Grantee Research Project Results
2005 Progress Report: Source-Apportionment of Primary Organic Carbon in the Eastern United States Combining Receptor-Models, Chemical Transport Models, and Laboratory Oxidation Experiments
EPA Grant Number: R832162Title: Source-Apportionment of Primary Organic Carbon in the Eastern United States Combining Receptor-Models, Chemical Transport Models, and Laboratory Oxidation Experiments
Investigators: Robinson, Allen , Donahue, Neil , Adams, Peter
Institution: Carnegie Mellon University
EPA Project Officer: Chung, Serena
Project Period: November 1, 2004 through October 31, 2007
Project Period Covered by this Report: November 1, 2004 through October 31, 2005
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:
The objectives of this research project are to: (1) measure oxidation kinetics of organic molecular markers in actual emissions (diesel engine, wood smoke, meat cooking) exposed to O3 and OH across a wide range of atmospheric conditions in a smog chamber; (2) apply receptor models (chemical mass balance [CMB], Positive Matrix Factorization [PMF], and Multilinear Engine [ME2]) to molecular marker data collected by the Supersites to apportion ambient organic carbon (OC) in the Eastern United States to sources of primary organic aerosol; (3) evaluate and improve existing emission inventories for primary OC by comparing predictions of photochemical transport models to receptor modeling results; (4) combine receptor and chemical transport to quantify the contribution of different source classes and geographic subregions to primary OC in the Eastern United States; and (5) assess the importance of photochemical aging on primary organic aerosol composition in the Eastern United States and its effects on source apportionment estimates.
Progress Summary:
The past year has been transformative. Research on this project has led to a new understanding of organic particle behavior in the atmosphere. This advance has drawn on support from other projects, notably Science To Achieve Results research on secondary organic aerosol (Donahue, Pandis, Robinson, Davidson, and Adams). The end product is a theoretical framework for the formation and transformation of organic particles, coupled with the realization that essentially all organic particulate matter (PM) is semivolatile. This realization blurs the dual notion of secondary and primary organic aerosol. We have, however, developed a framework to describe this behavior that facilitates understanding of semivolatile partitioning and also points to efficient modeling of the partitioning, including evolution of volatility caused by progressive chemical transformation during long-range transport of semivolatile material. Extensive publication of both experimental and modeling results is in progress.
A key finding during Year 1 of the project has been that primary organic aerosol emissions are semivolatile. The partitioning of this material follows according to the well-established treatment of Pankow and therefore varies with atmospheric conditions. This prevents defining a static emission factor and has important implications for the measurement and simulation of primary organic aerosol emissions. The partitioning behavior of emissions can be interpreted in terms of a “basis set” of compounds whose saturation concentrations span the full range of organic aerosol concentrations observed in the atmosphere (0.01 μg m-3 for a “nonvolatile” product to 1,000 μg m-3 for a “volatile” product, with intermediate-volatility products separated by factors of 10). The basis set representation puts all organic aerosol on a common framework, thus permitting a complete representation of semivolatile partitioning with five to seven highly constrained parameters. We are in the process of implementing this framework in Particulate Matter Comprehensive Air Quality Model with Extensions (PMCAMx). This permits:
- Uniform representation of semivolatile organics in models.
- Consistent evaluation of experimental data.
- Easy separation of source fractions for source attribution.
- Conceptualization and formal representation of aging effects attributed to secondary chemistry in both the vapor and condensed phases.
The practical implication of this is that in areas dominated by regional emissions, organic aerosol will not be represented by static “emissions factors” or “yields” but rather by emission of chemically active compounds whose products and thus whose partitioning evolves throughout long-range transport from the source to the receptor site.
Objective 1: Laboratory Measurements of Oxidation Kinetics of Organic Molecular Markers
Ongoing experiments have addressed oxidation of molecular markers in organic aerosols exposed to both O3 and OH. The experiments have been enhanced substantially by an Aerodyne aerosol mass spectrometer (AMS) and an Ionocon proton transfer mass spectrometer (PTRMS) obtained with a National Science Foundation Major Research Instrumentation grant. The following results have been obtained:
- We have developed a relative kinetics framework for interpretation of data on the heterogeneous oxidation of multicomponent organic aerosols. Common accommodation, diffusion, and deposition terms cancel in this formulation, and rate constants may be determined for many compounds simultaneously within an aerosol with a realistic composition. Finally, cross-phase relative rate constants with a gasphase reference compound and a condensed-phase target compound provide effective rate constants for use in atmospheric models.
