Revealing source signatures in ambient BTEX concentrations

Abstract

Management of ambient concentrations of Volatile Organic Compounds (VOCs) is essential for maintaining low ozone levels in urban areas where its formation is under a VOC-limited regime. The significant decrease in traffic-induced VOC emissions in many developed countries resulted in relatively comparable shares of traffic and non-traffic VOC emissions in urban airsheds. A key step for urban air quality management is allocating ambient VOC concentrations to their pertinent sources. This study presents an approach that can aid in identifying sources that contribute to observed BTEX concentrations in areas characterized by low BTEX concentrations, where traditional source apportionment techniques are not useful. Analysis of seasonal and diurnal variations of ambient BTEX concentrations from two monitoring stations located in distinct areas reveal the possibility to identify source categories. Specifically, the varying oxidation rates of airborne BTEX compounds are used to allocate contributions of traffic emissions and evaporative sources to observed BTEX concentrations.

Introduction

Volatile organic compounds (VOCs) are an important class of air pollutants found in any urban and industrial region. Some VOCs are toxic (e.g. benzene, 1,3-butadiene) while many are ozone precursors (Derwent, 1995, Na et al., 2005, Dollard et al., 2007, Kohder, 2007). VOCs are emitted by both natural and anthropogenic sources, with vehicle exhausts, evaporation of fuels, solvent usage, industrial processes, oil refining and bio-decomposition of wastes being the dominant anthropogenic sources (Derwent, 1995, Na et al., 2005). Aromatic hydrocarbons account for 20–40% of the total ambient VOCs in modern urban environments (Derwent et al., 2000), with benzene, toluene, ethylbenzene, and meta- (m), para- (p), and ortho- (o) xylene the most commonly VOCs continuously monitored in European Union member states (Larsen, 2004). The introduction of three-way catalyst and evaporative canisters in modern vehicles has been followed by a sharp decrease in urban VOC concentrations due to the large share that automotive sources used to have in VOC emissions (Dollard et al., 2007). The decrease in traffic emissions resulted in an increase in the relative shares of other VOC sources that hitherto have been relatively negligible.

In urban environments ozone production is often under a VOC-limited regime. Hence, management of VOC levels is the key for meeting the ozone concentration standards. For that purpose, automatic non-methane VOC monitoring is commonly being implemented. For example, routine automated monitoring of benzene, toluene, ethylbenzene and xylene isomers (BTEX) has been introduced to most of the air quality networks in Europe following the European directive 2002/69/EC to limit emissions of benzene at national legislations (Larsen, 2004). The datasets obtained from such a routine monitoring may contain large amounts of hidden information that can contribute to the understanding of processes and trends that govern dispersion and transformation of ambient pollutants (e.g. Yuval et al., 2007). However, gaining insight from the BTEX datasets is not simple since the recorded concentrations are often close to the detection limits of the monitoring instrumentation and naturally include large uncertainties.

Derwent et al. (2000) and Dollard et al. (2007) used simple correlation plots of observed concentrations of various VOC compounds to identify VOC sources in the United Kingdom. High correlations were found in locations where traffic was by far the dominant source whereas poor correlations were noted in locations in the vicinity of industrial sites. Yet this method is useful only as long as no more than two plausible sources are considered, and when the ambient concentrations are well above the detection limit of the monitoring devices. The Chemical Mass Balance (CMB; Friedlander, 1973, Watson et al., 2001), the Positive Matrix Factorization (PMF; Paatero and Tapper, 1994, Paatero, 1997, Xie and Berkowitz, 2006) and variants of Principle Component Analysis (PCA; Fukunaga, 1990, Na et al., 2005, Guo et al., 2007) are more sophisticated methods commonly used for allocation of VOCs in cases of multiple sources (Hopke, 1991). Using the CMB requires specifications of the complete emission inventory, which is not available in many regions. The PMF and PCA have been proved suitable for VOC source apportionment based on monitoring data (e.g., Xie and Berkowitz, 2006). However, their successful implementation requires sampling of a large number of VOC variables with reasonably small uncertainties. As stressed by Paatero and Hopke (2003), variables with a small signal to noise ratio should be discarded from PCA and PMF analyses. In places where the VOC levels are low this means elimination of most of the recorded data points from the analysis.

Ratios of ambient concentrations of different pairs of the BTEX family have been used for various purposes (Nelson and Quigley, 1983, Guicherit, 1997, Derwent et al., 2000, Monod et al., 2001, Na et al., 2005, Kohder, 2007). In all these studies certain BTEX concentration ratios were used as markers characterizing specific sources. In particular, for the purpose of assessing traffic related ratios and for source quantification the characteristic ratios are usually considered constants (Derwent et al., 2000, Na et al., 2005). In this study, however, a more circumspect approach is taken. Variations in concentration ratios among different BTEX compounds, resulting from their distinct reaction rates with ambient oxidants, are accounted for even in cases where the source is very close to the observation point. In the troposphere, oxidation of BTEX compounds is dominated by the reaction with OH radicals whereas reactions with NO₃ and O₃ can be neglected (Atkinson, 1990). The variation in the concentrations ratio due to distinct oxidation rates can be described by (Gelencser et al., 1997, Guicherit, 1997),

$${\frac{\left[A\right]}{\left[B\right]}|}_t={\frac{\left[A\right]}{\left[B\right]}|}_0{\mathrm{e}}^{\left(k_{\text{B}}-k_{\text{A}}\right)C_{\text{OH}}t}$$

where [A]/[B]₀ is the concentrations ratio at the emission source, [A]/[B]_t is the ratio of the ambient concentrations at the monitoring station, k_A and k_B (cm³ molecule⁻¹ s⁻¹) are the reaction rate constants with OH of compounds A and B, respectively, C_OH is the ambient concentration of OH radicals (molecule cm⁻³) and t is time (s). The reaction rates of the BTEX compounds with OH radicals are shown in Table 1. Eq. (1) suggests that since any two BTEX compounds that are emitted by the same source experience the same C_OH and the same aging time, their concentrations ratio should vary from the emission point to the monitoring station only in accordance with the parameter $\Delta \equiv k_{\text{B}}-k_{\text{A}}$.

This study presents a strategy for a qualitative BTEX source differentiation in low concentrations environments. Our approach is based on using the observed manifestation of seasonally and diurnally varying oxidation rates on ambient BTEX concentrations. Namely, the variation of the concentration ratios of paired BTEX compounds is related to corresponding variations in photochemical production of OH. The variations in the observed concentrations and in their mean ratios enable insight into the impact of various BTEX sources on ambient BTEX concentrations. The approach is demonstrated using BTEX data from two monitoring stations in Haifa Bay area, Israel.