EPA-600/2-76-039
February 1976
Environmental Protection Technology Series
CHEMISTRY OF FUEL NITROGEN CONVERSION
TO NITROGEN OXIDES IN COMBUSTION
Industrial Environmental Research Laboratory
Office of Research and Development
U.S. Environmental Protection Agency
Research Triangle Park. North Carolina 27711
-------
RESEARCH REPORTING SERIES
Research reports of the Office of Research and Development, U.S. Environmental
Protection Agency, have been grouped into five series. These five broad
categories were established to facilitate further development and application of
environmental technology. Elimination of traditional grouping was consciously
planned to foster technology transfer and a maximum interface in related fields.
The five series are:
1. Environmental Health Effects Research
2. Environmental Protection Technology
3. Ecological Research
4. Environmental Monitoring
5. Socioeconomic Environmental Studies
This report has been assigned to the ENVIRONMENTAL PROTECTION
TECHNOLOGY series. This series describes research performed to develop and
demonstrate instrumentation, equipment, and methodology to repair or prevent
environmental degradation from point and non-point sources of pollution. This
work provides the new or improved technology required for the control and
treatment of pollution sources to meet environmental quality standards.
EPA RE VIEW NOTICE
This report has been reviewed by the U.S. Environmental
Protection Agency, and approved for publication. Approval
does not signify that the contents necessarily reflect the
views and policy of the Agency, nor does mention of trade
names or commercial products constitute endorsement or
recommendation for use.
This document is available to the public through the National Technical Informa-
tion Service, Springfield, Virginia 22161.
-------
EPA-600/2-76-039
February 1976
CHEMISTRY OF FUEL NITROGEN CONVERSION
TO NITROGEN OXIDES IN COMBUSTION
by
A. E. Axworthy, G. R. Schneider,
M. D. Shuman, and V. H. Dayan
Rocketdyne Division
Rockwell International
6633 Canoga Avenue
Canoga Park, California 91304
Contract No. 68-02-0635
ROAPs No. 21ADG-08/21BCC-12
Program Element No. 1AB014
EPA Project Officer: G. Blair Martin
Industrial Environmental Research Laboratory
Office of Energy, Minerals, and Industry
Research Triangle Park, NC 27711.
Prepared for
U.S. ENVIRONMENTAL PROTECTION AGENCY
Office of Research and Development
Washington, DC 20460
-------
ABSTRACT
The report gives results of an experimental and analytical investiga-
tion of chemical mechanisms involved in the conversion of fuel nitro-
gen to NOX in combustion. The pyrolysis of fossil fuels and model .
fuel nitrogen compounds was investigated, droplet and particle combus-
tion models were developed, and premixed flat-flame burner experiments
were conducted to study the conversion of HCN and NH_ to NOX in low-
pressure CH.-O^-Ar flames. Decomposition rates and products were mea-
sured in helium from 850 to 1100 C for pyridine, benzonitrile, quino-
line, and pyrrole; products were measured for six No. 6 fuel oils, one
crude oil, and two coals. HCN was the major nitrogen-containing pyrol-
ysis product: the amount formed increased with temperature. NH_ was
a minor product and little if any N« was formed. The burner experi-
ments demonstrated that fuel NO forms relatively slowly above the
luminous zone in the same region where CO is oxidized to CO- or later.
Although HCN and NH gave similar yields of NO, the NH reacted very
early in the flame front; most of the HCN survived the luminous zone
and then reacted slowly. A mechanism was proposed in which fuel NO
forms via the reaction: 0 + NCO = NO + CO.
111.
-------
CONTENTS
Summary • 1
Phase IA - Theoretical Analysis 1
Phase IB - Fuel and Model Compound Decomposition Studies 3
Phase II - Burner Studies of Fuel NOX Formation 7
Introduction 17
Phase IA - Theoretical Analysis 19
Chemical Structures of Fuel Nitrogen Compounds 20
General Discussion of Fuel NOX Formation 25
Kinetics and Mechanisms of Pyrolysis of Organic Nitrogen Compounds 29
Analysis of N-Compound Combustion 34
Combustion Models 49
Phase IB - Fuel and Model Compound Decomposition Studies 95
Introduction 95
Phase IB: Experimental 98
Results and Discussion - Model Compound Pyrolysis 109
Results and Discussion - Inert Pyrolysis of Fuels 148
Phase II - Burner Studies of Fuel NOX Formation 161
Phase II: Experimental 161
Results and Discussion - Screening Experiments 177
Conclusions - Screening Experiments 199
Results and Discussion - Detailed Probing Experiments 200
Analysis of Results of Flame-Probing Experiments 267
Conclusions From Combustion Experiments 283
Appendix A
Chemical Analysis Techniques 285
Appendix B
Coal and Residue Analysis 301
Appendix C
Modified ASTM Method D2036-72, Hydrogen Cyanide in a Helium
Stream Containing Hydrogen Sulfide 303
-------
CONTENTS (Concluded)
Appendix D
Phenate Method for NH3 Measurement-Standard Methods for the
Examination of Water and Wastewater 307
Appendix E
