Abstract

A long term air quality study is being conducted in Roxana, Illinois, USA, at the fenceline of a petroleum refinery. Measurements include 1-in-6 day 24-hour integrated ambient fine particulate matter (PM2.5) speciation following the Chemical Speciation Network (CSN) sampling and analysis protocols. Lanthanoid elements, some of which are tracers of fluidized-bed catalytic cracker (FCC) emissions, are also measured by inductively coupled plasma – mass spectrometry (ICP-MS) after extraction from PM2.5

using hot block-assisted acid digestion. Lanthanoid recoveries of 80-90% were obtained for two ambient particulate matter standard reference materials (NIST SRM 1648a and 2783). Ambient PM2.5 La patterns could be explained by a two-source model

representing resuspended soil and FCC emissions with enhanced La/Ce ratios when impacted by the refinery. Nonparametric wind regression demonstrates that when the monitoring station was upwind of the refinery the mean La/Ce ratio is consistent with soil and when the monitoring station is downwind of the refinery the mean ratio is more than four times higher for bearings that corresponds to maximum impacts. Source

apportionment modeling using EPA UNMIX and EPA PMF could not reliably apportion PM2.5 mass to the FCC emissions. However, the weight of evidence is that such

contributions are small with no large episodes observed for the 164 samples analyzed.

18 This study demonstrates the applicability of a hot block-assisted digestion protocol for the extraction of lanthanoid elements as well as insights obtained from long-term monitoring data including wind direction-based analyses.

2.2. Introduction

Petroleum refinery operations have historically been associated with emission of a variety of volatile organic compounds (VOCs) from transport, processing and storage of gases and liquids, whereas their contribution to ambient particulate matter (PM) has been less studied. However, studies by Chellam and coworkers (Bozlaker et al., 2013; Kulkarni et al., 2007a; Kulkarni et al., 2006; Kulkarni et al., 2007b; Kulkarni et al., 2007c) and Moreno et al. (2008a) have drawn attention to particulate matter (PM) primary emissions from petroleum refinery fluidized-bed catalytic cracking (FCC) units which are used to crack the heavy, long-chain molecules in crude oil feedstock into lighter, shorter-chain products. While lanthanoid element concentrations in the atmosphere are typically dominated by emissions from natural sources such as resuspended crustal material, anthropogenic processes including but not limited to motor vehicle emissions, ceramic industries and refinery operations can be significant contributors (Bozlaker et al., 2013;

Kitto et al., 1992; Kulkarni et al., 2006; Moreno et al., 2008b). Particulate matter emitted from the FCC unit is mainly comprised of zeolite catalyst material enriched in crustal elements such as Al and Si as well as lanthanoid elements with a notable enhancement in lanthanum; thus, La/Ce ratios and La/Sm/Ce ratios have been used to identify ambient PM samples with FCC emissions impacts with supporting evidence provided by source apportionment modeling (Bozlaker et al., 2013; Kulkarni et al., 2007a; Kulkarni et al.,

19 2006) or air mass classifications derived from air mass back trajectories (Moreno et al., 2008a).

Chellam and coworkers have focused on measurements in the greater Houston, Texas (USA) area. In a one month study with 114 short-term (3 or 6 hour) PM10 samples they concluded that 70% of samples were negligibly impacted by FCC emissions (Bozlaker et al., 2013). Kulkarni et al. (2006; 2007b) apportioned 1-2% of the ambient fine PM (PM2.5) to refinery operations during days without FCC emission episodes for twenty-five 24-hour samples collected over a nominally three month period. In contrast,

contributions as high as 37% of ambient PM2.5 mass were estimated for emission episode days including one event estimated to have released about 57 kg of PM2.5 over two days in 2005 and another event that released about 45 tons of catalyst because of a break in the FCC unit cyclone of the FCC in 2006 (Kulkarni et al., 2007b). The role of transient emissions events and their reporting was further examined by Bozlaker et al. (2013).

Moreno et al. (2008a) conducted a 14-month sampling campaign that collected 110 PM10

samples and 111 PM2.5 samples at a monitoring station located about 3-4 km from the FCC unit at a petroleum refinery in Puertollano, Spain. Air mass back trajectories were used to classify each sampling day as crustal North African, crustal non-North African, local, regional and oceanic advective scenarios. Scenarios presumed to be dominated by local influences exhibited the highest La enrichment compared to Ce and Sm. Three days were deemed “lanthanum anomaly days” with particularly high La enrichment and in each case the prevailing winds were from the direction of the petroleum refinery.

20 The aforementioned studies demonstrate the utility of using lanthanoid elements to track FCC unit emissions impacts. Ambient PM samples, especially from routine monitoring networks, are commonly analyzed for elemental composition by x-ray fluorescence (XRF) but for the air volumes sampled the trace to ultra-trace level lanthanoid

concentrations are typically well below detection limits. Instrumental neutron activation analysis (INAA) and inductively coupled plasma-mass spectrometry (ICP-MS) have been widely accepted as accurate methods to quantify ambient PM lanthanoid elements

(Kowalczyk et al., 1982; Olmez and Gordon, 1985). ICP-MS has become the preferred method because of its easy accessibility, relatively low cost, and less stringent

requirement for laboratory infrastructures. High temperature, high pressure acid digestion is often used to extract a wide range of elements from ambient PM samples (Kotchenruther, 2013; Kulkarni et al., 2007a; Kulkarni et al., 2007c; Moreno et al., 2008a; Wu et al., 1996) and microwave-assisted acid digestion is commonly used when the focus is on lanthanoid elements (Celo et al., 2012; Kulkarni et al., 2006; Moreno et al., 2008a).

In this study, ambient PM2.5 data were already available for many elements from the routine XRF analysis and we sought a low-cost, high-throughput method to analyze lanthanoids in these samples. The prevailing microwave digestion systems usually provide a digestion environment with temperature of about 200C and pressure of about 200 psig (Danadurai et al., 2011; Kulkarni et al., 2007a; Kulkarni et al., 2007c; Wu et al., 1996). In contrast, we evaluated the capability of extracting lanthanoid elements using hot plate-assisted digestion which is conducted at moderately high temperature (90C in our case) and atmospheric pressure. Ambient PM2.5 filter samples collected over a

three-21 year period at the fenceline of a petroleum refinery were analyzed for lanthanoid

elements to track FCC emissions. Nonparametric wind regression was used to determine the La concentrations and La/Ce ratios when the sampling site was upwind and

downwind of the FCC unit.