Center on Airborne Organics

Annual Report 1995: Research Project Descriptions

SOURCE AND CONTROL

Combustion Chemistry of Polycyclic Compounds: J. B. Howard, Massachusetts Institute of Technology

Goal: The objective of this project is to contribute a predictive model of PAC formation suitable for use in source attribution studies and in the development of emission control strategies.

Rationale: Polycyclic aromatic compounds (PAC) are major contributors to air pollution from combustion sources. Basic understanding of the factors that govern the detailed chemical composition of the effluents from combustion systems is necessary for the identification of signatures for source attribution and the development of control strategies. The mechanistic and kinetic model being developed in this project will contribute to the basic understanding that is needed in the case of combustion-generated PAC.

Approach: A predictive model of PAC growth chemistry in flames is being developed using elementary reactions to describe the basic flame chemistry and PAH formation. Aerosol dynamics modeling is being done to describe all species (both PAC and soot) with masses above 400 amu. The aerosol model uses a sectional lumping method in which all material within the aerosol bin boundaries are considered as a single species, with the bin boundaries increasing by a constant factor (e.g., 2), allowing species with masses varying over several orders of magnitude to be described by inclusion of only tens of new species. The approach being used will combine elementary reaction modeling and aerosol dynamics modeling into a single calculation, by writing the aerosol dynamics equations in an equivalent elementary-reaction form. This combined modeling approach will allow for direct calculation of the effect of soot on concentrations of gas-phase species, including PAC. Previous methods for modeling soot formation have not provided such information.

Status: The theory for converting soot aerosol dynamics equations and equations expressing reactions of soot with gas-phase species (e.g., H, H2, C2H2, OH, O2, PAC) has been developed. The existing PAC formation mechanism has been expanded to allow prediction of PAC growth up to and including C32H14 (ovalene, 398 amu). The prototype soot aerosol dynamics equations have been written using 20 aerosol bins, each describing masses varying over a factor of two, so as to describe masses between 400 and 4.2*10^8 amu. Thermochemical properties have been generated using group additivity methods for the first thirty such soot bins (up to 4.3*10^11 amu), and have been converted into the NASA format used by the CHEMKIN subroutine package. All the conceptual difficulties in both describing the properties of soot as molecular properties and in the ability to incorporate the soot aerosol equations into elementary-reaction modeling have been addressed. The model also has the ability to account for the existence of multiple radical sites on the soot particle surface. The model also allows the mass balance to be conserved, while element balances are approximately conserved. The relevant features of the above theory have been coded as a FORTRAN computer program, which generates the reaction set for the soot model and writes a CHEMKIN-II format reaction file. The input file for the program allows variation of parameters such as soot particle size, which gas phase species (PAC and light species) react with the soot, and reaction parameters for all classes of reactions (collision efficiencies or Arrhenius parameters). Such a code facilitates use of the new modeling approach, since the soot model reaction set produced can be as large as 3000 to 10000 reactions. The soot reaction set is appended to the expanded PAC

formation model for combined modeling. Preliminary testing of the combined model demonstrated the ability to perform simulations using the CHEMKIN software, including features such as sensitivity and reaction path analysis, with large reaction sets for simplified flow fields. The data of Vaughn (1988) were used to test the ability to predict a critical equivalence ratio for soot formation in a well-stirred reactor. The range of predicted soot concentrations obtained by varying the input parameters bracketed the experimental concentrations. A decrease in PAC concentrations due to their reaction with soot was also predicted. The computational tools developed in this project are sufficiently powerful for use in addressing many facets of molecular weight growth processes in flames, since the entire range from H atom to soot is modeled. PAC formation rates are affected by many factors, all of which can be incorporated using the new model. This capability should allow further work on modeling PAC formation chemistry to be placed on a sound theoretical basis.


Deposit Effects Buildup on Engine Hydrocarbon Emissions: J. B. Heywood and S. Hochgreb, Massachusetts Institute of Technology

Goals: (1) To assess the effects of combustion chamber and intake-valve deposits on the hydrocarbon emissions from a modern production spark-ignition engine; (2) To develop and validate a model for the mechanism(s) by which combustion chamber deposits lead to additional HC emissions; (3) To determine whether the composition (and hence the reactivity) of the HC emissions is affected by deposit build-up; (4) To identify which HC compounds present in significant concentrations in commercial gasoline, lead to increased HC emissions due to deposit build-up; and (5) To study the effects of combustion chamber deposits on NOx emissions.