Derivation of NO Oxidation Equation 311
Appendix F
Thermocouple Radiation Correction 313
Appendix G
Flame-Probing Data 315
Appendix H
Derivation of One-Dimensional Flow Equation Including Diffusion
and Chemical Reaction 335
Appendix I
Application of Models to Predict Combustion Behavior 337
Nomenclature 345
References 349
VI
-------
ILLUSTRATIONS
No. Page
1 Comparison of Model Compound Decomposition Rates in Helium 5
2 Comparison of NO Flux Yield Profiles at Various Conditions 10
3 Distribution of Nitrogen in Wilmington Crude Oil 21
4 Representative Types of Basic Nitrogen Compounds Found in
Petroleum 23
5 Representative Types of Nonbasic Nitrogen Compounds Found in
Petroleum 24
6 Potential Paths for NO Formation in Fossil Fuel Combustion 28
7 Arrhenius Plots of Pyrolysis Rates for Picolines and
Pyridine Obtained by Hurd and Simon 31
8 Equilibrium Products as a Function of Equivalence Ratio for
CH1.6~Air Flame at 2100 K and 1 Atmosphere 43
9 Equilibrium Products as a Function of Equivalence Ratio for
CH1>6-Air Flame at 1600 K and 1 Atmosphere 44
10 Droplet Vaporization Model 52
11 Droplet Mass and Composition 59
12 Droplet Temperature 60
13 Schematic of Reactant Diffusion and Reaction in a Flame Zone
Surrounding a Fuel Droplet 61
14 Droplet Temperature 69
15 Vaporization Rate, Flame-Front Model 70
16 Critical Pressure for Hydrocarbons 71
17 Normal Boiling Point for Hydrocarbons 72
18 Schematic of Particle Burning Process 76
19 Volatile Matter Content 83
20 Fixed Carbon Content 83
21 Variation of Composition of Solid Matter With Degree of
Burnout 84
22 Kinetic/Diffusion Model 86
23 Preliminary Kinetic Diffusion Model Results With Flame-Front
Model Input 92
VII
-------
ILLUSTRATIONS (Continued)
No. Page
24 Schematic of Apparatus • 99
25 Model Compound Apparatus 102
26 Model Compound Apparatus 103
27 Fuel Pyrolysis Reactor 107
28 Model Compound Reactor, Temperature Profile, and Residence Time 111
29 Rate Data for Pyridine Pyrolysis in Helium 115
30 Pyridine Pyrolysis in Helium, Comparison of Experimental Results
With First-Order Rate Expressions 117
31 Pyridine Pyrolysis in Helium, Calculated Percent Decomposed as a
Function of Distance and Temperature 119
32 Pyridine Pyrolysis in Helium, Arrhenius Plot of Calculated
Decomposition Half-Life 120
33 Rate Data for Model Compound Pyrolysis in Helium, Liquid
Sample Injection 123
34 Pyridine Pyrolysis Results Using Vapor Injector 127
35 Effect of Oxygen on Decomposition Rate of Pyridine 128
36 Effect of Oxygen on Decomposition Rate of Benzonitrile 129
37 Effect of Oxygen Concentration on Rate of Pyridine Decomposition 130
38 Effect of Oxygen Concentration on Rate of Benzonitrile Decomposition 131
39 Organic Products of Inert Pyrolysis or Pyridine 134
40 Organic Products of Inert Pyrolysis of Quinoline 135
41 Organic Products of Inert Pyrolysis of Benzonitrile 136
42 Organic Products of Inert Pyrolysis of Pyrrole 137
43 Low-Pressure Flat-Flame Burner Apparatus 162
44 Flat-Flame Burner 164
45 Schematic of Chemiluminescent Analyzer Calibration System 172
46 Typical Calibration Curves for Molybdenum Converter at 800 C 175
47 Conversion of NHj to NOX as a Function of Equivalence Ratio, <(>,
and Flowrate in Screening Experiments at a Pressure of 76 torr 182
48 Effect of Argon Dilution on Temperature at Sampling Point 185
Vlll
-------
ILLUSTRATIONS (Continued)
No. Page
49 Effect of Argon Dilution on NOX Yield ' 186
50 Comparison of NOX Yield from NH3 and Temperature for Argon
Dilution Screening Experiments 187
51 Effect of Sampling Distance on NOX Yield From NH3 at a Pressure
of 76 torr 190
52 Effect of Pressure on the NOX Yield From NH3 193
53 Conversion of Additive to NOX as a Function of Equivalence
Ratio, , in Screening Experiments 195
54 Effect of Pressure on NOX Yield from HCN 198
55 Comparison of Mole Fraction and Flux Curves, Flame 1, NH3
Addition With = 0.8 207
56 Flame Temperature vs Time, Flame 1, NH3 Addition With = 0.8 211
57 Species Mole Fraction vs Time, Flame 1, NH3 Addition With =
0.8 212
58 Species Flux vs Time, Flame 1, NH3 Addition With.4> = 0.8 213
59 Species Reaction Rate vs Time, Flame 1, NH3 Addition With
= 0.8 215
60 Flame Temperature vs Time, Flame 2, HCN Addition With <}> = 0.8 219
61 Species Mole Fraction vs Time, Flame 2, HCN Addition With <|> = 0.8 220
62 Species Flux vs Time, Flame 2, HCN Addition With <(> = 0.8 221
63 Species'Reaction Rate vs Time, Flame 2, HCN Addition With
= 0.8 222
64 Flame Temperature vs Time, Flame 3, HCN Addition With = 1.5 225
65 Species Mole Fraction vs Time, Flame 3, HCN Addition With = 1.5 226
66 Species Flux vs Time, Flame 3, HCN Addition With <(> = 1.5 227
67 Species Reaction Rate vs Time, Flame 3, HCN Addition With $ = 1.5 228
68 Flame Temperature vs Time, Flame 4, NH3 Addition-With = 1.5 231
69 Species Mole Fraction vs Time, Flame 4, NH3 Addition With = 1.5 232
70 Species Flux vs Time, Flame 4, NH3 Addition With 4> = 1.5 233
71 Species Flux vs Time, Flame 4, NH3 Addition With = 1.5 234
IX
-------
ILLUSTRATIONS (Concluded)
No. Page
"" **'.
