University of Southern California Dissertations and Theses

Development of a novel heterogeneous flow reactor: soot formation and nanoparticle catalysis

(USC Thesis Other)

Development of a novel heterogeneous flow reactor: soot formation and nanoparticle catalysis

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content DEVELOPMENT OF A NOVEL HETEROGENEOUS FLOW REACTOR – SOOT FORMATION AND NANOPARTICLE CATALYSIS by Joaquin Camacho A Dissertation Presented to the FACULTY OF THE USC GRADUATE SCHOOL UNIVERSITY OF SOUTHERN CALIFORNIA In Partial Fulfilment of the Requirements for the Degree DOCTOR OF PHILOSOPHY (MECHANICAL ENGINEERING) December 2013 Copyright 2013 Joaquin Camacho ii Dedication This work is dedicated to my parents, Joaquin and Angelita, who gave me the strength and drive to succeed at my interests. iii Acknowledgements During my time at USC, I benefitted from support and guidance from a wide network of family, friends and colleagues. For example, I am sure that graduate life would have been a much greater challenge if I did not have the ability to drive across town to the comfort and welcoming feeling of visiting my parents and family. In addition, USC has been welcoming in the sense that many new friends have come to my aid and provided comradery during my courses and research endeavors. More importantly, my girlfriend and soon-to be Fiancee (as of time of publishing), Alison, has been the foundation for my happiness and inspiration. During my time at USC, I have been fortunate to have the comfort and satisfaction that I found the woman who will be the foundation of my family. Her intellect, beauty, sincere compassion and patience for my difficult side have allowed me to flourish. I would like to acknowledge her contribution to my life and future endeavors. I would also like to acknowledge my father and stepmother, Joaquin and Geli Camacho, who have raised me since elementary school in a loving and supportive environment. Their steady parenting was an ideal combination of discipline and freedom which allowed me to focus what could be a rebellious nature into more productive and powerful endeavors. My mother, Angie Del Rio, also raised me with patience and a sense of freedom during my early years in LA and Las Vegas. However, she also instilled a sense of confidence and high expectations in me which kept me on a true path. I have been fortunate to have graduate school in the same region as my numerous siblings. I am thankful for the ability to hang out with my brothers Danny, Eric and Eli for brotherly activities like Dodgers and Lakers games. I am also thankful for the touristy visits that my Mom and little sister Cindy frequently take to LA and San Diego with me. Being there to watch my younger siblings, Emilio and Evelina, grow up has also provided me a sense of pride iv and hope that I can be an example of success that they can possibly surpass. Christine and the rest of the Sanchez family have also been there throughout my college years. I will never take for granted the advantage of having my family close during my graduate years at USC. Another advantage for me at USC was the exceptional guidance and mentorship of my advisor, Prof. Hai Wang. I am grateful for the opportunity work under such a prolific and influential scientist. My productivity was inspired by many innovative ideas and fundamental research questions of Prof. Wang. Progress during my PhD was aided by clear objectives and the resources to reach those objectives. In addition, my personal, professional and presentation skills were greatly enhanced by his mentorship and example. I am fortunate to have worked with many good scientists and great people during my time in the USC Combustion Kinetics lab and AME department. I am very grateful to Aamir Abid for taking me on for my first project and introducing me to experimental life. In addition, he and David Sheen allowed me to help with what would be my first appearance on a scientific publication. David has since been a helpful scientific mentor and personal friend during my development. Enoch Dames has helped me in my scientific growth through many helpful discussions in the lab and he has also shown me many fun times as a personal friend. Erik Talmachoff and Saro Nikraz have also helped during my early years in the lab and have also been a lot of fun. A newer generation of co-workers in the Kinetics Lab have been very helpful such as Sydnie Lieb, Mohammad Janborgorgi, Stelio Kumlis, Rei Tenkgo, Yuxuan Xin, Duxing Du, Bin Yang, Tsutomu Shimizu, Gregory Proskrybeshev, Yujie Tao and Gao Shaokai. My committee was composed of professors helped my by guiding and challenging me. Prof. Egolfopoulos has instilled adeep appreciation in me about the foundations of my research. My v experimental abilities have been improved by his expectation that the fundamental theory should be strong. Prof. Veronica Elliasson and Prof. Ron Henry have given me helpful discussions on my academic and professional development. Within the AME department, Peter Veloo and Okjoo park have provided comradery and guidance to me during our experiences as combustion PhDs at USC and many excursions within the combustion community. More recently, Vyas Gururajan and Jaagan Chayachandran have tutored me on the math and physics of the counterflow flame and have been fun as drinking buddies. In addition, Leo Velasco, Hugo Burabano, DJ Lee, Chris Xiouris and Roe Burrel have been good colleagues and fun for much needed BBQs and happy hours. vi Table of Contents Dedication ………………………………………………………………………………….…. ii Acknowledgements …………………………………………………………………………... iii List of Tables ……....................................................................................................................... ix List of Figures ……..................................................................................................................... ix Abstract ……………………………………………………………………………………… xii Chapter 1: Introduction………................................................................................................... 1 1.1 Soot Formation Theory and Background…………………… ……......................................... 1 1.2 General View of the Role of Parent Fuel Structure in Premixed Flames …….……………... 4 1.3 Detailed Observations of the Fuel Structure Dependence in Premixed Flames …………..… 8 1.4 Classical Description of Soot Oxidation by O 2 …………………………………………..… 10 1.5 Combustion Enhancement by Freely Suspended Nanoparticle Catalyst …………………... 12 1.6 Dissertation Organization ……………………………………………………………… ..… 1 3 1.7 Chapter 1 References ………………………………………………… ……………………. 14 Chapter 2: Experimental Methods ........................................................................................... 18 2.1 Introduction ………………………………………………………………………………… 18 2.2 Burner-Stabilized Stagnation (BSS) flame configuration for nascent soot studies ………... 18 2.3. Oxidation rate of nascent soot in a novel aerosol flow reactor ………………………….… 22 2.4 in-situ Palladium nanoparticle synthesis for fundamental surface catalysis ………….. …… 25 2.5 Scanning Mobility Particle Sizing (SMPS) for nascent soot and Pd nanoparticles ……...… 28 2.6 Non-dispersive Infrared Spectroscopy for CH 4 and CO 2 ………………………………...... 32 2.7 Chapter 2 References …………………………………………………………………. …… 3 4 vii Chapter 3: Evolution of Size Distribution of Nascent Soot in n- and i-Butanol Flames .… 36 3.1 Introduction …………………………………………………………………………...….. 3 6 3.2 Experimental Methods ……………………………………………………………….... …. 3 7 3.3 Gas-phase kinetic model for high-temperature oxidation of butanol isomers …………..... 39 3.4 Experimental Results ……………………………………………………………..……….. 40 3.5 Numerical modeling of gas-phase species leading up to benzene ………….…...…….…... 47 3.6 Conclusion ………………………………………………………………………………… 50 3.7 Chapter 3 References ……………………………………………………………………. .. 51 Chapter 4: Time Dependant Soot Formation in Premixed C 6 Hydrocarbon Flames ..… 53 4.1 Introduction ………………………………………………………………………...……... 53 4.2 Experimental Methods ……………………………………………………………………. . 56 4.3 Experimental Results ………………………………………………………………..….…. 60 4.4 Numerical modeling of gas-phase species leading up to benzene ………………...………. 77 4.5 Conclusion ……………………………………………………………………………….… 84 4.6 Chapter 4 References …………………………………………………………………... ….. 85 Chapter 5: Kinetics of Nascent Soot Oxidation in a Flow Reactor ……………………...… 88 5.1 Introduction ………………………………………………………………………….....….. 88 5.2 Experimental Methods …………………………………………………………………..… . 91 5.3 Experimental Results ………………………………………………………………………. 92 5.4 Simplified Surface Oxidation Model ……………………………..…………………..…… . 95 5.5 Conclusion …………………………………………………………………………….…… 97 5.6 Chapter 5 References ………………………………………………………………...…… .. 98 viii Chapter 6: Catalytic Methane Oxidation by Freely Suspended Pd Nanoparticles …...… 101 6.1 Introduction ………………………………………………………………………………. 101 6.2 Experimental Methods ………………………………………………………………… …. 103 6.3 Results and Discussion .………………………………………………………………...… 106 6.4 Conclusion ………………………………………………………………………………... 110 6.5 Chapter 5 References …………………………………………………………………..…. 111 Chapter 7: Concluding Remarks and Future Work ……………………………………… 112 7.1 Concluding Remarks …………… ………………………………………………………... 112 7.2 Future Work ……….……………………………………………………………………... 114 Bibliography ………………………………………………………………………….……… 115 ix List of Tables Table 3.1 Summary of Flame Composition for the Premixed BSS Butanol flames …………. 32 Table 4.1 Summary of the premixed BSS flame compositions. The maximum flame temperature is 1800K for each flame ……………………………………………………………………… 52 Table 6.1 Flow Compositions for [CH 4 ] o = 1 %. Total flow rate = 43 SLPM ………………… 90 Table 6.2 Experimental Observation of Methane Oxidation Catalysis at [CH 4 ] o = 1% ……….. 90 List of Figures Figure 1.1 The critical equivalence ratio (  c ) as a function of the adiabatic flame temperature reported by Glassman and Takahashi [24,25] ………………………………………………….. 5 Figure 1.2 The correlation between the critical equivalence ratio (  c ) at T f = 2200 K and the number of C-C bonds reported by Takahashi and Glassman [24,25] ………………………….. 6 Figure 2.1 Schematic summary of the burner stabilized-stagnation (BSS) flame configuration 15 Figure 2.2 Schematic of the sequential burner - aerosol flow reactor setup for nascent soot oxidation ……………………………………………………………………………………….. 19 Figure 2.3 Schematic of the sequential Pd nanoparticle synthesis - aerosol flow reactor setup for methane oxidation catalyzed by freely suspended Pd ………………………………………….. 21 Figure 2.4 PSDF of the in-situ generated Pd nanoparticles. The number density and median particle diameter, thus surface area density, controlled by the precursor loading & carrier gas . 22 Figure 2.5 Schematic of the nano-DMA operating in dual flow mode [37] ………………….... 26 Figure 2.6 Schematic of a NDIR Gas analyzer configuration …………………………… ……. 28 Figure 3.1 Comparison of predicted and measured flames speeds (Su) of n-butanol/air flames at 1 atm and an unburned gas temperature of 343 K. The data are taken from [21] …………...… 35 Figure 3.2 Measured (symbols) and simulated (lines) temperature profiles. Open symbols and solid lines: i-butanol; filled symbols and dashed lines: n-butanol. The vertical error bars represent the limiting emissivity of 0.3 and 0.6 (see text) …………………………………………..……. 36 Figure 3.3 Measured (symbols) and simulated (lines) temperature profiles. Open symbols and solid lines: i-butane; filled symbols and dashed lines: n-butane. The vertical error bars represent the limiting emissivity of 0.3 and 0.6 (see text) ………………………………………..………. 37 Figure 3.4 Measured PSDF for i-butanol (open symbols) and n-butanol (filled symbols) flame 38 Figure 3.5 Comparison of PSDFs for i-butanol, n-butanol, i-butane and n-butane flames at selected burner-to-stagnation surface separations ……………………………………………. 39 x Figure 3.6 Volume fraction of nascent soot with Dp > 2.4 nm (symbols) measured at several Hp for all fuels tested. Lines are drawn to guide the eye. Inset: profiles shifted spatially to illustrate the mass growth rates beyond nucleation …………………………………………………….. 40 Figure 3.7 Mole fraction profiles of benzene computed at several Hp ……………….……… 43 Figure 3.8 Mole fraction profiles of C 2 H 2 (top) and C 3 H 3 (bottom) for computed at several H p 44 Figure 4.1 Measured (symbols) and simulated (lines) temperature profiles for the n-hexane flames at the given sampling locations, H p . The vertical error bars represent the uncertainty in thermocouple radiation corrections as described in the text …………………………………. 56 Figure 4.2 Measured (symbols) and simulated (lines) temperature for the C 6 hydrocarbon flames compared with H p = 1.2 cm. The thermocouple radiation correction for methylpentane was estimated from the flame composition and transport properties of n-hexane ………….…… 57 Figure 4.3 AFM images of soot from benzene (left) and hexane (right) flames taken from flame locations corresponding to 40 ms residence time ……………………………………..……. 58 Figure 4.4 Sphericity ratio and circularity vs. particle diameter for particles from benzene (solid symbol) flames and n-hexane (open symbol) flames ………………………………………. 59 Figure 4.5 Measured PSDFs for n-hexane (filled symbols) and 2-methylpentane (open symbols) flames expressed as a function of the particle surface area. Bi-modal distributions (solid lines) are fit to the PSDF at Hp = 1.0 and 1.2 cm to highlight nucleation size particles which persist late in the flame. As described in the text, recent morphology observations of nascent soot suggests that particles larger than D p = 5nm are not spherical thus the particle surface area is considered and the equivalent spherical diameter is shown on the upper x-axis ……………………….. 61 Figure 4.6 Measured PSDFs for n-hexene (open symbols) and cyclohexane (filled symbols) flames expressed as a function of the particle surface area. Bi-modal distributions (solid lines) are fit to highlight nucleation size particles which persist late in the flame ………….,……. 63 Figure 4.7 Measured PSDFs (symbols) for benzene flames fit to normal distributions (lines) which are expressed as a function of the particle surface area ………………………….…... 65 Figure 4.8 Representative 3-D plot of soot from the early growth stages, H p = 5.5mm, in a benzene flame. Particle 1 is a primary particle with a diameter of approximately 5nm and particle 2 is an early aggregate with a diameter of approximately 7nm ……………….……. 66 Figure 4.9 Volume fraction of nascent soot with Dp > 2.4 nm (symbols) measured at the given Hp for the C 6 H 14 fuels studied. Lines are drawn to guide the eye …………………………... 67 Figure 4.10 Volume fraction of nascent soot with Dp > 2.4 nm (symbols) measured for all the fuels studied as function of sampling distance and particle residence time. For the φ = 2.07 comparison, the benzene flame was not changed from φ = 1.7 because the fraction of soot increases beyond measurable limits in the benzene flame. Lines are drawn to guide the eye . 69 xi Figure 4.11 Volume fraction and number density of nascent soot with D p > 2.4 nm (symbols) measured for cyclohexane, n-hexene and a previously reported ethylene flame [22] as function of particle residence time. Lines are drawn to guide the eye ……………………………………. 71 Figure 4.12 Mole fraction profile of acetylene (top panel-solid lines), propargyl radical (top panel-dotted lines) and benzene (bottom panel-solid lines) computed at Hp = 1.2 cm for each of the fuels studied. The behavior of 2-methylpentane (not computed here) relative to n-hexane is assumed to be similar to the relative behavior of i-butane and n-butane fuels under comparable conditions [24] ………………………………………………………………………….……. 73 Figure 4.13 Reaction rate profiles computed for propargyl recombination (thin lines), butynyl + acetylene (thick lines) and dehydrogenation (dashed line) steps to benzene formation. Methylpentane and benzene fuels are not shown for clarity …………………………….…... 75 Figure 4.14 AFM image of a representative aggregate soot particle from benzene (a) and hexane (b) flames at residence time of 70 ms ………………………………………………………... 77 Figure 4.15 Summary of the time resolved evolution of nascent soot in terms of the PSDF. Soot nucleation is expected to stop later in the flame due to the depletion of soot precursors. The dashed line denotes flames such as benzene and cyclohexane flames which end nucleation relatively early ………………………………………………………………………….……. 79 Figure 5.1 Schematic of the coupled burner and flow reactor setup. Nascent soot is extracted at 0.8 cm from the BSS burner surface …………………………………………………………. 87 Figure 5.2 Measured PSDFs (symbols) of nascent soot from the ethylene BSS flame, probed in the flow reactor in an inert nitrogen flow at the initial (t = 0 sec, filled circles) and the final (t = 0.2 sec, open squares: 950 K; open triangles: 1000 K) residence times ……………………… 88 Figure 5.3 Measured PSDFs (symbols) of soot from the ethylene BSS flame as a function of the oxygen concentration (t = 0.22 sec). Left panel: 950 K; right panel: 1000 K. The lines are normal distributions which are fitted to the measured PSDF. The top plots show the PSDFs in inert nitrogen. Lines are log-normal fits to data ……………………………………………… 89 Figure 5.4 Measured specific oxidtation rate of nascent soot (symbols) compared to predictions by the NSC equation (solid lines). The dashed lines represent the NSC rates multiplied by a factor of 10 ……………………………………………………………………………………. 91 xii Abstract The development of novel experimental approaches to investigate fundamental surface kinetics is presented. Specifically, fundamental soot formation and surface catalysis processes are examined in isolation from other competing processes. In terms of soot formation, two experimental techniques are presented: the Burner Stabilized Stagnation (BSS) flame configuration is extended to isolate the effect of the parent fuel structure on soot formation and the fundamental rate of surface oxidation for nascent soot is measured in a novel aerosol flow reactor. In terms of nanoparticles, the physical and chemical properties of freely suspended nanoparticles are investigated in a novel aerosol flow reactor for methane oxidation catalyzed by palladium. The role of parent fuel structure within soot formation is examined by following the time resolved formation nascent soot from the onset of nucleation to later growth stages for premixed BSS flames. Specifically, the evolution of the detailed particle size distribution function (PSDF) is compared for butanol, butane and C 6 hydrocarbons in two separate studies where the C/O ratio and temperature are fixed. Under this constraint, the overall sooting process were comparable as evidenced by similar time resolved bimodal PSDF. However, the nucleation time and the persistence of nucleation with time is strongly dependent upon the structure of the parent fuel. For the C 6 hydrocarbon fuels, the fastest onset of soot nucleation is observed in cyclohexane and benzene flames and this may be due to significant aromatic formation that is predicted in the pre- flame region. In addition, the evolution of the PSDF shows that nucleation ends sooner in cylclohexane and benzene flames and this may be due to relatively quick depletion of soot precursors such as acetylene and benzene. Interestingly,within the butanol fuels studied the effect of the branched chain in i-butanol and i-butane was more significant than the presence of fuel xiii bound oxygen. A numerical analysis of the gas-phase chemistry for butanol and butane indicates the fuel structure effect is largely exhibited in the relative importance of C 2 versus C 3 intermediate species formed during the initial stage of fuel breakdown. Oxidation kinetics of soot are typically measured with carbon black or well aged soot as substrates. The soot surface is also assumed to be graphitic in theoretical soot oxidation rate calculations. However, recent experimental and theoretical studies show that nascent soot can have structures and surface composition drastically different from mature, graphitized soot. In the current study, oxidation of nascent soot by O 2 was observed at T= 950 and 1000K for oxygen concentrations ranging from 1000 to 7800 ppm in a laminar aerosol flow reactor at ambient pressure. Oxidation behavior of primary particles (D p < 20 nm) of nascent soot from a premixed BSS ethylene flame was observed by tracking the shift in the particle size distribution function (PSDF) under a given residence time. The measured rate of the surface reaction ranges from 1x10 6 -3x10 6 g/cm 2 s for nascent soot. The rate of oxidation observed at the given conditions is an order of magnitude faster than predicted by the classical Nagle Strickland-Constable (NSC) correlations derived from graphite oxidation. Heterogeneous surface reaction rates are highly sensitive to the surface composition. Thus the faster rate of surface reaction by the nascent soot observed currently suggests that the surface composition of nascent soot is more reactive than the conventional graphite surface. Catalytic activity in reacting flow laden with suspended nanoparticle catalyst is measured in a novel aerosol flow reactor. Similar to conventional gas phase kinetics, heterogeneous reactions are the product of collisions between the particle surface and surrounding gas. However, particles below 10 nm in diameter are in a transition region where collisions do not always result in perfectly elastic scattering. The inelastic scattering provides more opportunities for reaction to xiv occur than elastic scattering. It is this extra chemical behavior of nanoparticles which may serve as a novel parameter for tuning and optimizing catalytic activity. Specifically, the size dependence of catalytic activity is examined by observing methane oxidation catalyzed by freely suspended palladium nanoparticles. The role of the nanoparticle size is explored in flow reactor measurements and a novel in-situ process for rapid synthesis of aerosolized palladium nanoparticles is introduced. Under catalytic conditions, the temperature at which methane is oxidized is found to be 300K lower than conventional gas phase combustion. In addition, the preliminary measurements indicate that catalytic activity may be greater in nanoparticles relative to bulk surfaces. 1 Chapter 1: Introduction 1.1 Soot Formation Theory and Background Fundamental chemical kinetic processes of reacting flows involving suspended nanoparticles are examined with the focus on soot formation in laminar premixed flames and catalysis of fuel oxidation by freely-suspended nanoparticles. These heterogeneous reacting processes are rather complex and usually involve intricate interactions among various elementary processes, including gas-phase and gas-surface chemical reactions, molecular and particle transport and to an extent, fluid mechanics. The current difficulties in describing and modeling these processes are partly due to the lack of ability to isolate and observe fundamental surface reaction kinetics from the other simultaneously occurring processes. For this reason, a fundamental description that quantifies the kinetic, transport and flow processes that occur during soot nucleation and growth is far from being complete. Over the last century, many theories and mechanisms have been advanced to explain soot formation. In fact, the classic review by Street and Thomas recalls that in 1862 Berthelot proposed a mechanism involving simultaneous polymerization and dehydrogenation [1,2]. In 1981, Haynes and Wagner reviewed the state of knowledge about soot formation and for the first time introduced kinetic aspects of the formation process [3]. The overall scheme begins with particle inception due to both acetylene and aromatic precursors. The particles then grow by surface reactions with gas species where the volume of soot increases but the numbers of particles do not. Conversely, the particles also grow by coagulation where the volume stays constant and the number decreases. The actual particle growth is the net result of coagulation and surface reaction growth [3]. Coagulation reduces the particle surface area whereas surface 2 growth increases it, leading to intimately coupled kinetic competition that impact the level of soot emission. The detailed mechanism for the formation of soot precursors in flames was first described by Frenklach ( e.g., [4,5] ) using a phenomenological description of the kinetic processes of benzene formation and aromatics growth. A step towards a more rigorous treatment of the formation of the first benzene ring can be attributed to development of chemically activated reactions within combustion kinetics by Dean [4,5]. The pathways of chemically activated reactions for aromatic ring formation was subsequently explored by Westmoreland et al. [6]. Thermochemistry, transport and reaction kinetics for aromatic chemistry were then developed by Frenklach et al. [4,7-9]. The methods were applied in the development of a kinetic model for PAH formation that will be discussed next in detail [10,11]. As understanding of the formation of the primary aromatic rings became better, the use of chemically activated reactions became expanded [4,12]. In addition, a class of polycyclic aromatic hydrocarbons (PAHs), known as the Stein stabolimers [13], was found to be particularly stable and they provide the major paths to PAH growth in flames [4,12,13]. In particular, the well-established H-abstraction-C 2 H 4 -addition (HACA) mechanism was developed using these stable species, or stabilomers, as a backbone for the growth of aromatics towards soot. The mechanism is characterized as hydrogen abstraction from a site on the parent stabilomer followed by addition of carbon to the radical site. The mechanism, championed by Frenklach [14-15], captures the essence of the thermodynamic and kinetic requirements for the sooting process. 3 The kinetic processes driving the flame phenomena can be intimately related to soot formation. Within premixed flames, there are dominant kinetic and transport processes that govern the flame structure [4,12]. Regardless what initial fuel one uses, acetylene is favored heavily in the post-flame region of fuel rich flames because of thermodynamic considerations and is the most critical molecular building block of soot. The flame is sustained by chain branching that is driven by the H atom and the high-flame temperature facilitates high activation energy reactions. Yet, the overall sooting growth is suppressed towards high temperatures by thermal fragmentation of the soot precursors and by the increased reversibility of soot mass growth processes. The activation of the reactivity of the parent PAH is necessary because a PAH molecule is usually resistant to direct reaction with acetylene. The activation is achieved through H- abstraction to produce an aryl radical which in turn, reacts with gaseous hydrocarbon species like acetylene. All of these reactions have appreciable energy barriers and exhibit some extent of reversibility. For this reason, temperature plays a critical role within the HACA framework as the rates of molecular growth and fragmentation are strong function of temperature [4,12]. Currently, much of the work has been driven by the need of a predictive capability for soot formation and oxidation in engines. The goal is to combine computational fluid dynamics (CFD) with a fundamental chemistry modeling approach for predictive design of the next-generation efficient, clean-burning combustion engines. Rapid advances in fundamental chemistry models of real fuel combustion [16-18] and incorporation of soot formation models in CFD simulations have been made [19-22]. At the same time, there is a greater need to identify and fill out the gaps in current kinetic descriptions of soot formation. 4 For example, the mechanism of soot nucleation (or particle inception) still remains elusive and probing the chemical structure and composition of nascent soot in flames is in its infancy [4]. The elementary physio-chemical processes of soot nucleation are intricately coupled, thus a further understanding is better achieved by observing each process in some form of isolation. In the current work, the effect of the parent fuel on detailed soot formation processes will be assessed in an experimental configuration where the flow field and temperature are well defined. Detailed sooting properties such as the nucleation time are observed to assess the fuel structure effect on the flame chemistry leading to soot. 1.2 General Role of Parent Fuel Structure on Soot Formation in Premixed Flames Temperature is the dominant parameter governing soot formation [23]. In particular, soot growth processes are reversible and precursors begin to fragment at high temperatures in premixed flames [4]. Acetylene is the most abundant product of the high temperature flame zone in fuel rich premixed flames and this species also dominates the subsequent soot formation chemistry [4]. The structure of the parent fuel is considered to have a secondary effect because of temperature and acetylene production dominate chemistry the post-flame region. The structure of the parent fuel is known to have an indirect effect on the sooting behavior of premixed flames. For example, Takahashi and Glassman carried out systematic studies on the tendency of a particular fuel to form soot within premixed flames by observing the equivalence ratio (a form of fuel/O 2 ratio) at which the onset of soot occurs. The summary reported by Glassman is shown in Fig. 1.1 for both fuel / air flames and fuel / O 2 flames [23]. A low critical equivalence ratio indicates a high sooting tendency. The adiabatic flame temperature is fixed for 5 a given fuel / air mixture thus Glassman used fuel / O 2 flames such that the adiabatic flame temperature is controlled by the inert N 2 flow. Glassman reconciled previously conflicting reports of relative sooting tendencies in premixed and non-premixed flames by systematically accounting for the flame temperature [23,24]. Early reports on the sooting tendency of fuels in premixed flames did not consider the temperature as shown in the circled data points of Fig. 1.1. These early reports considered ethane to have the lowest threshold and acetylene was considered to have the highest threshold for soot formation. This early trend was problematic because it was in contradiction with the relative trend reported in non-premixed flames. In addition, the chemical kinetic mechanism postulated that fuels decompose down to ethylene and eventually to acetylene. As Fig. 1.1 shows, when the flame temperature effect was account for, the relative soot trends are resolved and agree with observations made in non-premixed flames. By accounting for temperature effects, Glassman resolved the sooting tendencies for both premixed and nonpremixed flames into a unified model which indicates that the fundamental soot mechanism is the same in all flames. Soot formation was analyzed in terms of the competition between the fuel pyrolysis (thermal breakdown w/o O 2 ) and soot precursor oxidation by Takahashi and Glassman [23,24]. The fuel structure effect on this competition was expressed in terms of the number of C-C bonds and the C/H ratio of the fuel. Glassman postulated that the specific fuel structure does not play a role on soot formation. Rather, the C-C bonds and C/H ratio of the fuel determine the tendency to form soot. A correlation, shown in Fig. 1.2, was developed by Glassman to quantify the fuel structure dependence on soot formation in premixed flames. As Fig. 1.2 shows, the sooting tendency is the same for different fuels if the number of C-C bonds and C/H ratio are the same. Based on other literature, Glassman also postulated that global properties such as the soot volume fraction 6 depend on this sooting tendency because subsequent growth properties are the same for all fuels [23,24]. Fig. 1.1. The critical equivalence ratio (  c ) as a function of the adiabatic flame temperature reported by Glassman and Takahashi [23,24]. + Fuel / Air Flames (different T flame ) 7 Fig. 1.2. The correlation between the critical equivalence ratio (  c ) at T f = 2200 K and the number of C-C bonds reported by Takahashi and Glassman [23,24]. 8 1.3 Detailed Observations of the Fuel Structure Dependence in Premixed Flames The generalized view of the role of the parent fuel structure has impacted overall kinetic soot models by emphasizing the temperature and acetylene formation in the growth of soot precursors. With some exceptions (e.g Reaction Design: Model Fuels Consortium [25]), the fuel specific gas phase chemistry leading to soot formation has not been emphasized in existing soot formation models. Rather, the larger scale processes are described in detail and gas-phase chemistry is left to more specialized chemical kineticists. Recognizing that the earlier view of the role of the parent fuel structure was established by observing global behavior of soot formation, and most notably by using visual appearance of soot luminosity or the measurement of soot volume fraction, the impact of the parent fuel structure on local, detailed sooting processes has not been examined carefully. Recently, experimental techniques have been advanced that can yield detailed particle size distribution of nascent soot, from nucleation and early stage of mass growth [26-28]. The recent advances were made possible by mobility sizing and probe sampling and allow us to look at the influence of the parent fuel structure (or the lack of it) at a more detailed level. In the current work, detailed sooting behavior including the time evolution of particle size distribution, was examined with the fuel structure as an emphasis. Specifically, the role of fuel bound oxygen was examined by comparing detailed sooting properties of flames of butanol isomers with the non-oxygenated hydrocarblon counterparts. The behavior of oxygenated fuels can reveal insights into mechanisms of soot nucleation and growth for all fuels in general. Fundamental combustion properties such as laminar flame speeds and ignition delay times have been determined and flow reactor measurements have been carried out to describe the kinetic behavior of i-butanol and n-butanol fuels over a wide range of conditions [26-40]. Measurements 9 of detailed flame structures have been made in low-pressure, burner stabilized flames of butanol isomers using photoionization mass spectrometry [41-43]. Much of the experimental work conducted thus far has been used to support the recent extensive development of flame chemistry for the butanol isomers (see, e.g., [26,31-35,37,38,43-45]). However, development of soot chemistry requires more direct observations of sooting behaviors because the nucleation and growth of soot is often impacted by competing kinetic processes [4]. One such competition originates from the fuel structure itself. In recent studies of intermediate species formed in low-pressure burner stabilized fuel-rich flames, Oßwald et al. [42] found a drastic sensitivity for the production of various intermediate species with respect to the fuel structure. In particular, tert-butanol and i-butanol flames yield significantly more aromatic species than in n- or 2-butanol flames. This effect is expected to propagate into soot nucleation and growth in butanol flames. In the current work, the effect of the fuel structure was investigated further by examining detailed sooting behavior of the C 6 hydrocarbons. Each class of hydrocarbon chemical structure is represented in C 6 fuels studied. The flame structure and sooting behavior of the C 6 hydrocarbons has been examined in previous experimental studies. Measurements of detailed flame structures has been made in low-pressure, burner stabilized flames of cyclohexane, benzene and other C 6 fuels by Hansen and co-workers using photoionization mass spectrometry [46-48]. Photoionization mass spectroscopy has been used also by Qi and co-workers to study the flame structure of premixed benzene flames and the pyrolysis of cyclohexane [49,50]. The detailed species profiles obtained by synchrotron photoionization allow for the chemistry leading to larger aromatics to be developed. 