- Smog chamber experiments this past year have focused on aging of model meat cooking emission aerosols. The experiments have focused on the oxidation of oleic acid, palmitoleic acid, and cholesterol, which are important molecular markers for meat cooking emissions. The oxidation of cholesterol and oleic acid strongly depends on mixture complexity. Furthermore, the experiments show that some saturated carboxylic acids, which do not react with ozone directly, decay in complex mixtures. By relating the decomposition of condensed-phase alkenoic acids to gasphase alkenes, we show that the reactive uptake of ozone by organic aerosol evolves as the aerosol is processed, decreasing by a factor of approximately 10 as the aerosol is aged.
- A method was developed for production of OH radicals is required for aerosol aging experiments. A new method was needed because of shortcomings of existing techniques in the context of heterogeneous oxidation. Existing methods typically require some combination of hard UV, high NOx, or extensive radical cycling. The new OH source is based on alkene ozonolysis. OH yield from ozonolysis is highly variable and alkene dependent; 2,3-dimethyl-2-butene (tetramethylethylene, TME) is the alkene of choice in our application because TME ozonolysis results in an OH yield near unity. The OH source can consistently produce radicals in the range of 5-7 x 106 molec cm-3 at high ozone concentrations. The main disadvantage to our method of OH production is that studying the chemistry of species that are reactive towards both OH radicals and ozone is impractical.
- Initial experiments also have been conducted to investigate the aging of motor vehicle emissions. Motor oil was exposed to typical summertime levels of OH in the smog chamber. Mass spectra collected using the AMS indicate significant oxidation of the motor oil. Future experiments will investigate the changes in molecular composition caused by this oxidation.
The practical implications of this work are that compounds used as molecular markers rapidly oxidize in simple systems, but that the oxidation rates strongly depend on mixture composition. Data are needed for more realistic systems to assess importance of aging on source apportionment analyses.
Objective 2: Application of Receptor Models To Apportion Ambient Organic Aerosol to Primary Sources in the Eastern United States
Individual organic compounds or molecular markers often are used in conjunction with the CMB model to apportion sources of primary organic aerosol. We have developed a methodology to visualize the organization of ambient molecular marker data, to compare the data to source profiles, and to better understand CMB solutions. It also can be used to assess chemical stability and aging. The core of the technique involves construction of plots of ratios of species concentrations (ratio-ratio plots) in which source profiles appear as points connected by linear mixing lines. The approach is illustrated using ambient measurements made in Pittsburgh, Pennsylvania, during a 1 year period of five, large polycyclic aromatic hydrocarbons (PAH) commonly used as molecular markers in CMB: benzo(b)fluoranthene, benzo(j)fluoranthene, and benzo(k)fluoranthene; benzo(e)pyrene; benzo[g,h,i]perylene; coronene; and indeno(1,2,3-cd)pyrene. In Pittsburgh, the ambient concentrations of these PAHs are correlated strongly, suggesting a single dominant source. These correlations underscore the significant source information contained in molecular marker concentrations. Ratio-ratio plots then are used to evaluate the potential contribution of gasoline exhaust, diesel exhaust, wood smoke, and coke production emissions to the ambient concentrations of the target PAHs. Coke production is the dominant source of these large PAHs in Pittsburgh. Ambient concentrations of these large PAHs provide little information on the gasoline-diesel split because of the strong influence of local emissions from coke production combined with potential photochemical decay of PAH in the regional air mass.
CMB analysis was performed to estimate the contribution of biomass smoke in Pittsburgh. Detailed comparisons are made between the ambient data and a large number of published source profiles. The fall and winter data are analyzed with fireplace and woodstove source profiles, whereas open burning profiles were used to analyze the spring and summer data. At the upper limit, biomass smoke is estimated to contribute 610 ± 130 ng C/m3 or 17 percent of the total OC in the fall, 225 ± 63 ng C m3 or 10 percent of the total OC in the winter, and 94 ± 29 ng C/m3 or 3 percent of the total OC in the summer. In the fall and winter, there is large day-to-day variability in the amount of OC apportioned to biomass smoke. The calculations face two major sources of uncertainty. First, the ambient data for levoglucosan, resin acids, and syringhaldehyde are relatively disorganized, showing that numerous sources with distinct source profiles contribute to marker concentrations. Second, the marker-to-OC ratio of available biomass smoke profiles is highly variable. This variability creates a bias of more than a factor of two in the amount of ambient OC apportioned to biomass smoke by different statistically acceptable CMB solutions. The marker-to-OC ratio is a critical parameter to consider when evaluating CMB solutions.
