1 Department of Informatics and Mathematical Modeling, Technical University of Denmark
This thesis introduces the reader to the concept of chemical environmental reference materials and their role in traceability and chemical analyses of the environment. A number of models and principles from the literature are described. Some suggestions are made as to how stability studies can be modelled when the length of the study is unknown. Experimental data has been collected from two stability studies of aqueous matrices. The first study regards the stability of TN NO² + ³-N and n in autoclaved wastewater samples over a period of 22 months. Data was collected specifically for this study with two purposes: 1) to investigate the stability of selected analytes in the chosen matrices and 2) to explore the applicability of various statistical models for the description of stability studies. Three univariate and three multivariate stability models have been applied to these data sets. The methods have been evaluated with regard to their robustness towards variations in the chemical analytical method and with regard to the number of times a significant out of control situation is indicated. The second study regards the stability of NH4-N and total phosphorous in autoclaved seawater samples. This study lasted 22 months as well. The samples were produced and stored according to a 2³ factorial design. The influences of storage temperature, UV radiation and ultra-filtration on the stability of NH4-N and total phosphorous have been investigated. A Youden plot method is suggested for the graphical evaluation of certification data. The plots illustrate consistency between replicate measurements on samples from a batch of reference material, carried out in a number of laboratories according to a staggered nested design. The development of a reference material is illustrated by a series of experiments with wastewater. The purpose was to improve ortho-phosphate (and total phosphorous) homogeneity. A procedure is suggested which includes freeze-drying and redissolving. All calculations have been performed in SAS® primarily by means of elementary procedures, analyses of variance procedures, SAS Insight and SAS IML (Interactive Matrix Language).