10 Development of sooting chemistry from the parent fuel requires more direct observations of sooting behavior because the nucleation and growth of soot is often dictated by competing kinetic processes [4]. One such competition originates from the fuel structure itself. In recent studies of intermediate species formed in low pressure burner stabilized fuel-rich flames, a drastic sensitivity for the production of benzene with respect to the fuel structure was reported. In particular, the C 6 H 12 isomers studied gave three distinct pathways to form benzene [47,48]. Such fuel dependence is expected to propagate again into soot nucleation and growth. A systematic approach must be taken to isolate the fuel structure effect. Detailed and global sooting behavior of the C 6 hydrocarbons has been observed experimentally but the contribution of the C 6 hydrocarbon structure has only recently been examined. For example, the flame structure of sooting premixed cyclohexane flames was examined by Ciajolo and compared to earlier reports of benzene and hexane flames [51]. The comparison indicates that cyclohexane soot nucleates faster due to the dehydrogenation of the carbon ring and this is the first indication of a strong effect of the fuel structure within these sooting hydrocarbon flames [52]. The detailed sooting behavior for premixed benzene flames has been investigated in terms of the evolution of the particle size distribution function (PSDF) by Lighty and co-workers and benzene doping was found to increase the size and volume of soot [53,54]. On the other hand, similar measurements of the PSDF for premixed ethylene flames carried out by Abid showed very little impact of benzene doping on the PSDF [55]. 1.4 Classical Description of Soot Oxidation by O 2 The above discussion highlights the challenge to linking gas phase chemistry to a fundamental description of soot formation. An additional challenge is to describe the composition and 11 evolution of the soot surface properties. The natural environment for soot is the flame environment but the surface reactivity of nascent soot is seldom studied under this condition because of the difficulties and uncertanties associated with non-invasive sampling techniques. Invasive probing techniques may alter the nascent soot surface structure and properties, thus their kinetic behavior. As such the observation may not extrapolate to actual flame conditions. The model for soot oxidation by O 2 is an example of the progression towards a more fundamental description despite experimental and conceptual challenges. Recent studies on soot morphology, particle size distribution function (PSDF) and composition have highlighted differences between nascent soot and mature, graphitized soot [56-59]. Specifically, it has recently been shown that nascent soot can contain significant contributions of aliphatic components [4,58,59]. Because the heterogeneous surface reaction kinetics and mechanism are expected to be highly sensitive to the surface composition, the oxidation kinetics of nascent soot surfaces is not expected to be the same as that of aged or graphitized carbon surfaces. The classical, empirical Nagle-Strickland Constable (NSC) equation [60] was developed largely for graphite or carbon black oxidation and nascent soot oxidation by O 2 should not behave as such. Early studies of soot oxidation by O 2 typically used aged soot or carbon black as the reactant. A kinetic theory for reaction of oxygen with a carbon surface was first proposed by Eyring and coworkers [61] to describe observations of graphite oxidation. Two distinct reaction sites were introduced to explain the maxima that were observed in the oxidation rate above 1400 K. The two reaction sites were also adopted in the development of the Nagle-Strickland Constable (NSC) expression which was derived from measurements of bulk pyrographite oxidation by O 2 12 [60]. The NSC expression has since been extended to describe high temperature carbon oxidation and used in models of soot formation and oxidation (see, e.g., [15]). In the present work, various obstacles to measuring the oxidation kinetics of nascent soot were removed by using fresh, freely suspended soot immediately extracted from flames as the reactant. The fundamental gas-surface kinetic process of oxidation by O 2 is isolated from competing flame processes such as particle-particle coagulation, surface growth and oxidation by OH. By doing so, the surface reactivity of nascent soot can be isolated and the validity of the classical formulation can be verified. 1.5 Combustion Enhancement by Freely Suspended Nanoparticle Catalyst A heterogeneous reacting flow process that is related to the above discussions is catalysis by freely suspendes nanoparticles. Catalysis is extensively used in chemical industries, but its extension to the use of freely suspended, nanoparticle catalysts has not been fully established. One related issue about the use of nanocatalyst is the hypothesized dependency of the catalytic activities on particle size. Evidently, such a hypothesis originates from work far removed from catalysis. In 2004, Li and Wang demonstrated that the Cunninghman slip correction to Stokes drag does not accurately extend to the description of smaller sizes and a more rigorous gas- kinetic analysis was applied where non-rigid collisions and van der Waals interactions were considered [62,63]. Two limiting cases of gas-particle collision were considered and a transition regime where both cases apply was identified. In specular scattering, the angle resulting from collision between the particle and gas is equal to the incident angle. In diffuse scattering, the resulting angle is random. 13 A transition region was identified and molecular dynamics simulations were used by Li and Wang to fundamentally describe the transition from specular to diffuse scattering [65]. Molecular dynamics showed that diffuse scattering is a result of gas molecule trapping on the particle surface. Gas-particle interactions and the particle’s energy accommodation behavior were found to cause gas molecules to be trapped on the surface. In addition to affecting momentum transfer, this nanoparticle theory is a novel interaction between gas and nanoparticle which may affect particle chemistry in a manner never before considered. In the current work, the oxidation of methane will be observed in the presence of freely suspended nanocatalyst to evaluate catalytic activity in this transition region. An observed dependence on the particle size is expected to reveal insights into gas-surface kinetic processes impacted by the fundamental scattering modes just discussed. If the scattering behaviors and processes can be controlled, a novel tuning ability to catalytic activity may be developed. 1.6 Dissertation Organization The approach taken in the current work is to study soot formation and nanoparticle catalysis with a focus on the surface kinetic processes. Surface chemistry and heterogeneous reacting flow processes of combustion and catalysis are developed via novel heterogeneous flow reactor designs which isolate surface chemistry from other competing processes. The experimental designs and methods are introduced in Chapter 2. The effect of parent fuel structure on the particle inception and early development of nascent soot structure is examined in isolation from temperature and time effects. Fuel bound oxygen and other fuel structures are examined in Chapter 3 in the systematic study of fuel structure effects on soot formation within butanol flames. A similar study is carried out in Chapter 4 in the investigation of time resolved nucleation processes in C 6 hydrocarbon flames. In Chapter 5, the oxidation kinetics of nascent 14 soot particles by O 2 at moderate temperature are measured in isolation from particle interaction effects and oxidation by OH. Beyond sooting flames, a novel in-situ nanoparticle synthesis method is combined with an aerosol flow reactor to measure the oxidation kinetics of methane in the presence of freely suspended Pd nanoparticle catalyst in Chapter 6. A conclusion and discussion of future work is in Chapter 7. Chapter 1 References [1] W. A. Bone, Flame and Combustion in Gases. Longmans, Green and Co., London, 1927. [2] J. C. Street, A. Thomas, Fuel 34 (1955) 4-36. [3] B. S. Haynes, H. G. Wagner, Prog. Energy Combust. Sci. 7 (1981) 229-273. [4] M. Frenklach, S. Taki, M. B. Durgrapasad, R. A. Matula, Combust. Flame 54 (1983) 81- 101. [5] M. Frenklach, J. Warnatz, Combust. Sci. Tech. 51 (1987) 265-283. [4] H. Wang, Proc. Combust. Inst. 33 (2011) 41-67. [5] A. M. Dean, J. Phys. Chem. 89 (1985) 4600-4608. [6] P. R. Westmoreland, A. M. Dean, J. B. Howard, J. P. Longwell, J. Phys. Chem. 93 (1989) 8171-8180. [7] H. Wang, M. Frenklach, J. Phys. Chem. 97 (1993) 3867-3874. [8] H. Wang, M. Frenklach, Combust. Flame 96 (1994) 163-170. [9] H. Wang, M. Frenklach, J. Phys. Chem. 98 (1994) 11465-11489. [10] H. Wang, M. Frenklach, Combust. Flame 110 (1997) 173-221. [11] J. Appel, H. Bockhorn, M. Frenklach, Combust. Flame 121 (2000) 122-136. [12] M. Frenklach, PCCP 4 (2002) 2028-2037. [13] S. E. Stein, A. Fahr, J. Phys. Chem. 89 (1985) 3714-3725. [14] M. Frenklach, J. Warnatz, Combust. Sci. Tech. 51 (1987) 265-283. [15] M. Frenklach, H. Wang, Proc. Combust. Inst. 23 (1991) 1559-1566. [16] M. Colket, J. T. Edwards, S. Williams, N. P. Cernansky, D. L. Miller, F. N. Egolfopoulos, P. Lindstedt, K. Seshadri, F. L. Dryer, C. K. Law, D. G. Friend, D. B. 15 Lenhert, H. Pitsch, A. F. Sarofim, M. D. Smooke, W. Tsang, 45th AIAA Aerospace Sciences Meeting, Jan 8-11, Reno, NV, 2007; pp AIAA Paper 2007-0770. [17] B. Sirjean, E. Dames, D. A. Sheen, X. You, C. J. Sung, A. T. Holley, F. N. Egolfopoulos, H. Wang, S. S. Vasu, D. F. Davidson, R. K. Hanson, H. Pitsch, C. T. Bowman, A. Kelley, C. K. Law, W. Tsang, N. P. Cernansky, D. Miller, A. Violi, R. P. Lindstedt, A high-temperature chemical kinetic model of n-alkane (up to n-dodecane), cyclohexane, and methyl-, ethyl-, n-propyl and n-butyl-cyclohexane oxidation at high temperatures, JetSurF version 2.0, September 19, 2010 (http://melchior.usc.edu/JetSurF/JetSurF2.0). [18] A. Violi, A. Kubota, T. N. Truong, W. J. Pitz, C. K. Westbrook, A. F. Sarofim, Proc. Combust. Inst. 29 (2002) 2343-2349. [19] D. O. Lignell, J. H. Chen, P. J. Smith, Combust. Flame 155 (2008) 316-333. [20] H. Pitsch, E. Riesmeier, N. Peters, Combust. Sci. Tech 158 (2000) 389-406. [21] L. Wang, D. C. Haworth, S. R. Turns, M. F. Modest, Combust. Flame 141 (2005)170- 179. [22] D. O. Lignell, J. H. Chen, P. J. Smith, T. Lu, C. K. Law, Combust. Flame 151 (2007) 2- 28. [23] I. Glassman, Combustion , Academic Press, San Diego, 1996. [23] F. Takahashi, I. Glassman, Combust. Sci. Tech. 37 (1984) 4-19. [24] I. Glassman, 22 nd Symp. (Int) Combustion (1988) 295-311. [25] Model Fuels Consortium: Science-based Soot Model Development, proprietary, Reaction Design, Inc. 2012. [26] B. Zhao, Z.W. Yang, M.V. Johnston, H. Wang, A.S. Wexler, M. Balthasar, M. Kraft, Combust. Flame 133 (2003) 173–188. [27] B. Zhao, Z.W. Yang, J.J. Wang, M.V. Johnston, H. Wang, Aerosol Sci. Technol. 37 (2003) 611–620. [28] B. Zhao, Z.W. Yang, Z.G. Li, M.V. Johnston, H. Wang, Proc. Combust. Inst. 30 (2005) 1441–1448. [26] J. T. Moss, A. M. Berkowitz, M. A. Oehlschlaeger, J. Biet, V. Warth, P.-A. Glaude, F. Battin-Leclerc, J. Phys. Chem. A 112 (2008) 10843-10855. [27] S. Vranckx, K. A. Heufer, C. Lee, H. Olivier, L. Schill, W. A. Kopp, K. Leonhard, C. A. Taatjes, R. X. Fernandes, Combust. Flame 158 (2011) 1444-1455. [28] P. S. Veloo, Y. L. Wang, F. N. Egolfopoulos, C. K. Westbrook, Combust. Flame 157 (2010) 1989-2004. [29] P. S. Veloo, F. N. Egolfopoulos, Proc. Combust. Inst. 33 (2011) 987-993. 16 [30] X. L. Gu, Z. H. Huang, S. Wu, Q. Q. Li, Combust. Flame 157 (2010) 2318-2325. [31] K. M. Van Geem, S. P. Pyl, G. B. Marin, M. R. Harper, W. H. Green, Ind. Eng. Chem. Res. 49 (2010) 10399-10420. [32] G. Black, H. J. Curran, S. Pichon, J. M. Simmie, V. Zhukov, Combust. Flame 157 (2010) 363-373. [33] P. Dagaut, S. M. Sarathy, M. J. Thomson, Proc. Combust. Inst. 32 (2009) 229-237. [34] C. Togbe, A. Mze, Amir, P. Dagaut, Energy Fuels 24 (2010) 5244-5256. [35] M. R. Harper, K. M. Van Geem, S. P. Pyl, G. B. Marin, W. H. Green, Combust. Flame 158 (2011) 16-41. [36] K. E. Noorani, B. Akih-Kumgeh, J. M. Bergthorson, Energy Fuels 24 (2011) 5834-5843. [37] R. Grana, A. Frassoldati, T. Faravelli, U. Niemann, E. Ranzi, R. Seiser, R. Cattolica, K. Seshadri, Combust. Flame 157 (2011) 2137-2154. [38] I. Stranic, D. P. Chase, J. T. Harmon, S. Yang, D. F. Davidson, R. K. Hanson, Combust. Flame (2011). [39] B. W. Weber, K. Kumar, Y. Zhang, C.-J. Sung, Combust. Flame 158 (2011) 809-819. [40] D. M. A. Karwat, S. W. Wagnon, P. D. Teini, M. S. Wooldridge, J. Phys. Chem. A 115 (2011) 4909-4921. [41] B. Yang, P. Oßwald, Y. Li, J. Wang, L. Wei, Z. Tian, F. Qi, K. Kohse-Höinghaus, Combust. Flame 148 (2007) 198-209. [42] P. Oßwald, H. Güldenberg, K. Kohse-Höinghaus, B. Yang, T. Yuan, F. Qi, Combust. Flame 158 (2011) 2-15. [43] N. Hansen, M. R. Harper, W. H. Green, Phys. Chem. Chem. Phys. 13 (2011) 20262- 20274. [44] C.-W. Zhou, J. M. Simmie, H. J. Curran, Combust. Flame 158 (2011) 726-731. [45] K. Kohse-Höinghaus, P. Oßwald, T. A. Cool, T. Kasper, N. Hansen, F. Qi, C. K. Westbrook, P. R. Westmoreland, Angew. Chem. Int. Edit. 49 (2010) 3572-3597. [46] M. E. Law, P. R. Westmoreland, T. A. Cool, J. Wang, N. Hansen, C. A. Taatjes, T. Kasper, Proc. Combust. Inst. 31 (2007) 565-573. [47] N. Hansen, T. Kasper, B. Yang, T. A. Cool, W. Li, P. R. Westmoreland, P. Oßwald, K. Kohse-Höinghaus, Proc. Combust. Inst. 33 (2011) 585-592. [48] N. Hansen, W. Li, M. E. Law, T. Kasper, P. R. Westmoreland, B. Yang, T. A. Cool, A. Lucassen, PCCP 12 (2011) 12112-12122. 17 [49] B. Yang, Y. Li, L. Wei, C. Huang, J. Wang, Z. Tian, R. Yang, L. Sheng, Y. Zhang, F. Qi, Proc. Combust. Inst. 31 (2007) 555-563. [50] Z. Wang, Z. Cheng, W. Yuan, J. Cai, L. Zhang, F. Zhang, F. Qi, J. Wang, Combust. Flame 159 (2012) 2243-2253. [51] M. Alfe, B. Apicella, R. Barbella, A. Tregrossi, A. Ciajolo, Proc. Combust. Inst. 31 (2007) 585-591. [52] A. Ciajolo, A. Tregrossi, M. Mallardo, T. Faravelli, E. Ranzi, Proc. Combust. Inst. 32 (2009) 585-591. [53] C. A. Echavarria, A. F. Sarofim, J. S. Lighty, A. D'Anna, Proc. Combust. Inst. 32 (2009) 705-711. [54] C. A. Echavarria, A. F. Sarofim, J. S. Lighty, A. D'Anna, Combust. Flame 158 (2010) 98- 104. [55] A. D. Abid, E. D. Tolmachoff, D. J. Phares, H. Wang, Y. Liu, A. Laskin, Proc. Combust. Inst. 32 (2009) 681-688. [2] B. Zhao, K. Uchikawa, H. Wang, Proc. Comb. Inst. 31 (2007) 851-860. [56] A. D Abid, N. Heinz, E. Tolmachoff, D.J. Phares, C. Campbell, H. Wang, Combust. Flame 154 (2008) 775-788. [57] J.P. Cain, P.L Gassman, H. Wang, A. Laskin, PCCP 12 (2010) 5206-5218. [58] J.P. Cain, J. Camacho, D.J. Phares, H. Wang, A. Laskin, Proc. Comb. Inst. 33 (2011) 533-540. [59] J.R. Walls, R.F Strickland-Constable, Carbon 1 (1964) 333-338. [60] G. Blyholder, J.S. Binford, H. Eyring, J. Phys. Chem. 62 (1958) 263-267. [61] Z.G. Li, H. Wang, Phys. Rev. E 68 (2003) 061206. [62] Z.G. Li, H. Wang, Phys. Rev. E 68 (2003) 061206. [63] Z.G. Li, H. Wang, Phys. Rev. E 70 (2004) 021205. [64] Z.G. Li, H. Wang, Phys. Rev. Letters 95 (2005) 014502. 18 Chapter 2: Experimental Methods 2.1 Introduction The work presented here is based on experimental investigation into fundamental surface chemistry and heterogeneous reacting flow processes of combustion and catalysis. Several experimental techniques were designed and updated during this work. In addition, commercially available instruments were utilized for characterization of reacting flow processes in the nanoparticle and gas phase. In terms of computational tools, novel methods were developed to connect experimental observations to fundamental chemical kinetic processes such as gas-phase combustion chemistry. The heterogeneous reacting flow processes discussed in the current work are soot formation within laminar premixed flames and catalysis on the surface of freely suspended nanoparticles. Both the flame and catalysis experiments are examples of reacting flow processes involving surface reactions on freely suspended nanoparticles. The emphasis of this work is to experimentally measure the rate of surface kinetic processes which occur during soot formation and catalysis in isolation from other competing processes. For example, the formation of soot within flames is known to be an interwoven process involving local temperature, local fuel/O 2 ratio and time. These competing factors must be accounted for in the investigation of fundamental soot formation processes and this work details such experimental investigations. 2.2 Burner-Stabilized Stagnation (BSS) flame configuration for nascent soot studies Elementary processes of chemical kinetics, molecular transport and fluid mechanics all interact within flames. Experimental configurations have been designed to observe and model each of these factors either collectively or individually. The simplest configuration for a premixed (fuel and oxidizer flow together) flame is a 1-D free stream flow with a stationary flame location. The 19 fuel is oxidized as the flow propagates towards the flame zone and the combustion products flow away from the flame. Such a configuration allows for the fluid mechanics to be greatly simplified allowing for the transport and chemistry processes to be studied. In the laboratory, these anchored flames are known as burner stabilized flames and they are achieved by stabilizing the premixed flame over a porous mesh. The opposed-jet / stagnation flow flame is a more robust flame configuration that allows for both the dynamics and structure (chemical species position) to be examined on a fundamental level. Much of the work in the combustion field is developed by cross-comparisons between theoretical models and experimental observations within these widely used flame configurations. In terms of soot, global properties such as the soot volume fraction were measured early on with the use of laser diagnostics. However, lasers and other optical methods such as small angle neutron scattering are not capable of detecting particles below 10 nm in diameter [1]. Many detailed processes which influence the total number density and mean particle size of sooting flames occur within the size range below 10 nm. The introduction of probe sampling lowered the observable size window to around 1 nm which lead to further understanding of processes leading to detailed properties such as the particle size distribution function (PSDF). Knowledge of the soot particle dynamics in the smaller size window has been improved by experimental probe sampling. Despite the wide-spread use of probe sampling in both sooting and non-sooting flames, care must be taken to avoid possible drawbacks to the method. First, the probe must be designed properly such that particle loss from both diffusion to the probe walls and coagulation among particles. A more fundamental drawback, however, is that the probe is inherently intrusive to the flame. Flame probes cause a significant local temperature drop and local flow stagnation. This 20 experimental problem has been traditionally remedied by applying a spatial shift to the distance that the flame was actually sampled. The shift is applied to more closely match the comparison with a model prediction which does not consider probe perturbation. A novel approach to probe sampling was introduced by Abid and co workers to resolve problems associated with flame perturbation [2]. The burner stabilized stagnation (BSS) flame configuration combines a sampling probe with a stagnation plate such that the combination simultaneously acts as a stagnation surface and a soot sample probe. A traditional burner stabilized flame is combined with flow stagnation as shown in Fig. 2.1. The stagnation surface/ probe temperature provides a rigorous definition of the boundary condition with probe intrusion and allows for less ambiguous comparison between the experimental observations and model. Even with probe perturbation, flame measurements sampled by tubular probes have been traditionally compared to the burner stabilized flame model (free stream-no downstream boundary is defined). In contrast, the BSS flame approach allows for the flame boundary conditions to be fully specified and the flame with perturbation can be modeled as a pseudo 1-D divergent reacting flow. Fig. 2.1. Schematic summary of the burner stabilized-stagnation (BSS) flame configuration. 21 The comparison between BSS flames and model is much less ambiguous than the comparison between burner stabilized flames and model because the boundary condition is more rigorous in BSS flames. In burner stabilized flames, the absence of a defined downstream boundary causes the model to be under-defined unless a temperature profile is provided or assumed. In BSS flames, the downstream boundary condition is defined as the measured probe temperarature. The BSS flames are premixed stagnation flames thus they can be modeled using the standard opposed-jet/stagnation reacting flow model known as OPPDIF. The stagnation flow field was solved early on by von Karman with the use of a similarity transformation that reduces the 2-D axis-symmetric flow field to a 1-D problem [3,4]. This can be done because the radial component of the velocity is linear in terms of r and this allows for the dependent variables to be expressed in terms of only the axial direction. The OPPDIF code presented here is based on the original formulation of R.J. Kee and co-workers [3,4] with modifications by Sheen and co-workers [2]. The detailed flame structure is computed using the burner temperature, probe temperature and unburned gas mass flux as input boundary conditions as will be discussed later in more detail. In the current work, the simulation of BSS flames is updated by evaluating the thermophoretic component of the soot particle velocity during probe sampling. The flame perturbation due to the sampling probe is included in the modified OPPDIF code by introducing a stagnation surface at x = H p in the form of a zero velocity boundary condition [2]. The flow stagnation causes the fluid velocity and hence the convective time for soot nucleation and growth to increase dramatically. For soot particles, the finite residence time is determined by considering the thermophoretic velocity which is driven by the significant temperature gradient, ∂ T /∂x, at the stagnation plate. In 22 the current work, the thermophoretic velocity is assumed to lie within the free molecular formulation introduced by Waldmann [5]: where λ is the thermal conductivity of the gas-phase calculated from transport properties of the flame gas in the multi-component formulation. The momentum accommodation factor, , in this limit is given by a value of 0.9 based on Millikan’s oil droplet experiments [5]. The number density of the gas phase molecules, N, the Boltzmann constant, k, and the local flame temperature, T also contribute. The thermophoretic velocity was calculated numerically and added to the total velocity. The residence time of the soot particles is defined as the time interval for the particle (or precursors) to traverse from the calculated location of the peak flame temperature to the location of the stagnation probe. 2.3. Oxidation rate of nascent soot in a novel aerosol flow reactor Measurements are ideally made with soot in the original aerosolized, flame state. In the present work, a novel aerosol reactor is introduced to examine the fundamental rate of nascent soot oxidation by molecular oxygen around 1000 K. As will be discussed in detail later, the measured oxidation rate of nascent soot is an improvement over the existing rates. The novel aerosol flow reactor is designed to isolate the surface oxidation of nascent soot by O 2 in isolation from other competing flame processes. The various obstacles to measuring the oxidation kinetics of nascent soot were minimized by using fresh, freely suspended soot immediately extracted from flames as the reactant. Specifically, a particle sample was continuously extracted from premixed burner 23 stabilized stagnation (BSS) flame and diluted immediately to reduce particle-particle coagulation. The particle sample was then introduced into the flow reactor. As discussed previously, BSS flames are in a pseudo 1-D configuration which allows for nascent soot to be probed under well defined boundary conditions [2]. The stagnation probe allows for the residence time of soot particles to be controlled in the flame, allowing for soot at the different growth stages to be examined. The electrical mobility was measured in the present study to determine the particle size distributions before and after oxidation occurs. Emphasis was placed on minimizing the transit time from the point where the soot was sampled to the flow reactor inlet. Changes in the original surface composition and morphology are minimized by using a short transit time and suspending the particles in an inert flow of cold nitrogen. As will be discussed in detail later, the aerosol flow reactor is designed to isolate the surface reactivity of nascent soot in isolation from other competing processes. A schematic summarizing the oxidation experiment is shown in Fig. 2.2. Nascent soot was directly extracted at a distance of 0.8 cm from the burner surface of a 15.1 % ethylene- 21.8 % oxygen-argon flame (φ = 2.07, Tf = 1850 K) using a probe sampling technique similar to previous characterization of PSDFs in BSS flames [2,6,7]. The flat flame burner is water cooled and the burner diameter was 5 cm. The mass flow rates of oxygen, argon and nitrogen were controlled by critical orifices. 24 Fig. 2.2. Schematic of the sequential burner - aerosol flow reactor setup for nascent soot oxidation. Nascent soot enters into the sampling probe orifice and a carrier flow of nitrogen with dilution ratio of approximately 100 dilutes and quenches the sample. The aerosol is carried to the flow reactor within a transmission time of 0.1 sec and mixed with a high-temperature stream of oxygen/nitrogen or nitrogen mixture. The high temperature stream is pre-heated to maintain an isothermal temperature profile in the flow reactor. The combined mixture enters into a quartz tubular flow reactor which has 1.8 cm inner diameter. The residence time is 0.2 seconds in the flow reactor. In the present study, the diagnostic for nascent soot oxidation is the change in the PSDF, which is measured with a TSI Scanning Mobility Particle Sizer (Electrostatic Classifier 3085and Condensation Particle Counter 3095). Sampling probes embedded at the inlet and the outlet of the aerosol reactor allows for the PSDF to be measured before and after a given oxidation time. 1. BSS Flame and Soot Sampling Exhaust Sampling N 2 Sampling orifice O 2 /N 2 2. Pre-Heated N 2 /O 2 Sampling N 2 Exhaust t = 0 sec TC TC TC t = t 1 3. Flow reactor (1 atm) ID = 1.8 cm, L= 90 cm t ~ 0.2 sec ΔP Kr 85 Nano- DMA ultra-fine condensation optics Premixed fuel/O 2 /Ar Scanning Mobility Particle Sizer (SMPS) 25 2.4 in-situ Palladium nanoparticle synthesis for fundamental surface catalysis observations Catalytic activity in reacting flow laden with suspended nanoparticle catalyst is measured in a novel aerosol flow reactor. Similar to conventional gas phase kinetics, heterogeneous reactions are the product of collisions between the particle surface and surrounding gas. However, particles below 10 nm in diameter are in a transition region where collisions do not always result in perfectly elastic scattering. The inelastic scattering provides more opportunities for reaction to occur than elastic scattering. It is this extra chemical behavior of nanoparticles which may serve as a novel parameter for tuning and optimizing catalytic activity. A novel experimental design is introduced in the current work to observe the catalytic oxidation of methane on the surface of freely suspended Pd nanoparticles. The surface catalysis process of methane oxidation is observed in isolation from other competing processes. The oxidation of methane on the surface of freely suspended palladium catalyst has been studied [8,9]. Shimizu observed the methane oxidation over the surface of in-situ generated Pd nanoparticles but the measurements may be effected by competing processes which occur in the same flow reactor. A development of this work is the systematic synthesis of Pd nanoparticles separate from methane oxidation. A sequential nanoparticle synthesis-flow reactor configuration, shown in Fig. 2.3, is introduced to study the surface kinetics of methane oxidation by O 2 . The focus of the current section is to introduce a novel method for the in-situ generation of Pd nanoparticles but the overall schematic gives the motivation for the synthesis design. As will be discussed in detail later, the new design is preferred because the particle synthesis occurs before the oxidation of methane is observed in 26 the flow reactor. The synthesis design is also optimized such that Pd nanoparticle is prepared in a stable manner to provide a steady flow of Pd catalysis for methane oxidation. Fig. 2.3. Schematic of the sequential Pd nanoparticle synthesis - aerosol flow reactor setup for methane oxidation catalyzed by freely suspended Pd The in-situ generation of Pd nanoparticles ensures that the catalyst surface is relatively free from contamination and poisoning. In the current work, nanoparticle generation is improved by de- coupling synthesis from methane oxidation. The in-situ process involves a Pd precursor solution which evaporates and decomposes in the high temperature jet stirred reactor shown in Fig. 2.3. Palladium vapors form and subsequent coagulation causes nanoparticles to form over time. The Pd nanoparticle precursor is a solution consisting of a Pd organometallic salt dissolved in an organic solvent. Palladium Acetate ( Pd(Ac) 2 ) is dissolved in either acetone or acetaldehyde in the current work. 1a. Palladium Acetate- Acetaldehyde Solution carrier N 2 sampling orifice pre-heated air 1b. Jet Stirred Reactor sampling N 2 exhaust t = 0 sec T in t = t 1 2. Flow reactor ID = 1.7 cm, L= 80 cm t ~ 0.1 sec CH 4 Scanning Mobility Particle Sizer exhaust Diagnostics Infrared CO 2 / CH 4 Detector cooling N 2 / O 2 ultrasonic transducer sub- μm precursor fog 1c. Cooling Vessel 27 Control of the surface area density and nanoparticle size is a novel aspect of the synthesis design. The Pd(Ac) 2 loading in the precursor solution and the flow rate of the carrier N 2 govern particle synthesis in the well-stirred reactor. The time scale of vaporization is relatively long and the stability of the vaporization step is improved by decreasing the precursor droplet size. The ultrasonic transducer design introduced here steadily generates sub micron precursor droplets thereby improving the distribution and vaporization of a Pd precursor solution. The performance of the synthesis method is summarized in Fig. 2.4 in terms of the PSDF of Pd nanopartlces. The smallest precursor loading in acetone is 0.1 wt % Pd(Ac) 2 and the resulting median diameter varies from 2-3 nm as the carrier N 2 flow is increased. At the solubility limit of the precursor, 3wt % Pd(Ac)2, the median particle diameter reached 15 nm. As Fig. 2.4 shows figures, the PSDF are reproducible and sensitive to the input parameters. Stability is improved because the spatial distribution of precursor is more uniform and the vaporization time is minimized. This novel in-situ Pd nanoparticle synthesis design allows for the catalytic oxidation of methane to be observed in isolation from other competing processes. . Fig. 2.4. PSDF of the in-situ generated Pd nanoparticles. The number density and median particle diameter, thus surface area density, is controlled by the precursor loading and carrier gas. 28 2.5 Scanning Mobility Particle Sizing (SMPS) for nascent soot and Pd nanoparticle aerosol The understanding of detailed processes surrounding the inception of condensed phases in flames has been aided by the development of probe sampling for scanning mobility particle sizing (SMPS). In sooting flames, the evolution of the particle size distribution function (PSDF) has been followed in studies of temperature and fuel structure effects on the soot formation [10-26,7] BSS]. The SMPS technique has also been used to analyze metal oxide nanoparticle formation in flames and to analyze the extent of surface reaction on freely suspended Pd catalyst [8,9,18, 27- 29]. In the present work, the SMPS is used to analyze fundamental processes of soot formation and nanoparticle catalysis. A major capability of the SMPS technique is to measure the PSDF of particles above 3 nm in diameter. The SMPS involves a classifying step where the sample containing many particles sizes (polydisperse aerosol) is processed into single flow with a narrow size distribution (monodisperse aerosol). A subsequent counting step builds the distribution from each bin count. The SMPS system in the current work is a TSI model consisting of an Electrostatic Classifier model 3085 and Condensation Particle Counter