72 Species Reaction Rate vs Time, Flame 4, NH3 Addition With =-1.5 235
73 Flame Temperature vs Time, Flame 5, HCN and Argon Addition With
(J> = 0.8 240
74 Species Mole Fraction vs Time, Flame 5, HCN and Argon Addition
With = 0.8 241
75 Species Flux vs Time, Flame 5, HCN and Argon Addition With
= 0.8 242
76 Species Reaction Rate vs Time, Flame 5, HCN and Argon Addition
With = 0.8 243
77 Flame Temperature vs Time, Flame 6, HCN and NO Addition With
= 0.8 246
78 Species Mole Fraction ys Time, Flame 6, HCN and NO Addition With
$ = 0.8 t 247
79 Species Flux vs Time, Flame 6, HCN and NO Addition With = 0.8 248
80 Species Reaction Rate vs Time, Flame 6, HCN and NO Addition With
= 0.8 249
81 Flame Temperature vs Time, Flame 7, HCN and NO Addition With
4> = 1.5 252
82 Species Mole Fraction vs Time, Flame 7, HCN and NO Addition With
= 1.5 . 253
83 Species Flux vs Time, Flame 7, HCN and NO Addition With = 1.5 254
84 Species Reaction Rate vs Time, Flame 7, HCN and NO Addition With
= 1.5 255
85 Flame Temperature vs Time, Flame 8, NO Addition With <|> = 1.5 257
86 Species Mole Fraction vs Time, Flame 8, NO Addition With = 1.5 259
88 Species Reaction Rate vs Time, Flame 8, NO Addition With = 1.5 260
89 Comparison of NO Flux Profiles for the Fuel-Rich Flames 3 and 8 . 262
90 Comparison of Flame 7 NO Flux Profile With Sum of NO Fluxes
From Flames 3 and 8 263
91 Species Mole Fractions in C-H.-O.-Ar Flame at Atmospheric
Pressure, NO Added Initially at 265 ppm, = 1.5 265
-------
TABLES
No. Page
i °
1 Comparison of Results of HCN and NH3 Addition • . 13
2 Types of Elementary Reactions That May Be Involved in Fuel NOX
Formation 47
3 Estimated Properties for Fuel Grade No. 2 74
4 Hydrocarbon and Thermal NO Reactions 91
5 Reactions for Fuel Nitrogen 93
6 Fuel Samples 98
7 Assigned Wall Temperature Profile for Model Compound Reactor 110
8 Experimental Data on Rate of Pyridine Decomposition in Helium 114
9 Experimental Data on Rate of Model Compound Decomposition in
Helium 122
10 Kinetic Parameters for Pyrolysis of Model Compounds in Helium 124
11 Experimental Data on Rate of Pyridine Decomposition in Helium 124
12 HCN Formation in Inert Pyrolysis of Model Compounds 139
13 Nitrogen Balance in Inert Pyrolysis of Pyridine 143
14 Product Mass Balances From Inert Pyrolysis 144
15 Residue Calculations 145
16 Summary of Experimental Data on HCN Formation From Inert
Pyrolysis of Oils 150
17 Comparison of HCN Formation From Various Oils Under Inert
Pyrolysis Conditions 151
18 Results of Measurements of N2 and NHg From Inert Pyrolysis of Oils 153
19 Summary of Percent Fuel-N Converted to NH, + N9 in Oil Pyrolysis 155
O £
20 Summary of Micrograms N2 per mg Oil From Oil Pyrolysis 155
21 Summary of NOX Measurements in Screening Experiments With
. Ammonia Additive 180
22 Effect of Argon Dilution on Oxygen Concentration of Reactant
Mixture and Combustion Products 188
23 Summary of Pressure Effect Screening Experiments With NHj
Additive 192
XI
-------
TABLES (Concluded)
No.
24 Summary of NOX Measurements in Screening Experiments With HCN
Additive 194
25 Summary of Pressure Effect Screening Experiments, NOX Yield With
HCN Additive 197
26 Conditions for Probing Experiments 201
27 Temperature Profiles 209
28 Conditions at Point of Maximum NO Formation Rate in Post-Flame
Gases 270
29 Reaction Scheme for Fuel NO Formation From Ammonia and Hydrogen
Cyanide 277
XII
-------
SUMMARY
The study reported here was carried out to investigate further the chemical
mechanisms involved in fuel NO formation to obtain information that might aid
A
in the development of new approaches for mitigating this problem. The program
was conducted under the following three tasks:
Phase IA - Theoretical Analyses
Phase IB - Fuel and Model Compound Decomposition Studies
Phase II - Burner Studies of Fuel NO Formation
A.