Rationale: Engine deposits (on intake valve and combustion chamber) increase HC emissions. Some recent data suggest that combustion chamber deposits increase NOx emissions. It is speculated that due to their low thermal conductivity, the combustion chamber deposits produce higher in-cylinder temperatures, and thus higher NOx emissions. To meet stringent future emission standards, the emissions due to deposits will have to be reduced. The first step towards that end is to better quantify these emissions and understand the mechanisms involved in their formation.

Approach: A four-cylinder, DOHC Saturn engine is subjected to a standardized deposit build-up cycle (1400 rpm @ 10% load for 6 minutes, and 220 rpm @ 20% load for 12 minutes). An additized fuel (which keeps the intake valves and ports clean and significantly increase the amount of combustion chamber deposits) is used for deposit accumulation and emissions testing, to isolate the effects of the combustion chamber deposits on the HC emissions. Starting with a ìcleanî engine, the HC emissions are continuously measured during the deposit build-up process. After the engine is cleaned, the HC emissions are again measured to ensure they return to the initial ìcleanî engine levels. An eddy-current probe is used to measure the deposit thickness at the center of the piston during deposit accumulation, and the deposit thickness profiles on the cylinder head and piston top at the end of the test. In addition, the cylinder head and piston top deposits are separately collected and weighed.

Status: Three deposit build-up tests (100, 50, and 25-hour tests) have already been conducted so far. In those tests, the HC emissions stabilized after about 25 hours. About 70% of the increase in HC emissions occurred in the first 10 hours. The HC emissions increased by an average of 14% (over the different tests and operating conditions) due to deposit build-up. The HC emissions returned to the ìcleanî engine baseline levels after cleaning the combustion chamber. Indolene was also used for HC emission measurements in the 100-hour test. The results show no significant difference in the HC emission levels between indolene and the deposit build-up fuel. HC emission were measured on the deposited and the ìcleanî engine (during the 50-hr test), using benzene, isooctane, and xylene. Benzene and isooctane emissions showed respectively, 45 and 28 % increase from the clean-engine emission levels, while xylene emissions showed no significant change. Fuel molecules get adsorbed in the deposit pores, during the intake, compression and combustion processes and desorbed during the exhaust process, thus escaping combustion. Adsorption isotherms of the deposit build-up fuel, isooctane, toluene, and the xylenes, are being measured on the deposits. The data will be used to evaluate some parameters in the model. The amount adsorbed depends on the deposit surface temperature throughout the engine cycle, which has been quantified from a simple, one-dimensional, heat transfer model. The adsorption model results are consistent with the emission measurements with benzene and isooctane. Absorption and desorption of fuel, through an oil layer on top of the deposit surface, is another possible mechanism through which deposits could lead to additional HC emissions. Early results from a simple, analytical model for fuel absorption/desorption from oil layers, developed at the MIT Sloan Lab, indicate that this is a possible mechanism for deposits to contribute to HC emissions.

Future work will concentrate on improving the adsorption model, to take into account the effect of diffusion of fuel through the boundary layer, and the deposit pores, and to be able to predict emission differences due to different fuels, operating conditions and deposit location (cylinder head or piston top deposits). In addition, another 25-hr deposit build-up test will be run. This test will be used to quantify the relative contribution of the piston and cylinder head deposits, to the HC emissions, by cleaning the combustion chamber in two stages. Also, the 25-hr test will be used to identify which single-component HCs (present in large concentrations in commercial gasoline), contribute to deposit-related HC emissions.


Soot Mass Growth in Stationary Combustion Systems: János M. Beér, Massachusetts Institute of Technology

Goal: To gain an improved understanding of the pathways leading to soot surface growth.

Rationale: The soot surface growth process which follows soot particle inception has been hypothesized to occur through two separate mechanisms: (i) acetylene (C2H2) addition and (ii) PAH coagulation. However, the relative contributions of these two mechanisms under various temperatures and equivalence ratios has not yet been determined. Since the soot nucleation process is believed to take place through PAH coalescence, the role of these compounds in the subsequent surface growth process cannot be underestimated. Their contribution, however, is expected to diminish with time since PAH concentrations in the soot growth zone tend to decrease due to soot scavenging. Acetylene concentrations, on the other hand, remain relatively constant throughout the soot growth process. This suggests that acetylene may play a more dominant role in the later stages of surface growth. As a result, a secondary objective of this study is the determination of the changes in the relative contributions of the two mechanisms with time.