model 3095. The Electrostatic Classifier operates with a nano-Differential Mobility Analyzer (DMA) model in dual flow mode as will be discussed in more detail. The current mobility sizer centers on the DMA device which analyzes the mobility of the particles.The DMA analyzes the particle size of an aerosol by interpreting the mobility of the particle within an electric field. A landmark device for the modern DMA is the Hewitt differential mobility analyzer because it introduced the capability to produce a monodisperse aerosol flow [30,31]. The operating principle of differential mobility analysis is that the trajectory of a particle can be controlled within a known geometry by applying a suitable electric 29 field. Moreover, a monodisperse aerosol flow is created by controlling the geometry in the DMA such that trajectory leads to a particle collection flow. The electrical mobility and other transport properties of nanoparticles in dilute gases are the product of collisions which occur between the particle and gas phase. A recent gas-kinetic theory analysis on the transport properties of nanoparticles has shown that perfect elastic collisions (specular scattering) occur in the molecular and sub-nm range [32]. Above this range, the scattering is diffuse which means that the angle resulting from collision is random and is no longer related to the incoming angle [32]. The mobility of a particle is a balance between the drag force over the surface and electric force. The electrical mobility based on this updated transport theory is summarized below [33]. where the subscript s/d denotes the specular or diffuse scattering limits, n is the number of charge in the particle, e is the elementary charge, μ is the reduced mass, k is the Boltzmann constant, T is the temperature, N is the number density of the gas, r p is the particle radius, and (1,1) s and (1,1) d are the reduced collision integrals of specular and diffuse scattering, respectively. The collisions integrals are fundamental parameters which are directly related to the corresponding Lennard-Jones potential parameters. The preceding analysis considers a rigorous gas-kinetic theory which is an update over the traditional Stokes-Cunningham formulation. In the traditional mobility analysis, the particle drag is assumed to follow a modified form of Stoke’s law. Mobility analysis occurs in the free molecular regime and the Cunningham silp correction accounts for continuum regime derivation 30 of Stoke’s law. However, this correction is limited for particles below 10 nm in diameter because the van der Waals gas-particle interactions and the transition between specular and diffuse scattering are unaccounted for [32,34,35]. The SMPS system in the current work does not consider this updated nanoparticle transport theory thus the following parameterized correlation between the mobility diameter, D p,SMPS and the true diameter, D p , is used for a carbonaceous particle [36]: where D p,SMPS is the measured mobility diameter in nm. In the correlation, particles larger than 10 nm in diameter are not corrected as much as smaller range of size. Particles in the small size range are primary particles which are known to be spherical. Larger particles may not be spherical and the mobility analysis of possible non-spherical particles will be discussed in detail when appropriate. A schematic which summarizes the current nano-DMA device is shown in Fig. 2.3. A polydisperse aerosol flow of known charge distribution flows from the outer shell of the DMA into the main section. The main section consists of two concentric metal cylinders and the sample flows down the annulus between the cylinders. The inner cylinder is maintained at a controlled negative voltage and the outer cylinder is electrically grounded. This creates an electric field between the two cylinders which causes positively charged particles to be attracted to the negatively charged cylinder. The trajectory of the particles depends on the particle electrical mobility, the flow rates in the DMA and the DMA geometry. Particles with a high electrical mobility impact along the upper portion of the inner cylinder and particles with a low electrical mobility impact the lower portion. A monodisperse aerosol flow forms at the bottom 31 slit because particles within a narrow range of electrical mobility all exit with the same trajectory. Fig. 2.5. Schematic of the nano-DMA operating in dual flow mode [37]. A significant advance in the overall mobility sizing scheme was introduced by Wang and Flagan in the form of the Scanning Electrical Mobility Spectrometer [38]. Wang and Flagan accelerated the classification process by replacing the traditional one size at time electrical mobility analysis with a more rapid voltage scan for the full range of particle sizes. This advancement has been incorporated into the SMPS described in the current work. Mobility sizing centers on the DMA device which analyzes the mobility of the particles. However, there are steps before the DMA converts the polydisperse aerosol into a monodisperse flow. Inertial impaction is carried out at the inlet of the classifier to remove larger particles. Specifically, the SMPS impactor nozzle is 0.071 cm in diameter which corresponds to a particle 32 diameter cutoff of approximately 500 nm. Electrical mobility classification is easier below this cutoff range because larger particles have a less clear charge distribution. A source of 85 Kr is introduced at the inlet of the DMA to achieve a known charge distribution of the polydisperse aerosol. This krypton isotope is a source of bipolar ions which collide with the sample flow. An equilibrium bipolar charge distribution is achieved by the ion source before mobility analysis. The charge distribution is calculated based on charging theory of bipolar ion atmospheres presented by Fuchs [38]. 2.6 Non-dispersive Infrared Spectroscopy for CH 4 and CO 2 In the current work, fundamental surface kinetic processes are observed by tracking the evolution of gas phase species such as CH 4 and CO 2 . A non-dispersive infrared (NDIR) spectroscopy configuration is used in the current experiments. A non-dispersive configuration is employed which means that the infrared source is designed to analyze the response of the sample to a single infrared wavelength. Molecules absorb infrared radiation in bands specific to the molecule. A comparison between the incident and resulting radiation at a known wavelength allows quantification of the sample concentration. A Beer-Lambert formulation is used to determine the desired gas concentration as a function of the measured absorption. A Caifornia Analytical Instruments model 200 CO 2 /CH 4 analyzer is used in the current work. A schematic of the device operation is shown in Fig 2.5. The infrared light source emits infrared light in all directions. The infrared light emitted backward is reflected by a reflecting surface and is added to the infrared light emitted forward. A chopper blade rotates in between the infrared source and the measuring cell to modulate the light beam at a regular frequency. The modulated infrared light beam then passes through the measuring cell which is filled with a sample gas 33 containing the species to be analyzed. The infrared energy is partially absorbed or attenuated before it reaches the front chamber of the detector. Fig. 2.6. Schematic of a NDIR Gas analyzer configuration. The detector consists of two sequential chambers which contain the specific gas component to be measured. Detection occurs by comparing the difference between gases in the chamber as the beam traverses from one chamber to the other. The infrared light energy is partially absorbed in the front chamber and residual light is absorbed in the rear chamber. The pressure in both chambers increases as the energy introduced. However, a pressure gradient forms between the two chambers because more energy is absorbed in the first chamber. A microflow is allowed to form between the two chambers and the amount of flow is proportional to the infrared energy absorbed by the gas sample. sample cell ΔP detection 34 Chapter 2 References [1] H. Wang, B. Zhao, B. Wyslouzil, K. Streletzky, Proc. Combust. Inst. 29 (2002) 2749– 2757. [2] A.D. Abid, J. Camacho, D.A. Sheen, H. Wang, Combust. Flame 156 (2009) 1862-1870. [3] R.J. Kee, J.A. Miller, G.H. Evans, G. Dixon-Lewis, Twenty Second Symposium (International) on Combustion (1988) 1479-1494. [4] A.E. Lutz, R.J. Kee, J. Grcar, F.M. Rupley, OPPDIF: A FORTRAN Program for Computing Opposed-Flow Diffusion Flames, Sandia National Laboratories, Albuquerque, New Mexico, 1996. [5] Z.G. Li, H. Wang, Phys. Rev. E 70 (2004) 021205. [6] A.D Abid, J. Camacho, D.A. Sheen, H. Wang, Energy Fuels 23 (2009) 4286-4294. [7] J. Camacho, S. Lieb, H. Wang, Proc. Combust. Inst. 34 (2013) 1853-1860. [8] T. Shimizu, A.D. Abid, G. Poskrebyshev, H. Wang, J. Nabity, J. Engel, J. Yu, D. Wickham, B. Van Devener, S.L. Anderson, S. Williams, Combust. Flame 157 (2010) 421-435. [9] T. Shimizu, H. Wang, Proc. Combust. Inst. 33 (2011) 1859-1866. [10] M.M. Maricq, S.J. Harris, J.J. Szente, Combust. Flame 132 (2003) 328–342. [11] B. Zhao, Z.W. Yang, M.V. Johnston, H. Wang, A.S. Wexler, M. Balthasar, M. Kraft, Combust. Flame 133 (2003) 173–188. [12] B. Zhao, Z.W. Yang, J.J. Wang, M.V. Johnston, H. Wang, Aerosol Sci. Technol. 37 (2003) 611–620. [13] M.M. Maricq, Combust. Flame 137 (2004) 340–350. [14] B. Zhao, Z.W. Yang, Z.G. Li, M.V. Johnston, H. Wang, Proc. Combust. Inst. 30 (2005) 1441–1448. [15] B. Oktem, M.P. Tolocka, B. Zhao, H. Wang, M.V. Johnston, Combust. Flame 142 (2005) 364–373. [16] M.M. Maricq, J. Aerosol Sci. 37 (2006) 858–874. [17] M.M. Maricq, Combust. Flame 144 (2006) 730–743. [18] B. Zhao, K. Uchikawa, H. Wang, Proc. Combust. Inst. 31 (2007) 851–860. [19] S.L. Manzello, D.B. Lenhert, A. Yozgatligil, M.T. Donovan, G.W. Mulholland, M.R. Zachariah, W. Tsang, Proc. Combust. Inst. 31 (2007) 675–683. 35 [20] L.A. Sgro, A. De Filippo, G. Lanzuolo, A. D’Aless io, Proc. Combust. Inst. 31 (2007) 631–638. [21] A. Ergut, Y.A. Levendis, H. Richter, J.B. Howard, J. Carlson, Combust. Flame 151 (2007) 173–195. [22] A.D. Abid, N. Heinz, E. Tolmachoff, D. Phares, C. Campbell, H. Wang, Combust. Flame 154 (2008) 775–788. [23] L.A. Sgro, A. Borghese, L. Speranza, A.C. Barone, P. Minutolo, A. Bruno, A. D’Anna, A. D’Alessio, Environ. Sci. Technol. 42 (2008) 859 –863. [24] L.A. Sgro, A.C. Barone, M. Commodo, A. D’Alessio, A. De Filippo, G. Lanzuolo, P. Minutolo, Proc. Combust. Inst. 32 (2009) 689–696. [25] A.D. Abid, E.D. Tolmachoff, D.J. Phares, H. Wang, Y. Liu, A. Laskin, Proc. Combust. Inst. 32 (2009) 681–688. [26] C.A. Echavarria, A.F. Sarofim, J.S. Lighty, A. D’Anna, Proc. Combust. Inst. 32 (2009) 705–711. [27] J.R. McCormick, B. Zhao, S. Rykov, H. Wang, J.G. Chen, J. Phys. Chem. B 108 (2004) 17398–17402. [28] P.S. Fennell, J.S. Dennis, A. Hayhurst, Combust. Flame 151 (2007) 560–572. [29] E.D. Tolmachoff, A.D. Abid, D.J. Phares, C.S. Campbell, H. Wang, Proc. Combust. Inst. 32 (2009) 1839–1845. [30] G.W. Hewitt, Trans. Am. Inst. Elect. Engrs 76 (1957) 300. [31] E. O. Knutson, K.T. Whitby, J. Aerosol Sci 6 (1975) 443-451. [32] Z.G Li, H. Wang, Phys. Rev. Letters 95 (2005) 014502. [33] Z. Li, H. Wang, J. Aerosol Sci. 37 (2006) 111–114. [34] Z.G. Li, H. Wang, Phys. Rev. E 68 (2003) 061206. [35] Z.G. Li, H. Wang, Phys. Rev. E 68 (2003) 061207. [36] J. Singh, R.I.A. Patterson, M. Kraft, H. Wang, Combust. Flame 145 (2006) 117–127. [37] TSI Inc. (2008) Series 3080 Electrostatic Classifiers Manual, Shoreview, MN. [38] S. C. Wang, R. C. Flagan, Aerosol Sci. Technol. 13 (1990) 230-240. 36 Chapter 3: Evolution of Size Distribution of Nascent Soot in n- and i-Butanol Flames 3.1 Introduction In the present study, the role of fuel structure on soot formation was investigated in a set of laminar premixed flames of n-butane, i-butane, n-butanol and i-butanol. The emphasis of the study was placed on probing the evolution of the detailed particle size distribution function (PSDF) for the alkanes and oxygenated fuels. Cross comparisons were made with respect to butanol isomeric branching structure and between butanols and their parent hydrocarbon analogs. A systematic approach was taken such that the effect of local flame temperature and carbon to oxygen ratio are isolated from the fuel structure effect. The burner stabilized stagnation (BSS) flame approach coupled with mobility sizing; described in detail elsewhere [1,2] was employed to investigate the evolution of PSDFs in nascent soot from particle nucleation to mass growth. The method is inherently intrusive to flame but our technique accounts for flame perturbation by the probe explicitly by treating it, experimentally and computationally, as the downstream boundary of the flame. With the flow field defined, the flame temperature and species concentrations can be directly modeled using a quasi one dimensional code without imposing a measured temperature profile or correcting for artificial probe perturbation [1]. A similar method without the use of a flow stagnation surface has been used by us earlier [3-6] and by other groups (see, e.g., [7-10]). To obtain reliable radiation correction for the measured temperature and to explore the fundamental kinetic causes for the fuel structure effects, a high temperature n- and i-butanol combustion model is used for numerical simulations. The model combines n- and i-butanol chemistry of Moss et al. [11] with existing C 1 -C 4 combustion chemistry in USC Mech II [12]. In 37 this way, the butane and butanol chemistry is made consistent to each other for computation of the temperature profiles of the flames studied. Additionally, analysis of the concentration profiles computed for benzene and other gas-phase species was made as such an analysis provides insights into soot nucleation and formation [13]. 3.2 Experimental Methods The BSS flame approach [1,2] was employed to probe nascent soot formation in butane and butanol flames. The BSS flame configuration can be simulated directly as a quasi one- dimensional problem because the stagnation surface acts as the sampling probe and flame boundary condition simultaneously. One lightly sooting flame was stabilized for each fuel considered at atmospheric pressure with nearly equal maximum flame temperature and flow conditions. Furthermore, the total C/O ratio of the flames was held fixed. The conditions of the flames are summarized in Table 1. The gas temperature profiles were measured with a Y2O3/BeO coated type-S thermocouple with radiation correction using a procedure discussed earlier [14]. The bead diameter was around 0.3 mm after coating. Table 3.1 Summary of Flame Conditions for the Premixed BSS Butanol flames. Mole fractions a Equivalence Velocity, b v 0 Maximum Fuel O 2 C/O ratio,  (cm/s) temperature, T f,max (K) n-C 4 H 9 OH 0.109 0.290 0.632 2.25 4.64 1790±70 i-C 4 H 9 OH 0.109 0.290 0.632 2.25 4.64 1790±70 n-C 4 H 10 0.0958 0.304 0.630 2.05 3.58 1750±70 i-C 4 H 10 0.0958 0.304 0.630 2.05 3.58 1790±70 a The balance gas is argon. b STP cold gas velocity. The flat flame burner is 5 cm in diameter and is uncooled because of potential condensation of the fuel in the porous material. Without water cooling, however, the pores tend to close in its center, thus modifying the local unburned gas velocity. For this reason, fresh porous material 38 was always used to keep the flame roughly one dimensional. A sheath of nitrogen shields the flame to prevent radial entrainment and diffusion of oxygen from ambient air. Liquid n-butanol and i-butanol, both acquired from Sigma-Aldridge (ACS Reagent grade, 99% purity) were vaporized and injected into the fuel line in a manner similar to a previous study [2]. The mass flow rates of n-butane, i-butane, oxygen, argon and nitrogen were measured by critical orifices and the flow of argon driving the fuel nebulizer was calibrated by a bubble meter. The butane isomers were C.P. grade, purchased from Gilmore gas. Particle size distributions were determined with a TSI 3080 SMPS using a sample dilution technique developed earlier and improved over time [3-6,14,15]. The sample gas entered the probe through an orifice and was immediately diluted with a cold nitrogen flow to prevent particle losses. The dilution range and calibration has been used before and care was taken to avoid diffusion losses, condensation of higher-molecular weight hydrocarbons, and probe- induced particle-particle coagulation during dilution [1]. Limitations of the Cunningham slip correction cause particles below 10 nm to be overestimated by mobility measurements and thus a nanoparticle transport theory was used for small particles to obtain more accurate particles sizes [16-18]. The experimentally measured temperature profiles are radiation corrected by using transport and flow properties that are calculated by a modified version of OPPDIF [19]. The ratio of the burner-to-probe separation to the burner diameter is much less than unity so the quasi one- dimensional assumption applies. The flame chemistry was calculated with USC Mech II [12] for the butane isomers. A butanol combustion model will be introduced below and used for simulation of the butanol flames. By energy conservation, the modified OPPIF code allows for the calculation of the temperature and species profile without the need for a measured 39 temperature profile as an input. The radiation corrected temperature profiles are compared to the calculated OPPDIF profile to test the validity of the experimental and numerical procedures. The temperature closest to the burner surface that can be measured is equal to one half of the thermocouple bead diameter (0.15 mm). The inlet temperature was extrapolated from the measured temperature profile immediately adjacent to the burner surface. The temperature variation is roughly linear with respect to the distance, as one would expect because in that region the dominant heat transfer mechanism is heat conduction. The probe temperature was measured with a type K thermocouple embedded on the stagnation surface. 3.3 Gas-phase kinetic model for high-temperature oxidation of butanol isomers The chemistry for high-temperature oxidation of the butanol isomers is extracted from Moss et al. [11] and superimposed over USC Mech II [12]. The two models were first combined by Veloo et al. [19] to model the laminar flame speed of n-butanol flames. The current work added additional chemistry of i-butanol. This subset of the model is, again, based on ref. 5. This approach ensures consistency when the flame chemistry of butanols is compared to their parent non-oxygenated hydrocarbon fuels, although the prediction accuracy of high-molecular weight species for all fuels tested has not been directly verified under the conditions tested. The combination of the two models was also motivated by the consideration that USC Mech II includes an adequate amount of fuel-rich chemistry leading to aromatics formation from small molecular fragments. The resulting model, comprised of 959 reactions and 136 species, is used here primarily for correction of radiative heat loss and should not be viewed or used as an independently proposed model for butanol combustion beyond the current purpose. 40 3.4 Experimental Results The BSS flame approach [1,2] was employed to probe nascent soot formation in butane and butanol flames. The BSS flame configuration can be simulated directly as a quasi one- To ensure an accurate prediction of the heat release rate in the BSS flames, the reaction model was first subject to validation against the laminar flame speed previously reported for n- and i-butanol-air mixtures at an unburned mixture temperature of 343 K [21,22]. The comparison between the measured and calculated n-butanol flame speed is shown in Fig. 3.1. It is seen that the combined model predicts the flame speed rather well. 10 20 30 40 50 60 0.8 1.0 1.2 1.4 S u (cm/s) Equivalence Ratio,  Fig. 3.1. Comparison of predicted and measured flames speeds (S u ) of n-butanol/air flames at 1 atm and an unburned gas temperature of 343 K. The data are taken from [21]. Local temperature is the dominant parameter which governs the soot chemistry. A comparison between the measured/radiation corrected and simulated temperature profiles is shown in Fig. 3.2 for the two butanol isomers at a series of burner-to-stagnation surface separation distances. The degree to which the stagnation probe causes heat loss is shown. In both fuels, the agreement between radiation corrected measurements and simulated temperatures is within thermocouple positioning uncertainty (±0.03 cm) and the temperature measurement uncertainty (±70 K around 41 the peak temperature region). As we discussed in ref. [14], literature emissivity values for Y/Be/O coating are between 0.3 and 0.6 [20,21]. The radiation-corrected temperature was estimated to be the average of the two limiting cases which also yielded the uncertainty bounds for the temperature shown in Fig. 3.2. Fig. 3.2. Measured (symbols) and simulated (lines) temperature profiles. Open symbols and solid lines: i-butanol; filled symbols and dashed lines: n-butanol. The vertical error bars represent the limiting emissivity of 0.3 and 0.6 (see text). The model solves the energy equation without the measured temperature profiles as an input. Thus, agreement between radiation corrected measurements and the simulation addresses uncertainty within the mechanism itself by yielding information on local heat release and loss rates [1]. Such information allows for the uncertainty within the simulated local temperature to be defined along with the resulting Arrhenius reaction kinetics and species transport. Furthermore, the agreement in measured temperatures and the model confirms that the flame conditions are nearly identical across all the flames studied. Temperature, T f (K) Height Above Burner, H (cm) 400 600 800 1000 1200 1400 1600 1800 0 0.2 0.4 0.6 0.8 1 1.2 1.4 42 A similar plot for the measured and simulated temperature profiles is shown in Fig. 3.3 for the n- and i-butane flames. These temperature profiles are very similar to those in the butanol flames. The only exception is for the n-butane flame which gives a maximum flame temperature of 1750 K, rather than 1790 K measured in other flames. The cause for the difference is the heat release rate. Under comparable conditions, n-butane flames tend to have a slightly higher flame speed, leading to a faster temperature rise in the preflame region. The increase in the temperature gradient causes an increased heat loss into the burner; and the maximum flame temperature is reduced accordingly. Fig. 3.3. Measured (symbols) and simulated (lines) temperature profiles. Open symbols and solid lines: i-butane; filled symbols and dashed lines: n-butane. The vertical error bars represent the limiting emissivity of 0.3 and 0.6 (see text). The PSDFs for sooting flames were measured for the butane and butanol isomers. The evolution of the PSDF from nucleation of soot to its growth for the two butanol isomers is summarized in Fig. 3.4. The evolution of the PSDFs is similar to previous measurements of ethylene and dodecane under comparable flame conditions [1,2]. At the early stage of soot formation, newly Temperature, T f (K) Height Above Burner, H (cm) 400 600 800 1000 1200 1400 1600 1800 0 0.2 0.4 0.6 0.8 1 1.2 1.4 43 nucleated particles burst into the lower end of the measurable size window at 2.4 nm. These particles grow in size, producing a shoulder in the PSDF, which grows into a log-normal distribution at larger burner-to-stagnation surface separations. Meanwhile, nucleation persists well into the large separation distances with the PSDF characterized by a strong tail throughout the particle size growth period. Fig. 3.4. Measured PSDFs for i-butanol (open symbols) and n-butanol (filled symbols) flames. Overall, the competition between nucleation and growth appears to be similar across the two butanol flames, with the differences being only quantitative and subtle. As Fig. 3.4 shows, the onset of nucleation is slightly delayed in the n-butanol flame compared to the i-butanol flame, but during the mass and size growth stages, and PSDFs become less distinguishable. At the largest separation distances probed (Hp = 1.2 and 1.4 cm), the lognormal part of the distribution 1.4 1.2 1.1 1.0 0.9 0.8 10 7 10 8 10 9 10 10 10 11 10 100 0.75 cm Diameter, D p (nm) dN/dlogD p (cm -3 ) 44 nearly overlaps each other, whereas the PSDF tails exhibit reproducible and subtle differences between the two flames. Detailed behaviors of nascent soot formation in all four flames probed are by all means similar. Fig. 3.5 provides a comparison of the PSDFs at three separation distances. The three positions represent three separate stages of sooting processes. The distributions at Hp = 0.8 cm is indicative of particle nucleation; those at 1.0 cm show the onset of primary particle formation; and the PSDFs at 1.4 cm illustrate the lognormal nature of the primary particle size distribution with median diameters around 30 nm. Overall, the differences among the four flames probed are only in the quantitative aspects of the PSDF and its evolution. Fig. 3.5. Comparison of PSDFs for i-butanol, n-butanol, i-butane and n-butane flames at selected burner-to-stagnation surface separations. The global sooting behavior for each flame can be determined in terms of the total soot volume fraction by integrating the PSDF over all particle sizes measured (> 2.5 nm). Obviously, particles smaller than the lower detection limit of the particle size do not contribute to volume fraction 10 8 10 10 i-C 4 H 9 OH n-C 4 H 9 OH 0.8 cm i-C 4 H 10 n-C 4 H 10 10 8 10 10 i-C 4 H 9 OH n-C 4 H 9 OH 1.0 cm i-C 4 H 10 n-C 4 H 10 10 100 10 8 10 10 i-C 4 H 9 OH n-C 4 H 9 OH 10 100 1.4 cm i-C 4 H 10 n-C 4 H 10 Diameter, D p (nm) dN/dlogD p (cm -3 ) Diameter, D p (nm) 45 substantially. The soot volume fraction as a function of burner to probe separation, Hp, is shown in Fig. 3.6. At the same C/O ratio, the butanol fuels have higher volume fraction than the alkane fuels throughout the flame. In addition, Fig. 4 shows that the branched alcohol and branched alkane have greater volume fractions relative to their straight chain counterparts. These observations are consistent with previous measurements for sooting tendency of the butanol fuels in doped co-flow diffusion flames [22,23]. In those studies, it was determined that the degree of branching within the fuel structure and the fuel carbon number controls the sooting behavior rather than the fuel bound oxygen. 10 -11 10 -10 10 -9 10 -8 10 -7 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4 Separation Distance, H p (cm) Volume Fraction, F v 10 -10 10 -9 10 -8 10 -7 0.8 1.0 1.2 1.4 Sihfted H p (cm) Volume Fraction, F v i-C 4 H 9 OH n-C 4 H 9 OH n-C 4 H 10 i-C 4 H 10 Fig. 3.6. Volume fraction of nascent soot with D p > 2.4 nm (symbols) measured at several H p for all fuels tested. Lines are drawn to guide the eye. Inset: profiles shifted spatially to illustrate the mass growth rates beyond nucleation. 46 i-Butanol and i-butane also show faster onset of nucleation than their straight chain counterparts. The difference in the nucleation rate is exhibited in the onset of volume fraction rise. Beyond the nucleation stage, the mass growth rates also differ. The difference may be assessed by an arbitrary shift of the volume fraction profiles spatially to match the nucleation part of the curve for i-butanol, as shown in the inset of Fig. 3.6. Clearly, the fuel giving rise to faster particle nucleation also yields faster size and mass growth rates. The final difference in the volume fraction is close to an order of magnitude between the sootiest i-butanol flame and the least sooty n-butane flame. As shown in Table 1, the cold gas velocity of the two butane flames is smaller than that of the butanol flames. At the same separation distance, the particle residence time in the butane flames is longer than that in the butanol flames. Hence, the observed differences in the nucleation and mass growth rates cannot be attributed to the difference in the reaction time. The higher volume of soot measured for the alcohol flames can be attributed, to a large extent, to the higher equivalence ratio than those of the butane flames. Under the same C/O ratio, however, our results indicate that alcohols do not always yield less soot to their alkane counterparts. Additional tests not shown here indicate that the butane flames are as sooty as the butanol flames if the equivalence ratio is made equal. In any case, any effect of the fuel bound oxygen on soot formation is overshadowed by the effect of the branched chain within the fuel structure. For the fuels studied the most conclusive observation of sooting behavior results from the straight versus branched chain within the fuel structure. Soot precursors calculated from the flame chemistry were examined to better understand the quantitative difference in soot formation among the four flames. The model presented does not extend to a fundamental description of 47 soot nucleation and growth. However, the sensitivity of soot precursor formation to the fuel structure can be evaluated. 3.5 Numerical modeling of gas-phase species leading up to benzene Species profiles and reaction rates were analyzed numerically for the BSS flames. In particular, the formation of benzene was analyzed to gain insight into the impact of the fuel bound oxygen and branched chains on soot formation. The species mole fraction profile calculated for benzene at several Hp is shown in Fig. 3.7 for the fuels studied. At the nucleation stage (Hp = 0.80 cm), the benzene concentration is predicted to be significantly higher in the branched chain fuels than in the straight chain fuels. The benzene concentration of the branched isomers is predicted to be 40% greater than the normal isomers in the post-flame region of the nucleation stage. These result are consistent with the earlier onset of soot nucleation observed in flames burning branched chain fuels. The relative concentration of benzene at the nucleation stage is also in agreement with observations of premixed burner stabilized flames [24] and doped co-flow diffusion flames where the peak benzene concentration ranked as i-butane > i-butanol > n-butane ~ n-butanol [23, 24]. 48 Fig. 3.7. Mole fraction profiles of benzene computed at several H p . Formation of the first aromatic ring depends on the formation of acetylene and propargyl radical. The species profile calculated for acetylene and propargyl at several Hp is shown in Fig. 3.8. The higher propargyl concentration in the i-butanol flame is attributable to a larger propene concentration. At Hp=0.8 cm, the peak propene concentration is about 3 times larger than the concentration in the n-butanol flames. Propene formed serves to increase the concentration of the propargyl radicals which can then recombine to form benzene. In contrast, acetylaldehyde and ethyl radical are the most significant intermediate products formed during the initial reactions of n-butanol. The above view is consistent with the experimental observations made in low- pressure burner stabilized flames by Oßwald et al. [24]. Comparing the n- and i-butanol flames they probed by molecular beam synchrotron photoionization mass spectrometry, the peak concentrations of propene, propyne, propargyl and consequently, benzene in the i-butanol flame are decidedly higher that those measured for the n-butanol flame. Hence, both the previous and current analyses suggests that the observed difference in the sooting behaviors is attributable to 0 5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Benzene x 10 4 Height Above Burner, H (cm) 0.8 cm 1.2 cm i-butanol n-butanol i-butane n-butane 1.4 cm 49 the competition of forming C2 versus C3 intermediates during initial attack on the fuel, and that i-butanol promotes the production of C3 species and benzene via propene. Fig. 3.8 Mole fraction profiles of C 2 H 2 (top) and C 3 H 3 (bottom) for computed at several H p . Unlike the nucleation stage, the observed sooting behavior at the mass growth stage is not directly explained by the predicted species profiles of soot precursors. The dependence on fuel structure for both the observed sooting behavior and the predicted benzene concentration becomes less clear at the mass growth stage. The branched fuels have greater benzene formation in the nucleation stage. However, Fig. 7 shows that n-butane flames have more significant benzene formation at the mass growth stage (Hp = 1.2 cm). This exchange in position indicates that propargyl recombination becomes competitive in the straight chain fuels towards the later stage of the flame, where the flame chemistry is more sensitive to the thermodynamic condition rather than the initial fuel structure. 0 5 10 15 20 25 i-butanol n-butanol i-butane n-butane 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 Propargyl x 10 5 Acetylene x 10 3 Height Above Burner, H (cm) 1.0 cm 1.2 cm 1.4 cm 50 3.6 Conclusion The evolution of the PSDF of nascent soot was examined in n-butanol, i-butanol, n-butane and i- butane flames to understand the impact of fuel bound oxygen and the fuel structure on the detailed sooting behavior of butanol fuels. The flames were probed under the same C/O ratio and nearly identical temperature. Under the same C/O ratio, butanol flames in fact nucleate soot earlier and gave greater soot volume fractions than the butane flames. In terms of fuel structure, the branched chain functionality has the most observable effect on soot formation. The onset of soot nucleation is faster in the branched fuels in comparison to the straight-chain counterparts. The faster nucleation rate also propagates into the mass growth stage. A combustion reaction model for i-butanol and n-butanol was applied to analyze the BSS configuration to elucidate the role of soot precursors on the observed sooting behavior. It is shown that for the fuel studied, the fuel structure effect is largely exhibited in the relative importance of C 2 versus C 3 intermediate species formed during the initial stage of fuel breakdown. 51 3.7 Chapter 3 References [1] A. D. Abid, J. Camacho, D. A. Sheen, H. Wang, Combust. Flame 156 (2009) 1862-1870. [2] A. D. Abid, J. Camacho, D. A. Sheen, H. Wang, Energy Fuels 23 (2009) 4286-4294. [3] B. Zhao, Z. Yang, M. V. Johnston, H. Wang, A. S. Wexler, M. Balthasar, M. Kraft, Combust. Flame 133 (2003) 173-188. [4] B. Zhao, Z. Yang, J. Wang, M. V. Johnston, H. Wang, Aerosol. Sci. Technol. 