You are on page 1of 141

ANALYSIS OF THE OXYACETYLENE FLAME UNDER DIAMOND DEPOSITION CONDITIONS USING FOURIER TRANSFORM INFRARED SPECTROSCOPY

by

ERIC SZETO

Submitted in partial fulfillment of the requirements for the degree of Master of Science

Thesis Adviser: Dr. Philip W. Morrison Jr.

Department of Chemical Engineering CASE WESTERN RESERVE UNIVERSITY January, 1999

ANALYSIS OF THE OXYACETYLENE FLAME UNDER DIAMOND DEPOSITION CONDITIONS USING FOURIER TRANSFORM INFRARED SPECTROSCOPY Abstract by ERIC SZETO

Diamond has many useful and attractive properties. These applications include scratchproof coatings, use as a cutting tool, and heat sinks in the microprocessor industry. Because of its many potential applications, many experiments have been performed to better understand the chemistry of diamond CVD using a low cost and high growth rate reactor system: the oxyacetylene torch. In order to get a better understanding of the chemistry in the oxyacetylene flame under diamond deposition conditions, Fourier transform infrared spectroscopy along with the HITRAN database were used to obtain quantitative information on various species in the flame. Unlike many researchers who have made qualitative or relative measurements of the species in the flame, this research gave a more quantitative determination of the species as well as the flame temperature. The oxyacetylene flame with a total flow rate of 2 slm was analyzed. One of the important parameters known to influence diamond deposition is the R ratio (flow of oxygen / flow of acetylene) and the axial distance of the substrate from the tip of the torch nozzle. Therefore, various R (0.90 to 1.10) and axial positions (0mm to 18mm from the tip of the nozzle) are analyzed for species concentration. The species

ii

concentrations at the position and R ratio that favors diamond deposition were compared to unfavorable conditions. In addition, open and enclosed flames were analyzed and compared since it is reported in literature that the enclosed flame is better for diamond growth.

iii

ACKNOWLEDGEMENTS

Many people should be acknowledged for their assistance, guidance, help, and support. First, I acknowledge the guidance and support of Dr. Philip W. Morrison Jr., whose efforts are immeasurable. I would also like to acknowledge the American Chemical Society for their financial support. I would also like to thank Wayne Schmidt for his help and suggestion for making and modifying the equipment used for this experiment. Cliff Hayman must be acknowledged for his help around the lab. Friends and family gave me tremendous support for this project. Special thanks goes to Shannon Tsang for her support throughout my project. In addition, I would like to thank my grandmother for taking care of me when I was young. Most of all, I would like to thank my mom and dad for giving me the opportunity and support for my education. Without them, I would have never been able to obtain a college education.

iv

TABLE OF CONTENTS Page List of Figures 1.0 Introduction 1.1 Diamond films and its applications 1.2 CVD Methods 1.3 The purpose of this work 2.0 Prior Research 2.1 Open Flames 2.2 Enclosed Flames 2.3 Diagnostic and Flames 2.3.1 Optical Emission Spectroscopy 2.3.2 Laser induced fluorescence 2.3.3 Absorption spectroscopy 2.3.4 Mass spectroscopy 2.4 Fourier Transform Infrared Spectroscopy (FTIR) - Background 2.5 FTIR Spectroscopy and CVD 2.5.1 Emission/transmission measurements 2.5.2 FTIR and oxyacetylene flames 2.6 HITRAN Database 3.0 Equipment Description 3.1 Torch Reactor viii 1 1 1 3 5 5 7 8 8 8 10 10 11 14 14 16 16 23 23

3.2 Spectrometer 3.3 Enclosed Flame Apparatus 3.4 Operational Procedure 4.0 Methods To Eliminate Flame Noise 4.1 Problem Statement 4.2 Methods to Eliminate Flame Noise 4.2.1 Two detector configuration 4.2.2 Pointing the flame upward 4.2.3 Apertures 4.2.4 Blocking the CO and CO2 absorptions 4.2.5 Long-pass infrared filters 5.0 Analysis of the Open Flame 5.1 E/T Temperature Measurements of the Open Flame 5.2 CO2 Measurements of the Open Flame 5.3 CO Measurements of the Open Flame 5.4 H2O Measurements of the Open Flame 5.5 OH Measurements of the Open Flame 5.6 C2H2 Measurements of the Open Flame 5.7 Discussion of Results 5.8 Comparison of Results to Previous Work 6.0 Analysis of the Enclosed Flame 6.1 Equipment Modifications

23 25 25 31 31 34 34 35 37 38 40 55 55 58 59 60 62 63 64 68 86 86

vi

6.1.1 Linear and non-linear detectors 6.1.2 Purging the arm of the spectrometer 6.2 Temperature Measurements in the Enclosed Flame 6.3 CO Measurements in the Enclosed Flame 6.4 CO2 Measurements in the Enclosed Flame 6.5 H2O Measurements in the Enclosed Flame 6.6 OH Measurements in the Enclosed Flame 6.7 C2H2 Measurements in the Enclosed Flame 6.8 CH4 Measurements in the Enclosed Flame 6.9 Discussion of the Enclosed Flame 6.10 Oxygen Entrainment in the Enclosed Flame 6.11 Comparison of the Free and Enclosed Flame 7.0 Conclusion 8.0 Future Work Comprehensive References Appendix I

86 87 88 91 92 92 92 93 93 94 95 96 112 115 116 123

vii

LIST OF FIGURES Page Figure 1.0-1 Regions of the acetylene flame. Figure 2.4-1 Components of the Michelson interferometer under transmission mode. Figure 2.4-2 Optical path in the interferometer under emission mode. Figure 2.6-1 HITEMP calculation of the water bands at 2500K, 0.055 atm cm in 1 atm. Figure 2.6-2 Smoothed spectrum after applying conv65.ab smoothing algorithm. Figure 3.1-1 Setup of the torch reactor. Figure 3.2-1 Bomem spectrometer with the extension arm and optics. Figure 3.3-1 Side view of enclosed flame apparatus. Figure 3.3-2 Top and bottom view of the enclosed flame apparatus. Figure 3.3-3 Water emission from the flame at 18 mm position and R=0.90 with and without the enclosed flame apparatus. Figure 4.1-1 Single scan interferogram of the background. Figure 4.1-2 Single scan interferogram of the flame at the nozzle. Total flow = 2 slm, R=1.0. Figure 4.1-3 Power spectrum of a single beam background without the flame. (one scan). Figure 4.1-4 Power spectrum of a single beam spectrum with the flame and globar on (one scan). 4 21 21 22 22 28 28 29 29 30

41 41 42 42

viii

Figure 4.1-5 Power spectrum of a single beam spectrum of the flame in the region of interest. (one scan) Total flow = 2 slm, R=1.0. Figure 4.1-6 Power spectrum of the single spectrum of the flame with 75 scans and globar on. Total flow = 2 slm, R=1.0. Figure 4.2-1 Two detectors used to collect noise spectra from the flame at 90 degrees. Colored bars represent collected IR. Figure 4.2-2 Flame noise detected by first DTGS detector at 0! using Labview for data acquisition. Figure 4.2-3 DTGS degree at 90! from the first one using Labview for data acquisition. Figure 4.2-4 Subtraction of the two spectra obtain at 90! from each other. Figure 4.2-5 Interferogram of the flame pointing down. Flame is at 2 slm total flow, R=0.94 at the tip of the torch. Single scan. Globar on. Figure 4.2-6 Interferogram of the flame pointing up. Flame is at 2 slm total flow, R=0.94 at the tip of the torch. Single scan with globar on. Figure 4.2-7 Power spectrum of the flame pointing down and globar off. Flame is at 2 slm total flow, R=0.94 at the tip of the torch. Single scan. Figure 4.2-8 Power spectrum of the flame pointing up with globar off. Flame is at 2 slm total flow, R=0.94 at the tip of the torch. Single scan. Figure 4.2-9 Power spectrum of single scan flame pointing down and globar on in the 2000-3000 cm-1 region. Flame is at 2 slm total flow, R=0.94 at the tip of the torch. Single scan. Figure 4.2-10 Power spectrum of single scan flame pointing up with globar on. Flame is at 2 slm total flow, R=0.94 at the tip of the torch. Single scan.

43

43

44 45 45 45 46

46

47

47

48

48

ix

Figure 4.2-11 Power spectrum of single scan flame without substrate. Figure 4.2-12 Power spectrum of single scan flame with substrate in 0 2000 cm-1 region. Figure 4.2-13 Power spectrum of flame with out substrate in 2000-4000 cm-1 region. Figure 4.2-14 Power spectrum of single scan flame with substrate in 2000-4000 cm-1 region. Figure 4.2-15 Power spectrum of a equivalent single scan of flame noise with a polyethylene bag using a DTGS detector. Globar was off. Figure 4.2-16 Power spectrum of equivalent single scan of flame noise with CO in bag using DTGS detector. Globar off. Figure 4.2-17 Power spectrum of the single scan flame without polyethylene bag and globar on. Flame is pointing down. Figure 4.2-18 Power spectrum of single scan flame with polyethylene bag filled with CO2 and globar on. Flame is pointing down. Figure 4.2-19 Transmission spectrum of the infared filter. This was taken at cm-1 using a DTGS detector. Single scan. Figure 4.2-20 Power spectrum of a flame pointing down without filter. Globar is on, single scan. Figure 4.2-21 Power spectrum of a single scan flame pointing down with filter. Globar is on, single scan. Figure 5.1-1 Calculated Radiance from the transmission data: (1-transmission)*blackbody(1900K). Flame conditions: Total flow = 2 slm R=0.90 at 0 mm position. Figure 5.1-2 Experimental radiance. Flame conditions: Total flow = 2 slm R=0.90 at 0 mm position. Figure 5.2-1 Experimental spectra of the CO2 hot emission bands. Flame conditions: Total flow = 2 slm R=0.90 at 0 mm position.

49 49 50 50 51

51 52 52

53 53 54 71

71 72

Figure 5.2-2 Overlap spectra of experimental and HITRAN92 (0.013 atm cm CO2 in 1 atm 2500K; Voigt lineshape) calculated. Flame conditions: Total flow = 2 slm R=0.90 at 0 mm position. Figure 5.3-1 Calculated Spectrum from HITRAN96 0.11 atm cm in 1 atm at 2700K; Voigt lineshape. Figure 5.3-2 Combined spectra of calculated CO (0.11 atm cm in 1 atm 2700K, HITRAN96 Voigt lineshape) and CO2 (0.013 atm cm 2500K, HITRAN92 Voigt lineshape). Figure 5.3-3 Comparing calculated spectrum to experimental spectrum. Calculated parameters: CO2: 0.013 atm cm in 1 atm 2500K Voigt lineshape. CO: 0.11 atm cm in 1 atm 2700K Voigt lineshape. Experimental conditions: Total flow = 2 slm R=0.90 at 0 mm position. Figure 5.3-4 Experimental Spectrum of flame: Total flow = 2 slm R=1.10 at 0 mm position. Figure 5.3-5 Difference of R = 0.90 minus R=1.10. Experimental conditions: Ft=2 slm at the 0 mm position. Figure 5.4-1 Experimental and calculated spectra in the water band region. Calculated paramters: HITEMP 0.03 atm cm in 1 atm Voigt lineshape at 3000K and 1900K Flame conditions: Ft=2 slm R=0.90 at 0 mm position. Figure5.4-2 Experimental (0 mm position, R=0.90, Ft=2 slm ) plotted with the calculated water bands (0.03 atm cm 3000K) Calculated spectrum is offset for clarity. Figure 5.4-3 Comparison of room temperature water with 3000K water transmission. Calculated parameters: 1 atms total, Voigt lineshape. HITEMP. Figure 5.4-4 Transmission spectrum of 0 mm position aith R =0.90 showing hot water bands.

72

73 73

74

74 75

75

76

77

77

xi

Figure 5.5-1 Experimental (0 mm position R=0.90 Open flame, Ft=2slm) plotted with the calculated spectra including the OH. bands H2O: 0.03 atm cm in 1 atm at 3000K HITEMP; Voigt. lineshape OH: 0.006 atm cm in 1 atm HITRAN 96; Voigt lineshape. Calculated spectrum is offset for clarity. Figure 5.5-2 Calculated spectrum minus experimental spectrum for the water band region. Taken from flame position 0 mm with R = 0.90 Ft=2 slm. Calculated parameters: H2O: 0.03 atm cm in 1 atm at 3000K HITEMP Voigt lineshape. OH: 0.006 atm cm in 1 atm HITRAN 96; Voigt lineshape; * denotes 3587 cm-1 band. Figure 5.6-1: Comparing calculated to experimental acetylene peak at 730 cm-1. Experimental conditions: 0.048 atm in 1 atm, 15 cm pathlength, and 300K. Calculated parameters: 0.048 atm, in 1 atm 15 cm, 300K, and Voigt lineshape. Figure 5.6-2: Comparing calculated to experimental acetylene peak at 3280 cm-1. Experimental conditions: 0.048 atm in 1 atm, 15 cm pathlength, and 300K. Calculated parameters: 0.048 atm in 1 atm, 15 cm, 300K, and Voigt lineshape. Figure 5.6-3 Experimental and calculated absorbance acetylene spectrum Flame conditions: 0 mm position, Total Flow = 2 slm and R=0.90 Calculated parameters: 0.007 atm cm in 1 atm 400K HITRAN 96; Voigt lineshape. Spectra shifted for clarity. Figure 5.6-4 Experimental minus Calculated absorbance spectra in the acetylene region for 0 mm position; from Figure 5.6-1. Figure 5.6-5 Comparing the acetylene bands at 300K with experimental. Calculated spectrum shifted for clarity. Calculation parameters: 0.005 atm in 1 atm, 1 cm pathlength, 300K, Voigt lineshape, HITRAN 92. Figure 5.6-6 Comparing Experimental Absorbance with calculated acetylene at 500K (0.005 atm cm in 1 atm Voigt lineshape HITRAN96). Calculated spectrum shifted for clarity. Figure 5.6-7 Comparing acetylene (400K) with experimental spectrum (0 mm position R=0.90) at the 730cm-1 acetylene peaks.

78

78

79

79

80

80 81

81

82

xii

Figure 5.6-8 Plot of concentrations as a function of distance for R=0.90 Ft=2 slm. Figure 5.6-9: Plot of concentration as a function of distance at R=0.94 Ft=2 slm. Figure 5.6-10: Plot of concentration versus distance for R=0.98. Figure 5.6-11: Plot of concentration versus distance for R=1.04. Figure 5.6-12: Plot of concentration versus distance for R=1.10. Figure 5.6-13: Plot of oxygen entrainment for different R ratios. Figure 5.6-14: Plot of calculated hydrogen for various R values and position. Figure 6.1-1: Single beam spectra of two different detectors (nonlinear and linearized). Taken at 50 scans at 1cm-1 resolution. Spectra are of the background. Figure 6.1-2: Absorbance spectra of the flame at total flow of 2 slm with R=0.90. Two different detectors are shown. Figure 6.1-3: Absorbance spectra of the standard and linearized detector in the CO2 region. It was taken through the flame with total flow rate of 2 slm and R=0.90 Figure 6.1-4: Two background spectrum (one with purge in arm and one without) shown here are taken with 2 scans at 1 cm-1 resolution. Figure 6.2-1: Emission spectrum of enclosed flame at Ft = 2 slm, 0 mm position, and R=0.90: Shown with a greybody calculated by (1-transmission spectra) * Blackbody at 450K. Figure 6.2-2: Transmission spectrum of the enclosed flame at Ft=2 slm, 0mm position, and R=0.90. Figure 6.2-3: Emission spectrum of the flame at Ft=2 slm, 6 mm position, and R=0.90. Figure 6.2-4: Transmission spectrum of the enclosed flame at 6 mm position and R=0.90.

82 83 83 84 84 85 85 99

99 100

100 101

101 102 102

xiii

Figure 6.3-1: Emission and transmission enclosed flame spectrum at Ft= 2 slm, 0 mm position, and R=0.90. Figure 6.3-2: E/T analysis of the enclosed flame for CO at Ft=2 slm, 0 mm position, and R=0.90. Shown above is the matching of the hot CO bands at 1200K. Figure 6.3-3: The enclosed flame at Ft=2 slm, 0 mm position, and R=0.90 showing that the hot CO bands indicates a temperature higher than 1200K. Simulated spectrum (top) does not match the experimental shape (bottom). Figure 6.3-4: The emission spectrum of the flame at Ft=2 slm, 0 mm position, and R=0.90 compared to calculated spectrum of 0.05 atm cm in 1 atm CO at 2500K using HITRAN96 Voigt. Figure 6.4-1: Enclosed flame at Ft=2 slm, 0 mm position, and R=0.90 showing the presence of hot CO2 bands. CO2 calculation parameters: 0.01 atm cm in 1 atm, 600K, HITRAN92, Lorentzian line shape. Figure 6.4-2: Enclosed flame at Ft=2 slm, 0 mm position, and R=0.90 showing the approximate concentration of CO2. Calculated CO2: 0.003 atm cm in 1 atm, 2000K, HITRAN92 Voigt lineshape. Figure 6.6-1: Experimental (Ft=2 slm, 0 mm position, R=0.90) and calculated OH (0.0032 atm cm in 1 atm at 3000K HITRAN96 Voigt) spectra. No OH is present in the flame at 0 mm position R=0.90. Figure 6.6-2: Enclosed flame at Ft=2 slm, 18 mm position ,and R=0.90 shows no observable OH peaks. Calculation parameters for OH: 0.0032 atm cm in 1 atm 3000K , HITRAN96 Voigt Figure 6.8-1: Methane at two different temperatures showing difference in side peaks. Calculation parameters: 0.01 atm cm in 1 atm, HITRAN92 Voigt at 500K and 800K. Figure 6.8-2: Enclosed flame at Ft= 2 slm, 0 mm position and R=0.90 in the methane region. Calculated parameters: 0.002 atm in 1atm 400K HITRAN92 Voigt Figure 6.9-1: Free enclosed flame: concentration versus position at R=0.90 Figure 6.9-2: Free enclosed flame: concentration versus position at R=0.94

103 103

104

104

105

105

106

106

107

107

108 108

xiv

Figure 6.9-3: Free enclosed flame: concentration versus position at R=0.98


Figure 6.9-4: Free enclosed flame: concentration versus position at R=1.04

109
109

Figure 6.9-5: Free enclosed flame: concentration versus position at R=1.10 Figure 6.9-6: Free enclosed flame: H2 calculated versus position Figure 6.9-7: Free enclosed flame: O2 entrainment calculation versus position

110 110 111

xv

1.0 INTRODUCTION 1.1 Diamond Films and Its Applications Diamond has many useful properties. It has a high hardness, high thermal conductivity, high electrical resistivity, high wear resistance, low coefficient of friction, is chemically inert, and is optically transparent from ultraviolet to far infrared. [1-4] With all these properties, diamond is a very attractive material for many applications. These applications include scratchproof coatings, cutting tool coating, and heat sinks in the electronic industry.

