The technique for on-line removal of buffers and for fraction selection to eliminate source clogging and suppression of electrospray ionization by unwanted constituents consisted of: (i) selecting the initial mobile phase composition to retain analytes on top of a capillary high-performance liquid chromatography (HPLC) column; (ii) washing the buffer isocratically to waste, using a two-way micro dump valve; (iii) starting the gradient and switching the dump valve to introduce analytes into the source. Peptide and protein mixtures were prepared in 0.05-0.5 M phosphate and 0.55 mM tris-based buffers, and water. After buffer removal, chromatograms and electrospray spectra were indistinguishable from those of aqueous controls, down to 1 pmol (acidic fibroblast growth factor) consumed, and up to 78 kDa (bovine transferrin) molecular weight. Aided by the dump valve, 100 fmol of angiotensin could be fractionated and identified on the slope of a 10 x 10(6) fmol of leucine enkephalin HPLC peak. On-line buffer removal and fraction selection eliminate the need for additional preparation steps than can lead to excessive sample losses.

No abstract

Flavor profile analysis (FPA) data for influent and effluent water samples from two Philadelphia treatment systems were studied, along with quality control protocols, in order to optimize the FPA method. The quality of the data was evaluated for (1) the qualitative description and (2) the quantitative response for each organoleptic descriptor.

A statistical method utilizing principal component factor analysis (PCFA) was developed for chemical/sensory taste and odor correlation. With this method, specific flavor descriptors can be correlated with specific chromatographic peaks. A background odor mechanism was assumed to explain the odors perceived at or below their odor threshold concentrations. PCFA was applied to a series of simulated data sets and chemical/sensory data obtained from drinking water samples. The simulated data sets were used to evaluate six types of chemical/sensory response equations. The response equation giving satisfactory correlation results was then used to evaluate the drinking water sample data sets. After merger of the chemical/ sensory data, the covariance between items was calculated, and PCFA was applied to the covariance structure followed by a target transformation of PCFA factors. Quality assurance evaluations of both sensory and chemical data were an integral part of the correlation procedure. The correlation study using simulated data showed that the PCFA correlation method using linear-additive (e.g., log-additive) data yielded better results (i.e., less type I and II errors) than nonlinear, nonadditive data. The quality of the drinking water sample correlation results was highly dependent on the sensory data quality.

The cause(s) of many taste and odor problems in drinking water remain unknown. If sufficient information about the problem is not available to develop a hypothesis, one can begin to develop a hypothesis to test by using correlation methods from data base information concerning the problem. This paper presents a method to develop a correlation between chemicals and sensory characteristics of.a set of water samples by Principal Component Factor Analysis. The correlations developed hypotheses for the causes of tastes and odors that must be further tested. A correlation indicates that a pattern is occurring between two data bases. In this case, the pattern is a change of a Chromatographie peak's detector response which parallels the sensory response of a flavor profile analysis panel for an organoleptic descriptor. The correlation must always be confirmed by sensory analysis. The results from an application of the method to data from the city of Philadelphia Water Department and Philadelphia Suburban Water Co. are shown. An evaluation of the results is described. The correlation between a sensory response and a chemical concentration in water was described by expanding the Weber-Fechner Law. The Weber-Fechner Law states that the odor intensity of a sensory descriptor is proportional to the logarithm of the concentration of the chemical associated with the odor. Two drinking water data sets from the Philadelphia Water Department and Philadelphia Suburban Water Co. were used to demonstrate the sensory-chemical correlation procedure. Correlations were observed. Clearly, a correlation indicates a possible relationship, a “presumptive result” that must be tested by sensory analysis to “confirm” if the relationship is true. A correlation indicates that a pattern is occurring between two data bases. In this case, the pattern is a change of a Chromatographie peak's detector response which parallels the sensory response of a flavor profile analysis panel for an organoleptic descriptor. The correlation must always be confirmed by sensory analysis as stated by the rules of scientific evidence.