Final Report: Near Real-Time Speciation of Organic Aerosols for Source Apportionment

EPA Grant Number: R832166
Title: Near Real-Time Speciation of Organic Aerosols for Source Apportionment
Investigators: Johnston, Murray V.
Institution: University of Delaware
EPA Project Officer: Chung, Serena
Project Period: January 1, 2005 through December 31, 2007 (Extended to December 31, 2008)
Project Amount: $450,000
RFA: Source Apportionment of Particulate Matter (2004) RFA Text |  Recipients Lists
Research Category: Air Quality and Air Toxics , Particulate Matter , Air

Objective:

Source apportionment of ambient organic aerosol is normally performed by quantification of specific molecular marker compounds by gas chromatography mass spectrometry (GC-MS) analysis of particulate samples collected over a 3-24 hour period. Although the molecular marker approach has proved to be quite effective, collection of a sufficient number of samples takes a long time, typically a month to a year. The range of compounds that can be used as markers is restricted because plumes from different sources have an adequate time to mix over each collection period. Therefore, markers must be sought that are specific to one and only one source. In principle, composition measurements with much higher time resolution would facilitate the apportionment process by substantially decreasing the time needed to collect a sufficient number of samples and broadening the range of compounds that can be used as markers. Because a natural break point in the wind power spectrum occurs around 1 hour, it stands to reason that performing molecular composition measurements on this timescale or shorter would be advantageous. In this project, highly time-resolved organic molecular composition data obtained with the photoionization aerosol mass spectrometer (PIAMS) were used to identify and apportion organic aerosol sources at an urban air quality monitoring site.

This project included five specific objectives: (1) develop the photoionization aerosol mass spectrometer (PIAMS) for near real-time measurement of trace organic constituents in fine particles; (2) couple PIAMS with an aerosol concentrator to reduce the sampling time of PIAMS to 2 minutes or less for ambient urban air; (3) acquire PIAMS source signatures (source profiles) for common sources of ambient organic aerosol; (4) use the concentrator and PIAMS to measure ambient organic aerosol constituents with high time resolution in Wilmington, Delaware, at the State of Delaware Department of Natural Resources and Environmental Control Air Quality Measurement Site; and (5) perform source apportionment of ambient organic aerosol using highly time-resolved organic molecular composition data.

Summary/Accomplishments (Outputs/Outcomes):

At the beginning of this project, the photoionization aerosol mass spectrometer (PIAMS) was reconfigured for field work, including (a) hardware and software development for automated sample collection and data acquisition, (b) an improved collection probe design that maintained robust alignment with the particle beam and permitted the probe surface to be cooled to inhibit evaporation of semivolatile compounds, (c) redesign of vacuum components for robust use in the field, and (d) development of a virtual impactor to more closely match the inlet flow rate of PIAMS with the outlet flow rate of the m-VACES (see below). PIAMS was then coupled with the mini-versatile aerosol concentration system (m-VACES) developed by the University of Southern California, and preliminary measurements of ambient aerosol were performed by drawing outside air into the research laboratory. The preliminary measurements showed that good quality mass spectra could be obtained with the m-VACES/PIAMS combination using a 2-minute aerosol sampling time, which corresponded to a total turnaround time of 3.5 minutes for both sampling and analysis. The mass spectra showed ion signals consistent with source signatures compiled previously with PIAMS. Additional signatures were obtained for biogenic secondary organic aerosol (Tolocka et al., 2006; Heaton et al., 2007).

Objectives 4-5. Field measurements and source apportionment of ambient aerosol were performed during two separate time periods, a 7-day period in June 2006 (2200 samples analyzed by PIAMS) and an 18-day period in October-November 2007 (6244 samples analyzed by PIAMS).  The results were reported in two separate publications (Dreyfus and Johnston, 2008; Dreyfus et al., 2009).

