<%BANNER%>

Interaction of Iron Species and Soot Particles in an Isooctane Diffusion Flame


PAGE 1

INTERACTION OF IRON SPECIES AND SOOT PARTICLES IN AN ISOOCTANE DIFFUSION FLAME By KIBUM KIM A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLOR IDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2006

PAGE 2

Copyright 2006 by Kibum Kim

PAGE 3

This work is dedicated to my family. Their support, encouragement and love made its completion possible.

PAGE 4

iv ACKNOWLEDGMENTS First and foremost, I am deeply grateful to Dr. David Hahn for his guidance and leadership during this study. Moreover, his encouragement always kept me looking at the bright side. His generosity and patience with my numerous mistakes in English and research allowed me to challe nge myself without hesitation. I would like to acknowledge the invaluable advice and suggestions of my committee members. I especially want to express my appreciation to Dr. Jill Peterson for her thoughtful concern about my school life. I would also like to thank all of my lab mates (Leia Shanyfelt, Prasoon Diwakar, Cary Henry, Brett Windom, Philip Jackson, Soupy Alexander, Amy Twining, Chris Macarian, and Jeff Crosby) for their help and assistance while I was conducting my research. Their solidarity and friendship ma de lab life more enjoyable, and gave me a great opportunity to learn Amer ican culture including spor ts activities and insightful conversations. Special thanks also go to Kathryn Maseillo and Prasoon Diwakar for their valuable input and cooperation in my combustion research. I would also like to thank my parents for their unconditional love and constant support. In addition, I sincerely thank my brother, Kee-Hoon for his concern and for stimulating me to even greater effort. La st but not least, words cannot express my gratitude to my lovely wife Yong-Soon; and my adorable son, Daniel. At all times, Yong-Soon was a great support emotionally and mentally as I went through the ups and downs of private and professional life. My sweet little boy motivates me to work

PAGE 5

v constantly harder. I am really thankful that I could be with my family during the entire period of my study overseas.

PAGE 6

vi TABLE OF CONTENTS page ACKNOWLEDGMENTS.................................................................................................iv LIST OF TABLES.............................................................................................................ix LIST OF FIGURES..........................................................................................................xii ABSTRACT....................................................................................................................xvii CHAPTERS 1 INTRODUCTION........................................................................................................1 1.1 Soot Formation......................................................................................................3 1.1.1 Formation of Soot Precursor Molecules......................................................4 1.1.2 Particle Coagulation and Growth................................................................7 1.1.3 Particle Agglomeration................................................................................9 1.1.4 Soot Oxidation...........................................................................................10 1.2 Soot Suppression with Transition Metallic Additives.........................................11 1.2.1 Manganese-Based Additives.....................................................................13 1.2.2 Iron-Based Additives.................................................................................14 1.2.2.1 Additives in premixed flames.........................................................14 1.2.2.2 Additives in diffusion flames..........................................................17 1.3 Studies of the Fractal Properties and the Struct ure of Soot Aggregates..............19 1.4 Spectroscopic Method.........................................................................................22 1.5 Objectives of Present Research...........................................................................25 2 FUNDAMENTAL SCIENCE A ND BACKGROUND THEORY............................27 2.1 Elastic Light Scattering Theory...........................................................................27 2.1.1 Rayleigh Scattering Theory.......................................................................29 2.1.2 Systems of Particles...................................................................................31 2.1.3 Rayleigh-Debye-Gans (RDG) Scattering Theory.....................................33 2.1.3.1 Rayleigh-Debye-Gans (RDG) scattering approximation................33 2.1.3.2 Evaluation of the extinction coefficient..........................................35 2.1.4 Sampling and Analyzing Soot Aggregate.................................................38 2.1.4.1 Thermophoretic sampling...............................................................38 2.1.4.2 Transmission electron microscopy..................................................39 2.1.4.3 Energy dispersive x-ray spectroscopy (EDS)..................................41

PAGE 7

vii 2.2 Spontaneous Raman Scattering Theory...............................................................42 2.3 Laser Induced Fluorescence Theory....................................................................47 3 EXPERIMENTAL APPARATUS AND METHODS...............................................51 3.1 Burner System.....................................................................................................51 3.2 Fuel Vaporization and Delivery System..............................................................53 3.3 Flame...................................................................................................................57 3.4 Optical Systems and Diagnostics.........................................................................60 3.4.1 Light Scattering System............................................................................60 3.4.2 Light Scattering Calibration......................................................................67 3.4.3 Signal Processing.......................................................................................70 3.5 Laser Power Measurement..................................................................................72 3.6 Transmission........................................................................................................73 3.7 Thermophoretic Sampling and Tr ansmission Electron Microscopy...................75 3.7.1 Thermophoretic Sampling.........................................................................76 3.7.2 Transmission Electron Microscope...........................................................77 3.8 Spectroscopic Techniques...................................................................................78 3.8.1 Preliminary CO Flame Study....................................................................79 3.8.2 Experimental Apparatus of Lase r Induced Fluorescen ce Spectroscopy...81 3.8.3 Experimental Apparatus of In Situ Raman Spectroscopy.........................83 3.8.4 Isooctane Flame Study..............................................................................86 4 INTEGRATED RESULTS AND DATA ANALYSIS..............................................90 4.1 Smoke Point Study..............................................................................................90 4.2 Elastic Light Scattering Results...........................................................................93 4.3 Transmission Results...........................................................................................97 4.4 Soot Characteristics Determined from RDG Theory...........................................99 4.4.1 Transmission Electron Microscopy...........................................................99 4.4.2 Fractal Properties of Soot Aggregates.....................................................104 4.4.2 Primary Soot Particle Size.......................................................................107 4.4.3 Number Density of Particles...................................................................109 4.4.4 Volume Fraction of Soot Particle............................................................110 4.4.5 The Extinction Coefficient of Soot Particle............................................112 4.4.6 Discussion of Results..............................................................................114 4.5 Spectroscopy......................................................................................................119 4.5.1 Laser Induced Fluorescence (LIF) Spectroscopy....................................122 4.5.2 In Situ Raman Spectroscopy....................................................................130 5 NUMERICAL ANALYSIS......................................................................................136 5.1 Thermodynamic Equilibrium Calculations........................................................136 5.1.1 Flame Temperature..................................................................................136 5.1.2 O2 Flow Rates..........................................................................................142 5.1.3 Fe(CO)5 Concentrations..........................................................................143

PAGE 8

viii 6 CONCLUSIONS AND FUTURE WORK...............................................................146 6.1 Summary and Conclusions................................................................................146 6.2 Future Work.......................................................................................................152 APPENDICES A ANALYSIS OF THE FLAME.................................................................................153 B RESULTS OF RDG CALCULATIONS..................................................................156 C ERROR ANALYSIS................................................................................................168 D STRAY LIGHT CONSIDERATION.......................................................................175 E SOOT REDUCTION MECHANISM......................................................................178 F PROPERTIES OF IRON PENTACARBONYL......................................................182 LIST OF REFERENCES.................................................................................................186 BIOGRAPHICAL SKETCH...........................................................................................193

PAGE 9

ix LIST OF TABLES Table page 1-1 Metallic additives in common..................................................................................12 1-2 Fractal dimension of various aggregates..................................................................23 2-1 Raman shifts and the emission wavelengths of common species............................47 3-1 Data collection heights.............................................................................................54 3-2 Summary of equipment for fuel vaporization and delivery system.........................58 3-3 Summary of gases and fuel used in the study..........................................................59 3-4 Description of the fl ame operating conditions.........................................................60 3-5 Components of scattering system apparatus............................................................62 3-6 Real optical densitie s for various ND filters............................................................65 3-7 The usage of the ND filters for individual height....................................................66 3-8 Average of the number densities, diff erential scattering cross sections, and scattering coefficient sets for methane and nitrogen calibration gases at 1 atm with standard deviation for 24 experimental............................................................69 3-9 Average results of calibration gas signal including stray light, a calibration ratio, stray light signal, the percentage of st ray light, and the ideal reference ratio along with the standard deviation over all scattering experiments..........................71 3-10 Summary of laser beam power propert ies for light scattering measurements.........73 3-11 Description of tr ansmission apparatus.....................................................................74 3-12 Components of spectroscopic system apparatus......................................................87 3-13 Data collection heights for spectroscopy.................................................................88 3-14 Components of the system appa ratus for absorption spectroscopy..........................89

PAGE 10

x 4-1 Average of 4 oxygen flow rates with their standard devi ation and relative standard deviation, the equivalence ra tio, and oxygen to fuel ratio for 10 different concentrations............................................................................................92 4-2 Average (N=10) K'VV results of the unseeded and seeded flame and standard deviations. Flame heights are measured from the burner lip..................................95 4-3 Average (N=6) transmission results of the unseeded and seeded flames and standard deviations. Flame heights are measured from the burner lip....................98 4-4 The summary of the fractal dimension at all heights in the unseeded and seeded flames.....................................................................................................................107 4-5 Diameters of primary soot particle at each height in the unseeded and seeded flames.....................................................................................................................108 4-6 The summary of number densities fo r the unseeded and iron-seeded flames........110 4-7 The volume fraction as a function of flame height for the unseeded and ironseeded flames.........................................................................................................112 4-8 The extinction coefficient of soot part icle as function of flame height for the unseeded and iron seeded flames...........................................................................113 4-9 Complex refractive indices for soot from various sources. (2001)........................118 4-10 EDS result atomic ratio of iron oxide.....................................................................121 4-11 Fe resonance transition wavelengths and corresponding fluorescence emission lines with their relative intensity............................................................................122 4-12 Reference to iron oxides Raman shift (cm-1).........................................................131 4-13 Fe atomic emission peaks.......................................................................................132 4-14 LIBS emission peaks..............................................................................................134 5-1 Mole of reactants used fo r input in the STANJAN code.......................................137 5-2 Products from STANJAN simulation....................................................................137 B-1 Measured radius of th e primary soot particle.........................................................156 B-2 The differential scattering cross section (cm2/sr)...................................................157 B-3 Summary of calculated resu lts for the unseeded flame..........................................158 B-4 Summary of calculated resu lts for the seeded flame..............................................158

PAGE 11

xi B-5 Uncorrected and corrected differe ntial scattering coefficients (cm-1sr-1)...............159 B-6 Differential scattering cross s ection for a fractal aggregate (cm2/sr).....................159 B-7 Number density of soot aggregates in the scattering volume (particles/cm3)........160 B-8 Total scattering cross section for a primary soot particle (cm2).............................161 B-9 Total scattering cross se ction of an aggregate (cm2)..............................................161 B-10 Absorption cross section of a primary particle (cm2).............................................162 B-11 Absorption cross sectio n of an aggregate (cm2).....................................................162 B-12 The extinction cross section of an aggregate (cm2)................................................163 B-13 The extinction coefficient (cm-1)............................................................................163 B-14 Number density of soot particles in the scattering volume (particles/cm3)............167 B-15 The volume fraction of soot par ticles in the scattering volume (cm3 soot/cm3)....167 C-1 Summary of the calculated parameters with Equations C-6 through C-9 for the unseeded flame.......................................................................................................169 C-2 Summary of the calculated parameters with Equations C-6 through C-9 for the seeded flame...........................................................................................................170 C-3 Results of the calculation using Equation C-12 for the unseeded flame................171 C-5 Summary of calculated errors at each height for the particle size and number density....................................................................................................................172 C-6 Summary of calculated parameters at each height for the unseeded flame............174 C-7 Summary of calculated parameters at each height for the seeded flame................174 F-1 Fe vapor pressure in Torr (mm Hg ) as a function of the flame height...................184

PAGE 12

xii LIST OF FIGURES Figure page 1-1 Transmission electron microscope (TEM ) images of soot aggregates from isooctane combustion. A) At 100 nm scale. B) At 0.2 m scale..............................4 1-2 Soot formation. Adapted with permi ssion from a reference (Bockhorn 1994)..........5 1-3 The H-abstraction-C2H2-addition mechanism acting on a biphenyl molecule...........8 1-4 Two processes of particle growth. A) Particle coagulation. B) Particle agglomeration...........................................................................................................10 1-5 Soot formation regimes in a diffusion fl ame, the axial soot concentration profile at the center of the flame, and the radial soot concentration prof ile at an arbitrary flame height..............................................................................................................12 2-1 Light scattering response to an incident light...........................................................28 2-2 Schematic of TEM...................................................................................................40 2-3 Energy level diagrams representing el astic scattering transitions and several inelastic Raman scattering transitions. A) Elastic scattering. B) Resonance Raman scattering. C) Stokes Raman scattering. D) Anti-Stokes Raman scattering..................................................................................................................43 2-4 Relationship between Rayleigh and Ra man scattered lines in a scattering spectrum. Source: Ingle and Crouch 1998...............................................................45 2-5 Energy level diagram of the fluoresce nce process for atoms or molecules.............48 3-1 Concentric diffusion burner schematic. A) Side view. B) Top view. Oxygen goes into the system through the annulus ar ray of ports whereas isooctane and nitrogen flow through th e tube in the center............................................................52 3-2 Concentric diffusion burner. A) Side view. B) Top view.......................................53 3-3 Data measurement heights.......................................................................................55 3-4 Fuel vaporization system schematic.........................................................................56 3-5 Fuel vaporization system..........................................................................................56

PAGE 13

xiii 3-6 Alicat Scientific digital flow meters employed for regulating the flow rates of nitrogen coflow and oxygen.....................................................................................58 3-7 Chemical structure of isooctane...............................................................................59 3-8 Photomultiplier tube. A series of dynodes between cathode and anode provide internal gain..............................................................................................................64 3-9 Sample scattering signals from methan e, nitrogen, and flame. Calibration gases are attenuated by a factor of 100.3 and flame signal is atte nuated by a factor of 105.43 for signal linearity...........................................................................................70 3-10 Top view of the transmission system setup..............................................................73 3-11 A setup of thermophoretic sampling a nd grid. A) Side view. B) Formvar carbon-supported 150 mesh copper grid..................................................................76 3-12 Photograph of the TEM system................................................................................78 3-13 Vaporization system of iron pent acarbonyl and a CO flame burner........................79 3-14 A photograph of the iron pentacarbonyl vaporization vessel and the heater...........80 3-15 Photographs of CO flame. A) uns eeded flame, B) iron seeded flame.....................81 3-16 The optical setup for laser-i nduced fluorescence spectroscopy...............................82 3-17 The optical set-up for in situ Raman spectroscopy..................................................83 3-18 A photograph of an optic al set-up inside OPO........................................................84 3-19 The optical set-up for absorption spectroscopy........................................................89 4-1 Plot of oxygen flow rate at th e smoke point as function of time.............................91 4-2 Smoke point, as measured by the co rresponding oxygen to fuel ratio and the equivalence ratio, as a function of iron pentacarbonyl concentration. Note that the equivalence ratio increases due to a reduction of the necessary oxygen quantity.....................................................................................................................93 4-3 Typical scattered signal response from photomultiplier tube measuring calibration gases and flames. Calibration gas signals are atte nuated by a factor of 100.3, and flame signals are attenuated by a factor of 105 to preserve signal linearity.....................................................................................................................94 4-4 Unseeded and seeded differential scatte ring coefficients in logarithmic scale. Error bars represent on e standard deviation.............................................................96 4-5 Transmission through the unseeded and seeded flames...........................................99

PAGE 14

xiv 4-6 Transmission electron micrographs of soot particles at different axial positions..100 4-7 A log-log plots of N versus Rg/dp 25 soot aggregates were sampled at the height 7 in the unseeded flame..........................................................................................106 4-8 A log-log plots of N versus Rg/dp 25 soot aggregates were sampled at the height 7 in the seeded flame..............................................................................................106 4-9 Diameters of the primary soot particle as a function of the flame height in the unseeded and seeded flames. A polynomi al curve fit was used for extracting more accurate values of dpar...................................................................................109 4-10 The number of primary soot particle as a function of the flame height in the unseeded and iron-seeded flames. A logarithmic curve fit was used for extracting more accurate values of Npar.................................................................111 4-11 Number density of the total soot pa rticle for the unseeded and iron-seeded flames.....................................................................................................................111 4-12 The volume fraction as a function of flame height for the unseeded and ironseeded flames. The error bar repr esents one standard deviation...........................113 4-13 The extinction coefficient of soot partic le as a function of flame height for the unseeded and iron-seeded flames...........................................................................114 4-14 Photographs of tips of the unseeded and seeded flames. Soot plume is seen in the unseeded flame while being not seen in seeded flame. A) the unseeded flame. B) the seeded flame.....................................................................................116 4-15 Photographs of the unseeded and seeded flames. A) the unseeded flame. B) the seeded flame...........................................................................................................117 4-16 TEM images of samples collected in th e Fe-seeded CO flame. A) sampled at the middle of the flame height. B) sampled at the flame tip........................................120 4-17 The typical signa l window of EDS........................................................................120 4-18 Energy level diagram of Fe atom. Bold font indicates the best combination.......123 4-19 Laser induced fluorescence peak for thr ee excitation sources at the two third of normalized CO seeded flame height Excitation lines are shown.........................124 4-20 Fe fluorescence corresponding to the excitation line of 296.69 nm as a function of the CO flame normalized four heights...............................................................124 4-21 Intensity of LIF measured in isoocta ne seeded flame at emission line of 373.49 nm corresponding to the exci tation line of 296.69 nm. Flam e tip is at height of 23.95 cm. Error bars represent one standard deviation.........................................125

PAGE 15

xv 4-22 Transmission of Fe atomic light pa ssing through the seeded flame at two different heights. Fe resonance trans ition line of 271.9 nm was chosen for this study. Flame tip is at th e height of 23.95 cm........................................................127 4-23 Transmission of Fe atomic light from the lamp as a function of 5 different heights of the seeded flame. Fe resonance transition line of 271.9 nm was chosen for this study. Flame tip is at the height of 23.95 cm................................127 4-24 Spectra measured as function of the in cident laser energy at flame tip using the 355 nm source in order to validate the LIBS effect on the LIF signal...................129 4-25 On-and-off resonant LIF signal induc ed by the excitation wavelength of 296.69 nm and 296.19 nm with th e same pulse energy.....................................................129 4-26 Energy level diagram of FeO molecule..................................................................130 4-27 A spectrum obtained from in situ Raman experiment of CO flame using 532 nm as an excitation source............................................................................................132 4-28 Spectra obtained from in situ Raman experiment of CO flame using 355 nm as an excitation source at four different heights.........................................................133 4-29 LIBS emission spectrum obtained from steel rod using 355 nm as an excitation source.....................................................................................................................134 5-1 Relative mass fraction of products as a function of temperature...........................138 5-2 Relative mass fraction of Fe speci es as a function of temperature........................139 5-3 Relative mass fraction of Fe as a function of temperature.....................................139 5-4 Flame temperature as a function of flame height...................................................140 5-5 Relative mass fraction of species as a function of temperature.............................141 5-6 The decrease in relative mass fractio n of the solid carbon as a function of the oxygen flow rate.....................................................................................................143 5-7 Relative mass fraction of the iron species as a function of the oxygen flow rate..144 5-8 Mass fraction of carbon as a function of the Fe(CO)5 concentration.....................145 6-1 Schematic of soot oxidation mechanism. A) Soot oxidation w ithout Fe. B) Soot oxidation with Fe....................................................................................................149 6-2 Schematic of surface reaction mechanis m of hydrogen oxidation. Three steps of the mechanism are adsorption, surface reaction, and desoprtion...........................150

PAGE 16

xvi B-1 The differential scattering coefficients measured along the radial positions at three different heights. Error bars represent one standard deviation......................164 B-2 The extinction coefficients as a func tion of the unseeded flame height. The extinction coefficients determined using RDG scattering theory for two different refractive index were comp ared with that from transmission experiments............165 B-3 The extinction coefficients as a func tion of the seeded flame height. The extinction coefficients determined using RDG scattering theory for two different refractive index were comp ared with that from transmission experiments............166 D-1 Source of stray light in the light scattering optical setup.......................................176 F-1 Fe(CO)5 vapor pressure as a function of temperature (Gilbert and Sulzmann 1974, Trautz and Badstubner, 1929)......................................................................183 F-2 Fe vapor pressure as a function of the flame height. Over the flame height of 15 cm which is the oxidation regime, the va por pressure is negligibly small.............185

PAGE 17

xvii Abstract of Dissertation Pres ented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy INTERACTION OF IRON SPECIES AND SOOT PARTICLES IN AN ISOOCTANE DIFFUSION FLAME By Kibum Kim August 2006 Chair: David W. Hahn Major Department: Mechanic al and Aerospace Engineering Metallic fuel additives have been considered for soot emission control over the last few decades. However, the exact mechanis ms of soot reduction are poorly understood and remain controversial. In response to the need for elucidating the correct chemical processes, elastic light scattering, laser-induc ed fluorescence, and thermophoretic sampling followed by transmission electron micr oscopy analysis were carried out in a laboratory-scale isooctane diffusion laminar flame seeded with 4000 ppm iron pentacarbonyl as the metallic additive. These measurements yielded the size, number density, and volume fraction of soot partic les throughout the flame, including formation and oxidation regimes. In comparison to th e scattering parameters extracted from the unseeded flame, the soot suppression effects of iron pentacarbonyl can be determined to act primarily in the regime of soot burnout or oxidation. It is concl uded that the additive has no direct effect on perturbation of soot in the soot growth zone of the flame, while

PAGE 18

xviii having a significant effect on soot in the bur nout zone of the flame, namely enhanced oxidation, realizing an overall soot suppression effect. In addition to the elastic scatte ring, laser-induced fluorescence and in situ Raman spectroscopy were performed to identify the state of the iron additive in the seeded flame. The results of the spectroscopic techniques reveal that the dominant iron species throughout the primary flame region was Fe, rather than any form of iron oxide. Moreover, elemental iron was observed to di minish through the soot oxidation region. The primary conclusion is that the catalytic effect of Fe atoms and possibly iron oxides enhanced soot oxidation in the burnout regime of the flame, thereby reducing the overall soot emissions. Consistent with this, the noted reduction in smoke point with the addition of iron was also observed.

PAGE 19

1 CHAPTER 1 INTRODUCTION Particulate matter (PM) is the term descri bing small particles found in the air such as dust, dirt, liquid droplets, smoke, and soot. These particles are emitted directly into the air from a variety of sources and are also formed in the ai r through chemical reactions. Sizewise, particles less than 2.5 m in diameter are called PM2.5 (or fine particulate matter). Because such fine particles are linked to both human health concerns and environmental issues, various efforts have b een made and many scientific studies have been done to find a way to decrease the production rates of fine particles. As a part of these efforts, the Environmental Protection Agency enacted National Ambient Air Quality Standards (NAAQS) for PM and declared that the annual average level of PM2.5 particles in the air should not exceed 15 micrograms per cubic meter (http://www.epa.gov/re gion4/sesd/pm25/p2.htm ) Consequently, sign ificant reductions have been achieved over the last two decades, however, more efforts are needed to ensure that the air is safe enough not to affect human health and the environment. As far as human health is concerned, inhaling PM causes a broad range of illness such as asthma, acute or chronic bronchit is, shortness of breath, painful breathing, respiratory and heart illness, diminished lung function, and even premature mortality (http://www.epa.gov/air/urbanair/pm/index.html ). Due to the small size of these particles, they are capable of penetrating and accumulating in the respiratory system. It is supported by a recent study that particulat e pollutants increase the incidence of cardiopulmonary diseases and ischemic heart attack (Pope et al 2004). A specific type

PAGE 20

2 of PM is soot particles, which are rich in amorphous carbon and polycyclic aromatic hydrocarbons (PAHs), and are known to be mutagenic and carcinogenic (Katsouyanni and Pershagen 1997, Farmer et al 2003). In addition, PM has a harmful influence on the environment in many ways. For example, it leads to atmospheric haze resulti ng in reduction of visibi lity in many parts of the US. It also may play a role in acid rain, which may be responsible for a range of problems. When PM settles on soil and wa ter, it changes the nutrient and chemical balance that are responsible for depleting ecosystems and ruining sensitive forests and farm crops. According to the latest studies, so ot is twice as potent as carbon dioxide in contributing to global warming resulting fr om the green house effect because it can darken snow and ice that results in absorp tion of solar energy rather than reflection (http://www.newscientist .com/article.ns?id=dn4508 ). Such harmful impacts of PM can impact the broad areas because it can travel long distance from the sources (US EPA. 2003). A major source of PM is soot, usually produced through incomplete combustion processes. Controlling these combustion processes is a key method to reduce soot production. There has been much interest in better understanding soot formation and methods of soot reduction. Soot reduction w ould benefit the health of those exposed to soot, for instance, ground crews working at the airport or on airc raft carriers. The moment a jet takes off, the engine thrust and fuel consumption rate are at maximum. As a result, soot emissions also are at ma ximum, and ground crews are exposed to high levels of soot in the exhaust gas from jet engines. Shortand long-term health effects of this exposure are serious concerns, and a means of reducing soot in turbine engines is of

PAGE 21

3 great interest. While the performance of the engine is preserved at an optimum, suppression of malignant soot em issions is most desirable. One approach to achieve this is to increase the temperature of the combus tion process, resulting in promotion of soot oxidation. Another way is to raise the local air-to-fuel ratio. However, these methods have shown the disadvantage of increasing the amount of NOx formed. As a potential solution, soot suppression via fuel additives is an alternative area fo r exploration. 1.1 Soot Formation Soot composed of carbonaceous particles is usually observed in flames and fires as orange luminescence during combustion of hydrocarbon fuels. Soot particles are mostly found as agglomerates of primary particles typically no larger than 500. The hydrogen to carbon ratio in soot ranges between 1:8 and 1:10. Physical characte ristics of soot are described in detail by Palm er and Cullis (1965:p265). The carbon formed in flames generally contai ns at least 1% by weight of hydrogen. On an atomic basis this represents quite a considerable proportion of this element and corresponds approximately to an empirical formula of C8H. When examined under an electron microscope, the deposited carbon appears to consist of a number of roughly spherical particles, strung together rather like pearls on a necklace. The diameters of these particles vary fr om 100 to 2000 and most commonly lie between 100 and 500 The smallest particles are found in luminous but nonsooting flames, while the largest are obtained in heavily sooting flames. A size distribution of individual soot pa rticles is well modeled by a log-normal distribution (Haynes et al 1981). The average diameter of soot particles corresponds to about one million carbon atoms. Figure 1-1 shows typical soot imag es taken as part of this study by transmission electron mi croscopy (TEM) at two magnifications. Soot formation is a kinetically governed process consisting of fuel pyrolysis and oxidation reactions, formation of the first ri ng (benzene) and then polycyclic aromatic hydrocarbons (PAH), inception of the first part icles, growth of soot particles due to

PAGE 22

4 reactions with gas phase species, particle coalescence, agglomeration and finally oxidation. Figure 1-2 illustrate s the soot formation process showing stages of formation on molecular and particulate scales (Bockhor n 1994). However, the process of soot formation has been more generally classifi ed according to the f our stages summarized below. 1. Formation of soot precursor molecules 2. Particle nucleation, co agulation and growth 3. Particle agglomeration 4. Soot oxidation. Sooting characteristics of a flame are complex due to the possible multiple mechanisms of soot formation. Thus, an unde rstanding of the process of soot formation is fundamental to the study of soot reduction in flames and practical combustion systems. 1.1.1 Formation of Soot Precursor Molecules Soot precursor species, most likely pol ycyclic aromatic hydrocarbons (PAH), are formed in the first stage of soot formation. These species act as nucleation sites for the formation of soot. It is presumed that this stage is the rate-limiting step in the soot A B Figure 1-1. Transmission electr on microscope (TEM) images of soot aggregates from isooctane combustion. A) At 100 nm scale. B) At 0.2 m scale.

PAGE 23

5 formation, and chemical kinetics play an im portant role in this stage. Numerous chemical mechanisms have been proposed to describe the formati on of these nucleation sites. All of these mechanisms generall y involve small aliphatic (open chained) compounds that form the first aromatic rings, typically benzene, C6H6. Acetylene, C2H2, is the most abundant aliphatic compound to ini tiate this process in the early stages of Figure 1-2. Soot formation. Adapted with pe rmission from a reference (Bockhorn 1994).

PAGE 24

6 combustion, and benzene leads to the producti on of more complex PAHs in the later stages (Frenklach 2002). One proposed mechan ism is an even-carbon-atom pathway that involves the addition of acetylene to n-C4H3 and n-C4H5 (Equations 1-1 and 1-2). n C4H3 C2H2 phenyl, (1-1) n C4H5 C2H2 benzene H. (1-2) It is proposed based on kinetic simulations of shock-tube acetylene pyrolysis that the reaction in Equation 1-1 plays an important role in forming the first aromatic ring (Frenklach et al. 1988). Moreover, the reaction in E quation 1-2 suggested by Bittner and Howard (1981) is an important pathway to ar omatic ring formation at low temperatures. On the other hand, Miller and Melius (1992) suggested an odd-carbon-atom pathway via combination of stable hydrocarbon radicals like propargyl radicals, C3H3 C3H3 benzeneorphenyl H. (1-3) They insisted that n-C4H3 and n-C4H5 are converted into thei r corresponding resonantly stabilized isomers very rapidly; thus, their conc entrations would not be adequate so that it could significantly impact the formation of aromatic ring. However, recent Monte Carlo theoretical studies predicted the higher stability of n-C4H3 radical and n-C4H5, supporting rather the even-carbon-atom pathway described by the reactions in Equations 1-1 and 1-2 than the odd-carbon-atom pathway. Another possible pathway for the initial ring formation is a combination of two reactant types, highly stable propargyl radical and the most abundant acetylene, to form a cyclopentadienyl radical by C3H3 C2H2 c C5H5. (1-4)

PAGE 25

7 The cyclopentadienyl radical is then rapidly converted benzene. By means of comparing reaction rates of Equation 1-4 w ith that of Equation 1-3, th e reaction of Equation 1-4 is predicted to proceed faster than that of Equation 1-3 by a factor of 2 to 103 (Frenklach 2002). It implies that the reaction 1-4 plays a dom inant role in forming the first aromatic ring. In addition to these pathways reviewed above, many others have been proposed to characterize the initial stage of soot formation, but have not been widely accepted. Soot inception is regarded as the most critical stage in soot formation, and is subject to perhaps the greatest debate. 1.1.2 Particle Coagulation and Growth The transition from molecular to particle properties occurs in the second stage of soot formation, namely particle coagulati on and growth. This transition occurs at a molecular weight of about 104 amu corresponding to an incipi ent soot particle diameter of about 3 nm. In this stage, soot par ticles collide with each other forming larger spherical particles. This is called the pro cess of coagulation, which dominates the early soot particle growth. The size of particles increases while the particle number density decreases in the coagulating process. Coagulation is limited to very small particles, on the order of ~18 nm or less. Aromatics play a role in growth toward soot particle, as gas phase species are attached to the surface of a pa rticle and become incorporated into the particulate phase. Frenklach (2002) described this mechanis m with a process of H-abstraction-C2H2addition (HACA), in which H atoms are ab stracted from aromatic compounds, and gaseous acetylene is incorporated to bri ng on growth and cyclization of PAHs. The process of H-abstraction-C2H2-addition is described by

PAGE 26

8 Ai H Ai H2, (1-5) Ai C2H2 products, (1-6) where the notation Ai is an aromatic molecule with i peri-condensed rings, and Ai is its radical. The repetitive reaction sequence of two principal steps in Equations 1-5 and 1-6 implies abstraction of a hydrogen atom fr om the reacting hydrocarbon by a gaseous hydrogen atom, and the formation of the ra dical site by adding a gaseous acetylene molecule respectively. Figure 1-3 represents an example of the aromatics growth via the process of H-abstraction-C2H2-addition that H abstraction fr om a biphenyl molecule and the subsequent addition of acetylene. Figure 1-3. The H-abstraction-C2H2-addition mechanism acting on a biphenyl molecule. A biphenyl molecule is formed in the pyrolysis of benzene, a H atom is abstracted from a biphenyl molecule, and the subsequent ad dition of acetylene occurs. It is possible for the growth of aromatic compounds to o ccur via different mechanisms specific to the fuel and flame conditions; however, us ing numerical simulations Frenklach et al. (1988) showed that these alternate methods quickly relax to the acetylene-addition mechanism. A process of HACA is sustained until the H atom concentration or the number of active sites on the soot partic le surface reduces in this st age. Eventually, the surface + + H + H2 + H + C2H2 + H

PAGE 27

9 growth rate of soot partic les declines and subsequently particle growth via these mechanisms ceases. Such phenomenon is termed soot surface aging. It was originally believed that the depletion of growth speci es was responsible for this phenomenon. Recently, it is now believed th at a decrease in the surface re activity of the soot is the main cause for the reduction of soot surface growth rate although it is not even fully understood how the soot particles lose surface reactivity (Harris et al. 1983). To support the theory, it is proposed that the decay of s oot surface reactivity is strongly connected to increase in the ratio of C to H atoms in the soot (Harris et al. 1983, Haynes et al. 1979). By describing the proposal in a chemical se nse, the surface reactions depend on a radical site formed by the abstraction of a H atom. Meanwhile, in physical sense, if it is assumed that the hydrogen in the particle is contained only at the edges of the aromatic ring, it can be seen that the C to H ratio will increase as the particle grows. As a result, the number of possible growth sites decreases. It is inco mplete to fully characterize the decay of soot surface reactivity with this method. Both these chemical and physical effects would lead to a direct proportionality between the H to C ratio and surface reactiv ity with this model; however, the C to H ratio decays 2 to 3 times more slowly than the surface reactivity (Dasch 1985). The molecular details underlyi ng the decay of the soot surface reactivity are under investigation to bette r understand this mechanism. 1.1.3 Particle Agglomeration When the viscosity of the particles in creases past a critical value due to dehydrogenation of the condensed phase, coag ulation transitions into chain-forming collisions (Prado et al. 1981). This is the third stage of soot formation, that is, particle agglomeration. When individual soot particle s collide, they stick to each other leading to fractal aggregates. Contrary to particle co agulation, the particles still preserve their

PAGE 28

10 original identity in agglomeration. Soot a ggregates have been analyzed in terms of fractal geometry. The fractal dimension, discussed in detail later, determined in numerous flames seems to be confined to a rather narrow range, namely 1.7-1.8. Individual aggregates of s oot particles generally contai n 30-1800 primary particles and are well characterized by a lognormal size distribution (Warnatz et al. 2001). Figure 1-4 elucidates the difference between coagulation and agglomeration. A B Figure 1-4. Two processes of pa rticle growth. A) Particle coagulation. B) Particle agglomeration. 1.1.4 Soot Oxidation Soot oxidation also called burnout, the final stage in soot formation, takes place at near the outer radii and the fl ame tip as oxygen diffuses into the combustion zone. In this stage, the soot particles are partially or co mpletely broken down, which yields CO or CO2 as a product. Oxidants in soot destructi on are O atoms and OH radicals as well as O2. According to studies by Warnatz et al. (2001), the concentration of O atoms is relatively low compared with that of other oxidant s in sooting flames. Consequently, the

PAGE 29

11 probability of reactions between O atoms and soot is also low. Therefore, it is assumed that OH radicals and O2 are primarily responsible for the oxidation of soot particles (Warnatz et al. 2001). A major source of soot is flames, which ma y be considered as either premixed or diffusion (or non-premixed) flames. In a pr emixed flame, fuels are premixed with oxidizers at the molecular level before any significant chemical reaction occurs. This type of flame is typically obser ved in Bunsen burner as well as the spark-ignition engine. This type of flame may have insignificant oxida tion of soot because most of the oxidizers are consumed before soot particles are fully -grown. In a diffusion flame, the reactants are initially separated, and then they are mi xed and react only at th e interface between the fuel and oxidizer. A classic example of a di ffusion flame is a candle. Soot oxidation in the diffusion flame is predominantly no ticeable at higher flame heights as oxygen diffuses into the combustion regime and encounter s mature soot particles. Therefore, the stages of soot formation can be divided mo re distinctly in the diffusion flame (Turns, 2000). Figure 1-5 shows soot formation regi mes in a diffusion flame, the axial soot concentration profile at the center of the flame, and the radial soot concentration profile at an arbitrary flame height. It can be seen that small quanti ties of soot are present in the inception regime while peak formati on occurs in the growth regime. 1.2 Soot Suppression with Tran sition Metallic Additives A wide variety of metallic additives in fuels has been studied to determine their effects on soot formation in many practical a nd laboratory scale combustion systems. In common, the alkali, alkaline earth and main tr ansition metals have been used as fuel additives to control soot em ission. Common metallic addi tives are summarized in Table 1-1.

PAGE 30

12 Figure 1-5. Soot formation regimes in a di ffusion flame, the axial soot concentration profile at the center of the flame, and th e radial soot concentration profile at an arbitrary flame height. Table 1-1. Metallic additives in common. Alkali Li, Na, K, Rb, Cs Alkaline earth Mg, Ca, Sr, Ba Transition Fe, Mn, Cr, Ni The mechanism of action of metallic fuel additives have been outlined in three different theories. Firstly, the fuel additive may affect nucleation mechanisms of soot formation in the early stage of soot particle inception. Secondly, the additive may enhance soot burnout as a result of rapid elimination of s oot precursors attributed to Soot concentration arb. uni t Relative axial p osition h/h 0 0 1 Relative radial p osition r/ r 0 Soot concentration arb. uni t 0 1 -1

PAGE 31

13 increase in hydroxyl radicals. Thirdly, the additive may accelerate the soot oxidation rate by occlusion within the soot phase. Obviously, all three mechanisms may be closely interrelated. The global and local effects of transition metallic additives were evaluated in many studies using a variet y of techniques from simple visual observations to novel laser diagnostic measurements. The review c onducted in this section will be limited to the key studies of transition me tallic additives in premixed and diffusion flames. In spite of the same type of combustion conditions, many studies often have yielded different conclusions. 1.2.1 Manganese-Based Additives Linteris et al. (2002) reported soot reduction effects of manganese and tin containing compounds by analyzi ng the burning velocity of me thane/air flames. Greater than 50% reduction of the burning velocity was shown in the seeded flame. In comparisons of the reduction efficiency w ith other suppressants, manganese-based additives showed about a factor of two less th an that of iron pentaca rbonyl, but twice as effective as bromine-based additives. This result is supported by a study of Wei and Lee (1999), who pyrolyzed polystyrene with manganese in a laboratory quartz reactor. Although results from several measurements varied slightly relying on the different conditions, overall 40% of reduction was obtained in the pyrolysis reacti on with manganese. They concluded that the addition of manganese sulfat e into the high temperature py rolysis of PS inhibited the formation of PAHs in the reaction. However, Feitelberg et al. (1993) found an adverse effect, namely that the additive increased soot volume fraction by approximately 50% in a study of a premixed ethylene flame seeded with manganese added in 140 ppm concentrations. They expected that

PAGE 32

14 manganese would exist in the gaseous phase as free metal atoms at high temperatures and form solid MnO through precipitation at residence times. Hayhurst and Jones (1989) al so investigated the effect s of metallic additives on ionization in premixed acetylene/oxygen/ar gon flames. It was found that manganese addition did not affect ion concen trations and soot particle size due to their relatively high ionization potentials that leads to the low ra tes of soot nucleation and particle growth. Consequently, they made a conclusion that manganese had no inhibition effect on soot production rates. 1.2.2 Iron-Based Additives While manganese is a known neurotoxin, ir on has relatively low toxicity; therefore, many combustion applications a nd laboratory studies have c oncentrated on the iron based additives such as ferrocene [(C5H5)2Fe] and iron pentacarbonyl [Fe(CO)5]. In many studies, they have been shown to be hi ghly effective soot suppressants (Bukewicz et al. 1974, Feitelberg et al. 1993). Iron pentacarbonyl was select ed in this research to study the effects of the iron on a laminar prevapori zed isooctane/oxygen diffusion flame. It is an organometallic solution that is soluble in liquid isooctane allowing for a simple means of regulating and delivering th e dopant to the com bustion system before vaporization of the fuel. This factor makes iron pentacar bonyl an ideal additive for this study. 1.2.2.1 Additives in premixed flames In a study of a laminar premixed ethylene flame seeded with 0.015-0.46% ferrocene by weight of the fuel, Ritrievi et al. (1987) studied the eff ects of the addition of ferrocene, Fe(C5H5)2, on inception and growth of soot par ticles. As particles moved from inception to growth regime, an increase in the diameter of soot particles in both seeded and unseeded flames was observed, and the di ameter of initial particles in the seeded

PAGE 33

15 flames was smaller, whereas final particles had a larger diameter than those in the unseeded flames. Contrary to the size, the num ber density of particles in both flames was reduced with height. The same trend on numb er densities was obser ved in the previous work achieved by Haynes et al. (1980). As for the volume fr action of soot, indiscernible change was shown at early residence time fo r both flames, and the final volume fraction was greater in the seeded flame at all the times. In addition, the spatial profile of the elements iron and carbon in the soot particles was determined with Auger electron spectroscopy. They found that the soot particles consisted of dense iron at the core and a thick carbon-rich layer at the outer surf ace. Mossbauer spectroscopy was used to determine the chemical state of the iron in the particles, and iron oxide, FeO, was found to be the stable species on the given flame conditions. It is noted that all analysis was done with sampled (i.e. extr acted) soot particles. In order to account for the different beha viors of particle in ception and growth shown for the seeded and unseeded flames at an early residence time, they proposed a hypothesis that FeO homogeneously nucleated early in the seed ed flame, prior to soot particle inception. This was able to illustrate the behavior s and the stratification of the soot particle at an early residence time in th e seeded flame very well. Additionally, they concluded that the carbon deposit ed on the particles was used for the direct reduction of FeO to metallic Fe. The consumption of carbon at the surface resulted in slower growth rates at the earlier growth region of the seeded flames and indicated that FeO is relatively less active in the later soot growth region. However, the catalytic effect of iron in the later residence time had an influence on growth of the particle surfaces.

PAGE 34

16 Feitelberg et al. (1993) also found that the addi tives increased overall soot formation in studies of a laminar premixed ethylene flame seeded with ferrocene. Iron was added to the fuel in 200 ppm concentratio ns by a molar basis. Overall, the soot volume fraction in seeded flame increased three times at later residence time, and particle size also increased with increasing residence time. This agreed well with Ritrievis conclusion (1987). However, they did not find any additive effect on the number density at an early soot inception region, while Ritrievis group found a significant additive effect. After analyzing the states of iron additive in the flame, Feitelberg concluded that the iron would initially exist in the gas phase as free metal atoms and precipitate out of the gas into metallic iron form at high temp eratures of about 1760 K, or at residence times of around 4 ms. In addition, it was concluded that thermodynamically FeO was not formed in the fuel rich flame, whereas it ex isted in the seeded flame in the Ritrievis work. To conclude, the role of the iron additi ves was not to affect soot particle inception but to increase the rate of gas-solid reactions leading to increase in the total mass of soot. As in Ritrievis work, they also paid atten tion on the catalytic effects of iron in the flame with a catalyzed acetylene addition model, and concluded that the iron additive played a role as a catalyst to carbon deposition via acetylene which increased the final particle size. Hahn (1992) assessed the role of iron pentacarbonyl, Fe(CO)5, in a premixed propane/oxygen flame with a fu el equivalence ratio of 2.4. Iron pentacarbonyl was added in concentrations of 0.32% by weight of iron to the fuel. The lower regions of the flames were not evaluated due to the limitation of in situ photocorrelation measurements.

PAGE 35

17 Within all regions of the flame studied, the iron additive had an effect to increase the overall amount of soot in the flame. That is the size, number density of soot particle and a volume fraction in the seeded flame were gr eater than those in the flame without the additive. Even though in situ analysis was not carried out, th e state of the iron in sampled particles was experimentally assessed using X-ray photoelectron spect roscopy rather than using prediction models. In this analysis, it was found that the form of iron oxide, Fe2O3, was a dominant species in the extracted soot part icles. Contrary to Ritrievis conclusion or Feitelbergs analysis, signi ficant quantities of elemental Fe or other iron oxide, FeO, were not identified in this study. They hypothesized that the role of the metal additives on the reduction of soot emission is to accelerate soot oxidati on rate in the burnout zone (Cotton et al. 1971, Hahn 1992). This region is not present in premixed flames; therefore, the full effect of the metal additive could not be seen. The foregoing studies of premixed flame have demonstrated that metallic additives made an increase in overall soot emission by either the catalystic effect of the metal in the later residence time, or acting as soot nucleation sites in the inception region. A complementary picture of the effect of additives can be investigated with the ad dition of the burnout regime in diffusion flames. 1.2.2.2 Additives in diffusion flames In addition to premixed flame studies, th ere are a lot of soot suppression studies using iron based additives in diffusion flames. Bonczyk (1991) studied the effect of an additive on soot production with a pre-vapor ized isooctane/air diffusion flame seeded with ferrocene added in 0.3% by weight of fuel In this study, he observed an increase in the diameter, the number density of particle s, and volume fraction in both seeded and

PAGE 36

18 unseeded flames in the regime close to the burner lib. In the burnout regime of the flames, these parameters however, decayed rapidly in the seeded flame while those kept slightly increasing in the unseeded flame. Th is net effect of soot reduction was visually noted as well when the smoke plume ex isting in the unseeded flame completely disappeared in the seeded flame. Soot sa mples were collected post-flame and subjected to an Auger-type chemical analysis so that th e species of iron present in the soot could be determined. From the Auger data, it was f ound that a condensate from the seeded flame with 0.3% ferrocene was determined to be Fe2O3 containing only negligible amounts of carbon. In contrast, the condensate was carbon retaining less than 2% of elemental iron when the percentage of ferrocene in the fuel was reduced to under 0.001%. Bonczyk concluded that the metal a dditive contributed to not only soot enhancement in soot inception zone but al so soot reduction in burnout zone. With respect to a qualitative illustration on the soot enhancement by additive in the early residence time, he supported conclusions of Cotton and Ritrievi that soot enhancement was a result of an increase in nucleation sites provided by solid FexOy particulates and an increase in the surface activity of particles resulting from a ca talytic effect of iron on the carbon deposited on the surfaces of soot partic les. The required Fe is produced by the reaction in Equation 1-7, FexOy yC xFe yCO (1-7) The presence of the metal additive enhances the soot reduction in the burnout zone as well. Iron oxide catalytically reinforces the removal of carbon by molecular oxygen, but this requires the iron metal to diffuse thr ough the soot matrix to the surface and its subsequent oxidation by

PAGE 37

19 xFe 1 2 yO2 FexOy. (1-8) In the combination of two reactions above, the net effect of car bon oxidation due to the metallic additive is expressed as C 1 2 O2 CO. (1-9) In short, the additive enhances carbon oxidation, and the result is a net reduction of soot in the burnout zone. The similar tendency of soot reduction via addition of ferrocene into ethylene coflowing diffusion flame was found by Zhang and Megaridis (1996). Ferrocene seeding accelerated soot inception, but enhanced soot ox idation in the tip of the flame. The soot volume fraction of the seeded flame was about an order of magnitude less than that of the unseeded flame. Besides, ferrocene affected the primary particle size at the flame terminus so that 33% net reduction of s oot was observed between the unseeded and seeded flames. Kasper et al. (1999) also reached the same conclusion in a study with ferrocene seeded methane/argon and acetylene/ar gon flames. The soot production rate of seeded flames was higher at the early residen ce time due to an increase in the surface of soot, but lower at the later residence time attributed to efficient soot oxidation by catalytic means of the additive. 1.3 Studies of the Fractal Properties and the Structure of Soot Aggregates Numerous studies concerning the physical pr operties of soot aggr egates have been reviewed by many researcher s. A research group led by Faeth has performed numerous work on fractal and structure properties of soot aggregat es using Rayleigh-Debye-Gans (RDG) scattering theory (Kyl et al. 1994&1995a&b, Farias et al. 1995, Wu et al. 1997,

PAGE 38

20 Krishnan et al 2000&2001). They accomplished it with both gaseous (acetylene, ethylene, propylene, and butadiene) and liquid fuels (benzene, cyclohexane, toluene, and n-heptane) for a variety of flame conditions, for example, laminar and turbulent flames, as well as premixed and diffusion flames. Th rough their diverse works, it was concluded that fractal properties of soot aggregate are relatively independent of fuel type, flame condition, and position. They obtained a fractal dimension of 1.82 and a fractal prefactor of 8.5, with experimental uncertainties (95% confidence) of 0.08 and 0.5, respectively. Fractal theory is discussed in detail below. They also carried out numerical simulations to create soot aggregates based on clustercluster aggregation. They computationally evaluated RDG theory for the optical propertie s of soot using the Iskander-Chen-Penner (ICP) approach in small scattering angle regi me and compared the results from the ICP approach with those from RDG theory. Th e results were in g ood agreement within numerical uncertainties. Fractal parameters used for the si mulation in their study were Df of 1.75 and Kf of 8.0 based on their proceeding information. In another study, they measured soot composition, density, struct ure, gravimetric volume fraction, and scattering and absorption properties for wave lengths between 350 and 800 nm in the fuellean region of buoyant turbul ent diffusion flames fueled with acetylene, propylene, ethylene, and propane burning in still air. Th en they analyzed these data to find soot fractal dimensions, refractiv e indices, refractive index functions, and dimensionless extinction coefficients using Rayleigh-De bye-Gans scattering for polydisperse mass fractal aggregates (RDG-PFA theory). They f ound both soot fractal dimensions of 1.77 in average and dimensionless extinction coeffici ents of 5.1 in average with a standard deviation of 0.04 and 0.5 respectively.