1.2 CVD Methods There are several methods of depositing diamond films by chemical vapor deposition including the use of a hot filament, microwave plasma, or oxyacetylene torch. In a hot filament reactor, methane and hydrogen are introduced near a hot filament heated to 2000-2500K. The purpose of the hot filament is to dissociate the gas mixture into radicals. This method is usually carried out at low pressures (about 20 Torr) and results in a low growth rate (approximately 0.1 to 1 m/hr) while attaining high quality diamond films. [1, 5] Microwave plasma is also used to deposit diamond films on various substrates. [1, 6, 7] Typically, a polished silicon substrate is placed in a cavity where the pressure is lowered to about 1 mTorr. Then a hydrogen plasma is ignited with the microwave power. After the plasma is ignited, the pressure is incrementally increased (to about 20 Torr) and methane and perhaps oxygen containing species are then added to the plasma for diamond deposition. [8, 9]

However, there has been interest to grow these diamond films at higher growth rates while minimizing the cost of the equipment. One method for diamond film synthesis that meets these criteria is the use of an oxyacetylene torch. Previous research has reported that at atmospheric pressure, the linear growth rate is approximately 100-200 m per hour. [10, 11] In the oxyacetylene torch method, mass flow controllers control the flow of oxygen and acetylene to the torch. The premixed gases then form the combustion flame at the nozzle. Figure 1.0-1 shows a schematic of the flame. There are three primary regions in an open flame: inner cone, feather, and the outer flame. In the inner cone, the gases are hot and not yet reacting significantly. In the region between the inner cone and feather, the gases react according to the overall reaction: C2H2 + O2 2 CO + H2 (1.2-1)

In the outer flame, secondary combustion produces carbon dioxide and water: 2 CO + H2 + 3/2 O2 2 CO2 + H2O (1.2-2)

The secondary combustion accounts for about 2/3 of the total heat released in the combustion of acetylene. A water-cooled substrate, usually molybdenum, is inserted inside the flame about one mm from the inner cone. [12] The temperature of the substrate is measured by a two-color pyrometer or a thermocouple. It has been determined that the oxygen to acetylene flow ratio (R) is one of the primary factors that determine the quality of the diamond films. The typical flow rate used is about 23 L/min total flow rate with an R ratio of about 0.93-0.99. [12]

Diamond films have been grown systematically in open and enclosed flames. [1216] In the case of enclosed flames, the secondary combustion in the outer flame disappears. Despite this large change in the chemistry (Equations 1.2-1 and 1.2-2), the only major effect of enclosing the flame is lowering the growth rate. [16]

1.3 The Purpose of This Work The purpose of this work is to better understand the chemistry of diamond CVD using the oxyacetylene torch system. The analysis technique is in situ Fourier transform infrared spectroscopy. This extends the work of Morrison et al. [17] in which they perform in situ FTIR measurements of the oxyacetylene flame using the emission and transmission method. The present work compares the tomographic reconstruction of FTIR measurements of a deposition flame. [18] A free flame without a substrate is analyzed in an open and enclosed atmosphere at various axial distances from the torch nozzle ranging from 0 mm to 18 mm. In addition, the R ratio is varied over a range of R = 0.90 to 1.10. The methodology combines the emission and transmission FTIR measurements to determine the species concentration and gas temperature as a function of R and position. Species under study include CO, CO2, H2O, OH, C2H2, and CH4. The HITRAN database with HITRAN-PC version 2.51 (developed by University of South Florida and distributed by Ontar Corporation) aids in characterizing these species.

Figure 1.0-1 Regions of the acetylene flame.

2.0 PRIOR RESEARCH 2.1 Open Flames Many authors have determined the range of deposition parameters for diamond growth, and Morrison and Glass [1] have reviewed much of the work prior to 1993. A brief summary of those works, as well as newer research, is presented here. Hirose and Amanuma [2] found that diamond growth can occur between R = 0.7 to 1.0. Amorphous carbon was formed when the R ratio was less than 0.7, and the etching of diamond occurred when R is greater than 1.0. The range of temperature for diamond deposition was found to be 370 1200 C. The best diamond films were grown when the R ratio was between 0.85 to 0.98. Wang et al. [3] found similar conditions for diamond growth using an R ratio between 0.89 and 0.99 with a temperature range of 650-1320C. On the contrary, Hanssen et al. [4] reported diamond growth under both fuel rich and oxygen rich conditions: R = 0.9 to 1.08. Schermer et al. [5] found that a supersaturation of 6 % (percentage of extra C2H2 flow compared to the neutral flame conditions) at a temperature of 1200 C to be optimum. The supersaturation is defined as: Supersaturaion = Fa Fan Fan (2.1-1)

where Fa = flow of acetylene and Fan = flow of acetylene of neutral flame. Solving for R (the ratio of oxygen flow to acetylene flow) the supersaturation is related to R via:

Rn 1 = Supersaturation R where Rn = R value at neutral flame condition. In the torch system used for this

(2.1-2)

experiment, the neutral flame occurs when R = 1.03. Therefore, a supersaturation of 6% yields an R value of 0.97. In addition, many authors have researched oxyacetylene CVD since 1993. Bang et al. [6] studied the effect of nozzle size and substrate temperature on the quality of diamond. At a fixed R = O2/C2H2 = 0.98, he found that the lower limit on temperature for any diamond deposition was 680 C. Two nozzle sizes were studied in these experiments. They found that the 0.939 mm diameter nozzle deposited better diamond films than the 1.067 mm diameter nozzle under identical growing conditions. Zeatoun and Morrison [7] found that high R ratio (0.97-0.99) along with either high (1000C) or low (800C) surface temperature, depending on the total flow rate, to be optimum. Marinkovic et al. [8] were able to deposit large diamond crystals, about 200 m in size, on a molybdenum substrate with the R ratio = 0.93 to 0.98, surface temperature of 700-765C, and a total flow rate of 3.8 to 5.2 L/min. The total deposition time was one to four hours. Zhu et al. [9] were able to grow high quality diamond at R = 0.96 and surface temperature of 750 C. Using Raman spectroscopy to analyze the film, their diamond has a FWHM of 4.5 cm-1, which is within the top quality range of synthetic CVD diamonds. Zhang et al. [10] calculated phase diagrams using thermodynamics for diamond growth in an oxyacetylene flames. They compared their results to many different authors experimental results and found that

the calculated phase diagrams correlate very nicely with the experimental results. In the diagram, it is predicted that diamond growth occurs when the R ratio is between 0.8 and 1.1 with the surface temperature between 1000-1250K. This is consistent with many experimental results. [2, 4, 11, 12]

2.2 Enclosed Flames There have been many studies on the effect of the atmosphere on the oxyacetylene flame under diamond growing conditions. Murakawa et al. [13] eliminated the outer flame by covering the flame and thus were able to increase the deposition area to 40 mm x 40 mm at a lower growth rate. Murakawa and Takeuchi [14] deposited diamond films on cemented carbide and compared the performance of the diamond films grown using an enclosed flame to that of an open flame. In addition, they found that the heat load decreased by a factor of 1/3 because the overall reaction in the enclosed flame is C2H2 + O2 instead of C2H2 + (5/2) O2 CO2 + H2O (1.2-2) CO + H2 (1.2-1)

The heat of reaction in (1.2-1) accounts for 1/3 of the heat of reaction in (1.2-2). In addition to the lower heat load, they found that the deposition area increased by 50% and the size of individual grains decreased. Doverspike et al. [15] shrouded the flame with a flow of argon to eliminate the outer flame. These diamond films had smooth (100) faces with more amorphous

carbon. Morrison et al [16] deposited diamond films with the enclosed flame on silicon. They found that the growth rate was approximately ten times slower than that of the open flame. This may have been due to an increased in deposition area, decreasing the linear growth rate. The deposition area, however, was about 2.5 times higher than in the open flame. They also found that the Raman spectra of the enclosed flame to be similar to that of cycling the R ratio. [17, 18] They found the best diamond growth in high temperature (900 C) and high R ratio (0.96-0.97).

2.3 Diagnostics and Flames 2.3.1 Optical Emission Spectroscopy Matsui et al. [19] found intense emission of the excited states CH* and C2* localized in the feather boundary. The OH* emission is localized in the intermediate zone, right outside of the feather. In other work, Yalamanchi et al. [20] found C*, C2*, CH*, molecular H2*, atomic H*, and atomic O* in the feather region. They also observed C2* and CH* in the intermediate zone, just outside of the feather. However, lower values of CH* were found in the intermediate zone compared to the primary combustion zone. In addition, when the torch was operated under fuel rich conditions, excess H*, OH*, and O* were also observed outside of the feather. The flame temperature reported here was 3237K. 2.3.2 Laser induced fluorescence Matsui et al. [21] used mass spectrometric techniques to find that the main gaseous species in the feather region were CO and H2. They then used LIF to analyze for C2

and OH. They found that the concentration of C2 decreases with radial distance at a fixed distance from the burner. They also observed that the concentration of OH peaks at a distance away from the center of the substrate. For the R ratio above 1.05, the oxygen containing radicals, O and OH are so numerous that part of the CO is oxidized to CO2, eliminating the feather region. For an R ratio of less than 0.96, carbon containing radicals dominate, and increased rapidly with decreasing R. They found that carbon radicals C2H, C3, C, C2, and CHx are the next highest in concentration. Using mass spectroscopy, other carbon radicals were found at mole fractions of 10-6 to 10-11: CH4, C2H3, C4, C5, C2H4, CH2O, C3O2, C2H5, CH4O, C2H6O, C2H6, C3H7, C3H8, C4H2. They found that the carbon containing species decreased linearly toward the feather tip because of inter-diffusion and reactions with oxygen radicals in the intermediate zones. These observations were consistent with the work of Snail et al. [11] where they found C2* and CH* in the flame front and feather using optical emission spectroscopy. Klein-Douwel et al. [22] used LIF to measure the distribution of C2, CH and OH during diamond deposition. They found ground and excited state of C2 and CH in the feather region of the flame, with little or no OH. The C2 concentration was higher under conditions that favor high quality diamond growth. The entire acetylene feather was filled with C2, and it decreased from the flame front to the substrate 1 mm away. Therefore, they concluded that C2 is an important precursor for diamond growth. OH was observed outside of the feather and was independent of combustion conditions. This finding contradicts that of Hirose et al. [2] The authors did not report on the

10

approximate concentrations of the species detected. They measured only the relative amounts of each species. Cappelli et al. [23] performed planar induced fluorescence on the flame. They also found similar trends, with C2H and CH in the feather region. 2.3.3 Absorption spectroscopy Welter et al. [24] determined column densities of CH, C2, CN, and OH radicals by absorption spectroscopy. They found that CH radicals were primarily produced in the inner cone and were gradually extinguished in the feather. The concentration of C2 mimicked that of the CH radicals. They also observed that OH was produced primarily in the outer flame. There was also detectable amount of CN, which was produced by atmospheric nitrogen diffusing into the flame and reacting with the radicals. It was observed that CN has higher column density when the supersaturation ratio is higher and closer to the substrate. When the supersaturation ratio is zero, the CN is below detectable limits. 2.3.4 Mass spectroscopy Matsui et al [21] used mass spectroscopy to detect H, O, C2, H2, and CH in the feather region of the flame. They found that the main species were CO and H2 in the feather and intermediate zones. CO2, N2, and C2O were detected outside of the feather. Marks et al. [25] found H2, CO, N2, and C2H2 in both laminar and turbulent oxyacetylene flames. In a turbulent flame, O2, CO2, and H2O were also detected.

11

They found that the concentration of the species did not vary much in the laminar flame from 0 to 4 mm away from the flame cone. Wolden et al. [26] performed in situ mass spectrometry on a flat flame 11mm away from the substrate at low pressure. They found six species that formed the majority of the stable gases in the system: CO, H2, Ar, H2O, C2H2, CO2, and CH4. Their results were similar to that of Marks et al. [25] since the concentration of the species did not vary much within the first 4 mm. However, the concentrations of the species change drastically beyond 4 mm away from the torch. In the region 11 mm away from the torch, the authors found that as the R ratio increased, that atomic hydrogen concentration was nearly constant while methyl radicals varied by about 15%. They also found that the acetylene concentration decreased 70% in response to a 10% change in the R ratio, from 0.95 to 1.05. In addition, they observed uniform increase in oxidation products, CO2 and H2O, as the R ratio was increased. There was also a 50% increase in hydroxyl radicals and 70% increase in atomic oxygen. However, their absolute concentration was only about 40 and 5 ppm respectively.

2.4 Fourier Transform Infrared Spectroscopy (FTIR) - Background In this research, Fourier Transform Infrared Spectroscopy (FTIR) is the method used to perform gas phase analyses of the diamond growing oxyacetylene flame. The spectrometer used to obtain the infrared spectrum consists of an infrared source, a beamsplitter, one fixed mirror, one movable mirror, and a detector. These components make up the Michelson interferometer. (See Figure 2.4-1) The optical

12

path is as follows: the infrared light originates from a source and strikes the beamsplitter, which is designed to transmit half of the radiation and reflect half of it. Then light transmitted by the beamsplitter strikes the fixed mirror, and the light reflected by the beamsplitter strikes the movable mirror. After reflecting off their respective mirrors, the light then recombines in the beamsplitter and leaves the interferometer to interact with the sample. After interacting with the sample, the light strikes the detector where the IR signals are converted to a voltage which is then digitized and processed by the computer. If the moving mirror and the fixed mirror are position such that they are exactly the same distance from the beamsplitter, the distance traveled by the two beams of light are the same. This condition is known as zero path difference (ZPD). When the movable mirror is moved from the ZPD position to a new position denoted by , an optical path difference () is introduced. Since the light has to travel to the mirror and back, the relationship between the mirror position () and the optical path difference is =2 (2.4-1)

When the position of the movable mirror is at ZPD, the light recombining at the beamsplitter are in phase, causing constructive interference. All wavelengths constructively interfere at the ZPD forming the centerburst observed in the interferogram. Constructive interference also takes place if the optical path difference () is exactly the same as the wavelength () of the light. In equation form, constructive interference occurs when

13

=n where n is any integer. Destructive interference occurs when the optical path difference is exactly half of the wavelength: = (n + )

(2.4-2)

(2.4-3)

When the optical path difference is different that any of the above, a combination of constructive and destructive interference occurs. A plot of the light intensity versus is called an interferogram. Since the mirror is moving with a constant velocity, the light going through the interferometer is time modulated, and the modulation frequency (Fv) for a given wavenumber is given by Fv = 2 V where V = moving mirror velocity, and = wavenumber of the light in the interferometer. Since each wavenumber causes a constructive and destructive interference at different optical path difference, light at different wavenumbers are then modulated at different frequencies. This is how the spectrometer distinguishes signals coming from different wavenumbers. Each different wavenumber of light gives rise to a sinusoidal wave of unique frequency measured by the detector. Since the infrared source is emitting a broadband of infrared light, information on all wavenumbers can be obtained in a single scan. Since the interferogram is the sum of sinusoidal waves, it can then be Fourier transformed to obtain the single beam spectrum. (2.4-4)

14

By performing multiple scans, the signal to noise ratio of the interferogram can be significantly improved. The reason is that the signal is added to every scan and the noise fluctuates randomly. Therefore, by signal averaging, the relative contribution of the noise to the spectrum are reduced. The signal to noise ratio is proportional to the square root of the number of scans. Figure 2.4-2 shows the light path in emission mode. The light from the flame travels in reverse direction back through the beamsplitter and into the emission detector located in the emission port of the spectrometer. During the emission mode, the globar is turned off so that only the IR from the flame are modulated by the moving mirror. This modulation are detected by the emission detector and converted to a spectrum the same way as the transmission detector.

2.5 FTIR Spectroscopy and CVD 2.5.1 Emission/transmission measurements Morrison and Haigis [27] were one of the first to perform emission and transmission IR measurements on a CVD system, in this case, plasma deposition of amorphous hydrogenated silicon. The authors used the relationship between the emission and transmission spectra to determine gas temperature in the plasma:

R ( ) R 0 ( ) = [1 ( , T g )] R bb ( , T g )

(2.5-1)

where R() = path corrected radiance spectrum in W/(m2 sr cm-1), = wavenumber in cm-1, Tg = temperature of the gas in K, Ro = background radiance from room

15

temperature surroundings in W/m2 sr cm-1, = transmission spectrum, and Rbb = Plancks blackbody radiation. This relationship assumes that the transmission detector is cooled to 77K, making the radiation from the detector negligible. The transmission spectrum ((,Tg)) and the path corrected emission spectrum (R()) are the measured parameters. Rbb(,Tg) is Plancks blackbody function given by: Rbb ( , T g ) = e C1 3
C 2 Tg

(2.5-2)

C1=1.191 x 10-8 W/m2 sr (cm-1)4 and C2 = 1.439 K cm and Rbb(,Tg) is in W/m2 sr cm-1. The temperature of the gas for the Plancks function is the temperature of the blackbody at which Equation 2.5-1 is satisfied. Given the high temperature of the flame, the emission at room temperature, Ro, is generally taken to be negligible. This method was also used by Morrison et al. [28] to study the aerosol synthesis of TiO2 particles. In this work, the authors used FTIR as an in situ diagnostic tool in the oxidation of TiCl4 in a premixed methane-oxygen flame. Transmission FTIR yielded quantitative information on HCl, CO2, and H2O by comparing experimental spectra to spectra calculated from spectroscopic constants found in the HITRAN database. From the emission and transmission spectra, the authors were also able to obtain temperature, composition, and particle characteristics in the electrically modified flames. This same technique are used in the present analysis of the oxyacetylene flame.

16

2.5.2 FTIR and oxyacetylene flames Preliminary measurements by Morrison et al [29] have shown that a combination of emission and transmission measurements yields species concentration (CO2, H2O, CO, and OH) and gas temperature. In later measurements, Morrison et al. [30] reported in situ measurements using the emission and transmission method together with tomography. They found that the feather of the flame was approximately 3000K and the temperature decreased slightly in the outer flame. In the radial direction, they observed that the concentration of CO decreases linearly from the inner cone to the outer cone, with a high concentration in the feather region. CO2 , H2O, and OH exist primarily in the outer flame. The authors did not observe any radical species using the FTIR, and the CO2, H2O, CO, and OH measurements were not quantitative.

