Objectives

The research is designed to provide experimental time-resolved data on soot particle size and structure during soot growth and burnout under well defined conditions pertinent to practical flames. The work builds on previous time-resolved measurements of soot mass concentration and gas composition, including PAH, and currently available modeling capability for handling the formation and interaction of PAH and soot. The specific objectives are: (1) to extend previous jet-stirred/plug-flow reactor (JSR/PFR) data to include measurements of particle size and structure at different residence times corresponding to different particle formation and burnout stages including any oxidative fragmentation; (2) to extend the present PAH-soot model to include oxidative enlargement of pores and fragmentation of soot particles; (3) to critically test the model against data so as to evaluate and improve the mechanistic understanding and accuracy of model predictions of soot particle and PAH emissions from combustion; and (4) to use the model to identify potential combustor design features and operation conditions for improved control of soot and PAH emissions.

Background

Soot formation in flames is a major source of the fine airborne particles of current health concern. The mechanism of soot formation involves particle nucleation or inception, mass addition to the particles by reaction with gas-phase molecules, coagulation of particles through particle-particle sticking collisions, mass removal from particles by pyrolytic elimination of functional groups accompanied by dehydrogenation and structural rearrangement or carbonization of the condensed material, and oxidation. Although the details of soot formation and oxidation are not completely understood, there is reason to believe that the fine soot particles of health concern may originate in combustors in different ways. The soot formation and oxidation processes may be terminated at different stages through dilution, thermal quenching or other mechanisms occurring in combustion equipment. The resulting particles found in the exhaust may have characteristics of the formation stage, the destruction state, or both, depending on the combustor design and operating conditions. For example, larger numbers of smaller particles could be produced by quenching the formation process at an early stage of formation before particle number is extensively reduced by coagulation, or by incomplete oxidation of particles under conditions where pore development leads to particle fragmentation. The detailed characteristics of environmental soot sources are poorly understood, and basic guidance is needed for the development and optimization of control strategies. To help provide the understanding and guidance that is needed, the present research is designed to provide improved understanding of the origin and characteristics of fine soot particles produced in combustion.

Method of Approach and Facilities

This work focuses on improving the basic information on the origin and characteristics of ultrafine particles in combustion. Time-resolved measurements of soot particle size and concentration are being performed for different flame conditions, and the results are used to study the mechanisms of soot particle generation and growth in flames. A Scanning Mobility Particle Size Analyzer (SMPS) and transmission electron microscopy (TEM) are used to determine soot particle size, morphology, and structure in the ultrafine region. A primary dilution sampling probe was designed for in-flame extraction of soot. A program was written in MATLAB to assist in particle sizing of the TEM images, and the results are compared with those obtained with the SMPS. TEM images also provide valuable information on soot structure and agglomerate formation, which is needed for complete characterization of the soot.

Summary of Progress and Accomplishments

The JSR/PFR was used to obtain samples over a range of temperatures and oxygen concentrations. Ethylene was burned premixed with different air/nitrogen mixtures at atmospheric pressure and equivalence ratio 2.2 in the JSR. In some runs oxygen was injected and rapidly mixed into the JSR effluent in the transition section leading to the PFR and larger nitrogen feed rates were used in the JSR so as to increase the oxygen concentration while holding temperature approximately constant in the PFR. Temperatures of 1520 K and 1620 K in the PFR were studied. Soot samples were taken by thermophoretic deposition on rapidly inserted and removed (exposure times of 25-500 ms) ambient temperature electron microscope grids at different distances or times along the axis of the PFR. Temperature and oxygen concentrations in the PFR were changed independently by the use of different feed rates of nitrogen into the JSR and oxygen into the PFR. Gas samples were taken with a water-cooled probe and analyzed for stable species by gas chromatography. Soot number density and particle size distribution at different axial positions were determined by transmission electron microscopy and the internal nanoscale structure of the soot was studied using high resolution transmission electron microscopy (HRTEM). Both the particle size and internal structure measurements were performed using in-house software.

