variable when establishing a calibration curve.
This study has used the Inductively Coupled Plasma (ICP) to analyze the standard samples of the aluminum alloy to get a figure, which is then put into calculation separately through the statistics methods ever been applied and being adopted by this study. Evidently, the latter has resulted in a more reasonable uncertainty to exclude the overestimation of the uncertainty analysis.
This research results show that the uncertainty measurement executed by statistics has been made on emphasizing the repeated analysis procedures, failing to form the calibration curve instrument signal deviations uncertainty and transmit it to the samples analysis. Our study, however, is able to correctly reflect the uncertainties both of the dependent and independent variables, giving acceptable uncertainty measurement to the analysis.
|