CMB analysis was performed to estimate the contribution of food cooking emissions to organic aerosol in Pittsburgh. We focus on five key molecular markers for cooking: cholesterol, palmitoleic acid, oleic acid, palmitic acid, and stearic acid. Ambient concentrations of subsets of these markers are correlated strongly, underscoring the source specificity of molecular markers. The organization of the ambient data allows one to infer potential source profiles to explain the data. The palmitoleic acid to oleic acid ratio in the ambient data, however, is roughly a factor of 10 greater than the ratio in essentially all published source profiles. The reason for this discrepancy is not clear, but it means that palmitoleic and oleic acid cannot be included in the CMB model at the same time. Based on comparison of source profiles with the ambient data, CMB analysis is performed using three different combinations of source profiles and molecular makers. The amount of OC apportioned to food cooking by these different scenarios varies by a factor of eight; this difference is caused by differences in the marker-to-OC ratios of the different source profile. The best CMB solution is based on three profiles: two meat cooking profiles with widely divergent marker to OC ratios and a seed oil cooking profile. This combination explains the significant variability in the ambient molecular data and apportions 320 ± 140 ng C m-3 or 10 percent of ambient OC on the study days to food cooking emissions.
We have used the CMB model and a large data set of ambient molecular marker concentrations to estimate the contribution of gasoline and diesel vehicles to ambient OC in Pittsburgh. To select source profiles for CMB analysis, we used a method of comparing the ambient ratios of marker species with published data for gasoline and diesel vehicle source emissions. The ambient winter data cluster on a hopanes/elemental carbon (EC) ratio-ratio plot, and a large number of source profile combinations can be used to explain the data. In contrast, the widely varying summer ambient ratios can be explained by a more limited number of source profile combinations. We present results for a number of different CMB scenarios, all of which perform equally well on the different statistical tests used to establish the quality of a CMB solution. After fitting the selected markers with the sources, CMB calculates the OC contributions from each source based on its marker-to-OC ratio. The hopanes-to-OC ratios in the different gasoline profiles vary more than two orders of magnitude, whereas EC/OC ratios for diesel profiles can differ by more than a factor of 10. This causes wide variations in the CMB predictions; these variations are tempered by averaging the different profiles using information on fleet composition. The vehicular contribution in the winter is bounded between 14 percent and 20 percent of the ambient OC (278 to 414 ng/m3). The variability in the diesel profiles, however, confounds a reasonable estimate of the gasoline-diesel split; one set of scenarios suggests gasoline vehicles dominate the vehicular OC contribution, whereas a second set predicts the opposite. The summer CMB solutions do not present a consistent picture given the seasonal shift in the ambient data relative to the source profiles.
PMF analysis was performed using 24 hours average concentrations of organic and inorganic species measured on 99 days between June 2001 and July 2002 at the Pittsburgh Air Quality Study main monitoring station. Sixty species were included in the model, including metals (as Ca, Ti, Fe, Zn), OC, EC, seven hopanes (vehicular markers), cholesterol (meat marker), levoglucosan (wood combustion marker), benzothiozole (tire wear marker), and a number of different alkanoic acids, alkenoic acids, PAHs, and biogenic oxidation products such as cis-pinionic acid. After examining many solutions, a total of seven factors were found to give the most useful results for this dataset. The high degree of source specificity enables one to compare factors directly with source profiles. Reasonable agreement was found between PMF results and literature profiles.
The practical implications of this work are that, even in a location strongly influenced by regional transport molecular markers, concentrations contain significant source information. The combination of photochemistry and variability in source profiles, however, create uncertainties for source apportionment analyses.
Objective 3: Evaluation of Emission Inventories Used by Chemical Transport Models
A source-resolved model has been developed to predict the contribution of eight different sources to primary organic aerosol concentrations. The model was applied to the Eastern United States during a 17 day pollution episode beginning on July 12, 2001. Primary organic matter (OM) and EC concentrations are tracked for eight different sources: gasoline vehicles, nonroad diesel vehicles, onroad diesel vehicles, biomass burning, wood burning, natural gas combustion, road dust, and all other sources. Individual emission inventories are developed from a modified version of the National Emissions Inventory (NEI) 1999 for each source and a three-dimensional chemical transport model (PMCAMx+) is used to predict the primary OM and EC concentrations from each source. The source-resolved model is simple to implement and faster than the existing source-oriented models.
The predictions of the source-resolved model are compared to measurements from the Speciation Trends Network (STN) and the Interagency Monitoring of Protected Visual Environments (IMPROVE) Network. Reasonable agreement is observed in the predicted total OM and the ambient data, but the model predicts EC concentrations three times higher than measurements from STN. Significant discrepancies exist if one compares the source-resolved predictions to the results of CMB models for Pittsburgh and Atlanta. Significant discrepancies exist between the source-resolved model predictions and the CMB model predictions for some of the sources. For EC, nonroad diesel, according to the emission inventory, is predicted to contribute more to EC in urban areas than onroad diesel. This overprediction suggests that the nonroad diesel emission inventory currently is too high. Although the nonroad diesel inventory should be reduced, the on-road diesel emission inventory also may need to be reduced. Natural gas, wood burning, and biomass burning are other sources that have emission inventory problems. The most striking problems are observed for natural gas; the primary OM emission inventory for natural gas should be reduced by at least 50 times the current value.
The practical consequence of this work will be improved emission inventories for chemical transport models used for State Implementation Plan development.
Objective 5: To Assess the Importance of Photochemical Aging on Primary Organic Aerosol Composition
An analysis of ambient molecular marker data found evidence that condensed-phase organic compounds are oxidized significantly in regional air masses and in locations affected by regional transport, especially during the summer. The core of the analysis involves examination of a large data set of ambient organic aerosol concentrations for removal of reactive compounds relative to less-reactive compounds. The approach allows visualization of both photochemistry and mixing of emissions from multiple sources to differentiate between the two phenomena. The focus is on hopanes and alkenoic acids, important markers for motor vehicle and cooking emissions. Ambient data from Pittsburgh and the southeastern United States contain evidence for significant photochemical oxidation of these compounds in the summertime. There is a strong seasonal pattern in the ratio of different hopanes to EC consistent with oxidation. In addition, measurements at rural sites indicate that hopanes are depleted severely in the regional air mass during the summer. Alkenoic acids also appear to be photochemically oxidized during the summertime; however, the oxidation rate appears to be much slower than that inferred from laboratory experiments. The significance of photochemistry is supported by rudimentary calculations, which indicate substantial oxidation by OH radicals and ozone on a time scale of a few days or so, comparable to time scales for regional transport. Oxidation is nonlinear; therefore, it represents a very substantial complication to linear source apportionment techniques such as the CMB model.
The practical implications of these findings are that oxidation likely alters source apportionment estimates using molecular markers in locations with significant regional transport. Assessing the extent of these problems will require the laboratory data being developed as part of Objective 1.
Future Activities:
During Year 2 of the project, we will focus on the following objectives:
- Conduct laboratory aging experiments using motor oil and meat cooking grease to measure oxidation kinetics of important molecular markers in complex surrogates.
- Implementation of the NEI 2002 emission inventory into PMCAMx
- Conduct source resolved simulations using PMCAMx and the NEI 2002 emission inventory for July 2001, October 2001, January 2002, and April 2002. Compare predictions with receptor modeling results and with measurements made by STN and IMPROVE networks as well as other monitoring sites.
- Complete analysis of Pittsburgh Supersite molecular marker data using PMF. Compare PMF results to CMB analysis.