2.6 HITRAN Database The HITRAN database was used to calculate high-resolution optical transmission spectra of the individual gases. The HITRAN database is a catalogue of spectroscopic constants of various gases found in the atmosphere including CO2, CO, H2O, OH, C2H2 , and CH4. A personal computer software package (HITRAN-PC ver. 2.51, Ontar Corporation) was developed by the University of South Florida to calculate gas spectra at a given partial pressure, total pressure, and temperature. A description of the procedures using the HITRAN database can be found in Morrison et al. [31]. The only change in methodology is the use of the 1996 HITRAN database and HITEMP

17

database in addition to the 1992 database. The 1996 database is an update of the 1992 database plus additional gases. The HITEMP database includes additional hot bands and rotational levels for CO2, CO, and H2O appropriate to at least 1000K. However, it was determined that the catalogued CO2 bands in HITEMP are incomplete at temperatures above 1500K. In addition, the HITRAN-PC software cannot handle the full set of the spectroscopic constants in the CO database. Therefore, H2O and sometimes CO (with a some spectral lines omitted) are calculated using the HITEMP database while the other gases are calculated using either the 1992 or 1996 database. Using HITRAN-PC, a spectrum of a gas can be calculated from spectroscopic constants in one of the HITRAN databases. The high-resolution spectra (typically better than 0.01 cm-1) can then be smoothed to a spectra with the same resolution as those obtained from the FTIR (typically 1cm-1). The method works as follows [31]: An inverse Fourier Transform is applied to the calculated HITRAN spectrum to obtain an interferogram. The interferogram is then truncated and apodized to the same resolution found in the FTIR. These steps mimic the processes that go on inside the FTIR. The modified spectrum is then Fourier transform back to obtain the 1 cm-1 spectrum. The spectra obtained can then be compared to the experimental spectra to determine the temperature and quantity of gases present in the flame. Morrison et al. [31] have verified the accuracy of the procedures outlined above using known quantities of gas. Morrison et al. have also successfully applied the method to flame measurements. [28] A sample of a spectra before and after conversion is shown in Figure 2.6-1 and 2.6-2. A copy of the smoothing program is in Appendix I.

18

In calculating the spectrum for gases in the flame, the wavenumber range was chosen such that it covers the region of interest and is consistent with the point to point resolution of the FTIR. In addition, the number of points must be a power of two. There are three different line shape calculations depending on the conditions and specie: Doppler, Voigt, and Lorentzian. The choice of line shape depends on the half width parameter at 296K, the temperature coefficient, the temperature, the total pressure, the molecular weight, and line position. If the ratio of the Lorentzian line width to the Doppler line width (Lorentz/Doppler) is less than 0.1, then the Gaussian line shape is appropriate. [31] If Lorentz/Doppler is between 0.1 and 5, then the Voigt calculation is appropriate. If the parameter Lorentz/Doppler is greater than 5, then the Lorentzian calculation is appropriate. Under most condition found in C2H2 / O2 flames, the line shape is typically Voigt. The Lorentz/Doppler parameter is defined as: L 296 n 1 = ( g( )) P D T 760(U ) (2.6-1)

where L/D = Lorentz/Doppler, g = half width parameter at 296 in (cm-1/atm), T = temperature of interest in (K), n= temperature coefficient, P= total pressure in (Torr), and U = Doppler half width half maximum T M

U = (3.58 x10 7 )( v o )

(2.6-2)

where o = line position in (cm-1), and M = molecular weight in (g / gmol).

19

The typical parameters used in HITRAN-PC are presented in Table 2.6-1.


Parameters Value

O2 lines Omitted Temperature dependence of atmospheric density,N Ideal Gas Path length (m) 0.01 Index of refraction 1.0000 Spectral calc margin (cm-1) 1 Number of saturated halfwidths 12 Extreme limit for line calculations (cm-1) 25 Optical depth threshold 1.00E-06 Use only air broaden line width from HITRAN Transmission graphs No continuum calculations.

Table 2.6-1 Table of parameters used to calculate spectra using HITRAN-PC and HITRAN database. The major species in the flame are CO, CO2, and H2O. The minor species are C2H2, CH4, and OH. Different calculations were done at different concentrations and different spectral windows. The path length was fixed at 0.01 m for all calculations with a total pressure of 1 atm. The temperature used for calculation was determined by one of two ways. In the first case, the emission and transmission method described in Section 2.5 gives the temperature of the species. However, if the temperature does not properly match the emission spectrum of the flame, we can also determine the temperature by matching the shape of the rovibrational band of the molecule. This is more clearly explained in later sections. The spectra calculated and used in this work are summarized in Table 2.6-2.

20

Table 2.6-2 Table of spectra calculated in this work. Note that some spectra presented in this work are scaled from calculated ones in order to obtain the necessary concentration. Filename
Partial T (K) Lineshape Actual Number Ending Pressure starting of points wavenumber (atm) wavenumber (cm-1) (cm-1) 0.0550 0.0130 0.1100 0.0300 0.0047 0.0047 0.0350 0.0300 0.0300 0.0060 0.0120 0.0070 0.0070 0.0050 0.0050 0.0480 0.0480 0.0500 0.0500 0.0100 0.0025 0.0065 0.0100 0.0100 0.0020 2400 Voigt 2500 Voigt 2700 Voigt 3000 Voigt 296 Voigt 296 Voigt 1900 Voigt 3000 Voigt 3000 Voigt 3000 Voigt 500 Voigt 400 Voigt 400 Voigt 500 Voigt 300 Voigt 300 Voigt 300 Voigt 1200 Voigt 2500 Voigt 600 Lorentzian 2000 Voigt 3000 Voigt 500 Voigt 800 Voigt 400 Voigt 3100.3441 2000.0354 1850.0809 2600.3353 3100.3441 3594.0847 2600.3353 3100.3441 3594.0847 2700.1442 3150.0075 610.4254 3150.0075 3150.0075 3150.0075 640.3199 3150.0075 1850.0809 1850.0809 2000.0354 2000.0354 2400.2353 2800.4352 2800.4352 2800.4352 262144 131072 65536 262144 262144 262144 262144 131072 262144 262144 32768 32768 16384 32768 32768 32768 32768 65536 65536 65536 131072 262144 65536 65536 65536 3594.0829 2493.7722 2343.8140 3094.0740 3594.0829 4087.8235 3094.0740 3594.0810 4581.5623 4675.0992 3396.8703 857.2882 3273.4351 3396.8703 3396.8703 887.1827 3396.8703 2343.8140 2343.8140 2493.7685 2493.7722 4375.1903 3294.1683 3294.1683 3294.1683 Database

29h2o 59co2 8CO 41h2o 34h2o 36h2o 38h2o 40h2o 42h2o 11oh 11c2h2 27c2h2 28c2h2 33c2h2 35c2h2 42c2h2 43c2h2 22CO 24CO 60co2 62co2 14oh 7ch4 8ch4 12ch4

HITEMP HITRAN92 HITRAN 96 HITEMP HITRAN96 HITRAN96 HITEMP HITEMP HITEMP HITRAN 96 HITRAN92 HITRAN96 HITRAN96 HITRAN92 HITRAN92 HITRAN92 HITRAN92 HITRAN 96 HITRAN 96 HITRAN92 HITRAN92 HITRAN 96 HITRAN92 HITRAN92 HITRAN92

21

Figure 2.4-1: Components of the Michelson interferometer under transmission mode. (optical path shown)

Figure 2.4-2: Optical path in the interferometer under emission mode.

22

Figure 2.6-1: HITEMP calculation of the water bands at 2500K, 0.055 atm cm in 1 atm and Voigt line shape

Figure 2.6-2: Smoothed spectrum after applying conv65.ab smoothing algorithm.

23

3.0 EQUIPMENT DESCRIPTION 3.1 Torch reactor Figure 3.1-1 shows the oxyacetylene torch reactor. [1] The system is composed of a cutting torch made by Victor with a size 0 tip positioned on an xyz translator. The 0 tip has a 1-mm diameter hole and was straightened slightly so that the inverted torch can be used with the FTIR extension arm. The torch was pointing upward to minimize the flicker noise in the IR spectra. (See Chapter 4 for a detailed analysis.) The flow rate to the torch was 2L/min total. MKS Instruments Inc. makes the mass flow controllers used to control the flow rates of acetylene and oxygen. Both controllers are types 1259C-10000RV that have a range of 0-10 slm of N2. The mass flow controllers was linked to a computer using Labview software (National Instruments). The software controlled the R ratio while depositing diamond films. A telescope made by Specell Model M820-S with a 1/20-mm scale was used to measure the distance between the inner cone and the substrate in addition to measuring the width of the flame.

3.2 Spectrometer The FTIR spectrometer used for this experiment is a Bomem model MB157 with a maximum resolution of 1cm-1 (See Figure 3.0-1 and 3.0-2). The range of sensitivity is 500 cm-1 to 6500 cm-1. The spectrometer has an internal globar that is similar to a blackbody at 1300K. A 2 mm aperture was placed in front of the globar to reduce the spot size and gain spatial resolution. The globar is capable of obtaining power from

24

an external power supply in the event that a hotter globar temperature is desired. An emission port is available on the spectrometer for placing an MCT detector to collect emission spectra. Two Mercury-Cadmium-Telluride (MCT) detectors, with a mid and near IR range and cooled by liquid nitrogen, are used. All spectra are averages of fifty scans at 15 scans per minute. An extension arm was placed over to spectrometer to direct the beam to the flame. Figure 3.2-1 presents a schematic drawing of the spectrometer and the arm. The beam from the globar travels through the aperture and a 3.5 focal length mirror inside the spectrometer. This parabolic mirror collimates the beam into the spectrometer. Then the collimated IR beam from the spectrometer reflects from two flat gold plated mirrors to divert and extend the beam. The beam is then sent to a parabolic mirror with a focal length of 3.5 inches. The flame was placed at the focal point at the 3.5 focal length mirror. Since the globar is behind a 2-mm aperture (which acts as the limiting aperture), the globar can be viewed as a 2-mm diameter source. The first parabolic mirror has a focal length of 3.5 inches and the parabolic mirror before the sample also has a 3.5 inch focal length resulting in a magnification of one. Therefore, the spot size is approximately the size of the internal globar, 2 mm in diameter. However, due to the imperfections of the optics, the measured size was 3 mm in diameter. Since flickering interferes with the FTIR measurements, a longwave bandpass filter (OCLI) was used for some experiments (see Chapter 4). Since most of the flickering occurs in the CO and CO2 bands (2100-2400 cm-1), the filter blocks

25

wavenumbers above 2100 cm-1. This filter was placed at the entrance of the transmission MCT detector.

3.3 Enclosed Flame Apparatus The drawing of the enclosing apparatus is shown in Figure 3.3-1 and 3.3-2. Because the IR transparent KBr windows used in the apparatus cannot withstand the high temperature (>100C), the windows were placed away from the flame. The torch entered the apparatus through the bottom opening. The bottom of the apparatus has two inlets for argon purge. The Ar flow of 6 slm purges the combustible gases from inside the apparatus. The hot gases flow out of the apparatus through an opening on top of the apparatus. The exit area (about 1.5 square inches) must be small enough to prevent oxygen from backflowing into the apparatus. Since the outer flame indicates the presence of entrained air, the apparatus was tested to see if the outer flame was eliminated. Figure 3.3-3 shows the spectrum of the flame at the water band with and without the enclosure. The enclosure was quite effective in shielding the flame.

3.4 Operational Procedure Before starting any experimental work, the MCT detectors were cooled by liquid nitrogen for 15 minutes. In the meantime, the IR source to the FTIR was turned on and the transmission detector was aligned, using the alignment feature on the Bomem/Grams 386 software. The location of the spot (focal point) was determined

26

by moving the torch on the xyz translator. If the amount of signal the detector is detecting decreased, then the torch nozzle was in the way of the beam. Once the spot was found, the globar was turned off and a blackbody was placed there to align the emission detector. The mirrors on the emission detector were adjusted until a maximum signal was achieved. Once both detectors were aligned, a spectrum of a blackbody at 700 C [2] was obtained in order to determine the path correction for the emission measurements. We calculate the path correction by dividing the spectrum obtained with the blackbody by the Plancks blackbody distribution at the measured temperature of the blackbody. [3] Once the path correction was obtained, the background spectrum was taken for the transmission detector. The path correction spectrum and the background spectrum were taken at the same resolution and number of scans as in the experiment: 1 cm-1 resolution and 50 scans at 15 scans per minute. Once the reference spectra were taken, the mass flow controllers were set for torch ignition. The enclosed flame apparatus was placed on the torch, with the torch pointing up. Then the torch is ignited. A top was placed on the apparatus to enclose the flame, sealed by clips around the edges. The telescope was used to measure the length of the inner cone. As the flame burns, the emission and transmission spectra were taken at various axial positions. In the case of the free flame (no substrate), the axial positions were 0, 3, 6, 12, and 18mm from the torch nozzle. The R ratio was also varied at each position (0.90 1.10) Once all data had been collected, the torch was moved away

27

from the substrate with the help of the xyz translator. The apparatus was then vented by removing the top. Then the flow of oxygen to the torch was increased to about 4 L/min to blow out the flame. The ranges of conditions are shown in Table 3.4-1. The transmission and emission IR spectra were divided by the background and path correction spectrum respectively. The resultant spectrum was used for data analysis in emission/transmission method and for fitting it with spectra calculated from HITRAN. Table 3.4-1 Table of range of conditions. Total flow rate is 2 L/min.
R value 0.90 0.90 0.90 0.90 0.90 0.94 0.94 0.94 0.94 0.94 0.98 0.98 0.98 0.98 0.98 1.04 1.04 1.04 1.04 1.04 1.10 1.10 1.10 1.10 1.10 Distance from torch tip (mm) 0 3 6 12 18 0 3 6 12 18 0 3 6 12 18 0 3 6 12 18 0 3 6 12 18

28

Figure 3.1-1: Setup of the torch reactor.

Figure 3.2-1: Bomem spectrometer with the extension arm and optics.

29

Figure 3.3-1: Side view of enclosed flame apparatus

Figure 3.3-2: Top and bottom view of the enclosed flame apparatus

30

Figure 3.3-3: Water emission from the flame at 18 mm position and R=0.90 with and without the enclosed flame apparatus.

31

4.0 METHODS TO ELIMINATE FLAME NOISE 4.1 Problem Statement The FTIR expects to detect just the modulated frequencies produced by the interferometer (Chapter Two). For the spectrometer used for this experiment, the Fourier frequency range lies between 105 1400 Hz (corresponding to 500 cm-1 to 6500 cm-1). However, the hot oxyacetylene flame also emits fluctuating IR radiation inside the range 0-500 Hz that interferes with the FTIR measurements. Therefore, a high level of noise is introduced to the infrared spectrum. Researchers postulate that the source of modulation (flicker) is a combination of turbulence, combustion instabilities, and buoyancy of the hot gases. Flicker may also result from the combustion reactions occurring simultaneously at the exit of the torch. As more reactions occur simultaneously, the amount of heat produce and the concentrations of the species vary. Since the rates of reactions are determined by the concentration, the reaction rates vary also. Therefore, the system never quite reaches equilibrium. Elfeky et al. [1] studied turbulent flames and suggested that the flames become noisy because of combustion instabilities. They suggested that the noise is associated with irregularities of the flame front and speed. The flame flicker is then the fluctuation of the amount of emitting species, pressure, or other characteristics of the flame front due to flow pulsation. Even though the flames used in our experiments were mainly laminar [2], these radiation fluctuations can still be present due to the combustion reactions occurring in the inner cone of the flame. It was observed that as the equivalence ratio increases (R decreasing), the amplitude of oscillation increased. [1]

32

They observed that lean limit of flammability was governed by low frequency oscillations (less than 30 Hz) and fuel rich conditions were governed frequencies greater than 30 Hz (30Hz corresponds to about 140 cm-1 on the Bomem MB151 spectrometer). Since it is necessary to be slightly fuel rich when depositing diamond, it is expected that higher frequencies will be observed in the oxyacetylene flame. Due to the fluctuating IR from the flame, the transmission interferograms taken through the flame have extra noise. Figure 4.1-1 shows an interferogram of a background spectrum (no flame present). The centerburst is clearly seen in the middle, at about 3900 x-units with the wings of the interferogram quickly falling to some very low value. In contrast, an interferogram taken through the flame is shown in Figure 4.1-2. Note that the centerburst is very hard to see due to the noise of the flame. Furthermore, the flicker intensity is the same magnitude as the centerburst. The frequency of the noise in the flame can be determined by measuring the flicker with the globar off and converting the interferogram to a power spectrum. Since each wavenumber has a distinct modulation frequency for a particular spectrometer, the detector detects these frequencies which is then Fourier transformed to intensity vs. frequency. (Since the frequency is proportional to wavenumber (Equation 2.5-4), intensity vs wavenumber is directly related to intensity vs. frequency). The power spectrum is nothing more than a complex Fourier transform of the noise interferogram, yielding a plot of amplitude vs. frequency. We then plot amplitude vs frequency as amplitude vs. wavenumbers to identify where the noise peaks appear in a flame spectrum. Any noise signal that shows up below 500 cm-1 is

33

clearly noise since the detector cannot detect below 500 cm-1. Therefore, a peak that appeared at 300 cm-1 is from the torch emitting IR modulated at a low frequency that corresponded to 300 cm-1 for the MB157 spectrometer. The modulation frequency of the flame can then be calculated knowing the conversion between frequency and wavenumber for the particular spectrometer. The spectrometer used in this experiment has the 7225 cm-1 modulated at 1500 Hz. This was calculated by putting a chopper in between a blackbody and a transmission detector. The chopper was modulating the blackbody signal at 1500 Hz and a peak at 7225 cm-1 was observed in the spectrum. The power spectrum of the background (globar on) in Figure 4.1-3 shows that there is very little amplitude below 500 cm-1. The power spectrum of a single scan through the flame with the globar on is shown in Figures 4.1-4 (low wavenumber) and 4.1-5 (full spectrum). The two figures show that there are many noise spikes at low wavenumbers (frequencies), and the spikes can be more than one hundred times the globar intensity. This means that the flame was heavily modulated at low frequencies, interfering with the low wavenumber FTIR measurements. In addition, the flicker raised the noise generally throughout the entire spectrum. If the modulated IR signal emitting from the flame was purely random noise, it should signal average away with multiple scans. The power spectrum of the same flame with 75 scans (5 minutes) shows that the frequency had been reduced significantly (See Figure 4.1-6), but the low wavenumber region is still contaminated with the flicker noise. The reason for this arises because the signal to noise ratio is proportional to the square