Approach: In order to qualitatively determine the contributions from the two mechanisms, it would be necessary to produce soot under conditions where one mechanism is dominant and then compare the soot formation data against data collected from experiments where the other mechanism dominates. Since PAH play such a vital role in soot nucleation, it is unlikely that soot can be produced in an environment with relatively little PAH. Consequently, it would be very difficult to devise an experiment where the acetylene mechanism is clearly more dominant. On the other hand, it is possible to produce soot in an "acetylene lean" environment. This can be done by injecting an aromatic fuel into the post flame zone of fuel-rich but non-sooting flame. The acetylene concentration is a strong function of fuel equivalence ratio (f) and can be expected to be significantly lower in a non-sooting flame than in a sooting flame. Therefore, by injecting an aromatic into a non-sooting post-flame zone, soot will be produced almost exclusively by PAH coagulation. Comparison of the soot formation rates and gas species concentrations obtained from such an experiment against those obtained from a sooting flame with no injection will allow qualitative assessment of the relative contributions of the mechanisms. A more quantitative approach will also be followed. This consists of calculating the collision rates between soot particles and PAH molecules and between soot particles and C2H2 molecules and comparing these results with the observed soot formation rates. The soot collected from all the experiments will be subjected to microscopic analysis to provide this procedure with particle size distribution information. The experimental apparatus which is used to carry out these experiments is a Jet-stirred/Plugflow reactor. This burner comprises a well-stirred toroidal reactor in series with a plugflow reactor. PAH and soot are formed in the isothermal plugflow section where the absence of radial concentration gradients simplifies analysis.

Status: Two distinct experiments were devised. An "acetylene rich" experiment, where the JS/PFR was operated at a f of 2.1, and an "acetylene-lean" experiment, where a baseline f of 1.3 was chosen but where benzene was injected to promote soot formation. With a benzene injection rate of 4.2 g/min (where 15% of carbon fed to JS/PFR came from benzene), an average soot growth rate approximately equal to that of the f = 2.1 case was obtained. A comparison of the data from these two experiments led to the conclusions that (i) in the "acetylene-lean" case, the dominant soot growth agent was PAH while (ii) in the "acetylene-rich" case, both acetylene and PAH were important contributors to soot growth. Soot-PAH collision efficiencies for the "acetylene-lean" case were calculated to lie between 0.03 and 0.12, depending on residence time in the PFR. The effect of raising the benzene injection rate while maintaining a constant baseline was to increases the PFR soot and PAH concentrations. Soot-PAH collision efficiencies corresponding to two injection rates were found to differ by 20%, while the soot concentrations differed by a factor of four. This led to the conclusion that, in the benzene injection experiments, the soot number density is not a strong function of the soot loading.


Fundamental Study on High Temperature Chemistry of Oxygenated Hydrocarbons as Alternate Motor Fuels and Additives: Joseph W , Chemical Engineering and Chemistry, New Jersey Institute of Technology.

Goal: Experimental and modeling studies are being performed in order to understand and characterize reactions of oxygenated hydrocarbons (OHC's) such as alcohols and ethers important to gasoline octane blending. Alcohols, such as methanol and ethanol, are scheduled for widespread use as additives and alternative motor fuels. Ethers, such as methyl tert-butyl ether (MTBE), are already in use as anti-knock components in gasolines and oxygenate additives. An elementary reaction model based on fundamental thermochemical kinetic principles allow facile calculations on effects and trends in performance resulting from these oxygenates.

Rationale: Experimental data are needed for model development and validation. The fundamental based model, calibrated by experimental data, will facilitate calculations of trends for future experiment testing and preferred fuel blends to reduce undesirable emissions e.g., HC's, CO, etc. while maintaining or improving engine performance.

Approach: Experimental: Gas mixtures are reacted in a uniform, high temperature tubular flow reactor. Reactor effluent is analyzed for products as a function of temperature, residence time, fuel equivalence ratio and initial concentration. Analysis is performed with on-line gas chromatography (GC), flame ionization detection (FID), Fourier Transform Infrared (FTIR) and GC / Mass Spectrometry.

The reaction mechanisms (models) are based upon fundamental principles of thermochemical kinetics, transition state theory, chemical activation, quantum Rice-Ramsperger-Kassel theory, for k(E) modified strong collision treatment for fall-off and accurate thermodynamic properties.

Status: Oxidation and pyrolysis experiments on neat methanol and methane, on four fuel equivalence ratios of methanol with methane / oxygen have been completed, pressure range 1 atmosphere to 10 atmospheres. Experiments on mixtures MTBE in methane / oxygen are completed at one atmosphere and in progress in the high pressure reactor, fuel equivalence ratios of 0.7 to 1.5. A thermo-dynamic data base and an elementary reaction mechanism has been assembled for evaluation and comparison to experimental data. The mechanism is continuing to be modified and improved. Experiment and model both predict that small amounts and up to 50% methanol significantly enhance combustion of methane. A detailed kinetic reaction model for isobutane inclusive of isobutene, propene and acetone has been developed and compared with experimental data. A separate model for neopentane oxidation has been constructed, and compared to experimental data.