Journal Articles on this Report : 11 Displayed | Download in RIS Format
| Other project views: | All 75 publications | 26 publications in selected types | All 26 journal articles |
|---|
| Type | Citation | ||
|---|---|---|---|
|
|
Donahue NM, Robinson AL, Huff Hartz KE, Sage AM, Weitkamp EA. Competitive oxidation in atmospheric aerosols: the case for relative kinetics. Geophysical Research Letters 2005;32:L16805. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit |
|
|
Donahue NM, Robinson AL, Stanier CO, Pandis SN. Coupled partitioning, dilution, and chemical aging of semivolatile organics. Environmental Science & Technology 2006;40(8):2635-2643. |
R832162 (2005) R832162 (2006) R832162 (Final) R831081 (2005) R831081 (Final) |
Exit Exit Exit |
|
|
Huff Hartz KE, Weitkamp EA, Sage AM, Donahue NM, Robinson AL. Laboratory measurements of the oxidation kinetics of organic aerosol mixtures using a relative rate constants approach. Journal of Geophysical Research-Atmospheres 2007;112(D4):D04204 (13 pp.). |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit |
|
|
Lambe AT, Zhang J, Sage AM, Donahue NM. Controlled OH radical production via ozone-alkene reactions for use in aerosol aging studies. Environmental Science & Technology 2007;41(7):2357-2363. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit Exit |
|
|
Lane TE, Pinder RW, Shrivastava M, Robinson AL, Pandis SN. Source contributions to primary organic aerosol:Comparison of the results of a source-resolved model and the chemical mass balance approach. Atmospheric Environment 2007;41(18):3758-3776. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit Exit |
|
|
Robinson AL, Subramanian R, Donahue NM, Bernardo-Bricke A, Rogge WF. Source apportionment of molecular markers and organic aerosol--1. Polycyclic aromatic hydrocarbons and methodology for data visualization. Environmental Science & Technology 2006;40(24):7803-7810. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit Exit |
|
|
Robinson AL, Subramanian R, Donahue NM, Bernardo-Bricker A, Rogge WF. Source apportionment of molecular markers and organic aerosol. 2. Biomass smoke. Environmental Science & Technology 2006;40(24):7811-7819. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit |
|
|
Robinson AL, Subramanian R, Donahue NM, Bernardo-Bricker A, Rogge WF. Source apportionment of molecular markers and organic aerosol. 3. Food cooking emissions. Environmental Science & Technology 2006;40(24):7820-7827. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit Exit |
|
|
Robinson AL, Donahue NM, Rogge WF. Photochemical oxidation and changes in molecular composition of organic aerosol in the regional context. Journal of Geophysical Research-Atmospheres 2006;111(D3):D03302 (15 pp.). |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit |
|
|
Shrivastava MK, Lipsky EM, Stanier CO, Robinson AL. Modeling semivolatile organic aerosol mass emissions from combustion systems. Environmental Science & Technology 2006;40(8):2671-2677. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit Exit |
|
|
Subramanian R, Donahue NM, Bernardo-Bricker A, Rogge WF, Robinson AL. Contribution of motor vehicle emissions to organic carbon and fine particle mass in Pittsburgh, Pennsylvania:effects of varying source profiles and seasonal trends in ambient marker concentrations. Atmospheric Environment 2006;40(40):8002-8019. |
R832162 (2005) R832162 (2006) R832162 (Final) |
Exit Exit Exit |
Supplemental Keywords:
ambient air, particulate matter, PM, atmospheric chemistry, model-based analysis, organic carbon, particulate organic carbon, source apportionment, air quality models, aerosol analyzers, particle size measurements,, RFA, Scientific Discipline, Air, Ecosystem Protection/Environmental Exposure & Risk, particulate matter, Air Quality, Environmental Chemistry, Monitoring/Modeling, Environmental Monitoring, Environmental Engineering, atmospheric dispersion models, particulate organic carbon, atmospheric measurements, model-based analysis, chemical characteristics, environmental measurement, source apportionment, emissions monitoring, air quality models, airborne particulate matter, air sampling, air quality model, analytical chemistry, speciation, particulate matter mass, modeling studies, chemical transport models, aerosol analyzers, atmospheric chemistry, chemical speciation sampling, real-time monitoring, particle size measurementRelevant Websites:
http://airquality.web.cmu.edu Exit
Progress and Final Reports:
Original AbstractThe perspectives, information and conclusions conveyed in research project abstracts, progress reports, final reports, journal abstracts and journal publications convey the viewpoints of the principal investigator and may not represent the views and policies of ORD and EPA. Conclusions drawn by the principal investigators have not been reviewed by the Agency.