Soot number density and particle size distribution at different axial positions in the PFR were also measured using a water-cooled high inert-gas dilution rate probe close coupled to a scanning mobility particle sizer (SMPS). The SMPS was coupled with a new sampling probe to analyze ultrafine soot particles (1000 nm down to 5 nm). The sampling probe was specially designed to withstand flame temperatures and quickly quench the extracted soot and combustion products. Rapid quenching prevents chemical reactions and suppresses aerosol dynamics by cooling and dilution. The sampling probe also prevents thermophoretic losses of soot by radial injection of the dilution gas inward so as to confine the sample to the center of the probe and away from the cool walls. A high dilution ratio of 60 to 100 is used to prevent soot particles from agglomerating and also to reduce the concentration of soot to within the limits of the SMPS. The dilution ratio is an important factor when comparing in-flame soot concentrations. The dilution ratio is measured by adding a tracer to the combustion gases and measuring its concentration in the diluted sample using gas chromatography/mass spectrometry. In this way, the in-flame concentrations of different samples can be compared even when the dilution ratio is varied.

The electrical mobility of a particle depends on the drag force between the particle and the surrounding gas. An irregular shaped particle or an agglomerate experiences a higher drag force that a spherical particle with the same volume. Therefore, SMPS gives larger particle sizes than the TEM method if the particles are highly agglomerated or irregularly shaped. Mostly spherical and non-agglomerated particles were observed in this study. The drag force on the particle also depends on the viscosity and hence the composition of the surrounding gas. Because of the high dilution ratios used in this work the surrounding gas is mainly nitrogen in all cases, and the influence of the dilution ratio on the measured particle diameter should be negligible. Also the sampling conditions were such that particle losses due to coagulation and inertial losses in sampling lines should be negligible and measured size distributions should be independent of the dilution ratio. However, the data showed that mean particle diameters measured with the SMPS decreased with increasing dilution ratio and with increasing SMPS inlet flowrates. The cause of this unexpected behavior has not yet been identified, but the problem seems to reside in the SMPS instrument itself, as was confirmed for polydisperse particles by the manufacturer of the instrument. Reproducibility of results is only good if the dilution ratio and the inlet flowrate are constant for all samples taken. Medium dilution ratios (~100) and low inlet flowrates (~ 1 liter/min) give the best reproducibility, but then measured particles sizes are about twice as large as TEM measurements.

The behavior described above also limits the accuracy of soot mass concentrations measured with the SMPS method. In this work, a gravimetric method based on sample collection by filtration and weighing with an analytical balance was used to calibrate the SMPS for soot mass concentration measurements. Without such calibration by a reliable method it does not appear possible to accurately measure absolute values of soot mass concentration or mean particle diameters with the SMPS system with its present performance. Nevertheless, it is possible to measure relative changes of soot particle diameter and soot mass concentration with PFR residence time if the dilution ratio, sample gas flowrate, and inlet flowrate are held constant. Using the electron microscopy method approximately 500 soot particles on each of several micrographs of thermophoretic samples from each of 10 sampling ports were sized for each set of combustion conditions studied. The results give time resolved data on the evolution of soot particle size distributions in the PFR for different temperatures and equivalence ratios. A lognormal size distribution with a standard deviation of about 1.6 was found to fit experimental distributions well. Although this method requires considerably more time than the SMPS method, it offers the strong advantage of giving accurate results.

From the time resolved measurements of soot mass concentration apparent rates of soot formation were determined. The apparent rate of soot formation oscillates with time in the PFR, in agreement with our previous studies. Oxygen injection increases soot mass concentration relative to the 1520 K baseline case but decreases soot mass concentration relative to the 1620 K baseline case. A net decrease of soot mass is observed in the higher temperature oxidation case where it appears that the enhancement of soot oxidation exceeds that of soot formation. At the lower temperature the rate of soot formation exceeds the rate of soot oxidation, and even more so when the JSR temperature is lowered so as not to exceed the baseline temperature in the PFR with oxygen injection.

A derivative analysis of the apparent soot formation rate was performed assuming a first order dependence of a surface growth reactant on this rate. The observed rate of soot formation can be explained if PAH are assumed to be the major surface growth reactant. In the higher temperature case, variations of PAH mass concentration correlate with the observed oscillations of the soot formation rate, hinting to a first order dependence of the soot formation rate on PAH concentration. An acetylene-addition mechanism alone cannot predict the experimental soot mass concentration profiles. Acetylene probably accounts for PAH mass growth while PAH are the intermediate species on the way from acetylene to soot.