The atmospheric chemistry and combustion communities will be able to use the elementary reaction rate constants and models in photochemical smog and combustion modeling. Applications relate to utility of oxygenated hydrocarbons as dedicated fuels, blend components, or additives; and aspects of their environmental impact regarding urban air pollutants such as ozone and peroxyacetyl nitrates (PAN's). The usefulness of OHC's will depend on their engine performance characteristics; for example, it is known that laminar burning velocities of iso-octane / methanol blends in air are lower than the velocity of either component alone. Data in this study has also shown some methane / methanol blends are faster than either component alone.


Simultaneous Removal of Soot and NOx from the Exhaust of Diesel Powered Vehicles Using a Rotating Fluidized Bed Filter: H. Shaw and R. Pfeffer, New Jersey Institute of Technology

Goal: The objective of this project is to develop a new approach for controlling the most problematic pollutants NOx and soot emissions from stationary and mobile Diesel engines. It is intended to specifically demonstrate the use of a rotating fluidized bed reactor (RFBR) containing an attrition resistant high surface area powder to promote the reaction of soot with NOx. One or more catalysts will also be evaluated for this application since there are indications in the literature that, e.g., copper exchanged ZSM-5 Zeolite promotes the, reaction of these pollutants. In addition, research will be conducted on the mechanism and competitive kinetics of soot reactions with the two oxidants, NO and O2. To the extent that catalysts are needed, various approaches to overcome the water and sulfur poisoning problems that have been reported in the literature will be evaluated. A mathematical model of the system will also be developed to facilitate design and application of the system to other types of Diesel engines.

Rationale: This project is designed to find ways to overcome some of the shortcomings of current technology consisting of small passage zeolite catalysts that are used to oxidize Diesel soot, or Pt based oxidation of NO to NO2 which is then used to oxidize soot collected on a filter. In the first case, the zeolite acts as a filter for soot capture and plugs when the catalytic surface becomes ineffective due to low temperature operating conditions of the Diesel engine (i.e., city driving). In the second case, much of the NO2 is reduced to NO rather than N2 emitting excessive quantities of NOx. The research is directed at investigating the use of a rotating fluidized bed reactor to filter soot which will not be subjected to rapid plugging, and hence, avoids large changes in pressure drop. The captured soot will be oxidized in situ by NOx to produce non hazardous CO2 and N2. Consequently, this system is self cleaning. The small size of the RFBR may allow convenient replacement of the filtration solids or catalyst charge over the long life of Diesel engines. Furthermore, the RFBR rate of rotation couples very well to that of the Diesel engine. Thus, at low load the bed will rotate relatively slowly and primarily act as a soot filter, and at high load the bed will rotate rapidly and promote the soot-NOx reaction. The envisioned RFBR has a relatively small footprint and is expected to fit into even an automobile engine compartment.

Approach: The research consists of both experiments and computer modeling. The experimental arrangement consists of two fluidized beds connected to a dynamometer driven automotive Diesel engine. One of the fluidized beds is a conventional vertical 7.5 cm by 1.0 m bed and the other is a 13 cm diameter by 15 cm long horizontal rotating bed. The effluent stream from the engine flows to one of the two fluidized beds. The beds are charged with various powdered media which will both filter soot and provide surface to promote the soot-NOx reaction. Continuous flue gas measuring equipment consisting of non-dispersive infrared CO and CO2 analyzers, electrochemical oxygen monitor, chemiluminescent NOx analyzer, and flame ionization unburned hydrocarbons analyzer are used to determine the flue gas composition. A gas chromatograph is used as needed to further identify the unburned hydrocarbons. Soot is characterized by monitoring particle size distributions. All analytical measurements are made before and after the fluidized beds. In addition, flow rate and pressure and temperature sensors are used to characterize the fluidized beds. All data will be automatically collected on a data acquisition and control unit. Under idle conditions, the soot stream is primarily filtered in one of the beds. At full power, soot is oxidized by NOx and O2. Catalyst deactivation will be monitored using the very sensitive propene-NOx reaction after each 20 hours of operation.

The modeling effort consists of studies of a single spherical particle on which carbon (soot) builds up at low temperatures and reacts at elevated temperature due to the reaction of carbon with the gas phase oxidants. This particle is considered independent of neighboring particles and represents a fluidized bed. A set of global kinetic rate constants will be obtained from the literature or determined in the laboratory. Soot is expected to catalytically promote its own oxidation. The combination of catalytic oxidation and poisoning on a particle with a moving external surface area has never been modeled and will require a new innovative approach.

