SHOCK TUBE MEASUREMENTS OF OXYGENATED FUEL COMBUSTION USING LASER ABSORPTION SPECTROSCOPY

Transcription

1 SHOCK TUBE MEASUREMENTS OF OXYGENATED FUEL COMBUSTION USING LASER ABSORPTION SPECTROSCOPY A DISSERTATION SUBMITTED TO THE DEPARTMENT OF MECHANICAL ENGINEERING AND THE COMMITTEE ON GRADUATE STUDIES OF STANFORD UNIVERSITY IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY King Yiu Lam June 2013

2 2013 by King Yiu Lam. All Rights Reserved. Re-distributed by Stanford University under license with the author. This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. This dissertation is online at: ii

3 I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. Ronald Hanson, Primary Adviser I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. Craig Bowman I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy. David Davidson Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost Graduate Education This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives. iii

4 iv

5 Abstract In the current engine development, fuel reformulation is considered as one of the potential strategies to improve fuel efficiency, reduce petroleum consumption, and minimize pollutant formation. Oxygenated fuels can be used as neat fuels or additives in spark-ignition and diesel engines to allow for more complete combustion. To understand the influence of oxygenated fuels on engine performance, accurate comprehensive kinetic mechanisms, which can consist of hundreds to thousands of elementary reactions, are needed to describe the chemistry of the combustion events, such as autoignition and pollutant formation. The primary objective of the research presented in this dissertation is to provide reliable experimental kinetic targets, such as ignition delay times, species time histories, and direct reaction rate constant measurements, using shock tube and laser absorption techniques in order to evaluate and refine the existing kinetic mechanisms for two different types of oxygenated fuels (i.e., ketones and methyl esters) and to reexamine the kinetics of the H 2 + OH reaction. The topics of this work are mainly divided into three sections: (1) H 2 + OH kinetics, (2) ketone combustion chemistry, and (3) methyl ester + OH kinetics. The reaction of OH with molecular hydrogen (H 2 ) H 2 + OH H 2 O + H (1) is an important chain-propagating reaction in all combustion systems, particularly in hydrogen combustion, and its direct rate constant measurements are discussed in the first part of this dissertation. The rate constant for reaction (1) was measured behind reflected shock waves over the temperature range of K at pressures of atm. OH radicals were produced by rapid thermal decomposition of tert-butyl hydroperoxide v

6 (TBHP) at high temperatures, and were monitored using the narrow-linewidth ring dye laser absorption of the well-characterized R 1 (5) line in the OH A X (0, 0) band near nm. Consequently, this work aims to report the rate constant for reaction (1) with a much lower experimental scatter and overall uncertainty (as compared to the data available in the literature). Ketones are important to a variety of modern combustion processes. They are widely used as fuel tracers in planar laser-induced fluorescence (PLIF) imaging of combustion processes due to their physical similarity to gasoline surrogate components. Additionally, they are often formed as intermediate products during oxidation of large oxygenated fuels, such as alcohols and methyl esters. In the second part of this dissertation, the combustion characteristics of acetone (CH 3 COCH 3 ), 2-butanone (C 2 H 5 COCH 3 ), and 3-pentanone (C 2 H 5 COC 2 H 5 ) are discussed in the context of the reflected shock wave experiments. These experiments were performed using different laser absorption methods to monitor species concentration time histories (i.e., ketones, CH 3, CO, C 2 H 4, CH 4, OH, and H 2 O) over the temperature range of K at pressures near 1.6 atm. These speciation data were then compared with the simulations from the detailed mechanisms of Pichon et al. (2009) and Serinyel et al. (2010). Consequently, the overall rate constants for the thermal decomposition reactions of acetone, 2-butanone, and 3-pentanone CH 3 COCH 3 (+ M) CH 3 + CH 3 CO (+ M) (2) C 2 H 5 COCH 3 (+ M) Products (+ M) (3) C 2 H 5 COC 2 H 5 (+ M) Products (+ M) (4) were inferred by matching the species profiles with the simulations from the detailed mechanisms at pressures near 1.6 atm. In addition, an O-atom balance analysis from the speciation data revealed the absence of a methyl ketene removal pathway in the original models. Furthermore, the overall rate constants for the reactions of OH with a series of ketones CH 3 COCH 3 + OH CH 3 COCH 2 + H 2 O (5) C 2 H 5 COCH 3 + OH Products (6) C 2 H 5 COC 2 H 5 + OH Products (7) C 3 H 7 COCH 3 + OH Products (8) vi

7 were determined using UV laser absorption of OH over the temperature range of K at pressures of 1-2 atm. These measurements included the first direct hightemperature measurements of the overall rate constants for reactions (6)-(8), and were compared with the theoretical calculations from Zhou et al. (2011) and the estimates using the structure-activity relationship (SAR) (1995). Biodiesel, which consists of fatty acid methyl esters (FAMES), is a promising alternative to fossil fuels. The four simplest methyl esters include methyl formate (CH 3 OCHO), methyl acetate (CH 3 OC(O)CH 3 ), methyl propanoate (CH 3 OC(O)C 2 H 5 ), and methyl butanoate (CH 3 OC(O)C 3 H 7 ), and their combustion chemistry is a building block for the chemistry of large methyl esters. In the third part of this dissertation, the rate constant measurements for the reactions of OH with four small methyl esters are discussed: CH 3 OCHO + OH Products (9) CH 3 OC(O)CH 3 + OH Products (10) CH 3 OC(O)C 2 H 5 + OH Products (11) CH 3 OC(O)C 3 H 7 + OH Products (12) These reactions were studied behind reflected shock waves using UV laser absorption of OH over K at pressures near 1.5 atm. This study presented the first direct hightemperature rate constant measurements of reactions (9)-(12). These measurements were also compared with the estimated values from different detailed mechanisms and from the structure-activity relationship (SAR). vii

8 viii

9 Acknowledgements First, I would like to thank my advisor, Prof. Ronald Hanson, for the opportunities and support he has offered throughout my graduate studies at Stanford. His critical thinking, carefulness, and wisdom have shaped me into a better and more careful researcher. I would like to thank Dr. David Davidson for offering numerous advice and guidance throughout my Ph.D. career. His willingness to help students at the lab has made my time at Stanford much smoother. I am also thankful to Prof. Bowman for serving on my qualification exam committee and my reading committee. I have been very fortunate to work with many outstanding students in the Hanson group. In particular, I am immensely grateful to Zekai Hong, Genny Pang, and Robert Cook for teaching me how to operate shock tubes and different laser equipment, allowing me to accomplish the work presented in this dissertation. I am very grateful to Wei Ren for his friendship and the collaborative efforts we have made together in research and course works. I am also grateful to many students in the Hanson group who have made my life at Stanford more meaningful and joyful. Finally, I am sincerely grateful to my parents for their support and encouragement in times of trouble and frustration. This work was supported by the U.S. Department of Energy, Basic Energy Sciences (DE-FG02-88ER13857) with Dr. Wade Sisk as program manager. ix

10 x

11 Table of Contents Abstract... v Acknowledgements... ix Table of Contents... xi List of Tables... xv List of Figures... xvii Chapter 1 Background and Motivation Introduction Background and Motivation H 2 + OH Kinetics Ketone Combustion Chemistry Methyl Ester + OH Kinetics Scope and Organization of Thesis...8 Chapter 2 Experimental Methods Shock Tube Facility Laser Absorption Methods UV Laser Absorption of OH UV Laser Absorption of Ketones UV Laser Absorption of CH IR Laser Absorption of CO CO 2 Laser Absorption of C 2 H IR Laser Absorption of H 2 O xi

12 2.2.7 IR Laser Absorption of CH Summary...20 Chapter 3 A Shock Tube Study of H 2 + OH H 2 O + H using OH Laser Absorption Introduction Experimental Details Kinetic Measurements Kinetic Mechanism Description H 2 + OH Kinetics Comparison with Earlier Work Summary...32 Chapter 4 Multi-Species Time History Measurements during High-Temperature Acetone and 2-Butanone Pyrolysis Introduction Experimental Details Mixture Preparation Species Absorption Coefficient Evaluations Results and Discussion Acetone Pyrolysis Butanone Pyrolysis Summary...53 Chapter 5 Shock Tube Measurements of 3-Pentanone Pyrolysis and Oxidation Introduction Experimental Details Mixture Preparation xii

13 5.2.2 Species Absorption Coefficient Evaluations Results and Discussion Pentanone Pyrolysis Pentanone Oxidation Comparisons of Ketone Oxidation Characteristics Summary Possible Future Work...87 Chapter 6 High-Temperature Measurements of the Reactions of OH with a Series of Ketones: Acetone, 2-Butanone, 3-Pentanone, and 2-Pentanone Introduction Experimental Details Kinetic Measurements Choice of Kinetic Mechanisms Acetone + OH Kinetics Butanone + OH Kinetics Pentanone + OH Kinetics Pentanone + OH Kinetics Comparison of Ketone + OH Kinetics Comparison with Low Temperature Data Comparison with Structure-Activity Relationship Summary Chapter 7 High-Temperature Measurements of the Reactions of OH with Small Methyl Esters: Methyl Formate, Methyl Acetate, Methyl Propanoate, and Methyl Butanoate Introduction Experimental Details xiii

14 7.3 Kinetic Measurements Choice of Kinetic Mechanisms Methyl Formate (MF) + OH Kinetics Methyl Acetate (MA) + OH Kinetics Methyl Propanoate (MP) + OH Kinetics Methyl Butanoate (MB) + OH Kinetics Comparison with Low Temperature Data Comparison with Structure-Activity Relationship Summary Chapter 8 Conclusions and Future Work Summary of Results H 2 + OH Kinetics Ketone Combustion Chemistry Methyl Ester + OH Kinetics Publications Recommendations for Future Work Ethyl Radical Diagnostics and Decomposition Pathway Methyl Ester Kinetics Appendix A Shock Tube Ignition Delay Time Measurements in Propane/O 2 /Argon Mixtures at Near-Constant-Volume Conditions Appendix B Ignition Delay Time Measurements of Normal Alkanes and Cycloalkanes Appendix C Multi-Species Time History Measurements during the Oxidation of n- Decane, iso-octane, and Toluene Bibliography xiv

15 List of Tables Table 3.1: Reactions Describing H 2 + OH Experiments at P = 1.3 atm Table 3.2: Rate Constant Data for H 2 + OH H 2 O + H Table 4.1: Summary of acetone unimolecular dissociation rate constant data Table 4.2: Summary of overall 2-butanone decomposition rate constant data Table 5.1: Summary of test gas mixture compositions and measured species Table 5.2: Kinetic parameters employed in the Serinyel et al. mechanism Table 5.3: Summary of overall 3-pentanone decomposition rate constant data Table 6.1: CH 3 COCH 3 + OH Products: Rate Constant Data Table 6.2: C 2 H 5 COCH 3 + OH Products: Rate Constant Data Table 6.3: C 2 H 5 COC 2 H 5 + OH Products: Rate Constant Data Table 6.4: C 3 H 7 COCH 3 + OH Products: Rate Constant Data Table 7.1: CH 3 OCHO + OH Products: Rate Constant Data Table 7.2: CH 3 OC(O)CH 3 + OH Products: Rate Constant Data Table 7.3: CH 3 OC(O)C 2 H 5 + OH Products: Rate Constant Data Table 7.4: CH 3 OC(O)C 3 H 7 + OH Products: Rate Constant Data Table 7.5: Comparison of the rate constants for channels (12a)-(12d) from Fisher et al. [48], Dooley et al. [54], and Hakka et al. [55] at 1133 K and 1300 K xv

16 xvi

17 List of Figures Figure 1.1: (a) Arrhenius plot for the reaction of OH with H 2 at all temperatures. (b) Arrhenius plot for the reaction of OH with H 2 at high temperatures ( K) Figure 1.2: Impact of the rate constant for the reaction of H 2 + OH H 2 O + H and its uncertainty (within a factor of 2) on the laminar flame speed predictions of H 2 -air mixtures at 298 K and 1 atm. Laminar flame speed simulations were performed using the preliminary CEFRC foundational fuel model [33] with the H 2 + OH reaction rate constant from Michael and Sutherland [18]... 5 Figure 2.1: Schematic of OH detection using a narrow-linewidth ring dye laser near nm. Figure adapted from refs. [23-24] Figure 2.2: High-temperature ketone absorption cross-sections at nm Figure 2.3: Schematic of CH 3 detection near nm using the frequency-quadrupled output of near-infrared radiation from a pulsed Ti:Sapphire laser. Figure adapted from ref. [63] Figure 2.4: Schematic of CO detection using cw quantum cascade laser near 4.56 µm. Figure adapted from ref. [66] Figure 2.5: Schematic of C 2 H 4 detection using cw CO 2 laser near 10.5 µm. Figure adapted from ref. [67] Figure 2.6: Schematic of H 2 O detection using cw DFB laser near 2.55 µm. Figure adapted from ref. [69] Figure 2.7: Schematic of CH 4 detection using a scanned-wavelength mid-ir laser near 3.4 µm. Figure adapted from ref. [70] Figure 3.1: OH sensitivity plot for the rate constant measurement of H 2 + OH at 1228 K and 1.29 atm xvii

18 Figure 3.2: Sample H 2 + OH rate constant measurement using the mixture of 1001 ppm H 2 with ~26 ppm TBHP (and 99 ppm water) in Ar at 1228 K and 1.29 atm. Simulation from USC-Mech v2.0 [72] for the best-fit rate constant, along with variations of ±20%, is also shown Figure 3.3: H 2 + OH rate constant measurements at various temperatures, along with the simulations from USC-Mech v2.0 for the best-fit rate constants Figure 3.4: Uncertainty analysis for the rate constant of H 2 + OH H 2 O + H at 1228 K and 1.29 atm Figure 3.5: Arrhenius plot for H 2 + OH (k 1 ) at temperatures above 833 K Figure 3.6: Comparison with previous studies at temperatures above 833 K Figure 4.1: CO sensitivity for 0.25% acetone in Ar using the Pichon et al. mechanism [89] Figure 4.2: CO rate of production (ROP) plot for 0.25% acetone in Ar using the Pichon et al. mechanism [89] Figure 4.3: Sample CO time histories: measured and calculated values Figure 4.4: Summary of acetone dissociation rate constant (k 2 ) Figure 4.5: Acetone sensitivity for 1% acetone in Ar using the Pichon et al. mechanism [89] Figure 4.6: Acetone time histories for 1% acetone in Ar: measured and simulated values Figure 4.7: CH 3 time histories for 0.25% acetone in Ar: measured and calculated values Figure 4.8: C 2 H 4 time histories for 1% acetone in Ar: measured and calculated values Figure 4.9: CH 4 time histories for 1.5% acetone in Ar: measured and calculated values Figure 4.10: Arrhenius plot of the rate constants for the reaction of CH 3 COCH 3 + CH 3 CH 3 COCH 2 + CH 4 from Saxena et al. [88], Sato and Hidaka [87], and Pichon et al. [89] Figure 4.11: 2-Butanone time histories for 1% 2-butanone in Ar: measured and simulated values Figure 4.12: 2-Butanone sensitivity for 1% 2-butanone in Ar Figure 4.13: Arrhenius plot for overall 2-butanone decomposition rate constant (k 3 ) Figure 4.14: CH 3 time histories for 0.25% 2-butanone in Ar: measured and calculated values xviii

19 Figure 4.15: CO rate of production (ROP) plot for 1% 2-butanone in Ar using the original Serinyel et al. mechanism (with the revised k 3 ) Figure 4.16: CO time histories for 1% 2-butanone in Ar: measured and calculated values Figure 4.17: C 2 H 4 time histories for 1% 2-butanone in Ar: measured and calculated values Figure 4.18: CH 4 time histories for 1.5% 2-butanone in Ar: measured and calculated values Figure 5.1: Comparison of CO mole fraction time histories at 1325 K and 1.60 atm with different absorption coefficients in Beer s law Figure 5.2: Comparison of OH mole fraction time histories at 1486 K and 1.52 atm with different absorption coefficients in Beer s law. The OH mole fractions by constant U, V and constant H, P are virtually indistinguishable for OH Figure 5.3: Measured and simulated 3-pentanone time histories for 1% 3-pentanone in Ar. Simulations used the Serinyel et al. mechanism Figure 5.4: 3-pentanone sensitivity analysis for 1% 3-pentanone in Ar at 1323 K and 1.6 atm Figure 5.5: CH 3 time histories for 0.1% 3-pentanone in Ar. Simulations were done using the Serinyel et al. mechanism Figure 5.6: CH 3 sensitivity analysis for 0.1% 3-pentanone in Ar at 1433 K and 1.6 atm Figure 5.7: (a) Best-fit 3-pentanone time histories and (b) best-fit CH 3 time histories using the Serinyel et al. mechanism with revised overall 3-pentanone decomposition rate constant (k 4 ) Figure 5.8: Arrhenius plot for the overall 3-pentanone decomposition rate constant (k 4 ) at 1.6 atm Figure 5.9: Measured 3-pentanone and CO time histories during 3-pentanone pyrolysis at 1248 K and 1.6 atm Figure 5.10: CO sensitivity analysis for 1% 3-pentanone in Ar at 1248 K and 1.6 atm Figure 5.11: CO time histories for 0.25% 3-pentanone in Ar: measured and calculated values from the (a) original and (b) modified Serinyel et al. mechanisms Figure 5.12: C 2 H 4 time histories for 0.25% 3-pentanone in Ar: measured and calculated values xix

20 Figure 5.13: Sample sidewall pressure and endwall OH* emission time histories recorded during an experiment of 3-pentanone ignition at 1113 K and 1.1 atm (3- pentanone/ 4.0% O 2 / Ar, Φ = 0.5). A tailored gas mixture of 60% helium/ 40% nitrogen was used as driver gas to achieve a long test time. For high fuel concentration mixtures, the definition of the endwall ignition delay time is shown in the figure Figure 5.14: Measured and simulated 3-pentanone ignition delay times at (a) Φ = 1.0 and (b) Φ = 0.5 and P 5 = 1.0 atm Figure 5.15: Comparison of model predictions between (a) the Serinyel et al. mechanism of NUI Galway [98] and (b) the modified mechanism on ignition delay time measurements from Serinyel et al Figure 5.16: OH sensitivity analysis for 400 ppm 3-pentanone with 0.28% O 2 in Ar (Φ = 1.0) at 1486 K and 1.52 atm Figure 5.17: H 2 O sensitivity analysis for 400 ppm 3-pentanone with 0.28% O 2 in Ar (Φ = 1.0) at 1486 K and 1.52 atm Figure 5.18: Comparisons of measured and simulated H 2 O time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3- pentanone with 0.28% O 2 in Ar (Φ = 1.0) Figure 5.19: Comparisons of measured and simulated H 2 O time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3- pentanone with 0.56% O 2 in Ar (Φ = 0.5) Figure 5.20: Comparisons of measured and simulated OH time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3- pentanone with 0.28% O 2 in Ar (Φ = 1.0). Inset figures are provided to show the early-time features over µs Figure 5.21: Comparisons of measured and simulated OH time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3- pentanone with 0.56% O 2 in Ar (Φ = 0.5). Inset figures are provided to show the early-time features over µs Figure 5.22: Comparison of ignition delay times for different ketones (acetone, 2- pentanone and 3-pentanone) xx

21 Figure 5.23: Comparison of OH time histories for the mixtures of ketone (i.e., acetone, 2- pentanone and 3-pentanone) with 0.525% O 2 in Ar at a pressure of 2.6 atm and an equivalence ratio of 1.0. An inset figure is provided to show the early-time features over µs Figure 6.1: OH sensitivity plot for the rate constant measurement of acetone + OH at 1148 K and 1.95 atm Figure 6.2: Sample acetone + OH rate constant measurement using the mixture of 304 ppm acetone with ~28 ppm TBHP (and 73 ppm water) in Ar at 1148 K and 1.95 atm. Simulation from the modified Pichon et al. mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown Figure 6.3: Uncertainty analysis for the rate constant of CH 3 COCH 3 + OH products at 1148 K and 1.95 atm Figure 6.4: Arrhenius plot for acetone + OH (k 5 ) at temperatures above 833 K Figure 6.5: OH sensitivity plot for the rate constant measurement of 2-butanone + OH at 1039 K and 1.41 atm Figure 6.6: Sample 2-butanone + OH rate constant measurement using the mixture of 152 ppm 2-butanone with ~14 ppm TBHP (and 41 ppm water) in Ar at 1039 K and 1.41 atm. Simulation from the modified Serinyel et al. mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown Figure 6.7: Arrhenius plot for 2-butanone + OH (k 6 ) at temperatures above 833 K Figure 6.8: OH sensitivity plot for the rate constant measurement of 3-pentanone + OH at 1188 K and 1.94 atm Figure 6.9: Sample 3-pentanone + OH rate constant measurement using the mixture of 213 ppm 3-pentanone with ~17 ppm TBHP (and 59 ppm water) in Ar at 1188 K and 1.94 atm. Simulation from the modified Serinyel et al. mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown Figure 6.10: Arrhenius plot for 3-pentanone + OH (k 7 ) at temperatures above 833 K Figure 6.11: Sample 2-pentanone + OH rate constant measurement using the mixture of 161 ppm 2-pentanone with ~15 ppm TBHP (and 45 ppm water) in Ar at 1186 K and 1.30 atm. Simulation from the modified Serinyel et al. xxi

22 mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown Figure 6.12: Arrhenius plot for 2-pentanone + OH (k 8 ) at temperatures above 900 K Figure 6.13: Arrhenius plot of the measured rate constants for reactions (5)-(8) at temperatures above 870 K Figure 6.14: Arrhenius plot for acetone + OH products (k 5 ) at all temperatures Figure 6.15: Arrhenius plot for 2-butanone + OH products (k 6 ) at all temperatures Figure 6.16: Arrhenius plot for 3-pentanone + OH products (k 7 ) at all temperatures Figure 6.17: Arrhenius plot for 2-pentanone + OH products (k 8 ) at all temperatures Figure 7.1: OH sensitivity plot for the rate constant measurement of methyl formate + OH at 1168 K and 1.40 atm Figure 7.2: Sample methyl formate + OH rate constant measurement using the mixture of 322 ppm methyl formate with ~26 ppm TBHP (and 70 ppm water) in Ar at 1168 K and 1.40 atm. Simulation from the Dooley et al. mechanism [49] for the best-fit rate constant, along with perturbations of ±50%, is also shown Figure 7.3: Uncertainty analysis for the rate constant of methyl formate + OH products at 1168 K and 1.40 atm Figure 7.4: Arrhenius plot for methyl formate + OH (k 9 ) at temperatures above 833 K Figure 7.5: OH sensitivity plot for the rate constant measurement of methyl acetate + OH at 1091 K and 1.37 atm Figure 7.6: Sample methyl acetate + OH rate constant measurement using the mixture of 384 ppm methyl acetate with ~28.5 ppm TBHP (and 73.5 ppm water) in Ar at 1091 K and 1.37 atm. Simulation from the Dooley et al. mechanism [49] for the best-fit rate constant, along with perturbations of ±50%, is also shown Figure 7.7: Arrhenius plot for methyl acetate + OH (k 10 ) at temperatures above 833 K Figure 7.8: OH sensitivity plot for the rate constant measurement of methyl propanoate + OH at 1208 K and 1.33 atm Figure 7.9: Sample methyl propanoate + OH rate constant measurement using the mixture of 281 ppm methyl propanoate with ~22 ppm TBHP (and 68 ppm water) in Ar at 1208 K and 1.33 atm. Simulation from the Dooley et al. xxii