PAGE 39

21 Kim and Choi et al. (1999&2003) measured the fr actal properties of silica aggregates generated in hydr ogen/oxygen coflow diffusion fl ame using light scattering, thermophoretic sampling and TEM observa tion. They also invented an in situ laser light scattering method for line measurement of aggr egate size and shape, and applied it for the measurement of silica aggregates produced in a methane/air premixed flat flame. The mean radius of gyration and fractal dimensi on of 1.7 were obtained and examined based on the RDG scattering theory for fractal aggregates. Wang and Sorensen (2002) compared scatteri ng cross sections of fractal aggregates predicted by using RDG scatteri ng theory with those that meas ured in an experiment and found a good agreement. For fractal aggregate aerosols of SiO2 and TiO2 formed fractal aggregates by diffusion-limited cluster aggreg ation, the fractal dimensions were roughly 1.75 and the number of primary particles per cluster was approximately 150. Mountain and Mulholland (1988) simulate d the growth of smoke agglomerates using the computer simulation technique of Langevin dynamics. In this study, 48 aggregates comprising between 10 and 687 primar y particles per cluster were created to characterize soot agglomerates and calculate th e light scattering from these agglomerates. The structural information and the results of the calculation were then used to obtain the fractal properties such as the primary particle size, the radius of gyration and the fractal dimension. In short, they discovered the fract al dimension of 1.9 a nd the fractal prefactor of 5.8. Dobbines and Megaridis (1991) investigated the abso rption, scattering, and differential scattering cross sections for polydisperse fractal aggregates with the prescribed fractal dimensions from 1.7 to 1.9 and uniform primary particle size.

PAGE 40

22 Kyl and coworkers (1995a&b) determined the fractal properties for carbonaceous soot and Al2O3 (alumina) agglomerates created from various flame conditions using angular light scattering (A LS) and thermophoretic sampling followed by analyzing transmission electron micrographs. Both procedures yielded the fractal dimension of 1.7 with the standard deviation of 0.15 and the fractal prefactor of 2.4 with the standard deviation of 0.4. Sorensen et al. (1995) sampled soot aggregates from a premixed methane/oxygen flame using thermophoretic sampling and anal yzed them with transmission electron microcopy (TEM) method. They obtained the fractal dimension of 1.74. An analysis with 36-aggregate samples of overfire soot from a laminar acetylene flame reported by Samson et al. (1987) yielded the fractal di mension of 1.4. However, it was regarded that the value was much skewed due to the lack of the number of samples. Sorensen (2001) reviewed scattering and absorption of light by fractal aggregates and concluded that the aggregates typically have the fractal dimension of approximately 1.75. Fractal dimensions determined from va rious sources are summari zed in Table 1-2. Even though the values of fractal dimensi on tabulated in Table 1-2 vary slightly depending on different measur ement techniques and flame conditions, the main fractal properties of soot are generally considered to be independent of the fuel and the flame conditions. 1.4 Spectroscopic Method For identifying the state of the me tallic additive without perturbing the characteristics of the flame, the mo st effective method is to use an in situ spectroscopic method. Having an advantage of high sensit ivity, Laser-induced fl uorescence has been

PAGE 41

23 Table 1-2. Fractal dimension of various aggregates. Investigator Fractal dimension Condition Method 1.82 soot from laminar and turbulent diffusion flames TEM 1.73~1.85 soot from various hydrocarbon fuels flame scattering and extinction measurements Faeth et al. 1.75 general soot aggregate computer simulation Wang et al. 1.75 aerosols of Si O2 and TiO2 Light scattering 1.77 silica aggregates generated in hydrogen/oxygen coflow diffusion flame Choi and Kim 1.7 silica aggregates produced in a methane/air premixed flat flame Light scattering and TEM Mountain et al. 1.9 smoke computer simulation Dobbines and Megaridis 1.62, 1.74 soot from laminar ethylene TEM Cai et al. 1.74 soot aggregates from a premixed methane/oxygen flame TEM Sorensen 1.75 general aggregate Review Samson et al. 1.4, 1.47 soot from laminar ethylene TEM 1.75, 1.86 Angular light scattering Kyl et al. 1.54~1.73 soot from various laminar and turbulent diffusion flames TEM widely used for measuring the concentration and temperature of gaseous phase species in combustion flows. Planar laser-induced fluorescence and Ra yleigh/Mie imaging measurements were conducted to investigate the mechanisms of pa rticle formation from gas phase species in a CH4/O2 premixed flame seeded with iron carbonyl (McMillin et al. 1996& Biswas et al. 1997). A XeCL excimer-pumped dye laser opera ting in 5 mJ pulse energy was used for FeO PLIF. While the excita tion laser was being scanning from 558.5 nm to 561.0 nm, the fluorescence was monitored near 586 and 618 nm with PMT and boxcar averager. They found that the concentrati on of vapor phase FeO rapidly ri ses at the flame cone (i.e.

PAGE 42

24 primary reaction zone) and FeO plays a role as precursors. To validate the experimental results, they developed a discrete secti onal model which accounted for precursor vapor concentrations and particle growth process. The simulation was in good agreement with the experiment. Son et al. (2000) conducted photolysis-probe ex periment to generate the ground state FeO molecules which were detected by LIF method. By means of directing an unfocused weak UV laser beam into the mixture of Fe(CO)5/(O2 or N2O)/(He or Ar), they created ground-state FeO molecule. The wa velength of the laser was in the range between 298 and 320 nm, and laser pulse en ergy of 0.5 ~ 1 mJ/pulse was used for photolysis. Then the fluorescence at 623.6 nm was detected when the FeO molecule was excited by a wavelength of 591.1 nm. Telle et al. (2001) combined LIF with LIBS to detect elements in non-accessible environment. They performed a parametric study with the combination of LIF and LIBS to investigate analytical selectivity and se nsitivity, and concluded that the combined technique is better than LIBS alone in sensitivity and selectivity. Nguyen et al. (1996) invented a combination of Raman-Rayleigh scattering and LIF to measure temperature and the concen tration of NO in a methane-air premixed flame under three different operating c onditions. Two fre quency-doubled Nd:YAGpumped dye laser systems were employe d for NO LIF. Then, the quenching was corrected with information from Raman-Rayl eigh scattering experiment. They observed that NO concentration reduced as th e equivalence ratio increased. As another common technique, in situ Raman spectroscopy has been employed for species identification and quantification. Maslar et al. (2000) observed various forms of

PAGE 43

25 iron oxide while investigating corrosion on th e surface of an electrolytic iron coupon in air saturated water at a pressure of 25.1 MP a and temperatures from 21 to 537 C using in situ Raman spectroscopy. The excitation so urce radiation was a 647.1 nm krypton ion laser. The in situ Raman spectra were compared with the ex situ spectra using the microRaman system having the excitation source of 785 nm. In this study, they realized that ex situ spectra were similar to the in situ spectra taken during coo ling but different from those taken during heating. 1.5 Objectives of Present Research Although the use of fuel additives as s oot suppressants has been known for over 40 years and widely studied, the mechanism of action of additives is poorly understood and still a subject of controversy. The primary objective of this pr oject is to quantitatively explore a role of the additive for soot suppr ession in the flame using the elastic light scattering technique along with therm ophoretic sampling followed by transmission electron microscopy (TEM) and in situ spectroscopy. In a ddition, Laser-induced fluorescence spectroscopy (LIF) and in situ Raman spectroscopy are used to identify the chemical state of the iron additive in th e flame. Finally, numerical simulation is performed to provide additional information on iron species in the flame. As another prevalent scheme, Laser-induced incandescence (LII) is a well-researched technique for analyzing and characterizing sooting flames a nd combustion processes. LII occurs when a very intense laser beam en counters particulate matter like soot. A soot particle can absorb energy from the beam, which leads an increase in the particle s temperatures of 4000-4500 K. If the energy absorption rate is su fficiently high, the te mperature will rise to levels where signific ant incandescence (essentiall y blackbody radiation) and vaporization can occur. LII technique wa s employed as a different approach for

PAGE 44

26 analyzing soot particle in other part of the same res earch project; however, a detailed treatment of the characterization of soot particles in the context of LII technique is beyond the scope of the present study, there by, will not be examined in this work. A laminar prevaporized isooctane/oxyge n diffusion flame was invented employing a laboratory scale diffusion burne r, and iron pentacarbonyl, Fe(CO)5 was used as an additive for all researches. Information retrie ved from this research will then be used in future application of soot reduction using pr actical combustion system like a real turbine engine and contribute ultimately to de veloping a solution to minimize health and environmental problems resulting from soot emission. Below are specific objectives. 1. To determine the differential scattering coefficient using in situ techniques of the light scattering and transmission measurements. 2. To evaluate scattering parameters such as the size, number density, and volume fraction of soot particles using Rayl eigh-Debye-Gans scattering theory in combination with thermophoretic sampling of soot followed by TEM. 3. To provide some insights into the role of additives by analyzing scattering parameters in the unseeded and seeded flames. 4. To implement in situ spectroscopic methods such as LIF and Raman scattering technique to probe the chemical state of iron additives species throughout the flame.

PAGE 45

27 CHAPTER 2 FUNDAMENTAL SCIENCE A ND BACKGROUND THEORY This chapter introduces fundamental th eories and background knowledge of the elastic light scattering techni que, spontaneous Raman, Laser Induced Fluorescence (LIF) spectroscopy, and electron microscopic schemes. The former techniques are the remarkable diagnostic methods that are nonintr usive and allow analysis of soot formation process in the diffusion flame w ithout intervening in the chemi cal and physical processes. In addition, the application of each spectrosc opic scheme to data analysis is discussed along with any limitations of those theories. 2.1 Elastic Light Scattering Theory Electromagnetic radiation can interact with a particle in several ways. That is, radiation can be reflected, scattered, abso rbed or emitted. These interactions are dependent on the nature of the heterogeneity: the shape of the particle, the material of the particle (i.e., refractive index), its relativ e size and the clearance between particles. Therefore, a certain particular system can be characterized using the way that electromagnetic radiation inter acts with the particles in the system. Information such as size and number density of the particles can be inferred from the scattering response. In this study, elastic laser light scattering was employed to determine the differential scattering coefficients of soot particles in the unseeded and iron seeded flames. The determined parameter will be used to calculate the number density and total volume fraction, in combination with the size of the soot particles obtained from TEM analysis.

PAGE 46

28 Elastic light scattering takes place in the case that an electromagnetic (EM) wave, the incident light, encounters a scattering particle. At the moment the EM waves collide with discrete particle, electrons oscillate with in the particle at the same frequency as the incident wave. The oscillation, called an i nduced dipole moment, is regarded as a source of light scattering. The energy of the incident light is either discharged by light radiation or extinguished by absorption within the partic le. When the frequency of the incident light is equal to that of scattered light consid ered, the process is called elastic scattering. In contrast, Raman scattering is considered an inelastic scatte ring process. More detailed explanation on Raman scattering will be gi ven later. Figure 2-1 shows the light scattering response to an incident electromagnetic light. Figure 2-1. Light scattering res ponse to an incident light. There are two kinds of categor ies in the elastic light sc attering. One is Rayleigh scattering theory that is app lied to a system with small, dielectric (non-absorbing) and spherical particles. The other is Mie scattering theory that is used for general spherical eElastically scattered light Incident light h incident=h Induced dipole moment Z X Y

PAGE 47

29 solution without a particular si ze limit; hence, it can be used for describing most spherical scattering particles, including Rayleigh s cattering particles. However, Rayleigh scattering theory is usually used as long as it is applicable due to complexity of Mie scattering solution. 2.1.1 Rayleigh Scattering Theory A valid scattering solution using Rayleigh theory for a spherical particle may be obtained under the following conditions: 1. The external electric field s een by the particle is uniform 2. The electric field penetrates faster than one period of incident electromagnetic radiation. These two conditions are sa tisfied for the case of 1 and |m | 1 respectively, where is the dimensionless size parameter given by 2 a (2-1) where a indicates the par ticle radius, and is the relative wavelength defined as o om (2-2) where o means the incident wa velength in vacuum and mo represents the refractive index of the surrounding medium. The refractive index, the pr operty of the material is defined as mni. (2-3) In this Equation, n indicates the common refraction of light while the complex term relates to absorption. The value of k is not exactly zero for any material, but materials with the value approaching to zero are termed dielectrics. The relative refractive index is defined as

PAGE 48

30 om m m (2-4) The magnitude of the relative refractive index, m is 1 22 2 on m m (2-5) In the Rayleigh regime, the vertical-verti cal differential scattering cross section (cm2/sr) indicated in Equation 2-6 means that both the incident light and the scattered light after the interaction are vertically polari zed with respect to th e same scattering plane (xy-plane), see Figure 2-1. 2 2 2 6 2 2 '2 1 4 m mVV (2-6) Simply, the horizontal-horizontal differential scattering cross section shown in Equation 2-7 means that both the incident li ght and the scattered light are polarized parallel to the scattering plane 2 'cosVV HH (2-7) In Equations 2-6 and 2-7, the first subscr ipt means the incident light, and the second subscript means the scattered light. Also, subscripted V and H, respectively, refer to the vertical and horizontal polarizat ion with respect to the scatte ring plane. Note that the vertical-vertical differential scattering cross section is independent of the observation angle while the horizontal-horizontal differe ntial scattering cross section has a minimum at 90 degrees. The tota l scattering cross section (cm2) and absorption cross section (cm2) are defined as

PAGE 49

31 2 2 2 6 22 1 3 2 m msca (2-8) 2 1 Im2 2 3 2m mabs (2-9) Finally, the total extinc tion cross section (cm2) is defined as a sum of the scattering and absorption cross section, namely, abs sca ext (2-10) As represented in Equations 2-8 and 2-9, the total scattering cross section scales with 6 whereas the absorption cros s section is proportional to 3. Compared to abs in the Rayleigh regime, sca is small enough to ignore the contribution of sca to ext ; hence, it can be assumed that abs ext for an absorbing particle. 2.1.2 Systems of Particles The light scattering theory is specifically applied in radiative analyses under the significant assumptions regarding the scattering particle. That is, the particle is assumed to be a single and spherical shape in the sy stem. However, to extend the assumption on the scattering of single particle to a system of particles premises three criteria as elucidated below (Jones, 1979): 1. The particles are spaced far enough to ha ve no electrical interactions between particles. 2. The light does not undergo multiple scattering. 3. There is no optical interferen ce between the scattered waves. The first criterion is fulfille d if individual partic les are placed in a di stance of 2 to 3 diameters from one another. The second criterion is met if the optical mean free

PAGE 50

32 pathlength is greater than the optical length of the system. The last criterion is satisfied if a large number of particles are randomly placed in the system. In this case, the total intensity of scattered light can be determ ined by directly adding the intensities of scattered light from each particle. There are two types of system of par ticles: monodisperse and polydisperse. A system containing uniformly sized particles is termed monodisperse. For such a system, the overall scattering and extin ction of light from partic les can be described by the differential scattering coefficient (cm-1sr-1) and extinction coefficient (cm-1), which are expressed respectively as ''VVVVKN (2-11) ext extN K (2-12) where N is the number density of par ticles (particles per volume). The transmission is defined as the intensity of the transmitted light through a system of particles per that of the incident light for a particular wavelength, namely, o d transmitteI I (2-13) A of unity results from 100% transmission of the incident light through a system of particles, whereas a of zero results from the entire incident light absorbed and/or scattered by the particles. The relationship between the transmission and the extinction coefficient can be correlated by the Beer-Lambert law, namely, ) exp(L Kext (2-14) where L is the optical length. The product L Kext is known as the turbidity, a measure of the ability of the system to extinguish incident light.

PAGE 51

33 The volume fraction vf is also an important parameter for characterizing the particles in a system. The volume fraction is defined as the volume of particles per unit volume, hence, it is dimensionless. For a monodisperse system, the volume fraction is given by N d fv 36 (2-15) where d is the diameter of the scattering particle. A polydisperse system is characterized by non-uniformly sized particles. A flame is a good example of the system, as the sizes of soot particles can vary throughout the sample region. The primary particle size must be determined to evaluate the scattering and extinction properties of a pol ydisperse system. This will be discussed in the next section. 2.1.3 Rayleigh-Debye-Gans (RDG) Scattering Theory As discussed in Chapter 1, the morphologica l characteristics of soot are to evolve into a fractal structured aggregate. Usi ng either Rayleigh scattering theory or Mie scattering theory itself is not a reliable a pplication here for th e large sized and open structured aggregates. Theref ore, a new theory is necessary to exam the size and the fractal dimension of soot a ggregates. Rayleigh-Debye-Gans (RDG) scattering theory has been found to be a suitable approximation for studying physical properties of soot aggregates. 2.1.3.1 Rayleigh-Debye-Gans (RDG) scattering approximation RDG theory has been used to interpret light scattering data to determine cluster parameters. Aggregates produced in the later stage of soot formation vary considerably in size and shape while they grow. Physical features of soot aggregate can be defined

PAGE 52

34 using Rayleigh-Debye-Gans (RDG) scattering th eory. Regarding physical properties of soot aggregates, this theory has the major assumptions described below: The size parameter of primary particles is sufficiently small such that individual primary soot particles satisfy the Rayleigh scattering theory. Soot aggregates are composed of monodisperse nonoverlapping spherical primary soot particles. Primary soot particles ju st touch one another. The number of primary partic les per aggregate satisfie s a lognormal probability distribution function. Soot aggregates are mass fractal-like obj ects with mass fractal dimension of less than 2. The fractal-like objects can be defined us ing the following power law relationship, namely, f D parfgparNkRd (2-16) where parNis the number of primary particles per aggregate, fkis the fractal prefactor, gR is the radius of gyra tion of an aggregate, pardis the primary particle diameter, and fDis the mass fractal dimension implying the ope nness of the soot aggregate. First of all, evaluation of the fractal propertie s requires determining optical properties. Not onlyN but also pardcan be directly determined usi ng transmission electron microscopy (TEM) analysis of post thermophoretic sampli ng of soot aggregates. The radius of gyration then can be calculated usin g either of two correlations (Kyl et al., 1995a). The first approach is to use only the ma ximum projected length of the aggregate,L. Alternatively, the radius of gyration can be evaluated us ing the geometric mean of

PAGE 53

35 maximum length, and the maximum projected width normal toL, W. Two correlations are mentioned below respectively, 49 1 ) 2 ( gR L, (2-17) 17 1 ) 2 ( ) (2 1gR LW. (2-18) With these parameters known, the fractal pr operties are obtained usi ng a linear regression method with a least squares appr oach. In other words, when N is plotted as function of Rg/dpar in logarithmic scale for a set of aggreg ates, the fractal dimension describes the slope, the fractal prefactor determines the magnitude of the least-squares straight line fit to the data. Determination of the radius of gyration re quires more attention because it directly affects scattering properties. The value of gR based on Equation 2-17 may be somewhat greater than that evaluated in Equation 2-18 unless the aggregat e is equilateral shape. As a result, the smaller values of Df are obtai ned when Equation 2-17 is used rather than Equation 2-18. To avoid such a conflict, so me researchers adopted 1.78 or 1.67 instead of taking1.49 in Equation 2-17. Even though sp ecific aggregate properties differ for the various flame systems, all the flames yield the same relationship between the number of primary particles in an aggregate and its ra dius of gyration. Th erefore, the fractal properties of aggregates are independent of various positions and flame conditions. 2.1.3.2 Evaluation of the extinction coefficient The next phase of the presen t evaluation of RDG theory is to consider absorption and total scattering cross sections. Up to this point, the scattering and extinction coefficients have been assumed spatially constant for a given system. However, in the case of a flame, these parameters may be hi ghly spatially dependent on the system. For

PAGE 54

36 instance, the concentration of soot varies very much based on the location in the system. In particular, determination of the extinction co efficient is required to pay more attention because the transmission data cannot be co llected only at representative points to determine it. To better char acterize this type of system the deconvolution techniques have been used if a given system is sufficie ntly large so that the system can be divided into several concentric regions. These t echniques construct information on the radial parameters based upon information in neighborin g sections. However, the technique may not be feasible in case a system area is too na rrow to be separated into a few sections. In such cases, the extinction coefficient for such cases can be determined using RayleighDebye-Gans scattering theory. The differential scattering cross section (cm2/sr) of an aggregate is defined as ) (' 2 'q S Npar par agg (2-19) while the differential scattering cross section of the primary soot particle in an aggregate par is determined using Equation 2-6 presen ted above, namely Rayleigh theory. Subscripts of agg and par mean an aggreg ate and the primary particle, respectively. A new parameter, ) ( q Sshown in Equation 2-19 is called either the stru cture factor or the angular scattering form factor given differently based on th e scattering angle variety. 1 1 ) ( gqR q S (2-20a) 1 3 1 ) (2 2 g gqR R q q S (2-20b) 1 ) ( ) ( g D gqR qR C q Sf. (2-20c) Under RDG scattering approximati on, the structure factor is nearly unity in the smallangle scattering regime, the so-called Guinie r regime. Consequen tly, scattering mainly

PAGE 55

37 depends upon the number of primary particles in the aggregates. Conversely, the structure factor depends st rongly on the values of Rg and Df in the large-angle scattering regime, the so-called power-law scattering regime. Herein qis the modulus of scattering vector (cm-1) defined as ) 2 / sin( 4 q. (2-21) The constant C refers to the cutoff of the density at the perimete r of the aggregate, and is approximately unity defined empirica lly. The total scattering cross section (cm2) for a fractal aggregate is also expressed as ) (2g sca par par sca aggkR G N (2-22) where 2 / 2 2) 3 4 1 ( ) (fD g f gR k D kR G (2-23) and the total scattering cross section of a single primary particle is simply given by Equation 2-8. Meanwhile, the absorption cro ss section of an aggr egate is defined as abs par par abs aggN (2-24) where it is assumed that absorption is not a ffected by aggregation, while the absorption cross section for a primary soot particle is determined by Equation 2-9. Finally, the total extinction cross section for an aggregate is th e sum of the total absorption of an aggregate and total scattering cross section of an aggregate by Equation 2-10. Note that the scattering volume is absolute ly greater than that of a typical soot aggregate; thus, a number of soot aggregates are involved in interactions with incident light. The number density of aggregate, aggN in the scattering volume can be calculated

PAGE 56

38 using the differential scattering coefficient, agg VVK obtained from the light scattering experiment. That is, the number density of aggregate in the scattering volume is readily determined as '' aggVVaggaggNK (2-25) Therefore, the total number density (particles/cm3) in the overall scattering volume is defined as par agg TotalN N N (2-26) and the extinction coefficient for the overa ll aggregates in the volume is finally determined using the following Equation, ext agg agg ext aggN K. (2-27) 2.1.4 Sampling and Analyzing Soot Aggregate Prior to describing all parameters char acterizing morphology of soot aggregate using RDG theory, thermophoretic sampling and analysis by TEM must be implemented. The principle of thermophoretic sampling and instrumentation of TEM will be reviewed in brief. 2.1.4.1 Thermophoretic sampling Thermophoretic sampling is employed by mean s of collecting particles with a thin film. As particles move across the temperat ure gradient existing between the hot flame and the cold mesh, a process known as therm ophoretic force causes them to move from the higher temperature of the flame to th e lower temperature of the mesh. This temperature gradient is readily established by introducing a mesh ini tially being at room temperature into the hot flame, and it drives the particles to the surface of the mesh where they are deposited. Recently, thermophoret ic deposition has provided a quantitative

PAGE 57

39 understanding of the drift of particles to the su rface that is at a lower temperature than the surrounding gases. 2.1.4.2 Transmission electron microscopy Information about soot aggregates obtain ed with thermophoretic sampling can be extracted using a high-resolu tion transmission electron micr oscope (TEM) equipped with photographic and data recording capabilities. The TEM is an evacuated metal cylinder (the column) about 2 meters high consisting of an electron gun, the illumination system, specimen, and the imaging system. A Ray di agram for TEM is schematized in Figure 22. As a virtual source, the electron gun at th e top of the microscope emits a stream of monochromatic electrons that travel through vacuum in the column of the microscope. The electron gun has a V-shaped tungsten he ating filament that is the cathode emitting electrons. When the cathode is heated, the accelerating voltage of between 40,000 to 100,000 volts is passed between the cathode a nd the anode placed just below the cathode. By using the high voltage, these negativel y charged electrons in the cathode are accelerated to an anode positively charged. The acceleration of electrons depends on the amount of high voltage. Some of electrons passing through a tiny hole in the anode form an electron beam which travels down the column. Electrons are high en ergy particles so that they could easily have an influence on the interaction with any matter. The interaction causes the emission of all the lower form s of energy such as x-rays, secondary electrons, ultraviolet, and heat energy. As a result, the microscope must be kept in a high vacuum of the order of 133 10-8 Pa. This electron stream is focused to a small, thin, coherent beam by the use of

PAGE 58

40 Figure 2-2. Schematic of TEM condenser lenses. The first lens largely dete rmines the spot size, and the second lens changes the size of the spot on the sample. The beam is restricted by the condenser aperture, blocking out electrons far from the optic axis. Specimens in the TEM are examined by passing the electron beam through them. Therefore the mass thickness of the specimens must be thin enough (50~100 nm) to allow electrons to pass through them. When the electrons strike th e specimen, they are either transmitted or scattered depending on the density of the atoms in the specimen. While some electrons are scattered, others are transmitted and hit a phosphorescent screen placed in the bottom of the microscope. It re sults in a contrast of the specimens that relies on both diffraction of electrons and the number of the atoms in the specimen. The Virtual Source Final image screen 1st condenser lens Projector lens 2nd condenser lens Condenser aperture Intermediate lens Selector aperture Intermediate aperture Objective aperture Objective lens Specimen

PAGE 59

41 higher the atomic number of the specimen, the more electrons are scattered and the greater the contrast. The electrons transmitted through the speci men are focused by the objective lens onto a phosphorescent screen to form an image. The quality of the objective lens plays a major role in determining the resolving power of the apparatus. Objective and selector apertures being right below the lens in a row are used to restrict the electron beam. The objective aperture blocks the unfocussed electrons, resulting in an enhancement of the image contrast. The periodic diffraction of elec trons is examined using selector aperture. The transmitted electrons ar e passed down the column through the intermediate and projector lenses. The intermediate lens can magnify the first intermediate image, and the projector lens can form a real image on th e fluorescent screen at the bottom of the microscope column. When the image st rikes the phosphor image screen, light is generated, which enables the image to be observed. The image can be analyzed directly by the operator or photographed with a camera. 2.1.4.3 Energy dispersive x-ray spectroscopy (EDS) The elemental composition of Iron seeded s oot particle can be identified using energy dispersive X-ray spectro scopy (EDX or EDS) that is a method used to determine the energy spectrum of X-ray radiation. The technique employs X-rays emitted from the atoms comprising the sample's surface when th e atoms are struck by electrons. In other words, when an electron from a higher shell fills in an electron vacancy, an X-ray is emitted to balance the energy difference betw een the two electrons. Qualitative and quantitative determinations of the elemen ts present in the sampled volume can be evaluated using the number of emitted X-rays versus their energy. It can be measured by an EDS X-ray detector that is a solid stat e device discriminating X-ray energies. The

PAGE 60

42 energy of the X-ray is characteristic of the element from which the X-ray was emitted. The detector is a semiconductor, usually silicon doped with lithium, and is polarized with a high voltage. When an X-ray photon hits the de tector, it creates elec tron-hole pairs that drift due to the high voltage. The electric charge is collected and a condensator is charged. Increase in the volta ge of the condensator is pr oportional to the energy of the photon so that the energy spectrum can be determined. The condensator voltage is reset regularly to avoid saturation. The detector is cooled to reduce the electronic noise using liquid nitrogen. 2.2 Spontaneous Raman Scattering Theory As discussed previously, el astically scattered radiati on shows the same frequency resulting in the same photon energy as the inci dent radiation. In contrast, inelastically scattered light governed by Raman scattering theo ry has certain shifts in frequency from the incident light. The inci dent radiation coupled into the induced dipole moment can create a quantum shift in the vibrational modes of the molecule. If the shift occurs to a lower photon frequency resulting in the lower ph oton energy, it is termed a Stokes shift. Conversely, a shift to a higher photon freque ncy resulting in the higher photon energy is referred to as an anti-Stokes shift. This shift can occur when a molecule excited via scattering interaction relaxes to a lower vibra tional energy state than initial state prior to excitation. In this case, energy from a molecule is added to the photon. These inelastic shifts of the incident wavelength are determined from Equations 2-28 and 2-29 respectively, 11Stokes oscat (2-28)

PAGE 61

43 11AntiStokes oscat (2-29) where o is the incident wavelength, scat is the wavelength of the scattered light, Stokes and AntiStokes are Stokes and anti-Stokes shifts, resp ectively. The Raman shift is usually quantified by wave number expressed in dimensions of cm-1. The shift in frequency of the scattered photons is species dependent, which enables Raman spectroscopy to be a powerful tool for species identification. Figure 2-3 presents elastic and inelastic scattering effect qualitatively in terms of molecular energy levels. A B C D Figure 2-3. Energy level diagrams representi ng elastic scattering transitions and several inelastic Raman scattering transitions. A) Elastic scattering. B) Resonance Raman scattering. C) Stokes Raman scattering. D) Anti-Stokes Raman scattering. E1 E0 h h inc E1 E0 h incident h inc virtual E1 E0 h anti-Stocks h inc virtual E1 E0 h Stocks h inc virtual

PAGE 62

44 As shown in Figure 2-3, the molecule abso rbs the incident photon and is excited to a virtual electronic state. The case of elasti c scattering transitions is that the molecule relaxes to the original vibrat ional level of the ground electr onic state. In the case of inelastic scattering transitions, however, the molecule does not come back to its original vibrational level, so that the shifts are create d. As presented in the case B in Figure 2-3, the molecule can reach stable electronic stat e past the virtual state when the incident photon energy exceeds an electronic transition ener gy. This process is called Resonance Raman scattering being closely related to fluorescence emission which will also be discussed later. Due to the similarity, fluor escence emission is a major source of noise in resonance Raman scattering technique. Compared to the fluorescence emission process, the resonance Raman process is nearly inst antaneous so that the Raman signal can be discriminated from the fluorescence. Consid ering such a condition, the resonance Raman effect is enhanced 102 to 104 times compared with spontaneous Raman effect. Figure 2-4 elucidates the rela tionship of the frequency and intensity between Raman and Rayleigh spectrum. Several thi ngs are noteworthy in Figure 24. First, two Raman lines are symmetric with respect to Rayleigh line because the energy gain and lose are the same amount for the Stokes line and anti-S tokes line. Second, the Stokes line is apparently more intense than the anti-Stokes line. The intensity of the Stokes line is typically 100 to 1000 times higher than that of th e anti-Stoke line. Th is is attributed to the fact that molecules are highly populated in the ground vibrational state at room temperature; hence, the chance that the incident photons encounter molecules is much more probable in the ground vibr ational state than in the exci ted vibrational states, which is governed by the Boltzman distribution. Fo r this reason, Stokes line is often adopted

PAGE 63

45 Figure 2-4. Relationship betw een Rayleigh and Raman scattered lines in a scattering spectrum. Source: Ingle and Crouch 1998. for signal detection in Raman spectroscopy. In case of elevated temp erature, the intensity of the anti-Stokes Raman signal is enhan ced. Considering the Rayleigh scattering intensity, it is usually thousands times more intense than Raman scattering intensity. The intensity of Raman scattere d radiation is expressed as 0i ikT Re IENg Q (2-30) where' is the differential Raman scattering cross section, 0E is the source irradiance, N is the number of gas molecules, ig is the vibrational degeneracy, Q is the vibrational partition function, i is energy level of a molecule in the initial vibrational state i, kis the Boltzmann constant, and Tis the temperature. Raman intensity is directly proportional to several parameters shown in Equation 2-30. The most significant parameter is the differential Raman scatteri ng cross section that depends on the forth power of 0 vib ; hence, higher photon energy increases the Raman scattering cross section. Therefore, Raman scattering intensity can be significantly increased by Intensity Stocks line Rayleigh line Anti-Stocks line incident vibration incident +vibration incident Frequency vib vib

PAGE 64

46 decreasing wavelength of the excitation light source. Higher photon energy is preferred for excitation, but gives rise to another conc ern about spatial resolu tion of detection for some species like Fe2O3. Some Fe2O3 Raman lines are very close to the Rayleigh line resulting in a difficulty of peak discrimination. Thus, highly efficient long pass filters are essential component to prevent elastically scattered stray lig ht from being detected in such an experiment. In addition, increase in the concentration of the active molecules in the excited volume and the intensity of the excitation source is helpful to obtain significant gains in Raman scattering intensity. As stated before, Raman scattering takes place when molecules are excited and deexcited between rotational states as well as vibrational states. Sin ce the polarizability of single atoms does not change with vibrati on or rotation, Raman scattering technique cannot be used for atomic identification. If the polarizability of a molecule does change during vibrational or rotational modes, the mo lecule is considered Raman-active. Each type of Raman-active molecule results in a particular shift in Raman spectrum allowing Raman spectroscopy to be useful for species identification. Ra man shifts and the emission wavelengths corresponding with the excitation wavelengths of 355 nm and 532 nm for common species are summarized in Table 2-1. Qualitative and quantitative information can be obtained using the wave numbers of the Raman shifts observed in molecules and the radiant power of Ra man scattering. In addition, structural informati on of molecules can be provide d by the depolarization ratio, defined as VH VV I I (2-31)

PAGE 65

47 Table 2-1. Raman shifts and the emission wavelengths of common species. Species Shift (cm-1) Emission 1* (nm) Emission 2* (nm) H2 4156 416.0 683.0 O2 1555 357.4 580.0 N2 2331 386.7 607.3 CO 2143 383.9 600.5 CO2 1285 371.6 571.0 CO2 1388 373.1 574.4 CH4 2917 395.6 629.7 CH3OH 2955 396.2 631.2 H2O 3650 407.5 660.2 Emission 1 and 2 are the emission wavele ngths corresponding w ith the excitation wavelengths of 355 nm and 532 nm respectively. where VH I and VV I denote the Raman radiant powers that are vertically and horizontally polarized scattered radiation with respect to the incident radiati on polarization. Nonspherical shaped molecule can cause the scatte red light to be depolar ized when struck by polarized light. If the vibrational mode is symmetric, the depolarization ratio would be almost zero. However, if the vibrational mode is non-symmetric, depolarization of the scattered radiation can occur. In such a case, the depolarization ratio is predicted as 0.75. The Raman scattering cross section ha s angular dependency on horizontally polarized incident radiation whereas it is i ndependent of vertically polarized incident radiation over all sca ttering angles; hence, a vertically polarized light is often used to avoid any angular dependence on the scattered radiation. 2.3 Laser Induced Fluorescence Theory Laser-induced fluorescence (L IF) is spontaneous emission from atoms or molecules that have been excited to higher levels by ab sorption of laser radiation. The fluorescence process for an atom or molecule is depicted in Figure 2-5. When atoms or molecules are resonantly stimulated by the laser source, they absorb photon energy and are subsequently excited to higher electronic energy states.

PAGE 66

48 Figure 2-5. Energy level diagram of the fluor escence process for atoms or molecules. Spontaneously excited atoms or molecule s decay in the ground state emitting photon energy which is lower than the incident photon energy. In ge neral, the intensity of the fluorescence is proportional to the species concentration, the gas temperature, and pressure. Compared to spontaneous Ra man spectroscopy, a great sensitivity is achievable for LIF due to the higher signal to noi se ratio. In addition, selectivity of a particular excitation source for a given species is capable of avoiding interferences with other species, which is another advantage of LIF spectroscopy. However, the difficulty rises when the elastic light sc attering interferes with the fluorescence signal as commonly presented in Raman scattering technique. Using high pass filt ers can eliminate the elastic light source from the fluorescence signal. The Stokes shift is important to eliminate such effects. The fluorescence signal can be used in quantitative measurements of species concentration, temperature, velocity and pressu re as well as qualitative analysis such as species identification. The concentration measurement is the most common one among these applications. The fluorescen ce signal (photons/s) is defined as exp0 FAFSKN (2-32) h F1 E1 E0 h F2 h inc

PAGE 67

49 where exp is an experimental constant, AK is a rate constant for stimulated absorption,0N is number density of species, and F is fluorescence quantum efficiency. exp includes the incident laser intensity, a focal volume, and a solid angle which can be determined through calibration with known source. AK is known from Einstein coefficient for stimulated absorption and Bo ltzmann distribution. In order to determine 0N, F remains the only unknown given by F F FICISCecK KKKK (2-33) where FK is fluorescence rate constant, ICK is the internal conve rsion rate constant,ISCK is intersystem crossover rate constant, and ecK is the external conversion rate constant defined as ecqKKQ. (2-34) By introducing new parameter in Equation 2-33, 0 F F FICISCK KKK (2-35) the number of unknowns can be reduced to 0 F and qK, which are determined from Stern-Volmer plot. Finally, 0Ncan be evaluated with all parameters determined. However, quantitative measurements are diffic ult as long as quenchi ng is present. The variation in collisional quenching is th e most common cause for uncertainty of fluorescence measurement. Accounting for th e collisional quenching is the very hard problem while the fluorescence signal is related to the absorbing species concentration. Among several approaches to account for quenching in fluorescence measurement, the saturated fluorescence technique is prevalent. Independent conditi on of quenching can be

PAGE 68

50 achieved by increasing the incident lase r energy until absorption and fluorescence dominate quenching. This technique wi ll not be examined here in detail.

PAGE 69

51 CHAPTER 3 EXPERIMENTAL APPARATUS AND METHODS Light scattering techniques and transmi ssion electron microscopy (TEM) analysis following thermophoretic sampling were employe d to characterize the concentration and morphology of soot in this study. In addition, two spectroscop ic techniques were utilized for in situ species detection and identification, namely, laser induced fluorescence spectroscopy (LIF) and spontane ous Raman spectroscopy. The aim of this chapter is to describe and explain each experimental ap paratus utilized, including the associated procedures and analysis methodologies employed. The experimental results are analyzed in Chapter 4. 3.1 Burner System A concentric diffusion burner composed of stainless steel tubing was employed for all experimentations. Figures 3-1 and 32 represent respectively a schematic and a picture of the burner shown from side and t op views. The burner dimensions are also shown in Figure 3-1. Gaseous Isooctane and nitrogen are supplied th rough the inner tube (0.15 cm ID) of the burner. A solid annular disk was press fit between the inner and outer tubes to maintain concentricity. The disk was perforated with 9, 0.03 cm-diameter holes, as shown. The oxygen flow was fed th rough the annular disk and out the nine ports at the burner outlet. Using a stai nless steel mesh flame holder does promote improvement of the flame stability, as show n in a previous study (Masiello, 2004). However, in the present study the oxidation regime must remain undisturbed so that the

PAGE 70

52 ID= ID= AB Figure 3-1. Concentric diffusi on burner schematic. A) Side view. B) Top view. Oxygen goes into the system through the annulus array of ports whereas isooctane and nitrogen flow through the tube in the center. soot characteristics can be extr acted purely from the flame; hence, no stabilizer was used. As recommended in the previous study, the exit area of burner was reduced to increase the flow velocity at the burner exit. As a result, a diffusion je t turbulent flame was created, and fluctuations of th e flame were decreased to an ac ceptable level. In addition, a shroud was placed around the burner to block ambient air currents as well as prevent stray light from infiltrating into the scattering detection optics. The shroud was made of semi-transparent Plexiglas, with dimensions of 25.4.7cm. In order to investigate the soot characteristics at various positions along the vertical axis of the flame, the burner was controlled with a vertical translation stage which allows the burner to move up and down. Twenty-five heights were selected and 0.95 cm 0.2 cm 4.3 cm 13.9 cm 0.31 cm Isooctane + N2 O2 Laser 0.513 cm 0.15 cm 0.696 cm 0.03 cm

PAGE 71

53 A B Figure 3-2. Concentric di ffusion burner. A) Side view. B) Top view. designated with the number 1 through 25, noting that the distance between heights was not necessarily constant. These heights were consistently reproducible, and corresponded to a number of specific rotations of the ve rtical stage knob. That is, one revolution corresponded to non-linear vertical motion. A first height designated was located in 9.4 cm above the burner tip as shown in Fi gure 3-1A. All measurements were made along the centerline at 25 diffe rent heights from bottom to top of the flame. The summary of these heights are tabulated in Table 3-1, and Figure 3-3 represents the relative positions of these heights in the flame. 3.2 Fuel Vaporization and Delivery System A fuel vaporization and delivery system was used for all experiments in this study. It was necessary to vaporize the liquid isooctane fuel before the fuel could be delivered to the burner in order to produce a stable, diffusi on flame, in the absence of liquid droplet,

PAGE 72

54 Table 3-1. Data collection heights. Position label Height above burner tip (c m) Distance between two heights (cm) 1 9.40 2 9.75 0.35 3 10.10 0.35 4 10.50 0.40 5 10.90 0.40 6 11.30 0.40 7 11.70 0.40 8 12.20 0.50 9 12.70 0.50 10 13.20 0.50 11 13.70 0.50 12 14.30 0.60 13 14.90 0.60 14 15.50 0.60 15 16.20 0.70 16 16.95 0.75 17 17.75 0.80 18 18.60 0.85 19 19.60 1.00 20 20.65 1.05 21 21.85 1.20 22 22.80 0.95 23 23.50 0.70 24 24.20 0.70 25 25.25 1.05 for the fuel to be burned more efficiently. By heating liquid isoocta ne to temperatures around 100 C, vaporization was achieved when the liquid isooctane passed through a vaporization system that consists of three main sections: the preheat zone, the vaporization zone, and the delivery line. The boiling point of is ooctane is about 99 C at atmospheric pressure. A schematic diagram and a picture of the fuel vaporization and delivery system are shown in Fi gures 3-4 and 3.5 respectively.

PAGE 73

55 Figure 3-3. Data measurement heights. All three sections of the vaporization system were wrapped with Omega heater tape and fiberglass insulating tape. The heaters we re controlled with tw o PID controllers set at a temperature of 100 C, which was above the boiling point for the liquid isooctane. As shown in Figure 3-4, a nitrog en gas with a flow rate of 0.8 liter per minute was introduced into the vaporization system at th e head of the preheat zone composed of a 0.625 inch diameter, 36 inch long 304 stainle ss steel tube packed with brass balls to promote heat transfer by increasing the surf ace area. The nitrogen gas was heated in

PAGE 74

56 Figure 3-4. Fuel vaporization system schematic. Figure 3-5. Fuel vaporization system.

PAGE 75

57 this region, and then it traveled into th e vaporization zone consisted of a 0.25 inch diameter, 36 inch long 304 stainless steel t ube. The liquid isooctane was supplied into this region via a variable flow Fisher Scient ific peristaltic pump at a flow rate of 0.0015 liter per minute and vaporized by the heated nitrogen gas as well as the direct heat from the hot surface of the tubing. By the nitrogen gas, the vaporized is ooctane was carried to the burner passing through a delivery line co mposed of roughly 50 inches of 0.25 inch diameter braided PTFE hose. This zone was also heated to eliminate the fuel condensation on the way to the burner. For warming the vaporization system up, the heaters were turned on and the nitrogen cofl ow at a flow rate of 0.4 liter per minute about one half an hour prior to any experimentation. This ensured that the vaporization system was adequately heated before the liqui d fuel was introduced into the system. The oxygen stream with a flow rate of 2.6 liter per minute and the isooc tane/nitrogen stream exited the burner to produce the diffusion flam e. The gas flow rates were regulated by Alicat Scientific digital flow controllers whose accuracy was 1% of full-scale. The maximum flow rates of the flow meters empl oyed were 1 liter per minute for the nitrogen and 10 liter per minute for the oxygen, respectivel y. These digital flow meters are shown in Figure 3-6. Table 3-2 summarizes the overall descripti on of the fuel vaporization and delivery system, and Table 3-3 tabulates the description of the gases and fuel. 3.3 Flame As discussed above, a roughly 30 cm long fl ame was created by the burner used in this study corresponding to the prevaporized isooctane/oxygen diffusion jet flame. The main fuel was isooctane, C8H18, which is characterized by a relatively low boiling point, promoting a stable fuel delivery system, and is compatible with a previous study

PAGE 76

58 (Masiello, 2004) that probed the soot in ception and growth regimes. Figure 3-7 elucidates the chemical structure of isooctane. Figure 3-6. Alicat Scientific digital flow me ters employed for regula ting the flow rates of nitrogen coflow and oxygen. Table 3-2. Summary of equipment for fuel vaporization and delivery system. Device Manufacturer Model Description Peristaltic pump Fisher Scientific 13-876-4 Variable flow peristaltic pump Heater tapepreheat zone Omega SRT101-060313 W Heater tapevaporization and delivery line Omega SRT051-060156 W Braided PTFE hose Swagelok SS-4BHT PTFE-lined stainless steel flexible hose Thermocouple Omega Type K Thermocouple Heater controller Omega CN9000A 2 PID controllers HEPA filter Gelman Labora tory 12144 2 HEPA filters Digital flow meter-N2 coflow 0-1 SLPM, accurate to 1% of full-scale Digital flow meter-O2 coflow Alicat Scientific 0-10 SLPM, accurate to 1% of full-scale Magnetic stirrers Fisher Isotemp 11-601-16S 60~1200 rpm, 120V, 50/60 Hz

PAGE 77

59 Table 3-3. Summary of gases and fuel used in the study. O2 Praxair UN 1072 >99% pure N2 Praxair UN 1066 >99% pure Isooctane Fisher Scientific O296-4 HPLC-grade CH4 (used for calibration) Praxair UN 1971 Ultrahigh purity, 99.97% pure Figure 3-7. Chemical structure of isooctane. In order to characterize the flame, Froude number and Reynolds number were calculated for the flame. The detailed calculations are attached in Appendix A. First, the Froude number was 11.8, which indicates that the flame was momentum-controlled rather than buoyancy-controlled, which wa s by design. Second, the Reynolds number was approximately 761.3, which denotes that the flame was laminar flow, as it was smaller than the critical va lue of 2300. Furthermore, th e fuel equivalence ratio was calculated as 1.08 based on the oxygen and is ooctane flow rates. While this value corresponds to fuel rich, significant additiona l oxygen is expected to diffuse into the flame. To better understand the stoichiometr y, a study was made to evaluate the smoke point for the unseeded flame. Oxygen flow ra tes were adjusted to obtain the fuel-rich flame that operated under the smoke point. Further discussion will be presented in

PAGE 78

60 Chapter 4 in detail. Visible smoke was emitted above the flame tip when the oxygen flow rate was below 4.24 L/min. The oxygen fl ow rate of 2.6 L/min was selected for all experiments, based on the need for a reasonabl e smoke load to explore soot suppression. Both the unseeded and iron seeded flames we re tested under the same conditions, except for the presence of the iron pentacarbonyl. Fo r the seeded flames, the iron pentacarbonyl was added to the liquid isooctane supply in 4000 ppm quantities by mass (~0.11% Fe per mass of fuel) and was delivered to the burner through the fuel stream. The selection of this value is discussed below. In all experi mental stages, a magnetic stirrer was used to prevent the iron pentacarbonyl from setting during experiments. The flame operating conditions are summarized in Table 3-4 below. Table 3-4. Description of the flame operating conditions. Stream Flow rate C8H18 (liquid) 1.5 mL/min N2 0.8 L/min O2 2.6 L/min Fe(CO)5 (seeded flame only) 4000 ppm 3.4 Optical Systems and Diagnostics 3.4.1 Light Scattering System Laser light scattering techniques were used to determine the scattering properties of soot particles in the unseeded and seeded flames. The optical setup for this scattering system is sketched in Figure 3-7, and the optical components are summarized in Table 35. For the laser light scattering experiment s, a frequency doubled Q-switched 532 nm Nd:YAG pulse laser (Continuum, Minilite ML-II) was used as a light source. The laser was operated at 10 Hz with a pulse energy of 0. 3 mJ/pulse. The laser was first turned 45

PAGE 79

61 with 532 nm dichroic mirror. The laser beam then passed th rough an aperture to cut out the Gaussian edge of the laser intensity profile befo re it was directed through a focusing lens. Finally, the focused beam passed through the center of the flame at desired vertical position above the burner lip, and was terminated at a beam dump. The cross sectional area of the beam was 0.0033 cm2 at the center of the flame. The scattered light from the soot particles was collected by a photomultiplier tube (PMT) at 90 degree angle from the incident beam. The scatte red light first passed through neutral density (ND) filters in the collection optics line. The ND filters were required to attenuate scattered beam intensity for maintaining signal linearity. After several ND filters, the Figure 3-7. Top view of the li ght scattering system setup. Beam Dump Beam Dump Oscilloscope Opaque Plexiglas Black tube Vertical Polarizer Photomultiplier tube Biconvex lens, f=100 mm Neutral density filters Plano-convex lens, f=250 mm Aperture 532 nm 45 dichroic mirror High Voltage supply Nd:YAG 532 nm pulsed laser Aperture Aperture 532 nm Laser PMT 532 nm bandpass filter Flame

PAGE 80

62 Table 3-5. Components of scattering system apparatus. Device Manufacturer Model Description Equipment 532 nm frequency doubled Nd:YAG Laser Continuum Minilite ML-11 Q-switched, 5 Hz, 2.4 mJ/pulse FWHM=5 ns, 20 mJ max Beam dump Kentek ABD-2 Beam dump Photomultiplier tube Hamamatsu R2949 Photomultiplier tube Photomultiplier tube housing Products for Research, Inc. PR1402CE Photomultiplier tube housing Oscilloscope LeCroy LT 372, WaveRunner 500 Hz, 4 GS/s digital oscilloscope with 50 termination Precision high voltage supply Stanford Research Instruments PS325 Digital high voltage power supply Double shielded BNC cable Pasternack Enterprises RG-223/U Double shielded coaxial cable to reduce line noise Translational stage Mitutoyo Micrometer-adjusted translational stages Optics 532 nm dichroic mirror CVI Laser 45 degree, 532 nm dichroic mirror Aperture Newport ID-1.0 2 apertures Plano-convex lens Newport KBX079AR.14 BBAR coated, 430-700 nm, 25.4 mm diameter, 250 mm focal length FDU-2.0 102.0 attenuation FDU-1.0 101.0 attenuation Neutral density filters Optics for Research FDU-0.3 100.3 attenuation (nominal values) Polarizer Newport 532 nm line filter Newport 10LF10-532 T > 50%, 25.4 mm diameter Aperture ID-0.5 2 apertures in collection optics Biconvex lens UV coating, 100 mm focal length, 25.4 mm diameter

PAGE 81

63 scattered light was passed thr ough a 532 nm band pass filter, a vertical polarizer, a first aperture, biconvex lens, and a second aperture in the coll ecting tube. The 532 nm band pass filter effectivel y removed all wavelengths except 532 nm. The vertical polarizer ensured that the scattered radiation observed was only vertically polarized, which matched the vertically polarized incident beam The polarizer contri butes to reduction of depolarized stray light, but it can also block th e scattered light because the arbitrary shape of the soot agglomerates can cause the scattere d light to be depolarized It is noted that RDG scattering theory as formulated relies on only vertically polarized scattering light. Two apertures reduced background noise (i.e., st ray light) and ensured that the scattered light was collected only from the small scat tering volume defined in the flame. The biconvex lens of 100 mm in focal length defined a solid collec tion angle of about 0.05 sr. The scattered light was finally incident on a PMT and the scattered intensity was recorded on a digital oscilloscope. A precision high vo ltage supply set to -650 V drove the photomultiplier tube. The photomultiplier tube (PMT) is a high gain de tector so that it is very useful for low light level detection. Phot ons are converted into an elec tric signal with a small load resister (less than 100 in common) by a phenomenon, namely the photoelectric effect. The photomultiplier tube contains a photosensi tive cathode and a collection anode that are separated by several elec trodes, called dynodes, providing electron multiplication or gain. When the photocathode is exposed to the electromagnetic radiation, a number of photoelectrons are ejected by the photocathode and hit the fi rst dynode. These electrons strike the next dynode, which results in releasing additional electrons. This

PAGE 82

64 multiplication process continues until electrons arrive at the anode A typical PMT is shown in Figure 3-8. Figure 3-8. Photomultiplier tube. A se ries of dynodes between cathode and anode provide internal gain. PMTs are generally able to output a lin ear response to a co ntinuous signal source over several decades. However, a pulsed nanosecond-scale laser can easily invoke a nonlinear response in the PMT due to the s udden flux of incident photons. Commonly, a PMT is limited to only about one decade of li nearity in such a syst em, therefore signal linearity must be carefully considered. When the strength of a scattered signal to a PMT exceeds the linear limits, the signal must be attenuated to bring the output of the PMT back into the linear response regime using neutral density (ND) filters. A ND filter is characterized by a broad and steady tran smission profile over a wide range of wavelengths. An x ND filter attenuates by a factor of 10x. For examples, a 0.3 filter should attenuate the signal by a factor of 100.3, or approximately 2; thus, the output signal was expected to drop by about one half if the PMT response was linear. The optical densities of the ND filters used were calibra ted previously and are summarized in Table 3-6. h An ode Dynodes Ph o t oc ath ode Eb R E0 E1 E3 E2 E4 E 6 E5 E7 E8 E9

PAGE 83

65 Table 3-6. Real optical dens ities for various ND filters Filter Optical density at 532 nm 0.3A 0.284 0.3B 0.299 1.0A 0.808 1.0B 0.806 2.0A 1.728 2.0B 1.717 As shown in Table 3-6, a total of 6 ND filters were available in a variety of attenuating strengths; thus, maintaining the PM T in its linear regime was simply a matter of adding and removing filters from the PMT incident path. Linearity was checked with nitrogen gas prior to every experiment by plac ing a 0.3 neutral density filter in front of the collection optics and ensuring a factor of 2 intensity reduction. The ND filters for individual height are tabulate d in Table 3-7, which were used to maintain a comparable and linear signal over all heights. As discussed in Chapter 2, the differentia l scattering coefficient is a key parameter for determining number densities of the scatteri ng particles in the flame. The scattering signal obtained by the PMT is defined as VVoVVSIVN (3-1) where Io is the incident laser intensity, is the efficiency of th e collection optics and the PMT detector, V is the scattering volume, and is the solid angle of observation. N and 'VV are number density and the differential scattering cross section as discussed previously. The parameters V, and may be measured although it is hard to determined them individually with the utmo st precision. By taking the ratio of a reference scatterer signal to the signal from th e scatterer of interest the direct evaluation

PAGE 84

66 Table 3-7. The usage of the ND filters for individual height. Position Height N.D. Filter Calibration 0.80 0.3B 1 9.40 2AB+1AB+0.3A 3 10.10 2AB+1AB+0.3A 5 10.90 2AB+1AB+0.3A 7 11.70 2AB+1AB+0.3A 9 12.70 2AB+1AB+0.3A 11 13.70 2AB+1AB+0.3A 13 14.90 2AB+1AB+0.3A 15 16.20 2AB+1AB+0.3A 16 16.95 2AB+1AB+0.3A 17 17.75 2AB+1AB+0.3A 18 18.60 2AB+1AB+0.3A 19 19.60 2AB+1AB+0.3A 20 20.65 2AB+1AB 21 21.85 2AB+1AB 22 22.80 2AB+1AB 23 23.50 2AB+1AB 24 24.20 2AB+1AB 25 25.25 2AB+1AB of the common terms (V) can be avoided. With this approach, Soot VVN' can be calculated using the following equation, 4 4, '' ,,1 ()()vvSoot VVSootVVCH vvCHmeasuredS NN SSL (3-2) Recall VVN is the differential scattering coefficient. In this equation, is the transmission through the flame which will be determined in a later section, SL is stray light obtained by a calibra tion, and the subscript CH4 refers to the methane calibration gas whose signal and the cross-section w ill be reported in the next section.