PHASE IA - THEORETICAL ANALYSIS
A theoretical study of fuel NO formation was conducted under this task to
A>
complement and to guide the experimental program. The objective of this analysis
was to identify potential areas for experimental studies and to aid in data
interpretation and development. The literature on nitrogen compounds in fossil
fuels was surveyed to determine the chemical structures of the most common fuel
nitrogen compounds.
The general reaction paths and physical processes most likely to be involved in
fuel NO formation were then considered. The remainder of the qualitative
analysis of fuel NO formation was divided into two areas: (1) consideration
A
of the pyrolysis (preflame) type reactions that the volatile fuel nitrogen com-
pounds will undergo (near the surface of the droplet or particle) before
approaching the flame front, and (2) consideration of the combustion reactions
of fuel nitrogen compounds and their reaction products. These areas relate,
respectively, to the Phase IB pyrolysis experiments and the Phase II burner
experiments.
-------
The major effort under this theoretical analysis task was the development of a
mathematical kinetic-diffusion model for the combustion of oil droplets and coal
particles. Such a model was required to establish the physical and chemical
conditions that will control the formation of fuel NO formation. In addition,
•A.
the model is prepared to accept chemical rate data, as they become available,
for the various reactions of fuel nitrogen compounds and their products. Thus,
the model will be useful in testing various possible fuel NO formation mech-
X
anisms under heterogeneous combustion conditions.
Heterogeneous Combustion Modeling
Models development under this program include a droplet vaporization model, a
droplet flame front model, a coal combustion model, and an average film kinetic/
diffusion model. Each of these models has essentially the same current status,
i.e., computer subprograms have been written, debugged, and checked out to ensure
their operability in the very simplest of flow situations, which is the insertion
of a single particle with prescribed initial conditions into a gas stream whose
conditions are known over the length of the flow. They are ready to be incorpo-
rated as subprograms in a decoupled computerized combustion model. The role of
the subprograms is to calculate spatial production of gaseous species that are
transferred from a size-distributed condensed phase to the gas flowfield.
The computer program to model the heating and vaporization of multicomponent
fuel droplets surrounded by a specified gas flowfield does not include any
reactions of the fuel vapor components with the surrounding gas. The model
accounts for changes in droplet density, latent heat of vaporization, vapor
pressure, and vapor thermal and transport properties that arise because of the
more rapid gasification of the more volatile droplet components during the
course of vaporization. The computer program produces descriptions of the
droplet vaporization rate, the average film (boundary layer) surrounding the
droplet, the droplet temperature, liquid composition, and droplet diameter from
ignition to final burnout.
-------
The droplet flame front model is for the heating and combustion of multicomponent
fuel droplets surrounded by a flame front and gives predictions for the combus-
tion rate, droplet heating rate, and the radial location and.temperature of the
flame front under convective conditions assuming infinite rate fuel kinetics. The
computer program produces descriptions similar to the droplet vaporization model
with particular attention given to the average film thickness, diffusion rates
through the film, vapor residence times within the film, and film temperature and
composition profiles.
The coal combustion model comprises a complementary set of computer programs for
analyzing the reactions attending combustion of a single condensed fuel particle
including particle devolatilization and heterogeneous combustion at the particle
surface. Calculated results include: local transient changes in each particle's
bulk weight, temperature, and composition; radial temperature and major reaction
species concentrations through the film surrounding the particle: and production
rates for combustion products.
The kinetic/diffusion model includes the kinetics and diffusion in the droplet
film of all compounds and intermediate species. Specifically, it accounts for
quasi-global finite rate combustion of mixtures of hydrocarbon compounds, some
of which may contain fuel nitrogen, and includes the production rates of fuel
NO , thermal NO , other pollutants, and combustion products. The model solves
J\ A.
simultaneously the energy equation, species continuity, and species diffusion
equations using boundary conditions of the droplet surface, and the temperature
and molar concentrations in the bulk gas stream.
PHASE IB - FUEL AND MODEL COMPOUND DECOMPOSITION STUDIES
The objective of this task was to conduct thermal decompositon studies at
elevated temperatures to investigate the fate of nitrogen bound in fossil fuels
(fuel-N) under conditions relevant to those existing in the initial phases of
the combustion process. Pyrolysis experiments were conducted with model fuel
nitrogen compounds to measure the kinetic parameters that determine under what
-------
conditions (i.e., at what point in the flame) typical fuel nitrogen structures
will decompose, and to identify the nitrogen-containing species that are formed.
Later in the program, fuel oils and coals were pyrolyzed under conditions
similar to those employed with the model compounds.
As representative of compounds having the common fuel-N structures, the model
compounds pyridine, quinoline, pyrrole, and benzonitrile were chosen for study.
After the major features of model compound pyrolysis had been established, fuel
oils and coals were pyrolyzed under similar inert pyrolysis conditions, and the
nitrogen-containing inorganic products were measured and compared with those
formed from the model compounds.
Model Compound Pyrolysis
The model compound study involved the measurement under inert pyrolysis condi-
tions of the decomposition rates, the nitrogen-containing inorganic products,
and the major organic products. The oxidative pyrolysis of pyridine and
benzonitrile was also investigated. The pyrolysis reactions were studied over
the temperature range of 850 to 1100 C in a quartz flow reactor with an ID of
2 mm, a volume of 1 cc, a nominal residence time of ^ second. and a sample size
of 0.2 (JL H of liquid. Organic decomposition products were measured by
temperature-programmed GC, and the inorganic products (NH,, hL and HCN) were
measured (at the microgram level) by a combination of wet-chemical and GC
techniques.