23 mechanism [54] for the best-fit rate constant, along with variations of ±50%, is also shown Figure 7.10: Arrhenius plot for methyl propanoate + OH (k 11 ) at temperatures above 870 K Figure 7.11: Chemical notations for fuel radicals from MCH + OH reactions used by Orme et al. [133] Figure 7.12: OH sensitivity plot for the rate constant measurement of methyl butanoate + OH at 1133 K and 1.37 atm Figure 7.13: Sample methyl butanoate + OH rate constant measurement using the mixture of 241 ppm methyl butanoate with ~20 ppm TBHP (and 60 ppm water) in Ar at 1133 K and 1.37 atm. Simulation from the Dooley et al. mechanism [54] for the best-fit rate constant, along with variations of ±50%, is also shown Figure 7.14: Arrhenius plot for methyl butanoate + OH (k 12 ) at temperatures above 870 K Figure 7.15: Arrhenius plots for methyl ester + OH reactions at temperatures above 250 K Figure 7.16: Comparison of the present rate constant measurements with the modified SAR estimations Figure A.1: Previous ignition delay time measurements for propane oxidation in air at Φ = 0.5. The constant U, V model calculations utilize the Curran et al. mechanism [100] Figure A.2: Comparison of pressure profiles for a mixture of 0.8% C 3 H 8 / 8% N 2 / Ar obtained with and without driver insert in the Stanford cm diameter shock tube. The fractional pressure rise without driver insert (over 20 ms) is approximately 20%, compared to ±3.0% local pressure variations with driver insert. The decay beginning at 25 ms is due to arrival of the rarefaction wave from the driver section Figure A.3: Comparison of pressure profiles for reactive mixture with and without LPST driver insert. Pressure rise without driver insert (over 10 ms) is 20%, compared to ±3.0% pressure variations with driver insert. Initial reflected xxiii

24 shock conditions: T 5 = 1034 K and P 5 = 7.1 atm (with dp 5 /dt ~ 2%/ms), T 5 = 1044 K and P 5 = 6.7 atm (with dp 5 /dt ~ 0%/ms) Figure A.4: Comparison of pressure profiles for reactive mixture with and without HPST driver insert. Pressure rise without driver insert (over 2.5 ms) is 17.5%, compared to ±1.0% pressure variations with driver insert. Initial reflected shock conditions: T 5 = 996 K and P 5 = 54.7 atm (with dp 5 /dt ~ 7%/ms), T 5 = 1008 K and P 5 = 53.7 atm (with dp 5 /dt ~ 0%/ms) Figure A.5: Ignition delay times for 0.8% C 3 H 8 / 8% O 2 / Ar mixture at P 5 = 6 atm, plotted at the initial post-shock T 5. Experimental data and calculated values from JetSurF v1.0 mechanism [155] and Curran et al. mechanism [100] Figure A.6: Low-pressure experimental data and CHEMSHOCK modeling using JetSurF v1.0 mechanism Figure A.7: High-pressure experimental data (at P 5 = 24 and 60 atm), along with CHEMKIN and CHEMSHOCK modeling using JetSurF v1.0 and Curran et al. mechanisms Figure B.1: n-pentane ignition delay time measurements at pressures of 1.8 and 3.6 atm and equivalence ratios of 1.0 and Figure B.2: n-hexane ignition delay time measurements at pressures of 1.8 and 3.6 atm and equivalence ratios of 1.0 and Figure B.3: n-octane ignition delay time measurements at pressures of 1.8 and 3.6 atm and equivalence ratios of 1.0 and Figure B.4: n-nonane ignition delay time measurements at pressures of 1.8 and 3.6 atm and equivalence ratios of 1.0 and Figure B.5: Cyclohexane (CH) ignition delay time measurements at pressures of 1.5 and 3.0 atm and equivalence ratios of 1.0 and Figure B.6: Methylcyclohexane (MCH) ignition delay time measurements at pressures of 1.5 and 3.0 atm and equivalence ratios of 1.0 and Figure B.7: n-butylcyclohexane (BCH) ignition delay time measurements at pressures of 1.5 and 3.0 atm and equivalence ratios of 0.88 and Figure C.1: OH and C 2 H 4 time history measurements for the mixture of 424 ppm JP-8 with 0.813% O 2 in Ar. Two JP-8 proposed surrogate models were employed. Simulations were done using the Dooley et al. mechanism [172] xxiv

25 Figure C.2: OH time histories for the mixture of ~360 ppm n-decane with 0.813% O 2 in Ar. Simulations were done using JetSurF v1.1 mechanism. An inset figure is also shown to provide the early-time features Figure C.3: C 2 H 4 time histories for the mixture of ~360 ppm n-decane with 0.813% O 2 in Ar. Simulations were done using JetSurF v1.1 mechanism Figure C.4: CO time histories for the mixture of ~360 ppm n-decane with 0.813% O 2 in Ar. Simulations were done using JetSurF v1.1 mechanism Figure C.5: OH time histories for the mixture of 511 ppm iso-octane with 0.813% O 2 in Ar. Simulations were done using LLNL mechanism (iso-octane mech. v3). An inset figure is also shown to provide the early-time features Figure C.6: CO time histories for the mixture of 511 ppm iso-octane with 0.813% O 2 in Ar. Simulations were done using LLNL mechanism (iso-octane mech. v3) Figure C.7: OH time histories for the mixture of 640 ppm toluene with 0.813% O 2 in Ar. Simulations were done using JetSurF v1.1 mechanism. An inset figure is also shown to provide the early-time features Figure C.8: CO time histories for the mixture of 640 ppm toluene with 0.813% O 2 in Ar. Simulations were done using JetSurF v1.1 mechanism Figure C.9: Comparison of ignition delay time measurements for JP-8, n-decane, isooctane, and toluene at 1.6 atm xxv

26 xxvi

27 Chapter 1 Background and Motivation 1.1 Introduction The path towards cleaner and more efficient fuel burning is highly desirable worldwide. To mitigate CO and particulate matter (PM) emissions in the United States, strict federal regulations have been implemented into the engine industries. For instance, PM and NO x emissions from heavy-duty diesel engines produced after 2010 must be reduced by 90% of the emission levels that were recorded in Owing to these strict regulations, further improvements in combustion chamber and fuel injection systems are required. In conjunction with these improvements, the use of oxygenated fuels to supplement petroleum-based fuels is also considered as one of the potential strategies in achieving higher fuel efficiency and reducing pollutant emissions. In particular, oxygenated fuels are known to assist in more complete combustion by adding oxygen as part of the fuel, thereby lowering CO and hydrocarbon emissions. Hence, optimizing the use of oxygenated fuels as neat fuels or additives is pertinent to the current engine development. To further understand the influence of oxygenated fuels on engine performance, a comprehensive model, that can predict important observables such as efficiency and pollutant emissions, is needed. This type of predictive model generally requires intensive knowledge in the areas of heat transport, fluid dynamics, and chemistry. Additionally, in some advanced combustion systems (e.g., homogeneous charge compression ignition (HCCI) engines), chemistry plays a critical role in governing the overall system performance, such as autoignition of the fuel-oxidizer mixtures, exhaust gas compositions 1

28 (i.e., CO, CO 2, and NO x ), and heat release rates. A detailed kinetic mechanism is typically used to describe the chemistry of these particular combustion events, and is comprised of hundreds to thousands of elementary reactions specified by rate constants, which are strong functions of temperature and pressure. Unfortunately, most of the existing mechanisms are likely to be error-prone, and require some sort of experimental validation via several kinetic targets. Typical kinetic targets for these mechanisms are ignition delay times, species concentration time histories, and direct reaction rate constant measurements [1]. In this dissertation, the kinetics of the H 2 + OH reaction is reexamined, and the combustion characteristics of two types of oxygenated fuels (i.e., ketones and methyl esters) are investigated. Several important kinetic targets are provided in order to evaluate and refine the existing kinetic mechanisms, and some definite conclusions regarding these mechanisms can be drawn from these experimental observations. 1.2 Background and Motivation H 2 + OH Kinetics The reaction of hydroxyl radicals with molecular hydrogen H 2 + OH H 2 O + H (1) is an important chain-propagating reaction in all combustion systems, particularly in hydrogen combustion. Its reverse reaction plays a critical role in the establishment of partial equilibrium in the post-combustion regime. Because of its important role in combustion, numerous direct rate constant measurements for reaction (1) have been conducted across a broad range of temperatures [2-21]. Figure 1.1 demonstrates some of the previous experimental determinations for reaction (1). Ravishankara and co-workers [8-9] measured the rate constant for the title reaction (under pseudo-first-order kinetic conditions) using the flash photolysis resonance fluorescence technique to monitor the temporal profiles of OH decays in a heated quartz cell over K. Their measurements confirmed the nonlinearity of the Arrhenius plot for reaction (1) over this wide temperature range. Recently, Orkin et 2

29 al. [15] reexamined the rate constant for reaction (1) using the flash photolysis resonance fluorescence technique in a Pyrex reactor over a narrower temperature range of K, and their results are in excellent agreement with the measured values from Ravishankara and co-workers [8-9]. At combustion-relevant conditions (T > 1000 K), the measurements for reaction (1) were generally carried out using shock tubes, and these high-temperature data have a much larger scatter (within a factor of 2) than the lowtemperature data, as illustrated in Figure 1.1. Frank and Just [17] measured the rate constants for the reactions of H + O 2 OH + O and H 2 + OH H 2 O + H by employing atomic resonance absorption spectrometry (ARAS) to monitor H- and O-atom concentration profiles behind reflected shock waves over K. Michael and Sutherland [18] studied the rate constant for the reaction of H + H 2 O H 2 + OH (reaction (-1)) using flash photolysis of H 2 O to generate H-atoms and using ARAS to monitor the temporal profiles of H-atom decays behind reflected shock waves over K. The rate constant for reaction (1) was calculated from the measured reverse rate constant and the equilibrium constant, and the resulting data were then compiled with the earlier experimental work from Ravishankara and co-workers [8-9] and Frank and Just [17] to form a three-parameter least-squares fit: k 1 (T) = T 1.51 exp(-3430 [cal/mol]/rt) cm 3 mol -1 s -1 over K. This expression is currently adopted in GRI-Mech 3.0 [22]. It is pertinent to note that the recent revised standard enthalpy of formation for OH at 298 K [23-24] seems to suggest a lower rate constant expression for reaction (1) (~15% lower) if the rate constant is evaluated from the reverse reaction and the revised equilibrium constant. Figure 1.1 also shows the revised evaluations from Michael and Sutherland [18] using the revised equilibrium constants. Similarly, Davidson et al. [19] studied the rate constant for the reverse reaction by using laser photolysis of H 2 O and UV laser absorption of OH near 307 nm behind reflected shock waves over K, and the rate constant for reaction (1) can be evaluated from their measured values and the revised equilibrium constants, as shown in Figure 1.1. Oldenborg et al. [20] conducted direct rate constant measurements for reaction (1) (under pseudo-first-order kinetic conditions) using the laser photolysis / laser-induced fluorescence technique to monitor OH radical concentration profiles in a heated cell over 3

30 K. Moreover, Krasnoperov and Michael [21] reexamined the rate constant for reaction (1) using a novel multi-pass absorption spectrometric detection technique to monitor OH species profiles at 308 nm in reflected shock wave experiments over K. Figure 1.1: (a) Arrhenius plot for the reaction of OH with H 2 at all temperatures. (b) Arrhenius plot for the reaction of OH with H 2 at high temperatures ( K). An accurate knowledge of the rate constant for reaction (1) is important in interpreting laminar flame speed measurements of H 2 -O 2 mixtures. Figure 1.2 presents the unstretched laminar flame speed measurements of H 2 -air mixtures as a function of equivalence ratio at standard initial temperature (298 K) and ambient pressure (1 atm) 4

31 [25-32], along with the simulations from the preliminary CEFRC foundational fuel model [33] with the rate constant for reaction (1) from Michael and Sutherland [18]. Note that the previous rate constant measurements of reaction (1) have a relatively large scatter (within a factor of 2) at elevated temperatures, and such large uncertainty in reaction (1) can affect the laminar flame speed predictions of H 2 -air mixtures from the model by approximately ±11%, as demonstrated in Figure 1.2. Figure 1.2: Impact of the rate constant for the reaction of H 2 + OH H 2 O + H and its uncertainty (within a factor of 2) on the laminar flame speed predictions of H 2 -air mixtures at 298 K and 1 atm. Laminar flame speed simulations were performed using the preliminary CEFRC foundational fuel model [33] with the H 2 + OH reaction rate constant from Michael and Sutherland [18] Ketone Combustion Chemistry There has been an increased interest in studying bio-derived oxygenated fuels because of their potential to help minimize fossil fuel consumption. Among these oxygenated fuels, methyl esters and alcohols have attracted a great deal of attention in the form of both theoretical and experimental studies. Other oxygenates, such as ketones, though not used as fuels, play a large role in the oxidation of the hydrocarbon and 5

32 oxygenated fuels. Fewer experimental studies at combustion conditions, however, have been carried out for ketones (e.g., acetone, 2-butanone, and 3-pentanone). Ketones are also used as fuel tracers for quantitative planar laser-induced fluorescence measurements (PLIF) of temperature and species concentration distributions in internal combustion engine research [34-37]. They are chosen for this purpose because they exhibit broad absorption spectra in the ultraviolet region (π* n transition) and sufficient quantum yields and strong fluorescence spectra in the visible region to be easily monitored. 3-pentanone, in particular, is a popular ketone fuel tracer due to its similar physical characteristics (e.g., boiling point) to that of the gasoline primary reference fuels (n-heptane and iso-octane). The impact of 3-pentanone in internal combustion engine studies requires more detailed information about its oxidation and pyrolysis chemistry in order to predict its influence as an additive on the ignition processes of the main fuel. Moreover, ketones are listed as a class of volatile organic compounds (VOCs), and are massively produced and used as solvents or polymer precursors in industries. As one of the common pollutants, some amounts of ketones are emitted into the atmosphere from a variety of natural and anthropogenic sources. The reactions with OH radicals are the primary removal pathways for ketones in the atmosphere, which may result in the formation of ozone and other components of the photochemical smog in urban areas [38]. In addition, these reactions of OH with ketones are one of the primary fuel consumption pathways during oxidation, and are poorly understood at high temperatures. Hence, an accurate knowledge of these H-atom abstraction reactions is needed in the development of successful detailed mechanisms suitable for high-temperature applications Methyl Ester + OH Kinetics Biodiesel is a promising alternative fuel because it has physical properties similar to conventional crude-oil-derived fossil fuels, and it provides the opportunity to reduce overall emissions of atmospheric pollutants [39]. Biodiesel is generally comprised of a mixture of extended alkyl chain methyl esters carbon atoms long [40], that are 6

33 typically derived from soybean oil in U.S. or rapeseed oil in Europe. Despite the complexity of these molecules, there has been a growing effort to develop comprehensive reaction mechanisms that can be used to describe the combustion of these large methyl esters [41-44]. In these detailed mechanisms, the reaction rate constants for large methyl esters are primarily based on the kinetic parameters of smaller methyl esters (e.g., methyl formate and methyl butanoate) [41, 42, 45-47]. Thus, accurate knowledge of the kinetic parameters for smaller methyl esters is crucial to the development of the detailed mechanism for practical biodiesel fuels. The combustion chemistry of small methyl esters has been a subject of interest for the past decade. Fisher et al. [48] developed the first comprehensive chemical kinetic mechanisms for the oxidation of methyl formate and methyl butanoate. However, the mechanisms were validated against only a limited set of low-temperature experimental data. Recently, Dooley et al. [49] have compiled a detailed mechanism for methyl formate oxidation, and the mechanism has been validated against a wide variety of experimental data, including shock tube ignition delay times, speciation data from a variable-pressure flow reactor, and laminar burning velocities of outwardly propagating spherical flames. Similarly, Ren et al. [50] conducted direct rate constant measurements of the initial dissociation pathways of methyl formate over K at pressures near 1.6 atm using shock tube/laser absorption techniques, and their measurements are in close accord with the estimated values adopted in the Dooley et al. mechanism [49]. Concurrently, Peukert et al. [51-52] investigated the high-temperature thermal decomposition and the H-atom abstraction reactions by H-atoms for methyl formate and methyl acetate over K at pressures around 0.5 atm using shock tube/atomic resonance absorption spectrometry technique. In addition, Westbrook et al. [53] developed a detailed mechanism for a group of four small alkyl esters, including methyl formate, methyl acetate, ethyl formate, and ethyl acetate. The mechanism was validated against the speciation data from fuel-rich, low-pressure, premixed laminar flames. Similarly, Dooley et al. [54] and Hakka et al. [55] developed two separate detailed mechanisms for methyl butanoate oxidation, and the mechanisms were tested against different sets of experimental data, including shock tube and rapid compression machine 7

34 ignition delay times and speciation data from a flow reactor, a jet-stirred reactor, and an opposed-flow diffusion flame. Moreover, numerous experimental studies [56-59] for methyl butanoate pyrolysis and oxidation were performed in order to improve the global performance of the existing detailed mechanisms. However, among most of these studies, little attention has been given to a better understanding of the elementary kinetics of these methyl esters. In particular, the H-atom abstraction reactions by OH radicals for methyl esters, which are one of the major fuel consumption pathways during oxidation, are not well known at combustion-relevant conditions. 1.3 Scope and Organization of Thesis The dissertation is organized as follows: 1) Chapter 2 describes the shock tube facility and different laser absorption techniques, which were utilized in this work to monitor key combustion radicals and intermediate species (i.e., OH, ketones, CH 3, CO, C 2 H 4, H 2 O, and CH 4 ). 2) Chapter 3 presents the high-temperature experimental determination of the important chain-propagating reaction in all combustion systems. H 2 + OH H 2 O + H (1) The present high-temperature measurements were also compared with the previous experimental determinations, the values employed in several detailed kinetic mechanisms, and the theoretical calculation using semi-classical transition state theory (SCTST). 3) Chapter 4 discusses the multi-species time history measurements during hightemperature acetone and 2-butanone pyrolysis. In this chapter, five different species, namely ketone, CH 3, CO, C 2 H 4, and CH 4, were presented and compared with the simulations from the detailed kinetic mechanisms. During acetone pyrolysis, the CO and acetone concentration time histories were used to infer the rate constant for acetone unimolecular decomposition reaction at pressures of atm: CH 3 COCH 3 (+ M) CH 3 + CH 3 CO (+ M) (2) 8

35 Similarly, during 2-butanone pyrolysis, the measured 2-butanone time histories were used to determine the overall rate constant (k 3 = k 3a + k 3b + k 3c ) for 2-butanone decomposition at pressures of atm: C 2 H 5 COCH 3 (+ M) C 2 H 5 + CH 3 CO (+ M) (3a) C 2 H 5 COCH 3 (+ M) CH 3 + C 2 H 5 CO (+ M) (3b) C 2 H 5 COCH 3 (+ M) CH 3 + CH 3 COCH 2 (+ M) (3c) In addition, using the measured 2-butanone and CO time histories and an O-atom balance analysis, a missing removal pathway for methyl ketene (one of the major products predicted by the model) was identified. 4) Chapter 5 discusses the shock tube measurements of 3-pentanone pyrolysis and oxidation. In this chapter, we provided six species time history measurements (i.e., 3- pentanone, CH 3, CO, C 2 H 4, OH, and H 2 O), along with 3-pentanone ignition delay time measurements. These measurements were also compared with the simulations from the detailed kinetic mechanism. More importantly, during 3-pentanone pyrolysis, the measured 3-pentanone and CH 3 time histories were used to determine the overall rate constant (k 4 = k 4a + k 4b ) for 3-pentanone decomposition at pressures of atm: C 2 H 5 COC 2 H 5 (+ M) C 2 H 5 + C 2 H 5 CO (+ M) (4a) C 2 H 5 COC 2 H 5 (+ M) CH 3 + C 2 H 5 COCH 2 (+ M) (4b) Similar to 2-butanone pyrolysis, an O-atom balance analysis from the measured 3- pentanone and CO time histories identified the absence of the methyl ketene decomposition pathway in the detailed mechanism. 5) Chapter 6 provides the direct high-temperature overall rate constant measurements of acetone + OH, 2-butanone + OH, 3-pentanone + OH, and 2-pentanone + OH reactions. CH 3 COCH 3 + OH CH 3 COCH 2 + H 2 O (5) C 2 H 5 COCH 3 + OH Products (6) C 2 H 5 COC 2 H 5 + OH Products (7) C 3 H 7 COCH 3 + OH Products (8) 9

36 6) Chapter 7 presents the first direct high-temperature overall rate constant measurements of methyl formate + OH, methyl acetate + OH, methyl propanoate + OH, and methyl butanoate + OH reactions. CH 3 OCHO + OH Products (9) CH 3 OC(O)CH 3 + OH Products (10) CH 3 OC(O)C 2 H 5 + OH Products (11) CH 3 OC(O)C 3 H 7 + OH Products (12) 7) Chapter 8 summarizes the present rate constant determinations of reactions (1)-(12), and proposes some future plans, which include multi-species time history experiments and direct rate constant measurements of H-atom abstraction reactions for large methyl esters (i.e., methyl decanoate). 8) At long shock tube test times, as are needed at low reaction temperatures, small gradual increases in pressure that result from incident shock wave attenuation and boundary layer growth can significantly shorten measured ignition delay times. In Appendix A, we investigated such pressure effects on propane ignition delay times at pressures of 6, 24, and 60 atm. 9) Appendix B presents the ignition delay time measurements of four n-alkanes (e.g., n- pentane, n-hexane, n-octane, and n-nonane) and three cycloalkanes (e.g., cyclohexane, methylcyclohexane, and n-butylcyclohexane) at various reflected shock temperatures and pressures (between 1240 and 1500 K and 1.5 and 3.8 atm) and at two equivalence ratios, namely Φ = 1.0 and Φ = ) Appendix C presents the species time history measurements of OH, C 2 H 4, and CO during the high-temperature oxidation of n-decane, iso-octane, and toluene, which are the proposed surrogate fuel components for JP-8. Concurrently, the measured ignition delay times of these surrogate fuel components were compared with the measured JP-8 ignition delay times. 10

37 Chapter 2 Experimental Methods This chapter discusses the shock tube facility and different laser diagnostic systems employed in this work. 2.1 Shock Tube Facility Experiments were performed in a stainless-steel, high-purity, low-pressure shock tube at Stanford. The shock tube is comprised of a 3.7-m driver section and a 10-m driven section, with an inner diameter of cm. The shock tube driver and driven sections are separated by a polycarbonate diaphragm of in thickness. Incident shock velocity measurements were made using a series of five piezoelectric pressure transducers (PCB 113A26 transducer, PCB 483B08 amplifier) over the last 1.5 m of the shock tube and linearly extrapolated to the endwall. Average shock velocity attenuation rates were between % per meter. Reflected shock temperatures and pressures were determined from the incident shock velocity at the endwall using standard normal shock relations, with uncertainties of approximately ±0.7% and ±1%, respectively, mainly due to the uncertainty in the measured shock velocity (±0.2%) [23]. For all experiments presented in this dissertation, vibrational equilibrium can be assumed immediately behind the incident and reflected shock waves. In addition to the five piezoelectric pressure transducers, a Kistler pressure transducer was utilized to measure the pressure time histories upon shock-heating. All laser absorption diagnostics, along with the Kistler pressure transducer, were located at a test section 2 cm from the driven section endwall. Concurrently, prior to every experiment, the shock tube and 11

38 mixing assembly were routinely turbomolecular pumped down to ~5 µtorr to ensure purity of the test mixtures, with a typical subsequent leak-plus-outgassing rate of less than 50 µtorr/min. Further details of the shock tube facility can be found elsewhere [60-62]. 2.2 Laser Absorption Methods UV Laser Absorption of OH OH radical concentration was measured using the frequency-doubled output of a narrow-linewidth ring dye laser near nm, as illustrated in Figure 2.1. The laser wavelength was tuned to the peak of the well-characterized R 1 (5) absorption line in the OH A X (0, 0) band. Visible light near nm was generated by pumping Rhodamine 6G dye in a Spectra Physics 380A laser cavity with the 5 W, cw output of a Coherent Verdi laser at 532 nm. The visible light was then intracavity frequency-doubled using a temperature-tuned AD*A nonlinear crystal to generate ~1 mw of light near nm. Using a common-mode-rejection detection scheme, a minimum absorbance of 0.1% can be detected, which resulted in the current experiments in a minimum detection sensitivity of ~0.2 ppm at 1400 K and 1.5 atm (with k λ = cm -1 atm -1 ). Further details of the OH laser diagnostic setup are discussed elsewhere [23-24]. The overall estimated uncertainty in the measured OH mole fraction (X OH ) is approximately ±3%, mainly due to the uncertainty in temperature (±0.7%). To check for the interference absorption, the laser was also tuned away from the narrow OH absorption line by approximately 4 cm -1. If the interference absorption was found, an OH off-line measurement was required for each OH on-line measurement. Under the assumption that the interfering species has wavelength-independent absorption near nm, the interference absorbance of the off-line measurement can be directly subtracted from the total absorbance of the OH online measurement. OH species concentration can then be calculated from Beer s law: -ln(i/i o ) corrected = -ln(i/i o ) online + ln(i/i o ) offline -ln(i/i o ) corrected = k OH X OH PL 12