PAGE 85

67 3.4.2 Light Scattering Calibration A calibration is necessary for determini ng and illustrating the amount of stray light presented in the system. Stray light, such as any ambient light and laser light reflected from surfaces, which enters through the scattering collection optics, can distort the experimental results considerably. The magni tude of stray light can be as large as the scattered signal, thus it is very crucia l to reduce errors cau sed by stray light. As mentioned briefly in the foregoing secti on, several efforts were already used to minimize stray light entering at the PMT. First, using apertures and lenses in the collection optics, a very small scattering vol ume was defined in the experiment. These apertures also played a role to block any st ray light from outside the scattering volume. Highly reflective surfaces of opt ical mounts were either pain ted in black or covered in black felt to minimize reflections of laser li ght from surfaces. In addition, by allowing the beam to pass through a black painted tube placed in the beam path, as well as the barrier with surrounding opaque Plexiglas cove red in black felt, any stray light from outside the scattering volume was largely blocked from introducing into the PMT. In spite of such efforts, it is impossible to completely get rid of stray light from the system. Therefore, the stray light calibration was necessary to compensate for stray light. The calibration was performed using methane (CH4) and nitrogen whose differential scattering cross s ections were already known as summarized in Table 3-8. The calibration gases shared the delivery line of the vaporization chamber, as depicted in Figure 3-4, thus the fuel ga s was expelled from the burner for calibration measurements. A series of brass plug valves were used to shut off the flow from the vaporization chamber during calibration, and vice versa during flame operation. The suppliers specifications of methane were listed in Ta ble 3-3 before. The calibration gases were

PAGE 86

68 injected into the measuring area 8 mm far from the burner lip. Both gases flowed at a rate of approximately 10 L/min, which wa s controlled by a rotameter (GE700 Gilmont) flow tube. The scattering measurement for the calib ration was carried using the identical configuration as used for the soot scattering study. A type K thermocouple was used to measure the temperature of gases exiting from the burner through the heated fuel delivery line. Over 24 sets of the experiment, the te mperatures were on average 346 K, with the standard deviation of 1.68 K for methane, a nd 340 K with the standard deviation of 1.37 for nitrogen, respectively. The number densit y N, differential scattering cross section, VV (cm2/sr) and differential scattering coefficient, 'K(cm-1sr-1) for each gas were determined using temperatures measured each time. Number densi ties were calculated using isobaric density data tabulated by the National In stitute of Standards and Technology (Lemmon et al., 2003). As for the differential scattering cross sections for each of these gases, they were previously reported to be 4.56E-28 cm2sr-1, and 2.12E-28 cm2sr-1, respectively (Rudder and Bach, 1968) at an incident wavelength of = 694.3 nm. With those values, VV at an incident wavelength of = 532 nm was calculated using the equation below, 2 21 12 4 '' 1 21 1VVVVn n (3-3) where n is the wavelength-dependent refr active index (Landolt-Bornstein and Funktionen, 1962). The number density, the differen tial scattering cross section and coefficients for calibration gase s at the wavelength of 532 nm are summarized in Table 38.

PAGE 87

69 Table 3-8. Average of the number densities, differential scattering cross sections, and scattering coefficient sets for methane and nitrogen calibrati on gases at 1 atm with standard deviation for 24 experimental. T (K) N (cm-3) 'VV (cm2/sr) K'VV (cm-1sr-1) Average 346.29 2.12E+19 1.35E-27 2.87E-08 Methane S.D.* 1.68 1.03E+17 1.40E-10 Average 340.04 2.16E+19 6.25E-28 1.35E-08 Nitrogen S.D. 1.37 8.32E+16 5.20E-11 S.D.*: Standard deviation The reference calibration ratio Rref for methane to nitrogen can be determined by the ratio of their differential scatteri ng coefficients as expressed below, 4 2' VV CH ref VV N referenceK R K (3-4) The reference ratio, Rref, can then be related to the measured signals and the corresponding stray light through the relationship, 4 2, ,() ()vvCHmeasured ref vvNmeasuredSSL R SSL (3-5) This value was used to determine the stray light contained in the measured scattering signals from the calibration gases. In the equation, measured CH vvS, ,4 and measured N vvS, ,4 are the measured scattering signals of methane and nitr ogen, respectively. With this relationship, the stray light in the system can be quantified as SLRrefSpp,N2Spp,CH4Rref1 (3-6) In the light scattering studies, stray light calibra tion measurements were taken prior to every flame study to determine an experiment-s pecific stray light valu e. Over the range of experimental measurements, the average contribution of stray light was around 38.5%

PAGE 88

70 of the methane calibration signal. The detaile d discussion and calculati on of stray light is represented in Appendix D. 3.4.3 Signal Processing For scattering analysis, a number of si gnals were measured, including the dark signal (no laser) for calibrations, the calibrati on signals, the flame dark signals (flame only, no laser), and the flame scattering signals The scattered signal from the calibration gases and the flame from a typical scattering experiment are represented in Figure 3-9. -0.004 -0.003 -0.002 -0.001 0 0.001 1.05 10-71.1 10-71.15 10-71.2 10-71.25 10-71.3 10-71.35 10-71.4 10-7 Calibration dark Methane with 0.3 ND filter Nitrogen with 0.3 ND filter Nitrogen without filter Flame dark Flame with 5.34 attenuationPMT response (V)Time (s) Figure 3-9. Sample scattering signals from methane, nitr ogen, and flame. Calibration gases are attenuated by a factor of 100.3 and flame signal is attenuated by a factor of 105.43 for signal linearity. The dark signals were recorded prior to th e measurement of the scattered signals to normalize the baseline of the flame and the ca libration gases. Such dark signals were integrated over a 50 ns full peak width. Si nce the PMT responses of the baselines to bulk signals were not consistent at all times, they were offset by subtrac ting the signal baseline

PAGE 89

71 from the integrated scattered signal. The o ffset dark signals were subtracted from the offset scattered signals, which were named th e calculated signal. The calculated signals then needed a correction for any attenuati on factor that was used to preserve PMT linearity. For the data presented in Figure 3-9, the calculated calibration signals were attenuated by a factor of 100.3, while the calculated flame signals was attenuated by a factor of 105.34. Therefore, the calculated signal stil l including the influence of stray light was corrected by multiplying it by the overall attenuation factor. The calculated calibration signals were found to be 0.249 V-ns for methane and 0.14 V-ns for nitrogen respectively, in this example. This yielde d a calibration ratio of 1.78, 16.4% deviation from the ideal reference ratio, Rref, of 2.128 for this experiment, which indicates the magnitude of the stray light. Using Equation 3-5 with the calculated calibration signals and the reference ratio, the stray light was dete rmined to be 0.043 V-ns in this case. The average results of calibration measurements over all experiments are summarized in Table 3-9 along with the standard deviat ion. The average calibration ratio was determined to be 1.51, with a standard deviation of 0.22, 29 % deviation from Rref. Overall, the stray light signal was approxima tely 40% of the methane gas signal for the typical experiment. Table 3-9. Average results of calibration ga s signal including stra y light, a calibration ratio, stray light signal, the percentage of stray light, and the ideal reference ratio along with the standard deviati on over all scattering experiments. Methane NitrogenRatio Stray LightPercent S.L.Ideal Ratio Average 0.30 0.21 1.51 0.13 38.48 2.126 S.D. 0.05 0.06 0.22 0.07 18.53 The final calibration signals were determined by subtracting the stray light from the recorded calibration signals. This yielded true calibrati on signals of 0.206 V-ns for

PAGE 90

72 methane and 0.097 V-ns for nitrogen, respectively, for the above example. In the same manner, the calculated flame signal was determ ined to be 23,648 V-ns. The stray light signal may be subtracted from this signal to calculate the true flame signal; however it may be neglected without signifi cantly altering the scattering results because it is orders of magnitude smaller than the flame signal. At any rate, the stray li ght signal calculated in this manner will be used to determine the differential scattering coefficient of the soot with Equation 3-2. All that remains to be de termined in order to extract the differential scattering coefficient from the scattering data is the transmission of laser light through the flame, which will be discussed in a later section. 3.5 Laser Power Measurement Accurate scattered signals may be obtained as long as modest laser pulse energies are used. However, a laser pulse can significan tly heat and vaporize so ot particles if the beam is focused to a small cross-sectional area. Therefore, it is important to determine suitable laser energy to eliminate any soot vaporization during the flame studies. According to Dasch (1984a and b), la ser fluences greater than 0.2 J/cm2 from a submicrosecond pulsed source can cause vaporiz ation of soot particles, resulting in reducing the light scattering and extinction characteristics of soot by an order of magnitude. Recent work by Yoder et al. (2005) quantified vaporization effects, and reported vaporization of soot partic les down to a fluence of 0.1 J/cm2. In order to ensure no vaporization effects, the laser output pow er was measured with a powermeter and adjusted to the appropriate energy for the spot size. The laser power was 3.04 mW on average wi th standard deviation of 0.29 over 10 measurements, which yielded 0.31 mJ/pulse fo r the laser pulse rate of 10 Hz. These correspond to a laser fluence of 0.092 J/cm2 based on a focal spot of 0.0033 cm2, where

PAGE 91

73 the term fluence refers to the laser energy per the beam area. The lowest laser energy was attained by simply re ducing the laser pump energy. The beam area was found by ablating ink off a slide, at increased pulse energy, and measuring the area in which the ink was removed. The beam spot was quite a perfect circle having a diameter of 0.65 mm. A summary of the laser beam powe r properties is given in Table 3-10. Table 3-10. Summary of laser beam power properties for lig ht scattering measurements. mW mJ Fluence(J/cm2) Average 3.04 0.304 0.0916 S.D. 0.29 3.6 Transmission In order to complete the calculation of the differential scattering coefficient using Equation 3-2, the transmission is required to be measured at each of the flame heights investigated. Figure 3-10 shows the experime ntal setup for the transmission system. Figure 3-10. Top view of the transmission system setup. Power meter receiver Aperture 532 nm line filter Power meter display Beam Dump Opaque Plexiglas Black tube Plano-convex lens, f=250 mm Aperture 532 nm 45 dichroic mirror Nd:YAG 532 nm pulsed laser 532 nm Laser Flame

PAGE 92

74 This setup is identical to the scattering system except that the laser does not terminate at a beam dump but encounters th e power meter receiver. Moreover, a narrow aperture and a 532 nm line filter were placed in front of the power meter receiver. The aperture played a role to prevent the forw ard scattered light from entering the power meter. The influence of the forward scatte red light could be li mited by distancing the power meter receiver from the flame as well. These efforts resulted in only a small solid angle of observation. In addition, the 532 nm line filter blocked all light except the 532 nm that was detected by the power meter. The transmission instrume ntation is described in Table 3-11 in detail. Table 3-11. Description of transmission apparatus. Device Manufacturer Model Description Equipment 532 nm frequency doubled Nd:YAG Laser Continuum Minilite ML-11 Q-switched, 5 Hz, 2.4 mJ/pulse FWHM=5 ns, 20 mJ max PM5200 Power meter Power meter Molectron PM3 Power meter head Optics 532 nm Dichroic mirror CVI Laser 45 degree, 532 nm dichroic mirror Aperture Newport ID-1.0 Aperture Plano-convex lens Newport KBX079AR.14 BBAR coated, 430-700 nm, 25.4 mm diameter, 250 mm focal length 532 nm line filter Newport 10LF10-532 T>50%, 25.4 mm diameter The measured power of the la ser through the flame was ratioed with the power measured at a position outside of the flame to obtain th e transmission through the flame. From the Beer-Lambert law, the transmission is generally defined as

PAGE 93

75 ) exp(0L K I Iext trans (3-7) where 0I is the incident intensity of the laser, as measured at a position outside of flame, transI is the transmitted intensity measured through the flame, L is the optical pathlength through the flame, and extK is the extinction coefficient. With the transmission data determined in this manner, the extinction co efficients could be si mply obtained only if the optical pathlengths are known. The optic al pathlengths corres ponding to 12 heights were determined statistically by taking 25 pi ctures of the unseeded and seeded flames each and analyzing them. The extinction coefficients evaluated in Equation 3-7 reflect an average value through the flame, which is co rrect as used to correct th e scattering coe fficient using Equation 3-7. However, if spatially resolv ed extinction data is desired, deconvolution techniques are often employed. In this study, the width of the flame used was not sufficient to apply these techniques; as a consequence, no deconvolution (e.g. Abel inversion) was used. 3.7 Thermophoretic Sampling and Tr ansmission Electron Microscopy The conventional light scatte ring technique is a non-intr usive diagnostic tool that can effectively infer optical propertie s from a system of particulates in situ; however, it has been limited to point measurements, and cannot extract fractal properties of aggregates such as the radius of gyration and fractal dimension. Therefore, it is required to use a complementary technique that can measure the size and shape of soot agglomerates. Such informa tion can be obtained utilizing ex situ transmission electron microscopy (TEM) following thermophoretic sampling.

PAGE 94

76 3.7.1 Thermophoretic Sampling The soot aggregates were extracted from the flame using thermophoretic sampling for the transmission electron microscope image analysis of size and morphology of aggregates. Soot samples were collecte d directly on formvar-carbon coated 150-mesh copper TEM grids (Electron Microscopy Sc iences, Hatfield, PA, Part No. FCF150CU50). A schematic of this devi ce is shown in Figure 3-11. A B Figure 3-11. A setup of thermophoretic sampli ng and grid. A) Side view. B) Formvar carbon-supported 150 mesh copper grid. Each film was held by a tw eezer attached to a holder. The sampling surface faced toward the flow direction. A film was in stalled outside the flame and swept through the flame allowing soot aggregates to deposit direct ly on the grid. Samples were collected at 12 different heights along the vertical axis of both the unseeded and seeded flames. Isooctane + N2 O2

PAGE 95

77 It is significant that the exposure time of the film shoul d be long enough to capture a reliable amount of sample, but short enough to avoid melting the grid or oversampling the aggregates. The sampling times were not controlled automatically (by e.g. using a double acting pneumatic cylinder, solenoid valves and timers) but manually in the present study. As a result, each sample had a slightly different residence time in the flame, but care was taken such that all expo sure times of the films were about 1 s. In spite of the absence of an automatic cont rolling device, it was concluded that the samples obtained in the present study were sufficient for TEM an alysis. Representative micrographs are presented in Chapter 4. Another issue of thermophoretic sampling is to obtain samples at a desired position in the flame. To do this, a shield is used for preventing the part icles from depositing on the grid while the sampling probe was out of a desired sampling location. Alternatively, the shorter transition time is used while the f ilm was traveling out of the desired region. The flame employed in the present study was so thin that undesired sampling region was relatively narrow enough to ignor e the quantity of undesired sa mples. After exposed to the flame producing soot aggregates, the grid was examined to observe the particle size and morphology with a TEM. 3.7.2 Transmission Electron Microscope The soot samples collected by thermophor etic sampling were observed using a JEOL 2010F analytical electron microscope system with a poi nt resolution of 0.19 nm. Figure 3-12 shows a photograph of the TEM system. The diffraction grating was used to calibrate the TEM. Magnifications used for the present measurements ranged from 50,000 to 330,000, corresponding to 500 nm and 50 nm sized scale bars, respectivel y. For analysis, each soot aggregate was randomly picked

PAGE 96

78 at low magnification, and then analyzed at the optimum magnificati on. Each micrograph has a scale bar indicating length, and particle diameters can be simply measured from the pictures, which will be discusse d in more details in Chapter 4. TEM micrographs of soot aggregates extracted from the center region of the diffusion flame were examined as a function of increasing height. In addition to TEM analysis the Energy Dispersive X-ray Spectroscopy (EDS) was performed for precis e detection of chemical components in samples, including iron species, using an Oxford INCA Energy TEM System. Figure 3-12. Photograph of the TEM system. 3.8 Spectroscopic Techniques Up to this point, all studies have focu sed on the quantitative viewpoint such as size and concentration of soot in the flame. Next, qualitative aspects will be examined such as the identification of species in the fl ame. The particular species present in the flame provide information on the mechanis ms of chemical reactions. Moreover,

PAGE 97

79 preliminary testing for eliminating the extern al environment interferences and variations was performed with a CO flame. These were summarized below. 3.8.1 Preliminary CO Flame Study CO flame was employed to pretest an avai lability of spectroscopic techniques such as LIF and in situ Raman spectroscopy before such techniques are applied to the isooctane flame study. It is well documented that the iron pentacarbonyl gas in a CO and O2 flame is thermally decomposed and believed to form Fe2O3 chain agglomerates (Cheng et al. 1991). Further, it has an advantage that the soot particle s do not interfere with a particular spec ies to be detected because carbon dioxide is the only product from the CO flame. Figure 3-13 depicts iron pe ntacarbonyl vaporization system and a burner used for the CO flame study. Figure 3-13. Vaporization sy stem of iron pentacarbonyl and a CO flame burner. The carbon monoxide gas at a flow rate of 0.45 liter per minute regulated by an Alicat Scientific digital flow controller was introduced into the fuel additive vaporization vessel through a tube. The fuel additive va porization system was designed to seed the O.D. = 0.25 I.D = 0.14 Heater CO+Fe(CO)5 CO or N2

PAGE 98

80 CO gas with iron pentacarbonyl. To increase the concentration of the additive in the CO gas, the outside of the vapor ization vessel was heated by Omega heater tape and its temperature was measured with a stainless steel type K thermo couple. A heater was used to maintain the constant temperature of the vessel at 60 C. Figure 3-14 shows a photograph of the iron pentacarbonyl va porization vessel and the heater. Figure 3-14. A photograph of the iron pentaca rbonyl vaporization ve ssel and the heater. The CO gas was passed through the Fe(CO)5 liquid and combined with Fe(CO)5 gas before being delivered to the burner. The delivery line is also ne cessary to be heated to prevent the condensation of Fe(CO)5. Otherwise, the condensed Fe(CO)5 blocks the gas flow which results in the flame ju mping up and down. The CO flame was approximately 6 cm tall reacted with air. The addition of iron pentacarbonyl visibly

PAGE 99

81 changed the flame from the blue CO flame to th e bright orange flame. This results from the blackbody radiation of the iron oxide par ticles present in the flame. The unseeded and seeded CO flames are shown in Figure 3-15. A B Figure 3-15. Photographs of CO flame. A) unseeded flame, B) iron seeded flame. After all experiments were over, the nitrog en gas was filled in the vessel to prevent the dissociation of Fe(CO)5. 3.8.2 Experimental Apparatus of Laser Induced Fluorescence Spectroscopy Laser-induced fluorescence spectroscopy (LIF) was employed to trace Fe atomic fluorescence emission in the seeded flame. The experimental apparatus for LIF is illustrated in Figure 3-16. A frequency tripled Q-switched 355 nm Nd :YAG laser was used as a pump laser source. The laser was operated at a 10 Hz repetition rate with around 200 mJ/pulse energy. For the Fe excitation, the frequency-tripled laser out put was tuned to the several Fe resonant transitions ba nd using Optical Parametric Oscillator (OPO). An OPO

PAGE 100

82 converts photon energy of a pump laser into lower energy by means of nonlinear optical interaction; thus, the higher phot on energy of a pump laser is ve ry essential. In addition, since the gain depends on pump power, suffi cient pump power is necessary to support oscillation. In other words, the oscillati on occurs only when the pump power reaches a particular threshold level. Above threshold, the gain is also dependent on the amplitude of the resonated wave. With pump wavelengt h of 355 nm, typical OP O yields the output wavelength ranging from 400 nm and 1000 nm. Th e OPO consists of not only an optical resonator and a nonlinear optical crystal, bu t also doubleing crystal so that the output range can be broadened from 200 nm to 1000 nm. In this study, the Fe excitation wavelength was 296.69 nm that was create d by frequency-doubling of 578 nm produced by means of nonlinear optical inte raction. In spite of high pump power of the laser (~200 mJ/pulse), the output power passed all th e way through optics in OPO drops around 2 Figure 3-16. The optical setup for lase r-induced fluorescence spectroscopy. Aperature Beam Dump Flame Collection lens Beam Dump Broadband mirror Broadband mirror Aperature 1064 nm Nd:YAG Laser iCCD 355 nm razor filter Spectrometer Computer O P O

PAGE 101

83 mJ/pulse. Upon exiting the OPO, the laser beam was turned by a 45 broadband mirror (200 ~ 400 nm). The output laser beam cont ains 3 sources of li ght having different wavelengths: one is tuned wavelength of 296. 69 nm, another is residual of 355 nm, and the third is resonant source of 578 nm. Thes e are dispersed by a prism. To block the unwanted beams, a 4 square aperture was placed in the beam path so that only tuned beam could pass through it and strike to the sample. 3.8.3 Experimental Apparatus of In Situ Raman Spectroscopy In situ Raman spectroscopy was employed to fi nd the molecular state of iron oxide in the flame. A schematic of optical a pparatus is depicted in Figure 3-17. Figure 3-17. The optical set-up for in situ Raman spectroscopy. Basically, the experimental setup was very similar with the LIFs set-up except several things. First, a fre quency tripled Q-switched 355 nm Nd:YAG laser was used as a laser source. Two broadband mirrors were replaced in two 355 nm dichroic mirror. Beam Dump Aperature Beam Dump Flame Collection lens 1064 nm Nd:YAG Laser iCCD 355 nm etch filter Spectrometer Computer O P O 355 nm 45 dichroic mirror Beam Dump Pierced Mirror 355 nm 45 dichroic mirror LIBS

PAGE 102

84 Instead of using non-linear opti cal crystal in OPO, a 355 nm dichoric mirror was put into the line path making 4 more dichoric mirrors be available. Since the output source of 355mn was created through double and triple crystals with original source of 1064 nm, the output laser still include s 1064 nm and 532 nm residual light. A dichroic mirror reflects certain wavelengths while transmitti ng others; hence, the dichoric mirror is essential component to get ri d of all residual wavelengths except 355 nm. Total eight dichoric mirrors were used to eliminate those unwanted lights efficiently. A dichroic mirror allows almost 99% of 355 nm light a nd only less one percent of others to be reflected on the surface; thus, us ing 8 dichoric mirrors (i.e. two in the laser, four in the OPO, two outside) can achieve that negligible amount of residual light reaches the flame. The optical set-up inside OPO is shown in Figure 3-18. Figure 3-18. A photograph of an optical set-up inside OPO.

PAGE 103

85 Since photofragmentation of species resu lting from high laser power can be a serious issue in Raman spectroscopy, the lase r power was attenuated using an attenuator between two 355 nm dichoric mirrors inside OP O, such that moderate laser power could be obtained in this study. In addition, a le ns was placed in the beam path in order to check laser induced breakdown spectrum from iron compounds, while the layout of other optical components was maintained as illustrated in Figure 3-17. For LIF spectroscopy, a 355 nm dichoric mi rror at the right side of the attenuator was taken out, and consequently the light passed through the optical resonator and nonlinear crystal such that tuned source of light could be generated. Considering the detection of either the fl uorescence or Raman scattered signal, the light was collected at an angle 0 with respect to the axis of the excitation light using a 4 square center-pierced mirror and then focused to a fiber optic using 4 diameter plano convex lens. The light passed through a 355-nm razor edge long-pass filter before entering the fiber optic. Since the saturation of the detector is a significant issue in spectroscopic techniques, blocking elastically scattered stray light with the filter can efficiently prevent iCCD from being saturate d by the elastic light. Using the sharp-edge filter is critical in cases where the Rama n spectrum is very close to the excitation wavelength. In addition to the usage of the filter, the saturation of iCCD can be avoided by setting the grating window such that the excitation wavelength is not on the detector. The detected light transf ers through the fiber optic into a 0.275-m grating spectrometer and is recorded with an intensifie d charge coupled device (iCCD) array detector. Regarding a setting for the detecti on of Raman scattered signal as well as the LIF signal, the iCCD gate width and the delay were 200 ns and 100 ns respectively,

PAGE 104

86 which centered them tempora lly on the laser wavelength. The spectral window for LIF signal was centered on 373 nm while that for Raman scattered signal was centered on 371.1 nm. Visualization and recording of emission spec tra was achieved using an Labview program. 3.8.4 Isooctane Flame Study Experimental setups used for isooctane flame study were basically the same as that discussed previously. The optic al components and specification for the spectroscopic system apparatus are listed in Table 3-12. A concentric diffusion burner shown in Figures 3-1 and 3-2 was employed for all experimentations. Fuel was the isooctane seeded with the iron pentacarbonyl of 4000 ppm by mass of the isooctane. For LIF, twenty three heights were newly select ed and designated with the number 1 through 23. Some positions of the flame overlapped with previous positions for the light scattering experiments, but mostly new pos itions in the soot inception regime were selected for spectroscopy. Th e first height designated was located in 0.1 cm above the burner lip. All measurements were made along the centerline, at 23 different heights from the bottom to top of the flame. The su mmary of these heights are listed in Table 313. The results of all experimentation will be presented in the next chapter. In addition to LIF, absorption spectroscopy was performed using a hollow cathode lamp to verify a ny possible LIF quenching phenom enon throughout the seeded isooctane flame. A schematic of apparatus for absorption spectroscopy is depicted in Figure 3-19. Fe atomic emission from th e hollow cathode lamp was focused through the flame using a focal lens. Then, the transmitte d light was collected us ing a collection lens and transferred through the fiber optic into the spectrometer. Finally, the signal was

PAGE 105

87 Table 3-12. Components of spect roscopic system apparatus. Device Manufacturer Model Description Equipment 1064 nm frequency doubled/tripled Nd:YAG Laser Continuum PRII 8000 532/355 nm, 10Hz repetition rate, pulse energy varied Optical Parametric Oscillator Continuum Panther Ex HEO Energy: 50 % pump source Wavelength:195 nm~2800 nm Spectrometer Action Research Corporation SpectraPro275, S/n 275995S 0.275 Meter Triple Grating Spectrometer iCCD Princeton Instruments Model: 1024MLDSE/1, N119302 Intensified CCD, 200 row chip Software Labview Metal Emissions Program Fiber Optic 6 foot, high optical grade, 17 fiber bundle, 1.5 mm diameter ACTON Optics Beam dump Kentek ABD-2 Beam dump 355 nm dichroic mirror CVI Laser corporation Y3-2037-45UNP 45 degree, 355 nm dichroic mirror, 2 diameter Aperture Newport ID-1.0 aperture Square Pierced Mirror Rolyn Optics 60.2475 100 mm mm mm, center pierced-0.5 inch.2" diameter Collection Lens Comar 160-PG-100 4 Plano-convex 355 nm Razor edge filter Semrock LP01-355RU25 355 nm 99% cutoff Focusing Lens CVI Laser Coporation PLCX-50.8130.8-UV355-532 Plano-convex, 250 mm focal length, 2 diameter Broadband Mirror Newport 10D20RM.2 Flat mirror, pyrex 25.4 mm diameter 300~400

PAGE 106

88 recorded with an iCCD. The optical co mponents and specification for spectroscopic system apparatus are tabulated in Table 3-14. Table 3-13. Data collection heights for spectroscopy. Positon label Height Above Burner tip (cm) Distance between two heights (cm) 1 0.10 2 0.60 0.50 3 1.20 0.60 4 1.80 0.60 5 2.40 0.60 6 3.00 0.60 7 3.70 0.70 8 4.50 0.80 9 5.25 0.75 10 6.10 0.85 11 7.10 1.00 12 8.30 1.20 13 9.40 1.10 14 10.60 1.20 15 11.60 1.00 16 12.60 1.00 17 13.80 1.20 18 15.10 1.30 19 16.65 1.55 20 18.50 1.85 21 20.55 2.05 22 22.20 1.65 23 23.95 1.75

PAGE 107

89 Figure 3-19. The optical set-up for absorption spectroscopy. Table 3-14. Components of the system apparatus for absorption spectroscopy. Device Manufacturer Model Description Equipment Hollow Cathode Lamp Perkin-Elmer intensitron lamp 1380 Operation 7 mA Element: Fe Precision high voltage supply Stanford Research Instruments PS325 Digital high voltage power supply Spectrometer Action Research Corporation SpectraPro-275, S/n 275995S 0.275 Meter Triple Grating Spectrometer iCCD Princeton Instruments Model: 1024MLDSE/1, N119302 Intensified CCD, 200 row chip Software Labview Metal Emissions Program Fiber Optic 6 foot, high optical grade, 17 fiber bundle, 1.5 mm diameter ACTON Optics Focusing Lens CVI Laser Coporation PLCX-50.851.5-UV Plano-convex, 100 mm focal length, 2 diameter Collection Lens CVI Laser Coporation PLCX-50.851.5-UV-355 Plano-convex, 100 mm focal length, 2 diameter iCCD Spectrometer Computer Hollow Cathode Lamp Flame Collection lens Resister High Voltage supply Focal lens

PAGE 108

90 CHAPTER 4 INTEGRATED RESULTS AND DATA ANALYSIS All data and results of the experiments obt ained from smoke point study, the elastic light scattering measurement, transmission, and transmission electron microscopy with thermophoretic sampling are presented and analyz ed in this chapter. Characteristic size, number density, and volume fraction data for soot particles in the unseeded and seeded flames will be extracted using all experimental data. In addition, the results invoked from Laser-induced fluorescence and in situ Raman spectroscopy are disc ussed in this chapter. 4.1 Smoke Point Study The quantity of stoichiometric oxidizer is a measure of an amount of oxidizer needed to completely burn a quantity of fuel If the oxidizer supplied is more than the stoichiometric quantity, the flame is called a fuel lean flame, while supplying less than the stoichiometric oxidizer in the flame lead s to a fuel-rich flame. The primary purpose of this project is to investigate the effect of the metal additives on soot suppression in the flame. This can be accomplished in part thr ough comparing soot profiles of the unseeded and seeded flame. This process may be optimized by working within a stoichiometric regime with a noticeable additive effect. To accomplish this, the smoke point of the unseeded flame was investigated. The smoke poi nt is defined as the point in which the soot plume in the flame tip (i.e. visible sm oke) visually disappeared as the oxygen flow was increased. It was noted that the smoke point kept changing duri ng the early period of the experiment, due presumably to the expa nsion of rubber pumping tube leading to an increase in fuel supply, and po ssibly to burner heating. This also resulted in a noticeable

PAGE 109

91 change in the flame height. To avoid this i ssue, it was first essential to find the period that the smoke point and the flame height we re consistent. Oxygen flow rate at the observed smoke point with respect to time is shown in Figure 4-1. 0 1 2 3 4 5 6Oxygen flow rate(LPM) 5 min.10min.15min. 2 hr. average(4.24) Time Figure 4-1. Plot of oxygen fl ow rate at the smoke point as function of time. Initially, oxygen flow rate increases and it converges after about 15 minutes. After that point, the oxygen flow rate was stable at a flow rate of 4.24 LPM corresponding to the smoke point. This measurement shows that at least 15 minutes warming up is required to reach the steady fl ame condition after ignition. To explore soot suppressing effects, it is then desirabl e to run the flame at or a bove the smoke point, as running significantly below the smoke point (i.e. excess oxygen) diminishes the sooting behavior. A study of an influence on the smoke poi nt by the concentration of the metal additive was then investigated to arrive at an optimal additive concentration. Due to the

PAGE 110

92 complexity of the experimental measurements a simple parametric study was avoided. Rather, it was determined to select a singl e best condition for in depth study. The concentration of iron additive and the resul ting smoke point was evaluated using fuels doped to 10 different co ncentration of Fe(CO)5. The smoke point of each additive concentration was found in the same manner e xplained above, and four experimental data sets over a number of days were collected fo r all different concentr ations. A summary of the oxygen flow rate, the equivalence ratio, a nd oxygen to fuel ratio at the smoke point with respect to 10 different concentration of Fe(CO)5 is tabulated in Ta ble 4-1 and plotted in Figure 4-2. Table 4-1. Average of 4 oxygen flow rates wi th their standard deviation and relative standard deviation, the equivalence ratio, and oxygen to fuel ratio for 10 different concentrations. Concentration of Fe(CO)5 (ppm) Average of O2 flow rate (LPM) Standard deviationRSD (%) A/F ratio 0 4.23 0.08 1.79 0.66 18.90 500 3.74 0.17 4.48 0.75 16.71 750 3.35 0.07 2.06 0.83 14.97 1000 3.23 0.39 12.19 0.87 14.44 2000 3.16 0.17 5.51 0.89 14.12 4000 3.05 0.05 1.65 0.92 13.63 6000 3.07 0.1 3.15 0.91 13.72 8000 3.1 0.16 5.03 0.90 13.85 15000 2.96 0.18 5.94 0.94 13.23 20000 2.96 0.17 5.7 0.94 13.23 As shown in Figure 4-2, the oxygen to fu el ratio decreases rapidly at lower concentrations, and is flattened out after r eaching the minimum at the concentration of 4000 ppm. Since reduced oxygen to fuel rati o corresponds to a reduced propensity to smoke, 4000 ppm was determined to be the best value. At the zer o concentration, the

PAGE 111

93 oxygen flow rate is determined to be 4.23, which is identical to the stoichiometric quantity of oxygen flow rate obtained in the fo regoing smoke point study, as it should be. 0.5 0.6 0.7 0.8 0.9 110 11 13 15 17 19 05000100001500020000 The oxygen to fuel ratioThe oxygen to fuel ratioFe(CO)5Concentration (ppm) Figure 4-2. Smoke point, as measured by th e corresponding oxygen to fuel ratio and the equivalence ratio, as a function of iron pentacarbonyl concentra tion. Note that the equivalence ratio increases due to a reduction of the necessary oxygen quantity. 4.2 Elastic Light Scattering Results The light emitted from the la ser cavity was vertically pola rized, and only vertically polarized scattered light was captured by a polarizer at the head of the scattering collection optics. Therefore, the vertical-vertical differential scattering coefficient, K'VV was the scattering parameter of interest. This parameter was extracted from the PMT raw signal output accurately using two calibrations One is a stray light calibration using methane and nitrogen to elimin ate the extraneous signal induc ed by the reflection of laser Breakpoint

PAGE 112

94 light from various surfaces (i.e. stray light). The other is a methane calibration to relate K'VV to the signal SVV. The same methane signal was used for both calibration operations. Figure 4-3 shows a typical PMT response signa l for the calibration gases and the flame signals. -0.006 -0.005 -0.004 -0.003 -0.002 -0.001 0 0.001 2 10-84 10-86 10-88 10-81 10-71.2 10-71.4 10-71.6 10-71.8 10-7 calibration dark CH4 flame dark flame signal N2PMT response (V)time (s) two dark N2 CH4 flame signal Figure 4-3. Typical scattered signal res ponse from photomultiplier tube measuring calibration gases and flames. Calibra tion gas signals are attenuated by a factor of 100.3, and flame signals are attenuated by a factor of 105 to preserve signal linearity. In Figure 4-3, calibration gas and flame si gnals have been attenuated by a factor of 100.3 and 105, respectively, to preserve linearity in the PMT response. The baseline of each signal contains a repeatable noise signal attributed to electri cal noise stemming from the laser Q-switch discharge. The repeatab le noise signal can be removed using dark signals taken for both the cal ibration gas and the flame signal shown in Figure 4-3. These dark signals were taken prior to each measurement of both the calibration gases and the flame signals, with the la ser blocked from the detector.

PAGE 113

95 Ten experimental raw data se ts (N=10) over a number of days were collected for the conditions of both the unseeded and iron s eeded flames respectively. With the PMT raw signals, the differential scattering coefficient was calculated by the following procedures. First, intensities of the scatte red and dark signals were integrated over the signal width of 50 ns. The integrated value of the dark signal was then subtracted from that of the scattered signal. The final step of the calculation was to consider the attenuation factors of the neutra l density filters and correct th e final value for stray light. The average values of the differential scatte ring coefficient in the unseeded and seeded flames for each height are summarized in Ta ble 4-2 along with the standard deviations. Table 4-2. Average (N=10) K'VV results of the unseeded and seeded flame and standard deviations. Flame heights are m easured from the burner lip. Unseeded K'VV (cm-1sr-1) Seeded K'VV (cm-1sr-1) Height (cm) Average Standard deviation Average Standard deviation 1 3.70E-03 8.59E-04 3.98E-03 7.31E-04 3 3.87E-03 4.71E-04 2.91E-03 9.92E-04 5 3.26E-03 1.11E-03 2.67E-03 8.99E-04 7 2.97E-03 1.11E-03 3.05E-03 1.08E-03 9 3.60E-03 1.15E-03 2.68E-03 5.83E-04 11 4.10E-03 6.99E-04 3.77E-03 8.65E-04 13 3.78E-03 4.21E-04 3.13E-03 1.23E-03 15 2.76E-03 3.54E-04 2.28E-03 8.58E-04 17 1.74E-03 4.36E-04 1.40E-03 3.25E-04 19 1.04E-03 2.04E-04 1.01E-03 2.62E-04 21 5.44E-04 1.06E-04 3.95E-04 1.20E-04 25 2.44E-04 4.27E-05 6.35E-05 4.07E-05 In spite of the consistent experimental conditions, the ra w (i.e. absolute) scattered light signals at each data point had a large st andard deviation (32% RSD on average) due to laser fluctuations and change s in the quantity of the stray light from day to day. As discussed in Chapter 2, this problem was elim inated in the procedure of the differential scattering coefficient calculation using the relative scattered intensities between the

PAGE 114

96 calibration scatterer (i.e. methane) and the scattered signal from the soot particles. The absolute signal intensities vary day in, day out but still remain in relative agreement with respect to the methane calibration signal. As shown in Table 4-2, the standard deviation for the differential scattering coefficient is much less than that in the daily absolute scattering signals. Consequen tly, the scattering data were characterized as remaining relatively consistent and repeatable over all experiments, with an average relative standard deviation of 27%. The differential scattering coefficients for the unseeded and seeded flames are plotted at each height in Figure 4-4, noting th e logarithmic scale. 10-50.0001 0.001 0.01 10.112.114.116.118.120.122.124.126.1 unseeded seededK' VV(cm-1sr-1)Height (cm) Figure 4-4. Unseeded and seeded differential sc attering coefficients in logarithmic scale. Error bars represent on e standard deviation. The scattering coefficients tend to peak in both cases near about 14 cm above the surface, and then steadily decrea se with additional height (i.e increasing residence time).

PAGE 115

97 This behavior indicates that the soot profile has transitione d from the growth regime to the oxidation regime, which is characterist ic of diffusion flames. Regarding the comparison of the unseeded and the iron-seed ed flames in the growth regime, the coefficients are almost identical to both flames at these early heights. Meanwhile, the prominent deviation between two flames in scattering coefficient is shown in the soot oxidation zone. It is conclude d that the additive does not have any noticeable influence on soot growth characteristics in the soot growth regime but does in the soot oxidation regime. This is consistent with the observ ations of Masiello ( 2004), following detailed investigation of the growth regime. It is also significant to note that intens ities of both the incide nt and scattered light were attenuated due to soot particles existi ng out of the scattering volume during radial passage of light in the flame. Therefore, the values of the scattering coefficient are required to be corrected by the transmission data presented in the following section. The absolute values of the coefficient were calculated by dividing th e original values by the transmission. Only the corrected coefficients for both flames were presented in Figure 44, the uncorrected values will be found in Appendix B. 4.3 Transmission Results Six transmission measurements (N=6) for the unseeded and seeded flames were carried out at the flame he ights corresponding to the pos itions in the scattering experiments. As discussed in Chapter 3, the transmission through th e flame is described by the ratio of the laser pulse power transmitted through the flame to a reference power measured from a position outside of the flame. The summary of the transmission results for the unseeded and seeded flames is tabul ated in Table 4-3. In addition, the transmission for the unseeded and seeded flames are plotted at each height in Figure 4-5.

PAGE 116

98 Table 4-3. Average (N=6) transmission results of the unseeded and seeded flames and standard deviations. Flame heights are measured from the burner lip. Unseeded Seeded Height (cm) average Standard deviation average Standard deviation 9.4 0.78 0.02 0.77 0.03 10.1 0.75 0.02 0.77 0.03 10.9 0.75 0.04 0.75 0.03 11.7 0.72 0.02 0.73 0.03 12.7 0.73 0.04 0.75 0.06 13.7 0.73 0.04 0.70 0.03 14.9 0.74 0.03 0.76 0.04 16.2 0.75 0.05 0.79 0.04 17.75 0.84 0.07 0.88 0.01 19.6 0.88 0.05 0.94 0.02 21.85 0.93 0.04 0.96 0.02 25.25 0.94 0.02 0.97 0.02 Transmission for both unseeded and seeded flam es are greater at the higher heights, which further corroborates the soot burnout regi me due to oxidation of soot as discussed above. Furthermore, while the transmissi on between the unseeded and seeded flame shows little deviation at the lower heights, th e transmission of the seeded flame is larger than that of the unseeded flames at the higher heights. This is evidence that soot in the seeded flame was additionally reduced within the oxidation zone. In general, the transmission provides an ove rall extinction coefficient for the line of sight through the flame using th e Beer-Lambert law if the optical pathlengths of the flame are known. However, this is not the proper way to determine the extinction coefficient if the path is not homogeneous. It can corre ctly be determined using deconvolution techniques (e.g. Abel inversion) in case th e flame width is sufficiently large, but such techniques were not us ed in this study.

PAGE 117

99 0.5 0.6 0.7 0.8 0.9 1 10.112.114.116.118.120.122.124.1 Unseeded SeededTransmissionHeight (cm) Figure 4-5. Transmission through the uns eeded and seeded flames. 4.4 Soot Characteristics Determined from RDG Theory 4.4.1 Transmission Electron Microscopy Soot aggregates were deposited on elect ron microscopy grids by thermophoretic sampling at 12 various heights in the unseed ed and seeded flame. By means of a transmission electron microscope (TEM), 25 digital photographs of soot samples on each grid were taken so that meas urements could be made to de termine the fractal morphology of soot aggregates, and characte rize scattering parameters of pr imary soot particle such as size, number density and volume fraction usi ng Rayleigh-Debye-Gans (RDG) scattering theory. TEM images of typical soot aggreg ate at different height s in the unseeded and seeded flames are represented in Figure 4-6.