Status: The primary objective of this year's effort was to design, fabricate, assemble, and calibrate the experimental equipment. This effort has been accomplished. The Diesel engine has been installed and connected to the dynamometer. Mobil Oil Co. has provided us a sample of 100% Cu-exchanged ZSM-5 with a silica/alumina binder designed to resist attrition in the fluidized beds. Mobil has also provided us with samples of the binder and ordinary ZSM-5 with the silica/alumina binder. All analytical monitoring equipment has been calibrated and is being installed near the Diesel engine. Calibration curves have been developed for flue gas temperature and flow rate as a function of engine load. The efficiency of the RFBR as a soot filter is currently being determined. On a small scale, laboratory experiments are being conducted using a Thermal Gravimetric Analyzer (TGA) and a 2.5 cm diameter by 0.4 m quartz reactor to obtain insight into the oxidation mechanisms. The quartz flow reactor is also being used to obtain carbon and nitrogen balances. Two masters theses are expected to be completed in 1995 on the model and the small scale laboratory research experiment.

It is anticipated that automobile and truck manufacturers, catalyst companies, environmental organizations, regulators, and the general public will be very interested in this novel approach for minimizing Diesel soot and NOx emissions. An industrial advisory board has been formed which includes two petroleum companies, a catalyst manufacturer, and two automobile/truck manufacturers. Discussions are taking place with another catalyst manufacturer and two automobile/truck manufacturers. Some of these companies are interested in developing the technology after completion of the basic research.

TRANSPORT AND TRANSFORMATION
Pathway Analysis and Elementary Reaction Mechanism for Atmospheric Reactions of Aromatics - Benzene and Toluene: Joseph W. Bozzelli and Tsan Lay, Department of Chemical Engineering and Chemistry, and Environmental Science, New Jersey Institute of Technology

Goal: Develop a model from reaction pathways, elementary reaction parameters and thermodynamic properties that accurately describes the photochemical oxidation of aromatic hydrocarbons (benzene, toluene, etc.) under atmospheric conditions. Validate the reaction mechanism against literature experimental data.

Rationale: The present study is providing theoretical estimates of the mechanisms and kinetics for the oxidation of aromatics which are the first steps in the gas-to-particle conversion experiments being carried out by Seinfeld and Flagan.

Status: Detailed thermodynamic kinetic analysis of reaction pathways and rate constants have been studied in detail for reactions of the major radical intermediates of benzene photochemical oxidation (phenyl, phenoxy, benzene-OH adduct, benzene-OH-O2 adduct and benzene-OH-O adduct) with O2, NO, and NO2. An elementary reaction mechanism for benzene photo-oxidation which includes microscopic reversibility for all reactions has been developed. Thermodynamic properties (enthalpies, entropies and heat capacities) are required in this simulation model and have been determined for all species in the reaction mechanism. The formation yields of products predicted by this model simulation are: phenol, 14.4% and nitrobenzene 4.2 %, compared to literature experimental values: 23.6% and 3~5%, respectively. A major uncertainty exists in barrier of the reaction of benzene-OH adduct with NO2 leading to nitrobenzene, to which the formation yield of nitrobenzene is very sensitive. This will be further studied using high level molecular orbital calculations. The reactions of phenyl and phenoxy radicals are not important to the overall benzene photochemical oxidation system. The development of reaction mechanism for toluene oxidation is well under way.


Atmospheric Transformation of Volatile Organic Compounds: Gas-Phase Photooxidation and Gas-to-Particle Conversion: J. H. Seinfeld, R. C. Flagan, California Institute of Technology

Goal: The objective of this project is to gain a better fundamental understanding of the atmospheric oxidation of volatile organic compounds (VOCs) important in urban and regional air quality. Specific aims are to determine the gas-phase mechanisms of reaction of important VOCs with the hydroxyl radical, the atmosphere's most ubiquitous oxidizing species, and to elucidate the mechanisms of formation of organic aerosols from the atmospheric oxidation of VOCs.

Rationale: Gas-to-particle conversion is a ubiquitous process in the atmosphere, determining the size and composition of particles from the polluted urban atmosphere to the remote marine boundary layer. Understanding the detailed chemistry and physics of atmospheric gas-to-particle conversion will allow us to predict the effects of primary gaseous and particulate emissions changes on airborne particulate matter, in the urban and regional setting, and the effects of sulfur and other species on the generation of cloud condensation nuclei in the remote atmosphere. A principal goal of the research program is the development of comprehensive air quality models based on the most complete description of atmospheric chemistry and physics. These models are forerunners of those that will eventually be used in the regulatory process. This research is aimed at developing the organic portion of advanced gas-aerosol models, and to advance the current state of understanding of molecular processes. The component of the proposed research on gas-phase photooxidation chemistry has the goals of adding to the body of kinetic and mechanistic data for atmospheric organics, with particular emphasis on those VOCs that are potential aerosol precursors.