The above interpretation of the data was tested in a kinetics modeling study using the technique of data incorporation in which chosen portions of the chemistry are replaced with functions of measured concentrations. This technique was used to isolate the process of PAH molecular weight growth and soot nucleation through the use of a discrete sectional model, and rate coefficients for hydrogen-atom abstraction, acetylene addition, and PAH-radical addition to PAH were obtained by comparisons to data from the 1620 K condition described above and, from earlier work, the same condition but with napthalene injection into the PFR. The data incorporation technique was then used to expand the discrete sectional model to include sections describing soot, and the experimental soot size distribution data described above was used with previously available PFR data to obtain values for rate coefficients of PAH addition to soot and coagulation of soot particles. Five PFR conditions were used to develop the soot formation model in these calculations, and the dominant mechanisms of soot formation present under these conditions appear to be present in the model. Quantitative agreement is obtained to all of the available data, including simultaneous agreement of soot mass and particle size, without significant deviation in the rate coefficients required to obtain agreement. Calculations were performed using both PAH and acetylene as the dominant soot surface growth reactant. It was found that PAH had far more consistent rate coefficient values (constant to within a factor of 4) than acetylene (constant to within a factor of 59) in describing the data for all of the conditions.

The above five sets of conditions in the PFR, an additional three for the PFR, and three for premixed one-dimensional flames of acetylene, ethylene, and benzene, for which data were available for concentrations of acetylene, PAH, and soot and, in the case of the one-dimensional flames, for soot particle size distributions, were analyzed with the aim of understanding the dominant characteristics of the soot surface growth reactant. Soot mass growth rates were calculated for all of the conditions, and marked deviations between the PFR and one-dimensional flames were observed. Soot growth rate increases and oscillates in the PFR and sharply declines in the one-dimensional flames in the region of soot growth after initial particle inception. Under all of these conditions, PAH show he characteristics required of the dominant surface growth reactant; increases and oscillations in the PFR and sharp declines in the one-dimensional flames. For acetylene to be the dominant surface growth reactant, anomalous behavior of acetylene-soot reactivity would be required that cannot be explained by soot aging or radical intermediates. This leads to the observation that the sometimes assumed decline in soot reactivity may instead be the effect of decreasing PAH concentration after soot inception.

Interactions with other CAO Projects

We have interacted with Professor Richard Flagan in the area of scanning mobility particle sizing of soot in flames, with Professor Joseph Bozzelli in the area of aromatics formation and oxidation, with Professor Wai Cheng in the area of soot formation, with Professor Paul Barton in the data incorporation method of kinetics modeling, and with Professor William Green in the calculation of PAH thermochemical properties.

Status of Original Plan

The extents of oxidation achieved, while large enough to show evidence of particle fragmentation in some cases, were in general too small (i.e., less than 70%) to allow a clear discernment or accurate characterization of the fragmentation process. The reason for this shortcoming is that the feed rate of nitrogen to the JSR to control the temperature increase resulting from oxygen injection in the PFR was found to be limited by the flow resistance in the exhaust system and the associated increase in pressure in the JSR/PFR. Consequently, the oxygen injection rates needed for the higher extents of soot oxidation could not be achieved. The exhaust system needs to be modified so that a new set of experiments with higher oxygen injection rates can be performed.

Future Plans

A proposal has been submitted for an additional year of work, which would involve the following tasks. The JSR/PFR experiments already performed would be extended to higher extents of oxidation by using larger oxygen injection rates into the transition region between the JSR and PFR, thereby increasing the oxygen concentration and extent of oxidation in the PFR. These experiments would be possible with a new exhaust system now available. The expected additional data would consist of soot particle size distributions and HRTEM images and curvature measurements for the soot structures, for extents of oxidation of up to 85% at temperatures of 1420 K to 1620 K. These data would be used to improve the knowledge of soot oxidation kinetics by including descriptions of particle fragmentation and the decrease of soot reactivity with time during burnout.

Handling of QA/ QC

The details of this research are recorded in the theses of David Kronholm and Thilo Lehre. The documentation includes specification of error limits and uncertainty. In the experimental work the error limits are computed from the errors associated with the instruments and the measurements. In the data analysis and modeling the error limits are associated with the uncertainty in the information used.

Key Personnel

Graduate Student:

Murray J. Height