37 (2003) 611-620. [5] B. Zhao, Z. Yang, Z. Li, M. V. Johnston, H. Wang, H. Richter, Proc. Combust. Inst. 30 (2005) 1441-1448. [6] B. Zhao, K. Uchikawa, H. Wang, Proc. Combust. Inst. 31 (2007) 851-860. [7] L. A. Sgro, A. De Filippo, G. Lanzuolo, A. D'Alessio, Proc. Combust. Inst. 31 (2007) 631-638. [8] L. A. Sgro, A. Borghese, L. Speranza, A. C. Barone, P. Minutolo, A. Bruno, A. D'Anna, A. D'Alessio, Environ. Sci. Technol. 42 (2008) 859-863. [9] L. A. Sgro, A. C. Barone, M. Commodo, A. D'Alessio, A. De Filippo, G. Lanzuolo, P. Minutolo, Proc. Combust. Inst. 32 (2009) 689-696 [10] C. A. Echavarria, A. F. Sarofim, J. S. Lighty, A. D'Anna, Proc. Combust. Inst. 32 (2009) 705-711. [11] J. T. Moss, A. M. Berkowitz, M. A. Oehlschlaeger, J. Biet, V. Warth, P.-A. Glaude, F. Battin-Leclerc, J. Phys. Chem. A 112 (2008) 10843-10855. [12] H. Wang, X. You, A. V. Joshi, S. G. Davis, A. Laskin, F. N. Egolfopoulos, C. K. Law USC Mech Version II. High-Temperature Combustion Reaction Model of H2/CO/C1-C4 Compounds., 2007. [13] H. Wang, M. Frenklach, Combust. Flame 110 (1997) 173-221. [14] A. D. Abid, N. Heinz, E. Tolmachoff, D. Phares, C. Campbell, H. Wang, Combust. Flame 154 (2008) 775-788. [15] A. D. Abid, E. D. Tolmachoff, D. J. Phares, H. Wang, Y. Liu, A. Laskin, Proc. Combust. Inst. 32 (2009) 681-688. [16] Z. G. Li, H. Wang, Phys. Rev. E 68 (2003) article 061206. [17] Z. G. Li, H. Wang, Phys. Rev. E 68 (2003) article 061207. J. Singh, R. I. A. Patterson, M. Kraft, H. Wang, Combust. Flame 145 (2006) 117-127. 52 [18] R. J. Kee, J. A. Miller, G. H. Evans, G. Dixon-Lewis, Symp. (Int.) Combust. 22 (1989) 1479-1494. [19] P. S. Veloo, 2010 Spring Technical Meeting, Western States Section of the Combustion Institute, University of Colorado, Boulder, CO, 2010, paper 10S-19. [20] P. S. Veloo, Y. L. Wang, F. N. Egolfopoulos, C. K. Westbrook, Combust. Flame 157 (2010) 1989-2004. [21] P. S. Veloo, F. N. Egolfopoulos, Proc. Combust. Inst. 33 (2011) 987-993. [22] R. C. Peterson, N. M. Laurendeau, Combust. Flame 60 (1985) 279-284. [23] C. R. Shaddix, Proceedings of the 33rd National Heat Transfer Conference, Albuquerque, NM, 1999, paper HTD99-282. [24] C. S. McEnally, L. D. Pfefferle, Proc. Combust. Inst. 30 (2005) 1363-1370. [25] C. S. McEnally, L. D. Pfefferle, Environ. Sci. Technol. 45 (2011) 2498-2503. [26] P. Oßwald, H. Güldenberg, K. Kohse-Höinghaus, B. Yang, T. Yuan, F. Qi, Combust. Flame 158 (2011) 2-15. 53 Chapter 4: Time Dependant Soot Formation in Premixed C 6 Hydrocarbon Flames 4.1 Introduction A fundamental description which quantifies the kinetic, transport and flow processes that occur during particle nucleation and soot growth does not exist [1]. This level of understanding is challenging even for the gas-phase chemistry leading to soot precursors and available kinetic mechanism focus on relatively small species [2,3]. An experimental approach is introduced here to examine fuel specific sooting chemistry on the basis detailed sooting behavior. The role of fuel structure on soot formation is investigated in a set of canonical laminar premixed flames of n-hexane, n-hexene, cyclohexane, methyl-pentane and benzene. The emphasis of the study was placed on probing the evolution of the detailed particle size distribution function (PSDF). A systematic approach was taken such that the effect of local flame temperature and carbon to oxygen ratio are isolated from the fuel structure effect. Cross comparisons of the detailed and global sooting behavior were made for the normal alkane, branched alkane, normal alkene, cycloalkane and aromatic fuels for the C6 hydrocarbon. In addition, the morphology of the nascent soot will be observed to supplement the particle distribution measurements and to assess any possible effects of the parent fuel on the soot nanostructure. The flame structure and sooting behavior of the C 6 hydrocarbons has been examined in previous experimental studies. Measurements of detailed flame structures has been made in low-pressure, burner stabilized flames of cyclohexane, benzene and other C 6 fuels by Hansen co-workers using photoionization mass spectrometry [4-6]. The work by Hansen has lead to development of quantitative knowledge of flame species leading to the formation of aromatics. Photoionization mass spectroscopy has also been used by Qi and co-workers to study the flame structure of premixed benzene flames and the pyrolysis of cyclohexane [7, 8]. The structure of co-flow 54 diffusion flames doped with benzene, cyclohexane and hexane has been studied by McEnally and Pfefferle [9-11]. Detailed and global sooting behavior of the C 6 hydrocarbons has been observed experimentally but the contribution of the C 6 hydrocarbon structure has only recently been examined. Ciajolo, Tregossi and co-workers examined the soot surface structure by UV-vis spectroscopy for premixed benzene and n-hexane flames and found that the soot nano-structure of benzene soot surface may be more ordered that the n-hexane soot surface [12]. The flame structure of premixed cyclohexane flames was also examined by Ciajolo and compared to the benzene and hexane flames. The comparison indicates that cyclohexane soot nucleates faster due to the dehydrogenation of the carbon ring and this is the first indication of a strong effect of the fuel structure within these sooting hydrocarbon flames [13]. The dominant role of local flame temperature was also confirmed by Ciajolo in the study of premixed benzene flames whereby the colder benzene flame favored soot formation [14]. The role of fuel structure was further probed by Ciajolo and co-workers by examining the development of the soot structure for premixed benzene and ethylene flames from soot inception to the later growth stages by FT-IR and UV-vis [15, 16]. The structure of benzene and hexane soot from inverse diffusion flames has been analyzed by Mondragon and Santamaria by FT-IR and the comparison shows that the soot surface for hexane may contain more aliphatic species than the benzene soot [17, 18]. The detailed sooting behavior for premixed benzene flames has been investigated in terms of the evolution of the PSDF by Lighty and co-workers and benzene doping was found to increase the size and volume of soot [19, 20]. On the other hand, similar measurements of the PSDF for premixed ethylene flames carried out by Abid showed very little impact of benzene doping on the PSDF [21]. The BSS flame configuration was introduced by Abid to measure the detailed 55 sooting behavior of premixed ethylene and dodecane flames in a pseudo 1-d configuration with well defined boundary conditions [22, 23]. The BSS configuration was extended by Camacho to study the impact of fuel bound oxygen on sooting behavior for butanol and butane isomers and it was shown that under comparable conditions, the fuel bound oxygen does not reduce soot formation [24]. Development of soot chemistry from the parent fuel requires these direct observations of sooting behavior because the nucleation and growth of soot is often dictated by competing kinetic processes [1].One such competition originates from the fuel structure itself. In recent studies of intermediate species formed in low pressure burner stabilized fuel-rich flames, a drastic sensitivity for the production of benzene with respect to the fuel structure was reported. In particular, the C 6 H 12 isomers studied gave three distinct pathways to form benzene [5, 6]. Such fuel dependence is expected to propagate into soot nucleation and growth within the C 6 hydrocarbons currently studied. In the current study, the burner stabilized stagnation (BSS) flame approach coupled with mobility sizing; described in detail elsewhere [22, 23]; is employed to investigate the evolution of PSDFs in nascent soot from particle nucleation to mass growth. The method is inherently intrusive to flame but our technique accounts for probe-flame perturbation explicitly by treating it, experimentally and computationally, as the downstream boundary of the flame. With the flow field defined, the flame temperature and species concentrations can be directly modeled using a quasi one dimensional code without imposing a measured temperature profile or correcting for artificial probe perturbation [22]. To obtain reliable radiation correction for the measured temperature and to explore the fundamental kinetic causes for the fuel structure effects, a high temperature combustion model for jet fuel surrogates is used for numerical simulations [3]. The gas-phase kinetic model begins with small hydrocarbon oxidation chemistry and ends with 1-ring 56 aromatic formation. Basic understanding of the competition between kinetic processes such as aromatics formation and fragmentation provides insights into soot formation [25]. In addition, the BSS flame configuration allows for the thermophoretic velocity of soot to be quantified within the domain thus allowing for sooting behavior to be compared in terms of residence time in the flame. The morphology of nascent soot is also observed by AFM to supplement the mobility measurements of the particle distribution and to assess any effect of the parent fuel on the development of nascent soot nanostructure. 4.2. Experimental Methods The BSS flame approach was employed to probe nascent soot formation in the flames of C 6 hydrocarbons summarized in Table 1. One lightly sooting BSS flame was stabilized for each fuel at atmospheric pressure with maximum flame temperature of 1800K. Furthermore, the total C/O ratio of the flames was held fixed at C/O = 0.69. The adiabatic flame temperature for benzene is much greater than the other flames because the equivalence ratio is closer to unity. Thus the cold gas velocity of the benzene flame is lower than the flow rate of the other fuels to match the 1800 K flame temperature constraint. The gas temperature profiles were measured with a Y 2 O 3 /BeO coated type-S thermocouple with radiation correction using a procedure discussed earlier [26]. The bead diameter was approximately 0.3 mm after coating. 57 Table 4.1 Summary of the premixed BSS flame compositions. The maximum flame temperature is 1800K for each flame. Mole fractions a C/O  Velocity b , ν o (cm/s) O 2 n-hexane n-C 6 H 14 0.0748 0.325 0.69 2.19 4.57 2-methylpentane i-C 6 H 14 0.0748 0.325 0.69 2.19 4.57 n-hexene n-C 6 H 12 0.0748 0.325 0.69 2.07 3.85 cyclohexane c-C 6 H 12 0.0748 0.325 0.69 2.07 4.87 benzene C 6 H 6 0.0748 0.325 0.69 1.72 3.41 a. The balance gas is argon (X Ar = 0.6) b. STP cold gas velocity The flat flame burner is 5 cm in diameter and is uncooled because of potential condensation of the fuel in the porous material. Without water cooling, however, the pores tend to close in its center over time, thus modifying the local unburned gas velocity. For this reason, fresh porous material was always used to keep the flame roughly one dimensional. A sheath of nitrogen shields the flame to prevent radial entrainment and diffusion of oxygen from ambient air. The C 6 hydrocarbon fuels, supplied by Sigma-Aldridge (ACS Reagent grade, 99% purity), were injected into the fuel line and vaporized in a manner similar to a previous study of dodecane BSS flames [23]. The mass flow rates of oxygen, argon and nitrogen were measured by critical orifices and the flow of argon driving the fuel nebulizer was calibrated by a bubble displacement. Particle size distributions were determined with a TSI 3080 SMPS (Electrostatic Classifier 3085 and UCPC 3080, AIM Software V.8.1) using a sample dilution technique developed earlier and 58 improved over time [21, 26-29]. The sample gas entered the probe through an orifice and was immediately diluted with a cold nitrogen flow to prevent particle losses. The dilution range and calibration has been used before and care was taken to avoid diffusion losses, condensation of higher-molecular weight hydrocarbons, and probe-induced particle-particle coagulation during dilution [22]. Limitations of the Cunningham slip correction cause particles below 10 nm to be overestimated by mobility measurements and thus a nanoparticle transport theory was used for small particles to obtain more accurate particles sizes [30-32]. The experimentally measured temperature profiles are radiation corrected by using transport and flow properties that are calculated by a modified version of OPPDIF [22, 33]. The ratio of the burner-to-probe separation to the burner diameter is much less than unity so the quasi one- dimensional assumption applies. The flame chemistry for the C 6 hydrocarbons was calculated with a reduced JetSurF mechanism [3] to be introduced below. By energy conservation, the modified OPPDIF code allows for the calculation of the temperature and species profile without the need for a measured temperature profile as an input. The radiation corrected temperature profiles are compared to the calculated OPPDIF profile to test the validity of the experimental and numerical procedures. The temperature closest to the burner surface that can be measured is equal to one half of the thermocouple bead diameter (0.15 mm). The inlet temperature was extrapolated from the measured temperature profile immediately adjacent to the burner surface. The temperature variation is roughly linear with respect to the distance, as one would expect because in that region the dominant heat transfer mechanism is heat conduction. The probe temperature was measured with a type K thermocouple embedded on the stagnation surface. The flame perturbation due to the sampling probe is included in the modified OPPDIF code by introducing a stagnation surface at x = H p in the form of a zero velocity boundary condition [22]. 59 The flow stagnation causes the fluid velocity and hence the convective time for soot nucleation and growth to increase dramatically. For soot particles, the finite residence time is determined by considering the thermophoretic velocity which is driven by the significant temperature gradient, ∂ T /∂x, at the stagnation plate. In this study, the thermophoretic velocity of the particle within the flame will be calculated in the hard sphere and free molecular formulation [34]: where λ is the thermal conductivity of the gas-phase calculated from transport properties of the flame gas in the multi-component formulation. The momentum accommodation factor, , in this limit is given by a value of 0.9 based on Millikan’s oil droplet experiments [34]. The number density of the gas phase molecules, N, the Boltzmann constant, k, and the local flame temperature, T also contribute. The thermophoretic velocity was calculated numerically and added to the total velocity. The residence time of the soot particles is defined as the time interval for the particle (or precursors) to traverse from the calculated location of the peak flame temperature to the location of the stagnation probe. The chemistry for high-temperature oxidation of the C 6 hydrocarbons is taken from the JetSurF 2.0 mechanism which incorporates an updated description of aromatic formation from cyclohexane [3]. For the present study, the mechanism includes an adequate amount of fuel-rich chemistry leading to aromatic formation from small molecular fragments, although the prediction accuracy of high-molecular weight species for all fuels tested has not been directly verified under the conditions tested. The full JetSurF model was reduced to include only the oxidation for the given parent fuels along with the subsequent benzene formation chemistry. As discussed in the acknowledgements section, this reduction was carried out by the primary author of JetSurF to 60 provide a more computationally efficient model for the fuel oxidation and benzene formation without compromising the validity of flame chemistry conclusions. The reduced JetSurF model, comprised of 1014 reactions and 175 species, allows for the sooting behavior of the flames to be examined in terms of the parent fuel chemistry leading up to aromatic formation. The reaction mechanism, thermochemical and transport databases are available in the Supplemental Material. Morphology measurements were performed using a Nanoscope Multimode V atomic force microscope (AFM) on soot from the benzene and n-hexane flames. These flames were chosen because they represent the extremes of the study in terms of differences in fuel structure, an aromatic fuel with a C/H ratio of 1 and an unsaturated straight-chained fuel with a C/H ratio of 3/7. Soot was sampled on freshly cleaved mica disks for AFM measurement. This substrate was chosen because it has a flat surface at the nanometer scale and is tolerant to heat. Soot was sampled by swiping the substrate horizontally across the flame. A stepper motor was used to perform the swipe in order to maintain consistent swipe speed and total exposure time in the flame for each sample. The typical exposure time for each swipe was 12 ms. Soot samples were chosen from two regions in the flames, the chosen regions corresponded to an early and a late residence time for particle growth to capture two different stages of the growth process. The samples were imaged immediately after collection to reduce instances of aging and oxidation in the ambient air. 4.3 Experimental Results Local temperature is the dominant parameter which governs the soot chemistry. A comparison between the measured/radiation corrected and the simulated temperature profile is shown in Fig. 1 for n-hexane at a series of burner-to-stagnation surface separation distances. The degree to 61 which the stagnation probe causes heat loss is shown. The boundary condition gives rise to a different flame at each sampling distance. However, the inlet and probe are the only temperatures required to model the flame for the given flow rate and sampling distance. The agreement between the radiation corrected measurements and the simulated temperatures is within the thermocouple position uncertainty (± 0.3 cm) and the temperature measurement uncertainty (± 70 K around the peak temperature region). The temperature profiles for the other fuels studied are similar at a series of sampling distances. Fig. 4.1. Measured (symbols) and simulated (lines) temperature profiles for the n-hexane flames at the given sampling locations, H p . The vertical error bars represent the uncertainty in thermocouple radiation corrections as described in the text. The measured and simulated temperature profile for all the fuels studied is shown in Fig. 2 for H p = 1.2 cm. Each fuel has a different inlet flow rate thus each residence time and local temperature 400 600 800 1000 1200 1400 1600 1800 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Temperature, (K) Distance from Burner, H (cm) 62 is slightly different for the given boundary conditions. For example, the calculated benzene flame temperature peaks slightly sooner but the post-flame region is colder due to the lower cold gas velocity. The model solves the energy equation without the measured temperature profiles as an input. Thus, agreement between radiation corrected measurements and the simulation addresses uncertainty within the calculated local heat release and loss rates [22]. Uncertainty within the simulated local flame temperature along with the resulting Arrhenius reaction kinetics and species transport can be defined with this information. In addition, the agreement between the measurement and computation confirms that the flame conditions are comparable and that sooting behavior can be observed under similar local temperature. Fig. 4.2 Measured (symbols) and simulated (lines) temperature for the C6 hydrocarbon flames compared with H p = 1.2 cm. The thermocouple radiation correction for methylpentane was estimated from the flame composition and transport properties of n-hexane. Detailed sooting behavior is examined in terms of nascent soot particle morphology and particle size for the sooting flames of C 6 hydrocarbons. In the current work, the primary method to 450 675 900 1125 1350 1575 1800 hexane nhexene cyclohexane benzene methylpentane 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Distance from Burner, H (cm) Temperature, T f (K) n-hexane n-hexene cyclohexane benzene 2-methylpentane 63 determine the particle size distribution is to measure the mobility of the soot via the SMPS system described above. However, novel microscopy techniques such as Helium-Ion Microscopy (HIM) and AFM have lead to observations which indicate that nascent soot particles begin to deviate from the spherical primary particle shape much earlier than expected [35,36]. This observation has immediate implications to the current work because the mobility measurements of the number density for a given particle size must be corrected for deviations from the spherical assumption or the distributions must be expressed in terms of particle surface area. The morphology of nascent soot during particle nucleation and early growth was observed to determine the general shape during these stages by observing many benzene and n-hexane soot particles using AFM. Particles in the early soot growth stages form about 40 ms after the flame zone qualitative deviations from spherical shape have been found. Representative AFM images of this behavior are shown in Fig. 3 with benzene soot in the right image and hexane soot on the left. As Fig. 3 shows, the nascent soot particles are asymmetrical with very few round particles. The morphology of particles larger 5 nm in dimension is more intricate than a simple sphere. Fig. 4.3: AFM images of soot from benzene (left) and hexane (right) flames taken from flame locations corresponding to 40 ms residence time. The deviation from spherical shape is quantified by calculating the sphericity ratio and circularity were for a set of particles from each flame. For a perfect circle the expected sphericity 64 ratio, Φ, is unity . The sphericity ratio for the measured particles is shown in Fig. 4, for particles from both benzene and n-hexane. For all of the particles observed, the sphericity ratio is between 0.6 and 0.75 and the sphericity ratio appears to be independent of particle size. The expected values of the circularity are 4π and 16 for a circle and square respectively. The values of circularity for the primary soot particles cluster around 15 and are also independent of particle size and parent fuel. The current finding that even small, primary particles are far from spherical is consistent with the recent morphology measurements [35, 36] and indicates that care must be taken when interpreting information from SMPS mobility measurements. Fig. 5.4 Sphericity ratio and circularity vs. particle diameter for particles from benzene (solid symbol) flames and n-hexane (open symbol) flames. During electrostatic classification, the particles are in low Reynold’s number flow and the particle mobility is determined from a balance between electrostatic and drag forces. The electrostatic force is known and the drag force is calculated from a slip corrected Stoke’s law 0 0.5 1 1.5 2 2.5 3 5 10 15 20 1000 2000 3000 4000 Sphericity Circularity Projected Area, A p (nm 2 ) 65 which is derived from the forces acting on the surface of a rigid sphere. Conventionally, this drag force is used to quantify the size of particles which pass through the electrostatic classifier. If the particles are not spherical, the mobility must be expressed in terms of the surface area which experiences the drag force, or the mobility must be corrected for shape. In the current work, the measurements of mobility will be analyzed in terms of the equivalent surface area of a spherical particle of diameter D p . Specifically, the number density distribution in terms of diameter, dN/dlogD p , will be converted to number density distribution in terms of particle surface area, dN/dlogS by expressing the diameter of the sphere in terms of the sphere surface area. The true surface area of the particle, S, is equal to the surface area corresponding to a spherical particle of diameter D p because the drag force is equal for a given surface area under the Stokes flow regime. The evolution of the PSDF from the onset of nucleation size particles to later growth stages is measured for the C 6 fuels. The development of the PSDF is summarized in Fig. 5 for the straight chain and branched chain isomers of hexane (C 6 H 14 ). It is well known that the local flame temperature and C/O ratio dominate the global sooting behavior of premixed flames, thus the constrained peak flame temperature and C/O ratio will isolate the fuel structure effect between the two isomers of hexane. At the early stage of soot formation, newly nucleated particles burst into the lower end of the measurable size window at 2.4 nm. These particles grow in size, producing a shoulder in the PSDF, which grows into a log-normal distribution at larger burner- to-stagnation surface separations. Meanwhile, nucleation persists well into the large separation distances with the PSDF characterized by a strong tail throughout the particle size growth period. 66 Fig. 4.5 Measured PSDFs for n-hexane (filled symbols) and 2-methylpentane (open symbols) flames expressed as a function of the particle surface area. Bi-modal distributions (solid lines) are fit to the PSDF at H p = 1.0 and 1.2 cm to highlight nucleation size particles which persist late in the flame. As described in the text, recent morphology observations of nascent soot suggests that particles larger than D p = 5nm are not spherical thus the particle surface area is considered and the equivalent spherical diameter is shown on the upper x-axis. Overall, the competition between the nucleation and growth processes of soot is similar across the two hexane isomers with the differences being only quantitative and subtle. As Fig. 5 shows, the onset of nucleation is slightly delayed in the n-hexane flame compared to the 2- methylpentane flame, but during the mass and size growth stages, and PSDFs become less distinguishable. At the largest separation distance probed (H p = 1.2), the lognormal part of the 10 7 10 8 10 9 10 10 10 11 10 2 10 3 10 4 10 5 K A 10 7 10 8 10 9 10 10 10 2 10 3 10 4 10 5 H p = 0.6 cm 0.7 1.2 1.0 dN/dlogS (cm - 3 ) Equivalent Diameter, D p (nm) 10 7 10 8 10 9 10 10 10 11 10 100 H p = 0.6 cm 10 7 10 8 10 9 10 10 10 2 10 3 10 4 10 5 H p = 0.6 cm 0.7 1.2 1.0 dN/dlogS (cm - 3 ) Particle Area, S (nm 2 ) 67 distributions nearly overlap each other, whereas the nucleation mode of the PSDF exhibits reproducible and subtle differences. The back panel of Fig. 5 shows the particle size assuming spherical particles and both fuels give similar median diameters under this assumption. The relative trend in nucleation and overall PSDF characteristics between branched and straight chain isomers observed in Fig. 5 has also been observed in previous observations of sooting C 4 premixed BSS flames. In the study of C 4 fuels, delayed nucleation and smaller particle size distributions were observed in the straight chain isomers when compared to the branched chain fuel [24]. The present comparison between methylpentane and n-hexane shows a similar trend but this fuel effect may be reduced by the extended carbon chain of C 6 H 14 which exists in both C 6 H 14 isomers. For the previous isomers, the evolution of nascent soot can be directly compared in terms of particle residence time for each H p because the boundary conditions and flow rates are the same. A similar comparison for the C 6 H 12 isomers (cyclohexane and n-hexene) is made but the comparison is not as direct because residence time is longer in the n-hexene flame for the given sampling locations. The adiabatic flame temperature of n-hexene is greater than cyclohexane for the given flame condition. Thus, the constraint of C/O = 0.69 and 1800K maximum flame temperature is satisfied for cylclohexane and the other fuels by controlling the cold gas velocity of the BSS flame. The particle residence time gives a direct comparison between the cycloalkane and alkene isomers for the given sampling locations. The development of the PSDF for a series of sampling locations and residence times is summarized in Fig. 6 for both C 6 H 12 isomers. Nucleation for each isomer occurs at Hp = 0.55 cm as observed by the onset of the PSDF. However, the residence time for the n-hexene soot is slightly longer due to the lower cold gas velocity. The evolution of the PSDF is similar for the 68 branched alkane, straight alkane, straight alkene and cylcoalkane analogs of C6. In all the flames studied, the onset of soot nucleation is followed by a bimodal PSDF due to persistent soot nucleation. In addition, the trough between the nucleation and growth particle sizes is roughly 10 nm in each flame which indicates that the monomer chemistry is similar [32]. Fig. 4.6. Measured PSDFs for n-hexene (open symbols) and cyclohexane (filled symbols) flames expressed as a function of the particle surface area. Bi-modal distributions (solid lines) are fit to highlight nucleation size particles which persist late in the flame. 10 7 10 8 10 9 10 10 H p = 0.55 cm t = 32 ms t = 28 ms C 6 H 12 10 7 10 8 10 9 10 10 H p = 0.70 cm 37 ms cyclohexane 44 ms n-hexene 10 7 10 8 10 9 10 10 44 ms 53 ms H p = 0.80 cm 10 7 10 8 10 9 10 10 10 2 10 3 10 4 10 5 72 ms 89 ms H p = 1.20 cm 10 7 10 8 10 9 10 10 58 ms 72 ms H p = 1.00 cm dN/dlogS (cm - 3 ) Particle Area, S (nm 2 ) 69 Quantitative differences in detailed sooting behavior between the C 6 H 12 isomers (cyclohexane and n-hexene) are not as subtle as the alkane isomers (n-hexane and methylpentane) previously discussed. The PSDF growth summarized in Fig. 6 indicates that nascent soot from n-hexene nucleates later and develops slower than cyclohexane soot. For example, the median diameter of the PSDF is 45 nm for both of the C 6 H 12 flames at Hp = 1.0 cm. However, it took nearly 20% less time for the median diameter to reach 45 nm in the cyclohexane flame. Another striking difference between nascent soot of C 6 H 12 isomers is the relative intensity of the nucleation mode within the PSDF. Nucleation persists late for both flames but the nucleation size particles become undetectable after 70 ms in the cyclohexane flame. Under comparable composition and flame conditions, the relative early end to nucleation in the cyclohexane is only attributable to the chemical structure difference in the fuel. For the fuels considered above, the most significant fuel structure effect is in cyclohexane flames where there is a relatively early start to soot nucleation followed by a relatively early end to nucleation. The chemical pathway towards soot formation in cyclohexane flames allows for soot to nucleate faster. On the other hand, soot nucleation is relatively weak later in the cyclohexane flame as the precursors are depleted and possibly scavenged by soot surface. The chemical pathway of cyclohexane will be explored shortly with regards to the observations of the detailed PSDF. The detailed sooting behavior of benzene was also examined in terms of the evolution of the PSDF. As Fig. 7 shows, the development of the PSDF does not follow the other fuels at the given sampling locations. The smallest sampling distance that could be probed in a stable manner is more limited in benzene than the other flames because the benzeme flame is more sensitive to heat loss to the stagnation probe. As shown in Fig. 2, the temperature profile for the benzene flame peaks at 1800 K but the post-flame region is weaker than the other flames due to 70 the lower cold gas velocity. The benzene flame is significantly disturbed by the sampling plate at Hp = 0.7 cm because the sharper temperature gradient of the post-flame causes more heat loss to the probe. Due to this limitation, the burst in nucleation size particles which occurs closer to the flame could not be sampled in the benzene flame. Detailed sooting properties such as the onset of nucleation and median particle size are different in the benzene flame. The onset of nucleation cannot be directly observed in the benzene flame but Fig. 7 shows that the bimodal distribution exists at the smallest sampling distance. Nucleation size particles become undetectable after Hp = 0.8 cm in the benzene flame and the PSDF transitions to a single log normal mode of larger particles. In addition, the median particle size of nascent soot is significantly greater in the benzene flames. Besides affecting the temperature profile, the lower cold gas velocity in the benzene flame also increases the residence time that particles undergo at the given sampling distances. The effect of longer residence time may contribute to the end of nucleation and larger particle size in the benzene flames as will be discussed shortly. Fig. 4.7. Measured PSDFs (symbols) for benzene flames fit to normal distributions (lines) which are expressed as a function of the particle surface area. 10 7 10 8 10 9 10 10 10 2 10 3 10 4 10 5 H p = 0.73 cm 0.8 1.2 1.0 dN/dlogS (cm - 3 ) Particle Area, S (nm 2 ) 10 7 10 8 10 9 10 10 10 11 10 2 10 3 10 4 10 5 K A 10 7 10 8 10 9 10 10 10 2 10 3 10 4 10 5 H p = 0.6 cm 0.7 1.2 1.0 dN/dlogS (cm - 3 ) Equivalent Diameter, D p (nm) 10 7 10 8 10 9 10 10 10 11 10 100 H p = 0.6 cm 71 The benzene flame was probed further by thermophoretic sampling to confirm whether a bimodal soot distribution similar to the other fuels exists. The sampling technique described above for AFM imaging is less intrusive and the benzene flame can be sampled at shorter soot residence times. A representative AFM image of nascent soot from the benzene flame at the early growth stages is shown in Fig. 8. The general distribution observed via AFM consists of both primary nucleation size particles and small aggregates. The particle labeled 1 is an example of a small primary particle, the height of this particle, which corresponds to the diameter, is roughly 5nm. Particle 2 is an aggregate with a height of 7nm. The distribution obtained via AFM supplements the evolution of the PSDF measured from mobility measurements. The morphology observations also confirm the existence of nucleation size particles with diameters below 10nm in benzene soot and a bimodal size distribution at early growth stages similar to the other fuels. Fig. 4.8. Representative 3-D plot of soot from the early growth stages, H p=5.5mm, in a benzene flame. Particle 1 is a primary particle with a diameter of approximately 5nm and particle 2 is an early aggregate with a diameter of approximately 7nm. Global sooting properties such as total soot volume fraction, F v , can be determined for each flame by integrating the PSDF over all particle sizes measured (Dp > 2.4 nm). The soot volume 72 fraction as a function of burner-to-probe separation distance, H p , is shown in Fig. 9 for n-hexane and 2-methylpentane flames. The flow rates and boundary conditions are identical between the alkane isomers thus the soot formation processes occur under the same residence time as shown in the upper x-axis of Fig. 9. Under this condition, the branched isomer has a faster onset of nucleation and greater soot volume fraction throughout. As discussed above, the difference in sooting behavior between the branched and straight chain alkanes is less dramatic than the differences observed previously in C 4 fuels [24]. The rise in volume fraction for the n-hexane flame occurs at the same rate as the 2-methylpentane flame after the delayed onset of nucleation. The branched chain increases sooting behavior in terms of faster nucleation time and higher final soot concentration. Fig. 4.9. Volume fraction of nascent soot with D p > 2.4 nm (symbols) measured at the given H p for the C 6H 14 fuels studied. Lines are drawn to guide the eye. 10 -10 10 -9 10 -8 10 -7 0.6 0.7 0.8 0.9 1.0 1.1 1.2 30 40 50 60 70 80 Volume Fraction, F v Separation distance, H p (cm) Residence Time, t (ms) methylpentane n-hexane C 6 H 14 T f,max = 1800 K C/O = 0.69 73 The global sooting behavior for all C 6 hydrocarbon flames studied is summarized in Fig. 10 in terms of total soot volume fraction. The volume fraction is shown as a function of probe to burner separation distance, H p , and particle residence time, t. Each BSS flame is sampled within the same range of H p as shown in the top plot of Fig. 10. However, the residence time that the soot particle undergoes in the post-flame region varies for each flame due to the contrasting flow rates. A basis formed from the reaction time of soot must be established by calculating the soot particle residence time at each H p . The middle plot of Fig. 10 shows that the global sooting behavior of all the fuels studied collapses on the basis of residence time. This is expected because the particle residence time is a more fundamental scale of soot formation than the sampling height. The onset of nucleation occurs 25 ms after the flame zone and the total soot volume fraction reaches 10-7 after 70 ms. The time development of detailed PSDFs and global properties is similar between the given C 6 hydrocarbon flames. This observation follows the established conclusion that global sooting behavior is dominated by flame temperature and C/O ratio in premixed flames rather than the structure of the fuel. However, comparisons of the fuel structure impact can be made between the given C 6 hydrocarbons on a more quantitative level if the timescale of soot formation is established. 