34

root of the number of scans, and the flicker noise level is large compared to the FTIR signal. If the noise is about 100 times the FTIR signal, it would take 10,000 scans to reduce the noise to the intensity level of the globar. At one cm-1 resolution, this would take about 11 hours. In summary, the low wavenumber region is most effected by the flame flickering. This is due to the flame emission modulating at low frequencies, which lie within the range of the modulated frequency for the FTIR used for this experiment. Since a lot of useful information is in the low wavenumber regions, we have investigated several ways to minimize the flicker noise

4.2 Methods to Eliminate Flame Noise Several techniques have been investigated to reduce the flicker noise. These include using a two detector configuration, pointing the flame upward, using apertures, blocking the CO and CO2 absorption, and using a long-pass infrared filter. 4.2.1 Two detector configuration The idea here was to use two detectors for data acquisition, one to detect noise and the other to detect signal plus noise. If both detectors saw the same flicker noise, then subtracting the two interferograms should result in a flicker free interferogram. To test that idea, two detectors were used to collect noise from the flame at positions 90 degrees from each other. (Figure 4.2-1) The same FTIR powered both detectors, and an A/D converter (Model AT-AO-10 with the software Labview 4.0 by National Instruments) collected voltage data from the two DTGS detectors. Prior to

35

measurements on the flame, the two detector system was tested using a light bulb as a modulated IR source. The two spectra collected from the two detectors overlapped. When used to measure the flame, however, the two flickering intererograms collected from the two detectors do not exactly match. (Figures 4.2-2 and 4.2-3) If the noises were identical, then a spectra subtraction of the two noise spectra should yield zero. However, that is not the case as shown in Figure 4.2-4. In summary, the noise in the flame is not symmetric since the noise at 90 from each other was different. In addition, the noise generated by the flame is very random, so the two noise interferograms detected at different times were also different. Therefore, the double detector configuration will not work. 4.2.2 Pointing the flame upward When pointing the flame upward, we observed that the noise level decreased significantly. As hot gases exit the nozzle, buoyant forces give the gases a tendency to travel upward. If the torch tip was pointing downward, the hot gases exiting the nozzle flow downward while buoyancy drives them upward. This causes disturbance in the gases resulting in oscillations. Figure 4.2-5 shows the single scan of the flame pointing downward. The signal to noise ratio (SNR) for this interferogram is about 6.8. In contrast, the single scan of the flame pointing upward is shown in Figure 4.26. The SNR for this interferogram is about 76, which is about 11 times better than the flame pointing down. The power spectra for the flickering interferograms (globar off) are shown in Figure 4.2-7 and 4.2-8 respectively. The amplitude of the noise is generally less with the torch pointing up than down. The dominant frequencies are

36

summarized in Table 4.2-1. Regular and bold fonts indicates the flame pointing down and up respectively.
Major peaks from the power spectrum Downward Upward Flow Flow frequency (Hz) frequency (Hz) 26.99 35.29 47.54 53.36 53.98 60.21 80.55 82.63 95.09 107.96 130.38 134.95 142.42 154.67

wavenumber (cm-1) 130 170 229 257 260 290 388 398 458 520 628 650 686 745

Table 4.2-1: Major frequencies of the flame. To quantify the flickering reduction, the power spectrum was integrated to calculate the area under the curve. The area under the power spectrum for the flame pointing up is about 6 times less than the flame pointing down. To verify further the noise reduction in pointing the flame upward, two power spectra are compared in the 20003000 cm-1 region. In Figure 4.2-9, a power spectrum of the flame pointing down with the globar on is shown. The noise level in the spectrum is about 2.0 x 10-5 units high. With the flame pointing up, shown in Figure 4.2-10, the noise level is about 1.0 x 10-5 units high, about 2 times less.

37

Another scenario to examine related to buoyancy driven flicker is the amount of flicker noise introduced by placing the substrate in the flame. The power spectrum of a single scan flame pointing up without the substrate is shown in Figure 4.2-11 and 4.2-13. The power spectrum of a single scan flame with the substrate is shown in Figures 4.2-12 and 4.2-14. The spectra without the substrate have lower amplitude noise in the low frequency region (ie. low wavenumbers) compared to the one with the substrate. However, at higher frequencies, the one with the substrate is slightly better. This means that the substrate is causing more low frequency flicker and less high frequency flicker. In summary, it is more advantageous to take FTIR spectra of the flame pointing up. The amplitude of the flicker at all frequencies is generally lower with the flame pointing up than it is down. In addition, when the substrate is placed over the flame to deposit diamond, the low frequency flickering was increased while the higher frequency noises were decreased. 4.2.3 Apertures Another method to reduce the flicker noise was to place an aperture between the flame and the transmission detector. The objective is to screen out flicker that may be collected by the detector optics with as little impact as possible on the FTIR signal. To study the effect of the SNR with apertures, apertures of different sizes were placed in the path of the IR between the flame and the transmission detector with the flame pointing up. The aperture was adjusted to see which size would yield the best SNR. The signal is taken to be the height of the centerburst in the interferogram and the

38

noise is taken to be the average height of the wings of the interferogram. The wings of the true interferogram should be small compared to the centerburst. However, the flame causes the wings to have a high intensity. For a typical single scan background spectrum, the ratio of the height of the centerburst to the average height of the noise (SNR) is about 450.
Signal / Noise Ratio for Various Aperture Diameters Aperture Size(mm diameter) 3 4 6 7 8 10 11 none 1 scan 4.3 9.6 11.1 11.8 11.3 12.9 13.8 13.8 10 scans 14.2 22.0 39.0 33.3 36.4 40.3 37.6 37.7

Table 4.2-2 Signal to noise ratio for various sizes of apertures.

As shown from Table 4.2-2, there is not a significant improvement in the SNR by using an aperture. Once the aperture was about 6 mm in diameter, there was no further increase in the SNR. The fluctuations observed in the table are due to estimates in calculating the SNR. Therefore, the use of apertures between the flame and the transmission detector in this experiment has no significant advantage. However, the use of an aperture between the flame and the FTIR can aid in improving spatial resolution. 4.2.4 Blocking the CO and CO2 absorptions In the oxyacetylene flame, the biggest peaks in the IR spectrum are the CO and CO2 bands. Thus, emission from the flicker in the hot CO and CO2 in the flame are

39

largely responsible for the modulated signal. Therefore, if a bag of CO or CO2 is placed between the flame and the detector, the bag absorbes most of the CO and CO2 emission from the flame, thereby reducing the noise level to the rest of the spectrum. However, the bag needs to be optically thick, so the major disadvantage of this is that all the CO and CO2 information is lost. For comparison, Figure 4.2-15 shows the power spectrum of the flicker noise from the flame going through an empty polyethylene bag. The power spectrum of the same bag filled with CO is shown in Figure 4.2-16. As shown in the figures, the CO in the bag was not very efficient in minimizing the noise in the power spectrum. The spectra here were taken with the DTGS detector instead of the MCT detector. Notice that the flickering frequencies are much higher than the ones with MCT (Figure 4.211) because the DTGS detector is much slower than the MCT detector. Therefore, the flickering frequencies appear higher when using the DTGS detector. Similarly, the power spectra of the flicker noise from the flame (globar on) without the bag and the bag filled with CO2 is shown in Figures 4.2-17 and 4.2-18 respectively. The spikes in the power spectrum have drastically decreased. The polyethylene bag has not decreased the amount of noise in the spectrum because if it did, the same trend would be observed in the CO bag experiment. Therefore, it must be the CO2 in the bag that has decreased the flicker noise received by the detector. In summary, the CO2 bag is much more efficient in reducing the spikes and noise in the spectra, especially in the 500-6500 cm-1 region. Thus, methods to filter out the IR emission from CO2 can also reduce the flicker noise.

40

4.2.5 Long-pass infrared filters The use of an infrared long-pass filter should also achieve the same effect as the CO2 bag. A long-pass infrared filter ( Optical Coatings Laboratory Inc.) is designed to filter out everything above 2100 cm-1, and this can remove the dominant CO and CO2 emission from the flame. The transmission spectrum of the filter is shown in Figure 4.2-19. Therefore, spectra taken with the filter can be used to obtain low wavenumber (less than 2100 cm-1) data, and the spectra taken without the filter can be used to analyze regions above 2100cm-1 (where flicker noise is unimportant). In order to prove this point, a power spectrum was again analyzed for both cases, with and without the filter (Figures 4.2-20 and 4.2-21). As shown, the noise spikes are greatly reduced as well as the general noise level (see regions near 1000 cm-1). The peak-to-peak value is approximately .0061 for the case with the filter and is about 0.013 for the case without the filter. The peak-to-peak value is the amplitude of noise in the 1200 cm-1 region of the spectrum. The noise level decreased by a factor of two with the use of a filter. In summary, two methods have worked well in minimizing the flame noise: pointing the flame up and the use of an infrared filter. Both methods reduced the general noise level. These two methods should allow for an increased in SNR in the low wavenumber region. Therefore, both methods were incorporated in obtaining data in this experiment.

41

Figure 4.1-1: Single scan interferogram of the background. The peak in the center is the centerburst.

Figure 4.1-2: Single scan interferogram of the flame at the nozzle (0 mm position). Total Flow = 2 slm, R=1.0.

42

Figure 4.1-3: Power spectrum of a single beam background without the flame (one scan).

Figure 4.1-4: Power spectrum of a single beam spectrum with the flame and globar on (one scan).

43

Figure 4.1-5: Power spectrum of a single beam spectrum of the flame in the region of interest. (one scan). Total flow = 2 slm, R=1.0 .

Figure 4.1-6: Power spectrum of the single beam spectrum of the flame with 75 scans and globar on. Total flow = 2 slm, R=1.0 .

44

Figure 4.2-1: Two detectors used to collect noise spectra from the flame at 90 degrees. Shaded areas represent collected IR.

45

Figure 4.2-2: Flame noise detected by first DTGS detector at 0 using Labview for data acquisition.

Figure 4.2-3: DTGS detector at 90 from the first one using Labview for data acquisition.

Figure 4.2-4: Subtraction of the two spectra obtain from detectors at 90 from each other.

46

Figure 4.2-5: Interferogram of the flame pointing down. Flame is at 2 slm total flow, R=0.94, at the tip of the torch. Single scan. Globar on.

Figure 4.2-6: Interferogram of the flame pointing up. Flame is at 2 slm total flow, R=0.94 at the tip of the torch. Single scan with globar on.

47

Figure 4.2-7: Power spectrum of the flame pointing down and globar off. Flame is at 2 slm total flow, R=0.94, and at the tip of the torch. Single scan.

Figure 4.2-8: Power spectrum of the flame pointing up with globar off. Flame is at 2 slm total flow, R=0.94, and at the tip of the torch. Single scan.

48

Figure 4.2-9: Power spectrum of single scan flame pointing down and globar on in the 2000-3000 cm-1 region. Flame is at 2 slm total flow, R=0.94, and at the tip of the torch. Single scan.

Figure 4.2-10: Power spectrum of single scan flame pointing up with globar on. Flame is at 2 slm total flow, R=0.94, and at the tip of the torch. Single scan.

49

Figure 4.2-11: Power spectrum of single scan flame without substrate.

Figure 4.2-12: Power spectrum of single scan flame with substrate in 0 2000 cm-1 region.

50

Figure 4.2-13: Power spectrum of flame without substrate in 2000 4000 cm-1 region.

Figure 4.2-14: Power spectrum of single scan flame with substrate in 2000 4000 cm-1 region.

51

Figure 4.2-15: Power spectrum of an equivalent single scan of flame noise with a polyethylene bag using a DTGS detector. Globar was off.

Figure 4.2-16: Power spectrum of equivalent single scan of flame noise with CO in bag using a DTGS detector. Globar off.

52

Figure 4.2-17: Power spectrum of the single scan flame without polyethylene bag and globar on. Flame is pointing down.

Figure 4.2-18: Power spectrum of single scan flame with polyethylene bag filled with CO2 and globar on. Flame is pointing down.

53

Figure 4.2-19: Transmission spectrum of the infrared filter. This was taken at 8cm-1 using a DTGS detector. Single scan.

Figure 4.2-20: Power spectrum of a flame pointing down without filter. Globar is on, single scan.

54

Figure 4.2-21: Power spectrum of a single scan flame pointing down with filter. Globar is on, single scan.

55

5.0 ANALYSIS OF THE OPEN FLAME

5.1 E/T Temperature Measurements of the Open Flame The emission/transmission analysis described in Section 2.5.1 was used to determine the average temperature of the flame. In order to obtain the temperature of the gases, the transmission spectrum was subtracted from one to obtain an absorption spectrum. Then the absorption spectrum was baseline corrected to make the absorption at 2500 cm-1 = 0 since the region near 2500 cm-1 had no absorption features. This baseline correction corrected for small errors in the transmission spectrum baseline due to variations in the spectrometer. After baseline correcting, the spectrum was then multiplied by a blackbody function of various temperatures until the CO2/CO peak corresponds to the CO2/CO peak found in the experimental emission spectrum. To demonstrate this, this procedure was carried out for the flame at total flow rate of 2 slm, 0 mm from the tip, and R = 0.90 ( Figure 5.1-1). In this figure, the absorption spectrum was multiplied by the blackbody function at 1900K to yield the calculated radiance. The corresponding experimental radiance spectrum for this flame is shown in Figure 5.1-2. The two spectra were compared to see if the shape and the intensity in the region 2400 1900 cm-1 were the same. Note that the temperature of the blackbody used to match the shape and intensity of the two spectra was the average temperature of the CO2 and CO peaks. The CO2 (2400-2200 cm-1) and CO (2200-1900 cm-1) peaks were used because they have the highest signal to noise ratio.

56

The water cannot be analyzed using the emission/transmission method because there was a significant amount of water in the background. In addition, the background level of water fluctuated, making it very hard to distinguish between room temperature and high temperature water in the transmission spectrum. This produced the inverted peaks in the transmission spectrum (negative peaks in the calculated radiance of Figure 5.1-1). Therefore, the water temperature was determined using the emission spectrum only (see below). Using the E/T method, we had determined the average flame temperature for different positions and R ratios in the flame (total flow rate = 2 slm). Table 5.1-1 shows the results. As shown in the table, the E/T temperature is about 1900K in the CO band and about 1850K in the CO2 band. From the E/T analysis, we observe that the average temperature is not a function of position and is independent of R. The average temperature does not vary by more than 100C. The E/T temperature is an average because both the emission and transmission are collected along a line of sight, so the cold portions of the flame pulls down the radiance while increasing the absorption (1 - transmittance). Thus, the observed E/T temperature is some average of the maximum temperature and room temperature. Therefore, the maximum temperature (highest temperature of the species) in the flame was used in the HITRAN analysis of concentration.

57

Table 5.1-1 Summary of temperature and concentrations at various values of R and position.
Axial Temperature Distribution for the Torch at Ft=2 slm Torch nozzle size zero. 3mm spot size. Position E/T HITRAN (mm) R=0.90 R=0.94 R=0.98 R=1.04 R=1.1 Temperature (K) Temperature (K) CO 0 0.110 0.110 0.110 0.110 0.110 1900 2700 3 0.242 0.242 0.242 0.242 0.230 1900 2700 6 0.220 0.220 0.210 0.210 0.200 1900 2700 12 0.220 0.220 0.210 0.210 0.200 1900 2700 18 0.220 0.210 0.210 0.200 0.190 1900 2700 H2O 0 0.030 0.030 0.030 0.030 0.030 N/A 3000 3 0.042 0.042 0.043 0.045 0.047 N/A 3000 6 0.060 0.060 0.060 0.060 0.055 N/A >3000 12 0.088 0.088 0.088 0.088 0.088 N/A >3000 18 0.090 0.090 0.090 0.088 0.088 N/A >3000 CO2 0 0.013 0.013 0.013 0.013 0.013 1850 2500 3 0.020 0.020 0.020 0.023 0.025 1850 2500 6 0.026 0.026 0.026 0.026 0.026 1850 2500 12 0.04 0.04 0.04 0.04 0.04 1850 2500 18 0.048 0.048 0.048 0.048 0.048 1850 2500 C2H2 0 0.007 0.006 0.005 0.005 0.004 N/A 400 3 <0.002 <0.002 <0.002 <0.002 <0.002 N/A N/A 6 <0.002 <0.002 <0.002 <0.002 <0.002 N/A N/A 12 <0.002 <0.002 <0.002 <0.002 <0.002 N/A N/A 18 <0.002 <0.002 <0.002 <0.002 <0.002 N/A N/A OH 0 0.006 0.006 0.007 0.008 0.009 N/A 3000 3 0.012 0.012 0.014 0.017 0.020 N/A 3000 6 0.020 0.020 0.021 0.021 0.022 N/A 3000 12 0.020 0.020 0.021 0.021 0.022 N/A 3000 18 0.025 0.025 0.025 0.025 0.026 N/A 3000

The adiabatic flame temperature can be used to estimate this maximum temperature. We consider only the primary reaction of C2H2 + O2 2 CO + H2

because the flame contains a high concentration of CO. We will assume that the primary combustion reaction dominates. The overall energy balance of the adiabatic process is:

58

H = 0 = Hf (298) + Hp

(5.1-1)

where Hf = heat of formation and Hp = enthalpy change of the products. Therefore, under adiabatic conditions, the heat of reaction must equal to the enthalpy change of the products from room temperature to the flame temperature. If chemical equilibrium is neglected and all the products are CO and H2, then the heat of reaction is 338 kJ. Based on this, the adiabatic flame temperature is about 3000K. This value is similar to the experimental temperature.

5.2 CO2 Measurements of the Open Flame Using the temperature from the E/T analysis of the CO/CO2 bands, the CO2 emission cannot be simulated correctly. The shapes of the calculated emission spectrum using HITRAN and HITEMP at the E/T temperature differs substantially from the experimental emission spectrum. Figure 5.2-1 shows the hot bands present in the experimental emission spectrum around the 2400 cm-1 region. These peaks do not appear in flames less than 2000K. Therefore, it is not possible that the CO2 is about 1800K, as predicted by the E/T measurements. These trends were also observed by Scutaru et al. [1] where they found that HITRAN92 is missing many high temperature lines below 2300 cm-1 when the temperature is above 700K. They compared experimental radiance spectra to those calculated using HITRAN92 in the temperature range of 750K to 2850K. Since the authors showed some spectra of experimental vs. HITRAN calculated spectra, those empirical correlations can be used to estimate the amount of CO2 in our flame. According to Scutaru et al. [1], the

59

HITRAN92 database did a reasonable good job at fitting the CO2 bands in the region of 2350 2400 cm-1 at high temperature. In addition, they found that the HITRAN92 calculated intensity of the peaks at 2200 2300 cm-1 is roughly 50% of the experimental values. Therefore, we used the region 2350 2400 cm-1 to fit the experimental spectrum to the HITRAN calculated spectrum. Restricting our analysis to that region only, we found that 0.013 atm cm of CO2 at 2500K was in the flame at 0 mm position, total flow rate of 2 slm and R=0.90 (see Figure 5.2-2). As shown in Figure 5.2-2, the 2350 2400 cm-1 region fits reasonably well, and the intensity of the calculated spectrum in the region 2200-2300 cm-1 is roughly 50% of the experimental spectrum, which corresponds to what others have observed. [1] Therefore, for all the other gases (CO, H2O, OH, C2H2 , CH4, CO2), the temperature was adjusted until the shapes in the radiance fits with the calculated shapes. The HITRAN temperature and concentrations for CO2 appear in Table 5.1-1 and Figures 5.6-8 to 5.6-12.

