1991 Volume 26 Issue 2 Pages 59-71
In this review receptor models, such as chemical mass balances (CMBs), factor analysis, multiple linear regression are described.
Good results of CMBs can be oftained only when the source profiles closely match those of components actually present. New markers have been identified; e. g. rare earths, which are associatedwith oil refineries and possibly late-model automobiles equipped with catalytic converters. Even now, however, source identification of trace elements in fine particles have not been sufficiently performed.
Regional sulfate in earstern United States is accompanied with specific trace elements. Concentration ratios of sixtrace elements (As, Sb, Zn, In, noncrustal V, and noncrustal Mn) to that of Se are used to identify particles from any of their five pollution source regions. The most promising hybrid regional-scale model is quantitative transport bias analysis (QTBA), which can treat receptor data to obtain contour maps of concentrations of sulfate and various trace elements at the receptor when air masses arrived from the surrounding areas.
Organic compounds are also used for source apportionment. Contributions from threeurban sources of PAHs, domestic heating, petrol vehicles and diesel v ehicles, were calculated by CMB from six PAHs that are present in the atmosphere only in particle form. On the other hand, several volatile organic compounds have been identified as candidates for tracers of fine-particulate carbon and extractable organic matter from mobile sources.
To save the need to identify and assay the hundreds of organic compound that may be mutagenic, multiple linear regression of revertant colonies vs. K and Pb was applied to apportion mutagens to motor vehicle emissions and wood smoke.
There is great potential for development of highly definitive identification of particles by analyzing the data obtained by using the scanning electron microscope equipped with the X-ray fluorescence system.
The use of accelerator mass spectrometry has enabled one to measure the 14C/total carbon ratio from the several ten micrograms carbon sample. This technique may be applied to assign clearly individual compounds and individual carbonaceous particles to fossil fuel or temporary carbon sources.
Factor analysis models have been widely applied to the screening for unexpected sources, source apportionment, and deriving source compositions, and many reasonable results have been obtained. But factor analysis cannot be successfully applied to all the data sets. The success of this method strongly depends partly on the adequate selection of the species to include in the data set and partly on the coordinate rotation technique. New models which include such complex algorithms as non-negative transformation have been developed actively because of the wide use of computers.