PAGE 118

100 A B C D E F Figure 4-6. Transmission elec tron micrographs of soot pa rticles at different axial positions.

PAGE 119

101 G H I J K L Figure 4-6. Continued

PAGE 120

102 M N O P Q R Figure 4-6. Continued

PAGE 121

103 S T U V Figure 4-6. Continued. A) H=9.4 cm, unseeded flame. B) H=9.4 cm, seeded flame. C) H=10.1 cm, unseeded flame. D) H=10.1 cm, seeded flame. E) H=10.9 cm, unseeded flame. F) H=10.9 cm, seeded flame. G) H=11.7 cm, unseeded flame. H) H=11.7 cm, seeded flame. I ) H=12.7 cm, unseeded flame. J ) H=12.7 cm, seeded flame. K) H=13.7 cm, unseeded flame. L) H=13.7 cm, seeded flame. M) H=16.2 cm, unseeded flame. N) H=14.9 cm, seeded flame. O) H=17.75 cm, unseeded flame. P) H=16.2 cm, seeded flame. Q) H=19.6 cm, unseeded flame. R) H=19.6 cm, seeded flame. S) H=21.85 cm, unseeded flame. T) H=21.85 cm, seeded flame. U) H=25.25 cm, unseeded flame. V) H=25.25 cm, seeded flame. Based on TEM analysis, the soot aggregates have a primary particle diameter ranging from 13.7 to 48.3 nm, and contain primary pa rticles ranging between about 10 and nearly

PAGE 122

104 1000 per aggregate. It is notew orthy that the primary particle diameters are generally less than 35 nm (see Figure 4-9), which yields the dimensionless size parameter no greater than 0.21 for the wavelength of 532 nm. It indicates that assumi ng individual primary particles as Rayleigh scattering particles is r easonable. In addition, soot aggregates shown in Figure 4-6 are approximately com posed of spherical primary particles, consistent with flame-generated soot. More agglomerated soot clusters were ofte n observed at a higher position; thus, the number density of primary soot particle per aggregate was greater while the size of individual particles was smaller, attributed to soot oxidation. Overall, TEM analysis has an advantage in observing the aggregate di rectly; however, there are difficulties in characterizing the size and mor phology of three-dimensional a ggregates from a projected image. It is a drawback of TEM analysis. It is also observed that the iron is concentr ated in the cores of the soot particles in TEM images for iron-seeded flame. Eviden ce presented here demonstrates that iron pentacarbonyl seeded flames yield particul ates that are soot/iron composites. Such effects were carefully investig ated. It is noted that when soot particles overlap one another, the transmission is re duced, resulting in dark regions that might be mistaken for iron-rich clusters. Therefor e, using EDS avoids such c onfusion between overlapping of particles and the inclusion of iron-rich species within the particle. Additional comments will be offered later relating to Figure 4-6 imag es as it relates to iron-cluster analysis. 4.4.2 Fractal Properties of Soot Aggregates The morphological fractal features of s oot aggregate can be characterized by a power law relationship between the number of primary partic les in an aggregate and its projected area on TEM image, as described in Equation 2-16,

PAGE 123

105 f D parfgparNkRd, (2-16) where the key parameters to fully characte rize aggregates are the fractal dimension, f D and the fractal prefactor, fk. In a TEM image, the number of primary particles per aggregate, parN was manually counted one by one, a nd the primary particle diameter, pard was measured using Measure tool in Phot oshop computer software. The number of pixels of both the particle diameter and the length of the absolute scale bar on the image were taken, and the pixel value of each particle diameter wa s then converted to absolute scale value. In the same manner, the ma ximum length of the aggregate and the normal width of it were measured to calculate the radius of gyration of the aggregate,gR, using Equation 2-18, 17 1 ) 2 ( ) (2 1gR LW. (2-18) The remaining parameters were determined using a linear regression method with a least squares approach with 25 known values of parN, pard, and gR. Two representative loglog plots of parN versus g par R d for determining these parameters are shown in Figures 4-7 and 4-8. The power law correlation is seen to provide an excellent fit of the data in Figure 4-8. Based on the correl ated equation of a least square linear fit in Figure 4-8, the fractal dimension and the fractal prefactor are determined to be 1.82 and 4.3 respectively. It is noted that these parameters are dime nsionless quantities. Because the fractal dimension directly affects dete rmining scattering parameters, more attention was paid to it rather than the fractal prefactor in this st udy. In the same manner, the fractal dimension was determined at all heights in the unseeded and seeded flames. This was summarized in Table 4-4.

PAGE 124

106 10 100 1000 110100 y = 7.4395 x^(1.7928) R= 0.91368 Log NLog Rg/d Figure 4-7. A l og-log plots of N versus Rg/dp 25 soot aggregates were sampled at the height 7 in the unseeded flame. 10 100 1000 110100 y = 4.3382 x^(1.8153) R= 0.97504 Log NLog Rg/d Figure 4-8. A l og-log plots of N versus Rg/dp 25 soot aggregates were sampled at the height 7 in the seeded flame.

PAGE 125

107 Table 4-4. The summary of the fractal dime nsion at all heights in the unseeded and seeded flames. Height (cm) Unseeded Seeded 1 1.93 1.96 3 1.82 1.75 5 1.78 1.63 7 1.79 1.82 9 1.72 1.90 11 1.93 1.64 13 -* 1.60 15 1.85 1.87 17 1.82 -* 19 1.84 1.72 21 1.73 1.99 25 1.99 1.89 *Analysis was skipped for the unseeded flame at height 13, and the seeded flame at height 17 due to the poor condition of the resulting soot samples. 25 soot aggregate samples were used to determine the fractal dimension for each height. The values fluctuate slightly becau se of the highly variant nature of soot agglomerates and limited sample size. While it is useful to explore the changes vs. height, the fractal dimension is essentially inde pendent of the various positions; thus, it is appropriate to examine an ensemble of a total of 550 aggregates on average. The statistical average of all was 1.82 along with the standard de viation of 0.11, which would be considered a more accurate value of the fractal dimension of the two flames. The fractal dimension of 1.82 agrees well with th ose obtained in other c onventional studies. The value averaged over all heights for the unseeded and seeded flames were 1.84.08 and 1.80.14, respectively, which are not statistically different. 4.4.2 Primary Soot Particle Size For each investigating position, samples of 375 soot primary particles randomly selected from 25 aggregates were used to find the mean value of the soot primary

PAGE 126

108 particles size within experimental uncertainty. From this measurement, the statistical averages of the primary soot particle diamet er at each height for the unseeded and seeded flames were calculated along with the standa rd deviation. The results are presented below in Table 4-5. In addition, Figure 49 represents the unseeded and seeded soot particle diameters graphi cally at all heights. Table 4-5. Diameters of primary soot particle at each height in the unseeded and seeded flames. Unseeded Seeded Height (cm) Average (nm) S.D. (nm) Average (nm) S.D. (nm) 9.40 34.1 9.3 27.1 5.4 10.10 29.7 8.2 25.9 6.2 10.90 32.4 5.7 29.4 5.7 11.70 28.8 4.8 28.9 6.1 12.70 31.4 5.9 28.9 3.5 13.70 32.1 5.0 32.2 2.9 14.90 31.7 4.5 31.4 5.7 16.20 31.2 4.0 29.9 3.6 17.75 30.4 4.7 28.4 5.0 19.60 26.6 4.2 26.8 6.3 21.85 24.9 5.2 24.8 4.5 25.25 23.5 5.7 20.1 4.9 *S.D. is standard deviation. Figure 4-9 shows that the primary particle size data fluctuate somewhat, especially at the lower height of flames. Such a trend is expected to result primarily from the experimental error, due to the nature of discrete aggl omerate analysis. To avoid propagation of these experimental varia tions, a smoothing routine was used. A polynomial curve fit was used to extract more accurate and consistent particle diameters from the experimental results. The curve fit had the effect of averaging over a greater number of agglomerates. Corrected data from the curve fit will be used for the calculation of volume fraction.

PAGE 127

109 0 10 20 30 40 50 60 10.112.114.116.118.120.122.124.126.1 Unseeded Seededdiameter (nm)Height (cm) Y=28.757+0.66583X-0.035825X2 for unseeded Y=7.3117+3.0508X-0.10196X2 for seeded Figure 4-9. Diameters of the primary soot partic le as a function of the flame height in the unseeded and seeded flames. A polynomi al curve fit was used for extracting more accurate values of dpar. 4.4.3 Number Density of Particles The number density of total primary soot particles in the scattering volume was determined based on the calculating procedure di scussed in Chapter 2. Note that this should not be confused with the nu mber of particles per aggregate, parN, determined above. Using the primary soot particle pa rameters in combination with RDG theory enables the average aggregate differential scat tering cross-section to be calculated. The ratio of the measured differential scatteri ng coefficient and the calculated differential scattering cross-section yields the soot aggregate number density, Nagg. Finally, the number density of primary soot particles is simply calculated by multiplying the number

PAGE 128

110 of particles per aggregate by th e number density of aggregates in the scattering volume. The summary of number densities is represented in Table 4-6. Table 4-6. The summary of number densitie s for the unseeded and iron-seeded flames. Unseeded Seeded Height (cm) Npri* Fit Npri Nagg* Ntotal* NpriFit Npri Nagg (cm-3) Ntotal (cm-3) 9.4 77 103 1.61E+091.65E+1164 99 2.60E+09 2.57E+11 10.1 108 115 1.34E+091.53E+1160 113 1.02E+09 1.16E+11 10.9 135 128 1.46E+091.86E+11116 128 1.74E+09 2.23E+11 11.7 133 139 8.12E+081.13E+11203 143 3.22E+09 4.59E+11 12.7 203 153 1.57E+092.40E+11211 159 1.70E+09 2.70E+11 13.7 164 166 1.08E+091.80E+11244 174 2.87E+09 4.99E+11 14.9 171 180 1.05E+091.89E+11286 191 1.69E+09 3.23E+11 16.2 178 194 8.06E+081.57E+11212 208 7.83E+08 1.63E+11 17.75 231 210 8.08E+081.69E+11256 226 6.10E+08 1.38E+11 19.6 346 226 5.95E+081.35E+11301 246 5.99E+08 1.47E+11 21.85 254 245 3.23E+087.91E+10223 268 2.25E+08 6.02E+10 25.25 247 269 2.14E+085.75E+10170 297 1.10E+08 3.26E+10 *Npar: the number of particles per aggregate; *Nagg(cm-3): the number density of aggregates in the scattering volume; *Ntotal(cm-3): the number density of total particles in the scattering volume. In a manner similar to what was done in determining the primary particle diameter, a logarithmic curve fit was used for extracting more accurate values of parN. Instead of using raw parN, new values from curve fitting were used to determine totalN. It is illustrated in Figure 4-10. Moreover, Figure 4-11 graphically represents totalN as a function of flame height for the unseeded and iron-seeded flames. 4.4.4 Volume Fraction of Soot Particle With the primary soot particle diameter and number density of total particle determined above, the overall soot volume fraction is calculated from Equation 4-1, agg agg vN V f (4-1)

PAGE 129

111 10 100 1000 10.112.114.116.118.120.122.124.126.1 unseeded seeded y = -351.39 + 462.04log(x) R= 0.85104 y = -218.98 + 355.54log(x) R= 0.6114 The number of primary particle per aggregateHeight (cm) Figure 4-10. The number of primar y soot particle as a function of the flame height in the unseeded and iron-seeded flames. A logarithmic curve fit was used for extracting more accurate values of Npar. 10101011101210.112.114.116.118.120.122.124.126.1 unseeded seededTotal number densitiy (particles/cm3)Height (cm) Figure 4-11. Number density of the total soot particle for the unseeded and iron-seeded flames.

PAGE 130

112 where the volume of an aggregate, aggV is given by par p aggN d V36 (4-2) Finally, the volume fraction which is represen tative of the total soot loading, as a function of flame position is compiled in Table 4-7 with errors determined through uncertainty analysis and graphically presen ted in Figure 4-12 for the unseeded and ironseeded flames. The detailed results and disc ussion of the analysis will be found in Appendix C. Table 4-7. The volume fraction as a function of flame height for the unseeded and ironseeded flames. Height (cm) Unseeded (cm3 soot/cm3) S.D.* (cm3 soot/cm3) Seeded (cm3 soot/cm3) S.D. (cm3 soot/cm3) % Reduction 9.4 2.78E-06 3.75E-06 2.64E-06 3.53E-06 5.0 10.1 2.58E-06 3.43E-06 1.29E-06 1.82E-06 50.0 10.9 3.11E-06 4.24E-06 2.70E-06 3.68E-06 13.2 11.7 1.87E-06 2.65E-06 5.91E-06 8.14E-06 -216.0 12.7 3.89E-06 5.42E-06 3.69E-06 4.97E-06 5.1 13.7 2.85E-06 3.85E-06 7.06E-06 9.79E-06 -147.7 14.9 2.87E-06 3.45E-06 4.64E-06 6.75E-06 -61.7 16.2 2.24E-06 3.06E-06 2.30E-06 3.16E-06 -2.7 17.75 2.23E-06 3.01E-06 1.83E-06 2.74E-06 17.9 19.6 1.55E-06 2.16E-06 1.69E-06 2.32E-06 -9.0 21.85 7.43E-07 1.00E-06 5.08E-07 6.92E-07 31.6 25.25 3.52E-07 4.78E-07 1.22E-07 1.82E-07 65.3 S.D.: Standard deviation 4.4.5 The Extinction Coefficient of Soot Particle In addition to the scatteri ng analysis, the extinction coefficient was determined from ext extN K (2-12)

PAGE 131

113 The extinction coefficient as a function of flame height is summa rized in Table 4-8 and graphically presented in Figure 4-13 for the unseeded and iron-seeded flames. 10-810-710-610-50.0001 10.112.114.116.118.120.122.124.126.1 unseeded seededVolume fraction (cm3/cm3)Height (cm) Figure 4-12. The volume fraction as a function of flame height for the unseeded and ironseeded flames. The error bar represents one standard deviation. Table 4-8. The extinction coefficient of soot particle as function of flame height for the unseeded and iron seeded flames. Height (cm) Unseeded (cm-1) Seeded (cm-1) 9.4 1.36E-01 1.29E-01 10.1 1.29E-01 6.72E-02 10.9 1.48E-01 1.26E-01 11.7 9.47E-02 2.61E-01 12.7 1.82E-01 1.68E-01 13.7 1.42E-01 3.13E-01 14.9 1.41E-01 2.10E-01 16.2 1.09E-01 1.09E-01 17.75 1.02E-01 8.39E-02 19.6 7.03E-02 7.56E-02 21.85 3.39E-02 2.33E-02 25.25 1.59E-02 5.37E-03

PAGE 132

114 0.001 0.01 0.1 1 10.112.114.116.118.120.122.124.1 unseeded seededKext(cm-1)Height (cm) Figure 4-13. The extinction coefficient of soot particle as a function of flame height for the unseeded and iron-seeded flames. 4.4.6 Discussion of Results In comparing all results for diameter number density, volume fraction, and extinction coefficient shown in Figures from 4-9 to 4-13, it is seen that all data have the same trend, within experimental uncertainty, and the data at the final position are always found to be greater in the unseeded flam e than in the iron seeded flame. Data can be specified in terms of two regimes regarding the comparison of the unseeded and iron-seeded flames, namely, the growth regime ( ~14 cm) and the oxidation regime ( ~14 cm). No distinct trends betw een the two flames were observed in the lower growth regime, below about 14 cm. This indicates that the iron additive has little effect on suppression of soot growth in this regime. As noted above, this is consistent with the work of Masiello (2004), in which detailed meas urements within the

PAGE 133

115 inception and growth led to the conclusion that iron particulates are quickly incorporated within soot agglomerates, giving way to t ypical soot growth. The inconsistent perturbation of the number density in this region, especially for the seeded flame in Figure 4-11, is linked to the propagation of error, including the differential scattering coefficient data and TEM data, as shown in Figur e 4-4. This perturbation directly affects volume fraction and the extinction coefficient data. The main error of measured data such as the differential scattering coefficient and transmission results from unsteady light signal caused by spatial excursions of the thin reaction zone. On the other hand, the iron-seeded flame data present a consis tent change with respect to the unseeded flames in the higher re gime (i.e. oxidation regime). In particular, the maximum deviation between the two flames is shown at the last height, which is near the flame tip, where the volume fraction is re duced by nearly 66%. In Figure 4-9, the average primary soot particle diameters at the lower heights, from 9.4 to 11.7 cm above burner, were 31 and 29 nm in the unseeded and seeded flames, respectively. The relative standard deviations were about 6.5% for these data; hence, the difference is on the order of the experimental uncertainty. In contrast at the greatest resi dence time investigated (H = 25 cm), the primary particle diamet ers were reduced to 24 and 20 nm in the unseeded and iron-seeded flames, respectively. Such a decrease in particle size is consistent with significant soot oxidation. Overall, the primary particle diameters and volume fraction data, as measured from th e TEM micrographs and scattering signal, depict the transition from soot growth to soot oxidation al ong the length of the flame. To explore the overall differences between the conditions of two flames, it is useful to examine the total soot volume fractions in detail.

PAGE 134

116 In Figure 4-12, the soot volume fraction is reduced from 3.52E-7 in the unseeded flame to a value of 1.22E-7 in the iron-seed ed flame, corresponding to a 66% reduction in soot emissions. This was visually obse rved as well. A pronounced smoke plume escaping from the flame tip disappeared when the iron seeded fuel was supplied, which was recorded with a video camera. Figur e 4-14 shows both the unseeded and seeded flames. A B Figure 4-14. Photographs of tips of the unseeded and seeded flames. Soot plume is seen in the unseeded flame while being not s een in seeded flame. A) the unseeded flame. B) the seeded flame. As evidenced above, it can be concluded that the iron additive has a significant role on soot reduction that takes pl ace in the oxidation regime, as evidenced by the data in Figures 4-9 and 4-12. With measured optical length of the flam e and the transmission data determined above, the extinction coefficient for the opt ical length can be calculated from BeerLambert law. Overall, the absolute values of it were about 43% higher than the extinction coefficient presented in Figure 4-13; however, the relative trend of two results

PAGE 135

117 is in good agreement. The agreement between these two values is excellent when one considers the uncertainty in the pathlength, spatial va riations along the path, and uncertainties in the optical properties. The detailed results of the analysis will be found in Appendix B. The effect of iron pentacarbonyl on the visi ble structure of the flame is shown in Figure 4-15. Iron pentacarbonyl addition is characterized by a decrease in the dark orange zone in the flame tip, and vague lateral outline of the flame. It may be attributed to the soot oxidation, but qualitative diagnos tics are required for mo re precise analysis. A B Figure 4-15. Photographs of the unseeded and seeded flames. A) the unseeded flame. B) the seeded flame. All analysis was carried out assuming a c onstant value for the complex index of refraction of soot. A value of m = 2.0-0.35 i was used for this study. Wide variations in the complex refractive index of soot have b een reported in the literature, and no single

PAGE 136

118 value has been generally accepted. The findings for m from key studies are listed in Table 4-9. Table 4-9. Complex refractive indices for soot from various sources. (2001) m = n ki Authors Type of soot Incident Wavelength (nm) n k Chippett and Gray (1978) acetylene visible 1.9-2.0 0.350.50 457.9 1.581.82 0.650.83 488 1.571.82 0.650.85 propane 514.5 1.541.71 0.670.87 Charalampopoulos and Chang (1988) soot 540 1.77 0.63 435.8 1.57 0.46 550 1.57 0.53 propane 650 1.56 0.52 Dalzell and Sarofim (1969) Standard value visible 1.57 0.56 435.8 1.56 0.46 550 1.56 0.46 acetylene 650 1.57 0.44 carbon 1.6-1.8 0.060.19 Pluchino et al. (1980) 8 um carbon sphere 488 1.7 0.7 Roessler and Faxvog Mean of review 515 1.75 0.5 Lee and Tien soot 633 1.8-2.0 0.450.65 Sloane soot 633 1.7 0.8 Mullins and Williams soot 633 1.85 0.4 Stagg and Charalampopoulos soot 633 1.53 0.38 Kyl and Faeth soot 515 1.54 0.48 Mulholland and Choi soot specific mass extinction 1.55 0.8

PAGE 137

119 The value chosen for this study is the one reported by Chippett and Gray (1978), which was based on a similar light scatteri ng analysis with a heavily sooting fuel. According to a study by Dobbins a nd Megaridis (1991), a value of m = 1.57-0.56i is typically recommended for car bonaceous soot when the fractal dimensions are ranged between 1.7 and 1.9. The same analysis with this refractive index will be presented in Appendix B. Overall, the conclusions are id entical for the alternative refractive index, although absolute values change by about 23%. 4.5 Spectroscopy This study has focused so far on the quantit ative points of view to find how much soot can be reduced by adding the additive. However, the additional information is required to elucidate how the soot can be re duced, and what mechanisms contribute to the reduction of soot. In order to determine the best configuration for species detection using spectroscopic methods, preliminary sets of experiments were conducted with the CO flame. The advantage for using the CO flame is to prevent the desired signal from interfering with soot particles. As stated previously, the intensity of Raman scattering is very weak; thus, a sensitive de tection configuration is ideal. In other words, evaluating the detection ability of the de sired species in the CO flame with spectroscopic techniques is the initial focus for a practical impl ementation in the isooctane flame. As done previously, products in the CO flame were collected via thermophoretic sampling at three different flame heights, recorded using TEM, and analyzed with EDS. Figure 4-16 represents TEM images of samp les from the CO flame. The shape of individual particle is hexa gonal, which is distinguishabl e from soot particles.

PAGE 138

120 A B Figure 4-16. TEM images of samples collected in the Fe-seeded CO flame. A) sampled at the middle of the flame height. B) sampled at the flame tip. It is well documented that the ir on pentacarbonyl gas in a CO and O2 flame is thermally decomposed, and presumably forms in Fe2O3 chain agglomerates (Cheng et al. 1991). EDS can be helpful for identifying th ese species more correctly. The typical signal window of EDS is shown in Figure 4-17. Figure 4-17. The typical signal window of EDS.

PAGE 139

121 The intensity of x-ray emission from el ements was represented at each energy level. It was also noted that Cu signal wa s attributed to the sampling grid. Table 4-10 summarizes the atomic ratio of iron and oxygen in the particles from EDS. Table 4-10. EDS result atomic ratio of iron oxide. Sample Flame Fe O ratio 1 24.7 54.34 1:2.20 2 4.97 20.17 1:4.06 3 4.71 7.7 1:1.63 4 5.57 0 5 4.38 0 6 10.09 16.72 1:1.66 7 4.96 2.8 1:0.56 8 0 2.14 9 34.27 0 10 Isooctane flame 22.67 0 11 26.94 50.49 1:1.87 12 25 49.34 1:1.97 13 37.36 36.4 1:0.97 14 26.31 32.46 1:1.23 15 26.01 40.45 1:1.56 16 14.97 26.15 1:1.75 17 25.86 30.51 1:1.18 18 6.93 20.19 1:2.91 19 18.43 40.02 1:2.17 20 22.89 49.98 1:2.18 21 CO flame 17.11 38.37 1:2.24 The ratio is quite random; therefore, it is very hard to explain that a particular formation of iron oxide (i.e. especially Fe2O3) is dominantly presen t in the CO seeded flame. As discussed later, iron oxides were likely formed after sampling was done. Since this analysis is an ex situ method, the sample may be affected by the oxygen post sampling process. The use of in situ analytical methods yields more real and accurate information on the chemical states.

PAGE 140

122 4.5.1 Laser Induced Fluorescence (LIF) Spectroscopy Laser induced fluorescence spectroscopy was us ed to detect the concentration of Fe along the axial direction of the seeded flame, which would give information on the role of the additive for the soot reduction. In order to find the strong resonance transition lines of Fe, the tunable laser was scanned across the particular wavele ngth ranging from 295 nm and 298 nm. During the scanning, laser induced fluorescence of Fe was found at three excitation wavelengths, which are list ed in Table 4-11, with fluorescence lines corresponding to each excitation wavelength. Table 4-11. Fe resonance transition wavelengths a nd corresponding fluorescence emission lines with their relative intensity. Wavelength (nm) Excitation Emission Relative Intensity 372.76 0.10 376.38 0.27 1 295.39 378.79 0.05 2 296.69 373.49 1.00 374.95 0.60 3 297.31 375.82 0.29 This can be also displayed in an energy level diagram as represented in Figure 418. In addition, Figure 4-19 describes th e laser induced fluorescence peaks for three excitation sources, all recorded within the CO flame (two third from the burner lip). These peaks are in agreement with the refe rence Fe atomic wavelength and relative peak intensity in the NIST database. The strongest fluorescence at 373.49 nm was obtained with the resonance transition of 296. 69 nm, which was chosen for all further experimentation. To ensure resonant exci tation, daily spectral calibration was necessary to ensure the correct output laser wavelength. Prior to all experiments, the laser was tuned into the resonance transi tion of 296.69 nm by stepping the laser in small increments

PAGE 141

123 across the transition line until the exact re sonance transition was calibrated using the spectrometer. First, Fe fluorescence was measured as a function of the normalized CO flame heights, namely, zero, 1/2, 2/3, and 1 (i.e. flame tip). Typical LIF signals are shown in Figure 4-20. As discussed previously, th e fluorescence signal is proportional to the concentration of species, temperature and pre ssure. Herein, the goal of LIF was to trace the Fe concentration in the flame. Sin ce CO flame is diffusion flame, the flame temperature would increase toward downstream. In spite of increase in temperature, the intensity of Fe fluorescence also decr eases toward downstream of the flame This may Figure 4-18. Energy level diagram of Fe atom. Bold font indicates the best combination. E=8154.710 cm-1 E=7985.780 cm-1 E=7728.056 cm-1 E=7376.760 cm-1 E=6928.266 cm-1 373.49nm 374.95nm 372.76nm 376.38nm 378.79nm 295.39nm 296.69nm 297.31nm 375.82nm E= 34547.207 cm-1 E= 34328.749 cm-1 E= 34039.513 cm-1 E= 33695.394 cm-1 E= 704.004 cm-1 E= 0 cm-1

PAGE 142

124 0 1000 2000 3000 4000 5000 6000 7000 8000 370372374376378380 295.39 nm 296.69 nm 297.31 nmIntensityWavelength (nm) Figure 4-19. Laser induced fluorescence peak fo r three excitation sour ces at the two third of normalized CO seeded flame height. Excitation lines are shown. 0 2000 4000 6000 8000 10000 12000 14000 16000 372372.5373373.5374374.5375375.5376 Height 0 of the flame Height 1/2 of the flame Height 2/3 of the flame tip of the flame the gas onlyIntensity (a.u.)Wavelength (nm) Figure 4-20. Fe fluorescence corresponding to the excitation line of 296.69 nm as a function of the CO flame normalized four heights. The gas only is the Fe fluorescence from the unreacted CO gas seeded with Fe(CO)5.

PAGE 143

125 illustrate that the concentration of Fe is reduced throughout the flame due to a certain chemical reaction, namely iron oxide formati on. It is noteworthy that Fe fluorescence was observed from the CO gas seeded with iron pentacarbonyl (i.e. non-reacting). Iron pentacarbonyl is very easily dissociated in even ambient temperature, so Fe atomic intensity is quite strong and even stronger than the signal obtained from the flame. The sensitive LIF Fe probe was then used to investigate the isooctane flame seeded with iron pentacarbonyl. Six measurements were performed at 23 different flame heights, and intensities of the signals were calculated. The average values of intensities of the Fe emission line at 373. 49 nm (integrated peak) for eac h height are presented in Figure 4-21. 0.1 1 10 100 1000 0510152025Relative IntensityHeight (cm) soot oxidation soot growth Flame tip Figure 4-21. Intensity of LIF measured in isooctane seeded flame at emission line of 373.49 nm corresponding to th e excitation line of 296.69 nm. Flame tip is at height of 23.95 cm. Error bars re present one standard deviation.

PAGE 144

126 The LIF data were normalized by the mean value of the initial five data points in order to minimize the large standard deviati on attributed to the varying OPO laser power from day to day. The LIF signal is higher ne ar the burner lip and is observed to decrease as the height increases. However, the Fe LIF signal is observed to undergo a more rapid drop with the oxidation regime, falling by mo re than 2 orders of magnitude. As discussed in Ch2, quenching is a common issu e in the LIF spectroscopy. Such a rapid change in quenching is unlikely to have occu rred in the flame, but any possibility of quenching was explored by absorption spectros copy using a hollow cathode lamp. While LIF is subject to quenching, the absorption sp ectroscopy is free from this phenomenon. The Fe atomic emission from the hollow cathode lamp passing through the Fe seeded flame was measured along the ve rtical axis of the flame. Then, the emission from the flame itself was subtracted from the overall emission of both the flame and the lamp. Finally, transmission was determined by divi ding the lamp signal excluding the flame signal by pure lamp emission. Since the Fe at omic emission is absorbed by the elemental Fe present in the seeded flame, the transmi ssion is low at Fe res onant transition lines. Two transmission measurements at two different heights are represen ted in Figure 4-22. A strong Fe transition line of 271.9 nm was c hosen for this study. This transition originated from the ground state; thus, the signal was more sensitive to absorption. Peak values at the wavelength of 271.9 nm were plot ted as a function of 5 different heights of the seeded flame in Figure 4-23. The amount of the elemental Fe is high in the soot growth regime; therefore, the Fe atomic photons are absorbed by the elemental Fe resulting in the low transmission of the light. However, the transmission of the light was high in the soot oxidation regime because the co ncentration in the elemental Fe is low.

PAGE 145

127 0 0.5 1 1.5 2 266268270272274276278 Flame height of 10 cm Flame tipTransmissionWavelength (nm) Figure 4-22. Transmission of Fe atomic light passing through the seeded flame at two different heights. Fe resonance trans ition line of 271.9 nm was chosen for this study. Flame tip is at the height of 23.95 cm. 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 0510152025TransmissionHeight (cm) Oxidation regime Flame tip Figure 4-23. Transmission of Fe atomic light from the lamp as a function of 5 different heights of the seeded flame. Fe re sonance transition line of 271.9 nm was chosen for this study. Flame tip is at the height of 23.95 cm.

PAGE 146

128 The transmission data was observed to be ne arly constant in the growth regime and than rapidly rise up throughout the oxidation regime. This trend is in an excellent agreement with the LIF result. Therefore, the absorption spectros copy ensured that the quenching was not responsible for Fe LIF signa l drop in the oxidation regime. The more possible explanations on the Fe LIF signal drop throughout the oxidation regime will be discussed in Chapter 6. Since the LIF signal could be influenced by laser-induced plasma excited atomic emission as discussed below, the LIF signal mu st be discriminated from such a signal, especially at the near flame tip. Spectra were measured as function of the incident laser power at the flame tip using the 355 nm s ource in order to validate the laser-induced plasma effect on the LIF signal. This is shown in Figure 4-24. Figure 4-24 elucidates that laser induced plasma (LIP) emission starts emerging when the laser energy is over 25 mJ/pulse. Because only 1.8 mJ/pulse was used for the LIF experiments, the laser power used for LIF was low enough not to be concerned about laser-induced plasma formati on and induced iron emission. A final test to ensure the validity of the LIF probe was made by tuning the OPO on and off the 296.69 nm transition line, while keeping the total laser energy constant. Offline measurements were taken at flame heights 7, 15, 21, and 25 with 296.19 nm to examine the presence of LIBS emission. Fi gure 4-25 represents the on and off resonant LIF signal induced by the excitation wavelengths of 296.69 nm and 296.19 nm. As shown in Figure 4-25, no signal was obs erved when the laser wavelength was tuned to 296.19 nm. Therefore, the LIF probe was concluded to accurately measure elemental Fe without perturbing the overall fl ame species, including any iron species.

PAGE 147

129 50 100 150 200 250 300 350 400 360362364366368370372374376 160 mJ at tip 95 mJ at tip 54 mJ at tip 25 mJ at tip 18 mJ at tip 10 mJ at tip 2 mJ at tip LIFIntensityWavelength (nm) Fe LIF O2 Raman Figure 4-24. Spectra measured as function of the incident laser energy at flame tip using the 355 nm source in order to validate the LIBS effect on the LIF signal. 50 100 150 200 250 371372373374375376 296.69 nm 296.19 nmIntensity (a.u.)wavelength (nm) Figure 4-25. On-and-off resona nt LIF signal induced by the excitation wavelength of 296.69 nm and 296.19 nm with the same pulse energy.

PAGE 148

130 Finally, LIF was also used for exploring th e presence of FeO in the flame. Several FeO resonant transition wa velengths (Mavrodineanu et al., 1965) used in this study are diagramed with corresponding emission wavelengt hs in Figure 4-26. However, no LIF of FeO was detected at transition wa velengths shown in Figure 4-26. Figure 4-26. Energy level di agram of FeO molecule. It is concluded that FeO ma y not be significantly present as an iron species in the flame configuration for this study. Alt hough FeO may exist in the flame, the amount would be below the detection limit. 4.5.2 In Situ Raman Spectroscopy In situ Raman spectroscopy was employed to ex tract the molecular information of iron oxides which may form throughout the flame. Initially, the CO flame was investigated for this purpose at 532 nm a nd 355 nm excitation wavelengths. According to the results from an ex situ study (Masiello, 2004), Fe2O3 was the most possible formation of iron oxide in the Fe-seeded flam e. However, such data was not recorded within the flame. Literature refere nce to the iron oxide Raman shifts (cm-1) are summarized in Table 4-12 (Faria et al., 1997 and Maslar et al., 2000). E=18,57 E=17,80 E=2,580 E=1,730 E=870 564.66 nm 590.30 nm 621.89 nm 627.89 nm 593.48 nm

PAGE 149

131 Table 4-12. Reference to iron oxides Raman shift (cm-1) FeO Fe2O3 Fe3O4 Faria et al. Maslar et al. Faria et al. Maslar et al. Faria et al. 224 225 243 247 290 293 292 300 406 412 494 498 524 532 607 613 652* 655 665 661 813 1055 1100 1310 1320 *Bold font means the strongest line. As discussed in Chapter 2, the Raman si gnal is proportional to the inverse fourth power of the excitation wavelength; however the background fluorescence also increases with decreasing wavelength, notably in the UV, causing increases in noise level. Another issue to be considered for the use of lowe r wavelength as an excitation source is the separation of Raman signal from the elastic scattering signal. Although the razor edge filter can sharply cut the elastic scattering light off, it is very hard to invoke the first Fe2O3 Raman peak when the 355 nm excitation laser was employed. Therefore, using 532 nm and 355 nm involves a trade-off betw een those issues. Figure 4-27 shows a spectrum obtained from in situ Raman experiment of CO flame using 532 nm as an excitation source with the laser pulse energy of 11 mJ/pulse. It is not clear that the spectrum show n in the Figure 4-25 represents the Fe2O3 Raman spectrum. Furthermore, neither FeO nor Fe3O4 Raman peaks are observed in the spectrum. Rather, the observed peaks correspond to Fe atom ic emission lines, which are listed in Table 4-13.

PAGE 150

132 50 100 150 200 250 300 535540545550555 200300400500600700800Intensity (a.u.)Wavelength (nm) Raman Shift (cm-1) 1 2 3 4 5 6 Figure 4-27. A spectrum obtained from in situ Raman experiment of CO flame using 532 nm as an excitation source. Table 4-13. Fe atomic emission peaks. Peak Label Wavelength (nm) Element 1 537.15 Fe I 2 539.32 Fe I 3 542.97 Fe I 4 544.69 Fe I 5 545.56 Fe I 6 557.70 Fe I In the same manner, the more intense study was achieved using 355 nm excitation laser. Several spectra taken at four norma lized CO flame heights are shown in Figure 428.

PAGE 151

133 0 500 1000 1500 2000 2500 3000 360364368372376 200400600800100012001400 Height 0 Height 1/2 Height 2/3 Height tipIntensity (a.u.)Wavelength (nm) Raman Shift (cm-1) Figure 4-28. Spectra obtained from in situ Raman experiment of CO flame using 355 nm as an excitation source at four different heights. Similar with the spectrum excited in 532 nm, peaks observed in Figure 4-28 may be Fe atomic emission peaks rather than possible iron oxide Raman peaks. For peak identification, a steel rod was examined using an increased laser fluence (achieved by focusing) to verify the Fe atomic emissi on lines produced by a la ser induced plasma. Not only was the pulse energy increased, but also a focal lens was positioned in a manner to generate a breakdown on the surface of th e steel rod. The same manner for data collection was used for LIBS. Figure 4-29 show s the atomic emissions from steel rod. Each peak is summarized in Table 4-14.

PAGE 152

134 0 1000 2000 3000 4000 5000 360364368372376 LIBS signalIntensity (a.u)wavelength (nm) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 Figure 4-29. LIBS emission spectrum obtai ned from steel rod using 355 nm as an excitation source. Table 4-14. LIBS emission peaks. Peak Label Wavelength (nm) Element 1 358.67 2 359.48 3 360.89 Fe I 4 361.88 Fe I 5 363.15 Fe I 6 364.78 Fe I 7 366.81 8 368.57 9 370.56 Fe I 10 371.99 Fe I 11 372.55 12 373.49 Fe I 13 374.56 Fe I 14 375.55 Fe I The profile of spectrum from the CO seeded flame is quite identical to that from steel rod; therefore, the spectrum obtained from in situ Raman spectroscopy is not iron oxide Raman signal but Fe atomic emi ssion most likely cause d by laser-induced

PAGE 153

135 breakdown. However, this result may be at tributed to breakdown of the iron oxides inside of the CO flame due to higher pulse energy. The effect of the sample breakdown must be very carefully considered during the Raman experiment. In order to ensure this, the laser energy was attenuated to find the optimum power sufficiently high to detect the iron oxide Raman signal, but simultaneously lo w to avoid breakdown of the sample. For the CO flame, no Fe signals were detected below the pulse energy of 10 mJ/pulse. Unfortunately, no Raman signals corresponding to iron oxides were noted. Additionally, it was noted in Figure 4-28 that the intensity of detected signal decreases as the flame height increases. This will be discussed later. The same experiment was then carried out using the isooctane iron seeded flame, and the similar trend was observed. In this experiment, Fe concentration is lower than that in CO flame; thus, the higher laser pulse energy still did not pr ovide a strong signal. As depicted in Figure 4-24, neither Fe nor iron oxides spectra were observed when the pulse energy was lower th an 25 mJ/pulse. Since O2 is present along the beam path as well as it was supplied in the fl ame, relatively strong Raman O2 signal was detected at Raman shift of 1543 cm-1. In short, Fe atomic emission and pronoun ced Fe LIF signals were observed using in situ LIF in this study. This is not consistent with the results from ex situ studies that Fe2O3 was the dominant species created in iron pentacarbonly seeded flame. However, a recent similar study accomplished by Kim et al. (2005) supports the re sult of this study. They analyzed the products from Fe(CO)5 seeded ethylene diffusion flame using a Laser Microprobe Mass Spectrometry, and found that Fe dominated over iron oxides such as FeO and Fe2O3. The detailed discussion will be presented in Chapter 6.

PAGE 154

136 CHAPTER 5 NUMERICAL ANALYSIS 5.1 Thermodynamic Equilibrium Calculations Thermodynamic equilibrium calculations were carried out using STANJAN code (3.89 version, Department of Mechanical Engineering, Stanford University, Reynolds 1990) so that mass fraction of products from the isooctane flame seeded with iron pentacarbonyl could be nume rically evaluated on seve ral given conditions: flame temperatures, O2 flow rates, and Fe(CO)5 concentrations. 5.1.1 Flame Temperature First of all, moles of each reactant were determined for the calculations using STANJAN code. They were calculated based on the quantities of reactants supplied in the seeded flame in the experiment. The calculations for each reactant are following below, N2: 3 3 5 31350 0.8100.96251.50810 60298mminKKgKg minsecKmsec (5-1) 541 1.508105.39310 28Kgmolmol secgsec, (5-2) O2: 3 35 31350 2.6101.15.610 60298mminKKgKg minsecKmsec, (5-3) 531 5.6101.7510 32Kgmolmol secgsec, (5-4) where the densities of N2 and O2 are 0.9625 and 1.1 3Kgmrespectively at the temperature of 350 K. For isooctane,

PAGE 155

137 C8H18: 3 65 31 1.5107031.757510 60mminKgKg minsecmsec, (5-5) 541 1.7575101.53910 114.23Kgmolmol secgsec, (5-6) where the density of C8H18 is 703 3Kgmat the temperature of 298 K. The mass flow rate is always consta nt. Since 1 g of C8H18 has 4 1000 g of Fe(CO)5, 584 1.7575107.0310 1000KgKg secsec, (5-7) 871 7.03103.5910 195.845Kgmolmol secgsec. (5-8) Moles of reactants used for calcu lation are tabulated in Table 5-1. Although the flame is diffusion flame, the air was excluded in reactant s to simplify the calculation. This might be a serious limitation, but nonetheless, the current results provide a starting point for theoretical analysis. Table 5-1. Mole of reactants used for input in the STANJAN code. N2 5393 O2 17500 C8H18 1539 Fe(CO)5 3.59 In the simulation, the following species considered to be equilibrium products were summarized in Table 5-2. Table 5-2. Products from STANJAN simulation. Name of Species Reactant C8H18, Fe(CO)5, O2, N2 phase 1 C(g*), CH4, CO, CO2, C3H8, C8H18, Fe(CO)5(g), Fe(OH)2(g), Fe (g), FeO(g), H, HO, H2, H2O, N, NO, NO2, N2, O, O2 Products phase 2 C(s*), Fe(CO)5(l*), Fe(OH)2(s), Fe(OH)3(s), Fe(a), Fe(l), Fe(r), FeCO3(s), FeO(l), FeO(s), Fe2O3(s), Fe3O4(s), Fe4N(s), H2O(l) *g:gas; l:liquid; s: solid; r:radical

PAGE 156

138 First, the mass fractions of all products were presented as a function of the adiabatic temperature. Figur es 5-1, 5-2 and 5-3 represen t the relative mass fraction of products obtained over the temperature ra nging from 400 K to 2000 K at atmospheric pressure. This is in good agreemen t with the result obtained by Yang (2004). 0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 0.80 0.90 1.00 400600800100012001400160018002000 Temperature (K)Mass fraction of species C CH4 CO CO2 C3H8 C8H18 H HO H2 H2O N NO NO2 N2 O O2 C(S) H2O(L) Figure 5-1. Relative mass fraction of pr oducts as a function of temperature. For comparison, the flame temperatures were measured with a type K thermocouple along the axial direction of th e flame. Due to radiation loss, the temperatures measured with the thermocoupl e should be corrected using the following equation, 44 0()()ftctchTTTT (5-9) where his the convection heat transfer coefficient, is the emissivity of the thermocouple, is the Stefan-Boltzman constant, f T is the temperature of the flame, tcT

PAGE 157

139 0.00E+00 5.00E-05 1.00E-04 1.50E-04 2.00E-04 2.50E-04 3.00E-04 3.50E-04 4.00E-04 4.50E-04 5.00E-04 400600800100012001400160018002000 Temperature (K)Mass fraction of species Fe(CO)5g Fe(OH)2g Fe(g) FeO(g) Fe(CO)5l Fe(OH)2s Fe(OH)3s Fe(a) Fe(l) Fe(r) FeCO3(s) FeO(l) FeO(s) Fe2O3(s) Fe3O4(s) Fe4N(s) Figure 5-2. Relative mass fraction of Fe species as a function of temperature. 0.00E+00 1.00E-06 2.00E-06 3.00E-06 4.00E-06 5.00E-06 6.00E-06 7.00E-06 8.00E-06 400600800100012001400160018002000 Temperature (K)Mass fraction of species Fe(g) Fe(a) Fe(l) Fe(r) Figure 5-3. Relative mass fraction of Fe as a function of temperature. is the temperature of the thermocouple, and 0T is the ambient temperature. An emissivity of thermocouple was selected in 0.3 based on a report of the Greene et al. (2000). A

PAGE 158

140 convection coefficient was calculated using an appropriate correlation given by Incropera and Dewitt (2002). 14 12230.42(0.4Re0.06Re)PrDD shD k (5-10) where Re D is the Reynolds number based on the diameter of the sphere, Pr is the Prandtl number, is the dynamic viscosity, s is the dynamic viscosity at the surface temperature, k is thermal conductivity, and D is the characteristic length. The calculated convection coeffi cient was from 294 to 374 2WmK The flame temperature is presented in Figure 5-4. 400 600 800 1000 1200 1400 1600 1800 2000 0510152025 Flame temperature (K)Flame temperature (K)Flame height (cm) Figure 5-4. Flame temperature as a function of flame height.

PAGE 159

141 Within this temperature range, the equilibri um calculation reveals that there are five major iron oxide species: Fe(OH)2(g), FeO(l), FeO(s), Fe2O3(s) and Fe3O4(s). Overall, FeO (s) is the dominant species among the iron oxides in the range from 800 to 1800 K. However, Fe3O4 is the dominant species over the other species at the temperature below 800 K, and Fe2O3 may be the dominant species at the ambient temperature. As shown in the Figure 5-3, Fe fraction d ecreases as flame height increases, which was proven during the LIF experi ment of both iron-seeded CO and isooctane flames. This result is in good agreement with the trend found in LIF experiments. 0 1 10-52 10-53 10-54 10-55 10-5800100012001400160018002000 Fe(g) FeO(g) Fe2O3(s) Fe3O4(s) O2 Fe(a)Relative mass fraction of speciesTemperature (K) Figure 5-5. Relative mass fraction of sp ecies as a function of temperature. With respect to FeO, gaseous phase FeO at the temperature of 2000 K is shown in Figure 5-5. McMillin et al. (1996) observed FeO LIF signal at the primary reaction zone in the premixed flame experimentally, and their numerical model showed the FeO was

PAGE 160

142 dominant species throug hout the flame (McMillin et al. 1996& Biswas et al. 1997). Since the temperature distribution of the pr emixed flame is normally from 2400 K to 1800 K, FeO could be formed in such an equilibrium condition. Son et al. (2000) also found FeO LIF signal from a ir on composite mixture, but they used photolysis of Fe(CO)5 to create FeO molecules. Accordi ng to Mitchell and Hackett (1990), the reaction between Fe atoms in the ground-state and O2 is endothermic reaction; therefore, the following reaction is unlikely to proceed in an appreciable rate at low temperature. 2Fe(g)O()FeO()Ogg (5-11) Gaseous FeO is not a very stable com pound under equilibrium at temperatures below 1800 K so that the formation of Fe O is not easy to occur under the flame configuration in this study. Therefore, the high concentrat ion of FeO (solid and liquid phase) shown in Figure 5-2 may not be a ppreciable in this study. 5.1.2 O2 Flow Rates The change in the concentration of car bon was examined at the temperature of 1800 K as the oxygen flow rate increases. Figure 5-6 shows the relative mass fraction of carbon as a function of the oxygen flow rates. Figure 5-6 shows the decreas e in soot as a function of the oxygen flow rate. Although this is little a bit different from th e smoke point of 4.25 LPM obtained from the smoke point study, this is in an excellent agreement with the oxyge n flow rate of 2.6 LPM employed for all experiments. In addi tion, the concentrati on of iron oxides is plotted as a function of the oxyge n flow rate in Figure 5-7. Figure 5-7 elucidates that FeO dominates in the fu el-lean condition while Fe2O3 and Fe3O4 start dominating in the fuel-rich condit ion. As the oxygen flow rate increases,

PAGE 161

143 the amount of Fe decreases, and Fe2O3 and Fe3O4 dominate FeO because the chance of reaction between Fe and O2 is high. It is noted here that the role of additional oxygen diffusion is neglected. 0 0.2 0.4 0.6 0.8 1 22.533.544.55 Mass fraction of carbonMass fraction of carbonOxygen flow rate (LPM) Figure 5-6. The decrease in re lative mass fraction of the so lid carbon as a function of the oxygen flow rate. 5.1.3 Fe(CO)5 Concentrations The change in the concentration of carbon was also examined at the temperature of 1800 K as the concentration of Fe(CO)5 increases. Figure 5-8 repr esents the relative mass fraction as a function of the iron pentacarbonyl concentration. In case of simulation on carbon mass fracti on, the amount of carbon increased as the mole of Fe(CO)5 increased, and mass fraction of carbon increased when iron

PAGE 162

144 pentacarbonyl was added to isooctane, which wa s not realistic. The increase in carbon appears to be attributed to increase in (CO)5 decomposed from iron pentacarbonyl. Overall, the numerical simulation does not provide the distinc tive transitions, but predict the relative stability and the pote ntial mixture of each species at a given temperature. For example, Figure 5-2 doe s not mean that FeO decomposes around 800 K and forms Fe3O4 at the lower temperature. Furthermore, the significant uncertainty of this simulation was to exclude the air in the reaction. It is also important to note that all species can partia lly evaporate at a given temper ature as long as they have appreciable vapor pressure. In addition to these uncertainties, the most significant 0 1002 10-54 10-56 10-58 10-51 10-41.2 10-41.4 10-422.533.544.55 Fe(a) FeO(s) Fe2O3(s) Fe3O4(s)Mass fraction of Fe speciesOxygen flow rate (LPM) Figure 5-7. Relative mass fraction of the ir on species as a function of the oxygen flow rate.

PAGE 163

145 discrepancy was that it was very difficult for species to reach thermal equilibrium condition inside of the flame. The residence time of the iron species inside of the flame would be about 50 milliseconds. This mean s the flame temperature drops down from 1800 K to 500 K within 50 ms. In such a cond ition, the equilibrium iron oxide particle formation is very unlikely to occur while the additive decomposition is still very active. The detailed discussion will be achieved in Chapter 6. Shortly, these numerical simulations can more correctly illustrate the state of species in the equilibrium condition. Therefore, this simulation results are limited of use to predict the experimental results in this study. 0 0.2 0.4 0.6 0.8 1 05000100001500020000 Mass fraction of carbonMass fraction of carbonConcentration of Fe(CO)5 Figure 5-8. Mass fraction of car bon as a function of the Fe(CO)5 concentration.