74 Fig. 4.10. Volume fraction of nascent soot with D p > 2.4 nm (symbols) measured for all the fuels studied as function of sampling distance and particle residence time. For the  = 2.07 comparison, the benzene flame was not changed from = 1.7 because the fraction of soot increases beyond measurable limits in the benzene flame. Lines are drawn to guide the eye. hexane 0.69 methylpentane 0.69 n-hexene cyclohexane benzene 10 -10 10 -9 10 -8 10 -7 0.6 0.7 0.8 0.9 1.0 1.1 1.2 Separation distance, H p (cm) Volume Fraction, F v n-hexane 2-methylpentane n-hexene cyclohexane benzene 10 -10 10 -9 10 -8 10 -7 30 40 50 60 70 80 90 100 Residence time, t (ms) T f,max = 1800 K C/O = 0.69 Volume Fraction, F v 10 -10 10 -9 10 -8 10 -7 30 40 50 60 70 80 90 100 Residence time, t (ms) T f,max = 1800 K = 2.07 Volume Fraction, F v  = 1.72 75 In terms of samping distance, H p , the detailed PSDF and soot volume fraction in the benzene flame are significantly larger than the other C 6 fuels. In terms of residence time, however, the detailed and global behavior of soot from benzene collapses closer to the rest of the fuels. The final soot volume fraction is the greatest in the benzene flame for the given conditions as seen in Fig. 4.10. This conclusion is expected based on the aromatic structure of benzene but the difference in final volume fraction observed between the other premixed BSS flames is only a fine detail. Hexane, methylpentane and cylcohexane flames approach the final soot volume fraction of benzene flame as seen in the middle plot of Fig. 4.10. The detailed PSDFs also show that the final median diameter is 50-60 nm for each flame with soot from the benzene flame consistently being the largest. A comparison on the basis of equivalence ratio is shown in the bottom plot of Fig. 9 for  = 2.07. On the basis of equivalence ratio, the volume fraction of soot increases with the C/O ratio of the flame and the benzene flame was not changed because heavy sooting occurs for the premixed benzene flame at C/O = 0.8. The impact of the parent fuel structure can be examined further if both the equivalence ratio and C/O ratio are identical. This basis of comparison is shown for in Fig. 11 for the cyclohexane, n- hexane and ethylene flames in terms of the soot volume fraction and number density. The ethylene flame is similar to the BSS flame reported previously with the flame temperature slightly decreased to match the current study [22]. On the basis of identical C/O and C/H ratio, the impact of the parent fuel structure is seen in the premixed flames in terms of contrasting final soot volume fraction, number density and nucleation time. The hierarchy of the soot volume fraction resembles the relative trend observed in non-premixed flames where aromatic structures and higher carbon number increase soot. The number density of the cyclohexane flame shown in 76 Fig. 4.10 is relatively small because nucleation size particles stop forming over time in the cyclohexane flames as discussed above. Fig. 4.11. Volume fraction and number density of nascent soot with D p > 2.4 nm (symbols) measured for cyclohexane, n-hexene and a previously reported ethylene flame [22] as function of particle residence time. Lines are drawn to guide the eye. Detailed sooting characteristics such as nucleation time and persistence of nucleation over time can be established if the residence time is known. As seen in the detailed PSDFs above, the persistence of soot nucleation varies for each fuel at the given conditions. The cyclohexane and benzene flames stop nucleating soot 70 ms after the flames zone. With factors such as flame temperature, C/O ratio and residence time eliminated, this indicates that soot precursor and monomer depletion occurs more readily in the aromatic and alkane ring fuels. However, the precise process in the benzene and cyclohexane flames which causes this behavior is unknown. Previous experimental and modeling studies have shown that competing kinetic processes of soot 10 8 10 9 10 10 30 40 50 60 70 80 Residence time, t (ms) Number Density, N (cm -3 ) 77 formation can be modeled in terms of the features of the bi-modal PSDF [32]. The transition from bimodal back to unimodal must also be included in cases such as cyclohexane and benzene where precursor depletion begins early and only growth processes remain. The time at which the onset of soot nucleation occurs can also be determined in the given BSS flames. The relative order in which nucleation begins may reveal insight into the fuel structure effect on soot precursor formation. As discussed above for n-hexane and methylpentane (alkane isomers), the fuel structure impacts the onset of nucleation under comparable conditions. All the flames studied show nucleation around 30 ms but the relative magnitude of the detailed PSDF at soot nucleation indicates that cyclohexane and benzene are significantly faster. Nucleation is next in methylpentane flames and the slowest onset of nucleation occurs in n-hexene flames. The competing kinetic processes leading to the nucleation of soot are linked to the underlying gas phase chemisty in a manner that is not completely understood. The formation of aromatic precursors is an important rate-limiting step and the onset of soot nucleation may be tied to the fuel specific chemistry leading to aromatic formation. Unlike soot formation processes, the fuel specific gas-phase chemistry is well developed. 4.4 Numerical modeling of gas-phase species leading up to benzene Species profiles and gas-phase reaction rates were analyzed numerically for the BSS flames with high-temperature oxidation chemistry containing fuel rich chemistry of small molecular fragments and larger fuels. The JetSurF gas-phase kinetic model does not extend to a fundamental description of soot nucleation and growth. However, the sensitivity of soot precursor formation to the fuel structure can be evaluated. In particular, the formation of acetylene, propargyl radical and benzene was analyzed to gain insight into the impact of fuel 78 structure within the C 6 hydrocarbon fuels. The computed species profiles are shown in Fig. 4.12 as a function of stagnation probe-to-burner separation distance, H. The profiles are characterized by the two distinct regions which come before and after the thin reaction zone (flame zone). Chemistry for 2-methylpentane is not fully developed but the behavior relative to n-hexane is assumed to be similar to the relative behavior of i-butane and n-butane fuels under comparable conditions [24, 37]. The faster nucleation time and greater final volume fraction in methylpentane discussed above may be influenced by greater benzene formation in the pre-flame and post-flame regions as observed in the butane study discussed previously. Fig. 4.12. Mole fraction profile of acetylene (top panel-solid lines), propargyl radical (top panel-dotted lines) and benzene (bottom panel-solid lines) computed at H p = 1.2 cm for each of the fuels studied. The behavior of 2-methylpentane (not computed here) relative to n-hexane is assumed to be similar to the relative behavior of i-butane and n-butane fuels under comparable conditions [24]. 0 5 10 15 20 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Separation Distance, H (cm) [ C 6 H 6 ] x 10 4 cyclohexane n-hexene benzene n-hexane T f = 1800 K 0 5 10 15 20 25 30 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Separation Distance, H (cm) [ C 2 H 2 ] x 10 3 [ C 3 H 3 ] x 10 5 cyclohexane n-hexene benzene n-hexane 79 The onset of nucleation is fastest in the cyclohexane flame and the pre-flame peak of benzene production shown in Fig. 4.12 may influence this detailed sooting behavior. The prediction of benzene at the peak temperature allows for a possible pathway to soot which starts in the pre- flame region. Pre-flame aromatic production is an exceptional case which does not exist in the classical mechanism of soot formation. It is well established that local flame temperature and acetylene production in the post flame are the main factors controlling soot precursor formation thus the parent fuel structure is only of secondary importance. In this case, the fuel structure of cylohexane has a primary impact in a manner not conventionally considered. Recently developed fuel chemistry shows that the formation of the first aromatic ring does not depend only on acetylene and propargyl radical production. This is possible in fuels that have structures close to the benzene ring such as cyclohexane. The reaction rates of benzene formation were analyzed numerically and 3 prominent pathways are summarized in Fig. 4.13 as function of distance from the burner, H. Propargyl recombination (C 3 path), Butynyl + Acetylene (C 2 path), and dehydrogenation (C 6 path) were the most dominant pathways to benzene for each of the fuels studied. The C 3 path does not vary significantly between the fuels studied. However, the rate of the C 2 path is predicted to be significantly greater in the pre-flame region of cyclohexane and the C 6 path only exists in cyclohexane flames. The C 2 and C 6 path which are specific to the cyclohexane flames may provide a pathway for soot formation which begins in the pre-flame region. If stable aromatics survive the flame, this early benzene production may influence the onset of soot nucleation in cyclohexane flames in a manner not conventionally considered before. 80 Fig. 4.13. Reaction rate profiles computed for propargyl recombination (thin lines), butynyl + acetylene (thick lines) and dehydrogenation (dashed line) steps to benzene formation. Methylpentane and benzene fuels are not shown for clarity. 0 3 6 8 0.1 0.2 0.3 0.4 Rate (mol/cm 3 -s) x 10 6 0 2 3 5 0.1 0.2 0.3 0.4 Rate (mol/cm 3 -s) x 10 6 c-C 6 H 7  C 6 H 6 +H C 4 H 5 +C 2 H 2  C 6 H 6 +H 2 C 3 H 3  C 6 H 6 cyclohexane n-hexene n-hexane 0.1 0.2 0.3 0.4 Distance from Burner, H (cm) 0.1 0.2 0.3 0.4 0 3 6 8 0.1 0.2 0.3 0.4 Rate (mol/cm 3 -s) x 10 6 0 3 6 8 0.1 0.2 0.3 0.4 Rate (mol/cm 3 -s) x 10 6 81 In addition, the detailed species profile of soot precursors shown in Fig. 4.12 may also explain nucleation behavior in later stages of the cyclohexane flame. Under comparable conditions, the PSDF shows the formation of nucleation size particles in cyclohexane flames ends earlier than the alkene counterpart. In cyclohexane flames, the volume of soot may stem from soot precursors formed before the flame, as described above, and these precursors may become depleted relatively early. As opposed to the pre-flame region, the post-flame region of the cyclohexane flame, shown in Fig. 4.12, is predicted to have 33% less benzene formation than the flames with nucleation sized particles. The predicted acetylene profile shown in Fig. 4.12 is comparable to the prediction of n-hexene but the volume of soot, shown in Fig. 10, is double that of n-hexene. Surface scavenging of acetylene may be more prevalent in the cyclohexane flame due to the higher volume of soot and this may also hinder particle nucleation. The relatively early start and early end to nucleation in the benzene flames can be understood under similar arguments. The existence of high concentrations of benzene in the pre-flame region, shown in Fig. 4.11, contributes to the overall volume of soot and to the early formation of nucleation sized particles. On the other hand, the PSDF of benzene shows that the nucleation size particles stop forming earlier than the aliphatic fuels and this behavior may also be tied to the detailed species profile of soot precursors. For the given C/O ratio, the benzene flame has a much lower equivalence ratio than the other fuels. This relatively low excess fuel in the benzene flame causes much lower amounts of soot precursors such as acetylene and benzene to form in the post-flame region. The limited formation of precursors may hinder nucleation of soot in the post-flame region of the benzene flame. Unlike conventional understanding, the parent fuel structure is shown to directly affect sooting behavior in premixed flames. 82 In terms of particle morphology, the nucleation size particles are thought to affect the development of the nascent soot nanostructure. One theory soot growth postulates that nucleation size particles “smooth out” gaps in between particles thus forming primary spheres and further coagulation of particles creates a fractal surface which forms after nucleation ends [38]. In the present work, the morphology of soot sampled during later soot growth is observed to examine the role of nucleation size particles and possibly confirm the preceding argument. The hypothesis is that the spherical monomers composing the aggregates in hexane should be larger than those in benzene due to the persistent nucleation far in to the soot growth region which is found in hexane but not in benzene. The particle morphology as observed by AFM, shown in Fig. 14, shows that there is no difference in size between the spherical monomers in the aggregates from the two fuels, in both cases the monomers are 15-20 nm in diameter. This result suggests that nucleation alone cannot be the primary driving force for the transition from spherical to fractal growth of soot particles. Fig. 4.14. AFM image of a representative aggregate soot particle from benzene (a) and hexane (b) flames at a residence time of 70 ms. 83 A summary of the general evolution of nascent soot in terms of the PSDF is shown in Fig. 4.15. As previously established, the nucleation of soot can be described by second order kinetics with respect to precursors and this propagates into a bimodal PSDF when growth and coagulation occur. Conventionally, the sooting processes which control the features of the detailed PSDF are temperature and flame composition. In the present work, the development of the PSDF was shown to be also affected by the parent fuel structure. Analysis of the time-resolved PSDFs shows that the cyclohexane flame is the fastest to nucleate soot for the fuels studied and this occurs around 28 ms after the flame zone. Nucleation occurs after 25 ms as reported in previous studies of comparable ethylene BSS flames. Other related BSS flames were also on the same time scale [22-24]. The time resolved summary shown in Fig. 15 uses t = 25 ms for the general prediction of the onset of nucleation. The bimodal distribution will transition back to a unimodal PSDF as nucleation is expected to end over time due to precursor depletion and surface scavenging. The fundamental mechanism and rate for this transition is unknown. In the present study, cyclohexane flames were observed to transition from a bimodal to unimodal PSDF after 70 ms. The time resolved observations made in the present work allow for the general picture of soot formation to be more clearly defined. 84 Fig. 4.15. Summary of the time resolved evolution of nascent soot in terms of the PSDF. Soot nucleation is expected to stop later in the flame. The dashed line denotes flames such as the benzene and cyclohexane flames studied which end nucleation relatively early. 4.5 Conclusion The time resolved formation of nascent soot from the onset of nucleation to later growth stages was examined for premixed burner stabilized stagnation (BSS) flames of C 6 hydrocarbons. The overall sooting process was comparable in the fuels as evidenced by similar time resolved bimodal PSDF and particle morphology observations. However, the nucleation time and the persistence of nucleation with time were strongly dependent upon the structure of the parent fuel. The fastest onset of soot nucleation was observed in cyclohexane and benzene flames and this may be due to significant aromatic formation that is predicted in the pre-flame region. In addition, the evolution of the PSDF showed that nucleation ends sooner in cylclohexane and benzene flames and this behavior is related to the relatively early depletion of precursors such as acetylene and benzene. The overall soot nanostructure development was similar for each flame despite the differences in soot nucleation behavior. t > 25 ms Particle Diameter , D p dN/dlog(D p ) t = 50 ms t > 70 ms 85 4.6 Chapter 4 References [1] H. Wang, Proc. Combust. Inst. 33 (2001) 41-67. [2] M. Colket, J. T. Edwards, S. Williams, N. P. Cernansky, D. L. Miller, F. N. Egolfopoulos, P. Lindstedt, K. Seshadri, F. L. Dryer, C. K. Law, D. G. Friend, D. B. Lenhert, H. Pitsch, A. F. Sarofim, M. D. Smooke, W. Tsang., AIAA Paper AIAA-2007- 0770, 2007. [3] B. Sirjean, E. Dames, D. A. Sheen, X. You, C. J. Sung, A. T. Holley, F. N. Egolfopoulos, H. Wang, S. S. Vasu, D. F. Davidson, R. K. Hanson, H. Pitsch, C. T. Bowman, A. Kelley, C. K. Law, W. Tsang, N. P. Cernansky, D. Miller, A. Violi, R. P. Lindstedt. A high-temperature chemical kinetic model of n-alkane (up to n-dodecane), cyclohexane, and methyl-, ethyl-, n-propyl and n-butyl-cyclohexane oxidation at high temperatures, JetSurF version 2.0, September 19, 2010 (http://melchior.usc.edu/JetSurF/JetSurF2.0). [4] M. E. Law, P. R. Westmoreland, T. A. Cool, J. Wang, N. Hansen, C. A. Taatjes, T. Kasper, Proc. Combust. Inst. 31 (2007) 565-573. [5] N. Hansen, T. Kasper, B. Yang, T. A. Cool, W. Li, P. R. Westmoreland, P. Oßwald, K. Kohse-Höinghaus, Proc. Combust. Inst. 33 (2011) 585-592. [6] N. Hansen, W. Li, M. E. Law, T. Kasper, P. R. Westmoreland, B. Yang, T. A. Cool, A. Lucassen, PCCP 12 (2011) 12112-12122. [7] B. Yang, Y. Li, L. Wei, C. Huang, J. Wang, Z. Tian, R. Yang, L. Sheng, Y. Zhang, F. Qi, Proc. Combust. Inst. 31 (2007) 555-563. [8] Z. Wang, Z. Cheng, W. Yuan, J. Cai, L. Zhang, F. Zhang, F. Qi, J. Wang, Combust. Flame 159 (2012) 2243-2253. [9] C. S. McEnally, L. D. Pfefferle, Combust. Flame 129 (2002) 305-323. [10] C. S. McEnally, L. D. Pfefferle, Combust. Flame 136 (2004) 155-167. [11] C. S. McEnally, L. D. Pfefferle, Combust. Flame 143 (2005) 246-263. [12] M. Alfe, B. Apicella, R. Barbella, A. Tregrossi, A. Ciajolo, Proc. Combust. Inst. 31 (2007) 585-591. [13] A. Ciajolo, A. Tregrossi, M. Mallardo, T. Faravelli, E. Ranzi, Proc. Combust. Inst. 32 (2009) 585-591. [14] C. Russo, F. Stanzione, A. Tregrossi, M. Alfe, A. Ciajolo, Combust. Flame 159 (2012), 2233-2242. [15] C. Russo, M. Alfe, J.N. Rouzaud, F. Stanzione, A. Tregrossi, A. Ciajolo, Proc. Combust. Inst. 34 (2013) 1885-1892. 86 [16] C. Russo, F. Stanzione, A. Ciajolo, A. Tregrossi, Proc. Combust. Inst. 34 (2013) 3661- 3668. [17] A. Santamaria, N. Yang, E. Eddings, F. Mondragon, Combust. Flame 157 (2010) 33-42. [18] M. Velasquez, F. Mondragon, A. Santamaria, Fuel 104 (2013) 681-690. [19] C. A. Echavarria, A. F. Sarofim, J. S. Lighty, A. D'Anna, Proc. Combust. Inst. 32 (2009) 705-711. [20] C. A. Echavarria, A. F. Sarofim, J. S. Lighty, A. D'Anna, Combust. Flame 158 (2010) 98- 104. [21] A. D. Abid, E. D. Tolmachoff, D. J. Phares, H. Wang, Y. Liu, A. Laskin, Proc. Combust. Inst. 32 (2009) 681-688. [22] A. D. Abid, J. Camacho, D. A. Sheen and H. Wang, Combust. Flame 156 (2009) 1862- 1870. [23] A. D. Abid, J. Camacho, D. A. Sheen, H. Wang, Energy Fuels 23 (2009) 4286-4294. [24] J. Camacho, S. Lieb, H. Wang, Proc. Combust. Inst. 34 (2013) 1853-1860. [25] H. Wang, M. Frenklach, Combust. Flame 110 (1997) 173-221. [26] A. D. Abid, N. Heinz, E. Tolmachoff, D. Phares, C. Campbell, H. Wang, Combust. Flame 154 (2008), 775-788. [27] B. Zhao, Z. W. Yang, J. J. Wang, M. V. Johnston, H. Wang, Aero. Sci. Tech. 37 (2003) 611-620. [28] B. Zhao, Z. W. Yang, M. V. Johnston, H. Wang, A. S. Wexler, M. Balthasar, M. Kraft, Combust. Flame 133 (2003) 173-188. [29] B. Zhao, K. Uchikawa, H. Wang, Proc. Combust. Inst. (2007) 851-860. [30] Z. G. Li, H. Wang, Phys. Rev. E 68 (2003) article 061206. [31] Z. G. Li, H. Wang, Phys. Rev. E 68 (2003) article 061207. [32] J. Singh, R. I. A. Patterson, M. Kraft, H. Wang, Combust. Flame 145 (2006) 117-127. [33] R. J. Kee, J. A. Miller, G. H. Evans, G. Dixon-Lewis, Symp. (Int.) Combust. 22 (1989) 1479-1494. [34] Z. G. Li, H. Wang, Phys. Rev. E 68 (2003) article 061207. [35] S. Lieb, H. Wang Comb. Flame, in preparation 87 [36] M. Schenk, S. Lieb, H. Vieker, A. Beyer, A. Golzhauser, H. Wang, K. Kohse- Hoinghause, Chem. Phys. Chem., submitted [37] H. Wang, X. You, A. V. Joshi, S. G. Davis, A. Laskin, F. N. Egolfopoulos, C. K. Law, USC Mech Version II: A high-temperature combustion reaction model of H2/CO/C1-C4 compounds, (http://ignis.usc.edu/USC_Mech_II.htm), May 2007. [35] P. Mitchell, M. Frenklach, Proc. Combust. Inst. (1998) 1507-1514. 88 Chapter 5: Kinetics of Nascent Soot Oxidation in a Flow Reactor 5.1 Introduction Critical gaps remain in the fundamental understanding of soot formation [1]. These gaps include the kinetics of soot oxidation. Recent studies on nascent soot morphology, particle size distribution function (PSDF) and composition have highlighted differences between nascent soot and mature, graphitized soot [2-5]. Specifically, it has been shown that nascent soot can contain a significant amount of aliphatic components [1,4,5]. Because the heterogeneous surface reaction kinetics and mechanism are expected to be highly sensitive to the surface composition, the oxidation kinetics of nascent soot surfaces is not expected to be the same as that of aged or graphitized carbon surfaces. In other words, the rate of nascent soot oxidation by e.g., molecular oxygen, can be substantially different from what is described by the classical, empirical Nagle- Strickland Constable (NSC) equation [6], which was developed largely for graphite or carbon black oxidation. Early studies of soot oxidation by O 2 typically used aged soot or carbon black as the reactant. A kinetic theory for reaction of oxygen with a carbon surface was first proposed by Eyring and coworkers [7] to describe observations of graphite oxidation. Two distinct reaction sites were introduced to explain the maxima that were observed in the oxidation rate above 1400 K. The two reaction sites were also adopted in the development of the NSC expression, derived from measurements of bulk pyrographite oxidation by O 2 [6]. The NSC expression has since been extended to describe high temperature carbon oxidation and used in models of soot formation and oxidation (see, e.g., [8]). 89 The understanding of soot oxidation has improved since the NSC expression was first proposed. The progress has been summarized by Stanmore, et al [9] and Lall and Zachariah [10]. The oxidation rate of soot has been measured at temperatures above 1700 K by aerosolizing carbon black in a shock tube [11-14]. Shock tubes allow for the temperatures relevant to flames to be accessed but these studies could not utilize fresh nascent soot as the reactants. The kinetics of soot oxidation has also been studied in flame/dual flame environments or vitiated oxidizing flows in which the rates of soot oxidation may be followed at high temperatures [15-19]. However, flame gas usually contains a large number of reactive species (e.g., hydroxyl and other radicals), making the fundamental kinetic processes and mechanisms difficult to isolate and probe. The oxidation behavior of soot at lower temperatures has been examined ex-situ. For example, oxidation kinetics has been studied gravimetrically for carbon black, flame soot and diesel soot in immobilized beds [20-25]. These studies developed the knowledge of the effects of temperature, oxygen concentration and the size of the particle on its oxidation kinetics. However, the collection and separation methods involved in preparing fixed bed reactors may alter the composition and reactivity of otherwise nascent soot. More recent studies have used spectroscopic techniques to examine soot oxidation but problems related to immobilizing the soot sample remain [26,27]. Electron microscopy techniques such as transmission electron microscopy (TEM) and scanning electron microscopy (SEM) have allowed for the soot morphology and size to be examined. The oxidation behavior of flame and diesel soot below 1000 K has also been followed under TEM by observing the change in size and morphology [28-31]. The TEM method gives an unmatched resolution for measuring the size and morphology changes, but the technique requires that the 90 soot sample be immobilized on a TEM grid. During TEM imaging, the low pressure may cause the soot sample to undergo some devolatilization and the high energy electron beam can cause sample damage [32]. These difficulties obscure a reliable measurement of oxidation kinetics of nascent soot. Measurements are ideally made with soot in the original aerosolized, flame state. The Tandem Differential Mobility Analyzer (TDMA) method has been introduced and used by Zachariah and coworkers to examine the oxidation behavior of soot at the tip of a coflow diffusion flame and from a diesel engine [33-37]. The TDMA method is capable of generating monodisperse soot particles to better examine the effect of particle size on soot oxidation the aerosolized state. In the present work, the fundamental rate of nascent soot oxidation by molecular oxygen was examined at around 1000 K in an aerosol flow reactor. The various obstacles to measuring the oxidation kinetics of nascent soot were minimized by using fresh, freely suspended soot immediately extracted from flames as the reactant. Specifically, a particle sample was extracted continuously from premixed burner stabilized stagnation (BSS) flame and diluted immediately to reduce particle-particle coagulation. The particle sample was then introduced into the flow reactor. As reported previously, BSS flames are in a pseudo 1-D configuration which allows for nascent soot be probed under well defined boundary conditions [38]. The stagnation probe allows for the residence time of soot particles to be controlled in the flame, allowing for soot at the different growth stages to be examined. Similar to Zachariah and coworkers, the electrical mobility was measured in the present study to determine the particle size distributions before and after oxidation occurs. Emphasis was placed on minimizing the transit time from the point where the 91 soot was sampled to the flow reactor inlet. By using a short transit time and suspending the particles in an inert flow of cold nitrogen, the changes in the surface composition and morphology may be minimized. A main difference between our work and that of Zachariah and coworkers is that the particles we examine are younger and have not undergone carbonization which may occur at the tip of their coflow diffusion flames. 5.2. Experimental Methodologies A schematic summarizing the oxidation experiment is shown in Fig. 1. Nascent soot was directly extracted at a distance of 0.8 cm from the burner surface of a 15.1 % ethylene- 21.8 % oxygen- argon flame (  = 2.07, T f = 1850 K) using a probe sampling technique similar to previous characterization of PSDFs in BSS flames [38-40]. The flat flame burner is water cooled and the burner diameter was 5 cm. The mass flow rates of oxygen, argon and nitrogen were controlled by critical orifices. Nascent soot enters into the orifice and was diluted and chilled in a carrier flow of nitrogen with the dilution ratio of approximately 100. The aerosol was carried to the flow reactor within a transmission time of 0.1sec and mixed with a high-temperature stream of oxygen/nitrogen or nitrogen mixture. The high temperature stream is pre-heated to maintain an isothermal temperature profile in the flow reactor. The combined mixture enters into a tubular flow reactor which has 1.8 cm inner diameter. The residence time was 0.2 seconds in the flow reactor. In the present study, the diagnostic for nascent soot oxidation is the change in the PSDF, which was measured with a TSI Scanning Mobility Particle Sizer (Electrostatic Classifier 3085 and Condensation Particle Counter 3095). Sampling probes embedded at the inlet and the outlet of the aerosol reactor allowed for the PSDF to be measured before and after a given oxidation time. 92 Fig. 5.1 Schematic of the coupled burner and flow reactor setup. Nascent soot is extracted at 0.8 cm from the BSS burner surface. 5.3. Experimental Results Oxidation of nascent soot was observed at T= 950 and 1000 K for oxygen concentrations ranging from 1000 to 7800 ppm. The total flow rate in the test section of the flow reactor was kept at 330 cm 3 /s (STP) for every oxygen concentration. The results shown in Fig. 2 are a comparison of the initial PSDF (t = 0 sec) and the final PSDF (t = 0.2 sec) in an inter nitrogen flow. Under inert conditions, there is no observable change in the normalized PSDF over the 0.2 sec flow time, indicating that particle-particle coagulation or particle thermal decomposition was suppressed. The suppression of these complicating factors enables a direct interpretation of the change expected for the particle size due to oxidation. The PSDF of nascent soot from the ethylene flame tested gives a median diameter of 10.9 nm. At this size, the particles are expected to be primary particles in the early stage of growth. 1. BSS Flame and Soot Sampling Exhaust Sampling N 2 Sampling orifice O 2 /N 2 2. Pre-Heated N 2 /O 2 Sampling N 2 Exhaust t = 0 sec TC TC TC t = t 1 3. Flow reactor ID = 1.8 cm, L= 90 cm t ~ 0.2 sec ΔP Kr 85 Nano- DMA ultra-fine condensation optics Premixed fuel/O 2 /Ar Scanning Mobility Particle Sizer (SMPS) 93 The PSDF under oxidation conditions was measured as a function of the oxygen concentrations at 950 and 1000 K, as shown in Fig. 3. For each condition, the experiments were repeated at least three times to ensure the reproducibility of the experiment. The dilution ratio has not been calibrated thus the PSDF are reported on a normalized basis. Observable reduction in the particle diameter occurs for >1000 ppm O 2 for T=950 K. As expected the reduction in particle size increases with an increase in the O 2 concentration and temperature. Fig. 5.2 Measured PSDFs (symbols) of nascent soot from the ethylene BSS flame, probed in the flow reactor in an inert nitrogen flow at the initial (t = 0 sec, filled circles) and the final (t = 0.2 sec, open squares: 950 K; open triangles: 1000 K) residence times. 0.0 0.5 1.0 1.5 2.0 2.5 3.0 2 4 6 8 10 30 50 (dN/dlogD p )/N <D p > = 10.9 nm s = 1.5 Particle Diameter, D p (nm) 94 Fig. 5.3 Measured PSDFs (symbols) of soot from the ethylene BSS flame as a function of the oxygen concentration (t = 0.22 sec). Left panel: 950 K; right panel: 1000 K. The lines are normal distributions which are fitted to the measured PSDF. The top plots show the PSDFs in inert nitrogen. Lines are log- normal fits to data 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 10.9 nm,  = 1.5 0 ppm O 2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 9.5 nm,  = 1.5 1700 ppm O 2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 7.7 nm,  = 1.6 3800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 2 4 6 8 10 30 50 <D p > = 7.3 nm,  = 1.6 7800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 7.4 nm,  = 1.6 5500 Diameter, D p (nm) (dN/dlogD p )/N 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 10.8 nm,  = 1.5 0 ppm O 2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 7.5 nm,  = 1.6 2300 ppm O 2 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 5.4 nm,  = 1.6 4500 2 4 6 8 10 30 50 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 4.5 nm,  = 1.6 7800 0.0 0.5 1.0 1.5 2.0 2.5 3.0 <D p > = 5.1 nm,  = 1.6 5500 Diameter, D p (nm) (dN/dlogD p )/N 95 5.4 Simplified Surface Oxidation Model The oxidation rate of nascent soot can be extracted from rate of PSDF reduction as long as the effective surface area undergoing oxidation is defined. Molecular oxygen may attack the soot particle in three distinct regimes depending on the degree of penetration. If the diffusion of molecular oxygen into the particle is infinitely fast then all of the internal surface area of soot undergoes oxidation. The diffusion of oxygen may also be limited such that the inner area is only partially active. A third regime applies when all of the surface reactions occur on the outer surface. In the present study, the outer surface area of the spherical particle is the only oxidation site considered. The observed median diameter is about 11 nm which suggests that the PSDF is composed of primary soot particles and that the mean free path of the gas is larger than the particle size. The penetration of oxygen into the primary particle is assumed to be unimportant. For a spherical particle with diameter, D p , in the puedo first order limit with respect to the molecular oxygen concentration, the mass disappearance rate on the basis of outer surface area consumption for a specific oxygen concentration is described by:   2 p dm D dt   . If the mass density of soot is a constant ( ρ s = 1.5 g/cm 3 taken from Zhao et al. [2], we obtain The above equation shows that the time variation in D p is directly related to the surface oxidation rate of soot under this simplified analysis. 