Aerosol sampling was performed at the State of Delaware Air Quality Monitoring Site on Martin Luther King Boulevard in Wilmington, Delaware. The site is located in a mostly urban area, within a short distance to a major highway (Interstate-95), numerous local roads, a transit bus depot, a railroad station, restaurants, apartments and small businesses. Major industrial facilities are located 5-10 km from the site, including a coal fired power plant, refinery, steel mill and other manufacturing facilities. Ambient air was sampled through an inlet with a 50% transmission cutoff of 2 µm before being split to the various instruments. The total sample flow through the inlet was 43 lpm:  30 lpm to the mini versatile aerosol concentration enrichment system (m-VACES; (Geller et al., 2005), 8 lpm to an elemental carbon/organic carbon analyzer (EC/OC), and 5 lpm to the scanning mobility particle sizer (SMPS). The concentrated minor flow exiting m-VACES at 1.0 lpm was split between PIAMS (0.1 lpm) and a small metering pump (0.9 lpm). The timing and sampling parameters for PIAMS were as follows: aerosol was sampled for 2 minutes; analysis followed for 20 seconds (repeated firings of the desorption and photoionization lasers at 10 Hz); and finally, the desorption laser was free-fired for an additional 70 seconds to clean off any remaining material before the next sample period began. The EC/OC instrument sampled aerosol through an in-line denuder to reduce positive artifacts; analysis was performed using the temperature steps of the NIOSH protocol. Gas-phase (CO, O3, NO, NO2, NOx), and PM2.5 measurements were collected hourly by the Delaware Department of Natural Resources and Environmental Control. Wind speed and direction were saved with each PIAMS spectrum. Positive matrix factorization was performed on the PIAMS dataset with the PMF2 algorithm. A total of 60 specific m/z species were selected for inclusion in the PMF model based on (1) molecular weights and prominent fragment ions of known markers for organic aerosol sources, and (2) prominence in the PIAMS mass spectra. Complete discussions of the experimental and computational procedures used in the ambient studies are given in the Dreyfus publications.

PMF modeling of the PIAMS data identified six factors that could be linked to specific sources or types of compounds based on prominent m/z species in the mass spectra, diurnal dependencies, wind rose plots, correlations with other gas and particle phase measurements at the site, and PIAMS source signatures. The first factor, identified as diesel exhaust representing 42% of total organic carbon (TOC), is dominated by alkyl fragment ions (e.g., 83 and 97 m/z) associated with aliphatic compound from unburned lubricating oil. Other prominent species associated with vehicular traffic apportioned to this factor include benzothiazole (138 m/z), benzenedicarboxylic acids/3,4-dimethoxybenzaldehyde (166 m/z), 3,4-dimethoxybenzoic acid (182 m/z) and hopanes (common fragment ion at 191 m/z). The diurnal dependence of this factor shows a substantial increase in the apportioned signal during the early morning rush hour followed by a gradual decrease during the remainder of the day back to its nighttime minimum level. The morning increase coincides with the time period that hydrocarbon nanoparticles are detected at this site with the nanoaerosol mass spectrometer as transit buses are stored at a depot ~0.5 km to the west northwest of the site depart on their daily routes. The second factor, car/road dust representing 26% of TOC, shows prominent ions associated with vehicle emissions (hopanes at 191 m/z and alkyl cyclohexanes at 308, 322, 336, 350 m/z) and road dust (benzothiazole at 135 m/z and triacontanoic acid at 452 m/z).  The diurnal dependence is much different from the diesel factor, showing a ~20% increase during the daytime into the early evening. This dependence is consistent with emission and/or atmospheric processing of semivolatile compounds whose concentrations build during the day, which drives more and more material to the particle phase. 

The third factor, meat cooking representing 12% of TOC, is identified by prominent ions at 256, 280 and 284 m/z, corresponding to palmitic, linoleic and stearic acids, respectively. The diurnal dependence of this factor shows two characteristic peaks, the first at noon where the apportioned signal increases ~40% from the early morning low and the second in the evening around 9:00 p.m. where the apportioned signal increases ~100% from the afternoon low. The signal decreases slowly through the night and returns to the morning low around 8:00 a.m. This dependence is consistent with typical meal times. The wind rose plot shows two peaks, one to the north in the direction of a fast food restaurant with intensive meat cooking ~0.4 km from the site and the other to the south in the direction of a few restaurants slightly further from the site that also are heavily engaged in meat cooking. The fourth factor, alkanes/alkanoic acids representing 16% of TOC, shows ions corresponding to alkanes (338, 450, 576 and 590 m/z) and alkanoic acids (340, 466, 550 m/z), and fragment ions that appear to be the products of aliphatic acids, esters and/or alcohols at 239, 265, 285 and 313 m/z. The 313 m/z ion is a known fragment ion of cholesterol. Alkanes and alkanoic acids are major constituents of several organic aerosol sources including cooking, vehicles, road dust, cigarette smoke, vegetative detritus, and residential and commercial heating. The diurnal dependence shows two maxima similar to the meat cooking factor, one around noon and the other in the evening. This dependence suggests that activities associated with food preparation and/or processing are the main sources of this factor. Unlike the meat cooking factor, the alkane/alkanoic acid factor does not appear to be restricted to a few specific restaurants near the site, since the wind rose plot shows no directional dependence. 

The fifth factor, phthalates representing 1% of TOC,  is characterized by fragment at 149, 167, and 279 m/z, which are prominent in the 70 eV EI mass spectra of isomeric dioctyl phthalates. Another ion, 183 m/z, is characteristic of substituted benzoic acid species. The diurnal dependence shows an increase in the nighttime hours (3-5 am) and the late afternoon. The wind rose plot shows an enhancement to the east southeast, where several industrial facilities are located ~5 km from the site. In a previous study of Wilmington aerosol with a single particle mass spectrometer, we identified several particle classes associated with industrial emissions that showed an enhancement in the diurnal dependence at 3-5 am and in the wind rose plot to the east southeast. These similarities with the phthalate factor suggest that industrial emissions contribute to this factor. The sixth factor, polycyclic aromatics (PAHs) representing 3% of TOC, is characterized by ions corresponding to both aliphatic and aromatic species: 252 (chrysene, benzanthracene), 276 (benzoperylene), 278 (benzochrysene; pentacene), 300 (coronene), 352 (dibenzoperylene; pentacosane), 354 (bianthracene, tricosanoic acid), 430 (dinapthylanthracene), 464 (tritriacontane) and 478 (tetratriacontane) m/z. The diurnal dependence shows that the apportioned signal for this factor is  ~60% higher at night than during the day, most likely caused by an increased partitioning to the particle phase at night owing to decreasing temperature and mixing height. 

Concurrently with the PIAMS measurements, time-resolved EC/OC and gas phase data (O3, NOx, CO) were used to distinguish primary (POC) and secondary (SOC) organic carbon. For this measurement period, almost all of the organic carbon was classified as primary and of this approximately one third could be assigned as combustion POC and the other two thirds as non-combustion POC. The combustion POC had a similar diurnal dependence to the diesel factor, showing a clear peak around 8 am during the morning rush hour. In contrast, noncombustion POC exhibited a minimum at 8 am, building up to much higher concentrations during the day and early evening. A dip in the noncombustion POC concentration in the mid/late afternoon was reminiscent of the meat cooking and alkane/alkanoic factors which showed relative maxima during the noon and evening meal times and a minimum in between.  These dependencies suggest that the combustion POC is represented by the diesel factor, while noncombustion POC is represented by the sum of the remaining factors. This interpretation is borne out quantitatively by correlations among the time dependencies of the various measurements. The mass loadings are consistent as well: From the PIAMS measurements, the diesel factor accounted for about 42% of total organic carbon which is similar to the 33% contribution of combustion POC. The mass loading of the sum of remaining PIAMS source factors (58%) is similar to that of the noncombustion POC (67%). It should be noted that semivolatile compounds emitted from a combustion source may be classified as noncombustion POC in the above analysis if they partition off primary particles into the gas phase and then re-partition back to the particle phase later in the day as gas phase emissions build up or in the evening when the temperature decreases. Processes such as these would explain why compounds normally associated with combustion activities (in particular, several m/z species in the PAH and Car/Road Dust factors) correlate more closely with noncombustion POC than combustion POC. 

Conclusions:

Organic molecular composition measurements with 3.5 minute time resolution were performed with the Photoionization Aerosol Mass Spectrometer (PIAMS) over two measurement periods, one in the summer (7 days; 2200 samples analyzed by PIAMS) and another in the late fall (18 days; 6244 samples analyzed by PIAMS) in Wilmington, Delaware. The time-resolved characteristics of organic molecular composition were similar for the two periods. The fall period included conventional EC/OC measurements that facilitated source apportionment. With the use of Positive Matrix Factorization (PMF), six factors were identified that could be linked to specific sources (diesel exhaust, car emissions/road dust, meat cooking) or types of compounds (alkanes/alkanoic acids, phthalates, PAHs). Owing to the inherent high time resolution of PIAMS, the temporal (diurnal) and wind direction dependencies of these factors were examined in detail to assess the impacts of point sources and atmospheric processes. The PMF results were combined with EC/OC data for source apportionment. The diesel and car/road dust factors together represented about two-thirds of TOC, while the alkane/alkanoic acid and meat cooking factors contributed most of the remaining one-third. The phthalate and PAH factors contributed very little, only a few percent of the total.  The diurnal variation and mass loading of the diesel factor correlated most strongly with combustion POC, while the sum of the remaining factors correlated well with noncombustion POC.    

In this study, highly time resolved measurements allowed source apportionment to be performed with a dataset compiled over days as opposed to months or a year with traditional methods. Although the time savings was substantial, one must remember that the results represent a snapshot in time of ambient aerosol composition. Therefore, one must take care to assure that the time period is not heavily influenced by extreme or atypical events.


Journal Articles on this Report : 5 Displayed | Download in RIS Format

Other project views: All 9 publications 5 publications in selected types All 5 journal articles
Type Citation Project Document Sources
Journal Article Dreyfus MA, Johnston MV. Rapid sampling of individual organic aerosol species in ambient air with the photoionization aerosol mass spectrometer. Aerosol Science and Technology 2008;42(1):18-27. R832166 (2007)
R832166 (Final)
  • Full-text: Taylor & Francis-Full Text HTML
    Exit
  • Abstract: Taylor & Francis-Abstract
    Exit
  • Other: Taylor & Francis-Full Text PDF
    Exit
  • Journal Article Dreyfus MA, Adou K, Zucker SM, Johnston MV. Organic aerosol source apportionment from highly time-resolved molecular composition measurements. Atmospheric Environment 2009;43(18):2901-2910. R832166 (Final)
  • Full-text: Science Direct-Full Text HTML
    Exit
  • Abstract: Science Direct-Abstract
    Exit
  • Other: ScienceDirect-Full Text PDF
    Exit
  • Journal Article Heaton KJ, Dreyfus MA, Wang S, Johnston MV. Oligomers in the early stage of biogenic secondary organic aerosol formation and growth. Environmental Science & Technology 2007;41(17):6129-6136. R832166 (2007)
    R832166 (Final)
  • Abstract from PubMed
  • Full-text: ES&T-Full Text HTML
    Exit
  • Abstract: ES&T-Abstract
    Exit
  • Other: ES&T-Full Text PDF
    Exit
  • Journal Article Tolocka MP, Heaton KJ, Dreyfus MA, Wang S, Zordan CA, Saul TD, Johnston MV. Chemistry of particle inception and growth during [alpha]-pinene ozonolysis. Environmental Science & Technology 2006;40(6):1843-1848. R832166 (2005)
    R832166 (2006)
    R832166 (2007)
    R832166 (Final)
  • Abstract from PubMed
  • Full-text: ES&T-Full Text HTML
    Exit
  • Abstract: ES&T-Abstract
    Exit
  • Other: ES&T-Full Text PDF
    Exit
  • Journal Article Zordan CA, Wang S, Johnston MV. Time-resolved chemical composition of individual nanoparticles in urban air. Environmental Science & Technology 2008;42(17):6631-6636. R832166 (Final)
  • Abstract from PubMed
  • Full-text: ES&T-Full-text HTML
    Exit
  • Abstract: ES&T-Abstract
    Exit
  • Other: ES&T-Full-text PDF
    Exit
  • Supplemental Keywords:

    Ambient air, particulates, PAH, organics, monitoring, analytical, measurement methods
    , RFA, Scientific Discipline, Air, Ecosystem Protection/Environmental Exposure & Risk, particulate matter, Environmental Chemistry, Monitoring/Modeling, Environmental Monitoring, Environmental Engineering, particulate organic carbon, atmospheric dispersion models, atmospheric measurements, model-based analysis, source apportionment, chemical characteristics, emissions monitoring, environmental measurement, airborne particulate matter, air quality models, air quality model, air sampling, speciation, particulate matter mass, analytical chemistry, modeling studies, monitoring of organic particulate matter, real-time monitoring, aerosol analyzers, chemical speciation sampling, particle size measurement

    Relevant Websites:

    Johnston Group Exit

    Progress and Final Reports:

    Original Abstract
  • 2005 Progress Report
  • 2006 Progress Report
  • 2007 Progress Report