PAGE 164

146 CHAPTER 6 CONCLUSIONS AND FUTURE WORK 6.1 Summary and Conclusions This study was designed to e xplore and document the role that iron has the effect on the soot burning processes in the iron-seeded diffusion fl ame. Since an earlier study was achieved on the soot inception regime, the primary focus was on the latter regime of the flame, namely soot growth and oxidat ion regimes. Elastic light scattering and transmission data coupled with TEM analysis of soot particles sampled thermophoretically were used to extract info rmation on size, number density, and volume fraction of soot particle in the unseeded and seeded flames. In comparison with all results, no distinct trends between the two flames were observed in the soot growth regime, which indicates that th e iron additive has little eff ect on soot growth in this regime. This is consistent with the work of Masiello (2004), in which detailed measurements within the inception and growth regimes led to the conclusion that iron particulates are quickly incorporated within soot agglomerates, gi ving way to typical soot growth. On the other hand, the iron-seeded flame data showed a consistent change with respect to the unseeded flames in the soot oxidation regime. In particular, the maximum deviation between the two flam es was observed near the flame tip where the final soot volume fraction is reduced by nearly 66%. Th e average primary soot particle diameters at the lower heights (from 9.4 to 11.7 cm above burner) were 31 and 29 nm in the unseeded and seeded flames, respectively. In contrast, at the greatest residence time

PAGE 165

147 investigated (H = 25 cm), the primary particle diameters were reduced to 24 and 20 nm in the unseeded and iron-seeded flames, respectivel y. Such a decrease in particle size is consistent with significant soot oxidation. Overall, the primary particle diameters and volume fraction data, as measured from th e TEM micrographs and scattering signal, depict the transition from soot growth to soot oxidation along the length of the flame. The reduction in soot emission was visual ly observed as well. A pronounced smoke plume escaping from the flame tip disappeared when the iron seeded fuel was supplied. Based on all evidence stated above, it can be concluded that the iron additive has a significant effect of soot reduction that takes place in the oxidation regime. The remaining questions are how this re duction occurs, and what mechanisms are involved in the process of soot reduction. In order to answer these questions, it is useful to identify the iron species present in the flame. While ex situ measurements of Masiello (2004) suggested that Fe2O3 would be the most probable st ate of the iron in the oxidized soot particle, a numerical calculation reveal s FeO may be a dominant iron species under conditions of thermodynamic equilibrium. Indeed, the residence time of iron species inside the flame is about 50 milisecond, which may be too short to reach the equilibrium condition; hence, the formation of FeO or Fe2O3 is unlikely to occur on the inside of the flame. This is consistent with the recent work of Kim et al. (2005), which found considerable elemental Fe in a ethylene di ffusion flame. Laser-induced fluorescence and in situ Raman spectroscopy were examined to identify the chemical state of iron additive inside the seeded flame in this study. With these techniques, any form of iron oxides was not observed in the flame. In contrast, a strong Fe LIF signal was found throughout the flame heights. As discussed before, the prim ary conclusion is that the iron species do not

PAGE 166

148 reach thermodynamic equilibrium conditions, so that Fe atoms are significant, notably in the growth regimes. However, the rapid loss of elemental Fe within the oxidation regime may suggest some oxide formation, which is presumably below the current detection limits. Based on the spectroscopic evidence, th e conclusion is that the soot is primarily reduced by a direct Fe catalytic effect. Theoretically, both iron oxides and atoms can contribute to the reduction of soot. They may lead to direct oxidation of the soot particle, which can be expressed in chemical reaction below, C + FeO CO + Fe (6-1) 23C + FeO CO + 2FeO. (6-2) McMillin et al. (1996) found that some Fe atoms were combined with oxygen radical in the inner cone of the premixed flame. In this case, the mechanism of soot reduction could be explained with Equati on 6-1. On the other hand, they may help to oxidize the soot particles as catalysts. The chemical reac tions 6-3 and 6-4 illustrate the catalytic soot oxidation processes, solid22C + O + Fe CO + Fe, (6-3) solidC + OH + Fe CO + H + Fe. (6-4) There are two types of catalysts, name ly homogeneous catalyst and heterogeneous catalyst. Homogeneous catalys t promotes the reaction via di ssolving into the gas phase or solution, while heterogeneous catalyst enhances the reaction by increasing probability of the reaction on the surface of the catalyst. Transition me tals (i.e. Fe) are typical heterogeneous catalysts. Figur e 6-1 illustrated heterogene ous catalytic mechanism of soot oxidation.

PAGE 167

149 A B Figure 6-1. Schematic of soot oxidation mechanism. A) S oot oxidation without Fe. B) Soot oxidation with Fe. Soot particles can be oxidized by thr ee major oxidizers such as O atoms, OH radicals, and O2. However, the concentration of O in sooting flame is very low compared with other oxidizers; hence, the primary oxida tion of soot is considered to be due to OH and O2. In particular, the role of OH as an oxidizer may be significant in the combustion of hydrocarbon fuel. Surface reactions are resp onsible for the creation of OH. The first step of this mechanism is the dissociativ e adsorption of oxygen molecules on the surface of the elemental Fe expressed in Equation 6-5, 22Fe(s) + O O (s) + O (s). (6-5) These adsorbed O atoms collide with ad sorbed H atoms, forming OH in surface reactions. Fe (s) + H H (s) (6-6) O (s) + H (s) OH (6-7) In these equations, Fe (s) denotes free su rface sites, and (s) behind O and H atoms indicate surface species. Finally, OH radicals are desorbed from the surface of Fe and attack the solid carbon as show n in Equation 6-4. It is noted that no H is consumed OH Fe O = O O = O O = O OH OH C C C C Fe Fe Fe C O = O O = O O = O OH OH OH C C C C C C C

PAGE 168

150 overall in the mechanism. The surface reaction mechanism of hydrogen oxidation was delineated in Figure 6-2. Experimental valid ation of the presence of OH inside of the flame will be achieved in the future research work. Figure 6-2. Schematic of surface reaction m echanism of hydrogen oxidation. Three steps of the mechanism are adsorption, surface reaction, and desoprtion. The mechanism of surface reaction produces H2O as well. Both iron and iron oxide are well known catalyst for rapid convers ion of CO to CO2, so called water-gas shift reaction, 222CO + HO CO + H. (6-8) Lowered CO concentrations in the gas phase may favor the reaction of solid carbon and oxygen. Finally, a simple analysis was performe d to support the conclusion that the catalysis effect of Fe is the primary reason on soot reduction. One mole of iron is able to oxidize at most 4 moles of carbon to satisfy th e direct oxidation m echanism. However, approximately 500 moles of carbon are oxidized pe r 1 mole of iron exist, as based on the realized soot reduction; hence, the iron addi tive must play a role of the catalyst to enhance the chemical reaction between soot a nd oxidizers rather than directly oxidize the soot in the flame, which is consistent with the conclusion from the experiments. Surface reaction Adsorption H H O2 O H H O OH OH OH OH Desorption H2O H2O Fe

PAGE 169

151 Regarding Fe LIF signal drop, there are so me possibilities to explain it. Although the overall concentration of Fe atom may be c onserved, the density of Fe in the scattering volume is lower in downstream of the flame due to the diffusion of Fe atoms. However, the most likely reason is the Fe atom loss due to oxidation, 2 2Fe + OFeO x yxy (6-9) Equilibrium may be delayed at lower heights due to several reasons. First, the flame residence time is limited, as noted. In addition, the iron is encapsulated within the soot, which protects it from oxidation. Finally, oxygen is limited due to the competition with hydrocarbon oxidation. Ho wever, at longer residence times, the oxidation of carbon exposes the iron to additional oxygen, oxygen c oncentrations are increased by diffusion, and time-scales are longer allowing the appro ach to equilibrium. Once iron transitions occur, additional catalytic path ways may emerge, such as solid2 2C+ FeO+ O CO + FeO x yxy, (6-10) that involved additiona l iron oxide species. In addition, the Fe LIF signal drop can be e xplained in terms of Fe vapor pressure. As shown in TEM images, iron species are concentr ated in the core of soot particles. In the soot growth regime, the same elemen tal iron exists in gas phase due to high temperature, so that they can diffuse through the soot partic le and into the gas. Such elemental Fe can be detected by LIF probe However, the flame temperature is significantly dropped throughout th e oxidation regime, which re sults in the reduction of Fe vapor pressure. As a result of this, the c oncentration of gas phase elemental Fe is also reduced. Even though Fe that remained in th e core would be somewhat exposed due to

PAGE 170

152 the loss of soot, the oxida tion by an enhanced amount of the oxygen would keep the concentration of Fe still low. Overall, the exact flame configuration, combustor residence times, and combustor stoichiometry are all expected to play key ro les in the soot suppressing effects realized with iron-based additives. 6.2 Future Work The current study has yielded quantitativ e and qualitative insights into the mechanisms of soot suppression with ironbased fuel additives. With further understanding of the iron mech anisms gained via this st udy, the next phase of the investigation is to examine such effects using the practical co mbustion engines. Future works should include: 1. Perform additional in situ probes for iron oxides (e.g. photofragmentation). 2. Perform OH LIF probe for validating the presence of OH radical in the flame. 3. Perform extractive sampling probe with on-line mass spectrometry. 4. Assess the role of iron a ddition in a laboratory-scale gas turbine engine applying all techniques used for lab studies.

PAGE 171

153 APPENDIX A ANALYSIS OF THE FLAME In order to evaluate Froude number, Reynolds number and eq uivalence ratio, a complete calculation was performed for the flame and presented here. With respect to Froude Number, it physically represents the ratio of the in itial jet momentum flow to the buoyant force experienced by the flame as given by g L v Frc e f 2, (A-1) where the flame length(cL) is around 0.31 m which was meas ured rather than calculated considering the flame temperature. Moreover, the gravity is 9.81 m/sec2, and ev is the exit velocity given by 2 22224()4 5.99sececenterannular O mix mixcOacvvv m m ddd m (A-2) In this equation, the volume flow rate was measured that LPM Vfuel0015 0 and the density of isooctane was 0.688 g/mL at 25C. Therefore, the mass flow rate is calculated below, 51.7210secfuelmKg, (A-3) 233 3 50.810 1.123 60sec 1.510secNmKg m m Kg (A-4)

PAGE 172

154 The density (2N ) of nitrogen is 1.123 Kg/m3 and the density (2O ) of oxygen is 1.284 Kg/m3 at the temperature of 300 K. The mo lecular weight (fuel + nitrogen) is defined as 23 142 28 23 1142 N fuel mixMW MW MW (A-5) 3101325 4.95 ()(8315/142.23)350mix umixP Kgm RMWT (A-6) 233 3 52.610 1.284 60sec 5.610secOmKg m m Kg (A-7) The diameter of the center hole of the burner (centerd ) was 1.5 10-3 m, and that of the nine annular hole s of the burner was 3 10-4 m. It is noteworthy that the annular holes were assumed as one concentric circ le having the outer diameter of 5.13 10-3 m. See Figure 3-1. 25551.72101.5103.2210secmixfuelNmmmKg. (A-8) Finally, the Froude number was determined to be 11.8. As for Reynolds number, d ve e Re, (A-9) 3101325 6.07 ()(8315/174.23)350mix umixP Kgm RMWT (A-10) m d d dannular center 310 13 5 (A-11)

PAGE 173

155 With respect to viscosity ( ),18 8H C is 4.710-4 Kg/ms and air is 210-5 Kg/ms. mix is assumed around 2.4510-5 Kg/ms. Therefore, the Reynolds number was 761.3. In addition, the equivalence ratio, was evaluated. The governing combustion stoichiometric reaction for this experiment is given by 81822225 89 2 CHOCOHO (A-12) In the experimental case, the volumetric fl ow rates of fuel a nd oxidizer were 4.38 cc per second and 50.9 cc per second, respectiv ely (see Appendix E). For an ideal gas, the volumetric ratio is equal to the molar ratio de fined as the ratio of th e actual air to fuel ratio to the stoichiometric air to fuel ratio. The equivalence ratio is greater than unity representing fuel lean combustion while be ing less than unity indicating fuel rich combustion. The equivalence ratio, was calculated as exp/ 252 1.08 /50.94.38stoichiometric erimentalAF AF. (A-13) In conclusion, first, the flame was not buoyancy-controlled because the Froude number was greater than unity. Second, it was laminar flow because the Reynolds number was less than ~2300. Third, the diffu sion flame employed in the experiment was run on the fuel rich side of stoichiometry.

PAGE 174

156 APPENDIX B RESULTS OF RDG CALCULATIONS In order to determine scattering paramete rs, Rayleigh Debye Gans scattering theory was employed. The detailed results are presen ted here. In additi on, new calculation was carried out based on a complex refractive index of m = 1.57-0.56 i Table B-1 shows the radius of the primary soot particle determined by TEM analysis. Curve fit data were used in the following calculation. Table B-1. Measured radius of the primary soot particle. Unseeded Seeded Height (cm) Radius (cm) Fit radius (cm) Radius (cm) Fit radius (cm) 9.4 1.71E-06 1.59E-06 1.36E-06 1.35E-06 10.1 1.48E-06 1.59E-06 1.29E-06 1.39E-06 10.9 1.62E-06 1.59E-06 1.47E-06 1.42E-06 11.7 1.44E-06 1.58E-06 1.45E-06 1.45E-06 12.7 1.57E-06 1.57E-06 1.45E-06 1.48E-06 13.7 1.60E-06 1.56E-06 1.61E-06 1.50E-06 14.9 1.58E-06 1.54E-06 1.57E-06 1.51E-06 16.2 1.56E-06 1.51E-06 1.50E-06 1.50E-06 17.75 1.52E-06 1.46E-06 1.42E-06 1.47E-06 19.6 1.33E-06 1.40E-06 1.34E-06 1.40E-06 21.85 1.25E-06 1.31E-06 1.24E-06 1.26E-06 25.25 1.17E-06 1.13E-06 1.00E-06 9.62E-07 The first step in determining the characteristics of the soot particles was to calculate the differential scattering cross section using the equation, 2 2 2 6 2 2 '2 1 4 m mVV (B-1) The result is listed in Table B-2.

PAGE 175

157 Table B-2. The differential sc attering cross section (cm2/sr). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 8.82E-16 6.83E-16 3.29E-16 2.55E-16 10.1 8.79E-16 6.81E-16 3.88E-16 3.00E-16 10.9 8.69E-16 6.73E-16 4.54E-16 3.52E-16 11.7 8.51E-16 6.59E-16 5.15E-16 3.99E-16 12.7 8.19E-16 6.34E-16 5.79E-16 4.48E-16 13.7 7.77E-16 6.02E-16 6.23E-16 4.83E-16 14.9 7.15E-16 5.54E-16 6.44E-16 4.99E-16 16.2 6.38E-16 4.94E-16 6.24E-16 4.83E-16 17.75 5.37E-16 4.16E-16 5.48E-16 4.24E-16 19.6 4.14E-16 3.21E-16 4.07E-16 3.15E-16 21.85 2.75E-16 2.13E-16 2.22E-16 1.72E-16 25.25 1.16E-16 9.02E-17 4.32E-17 3.35E-17 The radius of gyration and the structure f actor then can be calculated using the equations, 17 1 ) 2 ( ) (2 1gR LW, (B-2) 1 ) ( ) ( g D gqR qR C q Sf. (B-3) where the constant C is approximately one. In order to determine the total scattering cross section for a fractal aggregate, ) (gkR G is also calculated by 2 / 2 2) 3 4 1 ( ) (fD g f gR k D kR G (B-4) where the fractal dimension was 1.82 for all conditions. The results are summarized in Tables B-3 and B-4. In additi on, the number density of prim ary particles in an aggregate and curve fit values are shown in Tables. Th e differential scattering coefficients obtained from the light scattering expe riments and corrected values by considering transmission are shown in Table B-5.

PAGE 176

158 Table B-3. Summary of calculated results for the unseeded flame. Unseeded Height (cm) Rg Npri Fit Npri qRg S(q) G(kRg) 9.4 1.29E-05 77 103 2.16 0.25 0.18 10.1 1.28E-05 108 115 2.14 0.25 0.18 10.9 1.65E-05 135 128 2.76 0.16 0.12 11.7 1.37E-05 133 139 2.29 0.22 0.16 12.7 1.93E-05 203 153 3.22 0.12 0.09 13.7 1.56E-05 164 166 2.6 0.18 0.13 14.9 171 180 16.2 1.75E-05 178 194 2.93 0.14 0.11 17.75 2.24E-05 231 210 3.73 0.09 0.07 19.6 2.37E-05 346 226 3.95 0.08 0.07 21.85 2.10E-05 254 245 3.51 0.10 0.08 25.25 1.80E-05 247 269 3.00 0.14 0.11 Table B-4. Summary of calculated results for the seeded flame. Seeded Height (cm) Rg Npri Fit Npri qRg S(q) G(kRg) 9.4 8.98E-06 64 99 1.50 0.48 0.30 10.1 8.12E-06 60 113 1.36 0.57 0.34 10.9 1.43E-05 116 128 2.39 0.21 0.15 11.7 2.24E-05 203 143 3.75 0.09 0.07 12.7 2.04E-05 211 159 3.41 0.11 0.09 13.7 2.59E-05 244 174 4.33 0.07 0.06 14.9 2.42E-05 286 191 4.04 0.08 0.06 16.2 2.03E-05 212 208 3.4 0.11 0.09 17.75 256 226 19.6 2.62E-05 301 246 4.38 0.07 0.06 21.85 2.01E-05 223 268 3.36 0.11 0.09 25.25 1.69E-05 170 297 2.82 0.15 0.12 The differential scattering cross secti on of an aggregate is defined as ) (' 2 'q S Npar par agg (B-5) The results are shown in Table B-6.

PAGE 177

159 Table B-5. Uncorrected and corrected di fferential scattering coefficients (cm-1sr-1). Unseeded Seeded Height (cm) K vv, uncorrected K vv, corrected K vv, uncorrected K vv, corrected 9.4 2.87E-03 3.70E-03 3.08E-03 3.98E-03 10.1 2.90E-03 3.87E-03 2.24E-03 2.91E-03 10.9 2.43E-03 3.26E-03 1.99E-03 2.67E-03 11.7 2.16E-03 2.97E-03 2.24E-03 3.05E-03 12.7 2.63E-03 3.60E-03 2.02E-03 2.68E-03 13.7 2.98E-03 4.10E-03 2.65E-03 3.77E-03 14.9 2.81E-03 3.78E-03 2.38E-03 3.13E-03 16.2 2.07E-03 2.76E-03 1.80E-03 2.28E-03 17.75 1.47E-03 1.74E-03 1.24E-03 1.40E-03 19.6 9.17E-04 1.04E-03 9.44E-04 1.01E-03 21.85 5.04E-04 5.44E-04 3.78E-04 3.95E-04 25.25 2.29E-04 2.44E-04 6.16E-05 6.35E-05 Table B-6. Differential scattering cro ss section for a fractal aggregate (cm2/sr). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 2.30E-12 1.78E-12 1.53E-12 1.18E-12 10.1 2.89E-12 2.24E-12 2.84E-12 2.20E-12 10.9 2.24E-12 1.73E-12 1.54E-12 1.19E-12 11.7 3.66E-12 2.84E-12 9.48E-13 7.35E-13 12.7 2.30E-12 1.78E-12 1.58E-12 1.22E-12 13.7 3.78E-12 2.93E-12 1.32E-12 1.02E-12 14.9 3.60E-12 2.79E-12 1.86E-12 1.44E-12 16.2 3.42E-12 2.65E-12 2.92E-12 2.26E-12 17.75 2.16E-12 1.67E-12 2.30E-12 1.78E-12 19.6 1.75E-12 1.35E-12 1.68E-12 1.30E-12 21.85 1.68E-12 1.31E-12 1.76E-12 1.36E-12 25.25 1.14E-12 8.85E-13 5.77E-13 4.47E-13 The number density of aggregate, aggN in the scattering volume can be calculated using the equation, '' aggVVaggaggNK (B-6)

PAGE 178

160 The results are presented in Table B-7. Table B-7. Number density of soot aggregates in the sc attering volume (particles/cm3). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 1.61E+09 2.08E+09 2.60E+09 3.36E+09 10.1 1.34E+09 1.73E+09 1.02E+09 1.32E+09 10.9 1.46E+09 1.88E+09 1.74E+09 2.24E+09 11.7 8.12E+08 1.05E+09 3.22E+09 4.16E+09 12.7 1.57E+09 2.02E+09 1.70E+09 2.19E+09 13.7 1.08E+09 1.40E+09 2.87E+09 3.70E+09 14.9 1.05E+09 1.35E+09 1.69E+09 2.18E+09 16.2 8.06E+08 1.04E+09 7.83E+08 1.01E+09 17.75 8.08E+08 1.04E+09 6.10E+08 7.87E+08 19.6 5.95E+08 7.67E+08 5.99E+08 7.74E+08 21.85 3.23E+08 4.17E+08 2.25E+08 2.90E+08 25.25 2.14E+08 2.76E+08 1.10E+08 1.42E+08 The total scattering cross sections for a primary particle and an aggregate are, respectively, calculated by 2 2 2 6 22 1 3 2 m msca (B-7) ) (2g sca par par sca aggkR G N (B-8) The results are tabulated in Tables B-8 a nd B-9 respectively. The absorption cross sections for a primary particle and an a ggregate are, respectively, calculated by 2 1 Im2 2 3 2m mabs (B-9) abs par par abs aggN (B-10) The results are listed in Tables B-10 and B-11.

PAGE 179

161 Table B-8. Total scattering cross sec tion for a primary soot particle (cm2). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 7.39E-15 5.72E-15 2.75E-15 2.13E-15 10.1 7.36E-15 5.71E-15 3.25E-15 2.52E-15 10.9 7.28E-15 5.64E-15 3.80E-15 2.95E-15 11.7 7.13E-15 5.52E-15 4.31E-15 3.34E-15 12.7 6.86E-15 5.31E-15 4.85E-15 3.76E-15 13.7 6.51E-15 5.04E-15 5.22E-15 4.04E-15 14.9 5.99E-15 4.64E-15 5.39E-15 4.18E-15 16.2 5.34E-15 4.14E-15 5.22E-15 4.05E-15 17.75 4.50E-15 3.49E-15 4.59E-15 3.55E-15 19.6 3.47E-15 2.69E-15 3.41E-15 2.64E-15 21.85 2.30E-15 1.78E-15 1.86E-15 1.44E-15 25.25 9.76E-16 7.56E-16 3.62E-16 2.80E-16 Table B-9. Total scattering cro ss section of an aggregate (cm2). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 1.39E-11 1.08E-11 8.10E-12 6.28E-12 10.1 1.75E-11 1.36E-11 1.43E-11 1.11E-11 10.9 1.43E-11 1.11E-11 9.56E-12 7.41E-12 11.7 2.26E-11 1.75E-11 6.34E-12 4.91E-12 12.7 1.51E-11 1.17E-11 1.04E-11 8.08E-12 13.7 2.39E-11 1.85E-11 8.91E-12 6.91E-12 14.9 2.30E-11 1.79E-11 1.25E-11 9.68E-12 16.2 2.21E-11 1.72E-11 1.93E-11 1.50E-11 17.75 1.44E-11 1.12E-11 1.53E-11 1.19E-11 19.6 1.17E-11 9.09E-12 1.14E-11 8.83E-12 21.85 1.12E-11 8.66E-12 1.16E-11 8.99E-12 25.25 7.43E-12 5.76E-12 3.72E-12 2.88E-12 The total extinction cross section is de fined as a sum of the scattering and absorption cross section, namely, abs sca ext (B-11) The calculated results are presented in Table B-12.

PAGE 180

162 Table B-10. Absorption cross sec tion of a primary particle (cm2). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 6.86E-13 1.55E-12 4.19E-13 9.45E-13 10.1 6.85E-13 1.55E-12 4.55E-13 1.03E-12 10.9 6.81E-13 1.54E-12 4.92E-13 1.11E-12 11.7 6.74E-13 1.52E-12 5.25E-13 1.18E-12 12.7 6.61E-13 1.49E-12 5.56E-13 1.25E-12 13.7 6.44E-13 1.45E-12 5.77E-13 1.30E-12 14.9 6.18E-13 1.39E-12 5.86E-13 1.32E-12 16.2 5.84E-13 1.32E-12 5.77E-13 1.30E-12 17.75 5.36E-13 1.21E-12 5.41E-13 1.22E-12 19.6 4.70E-13 1.06E-12 4.66E-13 1.05E-12 21.85 3.83E-13 8.64E-13 3.44E-13 7.77E-13 25.25 2.49E-13 5.63E-13 1.52E-13 3.43E-13 Table B-11. Absorption cross se ction of an aggregate (cm2). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 7.04E-11 1.59E-10 4.13E-11 9.32E-11 10.1 7.86E-11 1.77E-10 5.14E-11 1.16E-10 10.9 8.69E-11 1.96E-10 6.32E-11 1.43E-10 11.7 9.40E-11 2.12E-10 7.47E-11 1.69E-10 12.7 1.01E-10 2.29E-10 8.84E-11 1.99E-10 13.7 1.07E-10 2.41E-10 1.00E-10 2.27E-10 14.9 1.11E-10 2.51E-10 1.12E-10 2.53E-10 16.2 1.13E-10 2.56E-10 1.20E-10 2.71E-10 17.75 1.12E-10 2.53E-10 1.22E-10 2.76E-10 19.6 1.06E-10 2.40E-10 1.15E-10 2.59E-10 21.85 9.38E-11 2.12E-10 9.23E-11 2.08E-10 25.25 6.71E-11 1.51E-10 4.51E-11 1.02E-10 The extinction coefficient for the overall aggregates in the volume is finally determined using the following equation, agg ext agg agg extN K. (B-12) The calculated results are summarized in Table B-13.

PAGE 181

163 Table B-12. The extinction cross section of an aggregate (cm2). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 8.43E-11 1.70E-10 4.94E-11 9.95E-11 10.1 9.61E-11 1.91E-10 6.57E-11 1.27E-10 10.9 1.01E-10 2.07E-10 7.27E-11 1.50E-10 11.7 1.17E-10 2.30E-10 8.11E-11 1.74E-10 12.7 1.16E-10 2.40E-10 9.88E-11 2.07E-10 13.7 1.31E-10 2.60E-10 1.09E-10 2.34E-10 14.9 1.34E-10 2.69E-10 1.24E-10 2.62E-10 16.2 1.36E-10 2.73E-10 1.39E-10 2.86E-10 17.75 1.27E-10 2.65E-10 1.38E-10 2.88E-10 19.6 1.18E-10 2.49E-10 1.26E-10 2.68E-10 21.85 1.05E-10 2.20E-10 1.04E-10 2.17E-10 25.25 7.45E-11 1.57E-10 4.88E-11 1.05E-10 Table B-13. The extinction coefficient (cm-1). Unseeded Seeded Height m=2.00.35i m=1.570.56i B-L Law* m=2.00.35i m=1.570.56i B-L Law 9.4 1.36E-01 3.53E-01 3.03E-01 1.29E-01 3.34E-01 2.45E-01 10.1 1.29E-01 3.29E-01 3.18E-01 6.72E-02 1.68E-01 2.43E-01 10.9 1.48E-01 3.90E-01 2.98E-01 1.26E-01 3.37E-01 2.65E-01 11.7 9.47E-02 2.41E-01 3.07E-01 2.61E-01 7.22E-01 2.77E-01 12.7 1.82E-01 4.86E-01 2.80E-01 1.68E-01 4.55E-01 2.48E-01 13.7 1.42E-01 3.64E-01 2.70E-01 3.13E-01 8.64E-01 3.02E-01 14.9 1.41E-01 3.64E-01 2.42E-01 2.10E-01 5.72E-01 2.33E-01 16.2 1.09E-01 2.84E-01 2.26E-01 1.09E-01 2.88E-01 1.99E-01 17.75 1.02E-01 2.76E-01 1.33E-01 8.39E-02 2.26E-01 1.07E-01 19.6 7.03E-02 1.91E-01 9.87E-02 7.56E-02 2.07E-01 5.68E-02 21.85 3.39E-02 9.18E-02 6.50E-02 2.33E-02 6.29E-02 4.11E-02 25.25 1.59E-02 4.33E-02 7.05E-02 5.37E-03 1.48E-02 3.30E-02 *B-L Law is Beer-Lambert law. The extinction coefficients for the optical length determined using Beer-Lambert law and transmission data are listed in Table B-13 as well. The optical lengths were obtained from 25 digital photographs of the uns eeded and seeded flames. In the same

PAGE 182

164 manner discussed for analyzing soot part icles in TEM images, the flame widths corresponding to 12 heights were measured using Measure tool in Photoshop computer software. Alternatively, it can be determin ed by using either the scattered signal or transmission data taken along the radial positi ons of the flame. Fi gure B-1 represents the differential scattering coefficients measured along the radial positions at three different heights. 0 0.001 0.002 0.003 0.004 0.005 -6-4-20246 H1 H11 H25K' VV (cm-1sr-1)Radial position (mm) Figure B-1. The differential sca ttering coefficients measured along the radial positions at three different heights. Error bars represent one standard deviation. This is the common way to determine th e optical pathlength; however, not only was the flame employed in this study very thin but also some what severe fluctuation of the flame made it hard to obtain reasonable da ta. Therefore, this data was not use for determining the extinction coefficient us ing Beer-Lamberts law in this study.

PAGE 183

165 Overall, the deviations of the absolute values of it were about 43% for the extinction coefficients determined with both th e former and the latter refractive indices. The extinction coefficients are presen ted in Figures B-2 and B-3. 0.01 0.1 1 10.112.114.116.118.120.122.124.126.1 m=2.0-0.35i m=1.57-0.56i ExperimentKext (cm-1)Height (cm) Figure B-2. The extinction coefficients as a function of the unseeded flame height. The extinction coefficients determined using RDG scattering theory for two different refractive index were comp ared with that from transmission experiments. Finally, the total number density in the overall scattering volume is defined as par agg TotalN N N (B-13) The volume of an aggregate is given by 36aggparparVdN (B-14) and the volume fraction vf is calculated by Equation B-15, agg agg vN V f (B-15)

PAGE 184

166 0.001 0.01 0.1 1 10.112.114.116.118.120.122.124.126.1 m=2.0-0.35i m=1.57-0.56i ExperimentKext(cm-1)Height (cm) Figure B-3. The extinction coefficients as a function of the seeded flame height. The extinction coefficients determined using RDG scattering theory for two different refractive index were comp ared with that from transmission experiments. The results are tabulated in Ta bles B-14 and B-15 respectively. The refractive index of m=1.57-0.56i yielde d about 23% greater results for the all scattering parameters and 56% greater results for absorpti on cross sections of both a primary particle and an aggregate, which resu lted in 63% of deviation for the extinction coefficient. However, the trend and the re lative ratios of the unseeded to seeded flame results were identical for the al ternative refractive indices.

PAGE 185

167 Table B-14. Number density of soot particles in the scattering volume (particles/cm3). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 1.65E+11 2.13E+11 2.57E+11 3.31E+11 10.1 1.53E+11 1.98E+11 1.16E+11 1.49E+11 10.9 1.86E+11 2.40E+11 2.23E+11 2.88E+11 11.7 1.13E+11 1.46E+11 4.59E+11 5.93E+11 12.7 2.40E+11 3.10E+11 2.70E+11 3.49E+11 13.7 1.80E+11 2.32E+11 4.99E+11 6.44E+11 14.9 1.89E+11 2.44E+11 3.23E+11 4.16E+11 16.2 1.57E+11 2.02E+11 1.63E+11 2.10E+11 17.75 1.69E+11 2.19E+11 1.38E+11 1.78E+11 19.6 1.35E+11 1.74E+11 1.47E+11 1.90E+11 21.85 7.91E+10 1.02E+11 6.01E+10 7.76E+10 25.25 5.75E+10 7.42E+10 3.26E+10 4.21E+10 Table B-15. The volume fraction of soot particles in the scattering volume (cm3 soot/cm3). Unseeded Seeded Height (cm) m=2.0-0.35i m=1.57-0.56i m=2.0-0.35i m=1.57-0.56i 9.4 2.78E-06 3.59E-06 2.64E-06 3.41E-06 10.1 2.58E-06 3.33E-06 1.29E-06 1.67E-06 10.9 3.11E-06 4.01E-06 2.70E-06 3.48E-06 11.7 1.87E-06 2.42E-06 5.91E-06 7.63E-06 12.7 3.89E-06 5.02E-06 3.69E-06 4.76E-06 13.7 2.85E-06 3.67E-06 7.06E-06 9.11E-06 14.9 2.87E-06 3.70E-06 4.64E-06 5.99E-06 16.2 2.24E-06 2.89E-06 2.30E-06 2.97E-06 17.75 2.23E-06 2.87E-06 1.83E-06 2.36E-06 19.6 1.55E-06 2.00E-06 1.69E-06 2.18E-06 21.85 7.43E-07 9.59E-07 5.08E-07 6.56E-07 25.25 3.52E-07 4.54E-07 1.22E-07 1.57E-07

PAGE 186

168 APPENDIX C ERROR ANALYSIS An error analysis was carried out for the vol ume fraction at each he ight. In general, for a function ,,,, x fabyz the standard deviation in x and x is determined from the propagation of the errors of a through z. Namely, 2 222 2222 x abyzxxxx abyz (C-1) The volume fraction is defined as 3 ' 42 323() 6 2211 31()VVagg Vparpar VVagg VVagg parparK fdN K m mNSqd (C-2) where the differential scattering cross section for an aggregate and primary particle is given by ) (' 2 'q S Npar par agg (C-3) 22 6 22 4 2 '6 22 2121 42 22p pard mm mm (C-4) Therefore, for the volume fraction, ,,,(),VVVaggparparffKNSqd the standard deviation is determined from p ar ,222 2 2222 ()d par()dV par VVaggVVVV fNsq K VVaggparffff KNSq ,(C-5)

PAGE 187

169 where, '3 ,()V VVaggparparf A KNSqd (C-6) 32 par1 ()dVVagg V parparAK f NSqN (C-7) 32 par1 ()d()VVagg V parAK f SqNSq (C-8) 4 p arpar3 1 d()dVVagg V parAK f NSq (C-9) In these equations, A is determined from 42 19 3222 6.1410 31m A m (C-10) The parameters calculated using these equations at each height were summarized in Tables C-1 and 2 for the unseeded and seeded flames respectively. Table C-1. Summary of the calculated parame ters with Equations C-6 through C-9 for the unseeded flame. Height (cm) ,VVVaggfK Vpar f N ()V f Sq p ardVf 9.4 7.52E-04 -2.71E-08 -1.12E-05 -2.63E+00 10.1 6.67E-04 -2.25E-08 -1.03E-05 -2.43E+00 10.9 9.53E-04 -2.44E-08 -1.96E-05 -2.94E+00 11.7 6.30E-04 -1.34E-08 -8.46E-06 -1.78E+00 12.7 1.08E-03 -2.54E-08 -3.26E-05 -3.72E+00 13.7 6.95E-04 -1.71E-08 -1.61E-05 -2.74E+00 14.9 7.59E-04 16.2 8.13E-04 -1.15E-08 -1.58E-05 -2.23E+00 17.75 1.28E-03 -1.06E-08 -2.44E-05 -2.28E+00 19.6 1.50E-03 -6.86E-09 -1.89E-05 -1.66E+00 21.85 1.37E-03 -3.04E-09 -7.26E-06 -8.51E-01 25.25 1.44E-03 -1.31E-09 -2.60E-06 -4.65E-01

PAGE 188

170 Table C-2. Summary of the calculated parame ters with Equations C-6 through C-9 for the seeded flame. Height (cm) ,VVVaggfK Vpar f N ()V f Sq p ardVf 9.4 6.63E-04 -2.68E-08 -5.51E-06 -2.94E+00 10.1 4.44E-04 -1.14E-08 -2.25E-06 -1.40E+00 10.9 1.01E-03 -2.10E-08 -1.31E-05 -2.84E+00 11.7 1.93E-03 -4.15E-08 -6.52E-05 -6.10E+00 12.7 1.37E-03 -2.32E-08 -3.42E-05 -3.73E+00 13.7 1.87E-03 -4.05E-08 -1.01E-04 -7.06E+00 14.9 1.48E-03 -2.43E-08 -5.87E-05 -4.61E+00 16.2 1.01E-03 -1.11E-08 -2.12E-05 -2.30E+00 17.75 1.30E-03 19.6 1.67E-03 -6.85E-09 -2.47E-05 -1.81E+00 21.85 1.29E-03 -1.90E-09 -4.61E-06 -6.03E-01 25.25 1.92E-03 -4.10E-10 -8.02E-07 -1.90E-01 The differential scattering coefficient co rrected for transmission was determined from the relationship, namely, ' ,VVraw VVcorrK K (C-11) The error in this calculation was determined from '' ,, ,2 2 '' ,, 22 2 2 22 21VVaggVVraw VVrawVVcorrVVcorr KK VVraw VVraw KKK K K (C-12) where, VVrawKand are standard deviations of ,VVrawK and which were obtained from experimental data. Tables C-3 and C-4 tabulate the experimental data and calculated parameters at each height for th e unseeded and seeded flame respectively.

PAGE 189

171 Table C-3. Results of the calculation usi ng Equation C-12 for the unseeded flame. Height (cm) '' ,,VVcorrVVrawKK VVrawK ,VVcorrK VVaggK 9.4 1.29E+00 6.67E-04 -4.77E-03 1.72E-02 8.63E-04 10.1 1.34E+00 3.53E-04 -5.16E-03 2.03E-02 4.82E-04 10.9 1.34E+00 8.30E-04 -4.37E-03 3.91E-02 1.12E-03 11.7 1.38E+00 8.03E-04 -4.10E-03 2.32E-02 1.11E-03 12.7 1.37E+00 8.38E-04 -4.92E-03 4.40E-02 1.17E-03 13.7 1.37E+00 5.09E-04 -5.62E-03 3.78E-02 7.31E-04 14.9 1.35E+00 3.13E-04 -5.08E-03 3.43E-02 4.56E-04 16.2 1.33E+00 2.66E-04 -3.67E-03 4.77E-02 3.95E-04 17.75 1.19E+00 3.68E-04 -2.07E-03 6.64E-02 4.57E-04 19.6 1.13E+00 1.80E-04 -1.18E-03 5.50E-02 2.14E-04 21.85 1.08E+00 9.78E-05 -5.87E-04 3.71E-02 1.08E-04 25.25 1.07E+00 4.01E-05 -2.60E-04 2.46E-02 4.32E-05 Table C-4. Results of the calculation us ing Equation C-12 for the seeded flame. Height (cm) '' ,,VVcorrVVrawKK VVrawK ,VVcorrK VVaggK 9.4 1.29E+00 5.66E-04 -5.15E-03 2.83E-02 7.46E-04 10.1 1.30E+00 7.64E-04 -3.78E-03 2.63E-02 9.97E-04 10.9 1.34E+00 6.71E-04 -3.58E-03 2.58E-02 9.04E-04 11.7 1.37E+00 7.91E-04 -4.17E-03 3.31E-02 1.09E-03 12.7 1.33E+00 4.38E-04 -3.57E-03 6.01E-02 6.21E-04 13.7 1.42E+00 6.09E-04 -5.36E-03 3.21E-02 8.82E-04 14.9 1.32E+00 9.37E-04 -4.12E-03 3.74E-02 1.24E-03 16.2 1.27E+00 6.78E-04 -2.89E-03 3.69E-02 8.65E-04 17.75 1.13E+00 2.86E-04 -1.59E-03 1.30E-02 3.25E-04 19.6 1.07E+00 2.45E-04 -1.08E-03 1.51E-02 2.62E-04 21.85 1.05E+00 1.15E-04 -4.13E-04 2.05E-02 1.20E-04 25.25 1.03E+00 3.95E-05 -6.55E-05 1.87E-02 4.08E-05 With respect to p arN, standard deviations for 25 samp les at each height are errors of propagation from large distri bution of them rather than true experimental errors. Thus, p arNwas estimated using deviation between expe rimental values and corrected values from curve fit, namely,

PAGE 190

172 expparparpar corr N par corrNN N (C-13) ,parparparNparNN corr unseededseededNaverage (C-14) Similar to the number density calculation, p ard was estimated. The average values of p arN and p ard were 0.216 and 0.038 respectively. Table C-5 shows the calculated errors for the particle size and number density. Table C-5. Summary of calculated errors at each height for the particle size and number density. p ard p arN True Height (cm) unseeded seeded unseeded seeded 9.4 1.20E-07 1.02E-07 2.21E+01 2.13E+01 10.1 1.20E-07 1.05E-07 2.47E+01 2.44E+01 10.9 1.20E-07 1.08E-07 2.75E+01 2.77E+01 11.7 1.19E-07 1.10E-07 3.01E+01 3.07E+01 12.7 1.19E-07 1.12E-07 3.31E+01 3.43E+01 13.7 1.18E-07 1.13E-07 3.58E+01 3.76E+01 14.9 1.16E-07 1.14E-07 3.89E+01 4.12E+01 16.2 1.14E-07 1.13E-07 4.19E+01 4.48E+01 17.75 1.11E-07 1.11E-07 4.52E+01 4.88E+01 19.6 1.06E-07 1.06E-07 4.88E+01 5.31E+01 21.85 9.90E-08 9.55E-08 5.28E+01 5.78E+01 25.25 8.58E-08 7.27E-08 5.80E+01 6.40E+01 Lastly, the structure factor of an aggregat e was given by the relationship, namely, ()() f D gSqCqR. (C-15) The error,() s q was determined from 2 2 ()()SqCSq C 2 2()qSq q 22 22()() g fRD gfSqSq RD (C-16)

PAGE 191

173 In this equation, 1()ffDD fg gSq DqR R (C-17) As for g R the radius of the gyration is determined from 17 1 ) 2 ( ) (2 1gR LW. (C-18) The error in this calculation was given by 22 22ggg RLWRR LW (C-19) where 11 221 41.17gR WL L (C-20) 11 221 41.17gR WL W (C-21) L and W were obtained from experimental data. Meanwhile, () lnfD g g fSq qRqR D (C-22) where f D was obtained from experimental data The calculated parameters were summarized in Tables C-6 and C-7 for the unseeded and seeded flame. C and q are zero because C and q are the constants; thus, ()Sq C and ()Sq q were not necessary to be determined.

PAGE 192

174 Table C-6. Summary of calculated parameters at each height for the unseeded flame. True Height (cm) () g Sq R g R () f Sq D f D ()Sq 9.4 -3.48E+04 9.26E-06 -1.90E-01 1.10E-01 3.23E-01 10.1 -3.53E+04 9.18E-06 -1.90E-01 1.10E-01 3.25E-01 10.9 -1.74E+04 1.18E-05 -1.60E-01 1.10E-01 2.05E-01 11.7 -2.93E+04 1.01E-05 -1.83E-01 1.10E-01 2.98E-01 12.7 -1.12E+04 1.41E-05 -1.39E-01 1.10E-01 1.59E-01 13.7 -2.06E+04 1.13E-05 -1.68E-01 1.10E-01 2.32E-01 14.9 1.10E-01 16.2 -1.47E+04 1.28E-05 -1.52E-01 1.10E-01 1.90E-01 17.75 -7.40E+03 1.60E-05 -1.20E-01 1.10E-01 1.19E-01 19.6 -6.30E+03 1.76E-05 -1.13E-01 1.10E-01 1.12E-01 21.85 -8.84E+03 1.51E-05 -1.28E-01 1.10E-01 1.34E-01 25.25 -1.37E+04 1.31E-05 -1.49E-01 1.10E-01 1.80E-01 Table C-7. Summary of calculated parameters at each height for the seeded flame. True Height (cm) () g Sq R g R () f Sq D f D ()Sq 9.4 -9.69E+04 6.44E-06 -1.94E-01 1.10E-01 6.24E-01 10.1 -1.29E+05 6.00E-06 -1.75E-01 1.10E-01 7.71E-01 10.9 -2.61E+04 1.02E-05 -1.78E-01 1.10E-01 2.67E-01 11.7 -7.32E+03 1.61E-05 -1.19E-01 1.10E-01 1.19E-01 12.7 -9.58E+03 1.46E-05 -1.32E-01 1.10E-01 1.41E-01 13.7 -4.87E+03 1.91E-05 -1.02E-01 1.10E-01 9.37E-02 14.9 -5.92E+03 1.83E-05 -1.10E-01 1.10E-01 1.09E-01 16.2 -9.66E+03 1.45E-05 -1.32E-01 1.10E-01 1.41E-01 17.75 1.10E-01 19.6 -4.72E+03 1.91E-05 -1.00E-01 1.10E-01 9.06E-02 21.85 -9.93E+03 1.44E-05 -1.33E-01 1.10E-01 1.44E-01 25.25 -1.63E+04 1.23E-05 -1.57E-01 1.10E-01 2.01E-01

PAGE 193

175 APPENDIX D STRAY LIGHT CONSIDERATION Stray light consideration is very crucial in the light sc attering experiment because stray light is able to dominate the scattering signal resulting in skew ing experimental data significantly; therefore, errors attributed to stray light must be reduced by calibration and minimization of stray light in the various wa ys discussed in Chapter 3. In addition, vertical polarizer contributes to blocking non-vertically polarized stray light from introducing into PMT. The influence of stray light was discussed here through determining theoretical photon arrival rates that photons of stray light could be incident on the PMT. The major sources of the stray light are lase r light reflected from surfaces of optical mounts and scattered light from dust particles in the beam pathway, which was illustrated in Figure D-1. First, the number of photon, ,0photonN of experimental laser flunces is calculated as 3 14 0 ,0 190.30410 8.1310 3.7410photonE J Nphoton hJphoton (D-1) where oE is the laser pulse energy m easured to be 0.304 mJ/pulse, h is Planks constant determined to be 6.6261 10-34 [Js], is frequency given by 8 141 9310 5.6410 53210 ms C s m (D-2)

PAGE 194

176 Figure D-1. Source of stray light in the light scattering optical setup. For the focusing lens, around 4% of light is reflected from bo th front and rear surfaces of focal lens. As an approximation 1% of light are reflected from beam dumps, aperture, and dichroic mirror. That is, 12% of the number of photon,,0photonN of experimental flunces may play a role as stray light. Compared to this, the contribution of scattered light from dust particles in the beam path is negligibly small. The influence of any ambient light was corrected by considering dark signal. Therefore, a total of 9.77 1013 photons have the potential to travel around the room and to be incident on the PMT after undergoing interactions in several wa ys such as reflectio n, scattering, and absorption. Some photons can be directly incident on the PMT al ong the collection optic line while others can enter into the phot on counting system from arbitrary angles. On the other hand, the number of photon in the scattering volume is calculated as Nphoton,0 Beam Dump Beam Dump Black tube Vertical Polarizer Neutral density filters Plano-convex lens, f=250 mm Aperture 532 nm Laser PMT 532 nm bandpass filter Nphoton,s

PAGE 195

177 ,0 343811 2 19 31 0.09168.11101.44102.86100.88 3.7410 7.210photonsVVNIVN h J srcmcmsr cm J photon photon (D-3) where o I is the fluences of the laser in the scattering volume measured to be 0.0916 J/cm2, is the solid angle, V is the scattering volume, is the efficiency of the system optics, and '()VVN is defined as the differential scattering coefficient of methane calibration gas which was determined to be 2.86-8 cm-1sr-1. The calculations for individual parameters are following below. 2 3 28.1110r sr S, (D-4) where the focal length is 25 cm, and the di ameter of the focal lens is 2.54 cm. 3 431.4410 6D Vcm, (D-5) where the diameter of the cross section of the beam is 0.065 cm. The efficiency of the system optics is given by 40.990.920.88. (D-6) Since the average contribution of the st ray light is around 40% of the calibration methane signal, which is the photon number of 2880, the probability that the stray light are incident on the PMT is around 2.95-11 (2880/9.77 1013). Such a dramatic reduction of stray light was ach ieved through all efforts for stray light minimization. Careless examination or lack of these efforts will tremendously increase this probability.

PAGE 196

178 APPENDIX E SOOT REDUCTION MECHANISM There are two possible mechanisms of soot suppression. The first process is direct oxidation of soot. The other process is catal ytic effect that iron species enhance the reaction between soot and oxidi zer. More possible mechanism can be experimentally found by investigating chemical state of iron species. In addition, it can be proved by determining the molar ratio of Fe to C in the flame tip. Calculations were accomplished in two different approaches here. In the first approach, the molar ratio of Fe to C was calculated by estimating the mass flow rates of soot and iron. Volume flow rates were 0.8, 2.6, and 1.5-3 liter per minute at the temperature of 298 K for N2, O2, and C8H18 respectively. At the burner lip where the temperature was assumed to be 350 K, volume flow rates became 15.7, 50.9, and 4.38 cc per second for N2, O2, and C8H18 respectively as calculated below. 3501min 0.815.7 min29860secseclKcc K (E-1) 3501min 2.650.9 min29860secseclKcc K (E-2) 0.71min 1.54.38 min0.00460secsec mlgcccc gcc (E-3) where the density of isooctan e at the temperature of 350 K was estimated from the ideal state equation, 101325 0.004 (8315/114)350g cc (E-4)

PAGE 197

179 Therefore, total volume flow rate of gases at the burner lip was 70.98 cc per second. To determine the volume flow rate at the flame tip, the temperature of the flame tip was assumed to be 1800 K, and it was assumed that air was added from outside of flame by 10 times greater volume flow rate than that of gases at the burner lip. 1800 781.224017.7 sec350secccKcc K (E-5) Volume fraction of soot at the flame tip for the unseeded flame was 3.52-7, and the density of soot is approximately 2 gram pe r cc. Therefore, the mass flow rate of soot at the flame tip was calculated as 3 73 34017.73.521022.8310 secsec cccmsootgg cmgascc (E-6) Since 66% reduction of s oot was observed, 1.56-4 mole of carbon was oxidized per second as calculated below. 341 2.83100.661.5610 sec12secgmolmol g (E-7) Meanwhile, Fe was supplied into the flame in 4000 ppm quantities by mass (~0.11% Fe per mass of fuel). Therefore, 3.4-7 mole of iron was supplied into the flame per second as calculated below. 5 6400055.8451min 11.910 min10195.84560secsec gfuelmolgg molg (E-8) 571 1.9103.410 sec55.845secgmolmol g (E-9) The molar ratio of Fe to C was 1 to 459.