5.3 CO Measurements of the Open Flame The calculated CO emission for 2700K at 0.11 atm cm is shown in Figure 5.3-1. This spectrum matches very nicely with the CO region of the experimental emission spectrum of the flame at 0 mm position, total flow rate of 2 slm and R=0.90 (Figure 5.2-1). By combining this with the calculated CO2 spectrum in Figure 5.2-2, the resulting spectrum is shown in Figure 5.3-2. The comparison between this calculated spectrum to the experimental emission spectrum is shown in Figure 5.3-3. These two

60

spectra are very close to each other except for the missing hot CO2 bands. Concentration and temperature results for CO at other conditions appear in Table 5.11 and Figures 5.6-8 to 5.6-12. The main result found in Table 5.1-1 is that the concentrations of CO and CO2 do not vary significantly with R ratio. For instance, consider the emission spectrum of the flame at 0 mm position, total flow rate of 2 slm, and R=1.10 shown in Figure 5.34 (Note that the neutral flame condition is at R=1.03). The two spectra of R=0.90 (Figure 5.2-1) and R=1.10 (Figure 5.3-4) do not differ very much as shown by the subtraction of the two spectra (Figure 5.3-5). The difference is about 0.1 at max, which is about an 8% difference. Noting the temperatures are roughly the same, this implies that the amount of CO2 is about 8% higher when the R ratio is 1.10 compared to 0.90. However, since the amount of CO2 in the flame is low, an 8% increase yields 0.014 atm cm for R = 1.10 vs. 0.013 atm cm for R=0.90. This increase is very small. However, there is a definite trend that when the R ratio increases, the amount of CO2 in the flame increases.

5.4 H2O Measurements of the Open Flame To obtain the amount of water in the flame, the emission spectra has been studied because the transmission spectrum is contaminated with water from the background. The amount of water in the background changes with time, making it hard to determine the amount of water in the flame. The emission spectra are better since the

61

water vapor in the room does not emit IR whereas hot water vapor in the flame emits significant IR. The best place to look when analyzing the water is in the 2600 3100 cm-1 region since both the H2O and OH bands are observed in that region. In addition, the curvature of the spectrum in this region is strongly dependent on the temperature of the water. Therefore, if the temperature of the water is correct, the shape should be the same. This is shown in Figure 5.4-1 where the water emission is compared the calculated (HITEMP Voigt) emission spectra at 1900K and 3000K. The experimental emission spectrum is much closer to the 3000K spectra. Therefore, the water in the flame should be about 3000K as shown in Figure 5.4-2. In the open free flame (Ft = 2 slm, R = 0.90 at the 0 mm position), there is ~0.03 atm cm of water. The calculated (HITEMP) spectrum fits the observed spectrum well. The water concentration does not vary much with R ratio at the 0 mm position. The water bands at R=0.90 and 1.10 are compared, there is no significant difference in the water emission. Since the water is at about 3000K, there should be hot water lines in the transmission spectrum. Even though the background spectrum is dominated by room temperature water from the atmosphere, the hot water bands should still appear in the transmission spectrum since hot water is much broader that room temperature water (shown in Figure 5.4-3). (Although the partial pressure is different in the two calculated spectra, the 0.03 atm cm of water at 3000K is equivalent to 0.003 atm cm water at 300K since the concentration and path length is constant). Using Figure 5.4-3

62

as a guide, we can observe these hot water bands in the experimental transmission spectrum (Figure 5.4-4). Similar water analyses were done at different position and R value of the flame. The results are tabulated in Table 5.1-1 and shown in Figures 5.6-8 through 5.6-12.

5.5 OH Measurements of the Open Flame There are OH features in the spectrum as shown in Figure 5.4-1 and 5.4-2. When trying to determine the temperature of the OH, different temperatures were compared to the experimental spectrum until there is a good fit. When the OH is added to the water spectrum (0.03 atm cm at 3000K), the 3000K OH spectrum fit better because the features below 3000 cm-1 are more intense, agreeing with the experimental spectrum. The comparison of the calculated H2O + OH spectra and the experimental spectrum is shown in Figure 5.5-1. The difference of the two spectra is shown in Figure 5.5-2. There is a slight difference in the baseline, with a maximum difference of 36% and an average error of 2%. This error is considered good. Some of the difference is that the temperature of the water is postulated to be above 3000K, but the database used for calculation was only available up to 3000K. Note that there is an extra line in Figure 5.5-2 (noted by an asterisk) at 3587 cm-1, which may be due to missing data in the HITRAN database. Similar analyses were done at other position and R values of the flame for OH. The results are tabulated in Table 5.1-1 and shown in Figures 5.6-8 through 5.6-12.

63

5.6 C2H2 Measurements of the Open Flame The two strongest absorption features for C2H2 are at 730 cm-1 and 3280 cm-1. Preliminary comparisons of HITRAN calculations and experiment found that the calculated acetylene peaks in the 730 cm-1 and 3280 cm-1 bands are not consistent with each other. Therefore, detailed experiments were carried out to compare the experimental and calculated acetylene spectra. A spectrum of acetylene flowing (at 0.048 atm and 300K) through a gas cell of 15 cm long was taken at 1 cm-1 resolution and appears in Figures 5.6-1 (730 cm-1 band) and 5.6-2 (3280 cm-1 band). Notice that the calculated spectrum at 730 m-1 is about six times higher than the experimental spectrum (Figure 5.6-1). The Q branch (730 cm-1) shown has an absorbance of 11.5 units while the experimental spectrum shows an absorbance of 1.7 units. When comparing the spectra at the 3280 cm-1 (Figure 5.6-2), the peaks are much more similar in intensity and shape. Therefore, HITRAN92 and HITRAN96 over predicts the peak at 730 cm-1 while the peak at 3280 cm-1 is more accurate. Therefore, the peak at 3280 cm-1 was used for quantitative analysis. The emission spectra do not show any acetylene at any R values or locations in the flame; C2H2 is observed only in the transmission. Therefore, the temperature of the acetylene must be low. Using the experimental transmittance alone, we find the best fit with 0.007 atm cm at 400K. A comparison of calculation and experimental spectra is shown in Figure 5.6-3 with the difference of the two spectra shown in Figure 5.6-4. The average difference is 0.0006, which is very small. The average percentage error is about 6%. Most of it is due to the noise in the transmission spectrum. Other

64

position and R values were calculated for C2H2 and are tabulated in Table 5.1-1 and shown in Figure 5.6-8 through 5.6-12. The temperature of the acetylene dictates the shape of its transmission spectrum. Therefore, if the experimental shapes of the acetylene peak (at 3280 cm-1) match the calculated shapes, then the temperature is correct. At 300K, the maxima in the P and R branches are not at the same location as the experimental C2H2 spectrum (Figure 5.6-5). The same scenario occurs when 500K acetylene is calculated (Figure 5.6-6). The 400K calculated spectrum matches up much better (Figure 5.6-3). At 0 mm position, acetylene is observed for all R ratios, and as the R ratio increases, the amount of acetylene decreases. The detection limit for acetylene is 0.002 atm cm, twice the noise level. At the 3, 6, 12, and 18 mm positions, the level of acetylene is below the detection limit. If acetylene is present in the flame, the acetylene peak at 730 cm-1 should also appear in the transmission spectrum. This peak is observed in the transmission spectrum using the longpass infrared filter to reduce the flicker noise (Figure 5.6-7). However, the calculated 730 cm-1 peak is bigger than the experimental, in line with experiments with the acetylene gas cell.

5.7 Discussion of Results In the following discussion, we assume that the species observed with the FTIR in this experiment are produced by the following four reactions: C2H2 + O2 H2 + O2 2 CO + H2 H2O (5.7-1) (5.7-2)

65

2CO + O2 H2+ O2

2 CO2 2OH

(5.7-3) (5.7-4)

Equation 5.7-1 is the primary combustion step while the other reactions account for secondary combustion processes. Now consider the following analysis for the data at 0 mm position and R = 0.90. Since 0.013 atm cm of CO2 is produced, there is 0.013 atm cm of CO converted to CO2 due to the 1:1 stoichiometry of Equation 5.7-1. Combined with measured CO, that means at least 0.123 atm cm of CO is produced in the primary combustion. From stoichiometry, that means at least 0.062 atm cm of H2 is produced in the primary reaction (Equation 5.7-1). However, the amount of H2 can be higher if some of the C2H2 converts directly to C or C2, which are not detectable by the FTIR. Since there are 0.03 atm cm of H2O present in the flame, at least 0.032 atm cm of H2 (or 0.064 atm cm of hydrogen radicals) is in the flame. Similar analysis for H2 at other positions and R values appears in Table 5.7-1. Since we know how much CO and CO2 is in the flame, it is also possible to determine how much C2H2 is consumed by the reactions. Based on the first reaction, there should be about 0.0615 atm cm of C2H2 consumed to produce 0.123 atm cm of CO. With a residual 0.007 atm cm at 400K, the corrected concentration is (assuming the CO and CO2 are at 2500K in the flame): 0.0615 2500 400 + 0.007 = C 2 H 2 (300 K ) = .5218 atm cm 300 300 (5.7-5)

where C2H2 (300K) is the equivalent amount of C2H2 entering the flame at a temperature of 300K. If the amount of oxygen / acetylene ratio is 0.90, then the amount of oxygen fed to the torch is about 0.47 atm cm 300K and 0.056 atm cm at

66

2500K. The oxygen entrainment is the difference between the amount of oxygen needed to produce the observed CO, OH, CO2, and H2O, and the amount fed to the torch. There should be about 0.086 atm cm necessary to produce the observed CO, OH, CO2, and H2O in the flame. This leads to an oxygen entrainment of 0.03 atm cm (2500K) at the 0 mm position. Since the oxygen entrainment comes from the air, there would be nitrogen in the flame that is not detectable by the FTIR. The nitrogen entrainment is calculated based on the air containing 79% nitrogen and 21 % oxygen. Value for different positions and R values are tabulated in Table 5.7-1. Note that the values there have the units of atm cm. Therefore, if the path length of each individual gases were known, then those values can be divided by its path length to yield partial pressure. Unfortunately, the path lengths are unknown, but some general trends are still clear (Figure 5.6-13 to 5.6-14). Note that the O2 diffusion is greater when it is farther from the nozzle. This makes very good intuitive sense since more oxygen will entrain the flame farther along the flow axis. There is a big jump between the 0 and 3 mm position because of the great concentration difference there. Based on the way the O2 entrainment is calculated, if this position contains a high level of C and C2 radicals that the FTIR cannot detect, it lowers the calculated amount of oxygen fed to the torch, thereby increasing the O2 entrainment. The amount of H2 is highest at the 3 mm position, which is the region in which diamond growth occurs. When the flow rate is 2 slm, the inner cone is about 4 mm in

67

length. Therefore, the deposition distance is about 5 mm form the torch nozzle. There are no sign of methane in the flame (as seen in the enclosed flame). The errors associated with this method of quantitative analysis are accurate to about 10%. For example, the difference between 0.1 atm cm of CO and 0.11 atm cm of CO is easily detectable. However, the difference between 0.013 atm cm of CO2 and 0.0135 atm cm of CO2 is not that obvious. In Figure 5.6-4, the fit between the experimental and calculated spectra have an error of about 6% (within experimental error). The maximum difference in the Figure is about 12%, which is about the error bars for this analysis. Table 5.7-1 Tabulated values of OH, C2H2 , calculated H2, N2, and O2 entrainment, in atm cm.
Concentration as a function of position and R ratio (in atm cm) Position R ratio (mm away from torch 0.90 0.94 0.98 1.04 1.1 nozzle) H2 0 0.029 0.029 0.028 0.028 0.027 3 0.083 0.083 0.081 0.079 0.070 6 0.053 0.053 0.048 0.048 0.047 12 0.032 0.032 0.027 0.027 0.021 18 0.032 0.027 0.027 0.024 0.018 O2 0 0.030 0.027 0.025 0.022 0.019 Entrained 3 0.050 0.045 0.041 0.037 0.033 2500K 6 0.065 0.060 0.056 0.049 0.040 12 0.087 0.082 0.077 0.070 0.063 18 0.095 0.089 0.084 0.076 0.069 N2 0 0.092 0.085 0.080 0.067 0.058 3 0.187 0.168 0.156 0.140 0.125 6 0.246 0.227 0.210 0.184 0.151 12 0.327 0.308 0.290 0.261 0.237 18 0.357 0.336 0.316 0.284 0.260

68

In summary, we found that different species have slightly different temperature because they are located in different parts of the flame. The major trends is that the concentration of CO2, H2O, and O2 entrained increases with increasing axial position. The amount of C2H2 decreases with increasing axial position, and CO and H2 have a maximum at position 3 mm from the torch nozzle. There is also a big jump in the partial pressure * pathlength at the 3 mm position. These results are shown in Figures 5.6-8 to 5.6-14. As the R ratio increases, the amount of CO is about constant at the 0 mm position and it decreases at positions greater than 0 mm. For the H2O, the amount is constant at 0 mm position, increases at the 3 mm position, then decreases at the 6 mm position, and finally levels off at positions greater than 12 mm. For the CO2, as R increases, the amount of CO2 remains constant at all positions except for the 3 mm, where it increases. For the C2H2 , the amount decreases with increasing R ratio. For OH, the amount increases with increasing R ratio. Finally for the H2 and O2 entrainment, the amount decreases with increasing R ratio.

5.8 Comparison of Results to Previous Work Morrison et al. [2, 3] performed in situ measurements of the oxyacetylene flame with the FTIR. The most extensive data set is the radial profiles of concentration and temperature at 9 mm from the torch nozzle. They found that the temperature of the flame is about 3000K at the tip of the torch, consistent with HITRAN temperature here. Since the temperature reported in this experiment is the average temperature

69

over a 3 mm spot size, then the temperatures reported by Morrison et al. must be average over the first 3 mm in order to draw a correct comparison. If the temperature in the first 3 mm is averaged, the temperature is very close. However, they did report a decrease in relative concentration of the CO species farther from the flame, which is again consistent with what is observed here. The authors did not observe any acetylene in the flame. Matsui et al. [4, 5] observed CH* and C2* by emission spectroscopy in the inner cone which were not observed in this experiment. However, C2* is not observable by the FTIR, and the concentration of CH* may be too low to be detected using the FTIR under the noisy flame. They estimated partial pressure of CH as 2 x 10-5 to 6 x 10-5 atm. [6] These concentrations are too low to be measured. Matsui et al. also observed OH in the flame at the outer boundary. The results in this experiment are consistent with that since we quantitatively showed OH is in higher amounts farther from the inner cone. Yalamanchi et al. [7] found C, CH, and H in the inner cone. Again, the C and H are not observable by the FTIR and the concentration of CH may be too low to be detected (the authors did not report on the concentration of CH). Matsui et al. [8] found high concentration of CO and H2 in the feather region of the flame at R = 0.90. This is also consistent with results presented here in this experiment, where high concentration of CO is found in the feather region (3mm from torch tip) (Table 5.1-1 and Figures 5.6-8 to 5.6-12 and Figure 5.6-14). The authors also found water and acetylene at lower levels than CO and H2 in the flame.

70

The same trend is observed in this experiment. They did not observe any acetylene in the intermediate zone (3 mm from torch tip). None is observed in this experiment too. Klein-Douwel et al. [9] found OH to be barely detectable inside the feather and was found primarily outside of the feather. In this experiment, small quantities of OH are observed in this region (0mm from tip) and higher amounts are found at distance greater than 0 mm from the torch tip. Welter and Menningen [10] determined the CH derived temperature at 6 mm from the torch to be 2700K, with the amount of OH increasing with increasing distance from the torch tip. This is consistent with the temperatures and OH concentrations found here. In general, the results obtained with the FTIR correspond with the results obtained by many authors.

71

Figure 5.1-1: Calculated radiance from the transmission data: (1-transmission)*blackbody(1900K). Flame conditions: Total flow = 2 slm, R=0.90, and at 0 mm position.

Figure 5.1-2: Experimental radiance. Flame conditions: Total flow = 2 slm, R=0.90, and at 0 mm position.

72

Figure 5.2-1: Experimental spectra of the CO2 hot emission bands. Flame conditions: Total flow = 2 slm, R=0.90, and at 0 mm position.

Figure 5.2-2: Overlap spectra of experimental and HITRAN92 (0.013 atm cm CO2 in 1 atm, 2500K; Voigt lineshape) calculated. Flame conditions: Total flow = 2 slm, R=0.90, and at 0 mm position.

73

Figure 5.3-1: Calculated spectrum from HITRAN96 with 0.11 atm cm in 1 atm at 2700K; Voigt lineshape.

Figure 5.3-2: Combined spectra of calculated CO (0.11 atm cm in 1 atm, 2700K, HITRAN96 Voigt lineshape) and CO2 (0.013 atm cm 2500K, HITRAN92 Voigt lineshape).

74

Figure 5.3-3: Comparing calculated spectrum to experimental spectrum. Calculated parameters: CO2 : 0.013 atm cm in 1 atm, 2500K, Voigt lineshape. CO : 0.11 atm cm in 1 atm, 2700K, Voigt lineshape. Experimental conditions: Total flow = 2 slm, R=0.90, and at 0 mm position

Figure 5.3-4: Experimental spectrum of flame : Total flow = 2 slm, R=1.10, and at 0 mm position

75

Figure 5.3-5: Difference of R = 0.90 minus R=1.10 . Experimental conditions: Ft=2 slm and at the 0 mm position.

Figure 5.4-1: Experimental and calculated spectra in the water band region. Calculated parameters: HITEMP, 0.03 atm cm in 1 atm, Voigt lineshape at 3000K and 1900K. Flame conditions: Ft=2 slm, R=0.90, and at 0 mm position.