Pyridine and pyrrole gave similarly shaped inert decomposition curves on plots of
percent undecomposed versus reactor temperature under conditions of constant mass
flowrate, Fig. 1, with slopes that remained steep until beyond 95-percent decom-
position. The pyrrole is less stable than pyridine, the 50-percent decomposition
temperature being lower by about 60 degrees C. Quinoline gave a decomposition
curve that was nearly linear with temperature. Quinoline is less stable than all
of the other compounds below 910 C, but is more stable than pyrrole above that
-------
CO
O
a.
o
o
LU
O
z
o
cc
100
90
80
70
60
50
30
20
10
PYRROLE
PYRIDINE
QU INCLINE
BENZONITRILE
P = 16 PSIG
FLOWRATE = A2.9 CC/MIN
DIAMETER = 2.2 MM ID
850
900
950 1000
TEMPERATURE, C
1050
1100
Figure 1. Comparison of Model Compound Decomposition Rates in Helium
(constant pressure and mass flowrate)
-------
temperature. Benzonitrile is unusual in that its decomposition curve remains
steep up to about 960 C and then tails out at high temperatures. In fact, 3 per-
cent remained undecomposed even at a temperature of 1100 C. Thus,.below 1010 C,
pyridine is more stable under these conditions than benzonitrile while, above
1010 C, the reverse is true.
The model compound decomposition rates were correlated with the following first-
order rate expressions:
Pyridine: k = 3.8 x 1012 exp (-70,000/RT), sec"1
Pyrrole: k = 7.5 x 1015 exp (-85,000/RT), sec'1
Quinoline: k = 2.4 x 108 exp (-45,000/RT), sec"1
Benzonitrile did not fit any simple rate expression and the rate constant for
pyridine was found to be a function of concentration. The extrapolated half-life
for pyridine is 50 microseconds at 1800 K.
Much of the model compound nitrogen was found to form HCN under inert pyrolysis
conditions and most of the remaining nitrogen appeared to be contained in the
carbonaceous residue that formed in the reactor. The amount of HCN formed
increased with temperatures as follows:
Percent N
Compound Temperature, C Converted to HCN
Pyridine 960 40
1105 • 102
Benzonitrile 955 50
1105 81
The complete conversion of pyridine to HCN at 1105 C is a very significant
result. These model compound results suggest that, at the higher temperatures
involved in actual combustion, most of the volatile fuel nitrogen may form HCN
-------
before reaching the flame front. It was found in Phase II of this study that HCN,
once formed, is stable until well into the flame zone. Less than 10 percent of
the model compound nitrogen is converted to NH, under inert pyrolysis conditions
and no N_ forms.
Fuel Pyrolysis
The pyrolysis of fuel oils under similar inert conditions in a flow reactor re-
sulted in 14 to 25 percent conversion of fuel-N to HCN at 950 C and 23 to 42 per-
cent conversion at about 1100 C. A Wilmington crude oil gave somewhat larger
yields of HCN. Interestingly, the two coal samples that were studied gave yields
of HCN similar to those obtained with fuel oils.
The amount of NH_ + N9 formed (per milligram of fuel) was nearly independent of
O £
temperature and fuel-N content. As with the model compounds, NH, was a minor
product.
Because much of the nitrogen in coals and fuel oils remains in the residue, the
volatile fuel nitrogen species may have inorganic product distributions similar
to those of the model compounds. The fuel pyrolysis results definitely indicate
that HCN is a likely important precursor to the formation of fuel NO in
J\.
combustion.
PHASE II - BURNER STUDIES OF FUEL NO FORMATION
X
HCN and NH, were added to premixed CH.-O^-Ar flames to investigate the combustion
kinetics involved in the formation of NO from these combustion intermediates. NO
was added initially in some experiments to determine its fate once formed. The
burner was water-cooled and an uncooled quartz probe and an Al_0,-coated thermo-
couple were used to measure species concentrations and temperature along the center-
line of the burner. The burner was enclosed in a glass envelope so that a pressure
of 0.1 atm could be maintained to spread the flame and permit more detailed probing.
The mole fractions of the major stable species were measured by mass spectrometry.
-------
A chemiluminescent analyzer (CA) measured NO directly and NO , HCN and NH after
z o
conversion of these species to NO. A molybdenum catalyst converter was used at
400 C to convert N02 and at 800 C to partially convert HCN and NH .. In most of
the experiments, the NHj or HCN additive was initially at 2500 ppm molar, based
on the total flowrate of all gases to the burner, and the Ar/02 ratio was the
same as the N /O ratio in air.
Screening Experiments
Before the detailed probing experiments were initiated, a series of screening
experiments was conducted with the probes positioned well'downstream (80 mm above
the burner) to study the effects of various parameters on the overall NO yield,
and to select the best conditions for detailed probing. The overall yields of NO
were nearly the same from HCN and NH_ when the conditions were identical. High
NO yields on the order of 80 percent were obtained in fuel-lean flames but the
yield dropped off continually when the flame was made more and more fuel rich.