39 where I and I o are the transmitted and incident laser intensities, k OH is the OH absorption coefficient, X OH is the OH mole fraction, P is the total pressure, and L is the path length (15.24 cm). Figure 2.1: Schematic of OH detection using a narrow-linewidth ring dye laser near nm. Figure adapted from refs. [23-24] UV Laser Absorption of Ketones Ketones (i.e., acetone, 2-butanone, 2-pentanone, and 3-pentanone) are known to have a near-uv absorption spectrum that corresponds to the symmetry forbidden electronic π* n transition where an electron from a non-binding orbital localized near the oxygen atom is excited to an anti-bonding orbital around the CO group [34-35]. At current experimental conditions, the spectrum is broad, lacks any fine structure, and varies gradually from 220 to 340 nm with peak absorption at around 295 nm. This is completely consistent with the interfering absorption seen in the OH measurements during ketone combustion studies. Taking advantage of this fact, we have measured ketones (off-line of OH) at nm using the same laser system as was used for OH measurements. Figure 2.2 shows the absorption cross-sections (at nm) of acetone, 2-butanone, and 3-pentanone over K at pressures near 1.5 atm, which were determined by measuring the absorption (from the mixtures of 1% ketone in Ar) immediately behind reflected shock waves when only ketone existed. The uncertainties 13

40 in ketone cross-sections were estimated to be ±5%. Using a common-mode-rejection detection scheme, a minimum ketone detection sensitivity of ~300 ppm at 1300 K and 1.5 atm can be achieved (with k λ 0.52 cm -1 atm -1 ). Figure 2.2: High-temperature ketone absorption cross-sections at nm UV Laser Absorption of CH 3 Methyl radical has a wide predissociatively broadened absorption feature (B 2 A 1 X 2 A 2 ) near 216 nm, with peak absorption at nm. In this work, CH 3 was measured using the frequency-quadrupled output of near-infrared radiation from a pulsed Ti:Sapphire laser, as illustrated in Figure 2.3. Stable mode-locking of the Ti:Sapphire laser (MIRA HP, Coherent Inc.) was obtained at a wavelength of nm with a peak output of 1 W, a repetition rate of approximately 76 MHz, and a pulse duration of approximately 2 picoseconds. Deep UV light was generated by frequency conversion, using fourth harmonic generation (FHG, Coherent Inc.), to obtain output at nm. Further details of this laser setup can be found elsewhere [63]. Similar to the other laser absorption methods described here, using a common-mode-rejection detection scheme, a minimum absorbance of ~0.1% can be detected, resulting in a minimum detection sensitivity of ~1 ppm at 1300 K and 1.5 atm (with k λ = 55.8 cm -1 atm -1 ). 14

41 It is also known that there is some interference absorption near 216 nm from the intermediate products of pyrolysis, such as ethylene, higher olefins and conjugated olefins. To account for the interference absorption, another wavelength at nm was also employed in this study. Here we assume that the interference absorbances at nm and nm are identical, based on the fact that these two wavelengths are very close to each other, and at high temperatures, these hydrocarbons show wide broad absorption features near 216 nm. CH 3 absorption coefficients at nm and nm were previously determined in our laboratory [64-65]. Thus, methyl radical concentration can then be calculated as follows: -ln(i/i o ) 217 = k CH3,217 X CH3 PL + k int X int PL -ln(i/i o ) 219 = k CH3,219 X CH3 PL + k int X int PL -ln(i/i o ) ln(i/i o ) 219 = (k CH3,217 k CH3,219 )X CH3 PL where k CH3 is the CH 3 absorption coefficient, X CH3 is the CH 3 mole fraction, P is the total pressure, and L is the path length. Based on the two-wavelength subtraction scheme, the interference absorption contributes about 25% to the total absorption signal at nm. Figure 2.3: Schematic of CH 3 detection near nm using the frequency-quadrupled output of near-infrared radiation from a pulsed Ti:Sapphire laser. Figure adapted from ref. [63]. 15

42 2.2.4 IR Laser Absorption of CO A quantum cascade laser (QCL) operating in cw mode has recently become an important diagnostic tool for many combustion products, such as CO, CO 2 and H 2 O. This mid-ir CO laser allows access to the R(13) transition line in the CO fundamental rovibrational band at 4.56 µm, where H 2 O and CO 2 absorption interference is minimal. As compared to previous CO diagnostics near 1.2, 1.5 and 2.3 µm, this new diagnostic scheme offers orders-of-magnitude greater sensitivity, resulting in ppm-level CO detectivity in shock tube measurements. A fixed-wavelength direct-absorption strategy was employed to monitor the peak intensity of the R(13) absorption line at cm -1 (with k λ = 12.1 cm -1 atm -1 at 1300 K and 1.5 atm), as illustrated in Figure 2.4. The spectroscopic parameters for the R(13) transition, including the line-strength and self-broadening coefficient, were taken directly from the HITRAN database. The collisional broadening coefficient for CO with argon (not available in HITRAN) was measured in the shock tube over the temperature range of K. Further details regarding the CO diagnostic setup are described elsewhere [66]. Figure 2.4: Schematic of CO detection using cw quantum cascade laser near 4.56 µm. Figure adapted from ref. [66]. 16

43 2.2.5 CO 2 Laser Absorption of C 2 H 4 C 2 H 4 was measured using CO 2 laser absorption at µm, as illustrated in Figure 2.5. This diagnostic takes advantage of the strong overlap of the P(14) line of the CO 2 laser transition with the strong Q-branch of the ν 7 ethylene band. It is capable of detecting 100 ppm levels of C 2 H 4 over a path length of 15 cm at 1200 K and 3 atm (with σ λ = 9.69 m 2 /mol or k λ = 0.98 cm -1 atm -1 ). Details regarding this diagnostic and the C 2 H 4 absorption cross-sections are described elsewhere [67]. There is some weak interference absorption directly from acetone, 2-butanone, and 3-pentanone at this wavelength. In addition, higher alkenes, such as propene and butene, have weak absorption features at µm. However, the pyrolysis of acetone, 2-butanone, and 3-pentanone generate negligible amounts of these higher alkenes, and thus, the primary interfering species are ketones. In the present study, C 2 H 4 was also measured at µm to account for ketone interference absorbance; C 2 H 4 absorption cross-sections at µm were found to be 4.0 ± 0.1 m 2 /mol over K at 1-6 atm [68]. Moreover, ketone absorption cross-sections at µm and µm were determined by measuring the absorption immediately behind reflected shock waves when only ketone existed. Thus, using these two wavelengths, C 2 H 4 species concentration can be found by solving the following two equations: -ln(i/i o ) = n C2H4 σ C2H4, L + n ket σ ket, L -ln(i/i o ) = n C2H4 σ C2H4, L + n ket σ ket, L where σ C2H4 and σ ket are the absorption cross-sections of C 2 H 4 and ketone, n C2H4 and n ket are the number densities of C 2 H 4 and ketone, and L is the path length. 17

44 Figure 2.5: Schematic of C 2 H 4 detection using cw CO 2 laser near 10.5 µm. Figure adapted from ref. [67] IR Laser Absorption of H 2 O H 2 O concentration was measured using a DFB (distributed feedback) diode laser at nm ( cm 1 ) within the ν 3 fundamental vibrational band, as illustrated in Figure 2.6. This absorption feature has been well-characterized previously in our laboratory [69]. During experiments, the beam path (outside the shock tube) was continuously purged with pure N 2 to minimize the laser attenuation due to ambient water. A minimum H 2 O detection sensitivity of ~40 ppm can be achieved at 1400 K and 1.5 atm (with k λ = 2.05 cm -1 atm -1 ). Figure 2.6: Schematic of H 2 O detection using cw DFB laser near 2.55 µm. Figure adapted from ref. [69]. 18

45 2.2.7 IR Laser Absorption of CH 4 CH 4 was measured near 3.4 µm using a scanned-wavelength mid-ir laser absorption diagnostic developed by Pyun et al. [70], as illustrated in Figure 2.7. Mid-IR light near 3.4 µm was generated using difference-frequency-generation (DFG) of a near- IR signal laser and a near-ir pump laser combined in a PPLN crystal. Due to the structural differences of the absorption spectrum of methane and other hydrocarbons near 3.4 µm, a differential absorption (peak minus valley) scheme was employed to obtain interference-free CH 4 concentration. To attain this scheme, the signal laser was current modulated at a 50 khz scanning frequency and a 0.5 V p-p scanning amplitude by a function generator. Then the modulated signal laser was combined with the pump laser through the PPLN crystal to create a modulated mid-ir laser light near 3.4 µm that included the peak and valley wavelengths in each scan. To maximize the signal-to-noise ratio of CH 4 and minimize that of the interfering species, cm -1 and cm -1 were selected as the optimal peak and valley wavelength pair in this work. Using this method, CH 4 concentration time histories with a time resolution of 20 µs and a minimum detection sensitivity of ~250 ppm at 1300 K and 1.5 atm were obtained (with a differential absorption coefficient of k peak-valley = 0.73 cm -1 atm -1 ). Figure 2.7: Schematic of CH 4 detection using a scanned-wavelength mid- IR laser near 3.4 µm. Figure adapted from ref. [70]. 19

46 2.3 Summary In this dissertation, a low-pressure shock tube facility and several laser absorption diagnostics were utilized to monitor key combustion radicals and intermediate species, including OH radicals near nm, ketones at nm, CH 3 radicals near 216 nm, CO near 4.56 µm, C 2 H 4 near 10.5 µm, H 2 O near 2.55 µm, and CH 4 near 3.4 µm. 20

47 Chapter 3 A Shock Tube Study of H 2 + OH H 2 O + H using OH Laser Absorption 3.1 Introduction As introduced in Chapter 1, the reaction of OH with molecular hydrogen H 2 + OH H 2 O + H (1) is an important chain-propagating reaction in all combustion systems, particularly in hydrogen combustion. Its reverse reaction plays a critical role in the establishment of partial equilibrium in the post-combustion regime. At elevated temperatures, the previous rate constant evaluations have an uncertainty factor of 2, and this relatively large uncertainty has a significant impact on the laminar flame speed predictions of H 2 -air mixtures from the detailed kinetic mechanism. Thus, there is motivation to reduce the existing experimental uncertainty in k 1 at combustion-relevant conditions. In this chapter, we aim to report the rate constant for reaction (1) with a much lower experimental scatter and overall uncertainty over the temperature range of K. 3.2 Experimental Details Test mixtures were prepared manometrically in a 40 liter stainless-steel tank heated uniformly to 50 o C and mixed with a magnetically-driven stirring vane. A doubledilution process was employed to allow for more accurate pressure measurements in the 21

48 manometrical preparation of a highly dilute mixture. A highly concentrated mixture was first prepared and mixed for at least 2 hours to ensure homogeneity and consistency, and the mixture was then further diluted with argon and mixed for additional 2 hours prior to the experiments. The gases utilized in this work were hydrogen (Research Grade) % and argon (Research Grade) %, which were supplied by Praxair and used without further purification. The liquid chemical was commercially available 70% tertbutyl hydroperoxide (TBHP) in water from Sigma-Aldrich, and was purified using a freeze-pump-thaw procedure to remove dissolved volatiles and air prior to mixture preparation. 3.3 Kinetic Measurements Kinetic Mechanism Description A series of 21 reflected shock wave experiments were conducted to determine the rate constant for the reaction of H 2 + OH H 2 O + H over the temperature range of K at pressures of atm. Dilute test mixtures with ppm TBHP (and water) and ppm H 2 in argon were used to minimize the temperature drop caused by the chemistry effects, and the temperature profile behind the reflected shock wave was nearly constant (<1 K change) over the time frame of the experiment. In the present study, the CHEMKIN PRO package [71] was used to simulate the consumption of OH radicals by molecular hydrogen under the standard constant energy and volume assumption, and a comprehensive reaction mechanism of Wang et al. (USC-Mech v2.0) [72] was selected as the base mechanism. This mechanism consists of 111 species and 784 elementary reactions, and has been validated against a series of shock tube ignition delay times, laminar flame speeds, and speciation data from a shock tube and a flow reactor during high-temperature oxidation of H 2, CO, and C1-C4 hydrocarbons. It is pertinent to note that the conclusions of the present study are effectively independent of the mechanism used, and near-identical results could be obtained using the GRI-Mech 3.0 [22]. 22

49 Tert-butyl hydroperoxide (TBHP or (CH 3 ) 3 CO OH) was chosen as an OH radical precursor in the present study, because it decomposes very rapidly to form an OH radical and a tert-butoxy radical, (CH 3 ) 3 CO, at temperatures greater than 1000 K [73]. (Note that the weakest bond in TBHP is the O O bond with the bond dissociation energy of ~47 kcal/mol at 298 K.) The tert-butoxy radical further decomposes to form acetone and a methyl radical. Additionally, TBHP reacts with OH to form other products, and hence a TBHP sub-mechanism was also incorporated into the base mechanism, i.e. (CH 3 ) 3 CO OH (CH 3 ) 3 CO + OH (13) (CH 3 ) 3 CO CH 3 COCH 3 + CH 3 (14) (CH 3 ) 3 CO OH + OH H 2 O + O 2 + tert-c 4 H 9 (15) (CH 3 ) 3 CO OH + OH H 2 O + HO 2 + iso-c 4 H 8 (16) The rate constants for reactions (13), (15) and (16) were adopted from Pang et al. [74], and the rate constant for reaction (14) was obtained from Choo and Benson [75]. The rate constants for reactions (13)-(16) are listed in Table 3.1. In addition, the thermodynamic parameters for TBHP and tert-butoxy radical were taken from the thermodynamic database from Goos et al. [76], and the standard enthalpy of formation for OH radical was updated with the measured value from Herbon et al. [23-24]. Table 3.1: Reactions Describing H 2 + OH Experiments at P = 1.3 atm. Rate Constant Reaction A [ ] b E [cal/mol] No. Reference H 2 + OH H 2 O + H see text 1 this work TBHP (CH 3 ) 3 CO + OH 3.57E E [74] (CH 3 ) 3 CO CH 3 COCH 3 + CH E E [75] TBHP + OH H 2 O + O 2 + tert-c 4 H E E [74] TBHP + OH H 2 O + HO 2 + iso-c 4 H E E [74] CH 3 + OH CH 2 (s) + H 2 O 1.65E [74] C 2 H 6 (+ M) CH 3 + CH 3 (+ M) 1.88E E [80] Low-Pressure Limit: 3.72E E+05 Troe centering: CH 3 COCH 3 + OH CH 3 COCH 2 + H 2 O 3.30E E+03 5 [82] CH 3 OH + M CH 3 + OH + M 5.62E E [77] Units of A are in s -1 for unimolecular reactions and cm 3 mol -1 s -1 for bimolecular reactions. 23

50 3.3.2 H 2 + OH Kinetics A local OH sensitivity analysis for the mixture of 1001 ppm H 2 with ~26 ppm TBHP (and 99 ppm H 2 O) in Ar at 1228 K and 1.29 atm is shown in Figure 3.1. The OH sensitivity is calculated as S OH = ( X OH / k i ) (k i /X OH ), where X OH is the local OH mole fraction and k i is the rate constant for reaction i. The analysis reveals that the OH time history is predominantly sensitive to reaction (1) over the time frame of the experiment, with some minor interference from the following secondary reactions: CH 3 + OH CH 2 (s) + H 2 O (17) C 2 H 6 (+ M) CH 3 + CH 3 (+ M) (18) CH 3 COCH 3 + OH CH 3 COCH 2 + H 2 O (5) The rate constant for reaction (17) was updated with the value of cm 3 mol -1 s -1 recently inferred by Pang et al. [74], which is in good agreement with the measurements from Srinivasan et al. [77] and Vasudevan et al. [78] and the theoretical calculation from Jasper et al. [79] (within ±35%). The rate constant for reaction (18) was updated with the measured values from Oehlschlaeger et al. [80], and the measurements from Oehlschlaeger et al. are in close accord with another experimental study from Kiefer et al. [81]. Recently, Lam et al. [82] have measured the rate constant for reaction (5) using UV laser absorption of OH near nm behind reflected shock waves over K at pressures near 2 atm, and their measured rate constant was adopted for reaction (5) in the present study. (For more details, please read Chapter 6.) In addition, the rate constant for the reaction of CH 3 OH + M CH 3 + OH + M (reaction (19)) was updated with the measured values from Srinivasan et al. [77] at ~ atm, and their values agree well with the theoretical calculation from Jasper et al. [79] and the measurements from Vasudevan et al. [78] at 1.3 atm. The rate constants for reactions (5) and (17)-(19) are also provided in Table

51 Figure 3.1: OH sensitivity plot for the rate constant measurement of H 2 + OH at 1228 K and 1.29 atm. Figure 3.2 shows an OH time history measurement for the mixture of 1001 ppm H 2 in argon at 1228 K and 1.29 atm, and the measured peak OH mole fraction is approximately 26 ppm. Due to wall adsorption and condensation of TBHP, the initial TBHP mole fraction was assumed to be the same as the measured peak OH mole fraction. This assumption is valid because the OH radical is formed very rapidly after the thermal decomposition of TBHP at T > 1000 K. More importantly, the presence of H 2 O in the mixture has a negligible influence on the simulated OH time history; hence, its presence would not affect our rate constant evaluation. As illustrated in Figure 3.2, a best-fit rate constant for reaction (1) of cm 3 mol -1 s -1 was obtained between the experimental data and the simulation at 1228 K and 1.29 atm. Simulations for the variations of ±20% in the inferred rate constant are also shown in Figure 3.2. Similarly, Figure 3.3 shows the measured OH time histories for different test mixtures at different temperatures, along with the simulations from USC-Mech v2.0 [72] for the best-fit rate constants. Interestingly, our measured values are identical to or very close to the values originally proposed by Michael and Sutherland [18], within ±4% at most temperatures, and their expression is currently adopted in GRI-Mech 3.0 [22]. Additionally, Table 3.2 summarizes the rate constant measurements of reaction (1) at T = K and P = atm. 25

52 Figure 3.2: Sample H 2 + OH rate constant measurement using the mixture of 1001 ppm H 2 with ~26 ppm TBHP (and 99 ppm water) in Ar at 1228 K and 1.29 atm. Simulation from USC-Mech v2.0 [72] for the best-fit rate constant, along with variations of ±20%, is also shown. Figure 3.3: H 2 + OH rate constant measurements at various temperatures, along with the simulations from USC-Mech v2.0 for the best-fit rate constants. 26

53 Table 3.2: Rate Constant Data for H 2 + OH H 2 O + H. T 5 [K] P 5 [atm] k 1 [cm 3 mol -1 s -1 ] 94 ppm TBHP (and water), 1516 ppm H 2, Ar E E E E E ppm TBHP (and water), 1500 ppm H 2, Ar E E E E E E E E E E E ppm TBHP (and water), 1001 ppm H 2, Ar E E E E E+12 A detailed error analysis was performed to estimate the overall uncertainty in the measured rate constant for reaction (1) at 1228 K. The following contributions were considered: (a) temperature (±1%), (b) OH absorption coefficient (±3%), (c) wavemeter reading in the UV (±0.01 cm -1 ), (d) fitting the data to computed profiles (±5%), (e) locating time-zero (±0.5 µs), (f) the rate constant for CH 3 + OH CH 2 (s) + H 2 O 27

54 (uncert. factor = 2), (g) the rate constant for C 2 H 6 (+ M) CH 3 + CH 3 (+ M) (±20%), and (h) the rate constant for CH 3 COCH 3 + OH CH 3 COCH 2 + H 2 O (±28%). As demonstrated in Figure 3.4, the individual error sources were introduced independently (within the estimated positive and negative bounds of their 2σ uncertainties) and their effects on the rate constant for the title reaction were studied. These uncertainties were combined in a root-sum-squared method to give an overall (2σ) uncertainty of ±17% at 1228 K. Similar error analysis was conducted for k 1 at 972 K, and the overall (2σ) uncertainty was also estimated to be ±17%. Figure 3.4: Uncertainty analysis for the rate constant of H 2 + OH H 2 O + H at 1228 K and 1.29 atm. Figure 3.5 shows the Arrhenius plot for the present rate constant measurements of reaction (1) at T = K and P = atm, along with the non-arrhenius expression originally proposed by Michael and Sutherland [18]. The measured values from the present study can be expressed in Arrhenius form as k 1 (T) = exp(- 3518/T) cm 3 mol -1 s -1 over K. As illustrated in Figure 3.5, the current data have a relatively low scatter (<7%). In the present study, three different mixture compositions were utilized to demonstrate that the inferred rate constants are not strongly 28

55 dependent on any secondary chemistry contributions, and the measured values from these mixtures are consistent with each other. It is interesting to note that the present measurements are in excellent agreement with the non-arrhenius expression proposed by Michael and Sutherland (within ±6%). Figure 3.5: Arrhenius plot for H 2 + OH (k 1 ) at temperatures above 833 K. 3.4 Comparison with Earlier Work Figure 3.6 presents the current data along with some earlier measurements of reaction (1) at temperatures above 833 K. Note that the present measurements have a much lower scatter (<7%) than the previous work. Frank and Just [17] investigated the rate constant for reaction (1) using the test mixtures with a few ppm N 2 O and ppm H 2 and O 2 in Ar behind reflected shock waves over K. In their experiments, the thermal dissociation of N 2 O upon shock-heating was used to generate O-atoms, followed by the reaction of O + H 2 OH + H to produce H-atoms and OH radicals. These OH radicals would then react with H 2 through reaction (1), and atomic resonance absorption spectrometry (ARAS) was used to monitor H- and O-atom concentration profiles simultaneously. These measured species time histories were also 29

56 quite sensitive to another major reaction (H + O 2 OH + O), which was one of the targeted reactions in their study. Their estimated uncertainty limit for k 1 was found to be less than ±40%. Michael and Sutherland [18] examined the reverse rate constant using the test mixtures of 0.1-1% H 2 O in Ar behind reflected shock waves over K. H-atoms were generated in the post-shock regions using the flash photolysis of H 2 O and were monitored using ARAS. In the present study, their measured values were also converted to the forward rate constants using the revised equilibrium constants (with the revised standard enthalpy of formation for OH from Herbon et al. [23-24]), as shown in Figure 3.6. The new calculated values for reaction (1) are approximately 15% lower than the old values evaluated by Michael and Sutherland. Such differences are solely due to the equilibrium constant evaluations. Nevertheless, their values for reaction (1) have a relatively large experimental scatter (within a factor of 2). Similarly, Davidson et al. [19] measured the reverse rate using the test mixtures of % H 2 O in Ar behind reflected shock waves over K. An ArF excimer laser at nm was employed to photo-dissociate a small amount of H 2 O in order to generate H-atoms and OH radicals, and a cw, narrow-linewidth ring dye laser at nm was then used to monitor the temporal evolution of OH radicals. The errors of their measurements varied from ±40% at 1600 K to ±12% at 2500 K. Their measured values were also converted to the forward rates using the revised equilibrium constants. As illustrated in Figure 3.6, the data from Frank and Just [17], Michael and Sutherland [18], and Davidson et al. [19] are in good agreement with each other (within their experimental scatter). Oldenborg et al. [20] also conducted direct rate constant measurements of reaction (1) using the laser photolysis / heated flow cell technique and using laser-induced fluorescence method to monitor the temporal profiles of OH decays at K. Their experiments were performed under pseudo-first-order kinetic conditions with an excess of H 2. In addition, Ravishankara et al. [9] studied reaction (1) using the flash photolysis resonance fluorescence technique (under pseudo-first-order kinetic conditions) in a heated quartz cell over K, and their high-temperature measurements (at 960 and 1050 K) are ~20% higher than the measurements from Oldenborg et al. [20]. Moreover, Krasnoperov and Michael [21] reexamined reaction (1) 30