76

Figure5.4-2: Experimental (0 mm position, R=0.90, Ft=2 slm ) plotted with the calculated water bands (0.03 atm cm at 3000K). Experimental spectrum is offset for clarity. Calculated spectrum: 0.03 atm cm in 1 atm, HITEMP, 3000K, Voigt lineshape.

77

Figure 5.4-3: Comparison of room temperature water with 3000K water transmission. Calculated parameters: 1 atm total, Voigt lineshape, HITEMP.

Figure 5.4-4: Transmission spectrum of 0 mm position with R =0.90 showing hot water bands.

78

Figure 5.5-1: Experimental (0 mm position, R=0.90, open flame, Ft=2 slm) plotted with the calculated spectra including the OH bands. H2O: 0.03 atm cm in 1 atm at 3000K HITEMP, Voigt lineshape. OH: 0.006 atm cm in 1 atm, HITRAN 96, Voigt lineshape. Experimental spectrum is offset for clarity.

Figure 5.5-2: Calculated spectrum minus experimental spectrum for the water band region. Taken from flame position 0 mm with R = 0.90 and Ft=2 slm. Calculated parameters: H2O: 0.03 atm cm in 1 atm at 3000K, HITEMP, Voigt lineshape. OH: 0.006 atm cm in 1 atm HITRAN 96; Voigt lineshape; * denotes 3587 cm-1 band.

79

Figure 5.6-1: Comparing calculated to experimental acetylene peak at 730 cm-1. Experimental conditions: 0.048 atm in 1 atm, 15 cm pathlength, and 300K. Calculated parameters: 0.048 atm in 1 atm, 15 cm, 300K, and Voigt lineshape. Experimental spectrum is offset for clarity.

Figure 5.6-2: Comparing calculated to experimental acetylene peak at 3280 cm-1. Experimental conditions: 0.048 atm in 1 atm, 15 cm pathlength, and 300K. Calculated parameters: 0.048 atm in 1 atm, 15 cm, 300K, and Voigt lineshape. Experimental spectrum is offset for clarity.

80

Figure 5.6-3: Experimental and calculated absorbance acetylene spectrum. Flame conditions: 0 mm position, total flow = 2 slm and R=0.90. Calculated parameters: 0.007 atm cm in 1 atm 400K HITRAN 96, Voigt lineshape. Spectra shifted for clarity.

Figure 5.6-4: Experimental minus calculated absorbance spectra in the acetylene region for 0 mm position; from Figure 5.6-1.

81

Figure 5.6-5: Comparing the acetylene bands at 300K with experimental. Calculated spectrum shifted for clarity. Calculation parameters: 0.005 atm in 1 atm, 1 cm pathlength, 300K, Voigt lineshape, HITRAN 92.

Figure 5.6-6: Comparing experimental absorbance with calculated acetylene at 500K (0.005 atm cm in 1 atm, Voigt lineshape, HITRAN96). Calculated spectrum shifted for clarity.

82

Figure 5.6-7: Comparing calculated acetylene spectrum (400K) with experimental spectrum (0 mm position and R=0.90) in the 730cm-1 region.

Free Open Flame R = 0.90


0.100 0.300 0.090

Partial Pressure * Path length (atm cm)

0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000 0 5 10 15 20

0.200

CO concentration (atm cm)

0.080

0.250

H2O CO2 C2H2 OH CO

0.150

0.100

0.050

0.000

Distance from Torch (mm)

Figure 5.6-8: Plot of concentrations as a function of distance for R=0.90 Ft=2 slm.

83

Open free flame R=0.94


0.100 Partial Pressure * Path length (atm cm) 0.090
CO concentration (atm cm)

0.300 0.250 0.200 0.150 0.100 0.050 0.000 0 5 10 15 20 Distance from torch (mm)

0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000

H2O CO2 C2H2 OH CO

Figure 5.6-9: Plot of concentration as a function of distance at R=0.94 Ft=2 slm

Open free flame R=0.98

Partial Pressure * Path length (atm cm)

0.100 0.080 0.060 0.040 0.020 0.000 0 5 10 15 20 Distance from torch (mm)

0.300 0.250 0.200 0.150 0.100 0.050 0.000


CO concentration (atm cm) H2O CO2 C2H2 OH CO

Figure 5.6-10: Plot of concentration versus distance for R=0.98

84

Open free flame R=1.04


0.100 Partial Pressure * Path length (atm cm) 0.090
CO concentration (atm cm)

0.300 0.250 0.200 0.150 0.100 0.050 0.000 0 5 10 15 20

0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000

H2O CO2 C2H2 OH CO

Distance from torch (mm)

Figure 5.6-11: Plot of concentration versus distance for R=1.04

Open free flame R=1.10


0.100 Partial Pressure * Path length (atm cm) 0.090
CO concentration (atm cm)

0.250

0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000 0 5 10 15 20

0.200

H2O CO2 C2H2 OH CO

0.150

0.100

0.050

0.000

Distance from torch (mm)

Figure 5.6-12: Plot of concentration versus distance for R=1.10

85

O2 entrainment
Partial Pressure * Path length (atm cm) 0.100 0.090 0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000 0 5 10 distance from torch tip (mm) 15 20 R=0.90 R=0.94 R=0.98 R=1.04 R=1.10

Figure 5.6-13: Plot of oxygen entrainment for different R ratios

Calculated H2
Partial Pressure * path length (atm cm) 0.100 0.090 0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000 0 5 10 Distance from torch tip (mm) 15 20

R=0.90 R=0.94 R=0.98 R=1.04 R=1.10

Figure 5.6-14: Plot of calculated hydrogen for various R values and position.

86

6.0 ANALYSIS OF THE ENCLOSED FLAME 6.1 Equipment Modifications 6.1.1 Linear and non-linear detectors In the experiment with the open flame, a standard MCT detector (Bomem Model #SMH-3200G) was used to obtain the spectrum. However, the detector was leaking after those sets of experiments were done. Therefore, another MCT detector (Bomem Model #FTIR-W22-1.00) was used. The difference is that model FTIR-W22-1.00 has a built-in linearizer while model SMH-3200G does not. The detector without the linearizer has more sensitivity but produces unwanted baseline shifts and has nonlinear output at higher wavenumbers. Two single beam spectra are shown in Figure 6.1-1, and both are done at the same gain with 50 scans and 1 cm-1 resolution. The intensity of the standard detector is much higher than that of the linearized detector. To compare the two detectors at experimental conditions, a spectrum of typical oxyacetylene flame used in this experiment (total flow rate of 2 L/min and R = 0.90) was taken with both detectors on the same day. An absorbance spectrum was calculated by dividing the single beam spectrum by the background spectrum for both of the detectors. The absorbance spectra are shown in Figure 6.1-2, and the spectra are not shifted for clarity. Both detectors detect the same absorption features, however, there is less noise in the 5000 cm-1 and 2500 cm-1 regions for the standard detector. The two spectra show a slightly different absorption for water because of differences in the amount of water in the background (3600 cm-1 and 5380 cm-1). As

87

shown also in Figure 6.1-2, there is an offset for the standard MCT detector, which is caused by the nonlinearity nature of the detector. Therefore, the standard MCT absorbance spectrum was baseline shifted in order to compare the shapes of the CO2 bands with the linearize detector. The baseline shifted spectra are shown in Figure 6.1-3. The two detectors show very similar absorbance for the CO2 region. In the 2360 cm-1 region, more features are detected with the standard detector. Thus the linear and nonlinear detectors produce similar absorbance spectra except the nonlinear detector has less noise but contains a spurious baseline shift.

6.1.2 Purging the arm of the spectrometer The extension arm used to extend the IR beam to the flame (Figure 3.2-1) was tested to see if purging it decreased the amount of water in the spectrum and thus better controlled the water levels in the background. The entire path of the IR, from the globar source to the focusing mirror before the flame was purged. Normally, when taking backgrounds at different times on the same day, the water levels changed substantially. Comparison of the background spectrum (2 scans at 1 cm-1 resolution) to a background spectrum after 5 minutes of purging with argon is shown in Figure 6.1-4. The figure shows the two background spectra divided by each other. The resultant spectrum should be a straight line at one if both conditions were identical. However, there are about 65% less CO2 and 50% less H2O in the spectrum with the purge. Therefore, purging the arm is an advantage when performing the enclosed flame analysis since one of the reason of enclosing the flame is to remove as much

88

water as possible from the spectrum. Another method to remove the water from the spectrum was to use a desiccant in the arm. However, the desiccant did not perform as well as purging the arm. Therefore, the arm of the spectrometer was purged with ultra high purity nitrogen (instead of argon because of availability) at 99.999% purity. The enclosed flame apparatus was purged with ultra high purity argon (99.999%) to keep the KBr windows cool and to remove the combustible gases that builds up inside.

6.2 Temperature Measurement in the Enclosed Flame The enclosed flame has particles at the 0 mm position. This is observed in the emission spectrum where all the gas emission features are riding on a greybody shape at about 450K as seen in Figure 6.2-1. The particles are also observed in the transmission spectrum where the entire transmission spectrum is baseline shifted from one (Figure 6.2-2). The shift in the transmission spectrum is caused by the particles blocking some of the light by scattering and absorption. The greybody shape seen in Figure 6.2-1 is calculated by taking (1 transmission) of Figure 6.2-2 and multiplying it with a blackbody at 450K. At positions 3 mm and greater, the particles disappeared because the transmission spectra do not show a baseline shift. This is shown in the emission spectrum of Figure 6.2-3 where the emission spectrum does not show a hump at the low wavenumber region. The transmission spectrum (Figure 6.2-4) also shows a baseline value of 1.

89

To calculate the gas temperature by E/T, the blackbody emissions due to the particle was subtracted from the emission spectrum before further analysis. Then the emission baseline was shifted to zero if necessary. Then the transmission spectrum was converted to absorbance, and the baseline due to the particles was removed. Afterward, the spectrum was converted back to the transmission spectrum. Then, the E/T analysis was done on the enclosed flame just like the open flames to determine the average temperature of the CO2 and CO bands. For the water, the temperature was determined by observing the 2100 2600 cm-1 region for curvature in the radiance spectra as describe in the open free flame section. The results are tabulated in Table 6.2-1. The temperature of the CO2 is much lower than all the other species when using the E/T method. One reason for this is that the CO2 may be closer to the edge of the flame. In addition, the CO temperature obtained is much cooler than that of the open free flame because of the CO being trapped inside the enclosed flame apparatus.

90

Table 6.2-1 Concentrations and temperatures as a function of position and R ratio (in atm cm). Position is axial distance from the torch nozzle (in mm) and R = O2/C2H2 Enclosed Flame Concentration as a function of position and R ratio (in atm cm) Position (mm) R ratio HITRAN E/T (away from torch nozzle) 0.90 0.94 0.98 1.04 1.1 Temperature Temperature CO 0 0.050 0.048 0.045 0.045 0.043 2500 1200 3 0.286 0.292 0.298 0.298 0.298 2500 1800 6 0.375 0.375 0.375 0.375 0.375 2500 1800 12 0.398 0.398 0.398 0.398 0.398 2500 1850 18 0.300 0.300 0.300 0.300 0.300 2500 1700 CO 0 0.025 0.020 0.019 0.019 0.017 >3000 N /A 3 0.200 0.210 0.221 0.221 0.221 >3000 N /A =2 6 0.180 0.180 0.180 0.180 0.189 >3000 N /A 12 0.187 0.187 0.197 0.197 0.197 >3000 N /A 18 0.112 0.112 0.118 0.121 0.125 >3000 N /A H2O 0 0.003 0.004 0.005 0.006 0.007 3000 N /A 3 0.011 0.013 0.017 0.024 0.034 3000 N /A 6 0.015 0.018 0.022 0.032 0.045 3000 N /A 12 0.015 0.020 0.025 0.038 0.049 3000 N /A 18 0.015 0.020 0.025 0.034 0.044 3000 N /A CO2 0 0.004 0.005 0.006 0.006 0.007 2000 570 3 0.003 0.003 0.004 0.006 0.007 2000 600 6 0.006 0.007 0.010 0.013 0.016 2000 640 12 0.008 0.010 0.011 0.015 0.019 2000 640 18 0.009 0.011 0.013 0.016 0.020 2000 650 C2H2 0 0.007 0.004 0.004 0.004 0.003 400 3 <.002 <.002 <.002 <.002 <.002 N /A 6 <.002 <.002 <.002 <.002 <.002 N /A 12 <.002 <.002 <.002 <.002 <.002 N /A 18 <.002 <.002 <.002 <.002 <.002 N /A OH 0 <.0015 <.0015 <.0015 <.0015 <.0015 N /A 3 <.0015 <.0015 <.0015 <.0015 <.0015 N /A 6 <.0015 <.0015 <.0015 <.0015 <.0015 N /A 12 <.0015 <.0015 <.0015 <.0015 <.0015 N /A 18 <.0015 <.0015 <.0015 <.0015 <.0015 N /A CH4 0 0.007 <.0032 <.0032 <.0032 <.0032 650 3 <.0032 <.0032 <.0032 <.0032 <.0032 N /A 6 <.0032 <.0032 <.0032 <.0032 <.0032 N /A 12 <.0032 <.0032 <.0032 <.0032 <.0032 N /A 18 <.0032 <.0032 <.0032 <.0032 <.0032 N /A 0.21 0.20 0.19 0.17 0.16 Soot 0 450 3 0 0 0 0 0 N /A 6 0 0 0 0 0 N /A 12 0 0 0 0 0 N /A 18 0 0 0 0 0 N /A

91

6.3 CO Measurements in the Enclosed Flame It is difficult to quantitatively analyze the CO2 and CO in an enclosed flame because much of the CO is trapped inside the box. A large cold band of CO is observed in the transmission spectrum: CO builds up in the box and obscures the CO spectrum. Thus, the cold CO band absorbs some of the emission from the hot CO in the flame as shown in Figure 6.3-1. Therefore, in order to get the concentration of the CO in the flame, only the hot band edge from the emission spectrum was analyzed. These hot bands are outside of the cold band in the transmission spectrum in regions between 1900 2000 cm-1. The temperature determined by E/T analysis (1200K) does not match the temperature of the CO needed to match the emission spectrum because E/T gives average temperature, which includes the CO outside the flame. These are shown in Figures 6.3-2 and 6.3-3. In Figure 6.3-3, it is shown that 1200K CO does not fit the emission CO bands very well. Part of the reason again is the E/T gives average temperature of the CO in the line of sight. If there are colder CO in the enclosed flame apparatus, it lowered the apparent temperature of the CO. The CO temperature in the flame is much closer to 2500K with a concentration of 0.05 atm cm at the 0 mm position when R = 0.90 (Figures 6.3-4). Similar analyses are done at other positions and R values of the enclosed flame. The results are tabulated in Table 6.2-1.

92

6.4 CO2 Measurements in the Enclosed Flame The temperature of the CO2 has the same problem as with CO when performing the E/T analysis. From E/T analysis, the CO2 should be about 600K. However, when CO2 is at 600K, it does not have the hot features shown in experimental emission spectrum (Figure 6.4-1). Those hot bands do not appear until it is above 2000K. Therefore, this is the approximate temperature of the CO2 in the flame. Again, some lines of the hot CO2 are missing as discussed in the open flame section (Section 5.2). The same procedures used in the open flame analysis were used for the enclosed flame to determine the concentration of CO2 at other axial positions and R values. Those results are tabulated in Table 6.2-1.

6.5 H2O Measurements in the Enclosed Flame It is again difficult to determine the temperature of the water in the flame using E/T analysis because the fluctuating amount of water in the atmosphere makes the transmission spectrum unreliable. Therefore, only the water in the emission spectrum was analyzed, especially the hot water bands. The methods used to analyze quantitatively the water are similar to the method used for the open free flame. Therefore, only the results are presented here in Table 6.2-1.

6.6 OH Measurements in the Enclosed Flame It is particularly easy to observe any OH peaks in the 2800 3000 cm-1 because there is little water emission in this region. The experimental and calculated OH

93

spectra are shown in Figure 6.6-1. No OH is observed in the flame at 0 mm position and R = 0.90. The detection limit is 0.0015 atm cm (twice the noise) for the enclosed flame. No observable OH is found in any region when R = 0.90 (See Figure 6.6-2 for the 18 mm position when R = 0.90) nor at any other positions and R values.

6.7 C2H2 Measurements in the Enclosed Flame Acetylene is only observed in the 0 mm position. The acetylene at 3200 cm-1 is fitted with the calculated spectrum to determine the concentration. The best temperature fit is 400K using an analysis similar to the open flame. The detection limit for acetylene is 0.002 atm cm. Therefore, all other positions have less than 0.002 atm cm of C2H2.

6.8 CH4 Measurements in the Enclosed Flame In some cases, a methane peak shows up in the transmission spectrum but not the emission spectrum. This indicates that the methane is cold and not readily emitting IR. The temperature of the methane is determined by the width of the main peak (Figure 6.8-1). The experimental spectrum is shown in Figure 6.8-2 with the width of the peaks matching up nicely as well as portions of the R branch. This corresponds to about 400K with a concentration of 0.002 atm cm. The detection limit for CH4 is 0.0008 atm cm in 1 atm at 400K.

94

6.9 Discussion of the Enclosed Flame As discussed in Chapter 5, the amount of hydrogen in the flame can be calculated using the overall reactions: C2H2 + O2 H2 + O2 2CO + O2 H2+ O2 C2H2 + 3 H2 2 CO + H2 H2O 2 CO2 2OH 2CH4 (6.9-1) (6.9-2) (6.9-3) (6.9-4) (6.9-5)

Table 6.9-1 shows the calculated amount of hydrogen in the flame similar to the analysis done with the open flame. Note there are entrainment of argon that is not accounted for in the table. The plots of concentration vs position for all species are shown in Figures 6.9-1 through 6.9-7.

Table 6.9-1 Calculated Concentration of H2 and O2 entrained in atm cm.


Species H2 Concentration in atm cm Position (mm from tip) 0 3 6 12 18 0 3 6 12 18 0.90 0.021 0.134 0.176 0.188 0.139 0.005 0.021 0.030 0.032 0.027 R ratio 0.94 0.022 0.135 0.173 0.184 0.136 0.005 0.017 0.024 0.027 0.024 0.98 0.021 0.134 0.171 0.179 0.131 0.005 0.014 0.020 0.022 0.022 1.04 0.020 0.127 0.162 0.168 0.124 0.004 0.009 0.015 0.018 0.019 1.10 0.017 0.118 0.150 0.159 0.115 0.004 0.005 0.011 0.013 0.016

O2 Entrained

The major results for the enclosed flame is that as the distance from the torch nozzle increases with the same R, the following trends are observed: CO has a maximum at 12 mm position, CO2 has a minimum at 3 mm position. The water in the

95

flame increases with increasing distance from the nozzle at low R but maximizes at 12 mm at high R. The acetylene decreases with increasing axial position with methane appearing only in the 0 mm position with R = 0.90. The OH is not detected at all. The O2 entrainment and H2 are at maximum at the 12 mm axial position. With the same position and increasing R, the following trends are observed: CO, H2, and C2H2 tend to decrease. CO2, O2 entrained, and H2O tend to increase. Table 6.11-1 summarizes these results.