Varying the pressure, burner feed rate or Ar/02 ratio usually affected the tem-
perature apparently by changing the position of the flame front and thereby the
rate of heat loss to the burner (argon dilution also decreased the adiabatic flame
temperature and the effects were partially compensating because the lower tempera-
ture and reaction rates moved the flame further from the burner decreasing the
heat loss). The observed NO yields appeared to correlate fairly directly with
temperature, i.e., increasing somewhat with increasing temperature. Changing
the reactant concentrations by varying the pressure or Ar/02 ratio had little
effect on the final NO yield if the effect of the attendant temperature change
was taken into account. This indicates that the reactions that form NO and N2
are of the same order -- undoubtedly second. The rate of NO formation is pre-
sumably a strong function of species concentration eventhough the yield apparently
is not.
-------
Detailed Probing Experiments
Based on the results of the screening tests, detailed probing experiments were
conducted at equivalence ratios, <}>, of 0.8 (fuel-lean) and 1.5. A total of
eight flame conditions were investigated. Flames 1 and 2 involved the addition
of NH_ and HCN, respectively, at = 0.8. HCN and NH, were then added under fuel-
«J 3
rich conditions ( = 1.5) in flames 3 and 4. In flame 5, the Ar/02 ratio was
increased 40 percent at = 0.8. Both HCN and NO (675 ppm) were added to flames 6
(4> = 0.8) and 7 ( = 1.5). In flame 8, NO was added alone (675 ppm) at = 1.5.
The measured profiles of species mole fraction vs distance from the burner are
affected by both the species reaction rate at various positions and the axial
diffusion of the species. The probe data were analyzed by the procedure of
Fristrom and Westenberg that corrects for diffusion permitting the species flux
profiles to be calculated. The flux profiles establish directly the chemical
reaction rates at various points in the flame. The importance of this diffusion
correction in these low-pressure flames is demonstrated by the flame 1 results
where the calculated NO flux is near zero at the top of the luminous zone even
though the NO mole fraction has risen to 1300 ppm at that point. Within the
accuracy of the rather large diffusion correction involved, these results estab-
lish that nearly all of the NO exiting the reactor forms above the luminous zone
as does the NO present in the flame front. The flame-front NO results from up-
stream diffusion caused by a steep NO concentration gradient.
Before summarizing the results from each flame, the NO flux yields obtained for
flames 1 through 4 are compared on a single plot in Fig. 2. The time scales were
adjusted so that the tops of the luminous zones coincide in this figure. In the
fuel-lean flames (<(> = 0.8), the NO forms rapidly and in high yield just above the
luminous zone. With NH_ additive, the NO forms more rapidly than with HCN (reach-
ing its maximum rate of formation just before CO- formation reaches its maximum
in the case of NH and just after with HCN). In fuel-rich flames, the NO forms
in low yield from either additive and most of the NO forms far above the luminous
zone. There is an apparent surge of NO formation just above the top of the
luminous zone.
-------
100
90
80
70
60
50
2 AO
>-
3 30
u_
§ 20
10
0
-10
-20
LUMINOUS ZONE
• 1.5)
NH3> $ = 0.8
1 2
TIME, ( = 0.8), MSEC
Figure 2. Comparison of NO Flux Yield Profiles
at Various Conditions
-------
Flame 1: NH3> (j) = 0.8. In the £uei-lean flame, all of the NH3 reacts below the
top of the luminous zone to form nitrogen species that are not measurable in the
probe-converter-CA system. Less than 20 percent of the NH_ is converted to N?
because the final N_ yield is 18 percent and N2 is not expected to react in these
moderate-temperature flames. The unidentified nitrogen intermediate (probably a
free radial) forms NO rapidly and in high yield just above the luminous zone
(Fig. 2).
Flame 2: HCN, (j> = 0.8. In contrast to NH , most of the added HCN apparently sur-
"" ' - - - I "I •-. - ^
vives the luminous zone in the fuel-lean flame. The "HCN"* reacts rapidly just
above the luminous zone and NO forms in high yield reaching its maximum rate about
0.1 msec after the reaction of HCN reaches it maximum (the rate of CO formation
reaches a maximum between these two maxima). The relatively slow reaction of HCN
appears to be the cause of the NO forming more slowly from HCN than from NH_
(Fig. 2). It can be shown that in this flame much and possibly all of the NO
is formed from an intermediate species that builds up above the luminous zone.
In particular, the maximum rate of NO formation is more than twice the rate of
HCN consumption at the point where the NO maximum rate occurs.
Flame 3: HCN,
-------
reacts slowly at the same rate as does "HCN" when HCN is the additive. Thus, the
mechanism of NO formation in the fuel-rich flame appears to be the same whether
the additive is NH, or HCN and involves the formation of NO from HCN.
o
Comparison of Results. The conclusions drawn from the addition of NH and HCN to
lean and rich methane flames are compared in Table 1. NH reacts much more rapidly
than does HCN in both flames. The similarity of the overall NO yields from NH3
and HCN in the rich flame (Fig. 2) apparently results from the fact that the NO
forms from HCN with either additive. The reason for the similarity of the NO
yields in the lean flame is not as apparent since a common stable intermediate is
not formed. However, a common free radical intermediate could be involved.
Effect of Argon Dilution. An experiment, flame 5, was conducted with HCN addition
at =0.8 in which the Ar/0_ ratio was increased by 40 percent and the total burner
feed rate was reduced to increase the heat loss to the burner. Although these
conditions reduced the maximum flame temperature by 150 degrees and the initial
reactant concentrations by 23 percent, the maximum reaction rates for NO, HCN,
CH. and CO only decreased by 30, 8, 42 and 29 percent, respectively. An analysis
of these results indicate that certain compensating effects control the maximum
reaction rates and that the measured rates do have a normal dependency on tempera-
ture and species concentration.
