Comprehensive chemical characterization of industrial PM_2.5 from steel industry activities

Highlights

• Exhaustive PM_2.5 chemical profiles emitted by steelworks subunits are investigated.

• Sulfate is emitted by the oxygen converter process but also by combustion processes.

• PAHs and sulfur containing PAH are emitted by combustion processes.

• Calcium is emitted by all subunits but differences are obtained for Ca/Ca²⁺ ratios.

• Al, Fe, Zn, Mn, Ti are emitted by all subunits, proportion dependent of processes.

Abstract

Industrial sources are among the least documented PM (Particulate Matter) source in terms of chemical composition, which limits our understanding of their effective impact on ambient PM concentrations. We report 4 chemical emission profiles of PM_2.5 for multiple activities located in a vast metallurgical complex. Emissions profiles were calculated as the difference of species concentrations between an upwind and a downwind site normalized by the absolute PM_2.5 enrichment between both sites. We characterized the PM_2.5 emissions profiles of the industrial activities related to the cast iron (complex 1) and the iron ore conversion processes (complex 2), as well as 2 storage areas: a blast furnace slag area (complex 3) and an ore terminal (complex 4). PM_2.5 major fractions (Organic Carbon (OC) and Elemental Carbon (EC), major ions), organic markers as well as metals/trace elements are reported for the 4 industrial complexes. Among the trace elements, iron is the most emitted for the complex 1 (146.0 mg g⁻¹ of PM_2.5), the complex 2 (70.07 mg g⁻¹) and the complex 3 (124.4 mg g⁻¹) followed by Al, Mn and Zn. A strong emission of Polycyclic Aromatic Hydrocarbons (PAH), representing 1.3% of the Organic Matter (OM), is observed for the iron ore transformation complex (complex 2) which merges the activities of coke and iron sinter production and the blast furnace processes. In addition to unsubstituted PAHs, sulfur containing PAHs (SPAHs) are also significantly emitted (between 0.011 and 0.068 mg g⁻¹) by the complex 2 and could become very useful organic markers of steel industry activities. For the complexes 1 and 2 (cast iron and iron ore converters), a strong fraction of sulfate ranging from 0.284 to 0.336 g g⁻¹) and only partially neutralized by ammonium, is observed indicating that sulfates, if not directly emitted by the industrial activity, are formed very quickly in the plume. Emission from complex 4 (Ore terminal) are characterized by high contribution of Al (125.7 mg g⁻¹ of PM_2.5) but also, in a lesser extent, of Fe, Mn, Ti and Zn. We also highlighted high contribution of calcium ranging from 0.123 to 0.558 g g⁻¹ for all of the industrial complexes under study. Since calcium is also widely used as a proxy of the dust contributions in source apportionment studies, our results suggest that this assumption should be reexamined in environments impacted by industrial emissions.

Introduction

Improvement of air quality is an important concern in many environments. In order to limit the impact of air quality on human health, public authorities need reliable and accurate information regarding the PM (particulate Matter) sources contributions. In the last two decades, the development of source apportionment approaches (Canonaco et al., 2013, Paatero and Tapper, 1994, Schauer et al., 1996) has considerably improved our knowledge of the relative impact of the various primary PM sources. One constant of the main sources apportionment approaches developed (Chemical Mass Balance, Positive Matrix Factorization or Multilinear Engine) is the a priori knowledge, at different extent of accuracy, of the chemical profiles of each emissions sources. However, comparisons between these different source apportionment approaches showed significant differences especially in regards to the industrial sources. For example, a comparative study between CMB and PMF approaches (Okamoto et al., 2012) showed that even if the sources contributions are well correlated, PMF attributed to the steel mill source about 2.5 times more PM mass than in results derived from a CMB analysis. This gap is explained by differences in the steel mill aerosol chemical profiles which mainly differ by the distribution of specific trace elements (ie. Ti and Fe, mostly). Similar discrepancies were observed by another inter-comparison study of source apportionment approaches (PMF, CMB and PCA) in an industrial area (Viana et al., 2008).