57 using a test mixture of ~5 ppm TBHP with 404 ppm H 2 in krypton behind reflected shock waves at lower temperatures ( K). TBHP was utilized as the OH precursor in their study, and a novel multi-pass absorption spectrometric detection method (with a MW discharge driven resonance OH lamp) was employed to measure OH species profiles. Although their experimental scatter is slightly high (±25%), their measured values are generally in good agreement with Oldenborg et al. [20] and Ravishankara et al. [9]. Concurrently, the present measurements are in excellent agreement with all previous studies. Figure 3.6 also shows the values of k 1 employed in three different comprehensive reaction mechanisms: GRI-Mech 3.0 [22], USC-Mech v2.0 [72], and Hong et al. [83]. The present measurements agree well with the values from GRI-Mech 3.0 and Hong et al. (within ±6%). However, the values of k 1 from USC-Mech v2.0 are ~20% lower than the present measurements. Concurrently, Nguyen et al. [84] computed the rate constant for the title reaction with semi-classical transition state theory (SCTST), which implemented non-separable coupling among all degrees of freedom (including the reaction coordinate) in the transition state region and multi-dimensional quantum mechanical tunneling along the curved reaction path. Their theoretical calculation (the black dashed line in Figure 3.6) is also in excellent agreement with the present and previous experimental studies, and their calculation is nearly indistinguishable from the rate constant adopted in GRI- Mech

58 Figure 3.6: Comparison with previous studies at temperatures above 833 K. 3.5 Summary The rate constant for the reaction of H 2 + OH H 2 O + H was studied behind reflected shock waves over the temperature range of K at pressures of atm using OH laser absorption. The current high-temperature data can be expressed in Arrhenius equation as k 1 (T) = exp(-3518/t) cm 3 mol -1 s -1 over the temperature range studied. A detailed error analysis was carried out with the consideration of both experimental and secondary chemistry contributions, and the overall (2σ) uncertainties in k 1 were found to be ±17% at 972 and 1228 K. Note that the experimental scatter from the present study is less than 7%, which is much lower than in previous work [17-21]. The present data are consistent with the previous measurements from Frank and Just [17], Michael and Sutherland [18], Davidson et al. [19], Oldenborg et al. [20], and Krasnoperov and Michael [21]. Additionally, the present measurements are in excellent agreement with the non-arrhenius expression from GRI-Mech 3.0 [22] and the recent theoretical calculation using semi-classical transition state theory (SCTST) from Nguyen et al. [84]. 32

59 Chapter 4 Multi-Species Time History Measurements during High- Temperature Acetone and 2- Butanone Pyrolysis 4.1 Introduction The pyrolysis of acetone (IUPAC name: propanone) has been studied by many researchers, particularly at temperatures below 1000 K. At temperatures above 1000 K, five recent studies have been performed. Capelin et al. [85] examined acetone pyrolysis utilizing flash vaporization into a heated reaction chamber, and suggested a pyrolysis mechanism with CH 3 COCH 3 (+ M) CH 3 COCH 2 + H (+ M) as the initiation reaction, based on the product distributions from gas chromatography. Ernst et al. [86] studied acetone pyrolysis using a shock tube and UV laser absorption technique. They recommended a similar mechanism with a different initiation reaction: CH 3 COCH 3 (+ M) CH 3 + CH 3 CO (+ M) (2) Similarly, Sato and Hidaka [87] investigated acetone pyrolysis and oxidation in a shock tube; they evaluated the rate constant (k 2 ) for reaction (2) by monitoring acetone concentrations using laser absorption at 200 nm and 3.39 µm. Saxena et al. [88] performed direct rate constant measurements of reaction (2) using laser-schlieren technique. And finally, Pichon et al. [89] developed a detailed kinetic mechanism for acetone, which was validated against their flame speed and ignition delay time measurements. 33

60 In contrast to acetone, fewer 2-butanone studies have been performed. Early lowtemperature 2-butanone oxidation static reactor studies were conducted by Bardwell and Hinshelwood [90-93]. Decottignies et al. [94] investigated 2-butanone oxidation using laminar premixed methane/air flames doped with different amounts of 2-butanone. Based on the product distributions from gas chromatography, they postulated a kinetic mechanism and suggested three initial decomposition pathways: 2-Butanone (+ M) C 2 H 5 + CH 3 CO (+ M) (3a) 2-Butanone (+ M) CH 3 + C 2 H 5 CO (+ M) (3b) 2-Butanone (+ M) CH 3 + CH 3 COCH 2 (+ M) (3c) with channel (3b) as the dominant pathway. Similarly, Serinyel et al. [95] developed a comprehensive kinetic mechanism, which was validated against their shock tube ignition delay times. However, they suggested that channel (3a) is the primary initial 2-butanone decomposition channel. This chapter presents high-temperature pyrolysis studies of acetone and 2- butanone behind reflected shock waves using laser absorption methods to measure time histories of five species: ketone, CO, CH 3, CH 4, and C 2 H 4. These measurements were used to determine the decomposition rate constants for both acetone and 2-butanone. 4.2 Experimental Details Mixture Preparation Test mixtures were prepared manometrically in a 40 liter stainless steel tank heated uniformly to 50 o C and mixed with a magnetically-driven stirring vane for at least 2 hours prior to the experiments. The mixture compositions were 0.25%, 1%, and 1.5% ketone in argon. Research grade argon (99.999%) from Praxair was used with the ACS spectrophotometric grade acetone ( 99.5%) and the CHROMASOLV grade 2-butanone ( 99.7%) from Sigma Aldrich. All liquid chemicals were purified using a freeze-pumpthaw procedure to remove dissolved volatiles and air. 34

61 4.2.2 Species Absorption Coefficient Evaluations Individual species time histories for dilute fuel mixtures (0.25% acetone or 2- butanone in Ar) were determined from Beer s law using a constant absorption coefficient value for each species, evaluated at the initial reflected shock temperatures and pressures. On the other hand, for the higher fuel concentration mixtures (1% and 1.5% ketone in Ar), the changes in temperature and pressure that occur during pyrolysis slightly perturbed the absorption coefficients of CO, C 2 H 4 and CH 4 up to 10%, 7% and 15%, respectively, for experiments with initial temperatures higher than 1400 K, and the use of constant absorption coefficients in determining the species mole fractions was not valid. To determine the experimental species time histories more quantitatively, the temperature and pressure profiles were first calculated by solving the energy equation under the standard constant energy (U) and volume (V) assumption (using CHEMKIN PRO [71]). The species mole fraction time histories were then inferred from the measured absorption data using values of the absorption coefficients evaluated at the simulated T and P. 4.3 Results and Discussion Acetone Pyrolysis High-temperature acetone pyrolysis was investigated using five species time history measurements over the temperature range of K at pressures around 1.6 atm. In the present study, the CHEMKIN PRO package [71] was used to simulate all species time histories under the standard constant energy and volume assumption. Fig. 4.1 shows the CO sensitivity analysis for the mixture of 0.25% acetone in argon at 1393 K and 1.55 atm simulated using the Pichon et al. mechanism of NUI Galway [89]. The CO time history is predominantly sensitive to reaction (2), with some minor interference from the reactions of C 2 H 6 (+ M) CH 3 + CH 3 (+ M), C 2 H 6 + H C 2 H 5 + H 2, and CH 3 + CH 3 C 2 H 5 + H. Similar result for the CO sensitivity analysis was obtained for the mixture of 1% acetone in Ar. 35

62 Figure 4.1: CO sensitivity for 0.25% acetone in Ar using the Pichon et al. mechanism [89]. Fig. 4.2 shows the CO rate of production (ROP) analysis for the mixture of 0.25% acetone in Ar at 1393 K and 1.55 atm simulated using the Pichon et al. mechanism [89]. The ROP analysis reveals that the primary CO formation pathway is via the reaction of CH 3 CO (+ M) CH 3 + CO (+ M) over the time frame of the experiment, and the acetyl (CH 3 CO) radical is directly formed from reaction (2). In particular, the CH 3 CO radical is rather short-lived at T > 1100 K; once it is formed, it will decompose nearinstantaneously to form a CH 3 radical and a CO molecule. According to the CO sensitivity and ROP analysis, the measured CO time histories can then be used to infer the acetone dissociation rate constant (k 2 ) over the temperature range of K at pressures of atm. The measured CO mole fractions were best-fit with the simulated profiles by varying the value of k 2 in the detailed kinetic mechanism of Pichon et al. 36

63 Figure 4.2: CO rate of production (ROP) plot for 0.25% acetone in Ar using the Pichon et al. mechanism [89]. Fig. 4.3 shows a sample measured CO concentration time history during acetone pyrolysis for the mixture of 0.25% acetone in argon at 1393 K and 1.55 atm, and the bestfit rate constant (k 2 ) of 3516 s -1, along with perturbations of ±30%, using the Pichon et al. mechanism. The best-fit simulation curve is virtually indistinguishable from the data. Figure 4.3: Sample CO time histories: measured and calculated values. 37

64 The present rate constant measurements (k 2 ) of reaction (2) (inferred from the CO profiles) at pressures near 1.6 atm are summarized in an Arrhenius diagram, Fig. 4.4, along with three recent evaluations [87-89]. It should be noted that the acetone concentration was varied from 0.25% to 1% in order to confirm that the current rate constant measurements were weakly dependent on any secondary reaction effects, and the measured values from these two mixtures were consistent with each other. Using a least-squares fit, the acetone dissociation rate constant can be expressed in Arrhenius form as k 2 = exp(-69.3 [kcal/mol]/rt) s -1 over the temperature range of K at pressures around 1.6 atm (see the dashed line in Fig. 4.4). The primary contributions to uncertainties in the rate constant were: temperature (±10%), CO absorption cross-section (±5%), fitting the data to computed profiles (±5%), and uncertainties resulting from secondary reactions (±5%), giving an overall uncertainty in k 2 of ±25%. Similar experiments performed at different pressures ( atm and 5 atm) show only a very weak pressure-dependence for k 2, confirming the assumption that these measurements were performed close to the high-pressure limit. In addition, Table 4.1 summarizes the rate constant measurements of acetone dissociation reaction that are inferred from the measured CO profiles over K at pressures of atm. Figure 4.4: Summary of acetone dissociation rate constant (k 2 ). 38

65 Similarly, as is evident in Fig. 4.5, the acetone sensitivity analysis shows that the acetone concentration time history is strongly sensitive to reaction (2), with some minor interference from the reactions of C 2 H 6 (+ M) CH 3 + CH 3 (+ M), CH 3 + CH 3 C 2 H 5 + H, and CH 3 COCH 3 + H CH 3 COCH 2 + H 2. Hence, the measured acetone time histories can also be used to determine the acetone dissociation rate constant (k 2 ). Figure 4.5: Acetone sensitivity for 1% acetone in Ar using the Pichon et al. mechanism [89]. Fig. 4.6 shows the measured acetone time histories, along with the simulations from the original Pichon et al. mechanism. Similar to the previous analysis based on the CO profiles, the experiments suggest much faster acetone removal rates than the original Pichon et al. mechanism. To infer the acetone dissociation rate constant, the measured acetone concentration time histories were also best-fit with the simulated profiles by varying the value of k 2 in the detailed kinetic mechanism of Pichon et al. [89], as depicted in Fig Table 4.1 also includes the rate constant measurements of reaction (2) that are inferred from the acetone profiles over K at pressures of atm. The inferred values based on the acetone profiles are consistent with those from the CO profiles. Note that the measurements seem to experience some slight non-arrhenius curvature at higher temperatures, as illustrated in Fig By compiling the rate constant 39

66 measurements based on the CO and acetone time histories, the acetone dissociation rate constant can be represented by the following three-parameter equation: k 2 ( atm) = T exp(-44,236/t) s -1 over the temperature range of K. Figure 4.6: Acetone time histories for 1% acetone in Ar: measured and simulated values. As illustrated in Fig. 4.4, the measurements of Saxena et al. [88] and the hightemperature measurements of Sato and Hidaka [87] are in good agreement with the current results (within 30% at T > 1450 K) at pressures near 1.6 atm. However, at lower temperatures, the determination from Sato and Hidaka departs significantly from the current study. For instance, their determination is at least three times slower than the measured rate constant from the current study at T < 1250 K. In addition, the theoretical estimate by Pichon et al. [89] using a chemical activation formulation based on Quantum Rice-Ramsperger-Kassel theory k 2, theoretical (1.6 atm) = T -8 exp(-43,400/t) s -1 recovers an activation energy similar to the current measurements, but its A-factor is approximately three times lower than the measured A-factor from the current study, as shown in Fig

67 Table 4.1: Summary of acetone unimolecular dissociation rate constant data. T 5 [K] P 5 [atm] k 2 [s -1 ] Initial Mixture: 0.25% Acetone / Ar (from CO) E E E E E E E E E E E E E+03 Initial Mixture: 1% Acetone / Ar (from CO) E E E E+04 Initial Mixture: 1% Acetone / Ar (from CH 3 COCH 3 ) E E E E E+05 Other species (CH 3, C 2 H 4, and CH 4 ) were also measured during pyrolysis, and comparisons with the simulations demonstrate how the current evaluated k 2 significantly improves the overall performance of the detailed mechanism. Fig. 4.7 displays the measured CH 3 time histories during acetone pyrolysis, along with the computed values from the original Pichon et al. mechanism and the modified mechanism (with our revised value for k 2 ). The computed CH 3 peak values from the original Pichon et al. mechanism are approximately half of the measured values. Additionally, the CH 3 sensitivity analysis reveals that CH 3 time histories are strongly sensitive to two reactions: CH 3 COCH 3 (+ M) 41

68 CH 3 + CH 3 CO (+ M) (reaction (2)) and the relatively well-established CH 3 + CH 3 (+ M) C 2 H 6 (+ M) [80-81]. Consequently, using the revised (higher) values for k 2 brings the simulations from the Pichon et al. mechanism into closer agreement with the measured CH 3 time histories. Note that the uncertainty in the CH 3 concentration during the first 100 µs was approximately ±20%, which was mainly attributed to the uncertainties in the absorption coefficient and the interference subtraction scheme. The CH 3 plateau levels (after 500 µs) had a much larger uncertainty of ±30%, owing to larger interference absorption (from other intermediate products, such as C 2 H 4 ). Figure 4.7: CH 3 time histories for 0.25% acetone in Ar: measured and calculated values. Ethylene is an important major product formed during acetone pyrolysis. Fig. 4.8 shows the measured C 2 H 4 time histories, along with the simulations from the original and modified Pichon et al. mechanisms. The original Pichon et al. mechanism clearly fails to predict the formation rates and the ultimate yields of ethylene. During acetone pyrolysis, C 2 H 4 is mainly formed from the direct competition between two reactions: CH 3 + CH 3 (+ M) C 2 H 6 (+ M) and CH 3 + CH 3 C 2 H 5 + H, immediately followed by C 2 H 5 (+ M) C 2 H 4 + H (+ M). The higher formation rates of CH 3 in the modified mechanism (via the higher values for k 2 ) significantly improve the simulations at early times. 42

69 Figure 4.8: C 2 H 4 time histories for 1% acetone in Ar: measured and calculated values. Methane is another major product formed during acetone pyrolysis. Fig. 4.9 shows the measured CH 4 time histories, along with the simulations from the modified Pichon et al. mechanisms (i.e., with the revised values for k 2 ) with two different CH 3 COCH 3 + CH 3 reaction rate constants adopted by Pichon et al. [89] and Saxena et al. [88]. The original Pichon et al. mechanism significantly underpredicted the CH 4 time histories by at least a factor of 4, and the modified Pichon et al. mechanism v1 (with revised k 2 and original acetone + CH 3 rate constant used by Pichon et al.) was still not able to capture the CH 4 formation rates at T < 1400 K. However, at T = 1556 K, the modified mechanism v1 predicted the CH 4 concentration reasonably well. Additionally, the CH 4 sensitivity analysis (not presented here) shows the importance of reaction (2), and the reactions of CH 3 plus CH 3, C 2 H 6, and CH 3 COCH 3. There are only limited uncertainties in the CH 3 + CH 3 and the CH 3 + C 2 H 6 reaction rate constants as given by Baulch et al. [96] (±20%). Thus, the rate constant for the reaction of CH 3 COCH 3 + CH 3 CH 3 COCH 2 + CH 4 in the original Pichon et al. mechanism k Pichon et al. = exp(-9741 [cal/mol]/rt) cm 3 mol -1 s -1 is likely to be too slow at T < 1400 K. In particular, the Saxena et al. mechanism uses a significantly different value for the rate constant of CH 3 COCH 3 + CH 3 (see Fig. 4.10), which is: 43

70 k Saxena et al. = T 4 exp(-8290 [cal/mol]/rt) cm 3 mol -1 s -1 The rate constant used by Saxena et al. [88] is at least three times faster than the value of the rate constant adopted by Pichon et al. [89] at 1250 K, and is approximately six times faster at 1538 K. When the rate constant from Saxena et al. is used in the modified Pichon et al. mechanism (labeled as modified mechanism v2), the simulated time histories show excellent agreement with the measured time histories at T < 1400 K. However, the modified mechanism v2 seems to overpredict the ultimate methane yield at T = 1556 K. This reveals that the activation energy of the acetone + CH 3 reaction likely still requires some fine adjustment. Figure 4.9: CH 4 time histories for 1.5% acetone in Ar: measured and calculated values. 44

71 Figure 4.10: Arrhenius plot of the rate constants for the reaction of CH 3 COCH 3 + CH 3 CH 3 COCH 2 + CH 4 from Saxena et al. [88], Sato and Hidaka [87], and Pichon et al. [89] Butanone Pyrolysis High-temperature 2-butanone pyrolysis was studied using multi-species time history measurements over the temperature range of K at pressures around 1.5 atm. Fig shows the measured 2-butanone time histories during 2-butanone pyrolysis, along with the simulations from the original Serinyel et al. mechanism [95]. Serinyel et al. postulated three initial 2-butanone decomposition pathways: 2-Butanone (+ M) C 2 H 5 + CH 3 CO (+ M) (3a) 2-Butanone (+ M) CH 3 + C 2 H 5 CO (+ M) (3b) 2-Butanone (+ M) CH 3 + CH 3 COCH 2 (+ M) (3c) As illustrated in Fig. 4.12, the 2-butanone sensitivity analysis reveals that the 2- butanone time history is predominantly sensitive to its three initial decomposition pathways, with channel (3a) as the primary decomposition channel. In addition, there is some minor interference from the reactions of 2-butanone + H CH 3 CHCOCH 3 + H 2 and CH 3 CHCOCH 3 CH 3 CHCO + CH 3. To determine the overall initial 2-butanone decomposition rates, the measured 2-butanone time histories were best-fit with the simulated profiles by adjusting the rate constants for channels (3a)-(3c), without 45

72 modifying their branching ratios. The best-fit simulated 2-butanone time histories are also shown on Fig At all temperatures, the modified Serinyel et al. mechanism captures the initial 2-butanone decomposition rates very accurately, at least for the first 500 µs. (Note that some simulations are nearly indistinguishable from the data traces.) Figure 4.11: 2-Butanone time histories for 1% 2-butanone in Ar: measured and simulated values. Figure 4.12: 2-Butanone sensitivity for 1% 2-butanone in Ar. The overall 2-butanone decomposition rate constant measurements (k 3 = k 3a + k 3b + k 3c ) are plotted on Fig. 4.13, along with the estimated values from the Serinyel et al. mechanism. Using a least-squares fit, the overall 2-butanone decomposition rate constant 46

73 was found to be k 3 = exp(-63.1 [kcal/mol]/rt) s -1 over the temperature range of K at pressures around 1.5 atm (see the dashed line in Fig. 4.13). (Note that the current data are the first direct high-temperature rate constant measurements for the initial 2-butanone decomposition.) The major contributions to uncertainties in the rate constants were: temperature (±10%), 2-butanone absorption cross-section (±5%), fitting the data to computed profiles (±5%), and uncertainties resulting from secondary reactions (±15%), giving an overall uncertainty in k 3 of ±35%. In addition, Table 4.2 summarizes the overall 2-butanone decomposition rate constant measurements over K at pressures near 1.5 atm. The effect of the branching ratios of the initial 2-butanone decomposition pathways on the measured rate constant was investigated by perturbing the branching ratio of channel (3a) from 0.70 to 0.50 at 1361 K, and no significant difference on the simulated 2-butanone time histories was found. In the following discussion, the original branching ratios of the 2-butanone decomposition pathways used by Serinyel et al. were retained. Our measured value for k 3 is approximately 30% faster than Serinyel et al. at 1119 K, and is approximately 100% faster at 1412 K. Our inferred rate constants for channels (3a)-(3c) at 1.5 atm can be expressed as follows: k 3a = exp(-59.9 [kcal/mol]/rt) s -1 k 3b = exp(-75.3 [kcal/mol]/rt) s -1 k 3c = exp(-72.9 [kcal/mol]/rt) s -1 Figure 4.13: Arrhenius plot for overall 2-butanone decomposition rate constant (k 3 ). 47

74 Table 4.2: Summary of overall 2-butanone decomposition rate constant data. T 5 [K] P 5 [atm] k 3 [s -1 ] Initial Mixture: 1% 2-Butanone / Ar E E E E E E E+04 Fig shows the measured CH 3 time histories during 2-butanone pyrolysis, along with the simulations from the original and modified Serinyel et al. mechanisms. The modified Serinyel et al. mechanism captures the initial CH 3 formation rates more closely, but the simulated CH 3 peak values are still underpredicted. According to the CH 3 sensitivity analysis, the CH 3 time histories are mainly sensitive to channels (3a)-(3c) and the relatively well-established reaction of CH 3 + CH 3 (+ M) C 2 H 6 (+ M) [80-81]. Thus, discrepancies between the measured and simulated CH 3 peak values may be attributed, at least partially, to uncertainties in the relative branching ratios of the initial 2-butanone decomposition pathways. (Note that the agreement here is not sacrificed by the subsequent addition of a methyl ketene decomposition channel, which will be discussed in the following section.) Figure 4.14: CH 3 time histories for 0.25% 2-butanone in Ar: measured and calculated values. 48

75 Similar to acetone pyrolysis, CO is another important stable species formed during 2-butanone pyrolysis. Fig shows the CO rate of production (ROP) analysis for the mixture of 1% 2-butanone in Ar at 1292 K and 1.58 atm simulated using the Serinyel et al. mechanism with the revised k 3, and the ROP analysis reveals that CO is mainly generated via two reaction pathways over the time frame of the experiment, which are CH 3 CO (+ M) CH 3 + CO (+ M) and C 2 H 5 CO C 2 H 5 + CO. The acetyl (CH 3 CO) radical can be formed via 2-butanone (+ M) C 2 H 5 + CH 3 CO (+ M) (channel (3a)) and 2-butanone + H CH 2 CH 2 COCH 3 + H 2, followed by the fuel radical decomposition (CH 2 CH 2 COCH 3 C 2 H 4 + CH 3 CO). Additionally, the propionyl (C 2 H 5 CO) radical can be formed via 2-butanone (+ M) CH 3 + C 2 H 5 CO (+ M) (channel (3b)). Figure 4.15: CO rate of production (ROP) plot for 1% 2-butanone in Ar using the original Serinyel et al. mechanism (with the revised k 3 ). Fig shows the measured and simulated CO time histories during 2-butanone pyrolysis. Based on the measured 2-butanone and CO time histories at T 1292 K, the measurements suggest ~87% conversion to CO from 2-butanone, while the original Serinyel et al. mechanism [95] predicts only about 57% conversion to CO. The model with the revised overall 2-butanone decomposition rate constant (k 3 ) generates slightly higher CO concentration (~67%), but still well below that measured. According to the simulations, the major species containing O atoms are 2-butanone, CO, and CH 3 CHCO 49