6.10 Oxygen Entrainment in the Enclosed Flame Theoretically, there should not be any oxygen entrainment in the enclosed flame if the apparatus is totally leak-proof. However, the apparatus may have a small leakage from air in the atmosphere. The amount of O2 entrainment in the enclosed flame, however, is only two to three times less than that of the free flame, if the amount of carbon radicals is neglected. This is because the amount of oxygen fed to the torch is based on the amount of acetylene back calculated from the observed species (CO2, and CO). If there are carbon radicals and/or C2 species in the flame, then the actual amount of acetylene fed to the torch would be higher, thereby increasing the calculated oxygen feed and decreasing the calculated oxygen entrainment. The assumption that there are very little carbon radicals and C2 may not be valid. Therefore, the amount of O2 entraiment here is the maximum amount of entrainment. The amount of O2 entrainment is lower at the 18 mm position for a reason. The carbon radicals that are not observable by the FTIR (but could be present) in the

96

regions closer to the torch nozzle are now reacted to form CO and CO2. Since the calculation for oxygen entrainment is based on the amount of CO and CO2 species observed, this will decrease the oxygen entrainment. This is because more of the carbon is now observable, therefore increasing the calculated amount of oxygen fed to the torch.

6.11 Comparison of the Free and Enclosed Flame No authors have done work in analyzing the concentration of the species in an enclosed flame. One of the reasons is that it is very difficult to carry out analytical techniques in an enclosed atmosphere. Therefore, no comparison to others can be made. Soot is observed in the enclosed flame at the 0 mm position. The temperature of the soot is close to the temperature of the acetylene. In the enclosed flame, there are no signs of OH in the flame under diamond growing conditions (fuel rich). This is consistent with the theory that OH is not important for diamond deposition [1] and the enclosed flame grows diamond films at a higher quality. [2] The amount of acetylene in the 0 mm position is about the same for open and enclosed flame. However, methane is observed in the enclosed flame at axial position close to the nozzle and at low R. Both observations are consistent with the fact that diamond is grown near the cone. There is much more CO in the enclosed flame at positions greater than 3 mm from the nozzle. Since there is no oxygen in the enclosed flame surroundings, the

97

secondary reaction is much slower than in the case of the free flame. In addition, the calculated hydrogen concentration in the flame is about three times more in the enclosed flame than in the open flame. Hydrogen is known to play a major role in diamond deposition. This may be the reason why diamond films grown by enclosed flame are of higher quality. In summary, for the same R ratio, the amount of H2O and CO2 increases with increasing distance from the torch nozzle. The amount of CO, O2 entrainment, and H2 reaches a maximum at the 12 mm axial position. The amount of C2H2 and CH4 decreases with increasing distance from the nozzle. And finally, no OH is observed at any position and R ratio. At the same position in the flame, the amount of CO is unchanged with increasing R ratio. The amount of CO*, H2O, CO2, tend to increase with increasing R ratio. On the other hand, the amount of H2 and O2 entrainment decreases with increasing R ratio. Table 6.11-1 summarizes the observations for the open and enclosed flame.

98

Table 6.11-1 Summary of the trends observed in the open and enclosed flame.

General Trends Observed in the Flame


Same R ratio Same Distance from torch Increasing Distance from Torch Increasing R Species Open Flame Enclosed Flame Open Flame Enclosed Flame CO2 Increase min at 3mm Same Increase H2O Increase Increase Same Increase O2 entrained Increase Max at 12 mm Decreases Decreases C2H2 Decrease Decrease Decrease Decrease CH4 N/A Decrease N/A Decrease OH Increase N/A Increase N/A CO Decrease Max at 12 mm Decrease Decrease/no change H2 Decrease Max at 12 mm Max at 3mm Decrease

99

Figure 6.1-1: Single beam spectra of two different detectors (nonlinear and linearized). Taken at 50 scans at 1cm-1 resolution. Spectra are of the background.

Figure 6.1-2: Absorbance spectra of the flame at total flow of 2 slm with R=0.90. Two different detectors are shown. Spectra are not shifted for clarity.

100

Figure 6.1-3: Absorbance spectra of the standard and linearized detector in the CO2 region. It was taken through the flame with total flow rate of 2 slm and R=0.90. Spectra are shifted to show clarity.

Figure 6.1-4: Two background spectrum (one with purge in arm and one without) shown here are taken with 2 scans at 1 cm-1 resolution.

101

Figure 6.2-1: Emission spectrum of enclosed flame at Ft = 2 slm, 0 mm position, and R=0.90: Shown with a greybody calculated by (1-transmission spectra) * Blackbody at 450K.

Figure 6.2-2: Transmission spectrum of the enclosed flame at Ft=2 slm, 0mm position, and R=0.90.

102

Figure 6.2-3: Emission spectrum of the flame at Ft=2 slm, 6 mm position, and R=0.90.

Figure 6.2-4: Transmission spectrum of the enclosed flame at 6 mm position and R=0.90.

103

Figure 6.3-1: Emission and transmission enclosed flame spectrum at Ft= 2 slm, 0 mm position, and R=0.90.

Figure 6.3-2: E/T analysis of the enclosed flame for CO at Ft=2 slm, 0 mm position, and R=0.90. Shown above is the matching of the hot CO bands at 1200K.

104

Figure 6.3-3: The enclosed flame at Ft=2 slm, 0 mm position, and R=0.90 showing that the hot CO bands indicates a temperature higher than 1200K. Simulated spectrum (top) does not match the experimental shape (bottom).

Figure 6.3-4: The emission spectrum of the flame at Ft=2 slm, 0 mm position, and R=0.90 compared to calculated spectrum of 0.05 atm cm in 1 atm CO at 2500K using HITRAN96 Voigt.

105

Figure 6.4-1: Enclosed flame at Ft=2 slm, 0 mm position, and R=0.90 showing the presence of hot CO2 bands. CO2 calculation parameters: 0.01 atm cm in 1 atm, 600K, HITRAN92, Lorentzian line shape. Spectra shifted for clarity.

Figure 6.4-2: Enclosed flame at Ft=2 slm, 0 mm position, and R=0.90 showing the approximate concentration of CO2. Calculated CO2: 0.003 atm cm in 1 atm, 2000K, HITRAN92 Voigt lineshape. Spectra shifted for clarity.

106

Figure 6.6-1: Experimental (Ft=2 slm, 0 mm position, R=0.90) and calculated OH (0.0032 atm cm in 1 atm at 3000K HITRAN96 Voigt) spectra. No OH is present in the flame at 0 mm position R=0.90. Experimental spectrum shifted for clarity.

Figure 6.6-2: Enclosed flame at Ft=2 slm, 18 mm position ,and R=0.90 shows no observable OH peaks. Calculation parameters for OH: 0.0032 atm cm in 1 atm 3000K , HITRAN96 Voigt

107

Figure 6.8-1: Methane at two different temperatures showing difference in side peaks. Calculation parameters: 0.01 atm cm in 1 atm, HITRAN92 Voigt at 500K and 800K. Spectra are offset for clarity.

Figure 6.8-2: Enclosed flame at Ft= 2 slm, 0 mm position and R=0.90 in the methane region. Calculated parameters: 0.002 atm in 1 atm 400 K HITRAN92 Voigt. Spectra are offset for clarity.

108

Free Enclosed Flame R = 0.90


0.016 0.450 0.400

Partial Pressure * Path length (atm cm)

0.014

0.012

0.300 0.250 0.200 0.150 0.100 0.050 0.000 0 5 10 15 20

0.010

0.008

0.006

CO concentration (atm cm)

0.350

H2O CO2 C2H2 CO CH4

0.004

0.002

0.000

Distance from Torch (mm)

Figure 6.9-1: Free enclosed flame: concentration versus position at R=0.90. Note: CO uses right hand scale.

Free Enclosed Flame R=0.94


0.025 Partial Pressure * Path length (atm cm) 0.450 0.400 0.020 0.350 0.300 0.015 0.250 0.200 0.150 0.005 0.100 0.050 0.000 0 5 10 15 20 Distance from torch (mm) 0.000 CO concentration (atm cm)

0.010

H2O CO2 C2H2 OH CO

Figure 6.9-2: Free enclosed flame: concentration versus position at R=0.94. Note: CO uses right hand scale.

109

Free Enclosed Flame R=0.98


0.030 Partial Pressure * Path length (atm cm) 0.025 0.020 0.015 0.010 0.005 0.000 0 5 10 15 20 Distance from torch (mm) 0.450 0.350 0.300 0.250 0.200 0.150 0.100 0.050 0.000
CO concentration (atm cm)

0.400

H2O CO2 C2H2 OH CO

Figure 6.9-3: Free enclosed flame: concentration versus position at R=0.98. Note: CO uses right hand scale.

Free Enclosed Flame R=1.04


0.040 Partial Pressure * Path length (atm cm) 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000 0 5 10 15 20 Distance from torch (mm) 0.450 0.400 0.350 0.300 0.250 0.200 0.150 0.100 0.050 0.000
CO concentration (atm cm)

H2O CO2 C2H2 OH CO

Figure 6.9-4: Free enclosed flame: concentration versus position at R=1.04. Note: CO uses right hand scale.

110

Free Enclosed Flame R=1.10


0.060 Partial Pressure * Path length (atm cm) 0.050 0.040 0.030 0.020 0.010 0.000 0 5 10 15 20 Distance from torch (mm) 0.450
CO concentration (atm cm)

0.400 0.350 0.300 0.250 0.200 0.150 0.100 0.050 0.000

H2O CO2 C2H2 OH CO

Figure 6.9-5: Free enclosed flame: concentration versus position at R=1.10. Note: CO uses right hand scale.

Calculated H2- Free Enclosed Flame


0.200 Partial Pressure * path length (atm cm) 0.180 0.160 0.140 0.120 0.100 0.080 0.060 0.040 0.020 0.000 0 5 10 15 20 Distance from torch tip (mm) R=0.90 R=0.94 R=0.98 R=1.04 R=1.10

Figure 6.9-6: Free enclosed flame: H2 calculated versus position

111

O2 entrainment-Free Enclosed Flame


Partial Pressure * Path length (atm cm) 0.035 0.030 0.025 0.020 0.015 0.010 0.005 0.000 0 5 10 distance from torch tip (mm) 15 20 R=0.90 R=0.94 R=0.98 R=1.04 R=1.10

Figure 6.9-7: Free enclosed flame: O2 entrainment calculation versus position

112

7.0 CONCLUSION Fourier transform infrared spectroscopy has been employed to determine species concentrations in an oxyacetylene flame. To obtain quantitative measurements of the oxyacetylene flame using the FTIR, methods to reduce the noise of the flame have been investigated. One method uses a two detector configuration where one detector detects the signal plus noise and the other detects just noise. However, the flame does not emit symmetrically, so this method does not work. A second method uses an aperture to reduce the amount of noise hitting the detector, but the signal to noise ratio does not improve using apertures. Another method that does reduce noise is inverting the torch such that the torch points upward. This method reduces the flame noise because the flow is now in the direction of buoyant forces. Finally, the use of an infrared filter to block out the CO2 and CO emission was also efficient in reducing the noise in the low wavenumber regions. Once the noise of the flame is minimized, emission/transmission FTIR analysis was used in conjunction with the HITRAN database to obtain quantitative information on various species in the flame. The average temperature of the species along with the maximum temperature have been determined. Quantitative analyses of the free oxyacetylene flame are carried out in an open and enclosed environment. Concentrations of species CO, CO2, H2O, OH, CH4, and C2H2 are found, and the amounts of H2 and O2 entrainment to the flame are calculated. We find that the enclosed flame contains much more CO and hydrogen than the open flame. This could be the reason why diamond films grown in the enclosed flame are higher in

113

quality. In addition to the increase amount of hydrogen and CO, the OH disappears and the hydrogen concentration reachs a maximum at 3 mm from the torch nozzle (where diamond deposition occurs). In the case of the open flame, we calculate that the oxygen entrainment increases with increasing distance from the flame. However, for the enclosed flame, the oxygen entrainment somewhat decreases with increasing distance. From the way the oxygen entrainment is calculated, this indicates that there were substantially higher amounts of carbon radical (or atoms) in the enclosed flame, which cannot be detected by the FTIR. These carbon radicals can play a major role in diamond deposition. The detection limits for acetylene, methane, and OH radicals have been determined. We find that acetylene is observed only at the 0 mm axial position, and methane is seen only at the 0 mm axial position (R = 0.90) in the enclosed flame. In addition to gases, particles of soot are also observed in the enclosed flame at the 0 mm axial position. Since this is a carbon rich region, it is not surprising that soot is formed here. The HITRAN database was a very useful tool. Using HITRAN, we are able to calculate quantitatively spectra close to that of the experimental data, making quantitative analysis possible. However, there are some shortcomings for the software. It was found that HITEMP, HITRAN do not correctly calculate the CO2 bands at high temperature, and the acetylene band at 730 cm-1 is overcalculated. Furthermore, the HITEMP CO database does not work well with HITRAN96 because

114

the HITEMP database contains numbers that are too small and causes an overflow in the program. Despite all the quantitative results, there are some limitations to using the FTIR. The exact path length is unknown, making the calculation of the species partial pressures impossible. In addition, since this is a line of sight measurement, the concentrations reported here are average and not point concentrations. Furthermore, with the FTIR method, it is not possible to analyze for carbon radicals in the flame. There may be substantial amounts of carbon radicals in the enclosed flame, as seen in the oxygen entrainment calculations.

8.0 FUTURE WORK In order to achieve a better understanding of how the flame varies with a substrate in place depositing diamond, additional experiments are needed. First, the FTIR should be used to make measurements of the flame with the substrate in place while depositing diamond films. The gas phase composition of the flame may change when diamond films are depositing due to further reactions on the surface of the substrate. The films should then be analyzed for its growth rate and quality in order to correlate diamond growth to species concentration. Secondly, the FTIR should be coupled with the cascade arc so that a higher intensity IR source can be used. This higher intensity source will enable an aperture to be placed in between the substrate and the IR source to gain spatial resolution. The 3 mm spot size used in these experiments is too large since the deposition distance is less than 1 mm from the inner cone. Therefore, it is necessary to obtain spatial resolution of less than 1 mm. If the cascade arc is used, it will be possible to probe the boundary layer. Another idea for future work is to use a faster FTIR so that the Fourier frequencies are greater than the flickering frequencies. This will aid in reducing the noise in the spectrum. With the use of the cascade arc, apertures, and faster FTIR, the signal to noise ratio and spatial resolution will substantially increase. Lastly, tomography should be considered for spatial resolution. Also, some of the problems encountered with CO in the enclosed flame apparatus may be minimized because the location of the cold CO will be more apparent.

115

116

Chapter One References


[1] [2] [3] J. C. Angus and C. C. Hayman, "Low-pressure, metastable growth of diamond and diamondlike phases," Science 241, 913 (1988). D. S. Knight and W. B. White, "Characterization of diamond films by Raman spectroscopy," J. Mater. Res. 4, 385 (1989). D. B. Oakes and J. E. Butler, "Diamond synthesis in oxygen-acetylene flames: Inhomogeneities and the effects of hydrogen addition," J. Appl. Phys. 69, 2602 (1991). L. M. Hanssen, K. A. Snail, W. A. Carrington, J. E. Butler, S. Kellogg, and D. B. Oakes, "Diamond and non-diamond carbon synthesis in an oxygenacetylene flame," Thin Solid Films 196, 271 (1991). J. C. Angus, Y. Wang, and M. Sunkara, "Metastable growth of diamond and diamondlike phases," Ann. Rev. Mat. Sci. 21, 221 (1991). J. C. Angus, M. Sunkara, S. R. Sahaida, and J. T. Glass, "Twinning and faceting in early stages of diamond growth by chemical vapor deposition," J. Mater. Res. 7, 3001 (1992). J.-S. Lee and K.-S. Liu, "Deposition of diamond films on SiO2 surfaces using a high power microwave enhanced chemical vapor deposition process," J. Appl. Phys. 81, 486 (1996). R. Locher, C. Wild, N. Herres, D. Behr, and P. Koidl, "Nitrogen Stabilized (100) Texture In Chemical-Vapor-Deposited Diamond Films," Appl. Phys. Lett. 65(1), 34 (1994). M. Kamo, Y. Sato, S. Matsumoto, and S. Setaka, "," J. Cryst. Growth 62, 642 (1983). Y. Matsui, H. Yabe, T. Sugimoto, and Y. Hirose, "The structure of acetylene flames for diamond synthesis," Diam. Related Mater. 1, 19 (1991). Y. Hirose and N. Kondo, Program and Book of Abstracts, Japan Applied Physics, 1988 Spring Meeting (Japanese Physical Society, Tokyo, 1988), p. 434. L. A. Zeatoun and P. W. Morrison Jr., "Optimizing diamond growth for an atmospheric oxyacetylene torch," J. Mater. Res. 12, 1237 (1997).

[4]

[5] [6]

[7]

[8]

[9] [10] [11]

[12]

117

[13]

K. Doverspike, J. E. Butler, and J. A. Freitas Jr., "Deposition of flame grown diamond films in a controlled atmosphere" in Materials Research Society Symposium Proceedings: Wide Band Gap Semiconductors, edited by T. D. Moustakas, J. I. Pankove, and Y. Hamakawa (Materials Research Society, Pittsburgh, PA, 1992), Vol. 242, p. 37. J. J. Schermer, J. E. M. Hogenkamp, G. C. J. Otter, G. Janssen, W. J. P. van Enckevort, and L. J. Giling, "Controlled deposition of diamond from an acetylene-oxygen combustion flame," Diam. Related Mater. 2, 1149 (1993). Y. Hirose and S. Amanuma, "The synthesis of high quality diamond in combustion flames," J. Appl. Phys. 68, 6401 (1990). P. W. Morrison Jr., A. Somashekhar, J. T. Glass, and J. T. Prater, "Growth of diamond films using an enclosed combustion flame," J. Appl. Phys. 78, 4144 (1995). P. W. Morrison Jr., J. E. Cosgrove, J. R. Markham, and P. R. Solomon, "In-Situ IR Spectroscopy of an Oxygen-Acetylene Flame During Diamond Film Growth," Carbon 28, 767 (1990). P. W. Morrison Jr., J. E. Cosgrove, J. R. Markham, and P. R. Solomon, "Tomographic reconstruction of FT-IR measurements of an oxygenacetylene flame during diamond film growth" in MRS International Conference Proceedings Series: New Diamond Science and Technology (Proceedings of the Second International Conference on New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 219.