96 The measured oxidation rate of nascent soot is shown in Fig. 5.4 as a function of oxygen concentration for all of the conditions tested. For comparison, the classical prediction of the oxidation rate by the NSC equation is also shown. Clearly, the NSC equation under predicts the oxidation rate of nascent soot by more than one order of magnitude. The reactivity of nascent soot towards oxidation by molecular oxygen is significantly higher than of graphite or carbonized soot. Fig. 5.4 Measured specific oxidtation rate of nascent soot (symbols) compared to predictions by the NSC equation (solid lines). The dashed lines represent the NSC rates multiplied by a factor of 10. 0.002 0.006 0.010 10 -8 10 -7 10 -6 P O2 (atm) NSC Oxidation Rate,  (g/cm 2 s) 955 K 1000 K NSC x 10 97 5.5 Conclusions Oxidation of nascent, uncarbonized soot by molecular oxygen was studied. Nascent soot was generated in an ethylene burner-stabilized stagnation flame. The rate of change in the median particle diameter was measured to range from −20 to − 40 nm/s depending on the temperature and O 2 concentration. In terms of surface reaction, the measured rate ranges from 1 to 3×10 6 g/cm 2 s. The observed rate of oxidation is much larger than that predicted by the classical NSC equation which suggests that the surface of nascent soot is more reactive than a traditional graphite or carbonized soot surface. 98 5.6. Chapter 5 References [1] H. Wang, Proc. Comb. Inst. 33 (2011) 41-67. [2] B. Zhao, K. Uchikawa, H. Wang, Proc. Comb. Inst. 31 (2007) 851-860. [3] A. D Abid, N. Heinz, E. Tolmachoff, D.J. Phares, C. Campbell, H. Wang, Combust. Flame 154 (2008) 775-788. [4] J.P. Cain, P.L Gassman, H. Wang, A. Laskin, PCCP 12 (2010) 5206-5218. [5] J.P. Cain, J. Camacho, D.J. Phares, H. Wang, A. Laskin, Proc. Comb. Inst. 33 (2011) 533-540. [6] J.R. Walls, R.F Strickland-Constable, Carbon 1 (1964) 333-338. [7] G. Blyholder, J.S. Binford, H. Eyring, J. Phys. Chem. 62 (1958) 263-267. [8] M. Frenklach, H. Wang, Proc. Comb. Inst. 23 (1991) 1559-1566. [9] B.R. Stanmore, J.F. Brilhac, P. Gilot, Carbon 39 (2001) 2247-2268. [10] A.A. Lall, M.R. Zachariah, “Size Resolved Soot Oxidation Kinetics”, in: Combustion Generated Carbonaceous Particles H. Bockhorn, H., D’Anna, A. F., Wang, H. (Eds.) KIT Scientific Publishing, 2009. [11] C. Park, J.P. Appleton, Combust. Flame 20 (1973) 369-379. [12] O. Brandt, P. Roth, J. Aerosol. Sci. 19 (1988) 863-866. [13] O. Brandt, P. Roth, Combust. Flame 77 (1989) 69-78. [14] P. Cadman, R.J. Denning, J. Chem. Soc. Faraday Trans. 92 (1996) 4159-4165. [15] K.B. Lee, M.W. Thring, J.M. Beer, Combust. Flame 6 (1962) 137-145. [16] C.P. Fenimore, G.W. Jones, J. Phys. Chem. 71 (1967) 593-597. [17] A. Feugier, Combust. Flame 19 (1972) 249-256. [18] A. Garo, G. Prado, J. Lahaye, Combust. Flame 79 (1990) 226-233. [19] C.A. Echavarria, I.C. Jaramillo, A.F. Sarofim, J.S. Lighty, Proc. Comb. Inst. 33 (2011) 659-666. [20] Otto, K., Sieg, M. H., Zinbo, M., and Bartosiewicz, SAE Technol. Pap. Ser No. 800336 (1980) 99 [21] F.A. Ahlstrahm, C.U. Odenbrand, Carbon 27 (1989) 475-483. [22] P. Gilot, F. Bonnefoy, F. Marcuccilli, G. Prado, Combust. Flame 95 (1993) 87-100. [23] J.P.A. Neeft, T.X. Nijhuis, E. Smakman, M. Makkee, J.A. Moulijn, J. A., Fuel 76 (1997) 1129-1136. [24] J.M. Encinar, J.F. González, E. Sabio, J.J. Rodríguez, J. Chem. Technol. Biotech. 75 (2000) 213-222. [25] K.O. Lee, H. Seong, S.M. Choi, Proc. Comb. Inst.34 (2011) 3057-3065. [26] J. Schmid, B. Grob, R. Niessner, N.P. Ivleva, Anal. Chem. 83 (2011) 1173-1179. [27] M.E. Schuster, M. Havecker, R. Arrigo, R. Blume, M. Knauer, N.P. Ivleva, D.S. Su, R. Niessner, R.J. Schlogl, Phys. Chem. A 115 (2011) 2568-2580. [28] T. Ishiguro, Y. Takatori, K. Akihama, Combust. Flame 108 (1997) 231-234. [29] R.L. Vander Wal, A.J. Tomasek, Combust. Flame 134 (2003) 1-9. [30] R.L Vander Wal, C.J. Mueller, Energy & Fuels 20 (2006) 2364-2369. [31] K. Yehliu, R.L. Vander Wal, O. Armas, A.L. Boehman, Combust. Flame 159 (2012) 3597-3606. [32] S. Lieb, H. Wang, “Height and Phase Mode Images of Nascent Soot using AFM”, 8th US National Combustion Meeting, Park City, UT, May 19-22, 2013, paper 070EN-0354. [33] K.J. Higgins, H. Jung, D.B Kittelson, J.T. Roberts, M.R. Zachariah, Environ. Sci. Technol. 37 (2003) 1949-1954. [34] K.J. Higgins, H. Jung, D.B Kittelson, J.T. Roberts, M.R. Zachariah, J. Phys. Chem. A 106 (2001) 96-103. [35] S.H. Kim, R.A. Fletcher, M.R. Zachariah, Environ. Sci. Technol. 39 (2005) 4021-4026. [36] A.M. Nienow, J.T. Roberts, M.R. Zachariah, J. Phys. Chem.B 109 (2005) 5561-5568. [37] H. Jung, D.B Kittelson, M.R. Zachariah, Environ. Sci. Technol. 40 (2006) 4949-4955. [38] A.D Abid, J. Camacho, D.A. Sheen, H. Wang, Combust. Flame 156 (2009) 1862-1870. [39] A.D Abid, J. Camacho, D.A. Sheen, H. Wang, Energy & Fuels 23 (2009) 4286-4294. 100 [40] J. Camacho, S. Lieb, H. Wang, Proc. Comb. Inst. 34 (2013) 1853-1860. 101 Chapter 6: Catalytic Methane Oxidation by Freely Suspended Pd Nanoparticles 6.1 Introduction The enhancement of combustion by freely suspended nanoparticle catalyst may be greater than the catalytic activity in bulk palladium. In the current, catalytic activity is observed in a novel aerosol flow reactor design which isolates the surface catalysis. Catalytic methane conversion by supported palladium is used in specialized combustion systems and the physical properties and chemistry has been reviewed [1]. However, catalytic behavior in the nanoparticle regime may be influenced by the unique physical and chemical properties of nanoparticles. An overview of the potential factors affecting catalysis in the nanoparticle regime has been carried out by H. Wang [2]. A summary of these factors is shown in Fig. 6.1 in terms of the surface area density, particle size and catalytic activity. The first benefit of nanoparticles is a basic geometric advantage. For the same surface area, the diffusion of gas to the surface of bulk materials is more limited than diffusion into a particle. As Fig. 6.1 shows, the diffusion limitation becomes significant in nanoparticles that are about 100 nm in diameter. As the particles become smaller, the catalytic activity is expected to increase until the molecular scale reached. The catalytic behavior under this regime is unknown and Wang identified potential controlling factors. In the mid-range (3-20 nm), the energy accommodation of the particle during the gas- particle collision can both increase and decrease the catalytic activity. The energy accommodation depends on whether elastic or specular scattering occurs. The equilibria between Pd and Pd-O phases may be different at the nanoparticle phases and the change in bond energy may affect catalysis. In addition, a phase-transition from solid to liquid may occur at this size. At the molecular scale, the ionic effects and structural sensitivity may impact catalysis in an 102 unknown manner. In the current work, the size dependence of the catalytic activity in methane oxidation will be evaluated with these factors in mind. Fig. 6.1 Summary of potential catalytic activity based on energy accommodation, phase change and bond energy changes in the nanoparticle regime [2]. Methane oxidation in the presence of freely suspended nanoparticles has been previously reported. An in-situ process for Pd nanoparticles was introduced and detailed chemical characterization indicated that an equilibrium between Pd and PdO phases exist and this surface is correlated to methane oxidation [3]. Further observations allowed for a gas-surface kinetic model to be developed based on experimental measurements [4,5]. In the current work, the in- Mean free Path Kn = 1 Catalytic Activity Molecular /cluster regime Regime (starts to be) Limited by diffusion Nanocatalysis regime ? ? 103 situ synthesis method will be extended with an emphasis on the size dependent properties discussed above. 6.2 Experimental Methods A schematic describing the particle synthesis and catalysis test section is shown if Fig. 6.2. The in-situ generation of Pd nanoparticles ensures that the catalyst surface is relatively free from contamination and poisoning. In the current work, nanoparticle generation is improved by de- coupling synthesis from methane oxidation. The in-situ process involves a Pd precursor solution which evaporates and decomposes in the high temperature jet stirred reactor shown in Fig. 6.2. Palladium vapors form and subsequent coagulation causes nanoparticles to form over time. The Pd nanoparticle precursor is a solution consisting of a Pd organometallic salt dissolved in an organic solvent. Palladium Acetate ( Pd(Ac) 2 ) is dissolved in either acetone or acetaldehyde in the current work. Fig. 6.2 Schematic summarizing sequential particle synthesis and methane oxidation sections. 1a. Palladium Acetate- Acetaldehyde Solution carrier N 2 sampling orifice pre-heated air 1b. Jet Stirred Reactor sampling N 2 exhaust t = 0 sec T in t = t 1 2. Flow reactor ID = 1.7 cm, L= 80 cm t ~ 0.1 sec CH 4 Scanning Mobility Particle Sizer exhaust Diagnostics Infrared CO 2 / CH 4 Detector cooling N 2 / O 2 ultrasonic transducer sub- μm precursor fog 1c. Cooling Vessel 104 The current configuration isolates methane oxidation of the palladium surface by constraining competing processes. Particle synthesis must be complete before the tube flow reactor (TFR) such that the palladium precursor (acetone) must be completely decomposed to CO 2 to eliminate competing oxidation reactions in the TFR. Catalytic methane oxidation occurs between 800 K and 1000 K, thus any methane oxidation above this range is attributed to gas (non-catalytic) chemistry. The median particle diameter should range from D p = 15- 20 nm because catalytic activity has been observed for D p = 15 nm with 0.05 cm -3 surface density. The boundary condition of the TFR should be a perfectly stirred with an isothermal temp. In addition, the H 2 O production from the solvent should be minimized because H 2 O covers the Pd surface. A detailed view of the jet stirred synthesis reactor is shown in Fig. 6.2 along with subsequent mixing/cooling stages. The novel design introduced here allows for the surface-kinetic process of methane catalysis to be observed in isolation. The characterization of the catalytic activity is dependent upon the temperature and the surface area density of palladium. Isothermal conditions were achieved in the TFR as shown in Fig. 6.3 for the three conditions studied. The surface area is well characterized by probing the PSDF at the inlet and the outlet of the reactor. During sampling, the PSDF may be skewed by coagulation and diffusion losses. As shown in Fig. 6.4, the dilution was optimized such that the measured PSDF is not sensitive as the dilution extends beyond 5x. 105 Fig. 6.3 Detailed summary of the particle synthesis apparatus and sequential mixing steps. pre-heated air Pd(Ac) 2 - acetaldehyde - N 2 aerosol T JSR1 T JSR2 cooling vessel JSR TFR sampling methane cooling N 2 106 Fig. 6.4 Optimization of dilution during sampling of in-situ generated Pd nanoparticles. 6.3 Results and Discussion The catalytic activity during methane oxidation was observed at 3 temperatures for a fixed initial methane concentration and surface area density. The flow composition for this of experiments is summarized in Table 6.1. A summary of the observed catalytic activity is shown in Table 6.2. in terms of the surface reaction rate: x25 x10 x5 x15 x50 0.5 1.0 1.5 2.0 2.5 3.0 4 6 8 10 30 0.5 wt % Pd(Ace) 2 in Acetaldehyde Q solution = 1.5 mL/min Diameter, D p (nm) (dN/dlogD p )/N 107 where p is the pressure, R is the universal gas constant, T is the temperature, x CH4 is the final methane concentration, σ is the measured surface area density and t is the residence time. Table 6.1 Flow Compositions for [CH 4 ] o = 1 %. Total flow rate = 43 SLPM. % CH4 1.3 CO2 2.3 O2 8.1 H2O 2.3 N2 85.9 Table 6.2 Experimental Observation of Methane Oxidation Catalysis at [CH 4 ] o = 1%. T TFR (K) D p (nm) area (cm -1 ) Δ[CH 4 ] time (s) rate (mol/cm 2 /s) conversion (%) 825 9.5 7.9E-03 ± 5.6E-04 -0.048 ± 0.012 0.096 ± 0.005 2.6E-05 ± 7.3E-06 3.6 ± 0.9 900 9.5 5.5E-03 ± 7.3E-04 -0.056 ± 0.016 0.089 ± 0.005 4.7E-05 ± 1.6E-05 4.2 ± 1.2 975 9 4.7E-03 ± 2.9E-04 -0.015 ± 0.009 0.084 ± 0.005 1.5E-05 ± 9.5E-06 2.8 ± 1.7 As Table 6.2 shows, the methane conversion is about 4% for the given temperatures and initial methane concentration. By considering the particle surface area and residence time the surface reaction rate is calculated. The PSDF observed at the 3 temperatures , shown in Fig. 6.5, shows that the surface area and median particle size are comparable. The measured surface reaction rate is an order of magnitude faster than the rate reported for bulk surface methane catalysis. However, the relatively high signal to noise ratio creates an experimental uncertainty which renders the measure surface reaction rate unreliable. The synthesis step must be optimized further such that the uncertainty in the measure rate decreases. 108 Fig. 6.5 PSDF of freshly generated Pd nanoparticles at the inlet of the methane oxidation reactor. Catalytic activity was observed in the reactor for several initial methane concentrations and the summary is shown in Fig. 6.6. The conversion of methane stays constant as the initial methane concentration is increased which indicates that the surface area of the catalyst is in excess. This indicate that the surface reaction is pseudo first order with respect to methane. Fig. 6.6 Methane conversion as a function of the initial methane concentration at T= 825 K. 0 1 x 10 9 2 x 10 9 3 x 10 9 4 x 10 9 5 x 10 9 6 x 10 9 4 6 8 10 30 T TFR = 825 K  = 8x10 -3 (cm -1 ) Diameter, D p (nm) dN/dlogD p (cm -3 ) 0 1 x 10 9 2 x 10 9 3 x 10 9 4 x 10 9 5 x 10 9 6 x 10 9 4 6 8 10 30 T TFR = 900 K  = 6x10 -3 (cm -1 ) Diameter, D p (nm) 0 1 x 10 9 2 x 10 9 3 x 10 9 4 x 10 9 5 x 10 9 6 x 10 9 4 6 8 10 30 T TFR = 975 K  = 5x10 -3 (cm -1 ) Diameter, D p (nm) 1.0 3.0 5.0 0.5 1.0 1.5 2.0 2.5 3.0 Initial Methane Concentration, [ CH 4 ] o (%) Conversion, 1-[ CH 4 ] / [ CH 4 ] o (%) T TFR = 825 K  Pd = 5x10 -3 cm -1 109 The overall flow rate of the process was decreased to control the growth of newly synthesized palladium nanoparticles. By observing multiple particle sizes, the size dependent factors described above may be assessed. The PSDF is shown in Fig. 6.7 for the two particle sizes observed. The median particle size increased from 9 nm to 13 nm and the surface area density is comparable. The resulting catalytic activity is summarized in Fig. 6.8 for several reactor temperatures. Pure gas-phase ignition behavior is well known and Fig. 6.8 shows that the expected ignition temperature of methane is observed. The catalytic activity of the smaller particles appears to be greater than the larger particles. This observation may support the notion that a peak particle size exists where the competing processes are optimized. Again, the results are preliminary and experimental uncertainties must be minimized for more definitive conclusions. Fig. 6.7 PSDF of Pd nanoparticles synthesized at controlled residence times. 0.0 5.0 x 10 8 1.0 x 10 9 1.5 x 10 9 2.0 x 10 9 2.5 x 10 9 3.0 x 10 9 3.5 x 10 9 4.0 x 10 9 2 4 6 8 10 30 50 t growth = 600 ms  Pd = 5x10 -3 (cm -1 ) Diameter, D p (nm) dN/dlogD p (cm -3 ) T TFR = 825 K t growth = 900 ms  Pd = 6x10 -3 (cm -1 ) 110 Fig. 6.8 Catalytic conversion of methane as a function of temperature for 2 particle sizes. The gas-phase ignition region was also verified experimentally. 6.4 Conclusion The catalytic oxidation of methane in the presence of freely suspended Pd nanoparticles was observed in isolation of other competing processes. A novel in-situ particle synthesis occurs before the catalytic methane oxidation thus measure rate is more specific to methane. The measured rate of surface reaction rate on the nanoparticle catalyst is on the order of 10 -5 mol/cm 2 /s which is significantly faster than the surface reaction on bulk Pd surfaces. In addition, a slight dependence on the particle size was observed. However, the 4% methane conversion is to noisy for a reliable conclusion. 1.0 3.0 5.0 800 900 1000 1100 1200 1300 Reactor Temperature, T TFR (K) Conversion, 1-[ CH 4 ] / [ CH 4 ] o (%) [CH 4 ] o = 0.03 <D p > = 13.5 nm  TFR = 6x10 -3 cm -1 [CH 4 ] o = 0.01 <D p > = 9.5 nm  TFR = 5x10 -3 cm -1 gas-phase ignition region 111 6.5 Chapter 6 References [1] D. Ciuparu, M.R. Lyubovsky, E. Altman, L.D. Pfefferle,A. Datye, Catal. Rev. Sci. Eng. 44 (2002) 593. [2] J. Camacho, S. Nikraz, S. Lieb, H. Wang, “ Combustion Enhancement by Palladium Nanoparticles – Experiments and Modeling” 2012 MURI Review Meeting, Georgia Tech, Dec. 18-19. [3] B. Van Devener, S. L. Anderson, T. Shimizu, H. Wang, J. Nabity, J. Engel, J. Yu, D. Wickham, S. Williams, J. Phys. Chem C 113 (2009) 20632–20639. [4] T. Shimizu, A.D. Abid, G. Poskrebyshev, H. Wang, J. Nabity, J. Engel, J. Yu, D. Wickham, B. Van Devener, S.L. Anderson, S. Williams, Combust. Flame 157 (2010) 421-435. [5] T. Shimizu, H. Wang, Proc. Comb. Inst. 33 (2011) 1859-1866. 112 Chapter 7 Conclusion and Future Work 7.1 Concluding Remarks Fundamental chemical kinetic processes of reacting flow laden with suspended nanoparticles were examined with the focus on soot formation and nanoparticle catalysis. These heterogeneous reacting processes are rather complex and usually involve intricate interactions among various elementary processes. Novel heterogeneous flow reactor designs were introduced which isolate surface chemistry from other competing processes. The effect of parent fuel structure on particle inception and early development of nascent soot structure were examined in isolation from temperature and time effects. Butanol and C 6 hydrocarbon fuel structures were examined to establish the impact of fuel bound oxygen, branched hydrocarbon chains and double bonds on the detailed soot behavior of premixed flames. The measured PSDF gave insights into detailed sooting properties such as the relative nucleation time and median particle diameter. Under the same C/O ratio, butanol flames in fact nucleate soot earlier and gave greater soot volume fractions than the butane flames. In terms of fuel structure, the branched chain functionality has the most observable effect on soot formation. The onset of soot nucleation is faster in the branched fuels in comparison to the straight-chain counterparts. The faster nucleation rate also propagates into the mass growth stage. For the C 6 hydrocarbon flames, the overall sooting process was comparable in the fuels as evidenced by similar time resolved bimodal PSDF and particle morphology observations. However, the nucleation time and the persistence of nucleation with time were strongly dependent upon the structure of the parent fuel. The fastest onset of soot nucleation was observed in cyclohexane and benzene flames and this may be due to significant aromatic formation that is predicted in the pre-flame region. In 113 addition, the evolution of the PSDF showed that nucleation ends sooner in cylclohexane and benzene flames and this behavior is related to the relatively early depletion of precursors such as acetylene and benzene. A link between the detailed sooting properties and gas-phase chemistry was established by eludicating the role of soot precursors on experimental observations. Gas-kinetic reaction models were applied for the butanol and C 6 hydrocarbon flames. Analysis of the gas-phase chemistry for the butanol flames indicates that the fuel structure effect is largely exhibited in the relative importance of C 2 versus C 3 intermediate species formed during the initial stage of fuel breakdown. In the C 6 hydrocarbon flames, the C 6 dehydrogenation pathway to benzene formation was found to occur only in sooting cyclohexane flames and this may explain the relatively fast nucleation time of soot. Oxidation of nascent, uncarbonized soot by molecular oxygen was studied with a focus on maintaining the surface composition and reactivity. The rate of change in the median particle diameter was measured to range from −20 to − 40 nm/s depending on the temperature and O 2 concentration. In terms of surface reaction, the measured rate ranges from 1 to 3×10 6 g/cm 2 s. The observed rate of oxidation is much larger than that predicted by the classical NSC equation which suggests that the surface of nascent soot is more reactive than a traditional graphite or carbonized soot surface. Beyond Soot, the catalytic oxidation of methane in the presence of freely suspended Pd nanoparticles was observed in isolation of other competing processes. A novel in-situ particle synthesis occurs before the catalytic methane oxidation the thus measured oxidation rate is more specific to methane. The measured rate of surface reaction rate on the nanoparticle catalyst is on 114 the order of 10 -5 mol/cm 2 /s which is significantly faster than the surface reaction on bulk Pd surfaces. In addition, a slight dependence on the particle size was observed. However, the 4% methane conversion is too noisy for a reliable conclusion. 7.2 Future Work This work introduces methods that can be generally applied to other heterogeneous reacting flows or the continued development of the issues discussed. A continuation of soot formation work can be extended to new experimental observations of many soot formation processes. For example, the HACA mechanism can be validated further by applying a similar flow reactor setup with a high temperature acetylene environment. There are also many questions related to the fuel structure effect on detailed soot formation properties. In the current work, only one temperature was considered and it would be interesting to observe the fuel structure dependence on precursor fragmentation at higher temperatures. Nanocatalysis of freely suspended nanoparticles is still an undeveloped field. Even in the current work, conclusive experimental evidence of the effect of particle size on catalysis was not attained. Further development of the particle synthesis is required to achieve stability and a larger signal. In terms of new applications, the methods discussed are useful in other new nanoparticle synthesis methods and fundamental energy research. 115 Bibliography Abid, A. D., Camacho, J., Sheen, D. A., and Wang, H. Energy Fuels 2009, 23 (9) 4286. Abid, A. D., Camacho, J., Sheen, D. A., and Wang, H. Combust. Flame 2009, 156 (10) 1862. Abid, A. D., Heinz, N., Tolmachoff, E., Phares, D., Campbell, C., and Wang, H. Combust. Flame 2008, 154 (9) 775. Alfè, M., Apicella, B., Barbella, R., Rouzaud, J. N., Tregrossi, A., and Ciajolo, A. Proc. Combust. Inst. 2009, 32 (1) 443. Alfè, M., Apicella, B., Barbella, R., Tregrossi, A., and Ciajolo, A. Proc. Combust. Inst. 2007, 31 (1) 585. Appel, J., Bockhorn, H., and Frenklach, M. Combust. Flame 2000, 121 (1) 122. Blyholder, G., Binford, J. S., and Eyring, H. J. Phys. Chem.1958, 62 (3) 263. Bone, W. A., Flame and Combustion in Gases. 1st ed.: Longmans, Green and Co: London, 1927. Brandt, O., Rajathurai, A. M., and Roth, P. Experim. Fluids 1987, 5 (2) 86. Brandt, O., and Roth, P. J. Aerosol. Sci. 1988, 19 (7) 863. Brandt, O., and Roth, P. Combust. Flame 1989, 77 (1) 69. Cadman, P., and Denning, R. J. J. Chem. Soc. Farad. Trans. 1996, 92 (21) 4159. Cain, J. P., Camacho, J., Phares, D. J., Wang, H., and Laskin, A. Proc. Combust. Inst. 2011, 33 (1) 533. Cain, J. P., Gassman, P. L., Wang, H., and Laskin, A. Phys. Chem. Chem. Phys. 2010, 12 (20) 5206. Camacho, J., Lieb, S., and Wang, H. Proc. Combust. Inst. 2013, 34 (1) 1853. 116 Ciajolo, A., Tregrossi, A., Mallardo, M., Faravelli, T., and Ranzi, E. Proc. Combust. Inst. 2009, 32 (1) 585. Ciuparu, D., Lyubovsky, M. R., Altman, E., Pfefferle, L. D., and Datye, A. Catalysis Rev. 2002, 44 (4) 593. Colket, M., Edwards, J. T., Williams, S., Cernansky, N. P., Miller, D. L., Egolfopoulos, F. N., Lindstedt, P., Seshadri, K., Dryer, F. L., Law, C. K., Friend, D. G., Lenhert, D. B., Pitsch, H., Sarofim, A. F., Smooke, M. D., and Tsang, Development of an Experimental Database and Kinetic Models for Surrogate Jet Fuels, AIAA Paper 2007-0770, 2007. Dean, A. M. J. Phys. Chem. 1985, 89 (21) 4600. Devener, B. V., Anderson, S. L., Shimizu, T., Wang, H., Nabity, J., Engel, J., Yu, J., Wickham, D., and Williams, S. J. Phys. Chem C 2009, 113 (48) 20632. Echavarria, C. A., Jaramillo, I. C., Sarofim, A. F., and Lighty, J. S. Proc. Combust. Inst. 2011, 33 (1) 659. Echavarria, C. A., Sarofim, A. F., Lighty, J. S., and D'Anna, A. Proc. Combust. Inst. 2009, 32 (1) 705. Echavarria, C. A., Sarofim, A. F., Lighty, J. S., and D'Anna, A. Combust. Flame 2010, 158 (1) 98. Encinar, J. M., González, J. F., Sabio, E., and Rodríguez, J. J. J. Chem. Tech. Biotech. 2000, 75 (3) 213. Ergut, A., Levendis, Y. A., Richter, H., Howard, J. B., and Carlson, J. Combust. Flame 2007, 151 (1-2) 173. Fenimore, C. P., and Jones, G. W. J. Phys. Chem.1967, 71 (3) 593. 117 Fennell, P. S., Dennis, J. S., and Hayhurst, A. Combust. Flame 2007, 151 (4) 560. Feugier, A. Combust. Flame 1972, 19 (2) 249. Fredrik AhlstrÃ¶m, A., and Ingemar Odenbrand, C. U. Carbon 1989, 27 (3) 475. Frenklach, M. Phys. Chem. Chem. Phys. 2002, 4 (11) 2028. Frenklach, M., Taki, S., Durgaprasad, M. B., and Matula, R. A. Combust. Flame 1983, 54 (1-3) 81. Frenklach, M., and Wang, H. Proc. Combust. Inst. 1991, 23 1559. Frenklach, M., and Warnatz, J. Combust. Sci. Technol. 1987, 51 (4-6) 265. Garo, A., Prado, G., and Lahaye, J. Combust. Flame 1990, 79 (3â€“4) 226 . Gilot, P., Bonnefoy, F., Marcuccilli, F., and Prado, G. Combust. Flame 1993, 95 (1) 87. Glassman, I. Proc. Combust. Inst. 1988, 22 295. Glassman, I. Combustion, 1st Ed.; Academic Press: San Diego, 1996. Grana, R., Frassoldati, A., Faravelli, T., Niemann, U., Ranzi, E., Seiser, R., Cattolica, R., and Seshadri, K. Combust. Flame 2011, 157 (11) 2137. Gu, X. L., Huang, Z. H., Wu, S., and Li, Q. Q. Combust. Flame 2011, 157 (12) 2318. Hansen, N., Kasper, T., Yang, B., Cool, T. A., Li, W., Westmoreland, P. R., Oßwald, P., and Kohse-Höinghaus, K. Proc. Combust. Inst. 2011, 33 (1) 585. Hansen, N., Li, W., Law, M. E., Kasper, T., Westmoreland, P. R., Yang, B., Cool, T. A., and Lucassen, A. Phys. Chem. Chem. Phys. 2011, 12 (38) 12112. Harper, M. R., Van Geem, K. M., Pyl, S. P., Marin, G. B., and Green, W. H. Combust. Flame 2011, 158 (1) 16. Haynes, B. S., and Wagner, H. G. Prog. Energy Combust. Sci. 1981, 7 (4) 229. 118 Higgins, K. J., Jung, H., Kittelson, D. B., Roberts, J. T., and Zachariah, M. R. J. Phys. Chem. A 2001, 106 (1) 96. Higgins, K. J., Jung, H., Kittelson, D. B., Roberts, J. T., and Zachariah, M. R. Environ. Sci. Technol. 2003, 37 (9) 1949. Ishiguro, T., Takatori, Y., and Akihama, K. Combust. Flame 1997, 108 (1-2) 231. Jung, H., Kittelson, D. B., and Zachariah, M. R. Environ. Sci. Technol. 2006, 40 (16) 4949. Kazakov, A., Wang, H., and Frenklach, M. J. Phys. Chem. 1994, 98 (41) 10598. Kee, R. J., Miller, J. A., Evans, G. H., and Dixon-Lewis, G. Symp. (Int.) Combus. 1989, 22 (1) 1479. Kim, C. H., Xu, F., and Faeth, G. M. Combust. Flame 2008, 152 (3) 301. Kim, S. H., Fletcher, R. A., and Zachariah, M. R. Environ. Sci. Technol. 2005, 39 (11) 4021. Lall, A. A., and Zachariah, M. R. in Combustion Generated Fine Carbonaceous Particles, Bockhorn, H., D’Anna, Sarofim, A.F., and Wang, H. (Eds.); KIT Scientific Publishing, Karlsruhe, 980. Law, M. E., Westmoreland, P. R., Cool, T. A., Wang, J., Hansen, N., Taatjes, C. A., and Kasper, T. Proc. Combust. Inst. 2007, 31 (1) 565. Lee, K. B., Thring, M. W., and Beer, J. M. Combust. Flame 1962, 6 (3) 137. Lee, K. O., Seong, H., and Choi, S. M. Proc. Combust. Inst. 34 (2) 3057. Li, Z. G., and Wang, H. Phys. Rev. E 2003, 68 (6) 061206. Li, Z. G., and Wang, H. Phys. Rev. E 2003, 68 (6) 061207. Li, Z. G., and Wang, H. Phys. Rev. E 2004, 70 (2) 021205. 119 Li, Z. G., and Wang, H. Phys. Rev. E 2004, 70 (6) 021206. Li, Z. G., and Wang, H. Phys. Rev. Letters 2005, 95 (1) 14502. Lieb, S., and Wang, H., Height and Phase Mode Images of Nascent Soot Using AFM, US National Meeting of the Combustion Institute, 2013, 070EN-0354 Lignell, D. O., Chen, J. H., Smith, P. J., Lu, T., and Law, C. K. Combust. Flame 2007, 151 (1-2) 2. Lutz, A. E., Kee, R. J., Grcar, J., and Rupley, F. M. 1996, OPPDIF: A FORTRAN Program for Computing Opposed-Flow Diffusion Flames, 1996. Manzello, S. L., Lenhert, D. B., Yozgatligil, A., Donovan, M. T., Mulholland, G. W., Zachariah, M. R., and Tsang, W. Proc. Combust. Inst. 2007, 31 675. Maricq, M. M. Combust. Flame 2004, 137 (3) 340. Maricq, M. M. Combust. Flame 2006, 144 (4) 730. Maricq, M. M., Harris, S. J., and Szente, J. J. Combust. Flame 2003, 132 (3) 328. McEnally, C. S., Lisa, D. P., Burak, A., and Kohse-Hoinghaus, K. Prog. Energy Combust. Sci. 2006, 32 (3) 247. McEnally, C. S., and Pfefferle, L. D. Combust. Flame 2004, 136 (1-2) 155. McEnally, C. S., and Pfefferle, L. D. Combust. Flame 2005, 143 (3) 246. Mitchell, P., and Frenklach, M. Symp. (Int.) Combust. 1998, 27 (1) 1507. Neeft, J. P. A., Nijhuis, T. X., Smakman, E., Makkee, M., and Moulijn, J. A. Fuel 1997, 76 (12) 1129. Nienow, A. M., Roberts, J. T., and Zachariah, M. R. J. Phys. Chem. B 2005, 109 (12) 5561. 120 Noorani, K. E., Akih-Kumgeh, B., and Bergthorson, J. M. Energy Fuels 2011, 24 5834- 5843. Oktem, B., Tolocka, M. P., Zhao, B., Wang, H., and Johnston, M. V. Combust. Flame 2005, 142 (4) 364. Osswald, P., Guldenberg, H., Kohse-Hoinghaus, K., Yang, B., Yuan, T., and Qi, F. Combust. Flame 2011, 158 (1) 2. Otto, K., Sieg, M. H., Zinbo, M., and Bartosiewicz SAE Technol. Pap. Ser 1980, No. 800336 Park, C., and Appleton, J. P. Combust. Flame 1973, 20 (3) 369. Russo, C., AlfÃ¨, M., Rouzaud, J.-N. l., Stanzione, F., Tregrossi, A., and Ciajolo, A. Proc. Combust. Inst. 2013 34 (1) 1885. Russo, C., Stanzione, F., Ciajolo, A., and Tregrossi, A. Proc. Combust. Inst. 2013 34 (2) 3661. Russo, C., Stanzione, F., Tregrossi, A., AlfÃ¨, M., and Ciajolo, A. Combust. Flame 2012, 159 (7) 2233. Santamaria, A., Yang, N., Eddings, E., and Mondragon, F. Combust. Flame 2010, 157 (1) 33-42. Schmid, J., Grob, B., Niessner, R., and Ivleva, N. P. Analytical Chem. 2011, 83 (4) 1173. Schuster, M. E., Havecker, M., Arrigo, R., Blume, R., Knauer, M., Ivleva, N. P., Su, D. S., Niessner, R., and Schlogl, R. J. Phys. Chem. A 2011, 115 (12) 2568. Sgro, L. A., Barone, A. C., Commodo, M., D'Alessio, A., De Filippo, A., Lanzuolo, G., and Minutolo, P. Proc. Combust. Inst. 2009, In Press, Corrected Proof 121 Sgro, L. A., Borghese, A., Speranza, L., Barone, A. C., Minutolo, P., Bruno, A., D'Anna, A., and D'Alessio, A. Environ. Sci. Technol. 2008, 42 (3) 859. Sgro, L. A., De Filippo, A., Lanzuolo, G., and D'Alessio, A. Proc. Combust. Inst. 2007, 31 631. Singh, J., Patterson, R. I. A., Kraft, M., and Wang, H. Combust. Flame 2006, 145 (1-2) 117. Sirjean, B., Dames, E., Sheen, D. A., You, X., Sung, C. J., Holley, A. T., Egolfopoulos, F. N., Wang, H., Vasu, S. S., Davidson, D. F., Hanson, R. K., Pitsch, H., Bowman, C. T., Kelley, A., Law, C. K., Tsang, W., Cernansky, N. P., Miller, D., Violi, A., and Lindstedt, R. P. 2010, A high-temperature chemical kinetic model of n-alkane (up to n-dodecane), cyclohexane, and methyl-, ethyl-, n-propyl and n- butyl-cyclohexane oxidation at high temperatures, JetSurF version 2.0, September 19, 2010 (http://melchior.usc.edu/JetSurF/JetSurF2.0). Stanmore, B. R., Brilhac, J. F., and Gilot, P. Carbon 2001, 39 (15) 2247. Stein, S. E., and Fahr, A. J. Phys. Chem. 1985, 89 (17) 3714. Street, J. C., and Thomas, A. Fuel 1955, 34 4. Takahashi, F., and Glassman, I. Combust. Sci. Technol. 1984, 37 (1-2) 1. TogbeÌ� , C., MzeÌ� -Ahmed, A., and Dagaut, P. Energy & Fuels 2011, 24 (9) 5244. Van Geem, K. M., Pyl, S. P., Marin, G. B., Harper, M. R., and Green, W. H. Ind. Eng. Chem. Res. 2010, 49 (21) 10399. Vander Wal, R. L., and Mueller, C. J. Energy & Fuels 2006, 20 (6) 2364. Vander Wal, R. L., and Tomasek, A. J. Combust. Flame 2003, 134 (1â€“2) 1 . Velasquez, M., Mondragon, F., and Santamaria, A. Fuel 2013, 104 (0) 681. 122 Veloo, P. S., and Egolfopoulos, F. N. Proc. Combust. Inst. 2011, 33 (1) 987. Veloo, P. S., Wang, Y. L., Egolfopoulos, F. N., and Westbrook, C. K. Combust. Flame 2011, 157 (10) 1989. Veloo, P. S. W., Y. L; Egolfopoulos, F. N 2010 Spring Technical Meeting, Western States Section of the Combustion Institute 2010, 10S-19. Violi, A., Kubota, A., Truong, T. N., Pitz, W. J., Westbrook, C. K., and Sarofim, A. F. Proc. Combust. Inst. 2002, 29 2343. Vranckx, S., Lee, C., Chakravarty, H. K., and Fernandes, R. X. Proc. Combust. Inst. 2013 34 (1) 377. Walls, J. R., and Strickland-Constable, R. F. Carbon 1964, 1 (3) 333. Wang, H. Proc. Combust. Inst. 2011, 33 (1) 41. Wang, H., and Frenklach, M. Combust. Flame 1991, 87 (3-4) 365. Wang, H., and Frenklach, M. J. Phys. Chem. 1993, 97 (15) 3867. Wang, H., and Frenklach, M. Combust. Flame 1994, 96 (1-2) 163. Wang, H., and Frenklach, M. Combust. Flame 1997, 110 (1-2) 173. Wang, H., You, X., Joshi, A. V., Davis, S. G., Laskin, A., Egolfopoulos, F. N., and Law, C. K. 2007, USC Mech Version II. High-Temperature Combustion Reaction Model of H2/CO/C1-C4 Compounds. 2002. Wang, H., Zhao, B., Wyslouzil, B., and Streletzky, K. Proc. Combust. Inst. 2003, 29 2749. Wang, S. C., and Flagan, R. C. Aerosol Sci. Technol. 1990, 13 (2) 230-240. Wang, Z., Cheng, Z., Yuan, W., Cai, J., Zhang, L., Zhang, F., Qi, F., and Wang, J. Combust. Flame 2013 159 (7) 2243. 123 Westmoreland, P. R., Dean, A. M., Howard, J. B., and Longwell, J. P. J. Phys. Chem. 1989, 93 (25) 8171. Yang, B., Li, Y., Wei, L., Huang, C., Wang, J., Tian, Z., Yang, R., Sheng, L., Zhang, Y., and Qi, F. Proc. Combust. Inst. 2007, 31 (1) 555. Yehliu, K., Vander Wal, R. L., Armas, O., and Boehman, A. L. Combust. Flame 2012, 159 (12) 3597. Zhao, B., Uchikawa, K., and Wang, H. Proc. Combust. Inst. 2007, 31 851. Zhao, B., Yang, Z. W., Johnston, M. V., Wang, H., Wexler, A. S., Balthasar, M., and Kraft, M. Combust. Flame 2003, 133 (1-2) 173. Zhao, B., Yang, Z. W., Li, Z. G., Johnston, M. V., and Wang, H. Proc. Combust. Inst. 2005, 30 1441. Zhao, B., Yang, Z. W., Wang, J. J., Johnston, M. V., and Wang, H. Aerosol Sci. Technol. 2003, 37 (8) 611.