76 (methyl ketene), and the model seems to predict significant amounts of methyl ketene formed during 2-butanone pyrolysis. This implies that some of the CO formation pathways might be incomplete in the model, particularly the methyl ketene submechanism. Methyl ketene is produced through the reaction of 2-butanone + H CH 3 CHCOCH 3 + H 2, followed by CH 3 CHCOCH 3 CH 3 CHCO + CH 3. According to the model, the removal pathway of methyl ketene is only through the H-abstraction reaction from methyl ketene (CH 3 CHCO + H C 2 H 5 + CO). Since methyl ketene is not a stable species, it should undergo a unimolecular decomposition process, which is not included in the original Serinyel et al. mechanism [95]. In the present analysis, a methyl ketene decomposition pathway (CH 3 CHCO (+ M) C 2 H 4 + CO (+ M)) was incorporated, and the corresponding rate constant was assumed to be the same as the value for ketene decomposition (CH 2 CO (+ M) CH 2 + CO (+ M)). The modified mechanism (with revised k 3 and added methyl ketene decomposition reaction) simulates the CO concentrations rather accurately, as illustrated in Fig Figure 4.16: CO time histories for 1% 2-butanone in Ar: measured and calculated values. Fig displays the measured C 2 H 4 time histories, along with the simulations from the original and modified Serinyel et al. mechanisms. The measurement suggests the ethylene yield (defined as the ratio of the long-time C 2 H 4 concentration to the initial 50

77 2-butanone concentration) to be ~0.88 at 1412 K, while the original Serinyel et al. mechanism predicts the yield to be ~0.73 and the modified mechanism predicts the yield of ~0.87. In general, the modified mechanism is able to accurately simulate the ultimate yields of C 2 H 4. (Note that the simulated C 2 H 4 time history from the modified mechanism lies exactly on top of the measured time history at 1252 K.) The C 2 H 4 sensitivity analysis shows the significance of the initial 2-butanone decomposition pathways (channels (3a)- (3c)) and the H-abstraction reactions from 2-butanone. C 2 H 4 is initially formed through the following processes: (i) 2-Butanone (+ M) C 2 H 5 + CH 3 CO (+ M), followed by C 2 H 5 (+ M) C 2 H 4 + H (+ M); (ii) 2-Butanone + H CH 2 CH 2 COCH 3 + H 2, followed by CH 2 CH 2 COCH 3 C 2 H 4 + CH 3 CO. Hence, fine refinement on the branching ratios of the initial 2-butanone decomposition pathways and the H-abstraction reaction rate constants appears needed to perfectly match the initial formation rates of C 2 H 4. Figure 4.17: C 2 H 4 time histories for 1% 2-butanone in Ar: measured and calculated values. In addition to CO and C 2 H 4, methane is another major product formed during 2- butanone pyrolysis. A plot of the measured methane time histories, along with the 51

78 computed values from the modified Serinyel et al. mechanism, is illustrated in Fig As compared to the measurements, the modified mechanism underpredicts the methane concentrations by at least a factor of 2 at temperatures less than 1400 K. However, at temperatures higher than 1400 K, the modified mechanism is able to capture the methane formations reasonably well. Methane is mainly formed through the H-abstraction reactions from 2-butanone by CH 3 radicals, which are: 2-Butanone + CH 3 CH 2 CH 2 COCH 3 + CH 4 2-Butanone + CH 3 CH 3 CHCOCH 3 + CH 4 2-Butanone + CH 3 C 2 H 5 COCH 2 + CH 4 Such huge differences between the measurements and simulations may be caused by the inaccurate model predictions for CH 3 concentrations and the uncertainties in the rate constants for 2-butanone + CH 3 reactions, particularly their activation energies. Further increases in the rate constants for 2-butanone + CH 3 reactions at temperatures less than 1400 K appear needed in order to improve the model predictions on methane. Figure 4.18: CH 4 time histories for 1.5% 2-butanone in Ar: measured and calculated values. 52

79 4.4 Summary High-temperature acetone and 2-butanone pyrolysis was investigated behind reflected shock waves using multi-species time history measurements (acetone/2- butanone, CO, CH 3, C 2 H 4, and CH 4 ). Direct determinations of the acetone dissociation rate constant (k 2 ) and the overall 2-butanone dissociation rate constant (k 3 = k 3a + k 3b + k 3c ) were made by taking advantage of the measured species time histories for CO (and acetone) and 2-butanone, respectively. In the 2-butanone pyrolysis system, an analysis of the O-atom balance based on the simulated and measured 2-butanone and CO time history measurements revealed pooling of methyl ketene in the simulations. The addition of the methyl ketene decomposition pathway to remove this pooling significantly improved the mechanism s performance. Further improvement in the 2-butanone mechanism will require a better understanding and refinement of the branching ratios of the initial 2-butanone decomposition pathways and the rate constants for the H-abstraction reactions from 2- butanone. 53

80 54

81 Chapter 5 Shock Tube Measurements of 3-Pentanone Pyrolysis and Oxidation 5.1 Introduction In contrast to acetone and 2-butanone, very little experimental data are available for high-temperature 3-pentanone combustion studies. Three studies of this fuel are of note. Davidson et al. [97] measured shock tube ignition delay times for a series of oxygenated fuels, including 3-pentanone, over temperatures of K and a pressure of ~1.8 atm. From their experiments, they concluded that 3-pentanone has much faster ignition delay times than found for acetone, n-pentane, methyl butanoate, and butanal. Similarly, Serinyel et al. [98] performed shock tube ignition delay time measurements in the temperature range K, pressures near 1 atm, and equivalence ratios of for mixtures of % 3-pentanone in O 2 /argon. They also conducted laminar flame speed measurements in a spherical bomb for mixtures of 3- pentanone in air with various equivalence ratios at an initial temperature of ~305 K and an initial pressure of 1 atm. Through their flame speed measurements, they concluded that 3-pentanone has higher reactivity than acetone and 2-butanone. Finally, Hong et al. [99] examined the influence of oxygenates (such as 3-pentanone) on soot formation during fuel rich n-heptane oxidation at temperatures of K and pressures of atm. A significant reduction in the overall soot yield was discovered with the addition of small quantities of oxygenates. 55

82 In this chapter, we present high-temperature pyrolysis and oxidation studies of 3- pentanone behind reflected shock waves using laser absorption methods to measure time histories of six species: 3-pentanone, CH 3, CO, C 2 H 4, H 2 O, and OH. In addition, 3- pentanone oxidation behavior was compared with the oxidation behavior of two other ketones, acetone and 2-pentanone, by examining their ignition delay times and OH time histories. 5.2 Experimental Details Mixture Preparation Test mixtures were prepared manometrically in a 40 liter stainless steel tank heated uniformly to 50 o C and mixed with a magnetically driven stirring vane for at least 2 hours prior to the experiments. Research grade (99.999%) gases (from Praxair) and ReagentPlus grade ( 99%) 3-pentanone (from Sigma-Aldrich), which was further treated using a freeze-pump-thaw procedure, were used in mixture preparation. The mixture compositions from this study are summarized in Table 5.1, along with the measured species for the corresponding mixtures. For the lower fuel concentration mixtures (<0.25% 3-pentanone), a double-dilution method was used in mixture preparation to allow for more accurate mixture compositions. Mix # Table 5.1: Summary of test gas mixture compositions and measured species. Gas Compositions Species Time histories τ ign 3-Pent. O 2 Ar 3-Pent. CH 3 CO C 2 H 4 OH H 2 O A 1.00% % x x x B 0.25% % x x C 0.10% % x D 0.040% 0.280% 99.68% x x x E 0.040% 0.560% 99.40% x x x F 0.075% 0.525% 99.40% x x x G 0.571% 4.00% 95.43% x H 0.286% 4.00% 95.71% x I 0.875% 12.25% 86.88% x 56

83 5.2.2 Species Absorption Coefficient Evaluations Because of the endothermic nature of the pyrolysis reaction, there is a temperature drop in the reacting test gas mixture during the experiment, which increases with the initial 3-pentanone mole fraction. This change in temperature can perturb (generally increase) the absorption coefficients of individual species, and hence perturb the conversion of measured absorbance to mole fraction. More accurate species mole fraction time histories are obtained by accounting for this effect rather than assuming a constant coefficient evaluated at the initial temperature. To determine these approximate time-varying absorption coefficients, the temperature and pressure profiles were calculated using the Serinyel et al. mechanism of NUI Galway [98] under either constant energy (U) and volume (V) constraints or constant enthalpy (H) and pressure (P) constraints (using CHEMKIN PRO [71]). The species mole fraction time histories were then inferred from the measured absorption data using known values of the absorption coefficients evaluated at the simulated T and P. The 3-pentanone mole fraction time history for a 1% 3-pentanone/Ar mixture and an initial temperature of 1248 K and an initial pressure of 1.58 atm was calculated from the measured absorption data using three different approaches: (1) constant absorption coefficient for Beer s law (as has been common in the past), (2) T- and P-dependent absorption coefficients based on constant U, V calculation, and (3) T- and P-dependent absorption coefficients based on constant H, P calculation. The 3-pentanone mole fraction time histories from these three approaches are nearly indistinguishable. Hence, the measured 3-pentanone mole fraction time histories are effectively insensitive to any small variation in temperature and pressure change that is a result of the gasdynamic model used. In this chapter, constant absorption coefficients for Beer s law are thus employed for 3-pentanone time history measurements. On the other hand, the absorption coefficients of CO and C 2 H 4 can increase by up to 10% and 7%, respectively, during the pyrolysis of 1% 3-pentanone in Ar. Clearly, to minimize the effects of temperature drop during pyrolysis, lower fuel concentration mixtures (0.1% or 0.25% 3-pentanone in Ar) are preferred. The drop in temperature for 57

84 the mixture of 0.25% 3-pentanone in Ar is ~30 K (approximately 4 times less than 1% 3- pentanone mixture) at an initial temperature of 1325 K and an initial pressure of 1.60 atm. As shown in Fig. 5.1, the CO mole fraction time history for the mixture of 0.25% 3- pentanone in Ar at 1325 K and 1.60 atm was calculated using absorption coefficients based on three different gas dynamic models. The initial CO formation rates (for the first 400 µs) from these three approaches are very nearly identical, but the ultimate yield from the constant absorption coefficient approach (labeled as method 1 and generally employed in most past studies) is ~2% higher than the yields from the T- and P- dependent absorption coefficient approaches that are based on constant U, V (method 2) and constant H, P (method 3) calculations. To correct for this slight difference at later times, all the CO mole fraction time histories in the present work were calculated using the T- and P-dependent absorption coefficients based on constant U, V calculations; the estimated uncertainty in the measured yields using this approach is ±2-3%. Similarly, all the C 2 H 4 time histories were also calculated from the measured absorption data using the T- and P-dependent absorption coefficients based on constant U, V calculations. Figure 5.1: Comparison of CO mole fraction time histories at 1325 K and 1.60 atm with different absorption coefficients in Beer s law. During 3-pentanone oxidation, very dilute mixtures (e.g., 400 ppm 3-pentanone) are used for OH and H 2 O species time history measurements in order to minimize the 58

85 rapid energy release at the time of ignition, which increases the temperature (and reduces the absorption coefficient). As illustrated in Fig. 5.2, for the mixture of 400 ppm 3- pentanone with 0.28% O 2 in Ar, the early-time features of OH obtained from the measured absorption data using three different absorption coefficient approaches are effectively identical. Similar results can be observed from the H 2 O time history profiles. This is particularly important because these early-time features are unique to individual fuels, as will be discussed in the later section. However, the final plateau levels of OH and H 2 O from method 1 are lower than the levels from methods 2 and 3 by 4% and 2%, respectively. This is mainly due to the fact that there is a temperature rise of ~50 K at the time of ignition, and the use of a constant absorption coefficient for Beer s law is not strictly valid at these later times. Interestingly, the final plateau levels of OH and H 2 O obtained from methods 2 and 3 are indistinguishable. Hence, the temperature corrections on the measured species are independent of the specific gasdynamic model (const. U, V or const. H, P). Similar to CO and C 2 H 4 measurements (as described above), all the measured OH and H 2 O time histories are corrected using the T- and P-dependent absorption coefficients based on constant U, V calculations. Figure 5.2: Comparison of OH mole fraction time histories at 1486 K and 1.52 atm with different absorption coefficients in Beer s law. The OH mole fractions by constant U, V and constant H, P are virtually indistinguishable for OH. 59

86 5.3 Results and Discussion Pentanone Pyrolysis A high-temperature 3-pentanone pyrolysis study was performed behind reflected shock waves using four species time history measurements (fuel, CH 3, CO, and C 2 H 4 ) over K at a pressure of ~1.6 atm. The test mixtures were 0.1% to 1% 3- pentanone in balance argon. In the present study, the CHEMKIN PRO package [71] was used to simulate all species time histories under the standard constant energy and volume assumption (constant U, V), and the Serinyel et al. mechanism of NUI Galway [98] was chosen as the base mechanism. To the best of our knowledge, the Serinyel et al. mechanism is the only available detailed mechanism in the literature that is suitable for high temperature 3-pentanone combustion. The sub-mechanism of 3-pentanone was developed by Serinyel et al. and implemented into the well-established C4 mechanism of NUI Galway [100]. In particular, the rate constants for 3-pentanone unimolecular decomposition reactions were estimated in the reverse direction, and their high-pressure limit values were further assumed and treated using Quantum Rice-Ramsperger Kassel (QRRK) theory with a master equation analysis to include the pressure fall-off effects. In addition, the detailed mechanism of Serinyel et al. was then validated against their ignition delay time and laminar flame speed measurements. Fig. 5.3 shows the measured 3-pentanone time histories during pyrolysis of 1% 3- pentanone in argon, along with the simulations from the Serinyel et al. mechanism of NUI Galway [98]. The measured fuel time histories are inconsistent with the simulated profiles from the model, and the model significantly underpredicts the fuel removal rates at current experimental conditions. As illustrated in Fig. 5.4, the 3-pentanone sensitivity analysis was performed to determine which reactions are pertinent to 3-pentanone time histories. As expected, 3-pentanone time histories are primarily sensitive to the initial fuel decomposition pathways: C 2 H 5 COC 2 H 5 (+ M) C 2 H 5 + C 2 H 5 CO (+ M) (4a) C 2 H 5 COC 2 H 5 (+ M) CH 3 + C 2 H 5 COCH 2 (+ M) (4b) 60

87 In addition, there is some minor interference from the reactions of C 2 H 4 + H (+ M) C 2 H 5 (+ M), CH 3 + CH 3 C 2 H 5 + H, and the H-atom abstraction reactions from 3- pentanone by H radicals. The branching fraction of the initial fuel decomposition through reaction (4a) ranges from 0.59 to 0.53 over K at 1.6 atm [98]. This indicates that 3-pentanone undergoes unimolecular decomposition through these two reaction pathways at similar rates. Figure 5.3: Measured and simulated 3-pentanone time histories for 1% 3- pentanone in Ar. Simulations used the Serinyel et al. mechanism. Figure 5.4: 3-pentanone sensitivity analysis for 1% 3-pentanone in Ar at 1323 K and 1.6 atm. 61

88 Methyl radical (CH 3 ) is an important transient species during 3-pentanone pyrolysis. CH 3 radicals are first formed through reaction (4b) of the initial fuel decomposition pathways, and hence the initial CH 3 formation rates are a good measure of the initial fuel decomposition rates. Fig. 5.5 shows the measured CH 3 time histories during pyrolysis of 0.1% 3-pentanone in argon, along with the computed profiles from the Serinyel et al. mechanism. It should be noted that the uncertainty of the CH 3 concentration during the first 100 µs was approximately ±10%, which was mainly contributed from the uncertainties in the absorption coefficient and the interference subtraction scheme. At later times, the CH 3 plateau levels had a slightly larger uncertainty of ±20% due to larger interference absorption (primarily from C 2 H 4 ). Similar to the fuel time histories, the model fails to capture the initial CH 3 formation rates, and the predicted CH 3 peak values are approximately 30% lower than the measured values. However, the model does simulate the CH 3 removal rates reasonably well from which we can infer that the reaction rate constants for these CH 3 removal channels are reasonable. Figure 5.5: CH 3 time histories for 0.1% 3-pentanone in Ar. Simulations were done using the Serinyel et al. mechanism. As shown in Fig. 5.6, the CH 3 sensitivity analysis reveals that the CH 3 time histories are mainly sensitive to the initial fuel decomposition pathways (reactions (4a) 62

89 and (4b)) at early times. At later times, there is some interference from the secondary reactions, which are described as follows: C 2 H 4 + H (+ M) C 2 H 5 (+ M) (20) CH 3 + CH 3 C 2 H 5 + H (21) C 2 H 6 (+ M) CH 3 + CH 3 (+ M) (18) Of note is that the rate constants for reactions (18), (20), and (21) are relatively well-established. In particular, the reverse of reaction (18) is a primary CH 3 removal channel at high temperatures. In the present analysis, we updated the rate constant for reaction (18) with the values measured by Oehlschlaeger et al. [80], whose measured values are consistent with another recent study from Kiefer et al. [81]. The rate constants for reactions (18), (20), and (21) (and the H-atom abstraction reactions from 3-pentanone by H radicals) are also provided in Table 5.2. Additionally, the most uncertain rate constants among reactions (4a), (4b), (18), (20), and (21) are the initial fuel decomposition pathways, reactions (4a) and (4b). Hence, the measured 3-pentanone and CH 3 time histories can be used to infer the overall fuel decomposition rate constant (k 4 = k 4a + k 4b ) at the measured pressure. Figure 5.6: CH 3 sensitivity analysis for 0.1% 3-pentanone in Ar at 1433 K and 1.6 atm. 63

90 Table 5.2: Kinetic parameters employed in the Serinyel et al. mechanism. Rate Constant Reaction A [ ] b E [cal/mol] No. Reference C 2 H 4 + H (+ M) C 2 H 5 (+ M) 1.081E E [98] Low-Pressure Limit: 1.200E E+03 Troe centering: CH 3 + CH 3 C 2 H 5 + H 4.990E E [98] C 2 H 6 (+ M) CH 3 + CH 3 (+ M) 1.880E E [80] Low-Pressure Limit: 3.720E E+05 Troe centering: C 2 H 5 COC 2 H 5 + H C 2 H 5 COC 2 H 4 p + H E E+03 22a [98] C 2 H 5 COC 2 H 5 + H C 2 H 5 COC 2 H 4 s + H E E+03 22b [98] C 2 H 5 COC 2 H 4 p C 2 H 5 CO + C 2 H E E [98] C 2 H 5 COC 2 H 4 s CH 3 CHCO + C 2 H E E [98] CH 3 CHCO + H C 2 H 5 + CO 4.400E E [98] C 2 H 4 + CO (+ M) CH 3 CHCO (+ M) 8.100E this study Low-Pressure Limit: 2.690E E+03 Troe centering: 5.907E H + O 2 OH + O 3.547E E [98] C 2 H 4 + OH C 2 H 3 + H 2 O 1.800E E [98] Units of A are in s -1 for unimolecular reactions, cm 3 mol -1 s -1 for bimolecular reactions, and cm 6 mol -2 s -1 for termolecular reactions. Fig. 5.7 shows the measured 3-pentanone and CH 3 time histories, along with their best-fit profiles simulated from the Serinyel et al. mechanism by revising the overall 3- pentanone decomposition rates. As a result, the measured 3-pentanone time histories provided the values for k 4 over K at 1.6 atm, and the measured CH 3 time histories provided the values for k 4 over K at 1.6 atm, as are shown in Fig A best-fit expression for the overall 3-pentanone decomposition rate constant measurements can be given as k 4 = T -10 exp(-44,780/t) s -1 over K (see the dashed line in Fig. 5.8). (In addition, the measured values can be expressed in Arrhenius form as k 4 = exp(-34,640/t) s -1 over K.) The major contributions to the uncertainties in the overall rate constant were: temperature (±10%), 64

91 3-pentanone absorption coefficient (±5%) (or CH 3 absorption coefficient and interference subtraction scheme (±10%)), fitting the data to computed profiles (±5%), and uncertainties resulting from secondary reactions (±15%), giving an overall uncertainty in k 4 of ±35% over K from the measured 3-pentanone time histories (or ±40% over K from the measured CH 3 time histories). The values for k 4 are also summarized in Table 5.3. The influence of the branching ratio (k 4a /k 4 ) of reaction (4a) on the overall fuel decomposition rate constant was examined by perturbing the branching ratio from 0.4 to 0.7 and keeping k 4 constant, and the changes in the computed profiles were negligible. Hence, the measured 3-pentanone and CH 3 time histories are insensitive to the branching ratios of the initial fuel decomposition pathways. In the present analysis, the original branching ratios postulated from Serinyel et al. are utilized. As illustrated in Fig. 5.8, the measured overall 3-pentanone decomposition rate constant is approximately 3.5 times those of Serinyel et al. over K at 1.6 atm, and as such, the high-pressure limit rate constants for reactions (4a) and (4b) as given by Serinyel et al. are in need of revision. Note that the measured and predicted overall 3- pentanone decomposition rate constants both experience severe non-arrhenius curvature, which explains the large values inferred for the A-factor and pre-exponential temperature dependence. Table 5.3: Summary of overall 3-pentanone decomposition rate constant data. T 5 [K] P 5 [atm] k 4 [s -1 ] Initial Mixture: 0.1% 3-Pentanone / Ar E E E E+04 Initial Mixture: 1% 3-Pentanone / Ar E E E E E+03 65

92 Figure 5.7: (a) Best-fit 3-pentanone time histories and (b) best-fit CH 3 time histories using the Serinyel et al. mechanism with revised overall 3- pentanone decomposition rate constant (k 4 ). 66

93 Figure 5.8: Arrhenius plot for the overall 3-pentanone decomposition rate constant (k 4 ) at 1.6 atm. Fig. 5.9 shows the measured 3-pentanone and CO time histories during pyrolysis of 1% 3-pentanone in argon at 1248 K and 1.6 atm, along with the simulations from the Serinyel et al. mechanism with the revised k 4. Note that initially, all O atoms (100 percent) are present in 3-pentanone. At 750 µs, approximately 35% of the total O atoms remains in 3-pentanone, with about 55% of the O atoms in CO; together the O atoms from these species add up to ~90% of the total available O atoms. Similarly, at 1500 µs, there are approximately 23% of the O atoms in 3-pentanone and 67% of the O atoms in CO, and these O atoms also sum up to ~90%. Therefore, the measurements suggest approximately 90% conversion of 3-pentanone to CO. However, the model with the revised k 4 only predicts ~57% conversion of 3-pentanone to CO. 67

94 Figure 5.9: Measured 3-pentanone and CO time histories during 3- pentanone pyrolysis at 1248 K and 1.6 atm. Based on the CO sensitivity analysis (see Fig. 5.10), the CO time histories, and particularly the initial formation rates, are mainly sensitive to the initial 3-pentanone decomposition pathways (reactions (4a) and (4b)), the ethyl radical decomposition (reaction (20)), and the H-atom abstraction reactions from 3-pentanone, which are: C 2 H 5 COC 2 H 5 + H C 2 H 5 COC 2 H 4 p + H 2 (22a) C 2 H 5 COC 2 H 5 + H C 2 H 5 COC 2 H 4 s + H 2 (22b) Figure 5.10: CO sensitivity analysis for 1% 3-pentanone in Ar at 1248 K and 1.6 atm. 68