Approach: The integrated research program to determine the mechanisms of photooxidation and secondary organic aerosol formation in the atmosphere for a number of important anthropogenic and biogenic hydrocarbons is carried out in both indoor and outdoor reactors. Experiments in the indoor reactor are used to probe chemical mechanisms. The large outdoor smog chamber is employed to study the integrated gas-phase and gas-to-particle conversion dynamics.

Status: During this past year over 75 smog chamber experiments involving aromatic and biogenic hydrocarbons were carried out. A definitive data base has been assembled on the aerosol yields from hydrocarbons.


Direct Treatment of Uncertainties in Mathematical Models of the Transport and Fate of Airborne Organics: Gregory J. McRae, Department of Chemical Engineering, Massachusetts Institute of Technology

Goal: The goal of this research is to develop a new mathematical and computational framework for the systematic sensitivity and uncertainty analysis of the complex transport and transformation processes that control the concentration dynamics of airborne organics. A particular focus is the implementation of numerical procedures that are much more computationally efficient than even the best Monte Carlo sampling strategies. Once the tools have been developed, the approach is to carry out a detailed investigation of the photochemical oxidation mechanics for airborne organics

Rationale: One of the consequences of using models to describe the formation and transport of photochemical air pollution is that some approximations are involved. In addition there are also measurement errors in the data used to develop inputs and kinetic parameters for the reaction mechanisms. The key issue is not that uncertainties are involved, they will always be present, but to identify which of the inputs contributes most to the uncertainty in predictions. The present set of tools available to the research community are simply computationally intractable for the complex reaction schemes needed to describe the photochemistry of airborne organics.

Approach: In this research a new approach termed the Deterministically Equivalent Modeling Method (DEMM) has been developed. In this technique, uncertain parameters are treated as random variables that are in turn approximated using orthogonal basis function expansions in the probability space. The expansion is based on the use of standard Hermite polynomials and independent Gaussian distributed random variables for each independent coordinate in probability space. A complete description of the method is contained in Tatang (1994).

Status: Set out below are several highlights of the project work to date:

Development of a computational algorithm that is approximately three orders of magnitude faster than the best Monte Carlo sampling methods.

Emphasis on the practicality of reusing deterministic numerical solvers for solving corresponding stochastic models. In particular developing and applying a prototype language as an environment for carrying out uncertainty analysis.

Development of the collocation method in the DEMM for "black box" type models.

Application of DEMM to aspects of Carter's SAPRC mechanism to investigate uncertainties of photolysis rates, initial conditions, and product coefficients.

Identification of parameters that contribute to uncertainties of ozone predictions. These include the photolysis rates of NO2 and formaldehyde, initial conditions of NOx and VOC, and stoichiometric coefficients of radicals from lumped product species.

One major result from this project is the implementation of a prototype language for the DEMM. Such an implementation serves not only to demonstrate the ease of conducting uncertainty analysis for a wide range of chemical and environmental engineering problems. DEMM has been applied to a complex photochemical mechanism for organic compounds. The next phase of the research is to apply the DEMM technique to a detailed analysis of the oxidation of airborne aromatic compounds, and to investigate the uncertainties due to the approximations and assumptions of different models.


MONITORING AND SOURCE APPORTIONMENT

Methodology for Sampling and Analysis of Polar Organics: Somenath Mitra and Barbara Kebbekus, New Jersey Institute of Technology


Goal: The objective of this research is to develop sampling and analysis techniques for polar volatile organic compounds (VOCs) in air. Of particular interest are sampling system using membranes and, real-time monitoring of VOCs.

Rationale: Polar and oxygenated VOCs are an important class of pollutants, but methods for their sampling and analysis are not well established. Some of the problems encountered in the sampling and analysis of these compounds are being addressed in this project. Gases such as H2O and CO2 are always present in air especially in most emission sources and interfere in air analysis. Another general problem in the area of VOCs analysis is that techniques for continuous monitoring at trace levels is not available.

Approach: An air sampling device is being developed where the air is passed through a hollow fiber membrane. The VOCs selectively permeate through the membrane, and are either analyzed on-line using the microtrap, or sampled into a canister. The canister sampling has produced some excellent results, and it was possible to demonstrate that the VOCs can actually be concentrated in the canister. This significantly lowers the detection limits of canister analysis. Canister sampling through a membrane has another advantage. The membrane has high selectivity towards the VOCs with respect to water. Consequently, it serves as a water management device. Most canister analysis is limited by the moisture content of the air samples, as moisture freezes cryogenic traps and interferes in GC/MS analysis.

For continuous GC monitoring of VOCs at trace levels a microtrap as an concentration cum injection device is being used. Much progress has been accomplished in this area. Present work involves working with microtraps containing layers of adsorbents to prevent breakthrough of volatile compounds.

Status: Various parameters that can effect enrichment factors during membrane sampling using the canister are being studied. Different membranes are being evaluated, that can not only enhance enrichment factor but also provide higher selectivity with respect to moisture. The use of the membrane sampler and the microtrap for continuous monitoring of VOCs has also been combined. Presently under study are the trapping and desorption characteristics of different adsorbents for different VOCs. A paper describing this application is being prepared for publication.

The Character of Primary Organic Aerosol Emissions and their Relation to Atmospheric Samples: Glen R. Cass, California Institute of Technology

Goal: The purpose of this research project is to create an expanded understanding of the chemical nature of organic particulate matter that is emitted form air pollution sources and how those effluents combine to produce the mixture of organic compounds observed in the atmosphere. In this work, particular attention is paid to determining source/ receptor relationships for those organic compounds in the airborne particle complex that are mutagenic.

Rationale: The airborne particle complex contains literally thousands of different organic compounds, many of which are present in trace amounts, and all of which are mixed in the atmosphere with a much larger amount of probably harmless but bulky material from other sources (e.g., airborne soil dust, sea salt). It is entirely likely that by concentrating attention on particle mass concentration, or even on the most plentiful or easily measured organic compounds, that those chemical components of the airborne particle complex that are actually capable of inducing the sort of biological changes that could affect health will be missed entirely. The purpose of the present study is to establish an understanding of source/receptor relationships for organic chemical compounds or compound classes in atmospheric particulate matter, with particular emphasis on those compounds that can produce measurable biological changes and those compounds that act as diagnostic markers for the presence of the effluent from particular sources within atmospheric samples.

Approach: A program of atmospheric sampling first will be undertaken in southern California to acquire large quantities of atmospheric fine (dp<2µm) particulate matter at sites representative of primary industrial emissions, primary motor vehicle emissions, downwind photochemical smog receptor sites, and upwind background air quality. Samples taken by comparable methods from the 15 most important primary sources of organic aerosol emissions in southern California also will be examined. These source and ambient samples will be processed by high performance liquid chromatography (HPLC) and by gas chromatography/mass spectrometry (GCMS) techniques such that the chemical composition of the source samples can be compared to the ambient samples. Mutagenicity-directed chemical analysis of both source and ambient samples will be used to try to identify organic compounds or compound classes in the source and ambient samples that are of particular interest because of their ability to produce measurable biological changes in those assays. Organic compounds will be sought that are characteristic of some sources but not others and thus can be used to trace the presence of the effluent from particular sources in an ambient sample. Then atmospheric modeling techniques will be used to make comparisons between the aerosol composition and mutagenicity expected if the source effluents were transported without chemical reaction versus the observed composition and mutagenicity of the actual atmospheric aerosol. In a parallel effort, the mechanisms by which tracer compounds emitted from particular sources react in the atmosphere will be studied to determine the time and distance scales over which they may be used to track the emissions from particular sources.

Status: The one-year atmospheric sampling experiment designed to collect large enough ambient aerosol samples for mutagenicity-directed chemical analysis was completed on schedule at the end of December, 1993. Samples were collected at Long Beach, Central Los Angeles, Azusa, and Rubidoux (Riverside), CA, at 6-day intervals using high volume dichotomous samplers. Periodic samples also were taken at San Nicolas Island upwind of Los Angeles to measure organic aerosol levels in the air entering the urban area. The ambient samples from the air monitoring locations within the South Coast Air Basin were combined to form 6 bimonthly composites at each site. The much smaller amount of organic carbon collected at the background site on San Nicolas Island was combined into a single composite sample in order to obtain sufficient material for the bioassay. Extraction and bioassay evaluation of these samples began in February, 1994, and was completed in October, 1994. During the past year the first round of data analysis from this experiment has been completed, and a journal article manuscript has been submitted for publication (Hannigan, et. al., 1995a). In that paper the spatial and temporal distribution of atmospheric organic carbon and elemental carbon particles is examined, as well as trends in the bacterial mutagenicity of that aerosol in the southern California atmosphere. The highest levels of mutagenicity per unit organic compound mass subjected to the bioassays and per unit volume of air sampled are found in the vicinity of the major primary emissions sources in Long Beach and in downtown Los Angeles rather than at the downwind areas that are traditionally perceived as receptor sites for photochemical smog transformation products. Mutagenicity aerosol levels are highest in the winter months at Los Angeles and at Long Beach, which is the pattern expected if primary emissions of mutagens directly from sources is controlling atmospheric mutagen levels. This also suggests that if secondary formation of mutagens by atmospheric chemical reactions is at all important, then those transformations must occur in winter months as well as during the more photochemically active summer months.

In order to begin to identify the compound classes that are most responsible for the observed mutagenicity of the source samples, extracts from source samples that tested positive during an initial pilot study were subdivided into four fractions of varying polarity on a gradient elution HPLC. The first fraction created contained normal PAH. Fraction 2 contained the nitro-PAH. Fraction 3 contained dinitro-PAH, HN-PAH, and polycyclic aromatic ketones (PAK). Fraction 4 includes the alcohols and organic acids. In all source samples tested to date, fraction 3 contains the most potent bacterial mutagens. Fraction 2 also shows significant bacterial mutagenicity in the heavy-duty diesel truck exhaust sample and in the natural gas-fired home appliance emissions. The ambient sample composites were separated into four fractions of varying polarity using a gradient elution HPLC as was described above for the source samples Fraction 3 containing the moderately polar compounds such as dinitro-PAH, HN-PAH and polycyclic aromatic ketones displayed the greatest bacterial mutagenicity, as was the case for the source samples. A second journal article manuscript is nearing completion at present (Hannigan, et. al., 1995b) that describes the distributions of bacterial mutagenicity of the southern California aerosol and aerosol sources between the various fractions that can be associated with organic compound classes of varying polarity.

The ultimate aim is to isolate and identify the most mutagenic compounds or groups of compounds in the atmospheric samples using successive fractionation on the HPLC and bioassays of each fraction to track down the most active compounds. A search for the same compounds in the corresponding source sample also will be conducted. This requires extensive chemical analysis of the atmospheric aerosol and source samples by GCMS. That chemical analysis is underway at present. Once the individual organic compound concentrations have been determined both for mutagenic compounds and for compounds that act as tracers for the sources, then source/receptor modeling methods will be used to try to relate the source emissions to the ambient samples.


Markers for Emissions from Combustion Sources: Adel F. Sarofim, J. B. Vander Sande, Massachusetts Institute of Technology

Goal: Soot particles generated and emitted by combustion processes have a microstructure and trace element composition which can be observed with high resolution transmission electron microscopes. Since the microstructure and chemical composition are functions of fuel type and combustion conditions they can be useful markers of the origin of soots. The goal in this project is to develop methods for quantifying the soot structure and elemental composition in order to determine their potential use as signatures for the source of particulate carbon in ambient air.

Rationale: Combustors are a major source of airborne organics which contribute, directly or indirectly, to health hazards and to visibility degradation. Fuel formulation and combustor design are expected to be steadily changing, and thus is it important to have reliable means for identifying the true sources of the organic contaminants in ambient air, so that effective corrective engineering and policy strategies can be formulated and implemented in a timely manner. To this end, the objective of the research is to identity, for combustion sources, chemical and structural characteristics of soots that are sufficiently unique and stable to be reliable signatures for the fuels and combustors responsible for their emission.

Approach: Work involves: (i) development of methods for obtaining quantitative measures of the soot microstructure and composition; (ii) determination of the relation of soot microstructure and composition to combustion conditions and fuel type through the use of well defined experiments, and (iii) development of a library of soot structures for use in emissions source attribution studies.

Status: To date, high resolution transmission electron microscopy and image analysis software (SEMPER™) has been used to quantify microstructural features in several soots including samples from mine and test diesel engines and well defined carbon blacks and graphites that have been produced from different model fuels and subjected to different oxidation histories. Soot structural features of interest for "fingerprinting" fuel type, oxidation history, and extent of thermal treatment, include interlattice spacing in graphitic layers, as well as the number, length, and area of lattice fringes. A method of presenting the data which shows the distinctive features of their origin is the population of the interlattice spacing or the lattice length. The interlattice spacing shows distinct mulimodal distributions with the different peaks providing insights on the curvature of the different lattice planes. The order of the structures is found to increase systematically with partial oxidation and heating, as reflected by the decrease in mean interlattice spacing, the decrease in the standard deviation of the lattice spacing, and the increase in lattice length. The position of the peaks of the multimodal distribution is, however, found to be insensitive of the degree of partial oxidation suggesting that these are more basic properties of the material. A complementary finding has been that the trace elements associated with the soots, not surprisingly, reflect the composition of the parent fuels. The distribution of the elements is, however, not always uniform, showing a tendency for iron and vanadium to concentrate in selected primary particles of the soot aggregates. This is observed for both ambient soot samples and soot particles taken from fuel oil combustors. The results to date have shown that the detailed structural and chemical characterization of soots provides ample material for use in establishing signatures. It also provides information on the mechanisms of formation of the particles.