Source apportionment of airborne PCDD/F at industrial and urban sites in Busan, South Korea

Highlights

• Long-term measurements of vapour and particulate PCDD/F at industrial and urban sites.

• Application of PMF to the industry dataset consisting of PCDD/F and PAH.

• Major PCDD/F sources are metallurgical industry and diesel vehicle exhaust.

• PAH congeners are significantly related to coal utilisation in metallurgical industry.

• Σ I-TEQ Sm⁻³ at a receptor is determined by composition of source profiles.

Abstract

Long-term atmospheric measurement of 17 total (gaseous and particulate) polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/Fs) has been made from 2007 to 2016 at five industrial and urban sites in Busan, South Korea, based on their persistence, bioaccumulation, and toxicity. In the present study, two pooled datasets covering a combination of 2 industry sites and 3 urban sites have been subjected to positive matrix factorization (PMF) to identify and quantify the major sources of PCDD/Fs. Additionally, PMF has been applied to the industrial urban dataset consisting of both polycyclic aromatic hydrocarbons (PAHs) and PCDD/Fs. The results show that the sum of PCDD/F mass (Σ17PCDD/Fs) at the industrial sites is determined by five major sources: non-ferrous metal production (33.7%), diesel vehicle emissions (30.2%), ferrous metal production (22.4%), other industrial emissions (11.1%), and traffic emissions (2.6%), while the PAH mass (Σ16PAHs) is predominantly associated with emissions from coal combustion, followed by traffic emissions. At the urban sites, the largest contribution to the Σ17PCDD/Fs was observed from transported emissions being emitted from metallurgical industry (75.5%), followed by diesel vehicle emissions (24.5%). The application of congener-specific toxicity to PCDD/F mass (Σ₁₇fg I-TEQ Sm⁻³) indicates enhanced contributions from the ferrous metallurgical emission factor associated with penta- and hexa-chlorinated furans across the study sites.

Introduction

Polychlorinated dibenzo-p-dioxins and polychlorinated dibenzofurans (PCDD/Fs) are ubiquitous compounds produced unintentionally via the chemical manufacture and combustion processes. They persist in the environment and bioaccumulate in the food chain. They have generated a great deal of concern over several decades due to their well-known toxic properties (Jones and de Voogt, 1999; Sweetman et al., 2000; WHO, 2010; Morales et al., 2014). According to the Stockholm Convention, estimates for PCDD/Fs emissions are required to identify the major sources and to prepare effective pollution control strategy, despite the high degree of complexity involved in developing an inventory of such emissions due to the source variability and sensitivity to specific plant-related operational conditions (Jones and de Voogt, 1999; Katsoyiannis et al., 2010).

According to the recent emission estimates, metal production activities, including smelting and refining have become the predominant industrial sources of airborne PCDD/Fs (Wang et al., 2003; Quaβ et al., 2004; Fiedler, 2007). Indeed, a substantial reduction in PCDD/Fs has been achieved following tighter regulatory controls on emissions from municipal solid waste (MSW) incineration plants, contributing predominantly to the total industrial PCDD/Fs burden in the early inventory (Eduljee and Dyke, 1996; Anderson and Fisher, 2002; Quaβ et al., 2004; Schuhmacher and Domingo, 2006). Hassanin et al. (2006) has reported that the change in the PCDD/F homologue pattern may occur either by altering the major sources over several decades or via abatement of various emissions. Recent interest has been outlined on the significance of unregulated atmospheric PCDD/Fs sources, such as traffic and domestic combustion of coal and wood (Alcock et al., 2001; Anderson and Fisher, 2002; Lavric et al., 2004; Piazzalunga et al., 2013).

Determination of toxicity weights to convert mixed congeners to a single value normalized to the toxicity of 2,3,7,8-TCDD (TEQ) has been widely used to calculate the toxic equivalency of a mixture. The sum of all the individual TEQ (Σ₁₇ mass TEQ year⁻¹) units has been widely used in the national inventory of PCDD/Fs emission (UNEP, 2003). However, single quantitative value masks show important differences in source-specific chemical composition of PCDD/Fs (DuarteDavidson et al., 1997). Individual PCDD/F congeners, having different numbers and positions of chlorine substitution may exhibit different degrees of atmospheric reactivity and congener-specific changes in their source signatures following emission. The lower chlorinated PCDD/F congeners exhibit a higher potential for reactive loss via OH oxidation, whilst the more chlorinated congeners are subjected to rapid removal by precipitation or deposition (Tysklind et al., 1993; Alcock et al., 2001, US EPA, 2012). The fact that airborne PCDD/Fs are predominantly associated with fine particles (<2.1 μm) and have long atmospheric residence times has promoted the consideration for long-range transport of size-segregated particles containing PCDD/Fs (Lohmann and Jones, 1998; Oh et al., 2002).

A few studies have reported the potential discrepancy between ambient PCDD/Fs emission estimates and atmospheric burden (Baker and Hites, 2000; Lohmann et al., 2006). Attempts were made to address individual congener-specific emissions from various PCDD/Fs sources (Alcock et al., 2001; Masunaga et al., 2003; Katsoyiannis et al., 2010) to match the source-specific contributions with the observed levels in ambient air. However, they are still highly unclear, expensive to measure, and highly variable across different sources. Similar PCDD/Fs homologue patterns have been reported among various thermal processes (Lohmann and Jones, 1998; Bright et al., 1999; Buekens et al., 2000; Lee et al., 2004; Oh et al., 2006). Additionally, the congener-specific PCDD/Fs source signatures can vary with the operating conditions (temperature, reaction time, and fuel-to-air ratio) of industrial thermal processes (Walker and Huntley, 1997; Kasai et al., 2001; Everaert and Baeyens, 2002; Lee et al., 2004; Ooi and Lu, 2011). Occasionally, it may be difficult to interpret source-specific contributions to the observed pollution levels, because measurements are determined by mixed sources with varing proportions.

The PCDD/Fs comprise a large number of individual compounds emitted in different proportions by various sources, suggesting the application of an advanced tool to the source identification for these compounds. Receptor modelling has been used widely in source apportionment studies for air quality data, as it is not based on the complex physicochemical processes of individual species. Several studies have applied factor analysis, such as principal component analysis (PCA) and cluster analysis, to separate chemical fingerprints of selected environmental PCDD/Fs according to their possible sources (Buekens et al., 2000; Ogura et al., 2001; Masunaga et al., 2003; Lee et al., 2004; Wu et al., 2010; Ho et al., 2016). However, PCA is limited by source-specific quantitative interpretation because of negative source contributions. This limitation has been overcome using positive matrix factorization (PMF) that is generally recognized as a more powerful technique than PCA. To date, few studies have reported source apportionment using ambient PCDD/Fs concentrations (Ho et al., 2016). A few researchers have successfully applied PMF for source identification and source-specific quantification of ambient PCDD/Fs (Ngo et al., 2018). However, relatively small PCDD/F datasets have been subjected to PMF and less intensive efforts made to identify sources based on published studies.

In the present work, PMF-derived congener profiles for ambient PCDD/Fs have been extensively analysed based on likely source categories. Long-term atmospheric measurements at five urban sites, including 17 individual 2,3,7,8-substituted PCDD/Fs and 17 PAH congeners produced predominantly through the thermal processes, have been assessed. Three datasets containing information on local emission sources have been subjected to PMF. Additionally, the source-specific contributions of PCDD/Fs using toxic equivalency factors (TEFs) have been assessed at the study sites in the aspect of health issue for these compounds.