95 Abstraction of hydrogen atom from the fuel molecule yields the formation of fuel radicals C 2 H 5 COC 2 H 4 p and C 2 H 5 COC 2 H 4 s, where p and s denote the primary and secondary sites, respectively. However, these reactions do not seem to significantly affect the final CO plateau values, and only minor changes in the computed CO plateaus are found if the rate constants for these reactions are increased by a factor of 3. One possible explanation for the discrepancy between the measurements and simulations is the incomplete CO formation pathways in the model. Based on the Serinyel et al. mechanism with the revised k 4, the major species that contain O-atoms are 3-pentanone, CO, and methyl ketene (CH 3 CHCO). Hence, we infer that the model overpredicts the concentration of methyl ketene. Methyl ketene is formed through the H-atom abstraction reaction (C 2 H 5 COC 2 H 5 + H C 2 H 5 COC 2 H 4 s + H 2 ), immediately followed by the decomposition of the fuel radical. C 2 H 5 COC 2 H 4 s CH 3 CHCO + C 2 H 5 (24) Methyl ketene is then removed via one pathway only in the original Serinyel et al. mechanism [98], as described in the following: CH 3 CHCO + H C 2 H 5 + CO (25) As discussed in Chapter 4, methyl ketene should also undergo thermal decomposition to form other stable species, as suggested in one previous study [101]. Here we also incorporate the methyl ketene unimolecular decomposition pathway into the Serinyel et al. mechanism: CH 3 CHCO (+ M) C 2 H 4 + CO (+ M) (-26) The rate constant for methyl ketene unimolecular decomposition was assumed to have the same value as the rate constant for ketene unimolecular decomposition (CH 2 CO (+ M) CH 2 + CO (+ M)). With this modification, the model can now predict higher CO concentrations and lower methyl ketene concentrations. In addition, based on the model, the remaining O-atom containing species is primarily ketene and an accurate knowledge of the ketene sub-mechanism becomes particularly important in predicting the CO plateau levels during 3-pentanone pyrolysis. Fig shows the measured CO time histories during 0.25% 3-pentanone in Ar, along with the simulations from the (a) original and (b) modified Serinyel et al. 69

96 mechanisms. As expected, the original Serinyel et al. mechanism underpredicts the CO time histories by at least 30% at current experimental conditions. With the revised overall 3-pentanone decomposition rate constant and the addition of the methyl ketene decomposition reaction, the computed profiles from the modified mechanism show much better agreement with the measurements at all temperatures. Similar agreement between the measurements and the simulations from the modified mechanism can also be found for the mixture of 1% 3-pentanone in Ar. As mentioned previously, the initial CO formation rates are sensitive to the H-atom abstraction reactions (reactions (22a) and (22b)). Hence, the current study supports the use of the Serinyel et al. values for these reaction rate constants. Figure 5.11: CO time histories for 0.25% 3-pentanone in Ar: measured and calculated values from the (a) original and (b) modified Serinyel et al. mechanisms. 70

97 During 3-pentanone pyrolysis, ethylene is mainly formed through either unimolecular decomposition pathways or H-atom abstraction reactions, and these H-atom abstraction reactions are particularly important at later times when H atoms are abundant in the system. The primary H-atom abstraction reactions are C 2 H 5 COC 2 H 5 + H C 2 H 5 COC 2 H 4 p + H 2 and C 2 H 5 COC 2 H 5 + H C 2 H 5 COC 2 H 4 s + H 2. These fuel radicals (C 2 H 5 COC 2 H 4 p and C 2 H 5 COC 2 H 4 s) can then form ethylene through the following processes: (i) C 2 H 5 COC 2 H 4 p C 2 H 5 CO + C 2 H 4 (23) (ii) C 2 H 5 COC 2 H 4 s CH 3 CHCO + C 2 H 5, followed by methyl ketene and ethyl radical decompositions. Thus, in addition to CO, ethylene is another major product during 3-pentanone pyrolysis, and each 3-pentanone molecule eventually turns into at least one C 2 H 4 molecule. As illustrated in Fig. 5.12, the measured ethylene yields (defined as the ratio of the long-time C 2 H 4 concentration to the initial 3-pentanone concentration) for the mixture of 0.25% 3-pentanone in Ar at K are approximately 1.4. When compared to the measurements, the Serinyel et al. mechanism fails to capture the initial C 2 H 4 formation rates, and the simulated ethylene yield at 1248 K is ~30% lower than the measured value. On the other hand, the modified mechanism simulates the initial C 2 H 4 formation rates and ultimate yields quite accurately at the measured temperatures. Figure 5.12: C 2 H 4 time histories for 0.25% 3-pentanone in Ar: measured and calculated values. 71

98 Pentanone Oxidation Ignition delay times were measured with mixtures varying in concentration from 0.040% to 0.875% 3-pentanone in O 2 /balance argon over K at a pressure of ~1 atm and equivalence ratios of 1.0 and 0.5. For high fuel concentration mixtures (X 3- pentanone > 0.1%), the endwall ignition delay time is defined as the time interval between the arrival of the incident shock and the initial rise in the OH* emission chemiluminescence trace at the endwall. The initial rise is located by linear extrapolation of the signal at the time of maximum rate of rise to the baseline. A representative ignition delay time plot is also provided in Fig On the other hand, for lower fuel concentration mixtures, the emission signal is rather weak, and a different definition of ignition delay time is employed. It is defined as the time to reach 50% of the peak OH concentration (measured using OH absorption method), with time zero being defined as the arrival of the reflected shock at the sidewall measurement location. Figure 5.13: Sample sidewall pressure and endwall OH* emission time histories recorded during an experiment of 3-pentanone ignition at 1113 K and 1.1 atm (3-pentanone/ 4.0% O 2 / Ar, Φ = 0.5). A tailored gas mixture of 60% helium/ 40% nitrogen was used as driver gas to achieve a long test time. For high fuel concentration mixtures, the definition of the endwall ignition delay time is shown in the figure. 72

99 Fig shows the measured ignition delay times from the current study (see the solid points) at Φ = 1.0 and 0.5, along with the simulations from the original and modified Serinyel et al. mechanisms under the assumption of constant internal energy and constant volume. In addition, Davidson et al. [97] performed ignition delay time measurements for the mixture of 0.571% 3-pentanone with 4% O 2 in Ar (Φ = 1.0) over K at pressures of atm. Their ignition delay time data were normalized to 1 atm using an overall correlation pressure dependence of P (see the hollow points), and their data are in good agreement with the current measurements within 7%. As is evident in Fig. 5.14, fuel-lean mixtures ignite much faster than stoichiometric mixtures, and this general trend is consistent with other hydrocarbon studies at similar temperatures and pressures [62, 97, ]. At high temperatures (T > 1100 K), the overall reactivity is mainly controlled by the chain branching reaction (H + O 2 OH + O), so that reactivity is very sensitive to molecular oxygen concentration. Therefore, at Φ = 0.5, the mixture of 0.875% 3-pentanone (with the highest O 2 content) ignites much faster than other lower fuel concentration mixtures. Interestingly, for the mixture of 0.875% 3-pentanone with 12.25% O 2 in Ar, the current measurements are approximately 30% faster than the measurements from Serinyel et al. [98]. 73

100 Figure 5.14: Measured and simulated 3-pentanone ignition delay times at (a) Φ = 1.0 and (b) Φ = 0.5 and P 5 = 1.0 atm. A linear regression analysis was performed on the current ignition delay time measurements, and these measurements can be expressed in a correlation with an R 2 value of 0.974: τ [µs] = P Φ 0.89 X O2 exp(20,450/t) where P is the total pressure [atm], Φ is the equivalence ratio, and X O2 is the oxygen mole fraction. As shown in Fig. 5.14, the simulated ignition times from the Serinyel et al. mechanism are in agreement with the measured values for low fuel concentration mixtures (400 ppm 3-pentanone) within ~35%. However, for higher fuel concentration mixtures, the computed ignition times are approximately twice those of the measurements. On the other hand, the modified Serinyel et al. mechanism shows much better agreement with the measurements at all concentrations (within 30%), and the revision of the overall 3-pentanone decomposition rate constant and the addition of the methyl ketene decomposition pathway significantly improve the general performance of the 3-pentanone oxidation chemistry model. When compared to the measurements from Serinyel et al. [98], the modified mechanism also shows improved agreement with their data (within 15%) over K at 1 atm. The simulations from the original and 74

101 modified mechanisms at the test conditions of Serinyel et al. are also provided in Fig Figure 5.15: Comparison of model predictions between (a) the Serinyel et al. mechanism of NUI Galway [98] and (b) the modified mechanism on ignition delay time measurements from Serinyel et al. In addition to ignition delay time measurements, OH and H 2 O time histories were acquired behind reflected shock waves using the mixtures of 400 ppm 3-pentanone with O 2 in balance argon over K at pressures around 1.6 atm and equivalence ratios of 1.0 and 0.5 (see Figs ). Low fuel concentration mixtures are preferred in species time history measurements during hydrocarbon oxidation in order to minimize 75

102 the rapid energy release at the time of ignition, which increases the reflected shock temperature. This would affect the analysis of species mole fractions, particularly if constant absorption coefficients for Beer s law are used, as has been common in the past. For the mixture of 400 ppm 3-pentanone with 0.28% O 2 in argon, there is approximately a 50 K increase in temperature after ignition for the case at 1377 K and 1.54 atm (under constant energy and volume constraints), which slightly perturbs (reduces) the OH and H 2 O absorption coefficients by up to 4% and 2%, respectively, an effect which has been accounted for in our data processing. Sensitivity analysis reveals that both OH and H 2 O time histories (for the mixture of 400 ppm 3-pentanone / 0.28% O 2 / Ar) are mainly sensitive to the following set of reactions (see Figs and 5.17): H + O 2 OH + O (27) C 2 H 5 COC 2 H 5 (+ M) C 2 H 5 + C 2 H 5 CO (+ M) (4a) C 2 H 5 COC 2 H 5 (+ M) CH 3 + C 2 H 5 COCH 2 (+ M) (4b) C 2 H 4 + H (+ M) C 2 H 5 (+ M) (20) C 2 H 4 + OH C 2 H 3 + H 2 O (28) CH 3 + CH 3 C 2 H 5 + H (21) Figure 5.16: OH sensitivity analysis for 400 ppm 3-pentanone with 0.28% O 2 in Ar (Φ = 1.0) at 1486 K and 1.52 atm. 76

103 Figure 5.17: H 2 O sensitivity analysis for 400 ppm 3-pentanone with 0.28% O 2 in Ar (Φ = 1.0) at 1486 K and 1.52 atm. During hydrocarbon oxidation, H 2 O is regarded as an important combustion progress marker, which gives nearly identical information to that of another combustion progress marker, CO. The H 2 O profiles for typical hydrocarbons [60-61, 102, 104] exhibit sequential features: an initial gradual formation of H 2 O is followed by a rapid increase indicating ignition, which is in turn succeeded by a very slow rise in H 2 O concentration. During 3-pentanone oxidation, the H 2 O profiles are slightly different from those of common hydrocarbons (n-alkanes and cycloalkanes). There is no obvious distinction between the initial gradual H 2 O formation and the rapid increase in H 2 O concentration during ignition. The formation of H 2 O is fast and steady starting from time zero, followed by a nearly constant H 2 O concentration. After this post-ignition H 2 O plateau level has been reached, all volatile hydrocarbons have been depleted and only small intermediates (i.e., H, O, OH, CO, CO 2, and H 2 ) would remain, gradually approaching chemical equilibrium. The kinetics that controls these small intermediates (after ignition) is well-established and subject to small uncertainties [22, 33, 72]. Additionally, this nearly constant H 2 O plateau level (after ignition) is primarily sensitive to the relatively well-established thermodynamic parameters, as suggested by Hong et al. [102]. This observation was also validated through the perturbation of the rate constants 77

104 for the above important reactions, following which the post-ignition H 2 O concentration level effectively remained the same. Thus, the post-ignition H 2 O plateau level can be used to confirm pre-shock fuel concentration. Fig shows the measured H 2 O time histories for the mixture of 400 ppm 3- pentanone with 0.28% O 2 in argon (Φ = 1.0), along with the computed profiles from the (a) original (top) and (b) modified (bottom) Serinyel et al. mechanisms. At a first glance, the measured and computed H 2 O plateau levels are consistent with each other, and this agreement justifies the accuracy of the initial fuel loading in our experiments. The original Serinyel et al. mechanism captures the general shape of the measured H 2 O profiles, but it does not predict the initial H 2 O formation rates accurately. In addition, the modified mechanism performs slightly better at higher temperatures (1486 K and 1542 K), and the computed profiles from the modified mechanism show much better agreement with the measurements at lower temperatures (1343 K and 1377 K). Similarly, Fig illustrates the measured H 2 O time histories for the mixture of 400 ppm 3-pentanone with 0.56% O 2 in argon (Φ = 0.5), along with the simulations from the (a) original (top) and (b) modified (bottom) Serinyel et al. mechanisms. Here also, the measured and computed H 2 O plateau levels (after ignition) are consistent with each other. For this fuel-lean mixture, the computed profiles from the original Serinyel et al. mechanism are quite different from the measured profiles in terms of the initial formation rates and the ignition delay times, especially when compared to the stoichiometric mixture. At 1217 K and 1.74 atm, the model significantly underpredicts the formation rate of H 2 O, and the computed ignition delay time based on the H 2 O time history is at least twice that of the measured value. On the other hand, the modified mechanism is able to capture the H 2 O formation rates reasonably well at all temperatures, particularly at 1217 K. Hence, the revision of the overall 3-pentanone decomposition rate constant and the addition of the methyl ketene decomposition pathway greatly improve the model predictions of H 2 O concentrations under oxidizing conditions. 78

105 Figure 5.18: Comparisons of measured and simulated H 2 O time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3-pentanone with 0.28% O 2 in Ar (Φ = 1.0). 79

106 Figure 5.19: Comparisons of measured and simulated H 2 O time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3-pentanone with 0.56% O 2 in Ar (Φ = 0.5). 80

107 OH species time histories provide another important kinetic target for the validation of detailed kinetic mechanisms during hydrocarbon oxidation. At early times, there is the rapid formation of OH simultaneous with the initial fuel decomposition. Based on the temperature, the OH species develops a well-defined plateau level during the induction period, especially at lower temperatures. At the time of ignition, the OH mole fraction rapidly increases to its post-ignition plateau level. More importantly, the early-time feature of OH appears to provide important information about the breakdown of the fuel, because this feature is governed by the fuel unimolecular decomposition and the H-atom abstraction reactions from the fuel (mainly by H radicals). At later times, the rapid OH rise at ignition is not unique to this fuel. At the time of ignition, fuel molecules have mostly decomposed into small intermediate radicals or molecules, such as H, H 2, C 2 H 4, C 3 H 6, etc., and these small fragments tend to control the rate of ignition (the rapid OH rise at ignition) during hydrocarbon oxidation, as suggested by Warnatz et al. [105]. Fig illustrates the measured OH time histories for the mixture of 400 ppm 3- pentanone with 0.28% O 2 in argon (Φ = 1.0), along with the simulated profiles from the (a) original (top) and (b) modified (bottom) Serinyel et al. mechanisms. At higher temperatures (1486 K and 1542 K), the original Serinyel et al. mechanism captures the rapid OH rise at the time of ignition reasonably well, but it underpredicts the initial OH formation rates and the initial OH plateau levels (see inset on Fig. 5.20). At lower temperatures (1343 K and 1377 K), the model underpredicts both the initial and final OH formation rates, but it is able to simulate the initial plateau levels quite well. When compared to the original Serinyel et al. mechanism, the modified model generally provides a much better agreement with the measurements, in terms of the initial plateau levels and both the initial and final formation rates. However, the modified model still cannot capture the slight overshoot prior to the formation of the initial plateau. The overshoot seems to be more obvious at higher temperatures, and such overshoot is quite sensitive to the H-atom abstraction reactions from 3-pentanone by H radicals and these rate constants may require some fine adjustment at higher temperatures. Fig shows the measured OH time histories for the mixture of 400 ppm 3- pentanone with 0.56% O 2 in argon (Φ = 0.5), along with the simulations from the (a) 81

108 original (top) and (b) modified (bottom) Serinyel et al. mechanisms. It should be noted that there is now no noticeable overshoot in OH concentration prior to the formation of the initial plateau for the fuel-lean mixture. Instead, there is a smooth transition from the initial OH rise to the first plateau level. In addition, with more O 2 molecules in the system, the initial and post-ignition plateau levels are much higher than those of the stoichoimetric mixture, and this observation is well-simulated by the model. This is mainly due to the fact that more oxygen molecules are available to undergo the chain branching reaction (H + O 2 OH + O). The Serinyel et al. mechanism underpredicts both the initial and final OH formation rates by at least 30% at all temperatures. Similarly, it does not simulate the initial plateau levels properly at temperatures greater than 1300 K. On the other hand, the modified model is able to simulate the OH time histories very accurately, in terms of the initial plateau levels and the initial and final formation rates, particularly at 1217 K. Hence, the revision of the overall 3-pentanone decomposition rate constant greatly improves the model predictions on the first OH plateau levels and its initial formation rates, and the branching ratio of reaction (4a) for the initial fuel decomposition pathways predicted by Serinyel et al. seems to be quite reasonable. More importantly, the methyl ketene decomposition pathway seems to promote the ignition behavior. One possible explanation is that more CO molecules are now introduced to the system, and each CO molecule reacts with an OH radical through the reaction of CO + OH CO 2 + H, which is an important exothermic reaction in combustion chemistry. As a result, more heat is generated during 3-pentanone oxidation, thereby promoting the ignition behavior. 82

109 Figure 5.20: Comparisons of measured and simulated OH time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3-pentanone with 0.28% O 2 in Ar (Φ = 1.0). Inset figures are provided to show the early-time features over µs. 83

110 Figure 5.21: Comparisons of measured and simulated OH time histories from the (a) original and (b) modified Serinyel et al. mechanisms for 400 ppm 3-pentanone with 0.56% O 2 in Ar (Φ = 0.5). Inset figures are provided to show the early-time features over µs Comparisons of Ketone Oxidation Characteristics In addition to 3-pentanone oxidation, the oxidation characteristics of acetone and 2-pentanone were examined and compared with that of 3-pentanone using ignition delay 84

111 time and OH time history measurements for the mixtures of ketone with 0.525% O 2 in balance argon at Φ = 1.0. Shown in Fig is a plot of the measured ignition delay times for these three ketone mixtures. Strikingly, the ignition delay times of 3-pentanone mixtures are only about half those of acetone and 2-pentanone mixtures. However, the apparent activation energies of these ketone mixtures are quite similar. Figure 5.22: Comparison of ignition delay times for different ketones (acetone, 2-pentanone and 3-pentanone). Shown in Fig is a plot of the measured OH time histories for the above ketone mixtures at T 1370 K and P = 2.6 atm. Note that the OH time history for the 3- pentanone mixture is at a slightly lower temperature (1354 K). Interestingly, the 3- pentanone mixture has a much higher initial OH plateau level than the acetone and 2- pentanone mixtures, while the initial OH plateau levels for the acetone and 2-pentanone mixtures are approximately the same. As discussed above, this early-time OH feature is unique to the individual fuel, and is primarily controlled by the initial fuel decomposition pathways and the H-atom abstraction reactions from the fuel by H radicals. In addition, as suggested by Davidson et al. [62], the initial OH plateau level is quite sensitive to the branching ratios of the fuel decomposition pathways. For instance, there is often a direct 85

112 competition between the C 2 H 5 and CH 3 formation channels during 3-pentanone decomposition. A slight increase in the rate constant for the C 2 H 5 channel can further increase the first OH plateau level, and a slight increase in the rate constant for the CH 3 channel can reduce the plateau level. Once the C 2 H 5 radical is formed, it is quickly followed by its decomposition to form an H radical. The H radical can then undergo the chain branching reaction (with O 2 ) to form an OH radical. On the other hand, if the CH 3 radical is formed, it tends to form ethane through the methyl-methyl recombination reaction and fewer H radicals are formed. In the case of acetone and 2-pentanone decomposition, CH 3 radicals are generally formed, and insufficient amounts of C 2 H 5 radicals are developed to give H radicals. Thus, the initial OH plateau levels of acetone and 2-pentanone are much less than that of 3-pentanone. More importantly, there seems to be a strong positive correlation between the initial OH plateau level and the ignition delay time. This correlation further explains why 3-pentanone has the fastest ignition delay times among these ketone mixtures at current experimental conditions. Figure 5.23: Comparison of OH time histories for the mixtures of ketone (i.e., acetone, 2-pentanone and 3-pentanone) with 0.525% O 2 in Ar at a pressure of 2.6 atm and an equivalence ratio of 1.0. An inset figure is provided to show the early-time features over µs. 86

113 5.4 Summary High-temperature 3-pentanone pyrolysis and oxidation studies were investigated using laser-based species time history measurements for 3-pentanone, CH 3, CO, C 2 H 4, H 2 O and OH. To our knowledge, these measurements are the first laser-based species time history measurements for high-temperature 3-pentanone pyrolysis and oxidation. Using these time histories and the Serinyel et al. 3-pentanone mechanism [98], improved determinations of the initial 3-pentanone unimolecular decomposition reactions were possible. As well, a comparison of the measured and modeled CO time history pathways identified the need to include the methyl ketene decomposition pathway to improve the simulations. These two modifications to the Serinyel et al. mechanism also significantly improved the agreement with ignition delay times and OH and H 2 O time histories during 3-pentanone oxidation. Finally, a comparison of OH time histories during the oxidation of 3-pentanone, acetone, and 2-pentanone showed that the initial OH plateau level of 3-pentanone was higher than that of acetone and 2-pentanone and this was consistent with the shorter ignition delay times seen with this fuel. 5.5 Possible Future Work More work is definitely needed to further improve the model predictions under 3- pentanone pyrolytic and oxidizing conditions. As discussed in Section 5.3.1, the original Serinyel et al. mechanism [98] appears to overpredict the methyl ketene concentration during 3-pentanone pyrolysis, and such effect cannot currently be observed experimentally (based on the O-atom balance). In the present analysis, the possible solution to such discrepancy is to introduce a unimolecular decomposition reaction for methyl ketene (reaction (26)) in order to remove the pooling of methyl ketene in the original model, and reasonable agreement can be obtained between the current measurements and the simulations from the modified model. Despite its improved predictive capability, the modified model is very likely to suffer from other deficiencies that have not been addressed in this dissertation. For instance, the rate constant for 87

114 reaction (24) (C 2 H 5 COC 2 H 4 s C 2 H 5 + CH 3 CHCO) is possibly too fast, resulting in the pooling of methyl ketene and prohibiting the CO formation. Therefore, a slower rate constant for reaction (24) is recommended. Additionally, as demonstrated in Chapters 4 and 5, the kinetics of methyl ketene is poorly understood, and more experimental and theoretical studies for methyl ketene are definitely required. In particular, the rate constant for reaction (26) (CH 3 CHCO (+ M) C 2 H 4 + CO (+ M)) was only estimated by analogy with the rate constant for ketene unimolecular decomposition in the present analysis, and there is a need for a more accurate rate constant expression. One of the major weaknesses in the modified mechanism is the fact that it is not able to accurately predict the ignition delay times for high fuel concentration mixtures (X 3-pentanone > 0.2%), as shown in Fig Ignition delay time is an important global kinetic target that is commonly used for the validation of the detailed models, but this global target is quite sensitive to many secondary chemistry reactions. In particular, the C2 chemistry, such as the ethyl radical decomposition reaction, is very crucial to the development of the successful 3-pentanone kinetic model suitable for high-temperature application. In the future, different base mechanisms, which might consist of different C2 chemistry sets, should also be considered. 88

115 Chapter 6 High-Temperature Measurements of the Reactions of OH with a Series of Ketones: Acetone, 2-Butanone, 3- Pentanone, and 2-Pentanone 6.1 Introduction Due to their significant roles in atmospheric chemistry, the rate constants for the reactions of OH radicals with a series of ketones, including acetone, 2-butanone, 3- pentanone and 2-pentanone, have been extensively studied by many researchers [38, ] over the temperature range of K. However, the kinetic data on ketones + OH at combustion-relevant conditions are generally scarce. There were a few experimental studies for the acetone + OH reaction rate constant over K. Yamada et al. [117] utilized two different OH precursors (HONO and N 2 O/H 2 O) and measured the rate constants for OH + CH 3 COCH 3 and CD 3 COCD 3 in a reactor over K using the pulsed laser photolysis/pulsed laser-induced fluorescence technique. Bott and Cohen [118] pioneered the use of tert-butyl hydroperoxide as an OH precursor and monitored the OH decay in a shock tube using the UV lamp absorption method at 309 nm in order to study the rate constant for acetone + OH reaction near 1200 K and 1 atm. Similarly, Vasudevan et al. [119] and Srinivasan et al. [77] both measured the acetone + OH rate constant using shock tubes and UV absorption methods over the combustion-relevant temperature range of K. These measurements are in good 89

116 agreement with each other. In contrast to acetone, there was only one experimental study available for larger ketone + OH kinetic data. Tranter and Walker [120] added small amounts of ketones (acetone, 2-butanone and 3-pentanone) individually to slowly reacting mixtures of H 2 + O 2 at 753 K, and measured the consumption of ketones and H 2 with the use of gas chromatography. This method allowed them to study the relative rate constants for the reactions of H and OH with ketones at 753 K. Furthermore, Zhou et al. [121] recently performed a theoretical study on the mechanism and kinetics of the reactions of OH with three methyl ketones: acetone, 2-butanone and isopropyl methyl ketone. They employed the computationally less expensive G3 and G3MP2BH&H methods to calculate the energy barriers, and utilized the Variflex code including Eckart tunneling corrections to compute the total rate constants over K. In addition, all possible abstraction channels have been accounted for in their calculation. However, except for acetone, their theoretical calculations have not been validated against any high-temperature experimental data. The overall rate constants for the reactions of OH with four ketones, namely acetone (CH 3 COCH 3 ), 2-butanone (C 2 H 5 COCH 3 ), 3-pentanone (C 2 H 5 COC 2 H 5 ) and 2- pentanone (C 3 H 7 COCH 3 ), were determined behind reflected shock waves over the temperature range of K at pressures of 1-2 atm: CH 3 COCH 3 + OH Products (5) C 2 H 5 COCH 3 + OH Products (6) C 2 H 5 COC 2 H 5 + OH Products (7) C 3 H 7 COCH 3 + OH Products (8) These measurements include the first direct high-temperature measurements of the overall rate constants for reactions (6)-(8). These high-temperature kinetic data, along with the earlier work [38, 77, ], are compared with the theoretical calculations (from Zhou et al. [121]) and the estimates using the group-additivity model. 90

117 6.2 Experimental Details Test mixtures were prepared manometrically in a 40 liter stainless steel tank heated uniformly to 50 o C and mixed with a magnetically-driven stirring vane. A doubledilution process was employed to allow for more accurate pressure measurements in the manometrical preparation of a highly dilute mixture. A highly concentrated mixture was first prepared and mixed for at least 2 hours to ensure homogeneity and consistency, and the mixture was then further diluted with argon and mixed for additional 2 hours prior to the experiments. The gas utilized in this study was argon (Research Grade) %, which was supplied by Praxair and used without further purification. The liquid chemicals were 70% tert-butyl hydroperoxide (TBHP) in water, CHROMASOLV grade acetone ( 99.9%), CHROMASOLV grade 2-butanone ( 99.7%), ReagentPlus grade 3- pentanone ( 99%), and ReagentPlus grade 2-pentanone ( 99%) from Sigma-Aldrich, and were purified using a freeze-pump-thaw procedure to remove dissolved volatiles and air prior to mixture preparation. The mixture composition was confirmed by sampling a portion of the mixture (from near the endwall) into an external multi-pass absorption cell with a path length of 29.9 m and monitoring the fuel concentration in the cell with a Jodon Helium-Neon laser at 3.39 µm. The details of the laser diagnostic set-up are discussed elsewhere [122]. Beer s law was used to convert the measured absorption data into the fuel mole fraction. The absorption cross-sections of ketones for Beer s law were directly obtained from the PNNL database [123], and the measured fuel concentrations were consistent with the values expected from the manometrical preparation within ±5%. 6.3 Kinetic Measurements Choice of Kinetic Mechanisms A total of 58 reflected shock wave experiments were performed to determine the overall rate constants for the reactions of OH with four ketones (acetone, 2-butanone, 3-91

118 pentanone and 2-pentanone) at near-pseudo-first-order conditions. Experiments were carried out over the temperature range of K at pressures of 1-2 atm using different initial fuel concentrations: acetone (304 ppm), 2-butanone (152 ppm, 161 ppm and 206 ppm), 3-pentanone (151 ppm and 211 ppm), and 2-pentanone (161 ppm). These ketones were prepared with ppm TBHP/water and diluted in argon. To properly simulate the consumption of OH radicals by ketones, the Pichon et al. mechanism of NUI Galway [89] with the revised k 2 was chosen as the base mechanism for acetone, and the Serinyel et al. mechanism of NUI Galway [95, 98] with the revised k 3 and k 4 and the addition of the methyl ketene decomposition pathway was utilized as the base mechanism for 2-butanone, 3-pentanone, and 2-pentanone. In addition, the tert-butyl hydroperoxide (TBHP) sub-mechanism was incorporated in these base mechanisms. (Please read Chapter 3 for more details on the TBHP chemistry.) Similarly, the thermodynamic parameters for TBHP and tert-butoxy radical were taken from the thermodynamic database from Goos et al. [76], and the thermodynamic parameters for OH were updated with the values from Herbon et al. [23-24]. As discussed in Chapters 4 and 5, the initial decomposition pathways for acetone, 2-butanone, and 3-pentanone can be described as follows: CH 3 COCH 3 (+ M) CH 3 CO + CH 3 (+ M) (2) C 2 H 5 COCH 3 (+ M) C 2 H 5 + CH 3 CO (+ M) (3a) C 2 H 5 COCH 3 (+ M) CH 3 + C 2 H 5 CO (+ M) (3b) C 2 H 5 COCH 3 (+ M) CH 3 + CH 3 COCH 2 (+ M) (3c) C 2 H 5 COC 2 H 5 (+ M) C 2 H 5 + C 2 H 5 CO (+ M) (4a) C 2 H 5 COC 2 H 5 (+ M) CH 3 + C 2 H 5 COCH 2 (+ M) (4b) For 2-butanone and 3-pentanone (and 2-pentanone), their initial decomposition pathways consist of multiple channels. High-temperature decomposition pathways for 2-butanone and 3-pentanone were first investigated by Serinyel et al. [95, 98]. Recently, Lam et al. [ ] have performed experimental studies during high-temperature acetone, 2- butanone, and 3-pentanone pyrolysis, as were discussed in Chapters 4 and 5. In their studies, they measured the rate constant for reaction (2) and the overall values for reactions (3) and (4) at pressures near 1.6 atm. At T > 1300 K, the consumption of 92

119 ketones in the present study is mainly controlled by the H-atom abstraction reactions by OH radicals and the ketone decomposition pathways. Hence, reactions (2)-(4) are pertinent to the determinations of the overall rate constants for reactions (5)-(7) at higher temperatures, and the rate constants for reactions (2)-(4) were updated with the values from Lam et al. [ ] (the rate constants from Chapters 4 and 5). In addition, a review of the literature shows that there is currently no experimental or theoretical study for high-temperature 2-pentanone pyrolysis. Thus, 2-pentanone decomposition pathways, along with the corresponding rate constants, are not known in this study, and the overall rate constant for 2-pentanone + OH reaction (reaction (8)) cannot be inferred accurately at T > 1300 K. A thorough theoretical or experimental study for 2-pentanone decomposition pathways is required. Nevertheless, at T < 1300 K, the consumption of 2- pentanone in the present study is predominantly controlled by the H-atom abstraction reactions by OH radicals, and the overall rate constant for reaction (8) was determined over a narrower temperature range K. In addition, all simulations were performed using the CHEMKIN PRO package [71] under the standard constant internal energy and volume assumption Acetone + OH Kinetics An OH radical sensitivity analysis for the mixture of 304 ppm acetone with 28 ppm TBHP (and 73 ppm H 2 O) in Ar at 1148 K and 1.95 atm is provided in Figure 6.1. The analysis reveals that the reaction of OH with acetone (reaction (5)) is the dominant reaction over the time frame of the experiment, with some minor interference from the secondary reactions: CH 3 + OH CH 2 (s) + H 2 O (17) C 2 H 6 (+ M) CH 3 + CH 3 (+ M) (18) CH 3 OH (+ M) CH 3 + OH (+ M) (19) The rate constants for reactions (17)-(19) were updated with the values from Table 3.1 (in Chapter 3). 93

120 Figure 6.1: OH sensitivity plot for the rate constant measurement of acetone + OH at 1148 K and 1.95 atm. Figure 6.2 shows a sample measured OH concentration time history for the mixture of 304 ppm acetone in Ar at 1148 K and 1.95 atm, and the measured peak OH mole fraction is ~28 ppm. Due to wall adsorption and condensation of TBHP, the initial TBHP mole fraction was assumed to be the same as the measured peak OH mole fraction, which was formed immediately after the decomposition of TBHP behind the reflected shock wave at T > 1000 K. Note that a 70%, by weight, solution of TBHP in water in the liquid phase corresponds, initially, to 69% water and 31% TBHP in the vapor phase, based on Raoult s law [126]. Therefore, a 101 ppm TBHP/water mixture should have at most 31.3 ppm TBHP. In the present study, the mixtures of 101 ppm TBHP/water consist of ~28-30 ppm TBHP, based on the measured peak OH yields, hence suggesting very little loss to the walls. In addition, the test mixtures were chosen such that the ratio of the initial acetone concentration to the initial TBHP concentration is ~10, thereby achieving near-pseudo-first-order conditions. For the conditions described in Figure 6.2, a best-fit overall rate constant for reaction (5) of cm 3 mol -1 s -1 was obtained between the experimental data and the simulation. Simulations for the perturbations of ±50% in the inferred rate constant are also illustrated in Figure 6.2. In addition, Table

121 summarizes the rate constant measurements of reaction (5) at K and atm. Figure 6.2: Sample acetone + OH rate constant measurement using the mixture of 304 ppm acetone with ~28 ppm TBHP (and 73 ppm water) in Ar at 1148 K and 1.95 atm. Simulation from the modified Pichon et al. mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown. Table 6.1: CH 3 COCH 3 + OH Products: Rate Constant Data. T 5 [K] P 5 [atm] k 5 [cm 3 mol -1 s -1 ] 101 ppm TBHP (and water), 304 ppm CH 3 COCH 3, Ar E E E E E E E E E E ppm TBHP (and water), 304 ppm CH 3 COCH 3, Ar E E+12 95

122 A detailed error analysis was performed to estimate the uncertainty limits of the measured rate constant for reaction (5) at 1148 K. The primary contributions to the uncertainties in the rate constant are: (a) temperature (±1%), (b) mixture composition (±5%), (c) OH absorption coefficient (±3%), (d) wavemeter reading in the UV (±0.01 cm - 1 ), (e) fitting the data to computed profiles (±5%), (f) locating time-zero (±0.5 µs), (g) the rate constant for CH 3 + OH CH 2 (s) + H 2 O (uncert. factor = 2), (h) the rate constant for CH 3 OH (+ M) CH 3 + OH (+ M) (uncert. factor = 2), and (i) the rate constant for C 2 H 6 (+ M) CH 3 + CH 3 (+ M) (±20%). As shown in Figure 6.3, the individual error sources were introduced separately and their effects on the rate constant for reaction (5) were determined. These uncertainties were combined in a root-sum-squared method to give an overall uncertainty estimate of ±28% at 1148 K. Figure 6.3: Uncertainty analysis for the rate constant of CH 3 COCH 3 + OH products at 1148 K and 1.95 atm. Figure 6.4 shows the Arrhenius plot for the present rate constant measurements of reaction (5) at T = K, along with the previous measurements of Vasudevan et al. [119] from the same laboratory. The current measurements agree well with the previous values (within ±5%). These measured values can then be expressed in Arrhenius form as k 5 = exp(-2437/t) cm 3 mol -1 s -1 over K. Bott and Cohen [118] also utilized TBHP as the OH precursor and employed both the shock tube 96

123 and UV lamp absorption method at 309 nm to monitor the OH decay and study reaction (5) near 1200 K and 1 atm. The current measurements are consistent with Bott and Cohen s measured value within 20%. In addition, Srinivasan et al. [77] used a similar method to investigate the rate constant for reaction (5) and determined a rate constant of cm 3 mol -1 s -1 over K. Their value is in close accord with our previous and current measurements. Figure 6.4 also shows the rate constants for reaction (5) adopted by two different detailed mechanisms: Pichon et al. [89] and Herbinet et al. [45]. The values of k 5 from the original Pichon et al. mechanism are approximately 24% and 43% faster than the current measurements at 1000 K and 1250 K, respectively; the values employed from the Herbinet et al. mechanism are in excellent agreement with the current measured values (within ±11%). Additionally, a theoretical calculation from Zhou et al. [121], which modeled all possible abstraction channels, was performed using the computationally less expensive G3 and G3MP2BH&H methods to calculate the energy barriers and using the Variflex code including Eckart tunneling corrections to compute the total rate constants for the reactions of OH with ketones (acetone, 2-butanone, and isopropyl methyl ketone) over K. As shown in Figure 6.4, the computed values from Zhou et al. are consistently lower than all high-temperature experimental data by ~55%. Figure 6.4: Arrhenius plot for acetone + OH (k 5 ) at temperatures above 833 K. 97

124 Butanone + OH Kinetics The OH sensitivity analysis was also carried out for the rate constant determination of 2-butanone + OH products (reaction (6)) using the mixture of 152 ppm 2-butanone with 14 ppm TBHP (and 41 ppm water) in Ar at 1039 K and 1.41 atm, as shown in Figure 6.5. Note that reaction (6) consists of 3 different abstraction channels, as described in the original Serinyel et al. mechanism [95, 98]: C 2 H 5 COCH 3 + OH CH 2 CH 2 COCH 3 + H 2 O (6a) C 2 H 5 COCH 3 + OH CH 3 CHCOCH 3 + H 2 O (6b) C 2 H 5 COCH 3 + OH C 2 H 5 COCH 2 + H 2 O (6c) At 1100 K, channel (6b) is the dominant pathway with a branching ratio of 0.53 due to the weaker C-H bond energy at the secondary site, and channel (6a) is the next most important pathway with a branching ratio of However, channel (6c) is nearly insignificant with a branching ratio of More importantly, reaction (6) is the most sensitive reaction at the conditions depicted in Figure 6.5, with some minor interference from the secondary reactions (reactions (13), (17) and (18)). As shown here, as temperature decreases, the reaction for TBHP decomposition becomes more important at the early times. Figure 6.5: OH sensitivity plot for the rate constant measurement of 2- butanone + OH at 1039 K and 1.41 atm. 98

125 Figure 6.6 shows an example of the overall rate constant measurement (k 6 = k 6a + k 6b + k 6c ) for reaction (6) at 1039 K and 1.41 atm. The mixture is 152 ppm 2-butanone in Ar, with the measured peak OH yield to be ~14 ppm. Thus, we infer that the initial TBHP mole fraction is 14 ppm. The model predictions from the modified Serinyel et al. mechanism with the best-fit overall rate constant of k 6 = cm 3 mol -1 s -1 and the variations of ±50% in the inferred rate constant are also shown in Figure 6.6. Due to the near-pseudo-first-order conditions, the measured overall rate constant should be insensitive to the branching ratios of the individual channels. The effect of the branching ratios on the rate constant determination was also investigated at 1039 K by interchanging the branching ratios of channels (6a) and (6b) while maintaining the total value, and a negligible change in the inferred rate constant was found. In addition, a detailed error analysis (similar to the analysis for reaction (5)) was performed for the rate constant measurement of reaction (6) at 1039 K and 1.41 atm, and the overall uncertainty was estimated to be ±22%. Table 6.2 summarizes the rate constant measurements of reaction (6) at K and atm. Three different mixture compositions were employed to confirm that the inferred rate constants are independent of any secondary chemistry effects. Figure 6.6: Sample 2-butanone + OH rate constant measurement using the mixture of 152 ppm 2-butanone with ~14 ppm TBHP (and 41 ppm water) in Ar at 1039 K and 1.41 atm. Simulation from the modified Serinyel et al. mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown. 99

126 Table 6.2: C 2 H 5 COCH 3 + OH Products: Rate Constant Data. T 5 [K] P 5 [atm] k 6 [cm 3 mol -1 s -1 ] 55 ppm TBHP (and water), 152 ppm C 2 H 5 COCH 3, Ar E E E E E E ppm TBHP (and water), 161 ppm C 2 H 5 COCH 3, Ar E E E E E ppm TBHP (and water), 206 ppm C 2 H 5 COCH 3, Ar E E E E E+13 Figure 6.7 shows the Arrhenius plot for the overall rate constant measurements of reaction (6) at T > 833 K, along with the estimated values adopted in the original Serinyel et al. mechanism [95, 98] and the theoretical values from Zhou et al. [121]. The measured values can be expressed in Arrhenius form as k 6 = exp(-2270/t) cm 3 mol -1 s -1 over K. The values used in the original Serinyel et al. mechanism are ~40% lower than the measurements. Interestingly, the theoretical values from Zhou et al. are in excellent agreement with the measurements within 10%. Note that the measurements and the theoretical calculations both exhibit some slight non-arrhenius curvature at the present test conditions. 100

127 Figure 6.7: Arrhenius plot for 2-butanone + OH (k 6 ) at temperatures above 833 K Pentanone + OH Kinetics As illustrated in Figure 6.8, the OH sensitivity reveals that the reactions of OH with 3-pentanone are the dominant pathways for the consumption of OH at 1188 K and 1.94 atm. In particular, reaction (7) consists of two channels, in which the OH radical can abstract the H-atom from 3-pentanone at the primary or secondary site. C 2 H 5 COC 2 H 5 + OH CH 2 CH 2 COC 2 H 5 + H 2 O (7a) C 2 H 5 COC 2 H 5 + OH CH 3 CHCOC 2 H 5 + H 2 O (7b) Based on the original Serinyel et al. mechanism [95, 98], the branching ratios of channels (7a) and (7b) are 0.42 and 0.58, respectively, at 1188 K. In addition, there is some minor interference from the following reactions at later times: CH 3 + OH CH 2 (s) + H 2 O (17) C 2 H 4 + H (+ M) C 2 H 5 (+ M) (20) CH 3 COCH 3 + OH CH 3 COCH 2 + H 2 O (5) In the current analysis, the rate constant for reaction (5) was updated with the Arrhenius expression from Section 6.3.2, with an uncertainty of approximately ±28%. In addition, 101

128 the rate constant for reaction (20) adopted by the original Serinyel et al. mechanism was used, and we assumed that its uncertainty is approximately a factor of 2. Figure 6.8: OH sensitivity plot for the rate constant measurement of 3- pentanone + OH at 1188 K and 1.94 atm. Figure 6.9 shows a representative OH time history trace at 1188 K and 1.94 atm using the mixture of 213 ppm 3-pentanone with 17 ppm TBHP (and 59 ppm H 2 O) in Ar. The model predictions from the modified Serinyel et al. mechanism with the best-fit rate constant of k 7 = cm 3 mol -1 s -1 and the variations of ±50% in k 7 are also shown in Figure 6.9. Note that the overall rate constant for reaction (7) is insensitive to the branching ratios of its individual channels due to the near-pseudo-first-order conditions. A detailed error analysis was then conducted for k 7 at 1188 K and 1.94 atm, and the overall uncertainty was estimated to be ±20%. 102

129 Figure 6.9: Sample 3-pentanone + OH rate constant measurement using the mixture of 213 ppm 3-pentanone with ~17 ppm TBHP (and 59 ppm water) in Ar at 1188 K and 1.94 atm. Simulation from the modified Serinyel et al. mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown. Table 6.3 summarizes the overall rate constant determinations of reaction (7) at T = K and P = atm. Note that two different mixture compositions were used to confirm that the inferred rate constants are free of any secondary chemistry effects, and the values determined from these two mixtures are consistent with each other. Figure 6.10 shows the Arrhenius plot for our measured values, along with the estimated values in the original Serinyel et al. mechanism [95, 98], at temperatures above 833 K. The measured values can be expressed in Arrhenius form as k 7 = exp(- 2361/T) cm 3 mol -1 s -1 over K. Interestingly, the values for reaction (7) from the original Serinyel et al. mechanism were estimated by analogy with the H-atom abstraction rate constants from alkanes [98], and their values are in close accord with the present measurements within ±5%. As is evident in Figure 6.10, the measurements and the estimated values from Serinyel et al. both experience slight non-arrhenius curvature at the current experimental conditions. Additionally, based on the theoretical study of the reactions of OH with ketones from Zhou et al. [121], the expressions of the group rate constants (on a per H-atom 103

130 basis) for different carbon atom sites (primary, secondary, and tertiary carbon atoms) were provided. In the present analysis, we can estimate the overall rate constant for reaction (7) using these group rate constants, and the estimated rate constant is k 7 = 6 k(ch 3 CH 2 C(O)) + 4 k( CH 2 C(O)), where k(ch 3 CH 2 C(O)) and k( CH 2 C(O)) are the group rate constants (per H-atom) for the primary carbon atom adjacent to the CH 2 C(O) group and for the secondary carbon atom adjacent to the C(O) group, respectively. As shown in Figure 6.10, the estimated values are in good agreement with the measurements within 15%. Table 6.3: C 2 H 5 COC 2 H 5 + OH Products: Rate Constant Data. T 5 [K] P 5 [atm] k 7 [cm 3 mol -1 s -1 ] 76 ppm TBHP (and water), 213 ppm C 2 H 5 COC 2 H 5, Ar E E E E E ppm TBHP (and water), 211 ppm C 2 H 5 COC 2 H 5, Ar E E E ppm TBHP (and water), 211 ppm C 2 H 5 COC 2 H 5, Ar E E E E E ppm TBHP (and water), 151 ppm C 2 H 5 COC 2 H 5, Ar E E E E E E

131 Figure 6.10: Arrhenius plot for 3-pentanone + OH (k 7 ) at temperatures above 833 K Pentanone + OH Kinetics As mentioned previously, a comprehensive mechanism for high-temperature 2- pentanone kinetics is not available in the literature. In the present work, we have assumed the pathways for the reactions of OH with 2-pentanone to be similar to those of methyl butanoate [54]. C 3 H 7 COCH 3 + OH C 2 H 4 + CH 3 COCH 2 + H 2 O (8a) C 3 H 7 COCH 3 + OH C 3 H 6 + CH 3 CO + H 2 O (8b) C 3 H 7 COCH 3 + OH C 2 H 5 CHCO + CH 3 + H 2 O (8c) C 3 H 7 COCH 3 + OH n-c 3 H 7 + CH 2 CO + H 2 O (8d) Channel (8a) describes the H-atom abstraction from 2-pentanone at the γ site to form a CH 2 CH 2 CH 2 COCH 3 radical and a H 2 O molecule. Through β-scission, the fuel radical decomposes very rapidly to form a C 2 H 4 molecule and a CH 3 COCH 2 radical. Due to the rapid decomposition of the fuel radical, we assumed that the products from the fuel radical are formed immediately after the H-atom abstraction. Similarly, channels (8b) and (8c) describe the H-atom abstraction from 2-pentanone at the β and α sites, respectively. Due to its similar structure to methyl butanoate, the rate constants for 105

132 channels (8a)-(8c) were approximated to be the same as the rate constants for methyl butanoate (MB) + OH reactions at the α, β and γ sites, and these values for MB + OH reactions were obtained from the Dooley et al. mechanism [54]. Among these three channels, channel (8b) (the H-atom abstraction at the β position) should be the fastest route for the removal of OH, which was also suggested in previous experimental studies [38, 107]. In addition, the rate constant for channel (8d) was assumed to be the same as that of channel (6c) (C 2 H 5 COCH 3 + OH C 2 H 5 COCH 2 + H 2 O). The resulting branching ratios of channels (8a)-(8d) at 1186 K are 0.23, 0.38, 0.37 and These four channels were then incorporated in the modified Serinyel et al. mechanism. As expected, the estimated branching ratios of channels (8a)-(8d) have no discernible effect on the determinations of the overall rate constant at near-pseudo-first-order conditions. In the present analysis, we also included the pathways for the reactions of H with 2-pentanone in the modified Serinyel et al. mechanism, which can be described as follows: C 3 H 7 COCH 3 + H C 2 H 4 + CH 3 COCH 2 + H 2 (29a) C 3 H 7 COCH 3 + H C 3 H 6 + CH 3 CO + H 2 (29b) C 3 H 7 COCH 3 + H C 2 H 5 CHCO + CH 3 + H 2 (29c) C 3 H 7 COCH 3 + H n-c 3 H 7 + CH 2 CO + H 2 (29d) In a similar way, the rate constants for channels (29a)-(29c) were assumed to be the same as the rate constants for the reactions of H with methyl butanoate at the α, β and γ sites, and these values were also adopted from the Dooley et al. mechanism [54]. Additionally, the rate constant for channel (29d) was assumed to be the same as that of 2-butanone (C 2 H 5 COCH 3 + H C 2 H 5 COCH 2 + H 2 ). Interestingly, the addition of reactions (29a)- (29d) has negligible influence on the overall rate constant determinations of reaction (8). Figure 6.11 shows a representative OH time history trace at 1186 K and 1.30 atm using the mixture of 161 ppm 2-pentanone with 15 ppm TBHP (and 45 ppm H 2 O) in Ar. The simulations from the modified Serinyel et al. mechanism with the best-fit rate constant of k 8 = cm 3 mol -1 s -1 and the variations of ±50% in k 8 were also illustrated. A detailed error analysis was then conducted to estimate the overall uncertainty in k 8 at 1186 K, and the uncertainty was found to be ±24%. 106