PAGE 198

180 Alternatively, the molar ratio of Fe to C could be directly estimated by calculating moles of carbon and iron. 1 g of C8H18 has 8.75-3 mole as calculated below. 31 18.7510 114.23mol gmol g (E-10) Since 1 g of C8H18 has 4 1000 g of Fe(CO)5, 541 2.0410 1000195.845mol gmol g for Fe(CO)5. (E-11) Therefore, the total mole of carbon was 0.07 moles (8.75-3). Since around 25% of carbon converts to soot at smoke point 0.0175 moles of soot were in the flame. 0.0116 moles of soot were destroyed, consider ing 66% soot reduction. The molar ratio of Fe to C was 1 to 569. For direct oxidation mechanism, several possible stoichiometric chemical reactions are given by FeOCFe+CO (E-12) 22FeOC2FeO+CO, (E-13) 23FeOC2FeO+CO, (E-14) 23FeO3C2Fe+3CO, (E-15) 2322FeOC4FeO+CO, (E-16) 2322FeO3C4Fe+3CO, (E-17) 23343FeOC2FeO+CO, (E-18) 34FeOC3FeO+CO, (E-19) 342FeO2C3Fe +2CO, (E-20)

PAGE 199

181 34FeO4C3Fe+4CO, (E-21) 3422FeOC6FeO+CO. (E-22) According to these chemical reactions, one mole of iron can oxidize no more than 4 moles of carbon. Since around 500 moles of car bon are oxidized per 1 mole of iron exist in the flame, based on calculations above, th e iron additive in s oot suppression plays a role of the catalyst to enhance the chemi cal reaction between soot and oxidizers rather than directly oxidize the soot in the flame.

PAGE 200

182 APPENDIX F PROPERTIES OF IRON PENTACARBONYL The properties of iron pentaca rbonyl are presented here after the data of Alfa Products (1980). A. Physical Data Molecular formula: Fe(CO)5 Boiling point: 103C Freezing/Melting point: -25C Vapor pressure: see Figure F-1 Solubility in water: insoluble Evaporation rate (butyl a cetate = 1): greater than 1 Vapor density (air = 1):6.75 Appearance: yellow to dark red liquid Odor: nearly odorless B. Safety Data Flammability: flammable Toxicity: Toxic. Iron pentacar bonyl may cause skin and eye irritation. Inhalation and/or in gestion may cause shock, loss of consciousness, and headaches occa sionally accompanied by fever, cyanosis and cough due to pulmonary edema.

PAGE 201

183 1 10 100 1000 -20020406080100120Vapor Pressure (mm Hg)Temperature (oC) Figure F-1. Fe(CO)5 vapor pressure as a function of temperature (Gilbert and Sulzmann 1974, Trautz and Badstubner, 1929).

PAGE 202

184 Table F-1. Fe vapor pressure in Torr ( mm Hg) as a function of the flame height. Flame Height (cm) Flame temper ature (C)Vapor pressure (mm Hg) 1 0.1 1084 5.5E-06 2 0.6 1275 3.0E-04 3 1.2 1444 5.0E-03 4 1.8 1579 4.0E-02 5 2.4 1444 5.0E-03 6 3 1420 4.0E-03 7 3.7 1404 3.5E-03 8 4.5 1381 3.0E-03 9 5.25 1373 2.6E-03 10 6.1 1376 2.7E-03 11 7.1 1368 2.5E-03 12 8.3 1352 2.3E-03 13 9.4 1314 6.0E-04 14 10.6 1275 3.0E-04 15 11.6 1247 2.7E-04 16 12.6 1210 8.0E-05 17 13.8 1137 1.8E-05 18 15.1 957 1.0E-07 19 16.65 891 -* 20 18.5 829 21 20.55 780 22 22.2 750 23 23.95 385 *Vapor pressure is negligible

PAGE 203

185 0 0.01 0.02 0.03 0.04 0.05 0246810121416Fe Vapor pressure in Torr (mm Hg)Height (cm) Figure F-2. Fe vapor pressure as a function of the flame height. Over the flame height of 15 cm which is the oxidation regime, the vapor pressure is negligibly small.

PAGE 204

186 LIST OF REFERENCES Agency for Toxic Substances and Disease Registry (ATSDR). Toxicology Profile for Jet Fuels JP-4 and JP-7. U. S. Department of Hea lth and Human Services, Public Health Service, Atlanta, GA (1995). Agency for Toxic Substances and Disease Registry (ATSDR). Toxicology Profile for Jet Fuels JP-5 and JP-8. U. S. Department of Hea lth and Human Services, Public Health Service, Atlanta, GA (1998). Baranska H., Labudzinska A., and Termpinski J. Laser Raman Spectrometry: Analytical Applications. Ellis Horwood Limited. Chichester, England (1987). Biswas P., Wu, C. Y., Zachar iah, M. R., and McMillin B. Characterization of iron oxide-sillica nanocomposites in flames: Pa rt II. Comparison of discrete-sectional model predictions to experimental data, Journal of Materials Research. 12:714723 (1997). Bittner J. D., and Howard J. B. Compos ition Profiles and Reaction Mechanisms in a Near-Sooting Premixed Ben zene/Oxygen/Argon Flame, 18th Symp. (Int.) Combust. 18:1105-1116, The Combustion In stitute, Pittsburgh (1981). Bockhorn H. (ed.). Soot Formation in Combustion: Mechanisms and Models. SpringerVerlag, Berlin (1994). Bonczyk P. A. Effect of Ferrocene on Soot in a Prevaporized Iso-Octane/Air Diffusion Flame, Combust. Flame. 87:233-244 (1991). Bulewicz E. M., Evans D. G., and Padley P. J. Effect of Metallic Additives on Soot Formation Processes in Flames, 15th Symp. (Int.) Combust. 1461-1470, The Combustion Institute, Pittsburgh (1974). Charalampopoulos T. T., and Chang H. In Situ Properties of Soot Particles in the Wavelength Range from 340 nm to 600 nm, Combust. Sci. Tech. 59:401-421 (1988). Cheng M. T., Xie G. W., Yang. M., and Shaw D. T. Experimental Characterization of Chain-Aggregate Aerosol by Electrooptic Scattering, Aerosol Science and Technology. 14:74-81(1991). Chippett S., and Gray W. A. The Size a nd Optical Properties of Soot Particles, Combust. Flame. 31:149-159 (1978).

PAGE 205

187 Choi M., Cho J., Lee J., and Kim H. W. Meas urement of silica aggregate particle growth using light scattering and thermophoretic sampling in a coflow diffusion flame, Journal of Nanoparticle Research. 1:169-183 (1999). Cotton D. H., Friswell N. J., and Jenkins D. R. Suppression of Soot Emission From Flames by Metal Additives, Combust. Flame. 17(1):87-98 (1971). Dalzell W. H., and Sarofim A. F. Optical Constants of Soot and Their Application to Heat-Flux Calculations, J. Heat Transfer. 91:100-104 (1969). Dasch C. J. Continuous-Wave Probe Laser I nvestigation of Laser Vaporization of Small Soot Particles in a Flame, Combust. Flame. 23(13):2209-2215 (1984a). Dasch C. J. New Soot Diagnostics in Fl ames Based on Laser Vaporization of Soot, 20th Symp. (Int.) Combust. 1231-1237, The Combustion Institute, Pittsburgh, (1984b). Dasch C. J. The Decay of Soot Surface Gr owth Reactivity and It s Importance in Total Soot Formation, Combust. Flame. 61(3):219-225 (1985). Dasch C. J. One-Dimensional Tomography A Comparison of Abel, Onion-Peeling, and Filtered Backprojection Methods, Combust. Flame. 31(8):1146-1152 (1992). de Faria D. L. A., Venncio Silvia S., and de Oliveira M. T. Raman Microspectroscopy of Some Iron Oxides and Oxyhydroxides, J. Raman Spec. 28(11):873-878 (1997). Dobbines R. A., and Megaridis C. M. Absorption and Scattering of Light by Polydisperse Aggregates, Applied Optics. 30(33):4747-4754 (1991). Espensheid W. F., Kerker M., and Matijevie E. Logarithmic Distribution Functions for Colloidal Particles, J. Phys. Chem. 68:3093-3097 (1964). Farias T. L., Carvalho M. G., Kyl ., and Faeth G. M. Computational Evaluation of Approximate Rayleigh-Debye-Gans/ Fractal-Aggregate Theory for the Absorption and Scattering Properties of Soot, J. Heat Transfer. 117:152-159 (1995). Farmer P. B., Singh R., Kaur B., Sram R. J., Binkova B., Kalina I., Popov T. A., Garte S., Taioli E., Gabelova A., and Cebulska-Wa silewska A. Molecular Epidemiology Studies of Carcinogenic Envi ronmental Pollutants Effects of Polycyclic Aromatic Hydrocarbons (PAHs) in Environmenta l Pollution on Exogenous and Oxidative DNA Damage, Mutation Research Review s in Mutation Research. 544(23):397-402 (2003). Feitelberg A.S., Longwell J.P., and Sarofim A.F. Metal Enhanced Soot and PAH Formation, Combust. Flame. 92(3):241-253 (1993).

PAGE 206

188 Frenklach M. Reaction Mechanism of Soot Formation in Flames, Phys. Chem. 4:20282037 (2002). Frenklach M., Yuan T., and Ramachandra M. K. Soot Formation in Binary Hydrocarbon Mixtures, Energy Fuels. 2(4):462-480 (1988). Gilbert A. G., and Sulzmann K. G. P. T he Vapor Pressure of Iron Pentacarbonyl, J. Electrochem. Soc. 121, 832-834. (1974). Goldstein J. I. Scanning Electron Microscopy and X-ray Microanalysis. Plenum Press, New York (1992). Greene G. A., Finfrock, C.C. and Irvine, T. F. Total hemispherical emissivity of oxidized Inconel 718 in the te mperature range 300-1000C, Experimental Thermal and Fluid Science. 22:145-153 (2000). Hahn D. W. Soot Suppressing Mechanisms of Iron in Premixed Hydrocarbon Flames. Ph.D. dissertation. Department of M echanical Engineering, Louisiana State University, Baton Rouge (1992). Harris S. J., and Weiner A. M. Surface Growth of Soot Part icles in Premixed Ethylene/Air Flame, Combust. Sci. Tech. 31:155-167 (1983). Harris S. J., and Weiner A. M. Determina tion of the Rate Constant for Soot Surface Growth, Combust. Sci. Tech. 32:267-275 (1983). Hayhurst A. N., and Jones H. R. N. The Eff ect of Metallic Additives on Ionization and Soot Formation in Oxy-Acetylene Flames, Combust. Flame. 78:339-356 (1989). Haynes B. S., Jander H., and Wagner H. G. The Effect of Metal Additives on the Formation of Soot in Premixed Flames, 17th Symp. (Int.) Combust. 1365-1373, The Combustion Institute, Pittsburgh (1979). Haynes B. S., Jander H., and Wagner H. G. O ptical Studies of Soot-Formation Processes in Premixed Flames, Ber. Bunsen. Phys. Chem. 84(6):585-592 (1980). Haynes B. S., and Wagner H. G. Soot Formation, Prog. Energy Combust. Sci. 7(4):229-273 (1981). Incropera F. P., and Dewitt D. P. Fundamentals of Heat and Mass Transfer. John Wiley& Sons, New York (2002). Ingle J. D., and Crouch S. R. Spectrochemical Analysis. Prentice-Hall, New Jersey (1988). Jones A. R. Scattering of Electromagnetic Radiation in Particul ate Laden Fluids, J. Prog. En. Comb. Sci. 5:73-96 (1979).

PAGE 207

189 Kasper M., Sattler K., Siegmann K., Matter U ., and Siegmann H. C. The Influence of Fuel Additives on The Formation of Carbon During Combustion, J. Aerosol Sci. 30 (2):217-225 (1999). Katsouyanni K., and Pershagen G. Ambie nt Air Pollution Exposure and Cancer, Cancer Causes and Control. 8:289-291 (1997). Kerker M. The Scattering of Light and Ot her Electromagnetic Radiation. Academic Press, New York (1969). Kim H. W., and Choi M. In situ Line Meas urement of Mean Aggr egate Size and Fractal Dimension along the Flame Axis by Planar Laser Light Scattering, Aerosol Science. 34:1633-1645 (2003). Kim S. H., Fletcher R. A., and Zacharia h M. R. Understanding the Difference in Oxidative Properties between Flame and Di esel Soot Nanoparticles: The Role of Metals, Environ. Sci. and Technol. 39(11):4021-4026 (2005). Kyl ., and Faeth G. M. Optical Propert ies of Overfire Soot in Buoyant Turbulent Diffusion Flames at Long Residence Times, J. Heat Transfer, 116:152-159 (1994). Kyl ., Faeth G. M., Fa rias T. L., and Carvalho M. G. Fractal and Projected Structure Properties of Soot Aggregates, Combust. Flame, 100:621-633 (1995). Kyl ., Xing Y., and Rosner D. E. F ractal Morphology Anal ysis of CombustionGenerated Aggregates Using Angular Li ght Scattering and Electron Microscope Images, Langmuir, 11:4848-4854 (1995) Krishnan S. S., Lin K. C., and Faeth G. M. Optical Propertie s in the Visible of Overfire Soot in large Buoyant Tur bulent Diffusion Flames, J. Heat Transfer, 122:517-524 (2000). Krishnan S. S., Lin K. C., and Faeth G. M. Extinction and Scattering Properties of Soot Emitted from Buoyant Turbulent Diffusion Flames, J. Heat Transfer., 123:331339 (2001). Landolt-Bornstein Zahlenwerte und Funkti onen aus Physik, Chemie, Astronomie, Goephysik und Technik. 11.8, 6-884 (1962). Lemmon E.W., McLinden M.O., and Friend D.G. Thermophysical Properties of Fluid Systems, NIST Chemistry WebBook, NIST Standard Reference Database No. 69, (Linstrom, P. J., and Mallard, W.G. (eds.)). National Institute of Standards and Technology, Gaithersburg MD (March 2003). http://webbook.nist.gov Linteris G. T., Knyazev V. D., and Babus hok V. I. Inhibition of Premixed Methane Flames by Manganese and Tin Compounds, Combust. Flame. 129:221-238 (2002).

PAGE 208

190 Masiello A. K. Study of Soot Formation in an Iron -Seeded Isooctane Diffusion Flame Using In Situ Light Scattering. Master Thesis. Department of Mechanical and Aerospace Engineering, University of Florida, Gainesville (2004). Maslar J. E., Hurst W. S., Hendr icks J. H., and Aquino M. I. In situ Raman spectroscopic investigation of aqueous iron corrosion at elevated temperatures and pressures, Journal of The Electrochemical Society. 147:2532-2542 (2000). Mavrodineanu R., and Boiteux H. Flame Spectroscopy. John Wiley& Sons, New York (1965). McMillin B. K., Biswas P ., and Zachariah M. R. In situ characterization of vapor phase growth of iron oxide-silica nanocomposites: Part I. 2-D planar laser-induced fluorescence and Mie imaging. Journal of Materials Research, 11:1552-1561 (1996). McNesby K. L., Daniel R. G., Morris J. B., and Miziolek A. W. Tomographic Analysis of CO Absorption in a Low-Pressure Flame, Appl. Opt. 34(18):3318-3324 (1995). Mie G. Beitrage zur Optik Truber Medien Speziell Kolloidal er Metallosungen, Ann. Physik. 25:377-445 (1908). Miller J. A., and Melius C. F. Kinetic and Thermodynamic Issues in the Formation of Aromatic-Compounds in Flames of Aliphatic Fuels, Combust. Flame. 91(1):2139 (1992). Mitchell S. A., and Hackett P. A. Chemical reactivity of iron atoms near room temperature, Chem. Phys. 93:7822-7829 (1990). Mountain R. D., and Mulholland G. W. L ight Scattering from Simulated Smoke Agglomerates, Langmuir. 4:1321-1326 (1988) National Academy of Sciences, December 2003, Soot Worse for Global Warming than Thought, Proceedings of the National Academy of Sciences, New Scientist.com News Service, http://www.newscientist.com/article.ns?id=dn4508 last accessed June 15, 2005. Nguyen, Q. V., Dibble, R. W., Carter, C. D., Fiechtner, G. J., and Barlow, R. S. RamanLIF measurements of temp erature, major species, OH, and NO in a Methane-air Bunsen flame, Combustion and flame. 105:499-510 (1996). Palmer H. B., and Cullis H. F. The Formation of Carbon from Gases. Chemistry and Physics of Carbon, Vol. 1, p. 265, Marcel Dekker, New York (1965). Pluchino A. B., Goldberg S. S., Dowling J. M., and Randall C. M. Refractive-Index Measurements of Single Micron -Sized Carbon Particles, Appl. Optics. 19(19):3370-3372 (1980).

PAGE 209

191 Pope C. A., Burnett R. T., Thurston G. D ., Thun M. J., Calle E. F., Krewski D., and Godleski J. J. Cardiovascular mortality and long-term exposure to particulate air pollution Epidemiological evidence of general pathophysiological pathways of disease, Circulation 109 (1):71-77 (2004). Prado G., and Lahaye J. Physical Aspects of Nucleati on and Growth of Soot Particles. Particulate Carbon (Siegla, D. C., and Smith G. W. (eds.)). Plenum Press, New York, 143 (1981). Rayleigh Lord. On the Light From th e Sky, Its Polarization and Colour, Philos. Mag., 41:107-120,274-279 (reprinted in Scientific Papers by Lord Rayleigh, Vol. I, No. 8, Dover, New York, 1964). Ritrievi K. E., Longwell J. P., and Sarofim A. F. The Effects of Ferrocene Addition on Soot Particle Inception and Growth in Premixed Ethylene Flames, Combust. Flame. 70(1):17-31 (1987). Rudder R. R., and Bach D. R. Raylei gh Scattering of Ruby-La ser Light by Neutral Gases, J. Opt. Soc. Amer. 58(9):1260-1266 (1968). Samson R. J., Mulholland G. W., and Gentry J. W. Structural Analysis of Soot Agglomerates, Langmuir. 3:272-281 (1987) Son H. S., Lee K., Shin S. K., and Ku, J. K. Radiative lifetimes of the FeO orange system, Chemical Physics Letter. 320:658-664 (2000). Sorensen C. M., Lu N., and Cai J. Fractal Cluster Size Distribution Measurement Using Static Light Scattering, Colloid Interface Sci. 174:456-460(1995). Sorensen C.M. Light Scatte ring by Fractal Aggregates: A Review, Aerosol Science and Technology. 35:648-687 (2001). Telle H. H., Beddows D. C. S., Morris G. W., and Samek O. Sensitive and selective spectrochemical analysis of metallic samp les: the combination of laser-induced breakdown spectroscopy and laser-in duced fluorescence spectroscopy, Spectrochimica Acta Part B. 56:947-960 (2001). Trautz V. M. and Badstubner W. Die Dampfdrucke und Verdampfungswarmen des Eisenpentacarbonyls Fe(CO)5. Z. Elektrochemie. 35, 799-802. (1929). Turns, S. R. An Introduction to Combustion, 2nd ed. McGraw Hill, Boston (2000). U. S. Environmental Protection Agency, Ma y 2003, EPA PM 2.5 Weighing Laboratory PM 2.5 Objectives and History, Region 4 Science & Ecosystem Support Division, East Coast Weighing Laboratory, http://www.epa.gov/region4/sesd/pm25/p2.htm last accessed June 15, 2005.

PAGE 210

192 U. S. Environmental Protection Agency, Se ptember 2003, PM How Particulate Matter Affects the Way We Live and Breathe, Office of Air Quality Planning and Standards, original publication November 2000, http://www.epa.gov/air/urbanair/pm/index.html last accessed June 15, 2005. U. S. Environmental Protection Agency, August 2003, 2002 Air Quality Trends Summary Report, EPA 454/K-03-001, Office of Air Quality Planning and Standards, Research Triangle Park, NC. Wang G., and Sorensen C. M. Experimental Test of the Rayleigh-Debye-Gans Theory for Light Scattering by Fractal Aggregates, Applied Optics. 41(22):4645-4651 (2002) Warnatz J., Mass U., and Dibble R. W. Combustion Physical and Chemical Fundamentals, Modeling and Simulation, Expe riments, Pollutant Formation, 3rd ed. Springer-Verlag, Berlin (2001). Wei Y. L., and Lee J. H. Manganese Sulfat e Effect on PAH Formation from Polystyrene Pyrolysis, The Science of the Total Environment. 228:59-66 (1999). Williams D. B., and Carter C. B. Transmission Electron Microscopy. Plenum Press, New York (1996). Wu J. S., Krishnan S. S., and Faeth G. M. R efractive Indices at Vi sible Wavelengths of Soot Emitted from Buoyant Turbulent Diffusion Flames, J. Heat Transfer. 119:230-237 (1997) Yang. G. S. Study on the effect of iron on PM10 formation and design of a particlegenerating system using a con centric diffusion burner flame, Journal of the Air& Waste Management Association. 54:898-907 (2004). Yoder G. D., Diwakar P. K., and Hahn D. W. Assessment of Soot particle Vaporization Effects during Laser-Induced Incande scence with Time-Resolved Light Scattering, Appl. Opt. 44(20):4211-4219 (2005). Zhang J., and Megaridis C. M. Soot Suppr ession by Ferrocene in Laminar Ethylene/Air Nonpremixed Flames, Combust. Flame. 105(4):528-540 (1996).

PAGE 211

193 BIOGRAPHICAL SKETCH Kibum Kim, the elder of Yong-Hyun Ki m and Pill-Young Kims two sons, was born in Dae-Jeon, Republic of Korea, on Febr uary 4, 1975. He grew up and spent most of his early life in the city. In 2000, he was awarded a bachelors degree in naval architecture and ocean engineering from C hung-Nam National University in the Republic of Korea. In his senior year, he built the first Korean human-powered ship with water foils. The next year, he was given the oppor tunity to experience American culture and study English at the University of Alabama. In August 2001, he began graduate study at the University of Florida. Kibum gradua ted with a Master of Science degree in mechanical engineering from the University of Florida in August 2003. He went on to earn his Ph.D. in mechanical e ngineering, also at the University of Florida, in August 2006.


Permanent Link: http://ufdc.ufl.edu/UFE0015080/00001

Material Information

Title: Interaction of Iron Species and Soot Particles in an Isooctane Diffusion Flame
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015080:00001

Permanent Link: http://ufdc.ufl.edu/UFE0015080/00001

Material Information

Title: Interaction of Iron Species and Soot Particles in an Isooctane Diffusion Flame
Physical Description: Mixed Material
Copyright Date: 2008

Record Information

Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
System ID: UFE0015080:00001


This item has the following downloads:


Full Text












INTERACTION OF IRON SPECIES AND SOOT PARTICLES IN AN ISOOCTANE
DIFFUSION FLAME















By

KIBUM KIM


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2006

































Copyright 2006

by

Kibum Kim

































This work is dedicated to my family. Their support, encouragement and love made its
completion possible.















ACKNOWLEDGMENTS

First and foremost, I am deeply grateful to Dr. David Hahn for his guidance and

leadership during this study. Moreover, his encouragement always kept me looking at

the bright side. His generosity and patience with my numerous mistakes in English and

research allowed me to challenge myself without hesitation.

I would like to acknowledge the invaluable advice and suggestions of my

committee members. I especially want to express my appreciation to Dr. Jill Peterson for

her thoughtful concern about my school life.

I would also like to thank all of my lab mates (Leia Shanyfelt, Prasoon Diwakar,

Cary Henry, Brett Windom, Philip Jackson, Soupy Alexander, Amy Twining, Chris

Macarian, and Jeff Crosby) for their help and assistance while I was conducting my

research. Their solidarity and friendship made lab life more enjoyable, and gave me a

great opportunity to learn American culture including sports activities and insightful

conversations. Special thanks also go to Kathryn Maseillo and Prasoon Diwakar for their

valuable input and cooperation in my combustion research.

I would also like to thank my parents for their unconditional love and constant

support. In addition, I sincerely thank my brother, Kee-Hoon for his concern and for

stimulating me to even greater effort. Last but not least, words cannot express my

gratitude to my lovely wife, Yong-Soon; and my adorable son, Daniel. At all times,

Yong-Soon was a great support emotionally and mentally as I went through the ups and

downs of private and professional life. My sweet little boy motivates me to work









constantly harder. I am really thankful that I could be with my family during the entire

period of my study overseas.
















TABLE OF CONTENTS



A C K N O W L E D G M E N T S ................................................................................................. iv

LIST OF TABLES ........... ........ ............................... .. ...... ........ ix

L IST O F FIG U R E S .... ...... ...................... ........................ .. ....... .............. xii

ABSTRACT ........................................ xvii

CHAPTERS

1 IN TR OD U CTION ............................................... .. ......................... ..

1.1 Soot F orm action ................ .... ....... ................ ... .............. ...... ............... 3
1.1.1 Formation of Soot Precursor Molecules.....................................................4
1.1.2 Particle C oagulation and G row th ........................................ .....................7
1.1.3 Particle A gglom eration.......................................... ........... ............... 9
1.1.4 Soot O xidation ................. ........ .............................. ... ... ... ............ 10
1.2 Soot Suppression with Transition Metallic Additives.......................................11
1.2.1 M anganese-Based A additives ........................................ ............... 13
1.2.2 Iron-B ased A dditives........................................... .......................... 14
1.2.2.1 A dditives in prem ixed flam es ............................... ............... .14
1.2.2.2 A dditives in diffusion flam es ........................... ....................... 17
1.3 Studies of the Fractal Properties and the Structure of Soot Aggregates .............19
1.4 Spectroscopic M ethod .............................................................. .....................22
1.5 Objectives of Present R research ........................................ ........ ............... 25

2 FUNDAMENTAL SCIENCE AND BACKGROUND THEORY ............................27

2.1 E plastic L eight Scattering T theory ..........................................................................27
2.1.1 R ayleigh Scattering Theory ............................................ ............... 29
2 .1.2 Sy stem s of P articles............... ....... .............................................. ......3 1
2.1.3 Rayleigh-Debye-Gans (RDG) Scattering Theory .......................... 33
2.1.3.1 Rayleigh-Debye-Gans (RDG) scattering approximation ................33
2.1.3.2 Evaluation of the extinction coefficient .......................................35
2.1.4 Sampling and Analyzing Soot Aggregate ...........................................38
2.1.4.1 Therm ophoretic sam pling .................................... ............... 38
2.1.4.2 Transm mission electron microscopy .................................................39
2.1.4.3 Energy dispersive x-ray spectroscopy (EDS)...............................41









2.2 Spontaneous Raman Scattering Theory............................................................42
2.3 Laser Induced Fluorescence Theory ................................................................ 47

3 EXPERIMENTAL APPARATUS AND METHODS .............. ............... 51

3.1 B urner Sy stem ..................................................... ................. 5 1
3.2 Fuel Vaporization and Delivery System .............................. .... ............53
3 .3 F la m e ...................................................................5 7
3.4 Optical System s and D iagnostics.................................... ......................... 60
3.4.1 Light Scattering System ........................................ ........................ 60
3.4.2 Light Scattering Calibration ............................ ................................... 67
3.4.3 Signal Processing.......... ...................................... ...... ...... .............. 70
3.5 Laser Pow er M easurem ent ............................................................................ 72
3.6 Transmission ...................... ............................... ... .... ......... 73
3.7 Thermophoretic Sampling and Transmission Electron Microscopy ...................75
3.7.1 Therm ophoretic Sam pling ........................................ ...... ............... 76
3.7.2 Transmission Electron M icroscope ................................. ............... 77
3.8 Spectroscopic Techniques ............................................................................78
3.8.1 Preliminary CO Flame Study ............................. .. .................. 79
3.8.2 Experimental Apparatus of Laser Induced Fluorescence Spectroscopy ...81
3.8.3 Experimental Apparatus of In Situ Raman Spectroscopy .......................83
3.8.4 Isooctane Flam e Study ........................................ ......................... 86

4 INTEGRATED RESULTS AND DATA ANALYSIS...........................................90

4 .1 Sm ok e P oin t Stu dy ................................................................... ................. .. 9 0
4.2 Elastic Light Scattering R esults...................................... ......................... 93
4.3 Transm mission R results .................................................... .................. ............... 97
4.4 Soot Characteristics Determined from RDG Theory..........................................99
4.4.1 Transm mission Electron M icroscopy ....................... ......... .....................99
4.4.2 Fractal Properties of Soot Aggregates...............................................104
4.4.2 Primary Soot Particle Size .............................................. ...............107
4.4.3 Num ber D density of Particles ....................................... ............... 109
4.4.4 Volum e Fraction of Soot Particle................. ...................... ...............110
4.4.5 The Extinction Coefficient of Soot Particle .........................................112
4.4.6 D discussion of R results ................................... ................ ..................... 114
4.5 Spectroscopy ............... ...... ..................... ............... ............ 119
4.5.1 Laser Induced Fluorescence (LIF) Spectroscopy ...............................122
4.5.2 In Situ Ram an Spectroscopy....................................... ....................... 130

5 NUM ERICAL ANALYSIS........................... ................................ ............... 136

5.1 Thermodynamic Equilibrium Calculations............................ ...............136
5.1.1 Flam e T em perature....................................................... ............... 136
5.1.2 0 2 Flow R ates .................. .............................. .. .. .... .. ............ 142
5.1.3 Fe(CO )5 Concentrations ........................................ ....... ............... 143









6 CONCLUSIONS AND FUTURE WORK ........................................................146

6.1 Sum m ary and C conclusions ....................................................... ..... .......... 146
6.2 Future W ork .................................................................... ........ 152

APPENDICES

A ANALYSIS OF THE FLAM E ...................................................... ...................153

B RESULTS OF RDG CALCULATIONS ..............................156

C E R R O R A N A L Y SIS ....................................................................... .................... 168

D STRAY LIGHT CONSIDERATION ............................................ ...............175

E SOOT REDUCTION M ECHANISM ............................................. ... ............ 178

F PROPERTIES OF IRON PENTACARBONYL................................... ................182

L IST O F R E F E R E N C E S ........................................................................ .................... 186

BIOGRAPH ICAL SKETCH .............................................................. ............... 193
































viii
















LIST OF TABLES


Table page

1-1 M etallic additives in com m on. ........................................................................... 12

1-2 Fractal dimension of various aggregates ................................................23

2-1 Raman shifts and the emission wavelengths of common species ..........................47

3-1 D ata collection heights. ................................................ ................................. 54

3-2 Summary of equipment for fuel vaporization and delivery system .......................58

3-3 Summary of gases and fuel used in the study. ......................................................59

3-4 Description of the flame operating conditions. ....................................................60

3-5 Components of scattering system apparatus. ........................................................62

3-6 Real optical densities for various ND filters .................................... .................65

3-7 The usage of the ND filters for individual height. .................................................66

3-8 Average of the number densities, differential scattering cross sections, and
scattering coefficient sets for methane and nitrogen calibration gases at 1 atm
with standard deviation for 24 experimental........................ ..................69

3-9 Average results of calibration gas signal including stray light, a calibration ratio,
stray light signal, the percentage of stray light, and the ideal reference ratio
along with the standard deviation over all scattering experiments ........................71

3-10 Summary of laser beam power properties for light scattering measurements. ........73

3-11 Description of transmission apparatus. ....................................... ............... 74

3-12 Components of spectroscopic system apparatus. .............................................. 87

3-13 Data collection heights for spectroscopy. ..................................... ............... 88

3-14 Components of the system apparatus for absorption spectroscopy ..........................89









4-1 Average of 4 oxygen flow rates with their standard deviation and relative
standard deviation, the equivalence ratio, and oxygen to fuel ratio for 10
different con centration s ........................................ .............................................92

4-2 Average (N=10) K'w results of the unseeded and seeded flame and standard
deviations. Flame heights are measured from the burner lip. ................................95

4-3 Average (N=6) transmission results of the unseeded and seeded flames and
standard deviations. Flame heights are measured from the burner lip ....................98

4-4 The summary of the fractal dimension at all heights in the unseeded and seeded
flam e s. .......................................................................... 10 7

4-5 Diameters of primary soot particle at each height in the unseeded and seeded
flam e s. .......................................................................... 10 8

4-6 The summary of number densities for the unseeded and iron-seeded flames........ 110

4-7 The volume fraction as a function of flame height for the unseeded and iron-
seeded flam es. .................................... ................................ ....... 112

4-8 The extinction coefficient of soot particle as function of flame height for the
unseeded and iron seeded flam es. .................................... ............. ............... 113

4-9 Complex refractive indices for soot from various sources. (2001)........................118

4-10 EDS result atomic ratio of iron oxide ............... ......... ...................121

4-11 Fe resonance transition wavelengths and corresponding fluorescence emission
lines w ith their relative intensity. ........................................ ....................... 122

4-12 Reference to iron oxides Raman shift (cm-1) ............................... ............... .131

4-13 Fe atom ic em mission peaks............................................... ............. ............... 132

4 -14 L IB S em mission p eak s ......... ............................................................................... 134

5-1 Mole of reactants used for input in the STANJAN code. .......... ...............137

5-2 Products from STANJAN simulation. ...................................... ............... 137

B-l M measured radius of the prim ary soot particle....................................................... 156

B-2 The differential scattering cross section (cm2/sr). .............................................157

B-3 Summary of calculated results for the unseeded flame............... ... ...............158

B-4 Summary of calculated results for the seeded flame ...........................................158









B-5 Uncorrected and corrected differential scattering coefficients (cm-sr-1) ...............159

B-6 Differential scattering cross section for a fractal aggregate (cm2/sr). ..................159

B-7 Number density of soot aggregates in the scattering volume (particles/cm3)........160

B-8 Total scattering cross section for a primary soot particle (cm2) .............................161

B-9 Total scattering cross section of an aggregate (cm2) ............................................161

B-10 Absorption cross section of a primary particle (cm2).........................................162

B-11 Absorption cross section of an aggregate (cm2).............................. .................. 162

B-12 The extinction cross section of an aggregate (cm2)................................................163

B-13 The extinction coefficient (cm -1) ............................................... ................. 163

B-14 Number density of soot particles in the scattering volume (particles/cm3)...........167

B-15 The volume fraction of soot particles in the scattering volume (cm3 soot/cm3). ...167

C-1 Summary of the calculated parameters with Equations C-6 through C-9 for the
unseeded flam e. ...................... ........................ .. .... ................. 169

C-2 Summary of the calculated parameters with Equations C-6 through C-9 for the
seeded flame. ...................................... .. .. ........... 170

C-3 Results of the calculation using Equation C-12 for the unseeded flame............. 171

C-5 Summary of calculated errors at each height for the particle size and number
density ............................................................................ 172

C-6 Summary of calculated parameters at each height for the unseeded flame............ 174

C-7 Summary of calculated parameters at each height for the seeded flame.............. 174

F-l Fe vapor pressure in Torr (mm Hg) as a function of the flame height................. 184
















LIST OF FIGURES


Figure page

1-1 Transmission electron microscope (TEM) images of soot aggregates from
isooctane combustion. A) At 100 nm scale. B) At 0.2 |tm scale. ..........................4

1-2 Soot formation. Adapted with permission from a reference (Bockhom 1994)..........5

1-3 The H-abstraction-C2H2-addition mechanism acting on a biphenyl molecule...........8

1-4 Two processes of particle growth. A) Particle coagulation. B) Particle
agglom eration............................................................................................. 10

1-5 Soot formation regimes in a diffusion flame, the axial soot concentration profile
at the center of the flame, and the radial soot concentration profile at an arbitrary
flame height ................ .......... ................. ......... 12

2-1 Light scattering response to an incident light...................................... ..................28

2-2 Schem atic of TE M ........................... ........ ..................... .. ...... .... ............40

2-3 Energy level diagrams representing elastic scattering transitions and several
inelastic Raman scattering transitions. A) Elastic scattering. B) Resonance
Raman scattering. C) Stokes Raman scattering. D) Anti-Stokes Raman
scatterin g ......................................................... ................ 4 3

2-4 Relationship between Rayleigh and Raman scattered lines in a scattering
spectrum. Source: Ingle and Crouch 1998. ................................... ............... 45

2-5 Energy level diagram of the fluorescence process for atoms or molecules. ............48

3-1 Concentric diffusion burner schematic. A) Side view. B) Top view. Oxygen
goes into the system through the annulus array of ports whereas isooctane and
nitrogen flow through the tube in the center. ................................ .................52

3-2 Concentric diffusion burner. A) Side view. B) Top view. .....................................53

3-3 D ata m easurem ent heights. ............................................. ............................. 55

3-4 Fuel vaporization system schem atic................................... .......................... 56

3-5 F uel vaporization sy stem ............................................................................ ... .... 56









3-6 Alicat Scientific digital flow meters employed for regulating the flow rates of
nitrogen coflow and oxygen. ........................................ .......................................58

3-7 Chem ical structure ofisooctane. ........................................ ......................... 59

3-8 Photomultiplier tube. A series of dynodes between cathode and anode provide
in te rn a l g a in ..............................................................................................................6 4

3-9 Sample scattering signals from methane, nitrogen, and flame. Calibration gases
are attenuated by a factor of 100.3 and flame signal is attenuated by a factor of
105.43 for signal linearity. ..... ........................... ......................................... 70

3-10 Top view of the transmission system setup................................... ............... 73

3-11 A setup ofthermophoretic sampling and grid. A) Side view. B) Formvar
carbon-supported 150 mesh copper grid. ...................................... ............... 76

3-12 Photograph of the TEM system ......................................................... .....................78

3-13 Vaporization system of iron pentacarbonyl and a CO flame burner......................79

3-14 A photograph of the iron pentacarbonyl vaporization vessel and the heater...........80

3-15 Photographs of CO flame. A) unseeded flame, B) iron seeded flame ...................81

3-16 The optical setup for laser-induced fluorescence spectroscopy .............................82

3-17 The optical set-up for in situ Raman spectroscopy. ............................................83

3-18 A photograph of an optical set-up inside OPO ....................................................... 84

3-19 The optical set-up for absorption spectroscopy ......... ...................... ............... 89

4-1 Plot of oxygen flow rate at the smoke point as function of time. ............................91

4-2 Smoke point, as measured by the corresponding oxygen to fuel ratio and the
equivalence ratio, as a function of iron pentacarbonyl concentration. Note that
the equivalence ratio increases due to a reduction of the necessary oxygen
q u an tity ...................... .. .. ......... .. .. ............................................... 9 3

4-3 Typical scattered signal response from photomultiplier tube measuring
calibration gases and flames. Calibration gas signals are attenuated by a factor
of 100.3, and flame signals are attenuated by a factor of 105 to preserve signal
linearity..................................... .................. ................ ......... 94

4-4 Unseeded and seeded differential scattering coefficients in logarithmic scale.
Error bars represent one standard deviation. .......................................................96

4-5 Transmission through the unseeded and seeded flames ........................................99









4-6 Transmission electron micrographs of soot particles at different axial positions. .100

4-7 A log-log plots ofNversus Rg/dp 25 soot aggregates were sampled at the height
7 in the unseeded flam e. ............................................... ............................... 106

4-8 A log-log plots ofNversus Rg/dp 25 soot aggregates were sampled at the height
7 in the seeded flam e. ........................... .................... .. ......... ...............106

4-9 Diameters of the primary soot particle as a function of the flame height in the
unseeded and seeded flames. A polynomial curve fit was used for extracting
m ore accurate values of dpar. .................................. ................................... 109

4-10 The number of primary soot particle as a function of the flame height in the
unseeded and iron-seeded flames. A logarithmic curve fit was used for
extracting more accurate values of Np ............. .............................................. 111

4-11 Number density of the total soot particle for the unseeded and iron-seeded
flam es. .................................... ................................... ............... 111

4-12 The volume fraction as a function of flame height for the unseeded and iron-
seeded flames. The error bar represents one standard deviation...........................113

4-13 The extinction coefficient of soot particle as a function of flame height for the
unseeded and iron-seeded flam es. ........ ............................................................... 114

4-14 Photographs of tips of the unseeded and seeded flames. Soot plume is seen in
the unseeded flame while being not seen in seeded flame. A) the unseeded
flam e. B ) the seeded flam e. ........................................................................ ....... 116

4-15 Photographs of the unseeded and seeded flames. A) the unseeded flame. B) the
seeded flam e ............................................................................................ 117

4-16 TEM images of samples collected in the Fe-seeded CO flame. A) sampled at the
middle of the flame height. B) sampled at the flame tip ....................................... 120

4-17 The typical signal window of EDS. ............................................ ............... 120

4-18 Energy level diagram of Fe atom. Bold font indicates the best combination. ......123

4-19 Laser induced fluorescence peak for three excitation sources at the two third of
normalized CO seeded flame height. Excitation lines are shown.......................124

4-20 Fe fluorescence corresponding to the excitation line of 296.69 nm as a function
of the CO flame normalized four heights.................................. .................. .....124

4-21 Intensity of LIF measured in isooctane seeded flame at emission line of 373.49
nm corresponding to the excitation line of 296.69 nm. Flame tip is at height of
23.95 cm. Error bars represent one standard deviation......................................125









4-22 Transmission of Fe atomic light passing through the seeded flame at two
different heights. Fe resonance transition line of 271.9 nm was chosen for this
study. Flam e tip is at the height of 23.95 cm ...................................................... 127

4-23 Transmission of Fe atomic light from the lamp as a function of 5 different
heights of the seeded flame. Fe resonance transition line of 271.9 nm was
chosen for this study. Flame tip is at the height of 23.95 cm.............................127

4-24 Spectra measured as function of the incident laser energy at flame tip using the
355 nm source in order to validate the LIBS effect on the LIF signal................... 129

4-25 On-and-off resonant LIF signal induced by the excitation wavelength of 296.69
nm and 296.19 nm with the same pulse energy. ............. .................................... 129

4-26 Energy level diagram of FeO molecule....................................... ............... 130

4-27 A spectrum obtained from in situ Raman experiment of CO flame using 532 nm
as an excitation source............. .... ................................................ .. .... .... ..... 132

4-28 Spectra obtained from in situ Raman experiment of CO flame using 355 nm as
an excitation source at four different heights. .................. ....................... 133

4-29 LIBS emission spectrum obtained from steel rod using 355 nm as an excitation
so u rce .......................................................................... 13 4

5-1 Relative mass fraction of products as a function of temperature...........................138

5-2 Relative mass fraction of Fe species as a function of temperature. .......................139

5-3 Relative mass fraction of Fe as a function of temperature................................ 139

5-4 Flame temperature as a function of flame height...............................................140

5-5 Relative mass fraction of species as a function of temperature. ............................141

5-6 The decrease in relative mass fraction of the solid carbon as a function of the
oxygen flow rate .......................................................... .. ........ .... 143

5-7 Relative mass fraction of the iron species as a function of the oxygen flow rate.. 144

5-8 Mass fraction of carbon as a function of the Fe(CO)5 concentration...................145

6-1 Schematic of soot oxidation mechanism. A) Soot oxidation without Fe. B) Soot
oxidation w ith Fe .......................................... ....... ........ .. ........ .. .. 49

6-2 Schematic of surface reaction mechanism of hydrogen oxidation. Three steps of
the mechanism are adsorption, surface reaction, and desoprtion.........................150









B-l The differential scattering coefficients measured along the radial positions at
three different heights. Error bars represent one standard deviation......................164

B-2 The extinction coefficients as a function of the unseeded flame height. The
extinction coefficients determined using RDG scattering theory for two different
refractive index were compared with that from transmission experiments............165

B-3 The extinction coefficients as a function of the seeded flame height. The
extinction coefficients determined using RDG scattering theory for two different
refractive index were compared with that from transmission experiments............166

D-1 Source of stray light in the light scattering optical setup. ......................................176

F-l Fe(CO)5 vapor pressure as a function of temperature (Gilbert and Sulzmann
1974, Trautz and Badstubner, 1929). ........................................ ............... 183

F-2 Fe vapor pressure as a function of the flame height. Over the flame height of 15
cm which is the oxidation regime, the vapor pressure is negligibly small.............85















Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy

INTERACTION OF IRON SPECIES AND SOOT PARTICLES IN AN ISOOCTANE
DIFFUSION FLAME

By

Kibum Kim

August 2006

Chair: David W. Hahn
Major Department: Mechanical and Aerospace Engineering

Metallic fuel additives have been considered for soot emission control over the last

few decades. However, the exact mechanisms of soot reduction are poorly understood

and remain controversial. In response to the need for elucidating the correct chemical

processes, elastic light scattering, laser-induced fluorescence, and thermophoretic

sampling followed by transmission electron microscopy analysis were carried out in a

laboratory-scale isooctane diffusion laminar flame seeded with 4000 ppm iron

pentacarbonyl as the metallic additive. These measurements yielded the size, number

density, and volume fraction of soot particles throughout the flame, including formation

and oxidation regimes. In comparison to the scattering parameters extracted from the

unseeded flame, the soot suppression effects of iron pentacarbonyl can be determined to

act primarily in the regime of soot burnout or oxidation. It is concluded that the additive

has no direct effect on perturbation of soot in the soot growth zone of the flame, while









having a significant effect on soot in the burnout zone of the flame, namely enhanced

oxidation, realizing an overall soot suppression effect.

In addition to the elastic scattering, laser-induced fluorescence and in situ Raman

spectroscopy were performed to identify the state of the iron additive in the seeded flame.

The results of the spectroscopic techniques reveal that the dominant iron species

throughout the primary flame region was Fe, rather than any form of iron oxide.

Moreover, elemental iron was observed to diminish through the soot oxidation region.

The primary conclusion is that the catalytic effect of Fe atoms and possibly iron oxides

enhanced soot oxidation in the burnout regime of the flame, thereby reducing the overall

soot emissions. Consistent with this, the noted reduction in smoke point with the addition

of iron was also observed.


xviii














CHAPTER 1
INTRODUCTION

Particulate matter (PM) is the term describing small particles found in the air such

as dust, dirt, liquid droplets, smoke, and soot. These particles are emitted directly into the

air from a variety of sources and are also formed in the air through chemical reactions.

Sizewise, particles less than 2.5 |tm in diameter are called PM2.5 (or fine particulate

matter). Because such fine particles are linked to both human health concerns and

environmental issues, various efforts have been made and many scientific studies have

been done to find a way to decrease the production rates of fine particles. As a part of

these efforts, the Environmental Protection Agency enacted National Ambient Air

Quality Standards (NAAQS) for PM and declared that the annual average level of PM2.5

particles in the air should not exceed 15 micrograms per cubic meter

(http://www.epa.gov/region4/sesd/pm25/p2.htm). Consequently, significant reductions

have been achieved over the last two decades, however, more efforts are needed to ensure

that the air is safe enough not to affect human health and the environment.

As far as human health is concerned, inhaling PM causes a broad range of illness

such as asthma, acute or chronic bronchitis, shortness of breath, painful breathing,

respiratory and heart illness, diminished lung function, and even premature mortality

(http://www.epa.gov/air/urbanair/pm/index.html). Due to the small size of these

particles, they are capable of penetrating and accumulating in the respiratory system. It is

supported by a recent study that particulate pollutants increase the incidence of

cardiopulmonary diseases and ischemic heart attack (Pope et al. 2004). A specific type









of PM is soot particles, which are rich in amorphous carbon and polycyclic aromatic

hydrocarbons (PAHs), and are known to be mutagenic and carcinogenic (Katsouyanni

and Pershagen 1997, Farmer et al. 2003).

In addition, PM has a harmful influence on the environment in many ways. For

example, it leads to atmospheric haze resulting in reduction of visibility in many parts of

the US. It also may play a role in acid rain, which may be responsible for a range of

problems. When PM settles on soil and water, it changes the nutrient and chemical

balance that are responsible for depleting ecosystems and ruining sensitive forests and

farm crops. According to the latest studies, soot is twice as potent as carbon dioxide in

contributing to global warming resulting from the green house effect because it can

darken snow and ice that results in absorption of solar energy rather than reflection

(http://www.newscientist.com/article.ns?id=dn4508). Such harmful impacts of PM can

impact the broad areas because it can travel long distance from the sources (US EPA.

2003).

A major source of PM is soot, usually produced through incomplete combustion

processes. Controlling these combustion processes is a key method to reduce soot

production. There has been much interest in better understanding soot formation and

methods of soot reduction. Soot reduction would benefit the health of those exposed to

soot, for instance, ground crews working at the airport or on aircraft carriers. The

moment a jet takes off, the engine thrust and fuel consumption rate are at maximum. As

a result, soot emissions also are at maximum, and ground crews are exposed to high

levels of soot in the exhaust gas from jet engines. Short- and long-term health effects of

this exposure are serious concerns, and a means of reducing soot in turbine engines is of









great interest. While the performance of the engine is preserved at an optimum,

suppression of malignant soot emissions is most desirable. One approach to achieve this

is to increase the temperature of the combustion process, resulting in promotion of soot

oxidation. Another way is to raise the local air-to-fuel ratio. However, these methods

have shown the disadvantage of increasing the amount of NOx formed. As a potential

solution, soot suppression via fuel additives is an alternative area for exploration.

1.1 Soot Formation

Soot composed of carbonaceous particles is usually observed in flames and fires as

orange luminescence during combustion of hydrocarbon fuels. Soot particles are mostly

found as agglomerates of primary particles typically no larger than 500A. The hydrogen

to carbon ratio in soot ranges between 1:8 and 1:10. Physical characteristics of soot are

described in detail by Palmer and Cullis (1965:p265).

The carbon formed in flames generally contains at least 1% by weight of hydrogen.
On an atomic basis this represents quite a considerable proportion of this element
and corresponds approximately to an empirical formula of CsH. When examined
under an electron microscope, the deposited carbon appears to consist of a number
of roughly spherical particles, strung together rather like pearls on a necklace. The
diameters of these particles vary from 100 to 2000 A and most commonly lie
between 100 and 500 A. The smallest particles are found in luminous but
nonsooting flames, while the largest are obtained in heavily sooting flames.

A size distribution of individual soot particles is well modeled by a log-normal

distribution (Haynes et al. 1981). The average diameter of soot particles corresponds to

about one million carbon atoms. Figure 1-1 shows typical soot images taken as part of

this study by transmission electron microscopy (TEM) at two magnifications.

