Klaus Doerffel

Beurteilung von Analysenverfahren und -Ergebnissen

Die analytische Chemie ist eine Kunst, sie zu beherrschen erfordert theoretische Kenntnisse, handwerkliches Konnen und personliche Erfahrung. Ohne umfassende theoretische Kenntnisse geht der Uberblick uber die mannigfachen Moglichkeiten und uber die Grenzen dieses Gebietes verloren, ohne sauberes handwerkliches Konnen last sich keine noch so einfache Analyse einwandfrei durchfuhren, ohne langjahrige personliche Erfahrung ist keine Beurteilung eines analytischen Problems und keine Bewertung von Analysenergebnissen moglich.

Assuring trueness of analytical results

SummaryA true analytical result presumes an analytical procedure without systematic errors. They can be detected by comparing the analytical results with the true value or with an accepted reference. Systematic errors are always superimposed by the random error, therefore such a comparison must include statistical tests. The paper describes suitable test-models for typical analytical problems, as e.g. the validation of an analytical procedure or the (current) control of analytical work. Furthermore, explanations will be given for the interpretation of the test results due to the trueness of analytical data and for the consequences concerning analytical work.

Investigations of the chemical homogeneity of solids

Abstract Chemical homogeneity is a relative property of solid-state materials; it depends not only on their ultimate purposes, but also on the parameters of the analytical procedure applied in the investigation. These features are often given inadequate attention in the characterization of solid materials. A statistical model for representative sampling is used here as the starting point to derive relationships which demonstrate quantitatively the increasing rigour of the homogeneity test with increasing spatial resolution and precision of the analytical procedure. It is also shown that chemical homogeneity can be confirmed at a calculated risk, when the tests do not show the material to be significantly inhomogeneous and when maximum permissible deviations in concentration can be given for a practical application.

The Fermi doublet V2−V3 + V4 of CH3CN as a probe of molecular interactions

Abstract The Fermi doublet V 2 − V 3 + V 4 of CH 3 CN in basic, inert and acidic solvents has been studied by IR and Raman spectroscopy. The values of W , the Fermi coupling coefficient, obtained from IR spectra varies with the nature of the solvent while W evaluated from Raman data remains constant at 12.5 ± 0.5 cm −1 . The similar effects of Bronsted and Lewis acids on the band frequencies and intensities is evidence that the CN group complexes with acids via the N atom “ n ” electron pair and not the π bond.

Nachweis systematischer Fehler durch gewichtete Regression

SummaryEach analytical method is to be tested for the absence of systematic errors. This is usually done by Youdens regression model, which is limited to precise “given” values. Furthermore, weighted regression allows the detection of systematic errors even if the random error of the “given” values is not to be neglected. This is demonstrated by simulation experiments and by testing a new spectrometric method for the determination of platinum.It must be taken into account that statistical tests can give information only about the presence of systematic errors (and not about their absence). Accuracy must be seen as a qualitative characteristic feature (yes/no); accuracy can never be given as a quantitative measure (more/less).

The shape of CUSUM — an indicator for tendencies in a time series

SummaryCUSUM (cumulative sum) can be seen as integral of the attached time series. Therefore the shape of CUSUM allows conclusions as to the type of an instationarity inherent in the time series to be made. This indirect way is advantageous for the evaluation of time series with high random fluctuations. The theoretical background for the information to be derived from the shape of CUSUM as well as examples for application (time series, distribution analysis) are shown in this paper.

Improving signal-to-noise by evaluation of correlation functions

ZusammenfassungDie Autokorrelationsfunktion und die Kreuzkorrelationsfunktion erweisen sich als wertvolle Werkzeuge zur Verbesserung analytischer Messungen. Die Autokorrelationsfunktion beschreibt die Eigenschaften des Rauschens und hilft, Fehler zu einem frühen Zeitpunkt zu entdecken. Mit Hilfe der Kreuzkorrelation können sehr schwache Signale in starkem Rauschen delektiert werden. Auf diese Weise werden Messungen unterhalb der normalen Nachweisgrenze möglich. Durch die Benutzung von Korrelationsfunktionen können Strategien für vorteilhafte analytische Messungen abgeleitet werden.SummaryThe autocorrelation function and the cross-correlation function are to be seen as valuable means for improving analytical measurements. The autocorrelation function describes the properties of noise and helps to detect errors at an early stage. By means of cross-correlation, very weak signals in a strong noise can be detected. In this way measurements below the normal limit of detection become possible. By using correlation functions strategies for carrying out analytical measurements advantageously can be derived.

Treatment of noisy data from distribution analysis using models from time-series analysis

SummaryData from distribution-analysis are often affected by a high noise. The described information about elemental distribution is frequently hidden in this noise far below Kaisers limit of detection with signal to noise of 3. To gain this useful information under these conditions (S/N ≪3) models from time-series-analysis (especially quality-control, process-control) have proven a powerful tool. The adaption of these models for the treatment of 1d-, 2d-, and 3d- distribution-analyses as well examples for application are presented.

Time-series analysis in analytical process control

Abstract Analysis of time series tries to extract tendencies from measured values dependent on time. For this purpose the cusum technique has proved to be a very sensitive tool for the evaluation of both current and completed time series. Even very weak tendencies can be detected at a high level of noise. Time-series analysis further tries to predict values to come from hitherto performed measurements. As a very flexible model exponential smoothing could be successfully used. Even for processes with a high extent of non-stationarity this model allowed a good prediction owing to the dynamics of the process. Three types of time-series analysis, i.e., evaluation of current measurements, retrospective evaluation and prediction of data (also known as “in vivo”, “post mortem” and “in futurum” time-series analysis) are demonstrated for problems stemming from analytical process control.

Zur multivariaten Auswertung von Ringversuchen

SummaryThe advantages and drawbacks of some graphical methods as interpreting aids for cooperative tests are demonstrated using results of slag sample analyses (5 laboratories, 7 components). In cases of low dimensions (components) the method of regular polygons has been found most informative. For higher dimensions modern data analysis techniques, such as principal components analysis (pca), correspondence analysis and discriminant analysis are to be preferred. From pca of correlation matrices one can also derive best choices of feature subsets to construct good polygons. The sensitivity of partial correlation coefficients due to deviating laboratory manners can be assessed by leaving-one-out comparisons.