Effect of NO Addition. When NO was added initially to the fuel-rich flame along
with HCN (flame 7) or alone (flame 8), rapid consumption of NO'occurred just be-
low the top of luminous zone. NO then formed rapidly just above the luminous zone
and decayed rapidly again further downstream. In the absence of added HCN, the
consumed NO was partially converted (40 percent) to a measurable nitrogen species
that was presumably HCN. When NO was added to the fuel-lean flame along with
HCN (flame 6), a small consumption of NO occurred in the luminous zone before the
formation of NO from HCN obscured any further reaction of the added NO.
12
-------
TABLE 1. COMPARISON OF RESULTS OF HCN AND NH3 ADDITION
Additive
4> » 0.8
(High NO Yields)
* = 1.5
(Low NO Yields)
HCN
NH.
65% HCN survives LI*
HCN reacts rapidly above
LZ
Al1 NO forms rapidly
above LZ
NO from intermediate
Al 1 NH-j reacts below
top LZ
NO forms rapidly above
LZ
Al 1 NO forms from
intermediate
70% HCN survives LZ
HCN reacts slowly above LZ
All NO forms slowly above
LZ
NO from intermediate and/or
HCN
All NHj reacts below top LZ
(one-half converted to HCN)
"HCN" reacts slowly above LZ
Most NO forms slowly above
LZ
NO from intermediate and/or
HCN
*LZ denotes luminous zone
13
-------
Reaction Mechanism
A possible reaction mechanism for the formation of NO and N_ from NH is presented.
The experimental results indicate that a rather long-lived nitrogen intermediate
forms in high concentration in the fuel-lean flame. A likely candidate that could
have the required characteristics of (1) low reactivity in the luminous zone and
(2) not forming NO in the probe-converter system is the NCO radical.*
According to the proposed reaction scheme, CN radicals form in fuel-lean flames
from NH_ and (more slowly) from HCN and then react to form NCO
CN + 02 = NCO + 0
CN + OH = NCO + H. '
The NCO radical is postulated to be relatively stable and react rapidly with
oxygen atoms above the luminous zone to form NO
NCO + 0 = NO + CO.
The most likely path for N. formation appears to be the reaction
NCO + NCO = N2 + 2 CO.
The amount of N» that can form via the reverse Zeldovich reaction
N + NO = N2 + 0
is limited by the competing reaction
N + 02 = NO + 0.
The mechanism of NO formation in fuel-rich flames may also involve the NCO radical
but there is little evidence relating to this point. The rapid consumption of NO
in the fuel rich flame front must result from reaction with hydrocarbon radicals.
It can be shown that the reverse Zeldovich reaction
0 + NO = N + 0_
is much too slow to cause the observed consumption of NO. Additional data and
theoretical modeling are required to test this somewhat speculative overall
mechanism for the formation of fuel NO.
•'NCO
could form N2 in the probe via 2 NCO = N2 + 2 CO.
14
-------
Measurements of NCL, made in the NH, addition experiments only, demonstrated that
£, ' O
the maximum N0_ mole fraction was only 8 ppm even though the NH was added at
2500 ppm. The amount of NCL that can form may be limited by the formation rate
of HCL radical, i.e., the HO concentration may become depleted by reaction with
NO to form N0_, preventing the formation of larger amounts of NO .
15
-------
INTRODUCTION
Nitrogen oxides (NO ) are one of the major air pollutants emitted by stationary
Jv
combustion sources.* Much of this NO is formed from the oxidation of N at peak
x 2
combustion temperatures ("thermal NO "), but a significant and probably increasing
J\.
fraction of the emitted NO is being formed from the chemically bound nitrogen
J\.
present in many fossil fuels ("fuel nitrogen") and is referred to as "fuel NOX."
A number of studies have been made of the effects of combustion conditions on the
extent of conversion of fuel nitrogen, and added nitrogen compounds, to NO . The
Jt
theoretical and experimental study reported here was carried out to investigate
further the chemical mechanisms involved in fuel NO formation and to obtain infor-
mation that might aid in the development of new approaches for mitigating this prob-
lem. The major portion of the program was devoted to the experimental studies.
Petroleum crudes produced in this country contain from 0.01 to 0.94 percent nitro-
gen, with California crudes being noted for their high nitrogen contents. Because
fuel nitrogen compounds are thermally stable and of rather low volatility, they
are concentrated in the heavy oil and .asphalt fractions during the refining of
petroleum. Therefore, natural gas, LPG, gasolines, and distillates (including
diesel and No. 2 fuel oils) generally contain little fuel nitrogen. Residual oils,
such as No. 6 fuel oils, however, contain typically from 0.2 to 1.4 percent nitro-
gen depending on the source of the crude. Coals have higher nitrogen contents
(on the average) than petroleum, with American coals ranging from 1.1 to 1.8 per-
cent nitrogen. European soft coals appear to contain even more nitrogen.
Flagan (Ref. 1 ) has estimated recently that in 1968 about one-fourth of all the
oxides of nitrogen emitted into the atmosphere of the United States, as the result
of human activities, was from the oxidation of fuel nitrogen during combustion.