Soot formation is a kinetically governed process consisting of fuel pyrolysis and

oxidation reactions, formation of the first ring (benzene) and then polycyclic aromatic

hydrocarbons (PAH), inception of the first particles, growth of soot particles due to









reactions with gas phase species, particle coalescence, agglomeration and finally

oxidation. Figure 1-2 illustrates the soot formation process showing stages of formation

on molecular and particulate scales (Bockhorn 1994). However, the process of soot

formation has been more generally classified according to the four stages summarized

below.

1. Formation of soot precursor molecules
2. Particle nucleation, coagulation and growth
3. Particle agglomeration
4. Soot oxidation.
Sooting characteristics of a flame are complex due to the possible multiple

mechanisms of soot formation. Thus, an understanding of the process of soot formation

is fundamental to the study of soot reduction in flames and practical combustion systems.

1.1.1 Formation of Soot Precursor Molecules

Soot precursor species, most likely polycyclic aromatic hydrocarbons (PAH), are

formed in the first stage of soot formation. These species act as nucleation sites for the

formation of soot. It is presumed that this stage is the rate-limiting step in the soot


3al tel


100 nu


-A B

Figure 1-1. Transmission electron microscope (TEM) images of soot aggregates from
isooctane combustion. A) At 100 nm scale. B) At 0.2 |tm scale.










formation, and chemical kinetics play an important role in this stage. Numerous

chemical mechanisms have been proposed to describe the formation of these nucleation

sites. All of these mechanisms generally involve small aliphatic (open chained)

compounds that form the first aromatic rings, typically benzene, C6H6. Acetylene, C2H2,

is the most abundant aliphatic compound to initiate this process in the early stages of


oxidation

agglomeration



coagulation and
growth



particle inception






formation of PAHs









soot precursor
molecules


OH 02 0 OH q 02 O H
02



.* .**o **o 0*
A Ah Ah~


c c
H2


particulate level






transition occurs
at 3 nm


$) molecular level


Figure 1-2. Soot formation. Adapted with permission from a reference (Bockhorn 1994).


c\









combustion, and benzene leads to the production of more complex PAHs in the later

stages (Frenklach 2002). One proposed mechanism is an even-carbon-atom pathway that

involves the addition of acetylene to n-C4H3 and n-C4H5 (Equations 1-1 and 1-2).

n-C4H3 + C2H2 phenyl, (1-1)

n-C4H5 + CH -- benzene+H. (1-2)

It is proposed based on kinetic simulations of shock-tube acetylene pyrolysis that the

reaction in Equation 1-1 plays an important role in forming the first aromatic ring

(Frenklach et al. 1988). Moreover, the reaction in Equation 1-2 suggested by Bittner and

Howard (1981) is an important pathway to aromatic ring formation at low temperatures.

On the other hand, Miller and Melius (1992) suggested an odd-carbon-atom pathway via

combination of stable hydrocarbon radicals like propargyl radicals,

C3H + C3H benzene or phenyl+H. (1-3)

They insisted that n-C4H3 and n-C4H5 are converted into their corresponding resonantly

stabilized isomers very rapidly; thus, their concentrations would not be adequate so that it

could significantly impact the formation of aromatic ring. However, recent Monte Carlo

theoretical studies predicted the higher stability of n-C4H3 radical and n-C4H5, supporting

rather the even-carbon-atom pathway described by the reactions in Equations 1-1 and 1-2

than the odd-carbon-atom pathway.

Another possible pathway for the initial ring formation is a combination of two

reactant types, highly stable propargyl radical and the most abundant acetylene, to form a

cyclopentadienyl radical by

C3H3 + CH -- c-CH,. (1-4)









The cyclopentadienyl radical is then rapidly converted benzene. By means of comparing

reaction rates of Equation 1-4 with that of Equation 1-3, the reaction of Equation 1-4 is

predicted to proceed faster than that of Equation 1-3 by a factor of 2 to 103 (Frenklach

2002). It implies that the reaction 1-4 plays a dominant role in forming the first aromatic

ring. In addition to these pathways reviewed above, many others have been proposed to

characterize the initial stage of soot formation, but have not been widely accepted. Soot

inception is regarded as the most critical stage in soot formation, and is subject to perhaps

the greatest debate.

1.1.2 Particle Coagulation and Growth

The transition from molecular to particle properties occurs in the second stage of

soot formation, namely particle coagulation and growth. This transition occurs at a

molecular weight of about 104 amu corresponding to an incipient soot particle diameter

of about 3 nm. In this stage, soot particles collide with each other forming larger

spherical particles. This is called the process of coagulation, which dominates the early

soot particle growth. The size of particles increases while the particle number density

decreases in the coagulating process. Coagulation is limited to very small particles, on

the order of-18 nm or less.

Aromatics play a role in growth toward soot particle, as gas phase species are

attached to the surface of a particle and become incorporated into the particulate phase.

Frenklach (2002) described this mechanism with a process of H-abstraction-C2H2-

addition (HACA), in which H atoms are abstracted from aromatic compounds, and

gaseous acetylene is incorporated to bring on growth and cyclization of PAHs. The

process of H-abstraction-C2H2-addition is described by









A, +H- A +H2, (1-5)

A + C2H2 -->products, (1-6)

where the notation Ai is an aromatic molecule with i peri-condensed rings, and Ai- is its

radical. The repetitive reaction sequence of two principal steps in Equations 1-5 and 1-6

implies abstraction of a hydrogen atom from the reacting hydrocarbon by a gaseous

hydrogen atom, and the formation of the radical site by adding a gaseous acetylene

molecule respectively. Figure 1-3 represents an example of the aromatics growth via the

process of H-abstraction-C2H2-addition that H abstraction from a biphenyl molecule and

the subsequent addition of acetylene.


O.+ -o+ H
+ H

+ H + H2



+ C2H2 -+ H


Figure 1-3. The H-abstraction-C2H2-addition mechanism acting on a biphenyl molecule.

A biphenyl molecule is formed in the pyrolysis of benzene, a H atom is abstracted

from a biphenyl molecule, and the subsequent addition of acetylene occurs. It is possible

for the growth of aromatic compounds to occur via different mechanisms specific to the

fuel and flame conditions; however, using numerical simulations Frenklach et al. (1988)

showed that these alternate methods quickly relax to the acetylene-addition mechanism.

A process of HACA is sustained until the H atom concentration or the number of

active sites on the soot particle surface reduces in this stage. Eventually, the surface









growth rate of soot particles declines and subsequently particle growth via these

mechanisms ceases. Such phenomenon is termed soot surface aging. It was originally

believed that the depletion of growth species was responsible for this phenomenon.

Recently, it is now believed that a decrease in the surface reactivity of the soot is the

main cause for the reduction of soot surface growth rate although it is not even fully

understood how the soot particles lose surface reactivity (Harris et al. 1983). To support

the theory, it is proposed that the decay of soot surface reactivity is strongly connected to

increase in the ratio of C to H atoms in the soot (Harris et al. 1983, Haynes et al. 1979).

By describing the proposal in a chemical sense, the surface reactions depend on a radical

site formed by the abstraction of a H atom. Meanwhile, in physical sense, if it is assumed

that the hydrogen in the particle is contained only at the edges of the aromatic ring, it can

be seen that the C to H ratio will increase as the particle grows. As a result, the number

of possible growth sites decreases. It is incomplete to fully characterize the decay of soot

surface reactivity with this method. Both these chemical and physical effects would lead

to a direct proportionality between the H to C ratio and surface reactivity with this model;

however, the C to H ratio decays 2 to 3 times more slowly than the surface reactivity

(Dasch 1985). The molecular details underlying the decay of the soot surface reactivity

are under investigation to better understand this mechanism.

1.1.3 Particle Agglomeration

When the viscosity of the particles increases past a critical value due to

dehydrogenation of the condensed phase, coagulation transitions into chain-forming

collisions (Prado et al. 1981). This is the third stage of soot formation, that is, particle

agglomeration. When individual soot particles collide, they stick to each other leading to

fractal aggregates. Contrary to particle coagulation, the particles still preserve their







original identity in agglomeration. Soot aggregates have been analyzed in terms of
fractal geometry. The fractal dimension, discussed in detail later, determined in
numerous flames seems to be confined to a rather narrow range, namely 1.7-1.8.
Individual aggregates of soot particles generally contain 30-1800 primary particles and
are well characterized by a log-normal size distribution (Wamatz et al. 2001). Figure 1-4
elucidates the difference between coagulation and agglomeration.


*









S 0 A B
Figure 1-4. Two processes of particle growth. A) Particle coagulation. B) Particle
agglomeration.
1.1.4 Soot Oxidation
Soot oxidation also called burnout, the final stage in soot formation, takes place at
near the outer radii and the flame tip as oxygen diffuses into the combustion zone. In this
stage, the soot particles are partially or completely broken down, which yields CO or CO2
as a product. Oxidants in soot destruction are O atoms and OH radicals as well as 02.
According to studies by Warnatz et al. (2001), the concentration of O atoms is relatively
low compared with that of other oxidants in sooting flames. Consequently, the









probability of reactions between O atoms and soot is also low. Therefore, it is assumed

that OH radicals and 02 are primarily responsible for the oxidation of soot particles

(Warnatz et al. 2001).

A major source of soot is flames, which may be considered as either premixed or

diffusion (or non-premixed) flames. In a premixed flame, fuels are premixed with

oxidizers at the molecular level before any significant chemical reaction occurs. This

type of flame is typically observed in Bunsen burner as well as the spark-ignition engine.

This type of flame may have insignificant oxidation of soot because most of the oxidizers

are consumed before soot particles are fully-grown. In a diffusion flame, the reactants

are initially separated, and then they are mixed and react only at the interface between the

fuel and oxidizer. A classic example of a diffusion flame is a candle. Soot oxidation in

the diffusion flame is predominantly noticeable at higher flame heights as oxygen

diffuses into the combustion regime and encounters mature soot particles. Therefore, the

stages of soot formation can be divided more distinctly in the diffusion flame (Turns,

2000). Figure 1-5 shows soot formation regimes in a diffusion flame, the axial soot

concentration profile at the center of the flame, and the radial soot concentration profile

at an arbitrary flame height. It can be seen that small quantities of soot are present in the

inception regime while peak formation occurs in the growth regime.

1.2 Soot Suppression with Transition Metallic Additives

A wide variety of metallic additives in fuels has been studied to determine their

effects on soot formation in many practical and laboratory scale combustion systems. In

common, the alkali, alkaline earth and main transition metals have been used as fuel

additives to control soot emission. Common metallic additives are summarized in Table

1-1.





















0 1
Relative axial position, h/ho


-1 0 1

Relative radial position, r/ro

Figure 1-5. Soot formation regimes in a diffusion flame, the axial soot concentration
profile at the center of the flame, and the radial soot concentration profile at
an arbitrary flame height.

Table 1-1. Metallic additives in common.
Alkali Li, Na, K, Rb, Cs
Alkaline earth Mg, Ca, Sr, Ba
Transition Fe, Mn, Cr, Ni


The mechanism of action of metallic fuel additives have been outlined in three

different theories. Firstly, the fuel additive may affect nucleation mechanisms of soot

formation in the early stage of soot particle inception. Secondly, the additive may

enhance soot burnout as a result of rapid elimination of soot precursors attributed to









increase in hydroxyl radicals. Thirdly, the additive may accelerate the soot oxidation rate

by occlusion within the soot phase. Obviously, all three mechanisms may be closely

interrelated. The global and local effects of transition metallic additives were evaluated

in many studies using a variety of techniques from simple visual observations to novel

laser diagnostic measurements. The review conducted in this section will be limited to

the key studies of transition metallic additives in premixed and diffusion flames. In spite

of the same type of combustion conditions, many studies often have yielded different

conclusions.

1.2.1 Manganese-Based Additives

Linteris et al. (2002) reported soot reduction effects of manganese and tin

containing compounds by analyzing the burning velocity of methane/air flames. Greater

than 50% reduction of the burning velocity was shown in the seeded flame. In

comparisons of the reduction efficiency with other suppressants, manganese-based

additives showed about a factor of two less than that of iron pentacarbonyl, but twice as

effective as bromine-based additives.

This result is supported by a study of Wei and Lee (1999), who pyrolyzed

polystyrene with manganese in a laboratory quartz reactor. Although results from several

measurements varied slightly relying on the different conditions, overall 40% of

reduction was obtained in the pyrolysis reaction with manganese. They concluded that

the addition of manganese sulfate into the high temperature pyrolysis of PS inhibited the

formation of PAHs in the reaction.

However, Feitelberg et al. (1993) found an adverse effect, namely that the additive

increased soot volume fraction by approximately 50% in a study of a premixed ethylene

flame seeded with manganese added in 140 ppm concentrations. They expected that









manganese would exist in the gaseous phase as free metal atoms at high temperatures and

form solid MnO through precipitation at residence times.

Hayhurst and Jones (1989) also investigated the effects of metallic additives on

ionization in premixed acetylene/oxygen/argon flames. It was found that manganese

addition did not affect ion concentrations and soot particle size due to their relatively high

ionization potentials that leads to the low rates of soot nucleation and particle growth.

Consequently, they made a conclusion that manganese had no inhibition effect on soot

production rates.

1.2.2 Iron-Based Additives

While manganese is a known neurotoxin, iron has relatively low toxicity; therefore,

many combustion applications and laboratory studies have concentrated on the iron based

additives such as ferrocene [(CsHs)2Fe] and iron pentacarbonyl [Fe(CO)5]. In many

studies, they have been shown to be highly effective soot suppressants (Bukewicz et al.

1974, Feitelberg et al. 1993). Iron pentacarbonyl was selected in this research to study

the effects of the iron on a laminar prevaporized isooctane/oxygen diffusion flame. It is

an organometallic solution that is soluble in liquid isooctane, allowing for a simple means

of regulating and delivering the dopant to the combustion system before vaporization of

the fuel. This factor makes iron pentacarbonyl an ideal additive for this study.

1.2.2.1 Additives in premixed flames

In a study of a laminar premixed ethylene flame seeded with 0.015-0.46%

ferrocene by weight of the fuel, Ritrievi et al. (1987) studied the effects of the addition of

ferrocene, Fe(CsH5)2, on inception and growth of soot particles. As particles moved from

inception to growth regime, an increase in the diameter of soot particles in both seeded

and unseeded flames was observed, and the diameter of initial particles in the seeded









flames was smaller, whereas final particles had a larger diameter than those in the

unseeded flames. Contrary to the size, the number density of particles in both flames was

reduced with height. The same trend on number densities was observed in the previous

work achieved by Haynes et al. (1980). As for the volume fraction of soot, indiscernible

change was shown at early residence time for both flames, and the final volume fraction

was greater in the seeded flame at all the times. In addition, the spatial profile of the

elements iron and carbon in the soot particles was determined with Auger electron

spectroscopy. They found that the soot particles consisted of dense iron at the core and a

thick carbon-rich layer at the outer surface. Mossbauer spectroscopy was used to

determine the chemical state of the iron in the particles, and iron oxide, FeO, was found

to be the stable species on the given flame conditions. It is noted that all analysis was

done with sampled (i.e. extracted) soot particles.

In order to account for the different behaviors of particle inception and growth

shown for the seeded and unseeded flames at an early residence time, they proposed a

hypothesis that FeO homogeneously nucleated early in the seeded flame, prior to soot

particle inception. This was able to illustrate the behaviors and the stratification of the

soot particle at an early residence time in the seeded flame very well. Additionally, they

concluded that the carbon deposited on the particles was used for the direct reduction of

FeO to metallic Fe. The consumption of carbon at the surface resulted in slower growth

rates at the earlier growth region of the seeded flames and indicated that FeO is relatively

less active in the later soot growth region. However, the catalytic effect of iron in the

later residence time had an influence on growth of the particle surfaces.









Feitelberg et al. (1993) also found that the additives increased overall soot

formation in studies of a laminar premixed ethylene flame seeded with ferrocene. Iron

was added to the fuel in 200 ppm concentrations by a molar basis. Overall, the soot

volume fraction in seeded flame increased three times at later residence time, and particle

size also increased with increasing residence time. This agreed well with Ritrievi's

conclusion (1987). However, they did not find any additive effect on the number density

at an early soot inception region, while Ritrievi's group found a significant additive

effect.

After analyzing the states of iron additive in the flame, Feitelberg concluded that

the iron would initially exist in the gas phase as free metal atoms and precipitate out of

the gas into metallic iron form at high temperatures of about 1760 K, or at residence

times of around 4 ms. In addition, it was concluded that thermodynamically FeO was not

formed in the fuel rich flame, whereas it existed in the seeded flame in the Ritrievi's

work. To conclude, the role of the iron additives was not to affect soot particle inception

but to increase the rate of gas-solid reactions leading to increase in the total mass of soot.

As in Ritrievi's work, they also paid attention on the catalytic effects of iron in the flame

with a catalyzed acetylene addition model, and concluded that the iron additive played a

role as a catalyst to carbon deposition via acetylene which increased the final particle

size.

Hahn (1992) assessed the role of iron pentacarbonyl, Fe(CO)5, in a premixed

propane/oxygen flame with a fuel equivalence ratio of 2.4. Iron pentacarbonyl was added

in concentrations of 0.32% by weight of iron to the fuel. The lower regions of the flames

were not evaluated due to the limitation of in situ photocorrelation measurements.









Within all regions of the flame studied, the iron additive had an effect to increase the

overall amount of soot in the flame. That is, the size, number density of soot particle and

a volume fraction in the seeded flame were greater than those in the flame without the

additive.

Even though in situ analysis was not carried out, the state of the iron in sampled

particles was experimentally assessed using X-ray photoelectron spectroscopy rather than

using prediction models. In this analysis, it was found that the form of iron oxide, Fe203,

was a dominant species in the extracted soot particles. Contrary to Ritrievi's conclusion

or Feitelberg's analysis, significant quantities of elemental Fe or other iron oxide, FeO,

were not identified in this study.

They hypothesized that the role of the metal additives on the reduction of soot

emission is to accelerate soot oxidation rate in the burnout zone (Cotton et al. 1971, Hahn

1992). This region is not present in premixed flames; therefore, the full effect of the

metal additive could not be seen. The foregoing studies of premixed flame have

demonstrated that metallic additives made an increase in overall soot emission by either

the catalystic effect of the metal in the later residence time, or acting as soot nucleation

sites in the inception region. A complementary picture of the effect of additives can be

investigated with the addition of the burnout regime in diffusion flames.

1.2.2.2 Additives in diffusion flames

In addition to premixed flame studies, there are a lot of soot suppression studies

using iron based additives in diffusion flames. Bonczyk (1991) studied the effect of an

additive on soot production with a pre-vaporized isooctane/air diffusion flame seeded

with ferrocene added in 0.3% by weight of fuel. In this study, he observed an increase in

the diameter, the number density of particles, and volume fraction in both seeded and









unseeded flames in the regime close to the burner lib. In the burnout regime of the

flames, these parameters however, decayed rapidly in the seeded flame while those kept

slightly increasing in the unseeded flame. This net effect of soot reduction was visually

noted as well when the smoke plume existing in the unseeded flame completely

disappeared in the seeded flame. Soot samples were collected post-flame and subjected

to an Auger-type chemical analysis so that the species of iron present in the soot could be

determined. From the Auger data, it was found that a condensate from the seeded flame

with 0.3% ferrocene was determined to be Fe203 containing only negligible amounts of

carbon. In contrast, the condensate was carbon retaining less than 2% of elemental iron

when the percentage of ferrocene in the fuel was reduced to under 0.001%.

Bonczyk concluded that the metal additive contributed to not only soot

enhancement in soot inception zone but also soot reduction in burnout zone. With

respect to a qualitative illustration on the soot enhancement by additive in the early

residence time, he supported conclusions of Cotton and Ritrievi that soot enhancement

was a result of an increase in nucleation sites provided by solid FexOy particulates and an

increase in the surface activity of particles resulting from a catalytic effect of iron on the

carbon deposited on the surfaces of soot particles. The required Fe is produced by the

reaction in Equation 1-7,

FeOy + yC -- xFe + yCO. (1-7)

The presence of the metal additive enhances the soot reduction in the burnout zone as

well. Iron oxide catalytically reinforces the removal of carbon by molecular oxygen, but

this requires the iron metal to diffuse through the soot matrix to the surface and its

subsequent oxidation by










xFe + yO, -- FeOy. (1-8)
2

In the combination of two reactions above, the net effect of carbon oxidation due to the

metallic additive is expressed as

C+ 10-- CO. (1-9)
2

In short, the additive enhances carbon oxidation, and the result is a net reduction of soot

in the burnout zone.

The similar tendency of soot reduction via addition of ferrocene into ethylene

coflowing diffusion flame was found by Zhang and Megaridis (1996). Ferrocene seeding

accelerated soot inception, but enhanced soot oxidation in the tip of the flame. The soot

volume fraction of the seeded flame was about an order of magnitude less than that of the

unseeded flame. Besides, ferrocene affected the primary particle size at the flame

terminus so that 33% net reduction of soot was observed between the unseeded and

seeded flames. Kasper et al. (1999) also reached the same conclusion in a study with

ferrocene seeded methane/argon and acetylene/argon flames. The soot production rate of

seeded flames was higher at the early residence time due to an increase in the surface of

soot, but lower at the later residence time attributed to efficient soot oxidation by

catalytic means of the additive.

1.3 Studies of the Fractal Properties and the Structure of Soot Aggregates

Numerous studies concerning the physical properties of soot aggregates have been

reviewed by many researchers. A research group led by Faeth has performed numerous

work on fractal and structure properties of soot aggregates using Rayleigh-Debye-Gans

(RDG) scattering theory (Koylu et al. 1994&1995a&b, Farias et al. 1995, Wu et al. 1997,









Krishnan et al 2000&2001). They accomplished it with both gaseous (acetylene,

ethylene, propylene, and butadiene) and liquid fuels (benzene, cyclohexane, toluene, and

n-heptane) for a variety of flame conditions, for example, laminar and turbulent flames,

as well as premixed and diffusion flames. Through their diverse works, it was concluded

that fractal properties of soot aggregate are relatively independent of fuel type, flame

condition, and position. They obtained a fractal dimension of 1.82 and a fractal prefactor

of 8.5, with experimental uncertainties (95% confidence) of 0.08 and 0.5, respectively.

Fractal theory is discussed in detail below. They also carried out numerical simulations

to create soot aggregates based on cluster-cluster aggregation. They computationally

evaluated RDG theory for the optical properties of soot using the Iskander-Chen-Penner

(ICP) approach in small scattering angle regime and compared the results from the ICP

approach with those from RDG theory. The results were in good agreement within

numerical uncertainties. Fractal parameters used for the simulation in their study were Df

of 1.75 and Kf of 8.0 based on their proceeding information. In another study, they

measured soot composition, density, structure, gravimetric volume fraction, and

scattering and absorption properties for wavelengths between 350 and 800 nm in the fuel-

lean region of buoyant turbulent diffusion flames fueled with acetylene, propylene,

ethylene, and propane burning in still air. Then they analyzed these data to find soot

fractal dimensions, refractive indices, refractive index functions, and dimensionless

extinction coefficients using Rayleigh-Debye-Gans scattering for polydisperse mass

fractal aggregates (RDG-PFA theory). They found both soot fractal dimensions of 1.77 in

average and dimensionless extinction coefficients of 5.1 in average with a standard

deviation of 0.04 and 0.5 respectively.









Kim and Choi et al. (1999&2003) measured the fractal properties of silica

aggregates generated in hydrogen/oxygen coflow diffusion flame using light scattering,

thermophoretic sampling and TEM observation. They also invented an in situ laser light

scattering method for line measurement of aggregate size and shape, and applied it for the

measurement of silica aggregates produced in a methane/air premixed flat flame. The

mean radius of gyration and fractal dimension of 1.7 were obtained and examined based

on the RDG scattering theory for fractal aggregates.

Wang and Sorensen (2002) compared scattering cross sections of fractal aggregates

predicted by using RDG scattering theory with those that measured in an experiment and

found a good agreement. For fractal aggregate aerosols of SiO2 and TiO2 formed fractal

aggregates by diffusion-limited cluster aggregation, the fractal dimensions were roughly

1.75 and the number of primary particles per cluster was approximately 150.

Mountain and Mulholland (1988) simulated the growth of smoke agglomerates

using the computer simulation technique of Langevin dynamics. In this study, 48

aggregates comprising between 10 and 687 primary particles per cluster were created to

characterize soot agglomerates and calculate the light scattering from these agglomerates.

The structural information and the results of the calculation were then used to obtain the

fractal properties such as the primary particle size, the radius of gyration and the fractal

dimension. In short, they discovered the fractal dimension of 1.9 and the fractal prefactor

of 5.8.

Dobbines and Megaridis (1991) investigated the absorption, scattering, and

differential scattering cross sections for polydisperse fractal aggregates with the

prescribed fractal dimensions from 1.7 to 1.9 and uniform primary particle size.









Koylu and coworkers (1995a&b) determined the fractal properties for

carbonaceous soot and A1203 (alumina) agglomerates created from various flame

conditions using angular light scattering (ALS) and thermophoretic sampling followed by

analyzing transmission electron micrographs. Both procedures yielded the fractal

dimension of 1.7 with the standard deviation of 0.15 and the fractal prefactor of 2.4 with

the standard deviation of 0.4.

Sorensen et al. (1995) sampled soot aggregates from a premixed methane/oxygen

flame using thermophoretic sampling and analyzed them with transmission electron

microcopy (TEM) method. They obtained the fractal dimension of 1.74.

An analysis with 36-aggregate samples of overfire soot from a laminar acetylene

flame reported by Samson et al. (1987) yielded the fractal dimension of 1.4. However, it

was regarded that the value was much skewed due to the lack of the number of samples.

Sorensen (2001) reviewed scattering and absorption of light by fractal aggregates

and concluded that the aggregates typically have the fractal dimension of approximately

1.75. Fractal dimensions determined from various sources are summarized in Table 1-2.

Even though the values of fractal dimension tabulated in Table 1-2 vary slightly

depending on different measurement techniques and flame conditions, the main fractal

properties of soot are generally considered to be independent of the fuel and the flame

conditions.

1.4 Spectroscopic Method

For identifying the state of the metallic additive without perturbing the

characteristics of the flame, the most effective method is to use an in situ spectroscopic

method. Having an advantage of high sensitivity, Laser-induced fluorescence has been









Table 1-2. Fractal dimension of various aggregates.
Fractal
Investigator di Condition Method
dimension
soot from laminar and turbulent
1.82 TEM
diffusion flames
scattering and
Faeth et al. 1.731.85 soot from various hydrocarbon extinction
Faeth et al. 1.73-1.85 fe flameextinction
fuels flame
measurements
computer
1.75 general soot aggregate compute
simulation
Wang et al. 1.75 aerosols of Si02 and TiO2 Light scattering
silica aggregates generated in
1.77 hydrogen/oxygen coflow Light scattering
Light scattering
Choi and Kim diffusion flame ad
and TEM
1.7 silica aggregates produced in a
methane/air premixed flat flame
Mountain et al. 1.9 smoke computer
simulation
Dobbines and
ines and 1.62, 1.74 soot from laminar ethylene TEM
Megaridis
Cai et al. 1.74 soot aggregates from a premixed TEM
methane/oxygen flame
Sorensen 1.75 general aggregate Review
Samson et al. 1.4, 1.47 soot from laminar ethylene TEM
Angular light
1.75, 1.86 soot from various laminar and Angular light
Koylu et al. scattering
l et a. turbulent diffusion flames scattering
1.54-1.73 TEM


widely used for measuring the concentration and temperature of gaseous phase species in

combustion flows.

Planar laser-induced fluorescence and Rayleigh/Mie imaging measurements were

conducted to investigate the mechanisms of particle formation from gas phase species in

a CH4/02 premixed flame seeded with iron carbonyl (McMillin et al. 1996& Biswas et al.

1997). A XeCL excimer-pumped dye laser operating in 5 mJ pulse energy was used for

FeO PLIF. While the excitation laser was being scanning from 558.5 nm to 561.0 nm,

the fluorescence was monitored near 586 and 618 nm with PMT and boxcar average.

They found that the concentration of vapor phase FeO rapidly rises at the flame cone (i.e.









primary reaction zone) and FeO plays a role as precursors. To validate the experimental

results, they developed a discrete sectional model which accounted for precursor vapor

concentrations and particle growth process. The simulation was in good agreement with

the experiment.

Son et al. (2000) conducted photolysis-probe experiment to generate the ground

state FeO molecules which were detected by LIF method. By means of directing an

unfocused weak UV laser beam into the mixture of Fe(CO)5/(O2 or N20)/(He or Ar), they

created ground-state FeO molecule. The wavelength of the laser was in the range

between 298 and 320 nm, and laser pulse energy of 0.5 1 mJ/pulse was used for

photolysis. Then the fluorescence at 623.6 nm was detected when the FeO molecule was

excited by a wavelength of 591.1 nm.

Telle et al. (2001) combined LIF with LIBS to detect elements in non-accessible

environment. They performed a parametric study with the combination of LIF and LIBS

to investigate analytical selectivity and sensitivity, and concluded that the combined

technique is better than LIBS alone in sensitivity and selectivity.

Nguyen et al. (1996) invented a combination of Raman-Rayleigh scattering and

LIF to measure temperature and the concentration of NO in a methane-air premixed

flame under three different operating conditions. Two frequency-doubled Nd:YAG-

pumped dye laser systems were employed for NO LIF. Then, the quenching was

corrected with information from Raman-Rayleigh scattering experiment. They observed

that NO concentration reduced as the equivalence ratio increased.

As another common technique, in situ Raman spectroscopy has been employed for

species identification and quantification. Maslar et al. (2000) observed various forms of









iron oxide while investigating corrosion on the surface of an electrolytic iron coupon in

air saturated water at a pressure of 25.1 MPa and temperatures from 21 to 537 C using in

situ Raman spectroscopy. The excitation source radiation was a 647.1 nm krypton ion

laser. The in situ Raman spectra were compared with the ex situ spectra using the micro-

Raman system having the excitation source of 785 nm. In this study, they realized that ex

situ spectra were similar to the in situ spectra taken during cooling but different from

those taken during heating.

1.5 Objectives of Present Research

Although the use of fuel additives as soot suppressants has been known for over 40

years and widely studied, the mechanism of action of additives is poorly understood and

still a subject of controversy. The primary objective of this project is to quantitatively

explore a role of the additive for soot suppression in the flame using the elastic light

scattering technique along with thermophoretic sampling followed by transmission

electron microscopy (TEM) and in situ spectroscopy. In addition, Laser-induced

fluorescence spectroscopy (LIF) and in situ Raman spectroscopy are used to identify the

chemical state of the iron additive in the flame. Finally, numerical simulation is

performed to provide additional information on iron species in the flame. As another

prevalent scheme, Laser-induced incandescence (LII) is a well-researched technique for

analyzing and characterizing sooting flames and combustion processes. LII occurs when

a very intense laser beam encounters particulate matter like soot. A soot particle can

absorb energy from the beam, which leads an increase in the particle's temperatures of

4000-4500 K. If the energy absorption rate is sufficiently high, the temperature will rise

to levels where significant incandescence (essentially blackbody radiation) and

vaporization can occur. LII technique was employed as a different approach for









analyzing soot particle in other part of the same research project; however, a detailed

treatment of the characterization of soot particles in the context of LII technique is

beyond the scope of the present study, thereby, will not be examined in this work.

A laminar prevaporized isooctane/oxygen diffusion flame was invented employing

a laboratory scale diffusion burner, and iron pentacarbonyl, Fe(CO)5 was used as an

additive for all researches. Information retrieved from this research will then be used in

future application of soot reduction using practical combustion system like a real turbine

engine and contribute ultimately to developing a solution to minimize health and

environmental problems resulting from soot emission. Below are specific objectives.

1. To determine the differential scattering coefficient using in situ techniques of the
light scattering and transmission measurements.

2. To evaluate scattering parameters such as the size, number density, and volume
fraction of soot particles using Rayleigh-Debye-Gans scattering theory in
combination with thermophoretic sampling of soot followed by TEM.

3. To provide some insights into the role of additives by analyzing scattering
parameters in the unseeded and seeded flames.

4. To implement in situ spectroscopic methods such as LIF and Raman scattering
technique to probe the chemical state of iron additives species throughout the
flame.














CHAPTER 2
FUNDAMENTAL SCIENCE AND BACKGROUND THEORY

This chapter introduces fundamental theories and background knowledge of the

elastic light scattering technique, spontaneous Raman, Laser Induced Fluorescence (LIF)

spectroscopy, and electron microscopic schemes. The former techniques are the

remarkable diagnostic methods that are nonintrusive and allow analysis of soot formation

process in the diffusion flame without intervening in the chemical and physical processes.

In addition, the application of each spectroscopic scheme to data analysis is discussed

along with any limitations of those theories.

2.1 Elastic Light Scattering Theory

Electromagnetic radiation can interact with a particle in several ways. That is,

radiation can be reflected, scattered, absorbed or emitted. These interactions are

dependent on the nature of the heterogeneity: the shape of the particle, the material of the

particle (i.e., refractive index), its relative size and the clearance between particles.

Therefore, a certain particular system can be characterized using the way that

electromagnetic radiation interacts with the particles in the system. Information such as

size and number density of the particles can be inferred from the scattering response. In

this study, elastic laser light scattering was employed to determine the differential

scattering coefficients of soot particles in the unseeded and iron seeded flames. The

determined parameter will be used to calculate the number density and total volume

fraction, in combination with the size of the soot particles obtained from TEM analysis.








Elastic light scattering takes place in the case that an electromagnetic (EM) wave,

the incident light, encounters a scattering particle. At the moment the EM waves collide

with discrete particle, electrons oscillate within the particle at the same frequency as the

incident wave. The oscillation, called an induced dipole moment, is regarded as a source

of light scattering. The energy of the incident light is either discharged by light radiation

or extinguished by absorption within the particle. When the frequency of the incident

light is equal to that of scattered light considered, the process is called elastic scattering.

In contrast, Raman scattering is considered an inelastic scattering process. More detailed

explanation on Raman scattering will be given later. Figure 2-1 shows the light

scattering response to an incident electromagnetic light.


Elastically scattered light X

Y




.V Vi/

_> .:iw V


lent light


' Induced


dipole moment
dipole moment


hv|incident=hv|


Figure 2-1. Light scattering response to an incident light.
There are two kinds of categories in the elastic light scattering. One is Rayleigh

scattering theory that is applied to a system with small, dielectric (non-absorbing) and

spherical particles. The other is Mie scattering theory that is used for general spherical


Incic









solution without a particular size limit; hence, it can be used for describing most spherical

scattering particles, including Rayleigh scattering particles. However, Rayleigh

scattering theory is usually used as long as it is applicable due to complexity of Mie

scattering solution.

2.1.1 Rayleigh Scattering Theory

A valid scattering solution using Rayleigh theory for a spherical particle may be

obtained under the following conditions:

1. The external electric field seen by the particle is uniform

2. The electric field penetrates faster than one period of incident electromagnetic
radiation.

These two conditions are satisfied for the case of a <<1 and I m I a <<1 respectively,

where a is the dimensionless size parameter given by

a=r (2-1)


where a indicates the particle radius, and X is the relative wavelength defined as


A = (2-2)
mo

where o means the incident wavelength in vacuum and mo represents the refractive

index of the surrounding medium. The refractive index, the property of the material is

defined as

m = n + iK. (2-3)

In this Equation, n indicates the common refraction of light while the complex term

relates to absorption. The value of k is not exactly zero for any material, but materials

with the value approaching to zero are termed dielectrics. The relative refractive index is

defined as










m=-. (2-4)
mo

The magnitude of the relative refractive index, Im is


(n2 C2)2 (2-5)
mo

In the Rayleigh regime, the vertical-vertical differential scattering cross section

(cm2/sr) indicated in Equation 2-6 means that both the incident light and the scattered

light after the interaction are vertically polarized with respect to the same scattering plane

(xy-plane), see Figure 2-1.

2 m -2 2
=-- a--2 (2-6)
472 m m+ 2


Simply, the horizontal-horizontal differential scattering cross section shown in

Equation 2-7 means that both the incident light and the scattered light are polarized

parallel to the scattering plane

HH = Cv cos2 0. (2-7)

In Equations 2-6 and 2-7, the first subscript means the incident light, and the second

subscript means the scattered light. Also, subscripted V and H, respectively, refer to the

vertical and horizontal polarization with respect to the scattering plane. Note that the

vertical-vertical differential scattering cross section is independent of the observation

angle 0, while the horizontal-horizontal differential scattering cross section has a

minimum at 90 degrees. The total scattering cross section (cm2) and absorption cross

section (cm2) are defined as









-2 2
2A2 6m 1
2a =- a (2-8)
S3r m +2

-2
aabs '- Im m 1 (2-9)
Z m +2


Finally, the total extinction cross section (cm2) is defined as a sum of the scattering and

absorption cross section, namely,

ext = Usca + abs (2-10)

As represented in Equations 2-8 and 2-9, the total scattering cross section scales

with a6 whereas the absorption cross section is proportional to a3. Compared to ab in

the Rayleigh regime, usca is small enough to ignore the contribution of ao-, to ae ;

hence, it can be assumed that axt = abs, for an absorbing particle.

2.1.2 Systems of Particles

The light scattering theory is specifically applied in radiative analyses under the

significant assumptions regarding the scattering particle. That is, the particle is assumed

to be a single and spherical shape in the system. However, to extend the assumption on

the scattering of single particle to a system of particles premises three criteria as

elucidated below (Jones, 1979):

1. The particles are spaced far enough to have no electrical interactions between
particles.

2. The light does not undergo multiple scattering.

3. There is no optical interference between the scattered waves.

The first criterion is fulfilled if individual particles are placed in a distance of 2 to 3

diameters from one another. The second criterion is met if the optical mean free









pathlength is greater than the optical length of the system. The last criterion is satisfied if

a large number of particles are randomly placed in the system. In this case, the total

intensity of scattered light can be determined by directly adding the intensities of

scattered light from each particle.

There are two types of system of particles: monodisperse and polydisperse. A

system containing uniformly sized particles is termed monodisperse. For such a system,

the overall scattering and extinction of light from particles can be described by the

differential scattering coefficient (cm-1sr-) and extinction coefficient (cm-1), which are

expressed respectively as

K, = No (2-11)

Kex = No,, (2-12)

where N is the number density of particles (particles per volume).

The transmission is defined as the intensity of the transmitted light through a

system of particles per that of the incident light for a particular wavelength, namely,

S= transmitted (2-13)
Io

A c of unity results from 100% transmission of the incident light through a system of

particles, whereas a c of zero results from the entire incident light absorbed and/or

scattered by the particles. The relationship between the transmission and the extinction

coefficient can be correlated by the Beer-Lambert law, namely,

r = exp(-KeL), (2-14)

where L is the optical length. The product KexL is known as the turbidity, a measure of

the ability of the system to extinguish incident light.









The volume fraction f, is also an important parameter for characterizing the

particles in a system. The volume fraction is defined as the volume of particles per unit

volume, hence, it is dimensionless. For a monodisperse system, the volume fraction is

given by


f, =-d3N, (2-15)
6

where d is the diameter of the scattering particle.

A polydisperse system is characterized by non-uniformly sized particles. A flame

is a good example of the system, as the sizes of soot particles can vary throughout the

sample region. The primary particle size must be determined to evaluate the scattering

and extinction properties of a polydisperse system. This will be discussed in the next

section.

2.1.3 Rayleigh-Debye-Gans (RDG) Scattering Theory

As discussed in Chapter 1, the morphological characteristics of soot are to evolve

into a fractal structured aggregate. Using either Rayleigh scattering theory or Mie

scattering theory itself is not a reliable application here for the large sized and open

structured aggregates. Therefore, a new theory is necessary to exam the size and the

fractal dimension of soot aggregates. Rayleigh-Debye-Gans (RDG) scattering theory has

been found to be a suitable approximation for studying physical properties of soot

aggregates.

2.1.3.1 Rayleigh-Debye-Gans (RDG) scattering approximation

RDG theory has been used to interpret light scattering data to determine cluster

parameters. Aggregates produced in the later stage of soot formation vary considerably

in size and shape while they grow. Physical features of soot aggregate can be defined









using Rayleigh-Debye-Gans (RDG) scattering theory. Regarding physical properties of

soot aggregates, this theory has the major assumptions described below:

* The size parameter of primary particles is sufficiently small such that individual
primary soot particles satisfy the Rayleigh scattering theory.

* Soot aggregates are composed of mono-disperse nonoverlapping spherical primary
soot particles.

* Primary soot particles just touch one another.

* The number of primary particles per aggregate satisfies a lognormal probability
distribution function.

* Soot aggregates are mass fractal-like objects with mass fractal dimension of less
than 2.

The fractal-like objects can be defined using the following power law relationship,

namely,


Npa = kf (Rgdp)D (2-16)

where N, is the number of primary particles per aggregate, kf is the fractal prefactor,

Rg is the radius of gyration of an aggregate, dpa is the primary particle diameter, and

Df is the mass fractal dimension implying the openness of the soot aggregate. First of

all, evaluation of the fractal properties requires determining optical properties. Not

only N but also dpa can be directly determined using transmission electron microscopy

(TEM) analysis of post thermophoretic sampling of soot aggregates. The radius of

gyration then can be calculated using either of two correlations (Koylu et al., 1995a).

The first approach is to use only the maximum projected length of the aggregate, L.

Alternatively, the radius of gyration can be evaluated using the geometric mean of









maximum length, and the maximum projected width normal toL, W. Two correlations

are mentioned below respectively,

L/(2Rg)= 1.49, (2-17)

(LW)/2/(2Rg) = 1.17. (2-18)

With these parameters known, the fractal properties are obtained using a linear regression

method with a least squares approach. In other words, when N is plotted as function of

Rg/dpar in logarithmic scale for a set of aggregates, the fractal dimension describes the

"slope", the fractal prefactor determines the "magnitude" of the least-squares straight line

fit to the data.

Determination of the radius of gyration requires more attention because it directly


affects scattering properties. The value of R based on Equation 2-17 may be somewhat

greater than that evaluated in Equation 2-18 unless the aggregate is equilateral shape. As

a result, the smaller values of Df are obtained when Equation 2-17 is used rather than

Equation 2-18. To avoid such a conflict, some researchers adopted 1.78 or 1.67 instead

oftakingl.49 in Equation 2-17. Even though specific aggregate properties differ for the

various flame systems, all the flames yield the same relationship between the number of

primary particles in an aggregate and its radius of gyration. Therefore, the fractal

properties of aggregates are independent of various positions and flame conditions.

2.1.3.2 Evaluation of the extinction coefficient

The next phase of the present evaluation of RDG theory is to consider absorption

and total scattering cross sections. Up to this point, the scattering and extinction

coefficients have been assumed spatially constant for a given system. However, in the

case of a flame, these parameters may be highly spatially dependent on the system. For









instance, the concentration of soot varies very much based on the location in the system.

In particular, determination of the extinction coefficient is required to pay more attention

because the transmission data cannot be collected only at representative points to

determine it. To better characterize this type of system, the deconvolution techniques

have been used if a given system is sufficiently large so that the system can be divided

into several concentric regions. These techniques construct information on the radial

parameters based upon information in neighboring sections. However, the technique may

not be feasible in case a system area is too narrow to be separated into a few sections. In

such cases, the extinction coefficient for such cases can be determined using Rayleigh-

Debye-Gans scattering theory.

The differential scattering cross section (cm2/sr) of an aggregate is defined as

a = N r,,S(q), (2-19)

while the differential scattering cross section of the primary soot particle in an aggregate

-par is determined using Equation 2-6 presented above, namely Rayleigh theory.

Subscripts of"agg" and "par" mean an aggregate and the primary particle, respectively.

A new parameter, S(q) shown in Equation 2-19 is called either the structure factor or the

angular scattering form factor given differently based on the scattering angle variety.

S(q) = 1, qRg(( 1 (2-20a)

S(q) = 1 q R/3, qRg <1 (2-20b)

S(q)= C(qRg)-Df, qRg)1. (2-20c)

Under RDG scattering approximation, the structure factor is nearly unity in the small-

angle scattering regime, the so-called Guinier regime. Consequently, scattering mainly









depends upon the number of primary particles in the aggregates. Conversely, the

structure factor depends strongly on the values of Rg and Df in the large-angle scattering

regime, the so-called power-law scattering regime. Herein q is the modulus of scattering

vector (cm-) defined as

477
q -sin( / 2). (2-21)
A

The constant C refers to the cutoff of the density at the perimeter of the aggregate,

and is approximately unity defined empirically. The total scattering cross section (cm2)

for a fractal aggregate is also expressed as

aca = N 2 Ga (kR ), (2-22)
agg par par ,

where


G(kRg) = (1+- k2R) -D/2, (2-23)
3Df

and the total scattering cross section of a single primary particle is simply given by

Equation 2-8. Meanwhile, the absorption cross section of an aggregate is defined as

ab = N. abp (2-24)

where it is assumed that absorption is not affected by aggregation, while the absorption

cross section for a primary soot particle is determined by Equation 2-9. Finally, the total

extinction cross section for an aggregate is the sum of the total absorption of an aggregate

and total scattering cross section of an aggregate by Equation 2-10.

Note that the scattering volume is absolutely greater than that of a typical soot

aggregate; thus, a number of soot aggregates are involved in interactions with incident

light. The number density of aggregate, Nagg in the scattering volume can be calculated









using the differential scattering coefficient, K' agg obtained from the light scattering

experiment. That is, the number density of aggregate in the scattering volume is readily

determined as

gg = K=, agg /OR-agg (2-25)

Therefore, the total number density (particles/cm3) in the overall scattering volume is

defined as

NTtal = agg Npar, (2-26)

and the extinction coefficient for the overall aggregates in the volume is finally

determined using the following Equation,

Kex = N e .t (2-27)
agg agg agg

2.1.4 Sampling and Analyzing Soot Aggregate

Prior to describing all parameters characterizing morphology of soot aggregate

using RDG theory, thermophoretic sampling and analysis by TEM must be implemented.

The principle of thermophoretic sampling and instrumentation of TEM will be reviewed

in brief.

2.1.4.1 Thermophoretic sampling

Thermophoretic sampling is employed by means of collecting particles with a thin

film. As particles move across the temperature gradient existing between the hot flame

and the cold mesh, a process known as thermophoretic force causes them to move from

the higher temperature of the flame to the lower temperature of the mesh. This

temperature gradient is readily established by introducing a mesh initially being at room

temperature into the hot flame, and it drives the particles to the surface of the mesh where

they are deposited. Recently, thermophoretic deposition has provided a quantitative









understanding of the drift of particles to the surface that is at a lower temperature than the

surrounding gases.

2.1.4.2 Transmission electron microscopy

Information about soot aggregates obtained with thermophoretic sampling can be

extracted using a high-resolution transmission electron microscope (TEM) equipped with

photographic and data recording capabilities. The TEM is an evacuated metal cylinder

(the column) about 2 meters high consisting of an electron gun, the illumination system,

specimen, and the imaging system. A Ray diagram for TEM is schematized in Figure 2-

2. As a virtual source, the electron gun at the top of the microscope emits a stream of

monochromatic electrons that travel through vacuum in the column of the microscope.

The electron gun has a V-shaped tungsten heating filament that is the cathode emitting

electrons. When the cathode is heated, the accelerating voltage of between 40,000 to

100,000 volts is passed between the cathode and the anode placed just below the cathode.

By using the high voltage, these negatively charged electrons in the cathode are

accelerated to an anode positively charged. The acceleration of electrons depends on the

amount of high voltage. Some of electrons passing through a tiny hole in the anode form

an electron beam which travels down the column. Electrons are high energy particles so

that they could easily have an influence on the interaction with any matter. The

interaction causes the emission of all the lower forms of energy such as x-rays, secondary

electrons, ultraviolet, and heat energy. As a result, the microscope must be kept in a high

vacuum of the order of 133x10.8 Pa.

This electron stream is focused to a small, thin, coherent beam by the use of











Virtual Source


1 st condenser lens

2nd condenser lens
Condenser aperture

Specimen
Objective lens
Objective aperture

Selector aperture
Intermediate lens


Intermediate aperture
Projector lens
I-



Final image screen
Figure 2-2. Schematic of TEM

condenser lenses. The first lens largely determines the spot size, and the second lens

changes the size of the spot on the sample. The beam is restricted by the condenser

aperture, blocking out electrons far from the optic axis.

Specimens in the TEM are examined by passing the electron beam through them.

Therefore the mass thickness of the specimens must be thin enough (50-100 nm) to allow

electrons to pass through them. When the electrons strike the specimen, they are either

transmitted or scattered depending on the density of the atoms in the specimen. While

some electrons are scattered, others are transmitted and hit a phosphorescent screen

placed in the bottom of the microscope. It results in a contrast of the specimens that

relies on both diffraction of electrons and the number of the atoms in the specimen. The









higher the atomic number of the specimen, the more electrons are scattered and the

greater the contrast.

The electrons transmitted through the specimen are focused by the objective lens

onto a phosphorescent screen to form an image. The quality of the objective lens plays a

major role in determining the resolving power of the apparatus. Objective and selector

apertures being right below the lens in a row are used to restrict the electron beam. The

objective aperture blocks the unfocussed electrons, resulting in an enhancement of the

image contrast. The periodic diffraction of electrons is examined using selector aperture.

The transmitted electrons are passed down the column through the intermediate and

projector lenses. The intermediate lens can magnify the first intermediate image, and the

projector lens can form a real image on the fluorescent screen at the bottom of the

microscope column. When the image strikes the phosphor image screen, light is

generated, which enables the image to be observed. The image can be analyzed directly

by the operator or photographed with a camera.