133 Figure 6.11: Sample 2-pentanone + OH rate constant measurement using the mixture of 161 ppm 2-pentanone with ~15 ppm TBHP (and 45 ppm water) in Ar at 1186 K and 1.30 atm. Simulation from the modified Serinyel et al. mechanism for the best-fit rate constant, along with perturbations of ±50%, is also shown. Table 6.4 summarizes the overall rate constant measurements of reaction (8) at K and atm, and Figure 6.12 also presents the Arrhenius plot for these measured values. These measured values are expressed in Arrhenius form as k 8 = exp(-2020/t) cm 3 mol -1 s -1 over K. It should be noted that the overall rate constant measurements for 2-pentanone + OH reaction are quite similar to the values obtained for 3-pentanone + OH reaction at our experimental conditions. Additionally, Figure 6.12 presents the estimated overall rate constant for reaction (8) using the group rate constants for the reactions of OH with ketones developed by Zhou et al. [121], and the estimated values are in excellent agreement with the measurements within 7%. 107

134 Table 6.4: C 3 H 7 COCH 3 + OH Products: Rate Constant Data. T 5 [K] P 5 [atm] k 8 [cm 3 mol -1 s -1 ] 60 ppm TBHP (and water), 161 ppm C 3 H 7 COCH 3, Ar E E E E E ppm TBHP (and water), 161 ppm C 3 H 7 COCH 3, Ar E E E E E E+13 Figure 6.12: Arrhenius plot for 2-pentanone + OH (k 8 ) at temperatures above 900 K. 108

135 6.3.6 Comparison of Ketone + OH Kinetics Figure 6.13 presents the Arrhenius plot of the measured rate constants for reactions (5)-(8) at temperatures above 870 K. Among all four ketone + OH reactions, the reaction of OH with acetone has the lowest reactivity due to the fact that there are less C H bonds available for H-atom abstraction in acetone. As the number of C H bonds in the fuel molecule increases, the fuel + OH reaction becomes more reactive. As demonstrated in Figure 6.13, the rate constant for 3-pentanone + OH is much faster than that for 2-butanone + OH, because 3-pentanone has one more secondary carbon atom site than 2-butanone. In addition, the rate constant for 2-pentanone + OH is about the same as that for 3-pentanone + OH due to the fact that both molecules have the same number of primary and secondary carbon atom sites. Note that the rate constant measurements of the 2-pentanone + OH reaction are slightly higher than the values of the 3-pentanone + OH reaction at T < 1100 K. This can be explained by the fact that the secondary carbon atom site at the β position from 2-pentanone is supposed to be more reactive than that at the α position from 3-pentanone, as suggested in previous experimental studies [38, 107]. Figure 6.13: Arrhenius plot of the measured rate constants for reactions (5)-(8) at temperatures above 870 K. 109

136 6.4 Comparison with Low Temperature Data Figure 6.14 presents the current data along with some earlier measurements of reaction (5) at temperatures greater than 250 K. In the study from Wollenhaupt et al. [106], the rate constant for reaction (5) was measured over K at torr of Ar or N 2 bath gas using the pulsed laser photolysis technique to generate OH radicals from the sequential two-photon dissociation of NO 2 in the presence of H 2 at 439 nm, or from the photolysis of HONO at 351 nm. They monitored the OH radicals using either resonance fluorescence or laser-induced fluorescence detection scheme, and they also concluded that their measurements are independent of pressure. Similarly, Le Calvé et al. [38] and Gierczak et al. [109] studied k 5 over K by generating OH via pulsed laser photolysis and detecting it via laser-induced fluorescence, while Wallington and Kurylo [107] investigated k 5 over K using the flash photolysis/resonance fluorescence measurement technique. Moreover, Yamada et al. [117] examined k 5 over a wide temperature range K using the pulsed laser photolysis/pulsed laser-induced fluorescence technique. They then performed a detailed analysis using Variational Transition State theory and suggested that the dominant products of reaction (5) are CH 3 COCH 2 and H 2 O through direct abstraction at all temperatures (particularly above 450 K). Additionally, Tranter and Walker [120] added small amounts of acetone to slowly reacting mixtures of H 2 + O 2 at 753 K, and monitored the consumption of acetone and H 2 with the use of gas chromatography. This method allowed them to infer the relative rate constant for reaction (5) at 753 K. It is pertinent to note that these lowtemperature measurements are in excellent agreement with each other. As is evident in Figure 6.14, the rate constants employed in the comprehensive mechanisms of Pichon et al. [89] and Herbinet et al. [45] are able to predict the low-temperature data reasonably well over K, but not for T < 298 K. In particular, the rate constant from Herbinet et al. provides much better agreement with the existing high-temperature data. In addition to the values from the detailed mechanisms, the theoretical calculation from Zhou et al. [121] agrees well with earlier low-temperature measurements (at T < 500 K), but the calculated values are at least 40% lower than the measurements at T > 500 K. 110

137 Figure 6.14: Arrhenius plot for acetone + OH products (k 5 ) at all temperatures. Figure 6.15 shows the Arrhenius plot for the overall rate constant measurements of reaction (6) at temperatures greater than 250 K. Kinetic measurements of reaction (6) were performed at room temperature by different researchers using both relative [ ] and absolute [38, 107, ] methods. In general, these room temperature measurements are in close accord with each other, except for the value obtained from Atkinson et al. [112]. Concurrently, the rate constant for reaction (6) was examined as a function of temperature ( K) by Wallington and Kurylo [107] using the flash photolysis/resonance fluorescence technique and by Le Calvé et al. [38], Carr et al. [110] and Jimenez et al. [111] using the pulsed laser photolysis/laser-induced fluorescence technique. Their measurements are in excellent agreement with each other, and no pressure dependence can be found at their experimental conditions. Based on these lowtemperature data, k 6 exhibits only slight positive temperature dependence over K. In addition to the rate constant determination for acetone + OH reaction, Tranter and Walker [120] measured the relative rate constant for reaction (6) at 753 K. Figure 6.15 also presents the estimated values of k 6 from Serinyel et al. [95] and the theoretical values from Zhou et al. [121]. As described previously, the calculated values from Zhou et al. are consistent with the present high-temperature data (at T > 879 K) within 10%. However, the calculated values are faster than the earlier low-temperature data by a factor 111

138 of 2 at 500 K and by a factor of 6 at 250 K. Consequently, the theoretical study predicts a pronounced negative temperature dependence of k 6 over K, and this effect does not appear in the existing data. On the other hand, the estimated values from Serinyel et al. are ~40% lower than the current high-temperature data, and are in good agreement with the low-temperature data over K. In addition, the overall rate constant from Serinyel et al. does not exhibit any negative temperature dependence over K. Figure 6.15: Arrhenius plot for 2-butanone + OH products (k 6 ) at all temperatures. Figure 6.16 also shows the current high-temperature data (at T > 878 K) and three previous low-temperature measurements (at T < 800 K) for the reaction of OH with 3- pentanone, along with the rate constant from the original Serinyel et al. mechanism [98]. As compared to acetone and 2-butanone, fewer experimental and theoretical studies are available in the literature. Tranter and Walker [120] measured the relative rate constant for reaction (7) at 753 K (using the same approach as the one they employed for reactions (5) and (6)). Atkinson et al. [116] also measured the relative rate constant for reaction (7) at 299 K using methyl nitrite (CH 3 ONO) photolysis in air as a source of OH radicals. They monitored the organic reactants using gas chromatography with flame ionization detection. In their work, they took advantage of their previous knowledge on the rate constant for cyclohexane + OH reaction and inferred the rate constant for reaction (7) 112

139 from the ratio of k 7 /k cyclohexane+oh at 299 K. Moreover, Wallington and Kurylo [107] determined the absolute rate constant for reaction (7) over K using the flash photolysis/resonance fluorescence measurement technique, and they suggested that k 7 did not exhibit any temperature dependence at their test conditions. Furthermore, the rate constant from the original Serinyel et al. mechanism is able to predict the existing data rather accurately over K. Figure 6.16: Arrhenius plot for 3-pentanone + OH products (k 7 ) at all temperatures. Similarly, Figure 6.17 illustrates the current high-temperature data and previous low-temperature measurements [107, 111, 116] of reaction (8) over K. More importantly, the rate constant provided by Jimenez et al. [111] exhibits a pronounced negative temperature dependence over K, and this trend does not appear in the kinetic measurements of reactions (5)-(7). This pronounced negative temperature dependence of k 8 is mainly attributed to the H-atom abstraction from the CH 2 group in the β position, which is the predominant reaction pathway at low temperatures. For instance, Jimenez et al. [111] determined that the branching ratios of channels (8a)-(8d) are 0.04, 0.76, 0.18, and 0.02, respectively, at 298 K. Some researchers [107, 111] also postulated that the H-abstraction reaction could proceed via an OH-addition complex, 113

140 resulting in a six- or seven-membered ring complex which enhances the abstraction of an H atom from the CH 2 group in the β position. Figure 6.17: Arrhenius plot for 2-pentanone + OH products (k 8 ) at all temperatures. 6.5 Comparison with Structure-Activity Relationship The measured overall rate constants for reactions (5)-(8) over K can be compared with the estimated values using the structure-activity relationship (SAR) developed by Atkinson and his co-workers [ ]. Their method of calculating the rate constants for the reactions of OH with organic compounds is based on the estimation of primary ( CH 3 ), secondary ( CH 2 ), and tertiary ( CH<) group rate constants, and these group rate constants depend on the nature of the neighboring atoms (substituents bound to the groups). The group rate constants can be expressed as k(ch 3 X) = k prim F(X), k(y CH 2 X) = k sec F(X) F(Y), and k((z)ch(x)(y)) = k tert F(X) F(Y) F(Z), where k prim, k sec and k tert are the rate constants for the H-atom abstraction from CH 3, CH 2 and CH< groups, and F(X), F(Y) and F(Z) are the substituent factors. Recently, Pang et al. [74] have demonstrated that the SAR estimation accurately predicts the measured rate constants for n-alkane + OH reactions (i.e., n-pentane + OH, n-heptane + OH, and n- 114

141 nonane + OH) over K. In particular, the SAR estimation captures the temperature dependence of their measurements reasonably well. In addition, Kwok and Atkinson [129] provided a revised list of the substituent factors F(X) at 298 K, and they assumed that the temperature dependence of the substituent factors can be expressed in the form of F(X) = exp(e x /T). In the present analysis, the E x term in the preceding expression was calculated from the substituent factor at 298 K, thereby allowing us to determine the substituent factors at different temperatures. The estimated rate constants for reactions (5)-(8) based on the SAR approach are provided in Figures For instance, the estimated rate constant for 2-pentanone + OH reaction can be evaluated as k 8 = k prim F( CH 2 ) + k sec F(CH 3 ) F( CH 2 C(O)R) + k sec F( CH 2 ) F( C(O) ) + k prim F( CO ). It is pertinent to note that the substituent factor F( CH 2 C(O)R) is 3.9 at 298 K and is much higher than F( CH 2 ), which is 1.23 at 298 K. This confirms the increased reactivity of the H-atom abstraction at the β position, as observed by Atkinson et al. [116]. As expected, the SAR estimation shows good agreement with the kinetic measurements of reactions (5)-(8) at 298 K, but the estimated values are higher than the measured values over K. In particular, the estimated values are ~25% faster than the present high-temperature data over K. Nevertheless, the SAR estimation can capture the temperature dependence of reactions (5)-(8) reasonably well, implying that the pre-exponential factors for the group rate constants k prim and k sec should be reduced by ~25%, particularly for ketone + OH reactions. With this modification, the estimated rate constants for reactions (5)-(7) show excellent agreement with the measurements over a wide temperature range K (see Figures ). In addition, the modified SAR estimation for reaction (8) precisely predicts the present high-temperature data (at K), and is in close accord with the previous data from Wallington and Kurylo [107] and Jimenez et al. [111] at temperatures near 298 K, as seen in Figure

142 6.6 Summary The overall rate constants for the reactions of OH with acetone (k 5 ), 2-butanone (k 6 ), 3-pentanone (k 7 ) and 2-pentanone (k 8 ) were studied behind reflected shock waves over K at pressures of 1-2 atm using OH laser absorption. The present hightemperature measurements can be expressed in Arrhenius form as: k 5 = exp(-2437/t) cm 3 mol -1 s -1 k 6 = exp(-2270/t) cm 3 mol -1 s -1 k 7 = exp(-2361/t) cm 3 mol -1 s -1 k 8 = exp(-2020/t) cm 3 mol -1 s -1 Detailed error analyses, which account for both experimental and secondary chemistry contributions, yielded the uncertainty estimates of ±28% at 1148 K for k 5, ±22% at 1039 K for k 6, ±20% at 1188 K for k 7, and ±24% at 1186 K for k 8. In addition, the structureactivity relationship (SAR) from Atkinson and his co-workers [ ] was used to estimate the rate constants for reactions (5)-(8), and the estimated values are in good agreement with the present high-temperature data (within ~25%). 116

143 Chapter 7 High-Temperature Measurements of the Reactions of OH with Small Methyl Esters: Methyl Formate, Methyl Acetate, Methyl Propanoate, and Methyl Butanoate 7.1 Introduction In this chapter, the overall rate constants for the reactions of OH with four small methyl esters, namely methyl formate (CH 3 OCHO), methyl acetate (CH 3 OC(O)CH 3 ), methyl propanoate (CH 3 OC(O)C 2 H 5 ), and methyl butanoate (CH 3 OC(O)C 3 H 7 ), were determined behind reflected shock waves over the temperature range of K at pressures near 1.5 atm: CH 3 OCHO + OH Products (9) CH 3 OC(O)CH 3 + OH Products (10) CH 3 OC(O)C 2 H 5 + OH Products (11) CH 3 OC(O)C 3 H 7 + OH Products (12) We believe these are the first direct high-temperature measurements of the overall rate constants for reactions (9)-(12). These kinetic data were compared with the values adopted in several detailed kinetic mechanisms and the estimates using the structureactivity relationship (SAR) developed by Atkinson and co-workers [ ]. 117

144 7.2 Experimental Details Test mixtures were prepared manometrically in a 40 liter stainless-steel tank heated uniformly to 50 o C and mixed with a magnetically-driven stirring vane. A doubledilution process was employed to allow for more accurate pressure measurements in the manometrical preparation of a highly dilute mixture. A more concentrated mixture was first prepared and mixed for at least 2 hours to ensure homogeneity and consistency, and the mixture was then further diluted with argon and mixed for additional 2 hours prior to the experiments. The gas utilized in this study was argon (Research Grade) %, which was supplied by Praxair and used without further purification. The liquid chemicals were commercially available 70% tert-butyl hydroperoxide (TBHP) in water, methyl formate ( 99%), methyl acetate ( 99%), methyl propanoate ( 99%), and methyl butanoate ( 99%) from Sigma-Aldrich, and were purified using a freeze-pump-thaw procedure to remove dissolved volatiles and air prior to mixture preparation. The mixture composition was confirmed by sampling a portion of the mixture (from near the endwall) into an external multi-pass absorption cell with a path length of 29.9 m and monitoring the fuel concentration in the cell with a Jodon Helium-Neon laser at 3.39 µm [82, 122]. Beer s law was then used to convert the measured absorption data into the fuel mole fraction. The absorption cross-sections of methyl esters for Beer s law were directly obtained from the PNNL database [123], and the measured fuel concentrations were consistent with the values expected from the manometrical preparation within ±5%. 7.3 Kinetic Measurements A total of 52 reflected shock wave experiments were performed to determine the overall rate constants for the reactions of OH with four methyl esters (methyl formate, methyl acetate, methyl propanoate, and methyl butanoate) over K at pressures near 1.5 atm. Experiments were carried out using different initial fuel concentrations: methyl formate (322 ppm, 404 ppm), methyl acetate (323 ppm, 384 ppm), methyl 118

145 propanoate (~281 ppm), and methyl butanoate (241 ppm, 270 ppm). Test mixtures with individual methyl esters and ppm TBHP (and water) diluted in argon were utilized in the present study. Note that dilute mixtures were preferred in order to minimize the temperature change resulted from the chemistry effects, and the temperature profile behind the reflected shock wave (from the present study) was nearly constant (less than 1 K change based on the calculation from CHEMKIN PRO [71]) over the time frame of the experiment (the first 100 µs) Choice of Kinetic Mechanisms The CHEMKIN PRO package [71] was used to simulate the OH time histories under the standard constant energy and volume assumption. A comprehensive chemical kinetic mechanism of Dooley et al. [49] was chosen as the base mechanism for methyl formate and methyl acetate. This mechanism can successfully simulate shock tube ignition delay times, laminar burning velocities of outwardly propagating spherical flames, and speciation data from a shock tube and a variable-pressure flow reactor [49-50] during methyl formate pyrolysis and oxidation. Additionally, this kinetic mechanism incorporates the sub-mechanism for methyl acetate, which was previously developed by Westbrook et al. [53]. The sub-mechanism for methyl acetate consists of the unimolecular decomposition pathways and the H-atom abstraction reactions by H, OH, and CH 3 radicals, and was validated against speciation data from fuel-rich, low-pressure, premixed laminar flames. A detailed kinetic mechanism of Dooley et al. [54] was also selected as the base mechanism for methyl propanoate and methyl butanoate. This mechanism was originally developed to predict the autoignition of methyl butanoate in a shock tube and a rapid compression machine over a wide range of experimental conditions, and was further validated against speciation data available in the literature from a flow reactor, a jet-stirred reactor, and an opposed-flow diffusion flame. As were done in Chapters 3 and 6, tert-butyl hydroperoxide (TBHP or (CH 3 ) 3 CO OH) was used as an OH radical precursor at the present experimental conditions, and the TBHP sub- 119

146 mechanism was also implemented into the base mechanisms for these methyl ester + OH studies. (Please read Chapter 3 for more details on the TBHP chemistry.) Methyl Formate (MF) + OH Kinetics The reaction of OH with methyl formate consists of 2 different channels: CH 3 OCHO + OH CH 3 OCO + H 2 O (9a) CH 3 OCHO + OH CH 2 OCHO + H 2 O (9b) The branching ratios of channels (9a) and (9b) are 0.32 and 0.68, respectively, at 1168 K, based on the Dooley et al. mechanism [49]. In their analysis, the estimated rate constant for channel (9a) was assumed to be an intermediate value between typical primary and secondary C H bonds (as in propane) due to the weaker bond strength of the CH 3 OCO H position (100.1 kcal/mol at 298 K). Similarly, the estimated rate constant for channel (9b) (per H-atom) was assumed to be 5% faster than the value for a typical primary C H bond, and the corresponding bond strength was estimated to be kcal/mol at 298 K. An OH radical sensitivity analysis for the mixture of 322 ppm methyl formate with 26 ppm TBHP (and 70 ppm H 2 O) in Ar at 1168 K and 1.40 atm is shown in Figure 7.1. The analysis reveals that the reaction of OH with methyl formate (reaction (9)) is the dominant reaction over the time frame of the experiment, with some minor interference from the secondary reactions: CH 3 + OH CH 2 (s) + H 2 O (17) C 2 H 6 (+ M) CH 3 + CH 3 (+ M) (18) CH 2 O + OH HCO + H 2 O (30) In this modeling, the rate constants for reactions (5) and (17)-(19) in the Dooley et al. mechanism [49] were updated with the values in Table 3.1. The rate constant for reaction (30) was previously measured using UV laser absorption of OH near 307 nm behind reflected shock waves over K at pressures near 1.6 atm by Vasudevan et al. [126], and their measured rate constant was also adopted in the present study. 120

147 Figure 7.1: OH sensitivity plot for the rate constant measurement of methyl formate + OH at 1168 K and 1.40 atm. Figure 7.2 illustrates a sample measured OH concentration time history for the mixture of 322 ppm methyl formate in Ar at 1168 K and 1.40 atm, and the measured peak OH concentration is approximately 26 ppm. According to the measured peak OH yields, the mixtures with 96 ppm TBHP/water are comprised of ~25-28 ppm TBHP in the present study. It should also be noted that the presence of H 2 O in the test mixtures does not have any significant influence on the computed OH profiles. As shown in Figure 7.2, a best-fit overall rate constant for reaction (9) of cm 3 mol -1 s -1 was obtained between the experimental data and the simulation at 1168 K and 1.40 atm. The simulations for the perturbations of ±50% in the inferred rate constant are also shown in Figure 7.2. Additionally, the effect of the branching ratios for reaction (9) on the overall rate constant determination was tested at 1168 K by interchanging the branching ratios of channels (9a) and (9b) while maintaining the overall value, and no discernible effect could be observed from the simulated OH profiles. Hence, the original branching ratios proposed by Dooley et al. [49] were kept in our simulations. In addition, Table 7.1 summarizes the overall rate constant measurements (k 9 = k 9a + k 9b ) of reaction (9) at T = K and P = atm. 121

148 Figure 7.2: Sample methyl formate + OH rate constant measurement using the mixture of 322 ppm methyl formate with ~26 ppm TBHP (and 70 ppm water) in Ar at 1168 K and 1.40 atm. Simulation from the Dooley et al. mechanism [49] for the best-fit rate constant, along with perturbations of ±50%, is also shown. Table 7.1: CH 3 OCHO + OH Products: Rate Constant Data. T 5 [K] P 5 [atm] k 9 [cm 3 mol -1 s -1 ] 96 ppm TBHP (and water), 322 ppm CH 3 OCHO, Ar E E E E E E E E E E ppm TBHP (and water), 404 ppm CH 3 OCHO, Ar E E E E E

149 A detailed error analysis was conducted to estimate the overall uncertainty of the measured rate constant for reaction (9) at 1168 K. The primary contributions to the overall uncertainty in k 9 were considered: (a) temperature (±1%), (b) mixture composition (±5%), (c) OH absorption coefficient (±3%), (d) wavemeter reading in the UV (±0.01 cm -1 ), (e) fitting the data to the simulated profiles (±5%), (f) locating timezero (±0.5 µs), (g) the rate constant for CH 3 + OH CH 2 (s) + H 2 O (uncert. factor = 2), (h) the rate constant for CH 2 O + OH HCO + H 2 O (uncert. factor = 2), and (i) the rate constant for C 2 H 6 (+ M) CH 3 + CH 3 (+ M) (±20%). As demonstrated in Figure 7.3, the individual error sources were introduced separately (within the positive and negative bounds of their 2σ uncertainties) and their effects on the overall rate constant for reaction (9) were studied. These uncertainties were combined in a root-sum-squared method to give an overall (2σ) uncertainty of ±24% at 1168 K. Similar error analyses were performed for k 9 at 913 K and 1289 K, and the overall uncertainties were estimated to be ±29% and ±18%, respectively. Figure 7.3: Uncertainty analysis for the rate constant of methyl formate + OH products at 1168 K and 1.40 atm. Figure 7.4 presents the Arrhenius plot for the overall rate constant measurements of reaction (9) at T = K, along with the estimated values proposed by Fisher et 123