Abstract (if available)

Abstract The development of novel experimental approaches to investigate fundamental surface kinetics is presented. Specifically, fundamental soot formation and surface catalysis processes are examined in isolation from other competing processes. In terms of soot formation, two experimental techniques are presented: the Burner Stabilized Stagnation (BSS) flame configuration is extended to isolate the effect of the parent fuel structure on soot formation and the fundamental rate of surface oxidation for nascent soot is measured in a novel aerosol flow reactor. In terms of nanoparticles, the physical and chemical properties of freely suspended nanoparticles are investigated in a novel aerosol flow reactor for methane oxidation catalyzed by palladium. ❧ The role of parent fuel structure within soot formation is examined by following the time resolved formation nascent soot from the onset of nucleation to later growth stages for premixed BSS flames. Specifically, the evolution of the detailed particle size distribution function (PSDF) is compared for butanol, butane and C₆ hydrocarbons in two separate studies where the C/O ratio and temperature are fixed. Under this constraint, the overall sooting process were comparable as evidenced by similar time resolved bimodal PSDF. However, the nucleation time and the persistence of nucleation with time is strongly dependent upon the structure of the parent fuel. For the C₆ hydrocarbon fuels, the fastest onset of soot nucleation is observed in cyclohexane and benzene flames and this may be due to significant aromatic formation that is predicted in the pre-flame region. In addition, the evolution of the PSDF shows that nucleation ends sooner in cylclohexane and benzene flames and this may be due to relatively quick depletion of soot precursors such as acetylene and benzene. Interestingly, within the butanol fuels studied the effect of the branched chain in i-butanol and i-butane was more significant than the presence of fuel bound oxygen. A numerical analysis of the gas-phase chemistry for butanol and butane indicates the fuel structure effect is largely exhibited in the relative importance of C₂ versus C₃ intermediate species formed during the initial stage of fuel breakdown. ❧ Oxidation kinetics of soot are typically measured with carbon black or well aged soot as substrates. The soot surface is also assumed to be graphitic in theoretical soot oxidation rate calculations. However, recent experimental and theoretical studies show that nascent soot can have structures and surface composition drastically different from mature, graphitized soot. In the current study, oxidation of nascent soot by O₂ was observed at T= 950 and 1000K for oxygen concentrations ranging from 1000 to 7800 ppm in a laminar aerosol flow reactor at ambient pressure. Oxidation behavior of primary particles (Dp < 20 nm) of nascent soot from a premixed BSS ethylene flame was observed by tracking the shift in the particle size distribution function (PSDF) under a given residence time. The measured rate of the surface reaction ranges from 1x10⁶-3x10⁶ g/cm²s for nascent soot. The rate of oxidation observed at the given conditions is an order of magnitude faster than predicted by the classical Nagle Strickland-Constable (NSC) correlations derived from graphite oxidation. Heterogeneous surface reaction rates are highly sensitive to the surface composition. Thus the faster rate of surface reaction by the nascent soot observed currently suggests that the surface composition of nascent soot is more reactive than the conventional graphite surface. ❧ Catalytic activity in reacting flow laden with suspended nanoparticle catalyst is measured in a novel aerosol flow reactor. Similar to conventional gas phase kinetics, heterogeneous reactions are the product of collisions between the particle surface and surrounding gas. However, particles below 10 nm in diameter are in a transition region where collisions do not always result in perfectly elastic scattering. The inelastic scattering provides more opportunities for reaction to occur than elastic scattering. It is this extra chemical behavior of nanoparticles which may serve as a novel parameter for tuning and optimizing catalytic activity. Specifically, the size dependence of catalytic activity is examined by observing methane oxidation catalyzed by freely suspended palladium nanoparticles. The role of the nanoparticle size is explored in flow reactor measurements and a novel in-situ process for rapid synthesis of aerosolized palladium nanoparticles is introduced. Under catalytic conditions, the temperature at which methane is oxidized is found to be 300K lower than conventional gas phase combustion. In addition, the preliminary measurements indicate that catalytic activity may be greater in nanoparticles relative to bulk surfaces.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Catalytic methane ignition over freely-suspended palladium nanoparticles