2.1.4.3 Energy dispersive x-ray spectroscopy (EDS)

The elemental composition of Iron seeded soot particle can be identified using

energy dispersive X-ray spectroscopy (EDX or EDS) that is a method used to determine

the energy spectrum of X-ray radiation. The technique employs X-rays emitted from the

atoms comprising the sample's surface when the atoms are struck by electrons. In other

words, when an electron from a higher shell fills in an electron vacancy, an X-ray is

emitted to balance the energy difference between the two electrons. Qualitative and

quantitative determinations of the elements present in the sampled volume can be

evaluated using the number of emitted X-rays versus their energy. It can be measured by

an EDS X-ray detector that is a solid state device discriminating X-ray energies. The









energy of the X-ray is characteristic of the element from which the X-ray was emitted.

The detector is a semiconductor, usually silicon doped with lithium, and is polarized with

a high voltage. When an X-ray photon hits the detector, it creates electron-hole pairs that

drift due to the high voltage. The electric charge is collected and a condensator is

charged. Increase in the voltage of the condensator is proportional to the energy of the

photon so that the energy spectrum can be determined. The condensator voltage is reset

regularly to avoid saturation. The detector is cooled to reduce the electronic noise using

liquid nitrogen.

2.2 Spontaneous Raman Scattering Theory

As discussed previously, elastically scattered radiation shows the same frequency

resulting in the same photon energy as the incident radiation. In contrast, inelastically

scattered light governed by Raman scattering theory has certain shifts in frequency from

the incident light. The incident radiation coupled into the induced dipole moment can

create a quantum shift in the vibrational modes of the molecule. If the shift occurs to a

lower photon frequency resulting in the lower photon energy, it is termed a Stokes shift.

Conversely, a shift to a higher photon frequency resulting in the higher photon energy is

referred to as an anti-Stokes shift. This shift can occur when a molecule excited via

scattering interaction relaxes to a lower vibrational energy state than initial state prior to

excitation. In this case, energy from a molecule is added to the photon. These inelastic

shifts of the incident wavelength are determined from Equations 2-28 and 2-29

respectively,


VStokes = (2-28)
A, Ao sct






43


SAnStokes +- (2-29)
^o clscat

where ko is the incident wavelength, ,scat is the wavelength of the scattered light, VStokes

and VAntiStokes are Stokes and anti-Stokes shifts, respectively. The Raman shift is usually

quantified by wave number expressed in dimensions of cm-1. The shift in frequency of

the scattered photons is species dependent, which enables Raman spectroscopy to be a

powerful tool for species identification. Figure 2-3 presents elastic and inelastic

scattering effect qualitatively in terms of molecular energy levels.

El, E


---------- virtual

hvIm /"V \/V hVincident hvm v
E, E

A B
E E



L ---------- virtual ----- --------- virtual

hVmo S V hVstocks hV1nc /' n // hvnti-Stocks
/s cE0

C D


Figure 2-3. Energy level diagrams representing elastic scattering transitions and several
inelastic Raman scattering transitions. A) Elastic scattering. B) Resonance
Raman scattering. C) Stokes Raman scattering. D) Anti-Stokes Raman
scattering.









As shown in Figure 2-3, the molecule absorbs the incident photon and is excited to

a "virtual" electronic state. The case of elastic scattering transitions is that the molecule

relaxes to the original vibrational level of the ground electronic state. In the case of

inelastic scattering transitions, however, the molecule does not come back to its original

vibrational level, so that the shifts are created. As presented in the case B in Figure 2-3,

the molecule can reach stable electronic state past the virtual state when the incident

photon energy exceeds an electronic transition energy. This process is called Resonance

Raman scattering being closely related to fluorescence emission which will also be

discussed later. Due to the similarity, fluorescence emission is a major source of noise in

resonance Raman scattering technique. Compared to the fluorescence emission process,

the resonance Raman process is nearly instantaneous so that the Raman signal can be

discriminated from the fluorescence. Considering such a condition, the resonance Raman

effect is enhanced 102 to 104 times compared with spontaneous Raman effect.

Figure 2-4 elucidates the relationship of the frequency and intensity between Raman and

Rayleigh spectrum. Several things are noteworthy in Figure 2-4. First, two Raman lines

are symmetric with respect to Rayleigh line because the energy gain and lose are the

same amount for the Stokes line and anti-Stokes line. Second, the Stokes line is

apparently more intense than the anti-Stokes line. The intensity of the Stokes line is

typically 100 to 1000 times higher than that of the anti-Stoke line. This is attributed to

the fact that molecules are highly populated in the ground vibrational state at room

temperature; hence, the chance that the incident photons encounter molecules is much

more probable in the ground vibrational state than in the excited vibrational states, which

is governed by the Boltzman distribution. For this reason, Stokes line is often adopted











Intensity
Rayleigh line


<- Vlb -> -- Vb >

Stocks line
Anti-Stocks line


Incident -Vvibration Vincident Vincident +Vvibration

Frequency

Figure 2-4. Relationship between Rayleigh and Raman scattered lines in a scattering
spectrum. Source: Ingle and Crouch 1998.

for signal detection in Raman spectroscopy. In case of elevated temperature, the intensity

of the anti-Stokes Raman signal is enhanced. Considering the Rayleigh scattering

intensity, it is usually thousands times more intense than Raman scattering intensity.

The intensity of Raman scattered radiation is expressed as

e lkT
I, = ENg, (2-30)


where a is the differential Raman scattering cross section, E0 is the source irradiance,

N is the number of gas molecules, g is the vibrational degeneracy, Q, is the


vibrational partition function, E, is energy level of a molecule in the initial vibrational

i
state v, k is the Boltzmann constant, and Tis the temperature. Raman intensity is

directly proportional to several parameters shown in Equation 2-30. The most significant

parameter is the differential Raman scattering cross section that depends on the forth

power of 0 + v,,b; hence, higher photon energy increases the Raman scattering cross

section. Therefore, Raman scattering intensity can be significantly increased by









decreasing wavelength of the excitation light source. Higher photon energy is preferred

for excitation, but gives rise to another concern about spatial resolution of detection for

some species like Fe203. Some Fe203 Raman lines are very close to the Rayleigh line

resulting in a difficulty of peak discrimination. Thus, highly efficient long pass filters are

essential component to prevent elastically scattered stray light from being detected in

such an experiment. In addition, increase in the concentration of the active molecules in

the excited volume and the intensity of the excitation source is helpful to obtain

significant gains in Raman scattering intensity.

As stated before, Raman scattering takes place when molecules are excited and de-

excited between rotational states as well as vibrational states. Since the polarizability of

single atoms does not change with vibration or rotation, Raman scattering technique

cannot be used for atomic identification. If the polarizability of a molecule does change

during vibrational or rotational modes, the molecule is considered Raman-active. Each

type of Raman-active molecule results in a particular shift in Raman spectrum allowing

Raman spectroscopy to be useful for species identification. Raman shifts and the

emission wavelengths corresponding with the excitation wavelengths of 355 nm and 532

nm for common species are summarized in Table 2-1.

Qualitative and quantitative information can be obtained using the wave numbers

of the Raman shifts observed in molecules and the radiant power of Raman scattering. In

addition, structural information of molecules can be provided by the depolarization ratio,

p, defined as


p= --, (2-31)
IVI









Table 2-1. Raman shifts and the emission wavelengths of common species.
Species Shift (cm-1) Emission 1* (nm) Emission 2* (nm)
H2 4156 416.0 683.0
02 1555 357.4 580.0
N2 2331 386.7 607.3
CO 2143 383.9 600.5
CO2 1285 371.6 571.0
CO2 1388 373.1 574.4
CH4 2917 395.6 629.7
CH3OH 2955 396.2 631.2
H20 3650 407.5 660.2
Emission 1 and 2 are the emission wavelengths corresponding with the excitation
wavelengths of 355 nm and 532 nm respectively.

where Im and I, denote the Raman radiant powers that are vertically and horizontally

polarized scattered radiation with respect to the incident radiation polarization. Non-

spherical shaped molecule can cause the scattered light to be depolarized when struck by

polarized light. If the vibrational mode is symmetric, the depolarization ratio would be

almost zero. However, if the vibrational mode is non-symmetric, depolarization of the

scattered radiation can occur. In such a case, the depolarization ratio is predicted as 0.75.

The Raman scattering cross section has angular dependency on horizontally

polarized incident radiation whereas it is independent of vertically polarized incident

radiation over all scattering angles; hence, a vertically polarized light is often used to

avoid any angular dependence on the scattered radiation.

2.3 Laser Induced Fluorescence Theory

Laser-induced fluorescence (LIF) is spontaneous emission from atoms or molecules

that have been excited to higher levels by absorption of laser radiation. The fluorescence

process for an atom or molecule is depicted in Figure 2-5.

When atoms or molecules are resonantly stimulated by the laser source, they absorb

photon energy and are subsequently excited to higher electronic energy states.















ShVF1
hv,., /\AAN\ hvF2




Figure 2-5. Energy level diagram of the fluorescence process for atoms or molecules.

Spontaneously excited atoms or molecules decay in the ground state emitting photon

energy which is lower than the incident photon energy. In general, the intensity of the

fluorescence is proportional to the species concentration, the gas temperature, and

pressure. Compared to spontaneous Raman spectroscopy, a great sensitivity is

achievable for LIF due to the higher signal to noise ratio. In addition, selectivity of a

particular excitation source for a given species is capable of avoiding interference with

other species, which is another advantage of LIF spectroscopy. However, the difficulty

rises when the elastic light scattering interferes with the fluorescence signal as commonly

presented in Raman scattering technique. Using high pass filters can eliminate the elastic

light source from the fluorescence signal. The Stokes shift is important to eliminate such

effects.

The fluorescence signal can be used in quantitative measurements of species

concentration, temperature, velocity and pressure as well as qualitative analysis such as

species identification. The concentration measurement is the most common one among

these applications. The fluorescence signal (photons/s) is defined as

SF =exp K, No DF,, (2-32)









where exp is an experimental constant, KA is a rate constant for stimulated

absorption, No is number density of species, and F is fluorescence quantum efficiency.

7exp includes the incident laser intensity, a focal volume, and a solid angle which can be

determined through calibration with known source. KA is known from Einstein

coefficient for stimulated absorption and Boltzmann distribution. In order to determine

No, FF remains the only unknown given by


SKF (2-33)
KF +Kic +Kisc +Kec
where KF is fluorescence rate constant, Ki is the internal conversion rate constant, Ksc

is intersystem crossover rate constant, and Kec is the external conversion rate constant

defined as

Ke = K [Q]. (2-34)

By introducing new parameter in Equation 2-33,

SF (2-35)
KF + Kic + KIsc

the number of unknowns can be reduced to (o and K which are determined from

Stern-Volmer plot. Finally, No can be evaluated with all parameters determined.

However, quantitative measurements are difficult as long as quenching is present. The

variation in collisional quenching is the most common cause for uncertainty of

fluorescence measurement. Accounting for the collisional quenching is the very hard

problem while the fluorescence signal is related to the absorbing species concentration.

Among several approaches to account for quenching in fluorescence measurement, the

saturated fluorescence technique is prevalent. Independent condition of quenching can be






50


achieved by increasing the incident laser energy until absorption and fluorescence

dominate quenching. This technique will not be examined here in detail.














CHAPTER 3
EXPERIMENTAL APPARATUS AND METHODS

Light scattering techniques and transmission electron microscopy (TEM) analysis

following thermophoretic sampling were employed to characterize the concentration and

morphology of soot in this study. In addition, two spectroscopic techniques were utilized

for in situ species detection and identification, namely, laser induced fluorescence

spectroscopy (LIF) and spontaneous Raman spectroscopy. The aim of this chapter is to

describe and explain each experimental apparatus utilized, including the associated

procedures and analysis methodologies employed. The experimental results are analyzed

in Chapter 4.

3.1 Burner System

A concentric diffusion burner composed of stainless steel tubing was employed for

all experimentation. Figures 3-1 and 3-2 represent respectively a schematic and a

picture of the burner shown from side and top views. The burner dimensions are also

shown in Figure 3-1. Gaseous Isooctane and nitrogen are supplied through the inner tube

(0.15 cm ID) of the burner. A solid annular disk was press fit between the inner and

outer tubes to maintain concentricity. The disk was perforated with 9, 0.03 cm-diameter

holes, as shown. The oxygen flow was fed through the annular disk and out the nine

ports at the burner outlet. Using a stainless steel mesh flame holder does promote

improvement of the flame stability, as shown in a previous study (Masiello, 2004).

However, in the present study the oxidation regime must remain undisturbed so that the










Laser r
0.95 cm Laser-

0.31 cm < ,
0.2 cm 13.9 cm

4.3 cm 0.03 cm ID= 0.15cm











<-- 0.513 cm

Isooctane + N2 -ID= 0.696 cm
A B

Figure 3-1. Concentric diffusion burner schematic. A) Side view. B) Top view. Oxygen
goes into the system through the annulus array of ports whereas isooctane
and nitrogen flow through the tube in the center.

soot characteristics can be extracted purely from the flame; hence, no stabilizer was used.

As recommended in the previous study, the exit area of burner was reduced to increase

the flow velocity at the burner exit. As a result, a diffusion jet turbulent flame was

created, and fluctuations of the flame were decreased to an acceptable level. In addition,

a shroud was placed around the burner to block ambient air currents as well as prevent

stray light from infiltrating into the scattering detection optics. The shroud was made of

semi-transparent Plexiglas, with dimensions of25.4x26.7cm.

In order to investigate the soot characteristics at various positions along the

vertical axis of the flame, the burner was controlled with a vertical translation stage

which allows the burner to move up and down. Twenty-five heights were selected and
































Figure 3-2. Concentric diffusion burner. A) Side view. B) Top view.

designated with the number 1 through 25, noting that the distance between heights was

not necessarily constant. These heights were consistently reproducible, and corresponded

to a number of specific rotations of the vertical stage knob. That is, one revolution

corresponded to non-linear vertical motion. A first height designated "1" was located in

9.4 cm above the burner tip as shown in Figure 3-1A. All measurements were made

along the centerline at 25 different heights from bottom to top of the flame. The

summary of these heights are tabulated in Table 3-1, and Figure 3-3 represents the

relative positions of these heights in the flame.

3.2 Fuel Vaporization and Delivery System

A fuel vaporization and delivery system was used for all experiments in this study.

It was necessary to vaporize the liquid isooctane fuel before the fuel could be delivered to

the burner in order to produce a stable, diffusion flame, in the absence of liquid droplet,









Table 3-1. Data collection heights.
Position label Height above burner tip (cm) Distance between two heights (cm)
1 9.40
2 9.75 0.35
3 10.10 0.35
4 10.50 0.40
5 10.90 0.40
6 11.30 0.40
7 11.70 0.40
8 12.20 0.50
9 12.70 0.50
10 13.20 0.50
11 13.70 0.50
12 14.30 0.60
13 14.90 0.60
14 15.50 0.60
15 16.20 0.70
16 16.95 0.75
17 17.75 0.80
18 18.60 0.85
19 19.60 1.00
20 20.65 1.05
21 21.85 1.20
22 22.80 0.95
23 23.50 0.70
24 24.20 0.70
25 25.25 1.05


for the fuel to be burned more efficiently. By heating liquid isooctane to temperatures

around 1000C, vaporization was achieved when the liquid isooctane passed through a

vaporization system that consists of three main sections: the preheat zone, the

vaporization zone, and the delivery line. The boiling point of isooctane is about 99C at

atmospheric pressure. A schematic diagram and a picture of the fuel vaporization and

delivery system are shown in Figures 3-4 and 3.5 respectively.










































Figure 3-3. Data measurement heights.

All three sections of the vaporization system were wrapped with Omega heater tape

and fiberglass insulating tape. The heaters were controlled with two PID controllers set

at a temperature of 100C, which was above the boiling point for the liquid isooctane. As

shown in Figure 3-4, a nitrogen gas with a flow rate of 0.8 liter per minute was

introduced into the vaporization system at the head of the preheat zone composed of a

0.625 inch diameter, 36 inch long 304 stainless steel tube packed with brass balls to

promote heat transfer by increasing the surface area. The nitrogen gas was heated in










HEPA
02 from -
supply tank


peristaltic pump liquid isooctane




C1- ^


brass balls


N2 from
supply tank
I


IEPA
ilter iver lne



delivery line


preheat zone


HEPA u
filter


N2 from
supply tank


calibration gases


Figure 3-4. Fuel vaporization system schematic.


Figure 3-5. Fuel vaporization system.


L
isooctane vapor
+ N2 or
calibra iw.n p'dL'


to bumrr


f









this region, and then it traveled into the vaporization zone consisted of a 0.25 inch

diameter, 36 inch long 304 stainless steel tube. The liquid isooctane was supplied into

this region via a variable flow Fisher Scientific peristaltic pump at a flow rate of 0.0015

liter per minute and vaporized by the heated nitrogen gas as well as the direct heat from

the hot surface of the tubing. By the nitrogen gas, the vaporized isooctane was carried to

the burner passing through a delivery line composed of roughly 50 inches of 0.25 inch

diameter braided PTFE hose. This zone was also heated to eliminate the fuel

condensation on the way to the burner. For warming the vaporization system up, the

heaters were turned on and the nitrogen coflow at a flow rate of 0.4 liter per minute

about one half an hour prior to any experimentation. This ensured that the vaporization

system was adequately heated before the liquid fuel was introduced into the system. The

oxygen stream with a flow rate of 2.6 liter per minute and the isooctane/nitrogen stream

exited the burner to produce the diffusion flame. The gas flow rates were regulated by

Alicat Scientific digital flow controllers whose accuracy was 1% of full-scale. The

maximum flow rates of the flow meters employed were 1 liter per minute for the nitrogen

and 10 liter per minute for the oxygen, respectively. These digital flow meters are shown

in Figure 3-6.

Table 3-2 summarizes the overall description of the fuel vaporization and delivery

system, and Table 3-3 tabulates the description of the gases and fuel.

3.3 Flame

As discussed above, a roughly 30 cm long flame was created by the burner used in

this study corresponding to the prevaporized isooctane/oxygen diffusion jet flame. The

main fuel was isooctane, C8H18, which is characterized by a relatively low boiling point,

promoting a stable fuel delivery system, and is compatible with a previous study








(Masiello, 2004) that probed the soot inception and growth regimes. Figure 3-7

elucidates the chemical structure of isooctane.


Ut


Figure 3-6. Alicat Scientific digital flow meters employed for regulating the flow rates of
nitrogen coflow and oxygen.

Table 3-2. Summary of equipment for fuel vaporization and delivery system.
Device Manufacturer Model Description
A Variable flow peristaltic
Peristaltic pump Fisher Scientific 13-876-4 Variable flow peristaltic
pump
Heater tape- 313
preheat zone Omega SRT101-060 313W
preheat zone
Heater tape-
vaporization and Omega SRT051-060 156 W
delivery line
Braided PTFE PTFE-lined stainless steel
hose Swagelok SS-4BHT flexible hose
Thermocouple Omega Type K Thermocouple
Heater controller Omega CN9000A 2 PID controllers
HEPA filter Gelman Laboratory 12144 2 HEPA filters
Digital flow 0-1 SLPM, accurate to 1% of
meter-N2 coflow A t S c full-scale
Alicat Scientific
Digital flow 0-10 SLPM, accurate to 1%
meter-O2 coflow _of full-scale
60-1200 rpm, 120V, 50/60
Magnetic stirrers Fisher Isotemp 11-601-16S
Hz










Table 3-3. Summary of gases and fuel used in the study.
02 Praxair UN 1072 >99% pure
N2 Praxair UN 1066 >99% pure
Fisher
Isooctane ienifi 0296-4 HPLC-grade
Scientific
CH4 (used for Praxair UN 1971 Ultrahigh purity, 99.97% pure
calibration)


H

H--C --H
H H H H

H -C---C C C C H

H H H

H-C -H H-C -H
I I
H H
isooctane
2,2,4-trimethylpentane
C8H18

Figure 3-7. Chemical structure of isooctane.

In order to characterize the flame, Froude number and Reynolds number were

calculated for the flame. The detailed calculations are attached in Appendix A. First, the

Froude number was 11.8, which indicates that the flame was momentum-controlled

rather than buoyancy-controlled, which was by design. Second, the Reynolds number

was approximately 761.3, which denotes that the flame was laminar flow, as it was

smaller than the critical value of 2300. Furthermore, the fuel equivalence ratio was

calculated as 1.08 based on the oxygen and isooctane flow rates. While this value

corresponds to fuel rich, significant additional oxygen is expected to diffuse into the

flame. To better understand the stoichiometry, a study was made to evaluate the smoke

point for the unseeded flame. Oxygen flow rates were adjusted to obtain the fuel-rich

flame that operated under the smoke point. Further discussion will be presented in









Chapter 4 in detail. Visible smoke was emitted above the flame tip when the oxygen

flow rate was below 4.24 L/min. The oxygen flow rate of 2.6 L/min was selected for all

experiments, based on the need for a reasonable smoke load to explore soot suppression.

Both the unseeded and iron seeded flames were tested under the same conditions, except

for the presence of the iron pentacarbonyl. For the seeded flames, the iron pentacarbonyl

was added to the liquid isooctane supply in 4000 ppm quantities by mass (-0.11% Fe per

mass of fuel) and was delivered to the burner through the fuel stream. The selection of

this value is discussed below. In all experimental stages, a magnetic stirrer was used to

prevent the iron pentacarbonyl from setting during experiments. The flame operating

conditions are summarized in Table 3-4 below.

Table 3-4. Description of the flame operating conditions.
Stream Flow rate
CsHis (liquid) 1.5 mL/min
N2 0.8 L/min
02 2.6 L/min
Fe(CO)5 (seeded flame only) 4000 ppm


3.4 Optical Systems and Diagnostics

3.4.1 Light Scattering System

Laser light scattering techniques were used to determine the scattering properties

of soot particles in the unseeded and seeded flames. The optical setup for this scattering

system is sketched in Figure 3-7, and the optical components are summarized in Table 3-

5.

For the laser light scattering experiments, a frequency doubled Q-switched 532 nm

Nd:YAG pulse laser (Continuum, Minilite ML-II) was used as a light source. The laser

was operated at 10 Hz with a pulse energy of 0.3 mJ/pulse. The laser was first turned 450










with 532 nm dichroic mirror. The laser beam then passed through an aperture to cut out

the Gaussian "edge" of the laser intensity profile before it was directed through a

focusing lens. Finally, the focused beam passed through the center of the flame at

desired vertical position above the burner lip, and was terminated at a beam dump. The

cross sectional area of the beam was 0.0033 cm2 at the center of the flame. The scattered

light from the soot particles was collected by a photomultiplier tube (PMT) at 90 degree

angle from the incident beam. The scattered light first passed through neutral density

(ND) filters in the collection optics line. The ND filters were required to attenuate

scattered beam intensity for maintaining signal linearity. After several ND filters, the


Nd:YAG 532 nm pulsed lase Oscilloscope

S=00 High Voltage supply





Plano-convex lens, f=250 mm

Black tube
Black tube Opaque Plexiglas
Aperture F le

532 nm 45 dichroic mirror


Figure 3-7. Top view of the light scattering system setup.









Table 3-5. Components of scattering system apparatus.
Device Manufacturer Model Description
Equipment
532 nm frequency Q-switched, 5 Hz, 2.4
doubled Nd:YAG Continuum Minilite ML-11 mJ/pulse FWHM=5 ns,
Laser 20 mJ max
Beam dump Kentek ABD-2 Beam dump
Photomultiplier
Pho ltiper Hamamatsu R2949 Photomultiplier tube
tube
Photomultiplier Products for PR1402CE Photomultiplier tube
tube housing Research, Inc. housing
500 Hz, 4 GS/s digital
LT 372,
Oscilloscope LeCroy WaveRunner oscilloscope with 50 Q
WaveRunner r
termination
Precision high Stanford Research PS325 Digital high voltage
voltage supply Instruments power supply

Double shielded Pasternack Double shielded coaxial
RG-223/U
BNC cable Enterprises cable to reduce line noise
Translational Micrometer-adjusted
stage translational stages
Optics
532 nm dichroic CVI Laser 45 degree, 532 nm
CVI Laser
mirror dichroic mirror
Aperture Newport ID-1.0 2 apertures
BBAR coated, 430-700
Plano-convex lens Newport KBX079AR.14 nm, 25.4 mm diameter,
250 mm focal length
FDU-2.0 102.0 attenuation

Neutral density Optics for FDU-1.0 1010 attenuation
filters Research
FDU-0.3 100.3 attenuation (nominal
values)
Polarizer Newport
T > 50%, 25.4 mm
532 nm line filter Newport 10LF10-532
diameter
2 apertures in collection
Aperture ID-0.5
optics
UV coating, 100 mm
Biconvex lens focal length, 25.4 mm
diameter









scattered light was passed through a 532 nm band pass filter, a vertical polarizer, a first

aperture, biconvex lens, and a second aperture in the collecting tube. The 532 nm band

pass filter effectively removed all wavelengths except 532 nm. The vertical polarizer

ensured that the scattered radiation observed was only vertically polarized, which

matched the vertically polarized incident beam. The polarizer contributes to reduction of

depolarized stray light, but it can also block the scattered light because the arbitrary shape

of the soot agglomerates can cause the scattered light to be depolarized. It is noted that

RDG scattering theory as formulated relies on only vertically polarized scattering light.

Two apertures reduced background noise (i.e., stray light) and ensured that the scattered

light was collected only from the small scattering volume defined in the flame. The

biconvex lens of 100 mm in focal length defined a solid collection angle of about 0.05 sr.

The scattered light was finally incident on a PMT and the scattered intensity was

recorded on a digital oscilloscope. A precision high voltage supply set to -650 V drove

the photomultiplier tube.

The photomultiplier tube (PMT) is a high gain detector so that it is very useful for

low light level detection. Photons are converted into an electric signal with a small load

resister (less than 100 Q in common) by a phenomenon, namely the photoelectric effect.

The photomultiplier tube contains a photosensitive cathode and a collection anode that

are separated by several electrodes, called dynodes, providing electron multiplication or

gain. When the photocathode is exposed to the electromagnetic radiation, a number of

photoelectrons are ejected by the photocathode and hit the first dynode. These electrons

strike the next dynode, which results in releasing additional electrons. This









multiplication process continues until electrons arrive at the anode. A typical PMT is

shown in Figure 3-8.


h v r Photocathode
hv Anode
-I

I Dynodes



Eo E EZ E3 E4 E, E6 E7 Es E9


Eb R

Figure 3-8. Photomultiplier tube. A series of dynodes between cathode and anode
provide internal gain.

PMTs are generally able to output a linear response to a continuous signal source

over several decades. However, a pulsed nanosecond-scale laser can easily invoke a non-

linear response in the PMT due to the sudden flux of incident photons. Commonly, a

PMT is limited to only about one decade of linearity in such a system, therefore signal

linearity must be carefully considered. When the strength of a scattered signal to a PMT

exceeds the linear limits, the signal must be attenuated to bring the output of the PMT

back into the linear response regime using neutral density (ND) filters. A ND filter is

characterized by a broad and steady transmission profile over a wide range of

wavelengths. An x ND filter attenuates by a factor of 10x. For examples, a 0.3 filter

should attenuate the signal by a factor of 100.3, or approximately 2; thus, the output signal

was expected to drop by about one half if the PMT response was linear. The optical

densities of the ND filters used were calibrated previously and are summarized in Table

3-6.









Table 3-6. Real optical densities for various ND filters
Filter Optical density at 532 nm
0.3A 0.284
0.3B 0.299
1.0A 0.808
1.OB 0.806
2.0A 1.728
2.0B 1.717


As shown in Table 3-6, a total of 6 ND filters were available in a variety of

attenuating strengths; thus, maintaining the PMT in its linear regime was simply a matter

of adding and removing filters from the PMT incident path. Linearity was checked with

nitrogen gas prior to every experiment by placing a 0.3 neutral density filter in front of

the collection optics and ensuring a factor of 2 intensity reduction. The ND filters for

individual height are tabulated in Table 3-7, which were used to maintain a comparable

and linear signal over all heights.

As discussed in Chapter 2, the differential scattering coefficient is a key parameter

for determining number densities of the scattering particles in the flame. The scattering

signal obtained by the PMT is defined as

S, = Io /7(A VA)No-r, (3-1)

where Io is the incident laser intensity, r is the efficiency of the collection optics and the

PMT detector, AV is the scattering volume, and AQ is the solid angle of observation. N

and C' are number density and the differential scattering cross section as discussed

previously. The parameters r, AV, and AQ may be measured although it is hard to

determined them individually with the utmost precision. By taking the ratio of a

reference scatterer signal to the signal from the scatterer of interest, the direct evaluation









Table 3-7. The usage of the ND filters for individual height.
Position Height N.D. Filter
Calibration 0.80 0.3B
1 9.40 2AB+1AB+0.3A
3 10.10 2AB+1AB+0.3A
5 10.90 2AB+1AB+0.3A
7 11.70 2AB+1AB+0.3A
9 12.70 2AB+1AB+0.3A
11 13.70 2AB+1AB+0.3A
13 14.90 2AB+1AB+0.3A
15 16.20 2AB+1AB+0.3A
16 16.95 2AB+1AB+0.3A
17 17.75 2AB+1AB+0.3A
18 18.60 2AB+1AB+0.3A
19 19.60 2AB+1AB+0.3A
20 20.65 2AB+1AB
21 21.85 2AB+1AB
22 22.80 2AB+1AB
23 23.50 2AB+1AB
24 24.20 2AB+1AB
25 25.25 2AB+1AB


of the common terms rj(AVAQ) can be avoided. With this approach, N C'' can be
SSoot

calculated using the following equation,


(N -')5o, = (N.>-')CH, SSoot (
SCH4, measured SL Y


(3-2)


Recall N-. J is the differential scattering coefficient. In this equation, r is the

transmission through the flame which will be determined in a later section, SL is stray

light obtained by a calibration, and the subscript CH4 refers to the methane calibration

gas whose signal and the cross-section will be reported in the next section.









3.4.2 Light Scattering Calibration

A calibration is necessary for determining and illustrating the amount of stray

light presented in the system. Stray light, such as any ambient light and laser light

reflected from surfaces, which enters through the scattering collection optics, can distort

the experimental results considerably. The magnitude of stray light can be as large as the

scattered signal, thus it is very crucial to reduce errors caused by stray light.

As mentioned briefly in the foregoing section, several efforts were already used to

minimize stray light entering at the PMT. First, using apertures and lenses in the

collection optics, a very small scattering volume was defined in the experiment. These

apertures also played a role to block any stray light from outside the scattering volume.

Highly reflective surfaces of optical mounts were either painted in black or covered in

black felt to minimize reflections of laser light from surfaces. In addition, by allowing

the beam to pass through a black painted tube placed in the beam path, as well as the

barrier with surrounding opaque Plexiglas covered in black felt, any stray light from

outside the scattering volume was largely blocked from introducing into the PMT. In

spite of such efforts, it is impossible to completely get rid of stray light from the system.

Therefore, the stray light calibration was necessary to compensate for stray light.

The calibration was performed using methane (CH4) and nitrogen whose

differential scattering cross sections were already known as summarized in Table 3-8.

The calibration gases shared the delivery line of the vaporization chamber, as depicted in

Figure 3-4, thus the fuel gas was expelled from the burner for calibration measurements.

A series of brass plug valves were used to shut off the flow from the vaporization

chamber during calibration, and vice versa during flame operation. The supplier's

specifications of methane were listed in Table 3-3 before. The calibration gases were









injected into the measuring area 8 mm far from the burner lip. Both gases flowed at a

rate of approximately 10 L/min, which was controlled by a rotameter (GE700 Gilmont)

flow tube. The scattering measurement for the calibration was carried using the identical

configuration as used for the soot scattering study. A type K thermocouple was used to

measure the temperature of gases exiting from the burner through the heated fuel delivery

line. Over 24 sets of the experiment, the temperatures were on average 346 K, with the

standard deviation of 1.68 K for methane, and 340 K with the standard deviation of 1.37

for nitrogen, respectively. The number density N, differential scattering cross section,

Cu, (cm2/sr) and differential scattering coefficient, K' (cm-sr-1) for each gas were

determined using temperatures measured each time. Number densities were calculated

using isobaric density data tabulated by the National Institute of Standards and

Technology (Lemmon et al., 2003). As for the differential scattering cross sections for

each of these gases, they were previously reported to be 4.56E-28 cm2sr1, and 2.12E-28

cm2sr-1, respectively (Rudder and Bach, 1968) at an incident wavelength of = 694.3 nm.

With those values, C at an incident wavelength of = 532 nm was calculated using the

equation below,


v -4 I \ \ -- 122 (3-3)


where nx is the wavelength-dependent refractive index (Landolt-Bornstein and

Funktionen, 1962). The number density, the differential scattering cross section and

coefficients for calibration gases at the wavelength of 532 nm are summarized in Table 3-

8.









Table 3-8. Average of the number densities, differential scattering cross sections, and
scattering coefficient sets for methane and nitrogen calibration gases at 1 atm
with standard deviation for 24 experimental.
T (K) N (cm-3) vv (cm2/sr) K'vv (cmlsr1)
Methane Average 346.29 2.12E+19 1.35E-27 2.87E-08
Methane -
S.D. 1.68 1.03E+17 1.40E-10
Average 340.04 2.16E+19 6.25E-28 1.35E-08
Nitrogen
S.D. 1.37 8.32E+16 5.20E-11
S.D. : Standard deviation


The reference calibration ratio Rref for methane to nitrogen can be determined by

the ratio of their differential scattering coefficients as expressed below,


Rref = CH (3-4)
N2 reference

The reference ratio, Rref, can then be related to the measured signals and the

corresponding stray light through the relationship,

S(Svw,CH4 measured ) SL
Re/ = (3-5)
w,(SN2 measured)- SL

This value was used to determine the stray light contained in the measured scattering

signals from the calibration gases. In the equation, SCH4,measured and S,N4 measured are the

measured scattering signals of methane and nitrogen, respectively. With this relationship,

the stray light in the system can be quantified as


SL = Rref SppN2 SppCH4 (3-6)
Rref 1

In the light scattering studies, stray light calibration measurements were taken prior to

every flame study to determine an experiment-specific stray light value. Over the range

of experimental measurements, the average contribution of stray light was around 38.5%







70


of the methane calibration signal. The detailed discussion and calculation of stray light is

represented in Appendix D.

3.4.3 Signal Processing

For scattering analysis, a number of signals were measured, including the dark

signal (no laser) for calibrations, the calibration signals, the flame dark signals (flame

only, no laser), and the flame scattering signals. The scattered signal from the calibration

gases and the flame from a typical scattering experiment are represented in Figure 3-9.


0.001


0



--0.001

o


2 -0.002
CH


-0.003


_0 nn0A


1.05 10-7 1.1 10-7 1.15 10-7 1.2 10-7 1.25
Time (s)


10-7 1.3 10-7 1.35 10-7 1.4 10'


Figure 3-9. Sample scattering signals from methane, nitrogen, and flame. Calibration
gases are attenuated by a factor of 100.3 and flame signal is attenuated by a
factor of 105.43 for signal linearity.

The dark signals were recorded prior to the measurement of the scattered signals to

normalize the baseline of the flame and the calibration gases. Such dark signals were

integrated over a 50 ns full peak width. Since the PMT responses of the baselines to bulk

signals were not consistent at all times, they were offset by subtracting the signal baseline


-


-






"V / Calibration dark
S/ -- Methane with 0.3 ND filter
-y --------Nitrogen with 0.3 ND filter
Nitrogen without filter
-- Flame dark
-- Flame with 5.34 attenuation
, I , I ,, I ,









from the integrated scattered signal. The offset dark signals were subtracted from the

offset scattered signals, which were named the calculated signal. The calculated signals

then needed a correction for any attenuation factor that was used to preserve PMT

linearity. For the data presented in Figure 3-9, the calculated calibration signals were

attenuated by a factor of 100.3, while the calculated flame signals was attenuated by a

factor of 105.34. Therefore, the calculated signal still including the influence of stray light

was corrected by multiplying it by the overall attenuation factor. The calculated

calibration signals were found to be 0.249 V-ns for methane and 0.14 V-ns for nitrogen

respectively, in this example. This yielded a calibration ratio of 1.78, 16.4% deviation

from the ideal reference ratio, Rref, of 2.128 for this experiment, which indicates the

magnitude of the stray light. Using Equation 3-5 with the calculated calibration signals

and the reference ratio, the stray light was determined to be 0.043 V-ns in this case. The

average results of calibration measurements over all experiments are summarized in

Table 3-9 along with the standard deviation. The average calibration ratio was

determined to be 1.51, with a standard deviation of 0.22, 29% deviation from Rref.

Overall, the stray light signal was approximately 40% of the methane gas signal for the

typical experiment.

Table 3-9. Average results of calibration gas signal including stray light, a calibration
ratio, stray light signal, the percentage of stray light, and the ideal reference
ratio along with the standard deviation over all scattering experiments.
Methane Nitrogen Ratio Stray Light Percent S.L. Ideal Ratio
Average 0.30 0.21 1.51 0.13 38.48 2.126
S.D. 0.05 0.06 0.22 0.07 18.53


The final calibration signals were determined by subtracting the stray light from the

recorded calibration signals. This yielded true calibration signals of 0.206 V-ns for









methane and 0.097 V-ns for nitrogen, respectively, for the above example. In the same

manner, the calculated flame signal was determined to be 23,648 V-ns. The stray light

signal may be subtracted from this signal to calculate the true flame signal; however it

may be neglected without significantly altering the scattering results because it is orders

of magnitude smaller than the flame signal. At any rate, the stray light signal calculated

in this manner will be used to determine the differential scattering coefficient of the soot

with Equation 3-2. All that remains to be determined in order to extract the differential

scattering coefficient from the scattering data is the transmission of laser light through the

flame, which will be discussed in a later section.

3.5 Laser Power Measurement

Accurate scattered signals may be obtained as long as modest laser pulse energies

are used. However, a laser pulse can significantly heat and vaporize soot particles if the

beam is focused to a small cross-sectional area. Therefore, it is important to determine

suitable laser energy to eliminate any soot vaporization during the flame studies.

According to Dasch (1984a and b), laser fluences greater than 0.2 J/cm2 from a

submicrosecond pulsed source can cause vaporization of soot particles, resulting in

reducing the light scattering and extinction characteristics of soot by an order of

magnitude. Recent work by Yoder et al. (2005) quantified vaporization effects, and

reported vaporization of soot particles down to a fluence of 0.1 J/cm2. In order to ensure

no vaporization effects, the laser output power was measured with a powermeter and

adjusted to the appropriate energy for the spot size.

The laser power was 3.04 mW on average with standard deviation of 0.29 over 10

measurements, which yielded 0.31 mJ/pulse for the laser pulse rate of 10 Hz. These

correspond to a laser fluence of 0.092 J/cm2 based on a focal spot of 0.0033 cm2, where










the term fluence refers to the laser energy per the beam area. The lowest laser energy

was attained by simply reducing the laser pump energy. The beam area was found by

ablating ink off a slide, at increased pulse energy, and measuring the area in which the

ink was removed. The beam spot was quite a perfect circle having a diameter of 0.65

mm. A summary of the laser beam power properties is given in Table 3-10.

Table 3-10. Summary of laser beam power properties for light scattering measurements.
mW mJ Fluence(J/cm )
Average 3.04 0.304 0.0916
S.D. 0.29



3.6 Transmission

In order to complete the calculation of the differential scattering coefficient using

Equation 3-2, the transmission is required to be measured at each of the flame heights

investigated. Figure 3-10 shows the experimental setup for the transmission system.


Nd:YAG 532 nm pulsed laser Power meter display







Plano-convex lens, f=250 mm

Black tube Opaque Plexiglas
Aperture Flame 532 nm line filter

532 nm 45 dichroic mirror
Aperture
Power meter receiver


Beam Dump

Figure 3-10. Top view of the transmission system setup.









This setup is identical to the scattering system except that the laser does not

terminate at a beam dump but encounters the power meter receiver. Moreover, a narrow

aperture and a 532 nm line filter were placed in front of the power meter receiver. The

aperture played a role to prevent the forward scattered light from entering the power

meter. The influence of the forward scattered light could be limited by distancing the

power meter receiver from the flame as well. These efforts resulted in only a small solid

angle of observation. In addition, the 532 nm line filter blocked all light except the 532

nm that was detected by the power meter. The transmission instrumentation is described

in Table 3-11 in detail.

Table 3-11. Description of transmission apparatus.
Device Manufacturer Model Description
Equipment
532 nm frequency
532 nm frequency .Q-switched, 5 Hz, 2.4 mJ/pulse
doubled Nd:YAG Continuum Minilite ML-11 -It ns, 20 mJ m
FWHM=5 ns, 20 mJ max
Laser
PM5200 Power meter
Power meter Molectron
PM3 Power meter head
Optics
532 nm Dichroic L 45 degree, 532 nm dichroic
mirror CI L mirror
Aperture Newport ID-1.0 Aperture
BBAR coated, 430-700 nm,
Plano-convex lens Newport KBX079AR.14 25.4 mm diameter, 250 mm
focal length
532 nm line filter Newport 10LF10-532 T>50%, 25.4 mm diameter


The measured power of the laser through the flame was ratioed with the power measured

at a position outside of the flame to obtain the transmission through the flame. From the

Beer-Lambert law, the transmission is generally defined as










r =- =r exp(-KexL), (3-7)
'0

where I, is the incident intensity of the laser, as measured at a position outside of flame,

Itrans is the transmitted intensity measured through the flame, L is the optical pathlength

through the flame, and Kex, is the extinction coefficient. With the transmission data

determined in this manner, the extinction coefficients could be simply obtained only if

the optical pathlengths are known. The optical pathlengths corresponding to 12 heights

were determined statistically by taking 25 pictures of the unseeded and seeded flames

each and analyzing them.

The extinction coefficients evaluated in Equation 3-7 reflect an average value

through the flame, which is correct as used to correct the scattering coefficient using

Equation 3-7. However, if spatially resolved extinction data is desired, deconvolution

techniques are often employed. In this study, the width of the flame used was not

sufficient to apply these techniques; as a consequence, no deconvolution (e.g. Abel

inversion) was used.

3.7 Thermophoretic Sampling and Transmission Electron Microscopy

The conventional light scattering technique is a non-intrusive diagnostic tool that

can effectively infer optical properties from a system of particulates in situ; however, it

has been limited to point measurements, and cannot extract fractal properties of

aggregates such as the radius of gyration and fractal dimension. Therefore, it is required

to use a complementary technique that can measure the size and shape of soot

agglomerates. Such information can be obtained utilizing ex situ transmission electron

microscopy (TEM) following thermophoretic sampling.









3.7.1 Thermophoretic Sampling

The soot aggregates were extracted from the flame using thermophoretic sampling

for the transmission electron microscope image analysis of size and morphology of

aggregates. Soot samples were collected directly on formvar-carbon coated 150-mesh

copper TEM grids (Electron Microscopy Sciences, Hatfield, PA, Part No. FCF 150-

CU50). A schematic of this device is shown in Figure 3-11.












4 A



S02




Isooctane + N2 A


Figure 3-11. A setup of thermophoretic sampling and grid. A) Side view. B) Formvar
carbon-supported 150 mesh copper grid.

Each film was held by a tweezer attached to a holder. The sampling surface faced

toward the flow direction. A film was installed outside the flame and swept through the

flame allowing soot aggregates to deposit directly on the grid. Samples were collected at

12 different heights along the vertical axis of both the unseeded and seeded flames.









It is significant that the exposure time of the film should be long enough to capture

a reliable amount of sample, but short enough to avoid melting the grid or oversampling

the aggregates. The sampling times were not controlled automatically (by e.g. using a

double acting pneumatic cylinder, solenoid valves and timers) but manually in the present

study. As a result, each sample had a slightly different residence time in the flame, but

care was taken such that all exposure times of the films were about 1 s. In spite of the

absence of an automatic controlling device, it was concluded that the samples obtained in

the present study were sufficient for TEM analysis. Representative micrographs are

presented in Chapter 4.

Another issue of thermophoretic sampling is to obtain samples at a desired position

in the flame. To do this, a shield is used for preventing the particles from depositing on

the grid while the sampling probe was out of a desired sampling location. Alternatively,

the shorter transition time is used while the film was traveling out of the desired region.

The flame employed in the present study was so thin that undesired sampling region was

relatively narrow enough to ignore the quantity of undesired samples. After exposed to

the flame producing soot aggregates, the grid was examined to observe the particle size

and morphology with a TEM.

3.7.2 Transmission Electron Microscope

The soot samples collected by thermophoretic sampling were observed using a

JEOL 2010F analytical electron microscope system with a point resolution of 0.19 nm.

Figure 3-12 shows a photograph of the TEM system.

The diffraction grating was used to calibrate the TEM. Magnifications used for the

present measurements ranged from 50,000 to 330,000, corresponding to 500 nm and 50

nm sized scale bars, respectively. For analysis, each soot aggregate was randomly picked









at low magnification, and then analyzed at the optimum magnification. Each micrograph

has a scale bar indicating length, and particle diameters can be simply measured from the

pictures, which will be discussed in more details in Chapter 4. TEM micrographs of soot

aggregates extracted from the center region of the diffusion flame were examined as a

function of increasing height. In addition to TEM analysis, the Energy Dispersive X-ray

Spectroscopy (EDS) was performed for precise detection of chemical components in

samples, including iron species, using an Oxford INCA Energy TEM System.

























Figure 3-12. Photograph of the TEM system.

3.8 Spectroscopic Techniques

Up to this point, all studies have focused on the quantitative viewpoint such as

size and concentration of soot in the flame. Next, qualitative aspects will be examined

such as the identification of species in the flame. The particular species present in the

flame provide information on the mechanisms of chemical reactions. Moreover,









preliminary testing for eliminating the external environment interference and variations

was performed with a CO flame. These were summarized below.

3.8.1 Preliminary CO Flame Study

CO flame was employed to pretest an availability of spectroscopic techniques such

as LIF and in situ Raman spectroscopy before such techniques are applied to the

isooctane flame study. It is well documented that the iron pentacarbonyl gas in a CO and

02 flame is thermally decomposed and believed to form Fe203 chain agglomerates

(Cheng et al. 1991). Further, it has an advantage that the soot particles do not interfere

with a particular species to be detected because carbon dioxide is the only product from

the CO flame. Figure 3-13 depicts iron pentacarbonyl vaporization system and a burner

used for the CO flame study.


CO or N2 I CO+Fe(CO)5 I.D=014 "

O.D.=0.25 "









Heater




Figure 3-13. Vaporization system of iron pentacarbonyl and a CO flame burner.

The carbon monoxide gas at a flow rate of 0.45 liter per minute regulated by an

Alicat Scientific digital flow controller was introduced into the fuel additive vaporization

vessel through a tube. The fuel additive vaporization system was designed to seed the









CO gas with iron pentacarbonyl. To increase the concentration of the additive in the CO

gas, the outside of the vaporization vessel was heated by Omega heater tape and its

temperature was measured with a stainless steel type K thermocouple. A heater was used

to maintain the constant temperature of the vessel at 60 OC. Figure 3-14 shows a

photograph of the iron pentacarbonyl vaporization vessel and the heater.




























Figure 3-14. A photograph of the iron pentacarbonyl vaporization vessel and the heater.

The CO gas was passed through the Fe(CO)5 liquid and combined with Fe(CO)5

gas before being delivered to the burner. The delivery line is also necessary to be heated

to prevent the condensation of Fe(CO)5. Otherwise, the condensed Fe(CO)5 blocks the

gas flow which results in the flame jumping up and down. The CO flame was

approximately 6 cm tall reacted with air. The addition of iron pentacarbonyl visibly









changed the flame from the blue CO flame to the bright orange flame. This results from

the blackbody radiation of the iron oxide particles present in the flame. The unseeded

and seeded CO flames are shown in Figure 3-15.






















Figure 3-15. Photographs of CO flame. A) unseeded flame, B) iron seeded flame.

After all experiments were over, the nitrogen gas was filled in the vessel to prevent

the dissociation of Fe(CO)5.

3.8.2 Experimental Apparatus of Laser Induced Fluorescence Spectroscopy

Laser-induced fluorescence spectroscopy (LIF) was employed to trace Fe atomic

fluorescence emission in the seeded flame. The experimental apparatus for LIF is

illustrated in Figure 3-16.

A frequency tripled Q-switched 355 nm Nd:YAG laser was used as a pump laser

source. The laser was operated at a 10 Hz repetition rate with around 200 mJ/pulse

energy. For the Fe excitation, the frequency-tripled laser output was tuned to the several

Fe resonant transitions band using Optical Parametric Oscillator (OPO). An OPO










converts photon energy of a pump laser into lower energy by means of nonlinear optical

interaction; thus, the higher photon energy of a pump laser is very essential. In addition,

since the gain depends on pump power, sufficient pump power is necessary to support

oscillation. In other words, the oscillation occurs only when the pump power reaches a

particular threshold level. Above threshold, the gain is also dependent on the amplitude

of the resonated wave. With pump wavelength of 355 nm, typical OPO yields the output

wavelength ranging from 400 nm and 1000 nm. The OPO consists of not only an optical

resonator and a nonlinear optical crystal, but also doubling crystal so that the output

range can be broadened from 200 nm to 1000 nm. In this study, the Fe excitation

wavelength was 296.69 nm that was created by frequency-doubling of 578 nm produced

by means of nonlinear optical interaction. In spite of high pump power of the laser (-200

mJ/pulse), the output power passed all the way through optics in OPO drops around 2





iCCD Spectrometer 355 nm razor filter

Computer


Beam Dump 111o4 I in N d. .AG

Broadband mirror P
r I Collection lens


Aperature


Broadband mirror


Flame


team uump


Figure 3-16. The optical setup for laser-induced fluorescence spectroscopy.