[14]

[15] [16]

[17]

[18]

Chapter Two References


[1] P. W. Morrison Jr. and J. T. Glass, "Flame characteristics for combustion growth of diamond" in EMIS Datareview Series: Properties and Growth of Diamond, edited by G. Davies (Institution of Electrical Engineers, London, UK, 1994), Vol. 9, p. 368. Y. Hirose and S. Amanuma, "The synthesis of high quality diamond in combustion flames," J. Appl. Phys. 68, 6401 (1990).

[2]

118

[3]

R.-B. Wang, M. Sommer, and F. W. Smith, "Deposition of diamond films via the oxyacetylene torch: experimental results and thermodynamic predictions," J. Cryst. Growth 119, 271 (1992). L. M. Hanssen, K. A. Snail, W. A. Carrington, J. E. Butler, S. Kellogg, and D. B. Oakes, "Diamond and non-diamond carbon synthesis in an oxygenacetylene flame," Thin Solid Films 196, 271 (1991). J. J. Schermer, J. E. M. Hogenkamp, G. C. J. Otter, G. Janssen, W. J. P. van Enckevort, and L. J. Giling, "Controlled deposition of diamond from an acetylene-oxygen combustion flame," Diam. Related Mater. 2, 1149 (1993). K. Bang, A. J. Ghajar, and R. Komanduri, "The effect of substrate surface temperature on the morphology and quality of diamond films produced by the oxyacetylene combustion flame," Thin Solid Films 238, 172 (1994). L. A. Zeatoun and P. W. Morrison Jr., "Optimizing diamond growth for an atmospheric oxyacetylene torch," J. Mater. Res. 12, 1237 (1997). S. Marinkovic and S. Zec, "Formation of large diamond crystals in oxyacetylene flame chemical vapour deposition," J. Mat. Sci. 31, 5999 (1996). W. Zhu, B. H. Tan, J. Ahn, and H. S. Tan, "Studies of diamond films/crystals synthesized by oxyacetylene combustion flame technique," J. Mat. Sci. 30, 2130 (1995). D. W. Zhang, Z. J. Liu, Y. Z. Wan, and J. T. Wang, "Phase Diagram for diamond growth in atmospheric oxyacetylene flames," J. Appl. Phys. 66, 49 (1998). K. A. Snail, R. G. Vardiman, J. P. Estrera, J. W. Glesener, C. Merzbacher, C. J. Craigie, C. M. Marks, R. Glosser, and J. A. Freitas Jr., "Diamond growth in turbulent oxygen-acetylene flames," J. Appl. Phys. 74, 7561 (1993). R. Phillips, J. Wei, and Y. Tzeng, "High quality flame-deposited diamond films for IR optical windows," Thin Solid Films 212, 30 (1992). M. Murakawa, S. Takeuchi, and Y. Hirose, "An experiment in large area diamond coating using a combustion flame torch in its traversing mode," Surf. Coat. Technol. 43-44, 22 (1990).

[4]

[5]

[6]

[7] [8]

[9]

[10]

[11]

[12] [13]

119

[14] [15]

M. Murakawa and S. Takeuchi, "Diamond coating using multiple flame torches in an atmospheric chamber," Surf. Coat. Technol. 54/55, 403 (1992). K. Doverspike, J. E. Butler, and J. A. Freitas Jr., "Deposition of flame grown diamond films in a controlled atmosphere" in Materials Research Society Symposium Proceedings: Wide Band Gap Semiconductors, edited by T. D. Moustakas, J. I. Pankove, and Y. Hamakawa (Materials Research Society, Pittsburgh, PA, 1992), Vol. 242, p. 37. P. W. Morrison Jr., A. Somashekhar, J. T. Glass, and J. T. Prater, "Growth of diamond films using an enclosed combustion flame," J. Appl. Phys. 78, 4144 (1995). F. R. Sivazlian, J. A. von Windheim, and J. T. Glass, "Diamond growth in an oxy-acetylene flame by an alternating gas ratio technique" in Materials Research Society Symposium Proceedings: Novel Forms of Carbon, edited by C. L. Renschler, J. J. Pouch, and D. M. Cox (Materials Research Society, Pittsburgh, PA, 1992), Vol. 270, p. 329. J. A. von Windheim, F. Sivazlian, M. T. McClure, and J. T. Glass, "Nucleation and growth of diamond using a computer controlled oxy-acetylene torch," Diam. Related Mater. 2, 438 (1993). Y. Matsui, H. Yabe, T. Sugimoto, and Y. Hirose, "The structure of acetylene flames for diamond synthesis," Diam. Related Mater. 1, 19 (1991). R. S. Yalamanchi and K. S. Harshavardham, "Diamond growth in combustion flames," J. Appl. Phys. 68(11), 5941 (1990). Y. Matsui, A. Yuuki, M. Sahara, and Y. Hirose, "Flame structure and diamond growth mechanism of acetylene torch," Jpn. J. Appl. Phys. 28, 1718 (1989). R. J. H. Klein-Douwel, J. J. L. Spaanjaars, and J. J. ter Meulen, "Twodimensional distribution of C2, CH, and OH in a diamond depositing oxyacetylene flame measured by laser induced fluorescence," J. Appl. Phys. 78 (3), 2086 (1995). M. A. Cappelli and P. H. Paul, "An investigation of diamond film deposition in a premixed oxyacetylene flame," J. Appl. Phys. 67 (5), 2596 (1989).

[16]

[17]

[18]

[19] [20] [21]

[22]

[23]

120

[24]

M. D. Welter and K. L. Menningen, "Radical Density measurements in an oxyacetylene torch diamond growth flame," J. Appl. Phys. 82(4), 1900 (1997). C. M. Marks, H. R. Burris, J. L. Grun, and K. A. Snail, "Studies of turbulent oxyacetylene flames used for diamond growth," J. Appl. Phys. 73, 755 (1993). C. A. Wolden, R. F. Davis, and Z. Sitar, "In situ mass spectrometry during diamond chemical vapor deposition using a low pressure flat flame," J. Mater. Res. 12(10), 2733 (1997). P. W. Morrison Jr. and J. R. Haigis, "In situ infrared measurements of film and gas properties during the plasma deposition of amorphous hydrogenated silicon," J. Vac. Sci. Technol. A 11, 490 (1993). P. W. Morrison Jr., R. Raghavan, A. J. Timpone, C. P. Artelt, and S. E. Pratsinis, "in situ Fourier transform infrared characterization of electrically assisted flame synthesis of TiO2 particles," Chem. Mat. 9, 2702 (1997). P. W. Morrison Jr., J. E. Cosgrove, J. R. Markham, and P. R. Solomon, "In-Situ IR Spectroscopy of an Oxygen-Acetylene Flame During Diamond Film Growth," Carbon 28, 767 (1990). P. W. Morrison Jr., J. E. Cosgrove, J. R. Markham, and P. R. Solomon, "Tomographic reconstruction of FT-IR measurements of an oxygenacetylene flame during diamond film growth" in MRS International Conference Proceedings Series: New Diamond Science and Technology (Proceedings of the Second International Conference on New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 219. P. W. Morrison Jr., and O. Taweechokesupsin, Calculation of gas phase spectra for quantitative Fourier transform infrared spectroscopy of chemical vapor deposition, {iJ. Electrochem. Soc.}, {b145}, 3212 (1998).

[25]

[26]

[27]

[28]

[29]

[30]

[31]

Chapter Three References


[1] L. A. Zeatoun and P. W. Morrison Jr., "Optimizing diamond growth for an atmospheric oxyacetylene torch," J. Mater. Res. 12, 1237 (1997).

121

[2]

J. Wang, T. Wang, Z. Zhen, Y. Luo, M. R. Clench, and D. J. Mowthorpe, "The influence of the temperature of blackbody for calibrating FTIR system on the instrument response function," SPECTROSC. LETT. 30, 783 (1997). P. W. Morrison Jr. and J. R. Haigis, "In situ infrared measurements of film and gas properties during the plasma deposition of amorphous hydrogenated silicon," J. Vac. Sci. Technol. A 11, 490 (1993).

[3]

Chapter Four References


[1] S. Elfeky, A. Penninger, L. Karpati, and A. Bereczky, "Noise emission from open turbulent premixed flame," Periodica Polytechnica, Mechanical Engineering 40 (2), 85 (1995). L. A. Zeatoun and P. W. Morrison Jr., "Optimizing diamond growth for an atmospheric oxyacetylene torch," J. Mater. Res. 12, 1237 (1997).

[2]

Chapter Five References


[1] D. Scutaru, L. Rosenmann, and J. Taine, "Approximate Intensities of CO2 Hot Bands at 2.7, 4.3, and 12 micron for High Temperature and Medium Resolution Applications," J. Quant. Spectrosc. Radiat. Transfer 52, 765 (1994). P. W. Morrison Jr., J. E. Cosgrove, J. R. Markham, and P. R. Solomon, "Tomographic reconstruction of FT-IR measurements of an oxygenacetylene flame during diamond film growth" in MRS International Conference Proceedings Series: New Diamond Science and Technology (Proceedings of the Second International Conference on New Diamond Science and Technology, edited by R. Messier, J. T. Glass, J. E. Butler, and R. Roy (Materials Research Society, Pittsburgh, PA, 1991), p. 219. P. W. Morrison Jr., J. E. Cosgrove, J. R. Markham, and P. R. Solomon, "In-Situ IR Spectroscopy of an Oxygen-Acetylene Flame During Diamond Film Growth," Carbon 28, 767 (1990).

[2]

[3]

122

[4]

G. Zhimeng, Y. Sheng, and L. Heyi, "The roles of hydrogen in deposition of diamond film using hydrogen-oxyacetylene combustion flame," J. Mat. Sci. Lett. 14, 244 (1995). Y. Matsui, H. Yabe, T. Sugimoto, and Y. Hirose, "The structure of acetylene flames for diamond synthesis," Diam. Related Mater. 1, 19 (1991). P. F. Jessen and A. G. Gaydon, "Estimation of carbon radical concentrations in fuel rich acetylene-oxygen flames by absorption spectroscopy" in xxx Symposium (International) on Combustion (Combustion Institute, Pittsburgh, PA, 1968), p. 481. R. S. Yalamanchi and K. S. Harshavardham, "Diamond growth in combustion flames," J. Appl. Phys. 68(11), 5941 (1990). Y. Matsui, A. Yuuki, M. Sahara, and Y. Hirose, "Flame structure and diamond growth mechanism of acetylene torch," Jpn. J. Appl. Phys. 28, 1718 (1989). R. J. H. Klein-Douwel, J. J. L. Spaanjaars, and J. J. ter Meulen, "Twodimensional distribution of C2, CH, and OH in a diamond depositing oxyacetylene flame measured by laser induced fluorescence," J. Appl. Phys. 78 (3), 2086 (1995). M. D. Welter and K. L. Menningen, "Radical Density measurements in an oxyacetylene torch diamond growth flame," J. Appl. Phys. 82(4), 1900 (1997).

[5] [6]

[7] [8]

[9]

[10]

Chapter Six Reference


[1] R. J. H. Klein-Douwel, J. J. L. Spaanjaars, and J. J. ter Meulen, "Twodimensional distribution of C2, CH, and OH in a diamond depositing oxyacetylene flame measured by laser induced fluorescence," J. Appl. Phys. 78 (3), 2086 (1995). P. W. Morrison Jr., A. Somashekhar, J. T. Glass, and J. T. Prater, "Growth of diamond films using an enclosed combustion flame," J. Appl. Phys. 78, 4144 (1995).

[2]

123

Appendix I Conv65K.ab Basic Program

' This program converts a high resolution, simulated transmission ' spectrum into a simulated FTIR spectrum at a lower resolution. ' The input spectrum is calculated using the HITRAN-PC program called ' trans.exe at a high resolution (0.1 to 0.01 cm-1 depending on the ' pressure). CONV65K.AB performs an inverse FFT to generate an ' equivalent i-gram for the transmission spectrum and then truncates ' the i-gram to reduce the resolution and performs a triangular ' apodization on the truncated i-gram. CONV65K.AB then performs a ' FFT to convert the apodized i-gram to the simulated FTIR spectrum. ' Copyright 1995,1996 Philip W. Morrison, Jr. ' Case Western Reserve Univ. free if getright() < getleft() then xflip

' makes sure lo --> hi

' check number of points pts = npts #s ' pts = npts of HITRANPC file mm=log(pts)/log(2) xx = mm - int(mm) if xx<>0 then dialogbeg "# of points not = power of 2" : dialogend : stop ' check for correct resolution fi = getffp() : la = getflp() resinit = abs(la-fi)/(pts - 1)

'resolution of HITRANPC file

if abs(resinit* 32/0.4821686 - 1) <= 1e-6 then goto 100 if abs(resinit* 64/0.4821686 - 1) <= 1e-6 then goto 100 if abs(resinit*128/0.4821686 - 1) <= 1e-6 then goto 100 if abs(resinit*256/0.4821686 - 1) <= 1e-6 then goto 100 dialogbeg "Check resolution on HITRANPC spectrum" print "Spectrum resolution is "; resinit; " cm-1" print "Must be 0.0150677, 0.007534 cm-1, or 0.003767 cm-1" print "Probably wrong endpoint values" print " " dialogend stop 100 input "Enter desired cm-1 resolution (1, 2, 4, or 8): resfinal "

if resfinal = 1 then resratio = 1.32/.9643519 : reseff = 1.32 ' 0.9643519 = nominal resolution of 1 cm-1 MB157 ' 1.32 = effective resolution of 1 cm-1 MB157, MCT

124

if resfinal = 2 then resratio = 2.16/1.928674 : reseff = 2.16 ' 1.928674 = nominal resolution of 2 cm-1 MB157 ' 2.16 = effective resolution of 2 cm-1 MB157, DTGS if resfinal > 2 then resratio = 1 : reseff = resfinal ' at >= 4 cm-1 resolution, nominal resolution = effective resolution ' calculate reduction factor resfinal = resfinal/2 'nominal point to point spacing n = int(log(resfinal/resinit)/log(2)) 'nearest power of 2 n = 2^n 'convert transmission to i-gram dpts = pts * 2 ' dpts = pts * 2 dim xxx(dpts) la = pts - 1 : fi = pts + 1 xxx = 0 : xxx(0,la) = #s(#0,#la) ' 1st half = #s xxx(pts) = #s(#la) ' # of data points = pts + 1, ' so need (pts + 1)th point reverse #s xxx(fi,dpts-1) = #s(#0,#(la-1)) ' 2nd half = reversed #s reverse #s : fi = getffp() : la = getflp() 'newspc junk(dpts) : setffp 0,(dpts-1) : junk=xxx : stop dialogbeg "Converting to interferogram" print "Reducing # of points by a factor of "; n print "Desired resolution = "; resfinal*2; " cm-1" print "Effective resolution = "; reseff; " cm-1" print " " dialogend 20 dim complex(2,dpts) complex(0) = xxx 'real part of array complex(1) = 0 'imag part of array free xxx transpose complex 'cfft complex cifft complex transpose complex

' cfft converts igram --> frequency ' cifft converts frequency --> igram

dim igramold(dpts) igramold = complex(0) ' real part of i-gram 'newspc junk(dpts) : setffp 0,(dpts-1) : junk=igramold: stop free complex 'truncate i-gram trpts = dpts/n dim igramnew(trpts) midpt = trpts/2

'resolution increases by factor of n

125

igramnew(0,midpt-1) = igramold(0,midpt-1) igramnew(midpt,trpts-1) = igramold(dpts-midpt,dpts-1) free igramold 'newspc junk(trpts) : setffp 0,(trpts-1) : junk=igramnew: stop '******* apply **perfect triangular** apodization ******** 'dim triangle(trpts) : fillbeg 1 : fillinc -2/trpts 'triangle=fill(triangle) : triangle = abs(triangle)' 'igramnew = igramnew*triangle 'free triangle '************************************************************** '**** apply triangular apodization using effective resolution **** dim triangle(trpts) fillbeg resratio : fillinc (-2/trpts)*resratio triangle=fill(triangle) : triangle = abs(triangle) triangle = triangle - resratio+1 triangle = (triangle > 0)*triangle triangle = - triangle igramnew = igramnew*triangle free triangle 'newspc junk(trpts) : setffp 0,(trpts-1) : junk=igramnew: stop ' ************************************************************* '**** apply cosine apodization using effective resolution ***** 'cospts=int(trpts/(2*resratio)) 'trpts/(2*resratio) = eff. length i-gram 'dim cosapod(trpts), halfapod(cospts) 'cosapod = 0 'will become cosine apodization function 'fillbeg 0 : fillinc 1 'halfapod = fill(halfapod) ' will become effective cosine function 'halfapod = (cos(halfapod*pi_value*2*resratio/trpts) + 1)/2 'cosapod(0,cospts-1) = halfapod 'reverse cosapod 'cosapod(0,cospts-2) = halfapod(1,cospts-1) 'reverse cosapod 'igramnew = igramnew*cosapod 'free cosapod, halfapod ' ******************************** dim complex(2,trpts) complex(0) = igramnew complex(1) = 0 free igramnew transpose complex cfft complex 'cifft complex transpose complex dim specnew(trpts)

'real part of array 'imag part of array

' cfft converts igram --> frequency ' cifft converts frequency --> igram

126

specnew = complex(0)/n magnitude newspc Repart(trpts/2) Repart = specnew(0,trpts/2-1) la = la - ((la-fi)/(pts-1)*(n-1)) endpoints setffp fi,la setxtype 1 : setytype 128 dim rsln(8) string rsln(0,7) = resfinal*2 resolution string $rsln(1,7) = " cm-1" for i = 0 to 7 j = -125 - i portout j,rsln(i) next free rsln

'corrects for change in

'real part of complex 'correction for loss of 'during truncation

'resfinal*2 = nominal

'set resolution in new file

for i = 0 to 4 j = -120 - i portout j,clock(-7+i) next dblog #0, -3, "Apod=Triangular"

'set time stamp in new file

'set apodization in new file

free memo 'put memo in file dim memo(200) 'input "string for memo ", $memo setfile 2 : string $memo,-5 setfile 1 : string $memo,-6 savespc 'specnew = complex(1) 'newspc Impart(trpts/2) 'Impart = specnew(0,trpts/2-1) 'setffp fi,la

'imag part of complex

'newspc power(dpts) 'power = sqrt(Repart*Repart + Impart*Impart) 'setffp fi,la end

You might also like