PDF

Kinetic modeling of high-temperature oxidation and pyrolysis of one-ringed aromatic and alkane compounds

PDF

Development of a novel heterogeneous kinetics reactor and chemical characterization of soot particles

PDF

Experimental investigation of nascent soot physical properties and the influence on particle morphology and growth

PDF

Studies of combustion characteristics of heavy hydrocarbons in simple and complex flows

PDF

Experimental investigation of soot nucleation and growth in premixed flat flames

PDF

Experimental and kinetic modeling studies of flames of H₂, CO, and C₁-C₄ hydrocarbons

PDF

A flow-through membrane reactor for destruction of a chemical warfare simulant

PDF

Studies of siloxane decomposition in biomethane combustion

PDF

Studies of methane counterflow flames at low pressures

PDF

Plasmonic enhancement of catalysis and solar energy conversion

PDF

Applications of continuous flow reactors towards screening catalytically active nanoparticle syntheses

PDF

Experimental studies of high pressure combustion using spherically expanding flames

PDF

Development of novel techniques for evaluating physical, chemical and toxicological properties of particulate matter in ambient air

PDF

Wake modes of rotationally oscillating circular cylinder in cross-flow and its relationship with heat transfer

PDF

Novel methods for landfill gas and biogas clean-up

PDF

Design and characterization of metal and semiconducting nanostructures and nanodevices

PDF

Electrokinetic transport of Cr(VI) and integration with zero-valent iron nanoparticle and microbial fuel cell technologies for aquifer remediation

PDF

Nanomaterials under extreme environments: a study of structural and dynamic properties using reactive molecular dynamics simulations

PDF

Fabrication, deposition, and characterization of size-selected metal nanoclusters with a magnetron sputtering gas aggregation source

Asset Metadata

Creator Camacho, Joaquin (author)

Core Title Development of a novel heterogeneous flow reactor: soot formation and nanoparticle catalysis

School Viterbi School of Engineering

Degree Doctor of Philosophy

Degree Program Mechanical Engineering

Publication Date 11/20/2013

Defense Date 10/04/2013

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag catalysis,combustion,Energy,nanoparticles,OAI-PMH Harvest,soot

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Wang, Hai (committee chair), Campbell, Charles S. (committee member), Egolfopoulos, Fokion N. (committee member), Eliasson, Veronica (committee member), Henry, Ronald (committee member)

Creator Email camachojoaquin57@gmail.com,jocamach@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-349397

Unique identifier UC11296505

Identifier etd-CamachoJoa-2168.pdf (filename),usctheses-c3-349397 (legacy record id)

Legacy Identifier etd-CamachoJoa-2168.pdf

Dmrecord 349397

Document Type Dissertation

Format application/pdf (imt)

Rights Camacho, Joaquin

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA