Statistical analysis as a tool for discriminating dioxin formation pathways

Typical thermal fingerprints are rather evenly composed of all available polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/F) congeners, a feature also linked with CuCl₂-catalysis. Conversely, chlorophenol condensation patterns are supported strongly by CrCl₃-, ZnCl₂- and CdCl₂-catalysis; they feature selective formation of, e.g., 1,3,6,8- and 1,3,7,9-TCDD arising from mainly 2,4,6-trichlorophenol, a chlorophenol well-represented in incinerator flue gas. Just few, well-defined PCDD-congeners steadily emerge from chlorophenols. Surprisingly, it is experimentally established that these are not generally formed by parallel processes. Additionally, the formation of four important 2,3,7,8-substituted TCDD/F- and PeCDD/F-congeners is statistically quasi-unrelated. Apparently, external (α) or lateral (β)-chlorination occurs as a chance event, contrary to observations and hypotheses in earlier studies. These statements are corroborated by selected congener-to-congener representations. For performing this statistical analysis, simple techniques are followed. Cross-correlation matrices show a relationship for each pair of PCDD/F-congeners in a particular data set: PCDD/F-congener pairs statistically correlate, anti-correlate, or appear rather neutral. In this paper, these novel concepts were applied on data, established during tests on Model Fly Ash (MFA); principal component analysis was used to demonstrate the relevance of MFA-data in relation to municipal solid waste incineration signatures.