Numerous studies have been carried out to characterize the chemical source profiles of vehicular emissions (El Haddad et al., 2009, Liu et al., 2010, Lough et al., 2007, Rogge et al., 1993a, Rogge et al., 1993b, Schauer et al., 1999b, Schauer et al., 2002b), biomass burning (Robinson et al., 2006a, Rogge et al., 1998, Schauer et al., 2001, Simoneit et al., 1999, Nolte et al., 2001) and food cooking (Nolte et al., 1999, Robinson et al., 2006b, Rogge et al., 1991, Schauer et al., 1999a, Schauer et al., 2002a). Among the main primary aerosol anthropogenic sources, industrial emissions are the least documented in the literature. This lack is mainly due to the difficulty to get representative source profiles. The number of industrial sources associated with a wide range of processes which are, in most cases, not continuous accentuate this difficulty. For example, in metallurgy, two distinct processes exist to produce molten steel: the basic oxygen furnace and the electric arc furnace. Previous studies (Larsen et al., 2008, Yatkin and Bayram, 2008) have shown that, if both processes emit the same trace elements such as calcium (Ca), iron (Fe) and zinc (Zn), their proportions are significantly different according to the processes considered. The basic oxygen furnace emits more calcium while the electric arc furnace emits more iron and zinc. The Ca/Fe ratio is indeed 73 times higher for the basic oxygen furnace. Emission of lead (Pb) is also observed in high proportion for the electric arc furnace (0.08 g g⁻¹; Yatkin and Bayram, 2008) but is only weakly emitted by the basic oxygen furnace (0.001 g g⁻¹; Larsen et al., 2008). Insights in aerosol chemical composition of industrial activities have been provided using either field measurements conducted in the vicinity of an industrial complex or either measurements carried out directly in the stack (Riffault et al., 2015, Hleis et al., 2013, Baraniecka et al., 2010, Dall’Osto et al., 2008, Okamoto et al., 2012, Rogge et al., 1997a, Rogge et al., 1997b, Sánchez de la Campa et al., 2010, Weitkamp et al., 2005, Yang et al., 2002, Yang et al., 1998, Yoo et al., 2002, Leoni et al., 2016). Some studies highlighted the importance of trace elements and metals such as Al, Fe, Ca, Ni, V, Zn, Pb or Mg (Dall’Osto et al., 2008, Guinot et al., 2016, Hleis et al., 2013, Kfoury et al., 2016, Mbengue et al., 2017, Pokorná et al., 2015, Taiwo et al., 2014, Weitkamp et al., 2005, Yoo et al., 2002) while others revealed high emission rates of organic compounds such as Polycyclic Aromatic Hydrocarbons (PAHs) (Baraniecka et al., 2010, Leoni et al., 2016, Yang et al., 1998, Yang et al., 2002). The characterization of both inorganic and organic aerosol fractions is thus required in order to build comprehensive and representative industrial source profiles.

Industrial emissions profiles have mostly been established by mean of direct measurements in the stacks (Buonanno et al., 2011, Chen et al., 2013, Tsai et al., 2007, Yang et al., 1998, Yang et al., 2002). While this approach provides straightforward and detailed information of the composition of the emissions associated to one specific industrial process, it suffers from 2 biases that limit the use of the chemical profile obtained. Due to the high concentrations and temperatures prevailing in industrial stacks, emissions do not reach a thermodynamic equilibrium, thus the gas-particle partitioning cannot be considered as representative of the ambient atmosphere. This results mainly in an overestimation of the Organic Carbon (OC) and other semi volatile organic compounds emission factors. Furthermore, the global impact of an industrial complex cannot be assessed by only considering the emissions of the main stack exhausts. Diffuse and fugitive emissions can be captured by the study of the enrichments of atmospheric pollutants downwind from an industrial complex, using an upwind reference. This kind of methodology has been successfully adopted in several studies such as Weitkamp et al., 2005, Alleman et al., 2010, Dall’Osto et al., 2008 or Lim et al. (2010). Such enrichment based approaches are more difficult to implement and the choice of both up and downwind sites must be addressed with cautions. While the upwind measurements site must be representative of the regional background air pollution, the downwind site must be located close to the studied sources in order to avoid interferences from other sources but far enough to capture the diversity of the industrial emissions (direct, diffuse and fugitive).

Here we report 4 chemical profiles of PM emitted by 4 subunits of a vast metallurgical complex obtained by mean of an enrichment based approach. A particular emphasis has been put in the chemical characterization of aerosol which combines, in addition to the major fractions, a large array of trace elements, metals and organic markers.