The numbers used in his estimate indicate that nearly one-half of the total NO
A
was from electric utility and industrial combustion sources, and a little more
than one-half of this stationary source NO was fuel NO . Flagan's estimates are
A. J\
*The NO formed in most combustion processes accounts for about 95-percent of the
NO and the remainder is mainly NO-.
17
-------
based, in part, on the assumptions that fossil fuels have an average nitrogen
content of 1 percent, and 20 percent of this is converted, on the average, to NO.
It is likely that the importance of fuel NO has increased since 1968 because
A
some of the combustion techniques that are effective in reducing thermal NO do
J\.
not reduce fuel NO . Furthermore, low nitrogen content fossil fuels are in shorter
Jt
supply.
Most of the information on fuel NO that was available at the start of this program
yv-
is discussed in an excellent review by Sternling and Wendt (Ref. 2) and in Ref. 3
through 5. During the course of this program, four technical symposia were held
that highlighted the subject of fuel NO (Ref. 6 through 9). Several papers were
A
also presented at the 14th and 15th Combustion Symposia (International) relating
to this subject.
The present program was conducted under the following tasks:
Phase IA - Theoretical Analysis
Phase IB - Fuel and Model Compound Decomposition Studies
Phase II - Burner Studies of Fuel NO Formation
x
The Phase IA theoretical study was conducted to complement the experimental tasks.
The major effort under this task was the development of a mathematical kinetic-
diffusion model for droplet and particle combustion that predicts the conditions
that will be encountered by fuel nitrogen compounds during combustion. In Phase IB,
fuel oils, coals, and model fuel nitrogen compounds were pyrolyzed to investigate
the nature of the decomposition reactions which the nitrogen compounds undergo in
the early stages of combustion. Finally, the combustion kinetics of the potential
fuel NO intermediates, HCN and NH~, were investigated in subatmospheric, premixed
A O
CH4-02-Ar flames in the Phase II study. The reduced pressure (0.1 atm) was used
to spread the reaction sufficiently to permit detailed probing of the flame.
18
-------
PHASE IA - THEORETICAL ANALYSIS
A theoretical study of fuel NOX formation was conducted under this task to comple-
ment and to guide the experimental program. The objective of this analysis was to
identify potential areas for experimental studies and to aid in data interpreta-
tion and development.
The literature on nitrogen compounds in fossil fuels was surveyed to determine the
chemical structures of the most common fuel nitrogen compounds. The general reac-
tion paths and physical processes most likely to be involved in fuel NOX formation
were then considered. The remainder of the qualitative analysis of fuel NOX for-
mation was divided into two areas: (1) consideration of the pyrolysis (preflame)
type reactions that the volatile fuel nitrogen compounds will undergo (near the
surface of the droplet or particle) before approaching the flamefront, and (2)
consideration of the combustion reactions of fuel nitrogen compounds and their
reaction products. These areas relate, respectively, to the Phase IB pyrolysis
experiments and the Phase II burner experiments.
The major effort under this theoretical analysis task was the development of a
mathematical kinetic-diffusion model for the combustion of oil droplets and coal
particles. Such a model was required to establish the physical and chemical con-
ditions which will control the formation of fuel NOX formation. The types of in-
formation the program can provide that are directly relevant to this project in-
clude: (1) the temperature-time history of the nitrogen species, (2) the concen-
tration profiles of 02 and oxygen-containing radicals, and (3) the integrated
rates of the various reactions in the extended Zeldovich mechanism for the forma-
tion of thermal NOX (which will also be important in the formation of fuel NOX if
it turns out that N atoms are a principal intermediate in this process). In ad-
dition, the model is prepared to accept chemical rate data, as they become avail-
able, for the various reactions of fuel nitrogen compounds and their products.
Thus, the model will be useful in testing various possible fuel NOX formation
mechanisms under heterogeneous combustion conditions.
19
-------
CHEMICAL STRUCTURES OF FUEL NITROGEN COMPOUNDS
Ball, et al. (Ref. 10) have demonstrated an apparent correlation between the ni-
trogen content of petroleum crudes, the carbon residue of the crude, and the geo-
logical period in which the petroleum was formed. For a given geological period,
the nitrogen content of the crude is approximately proportional to the carbon res-
idue with the ratio of nitrogen to carbon residue increasing in the order: Cre-
taceous period (80 to 150 million years ago), Carboniferous (250 to 300 million
years), and Tertiary (50 to 80 million). Having been formed in the Tertiary
geological period, and having high carbon residues, the California crudes contain
the highest average amounts of nitrogen (up to 0.65 weight percent).
Although the total amount of chemically bound nitrogen that is present in fossil
fuels is quite well established, the identity and distribution of the chemical
types of nitrogen are not. This is particularly true for coals because many of
the coal-nitrogen compounds cannot be extracted from the solid in an unreacted
form for identification and analysis. Most of the information on the structure
of coal nitrogen compounds has been deduced from the nitrogen compounds that are
obtained by extraction and from distillation. It has been established, however,
that much of the nitrogen in heavy oils is in the form of heterocyclic organic
compounds and the indications are that the same is true for coal nitrogen.
No comprehensive review is available, at present, of the literature on the mea-
surement and identification of fuel nitrogen compounds. Although not intended
as a comprehensive literature' survey, some 46 publications relating to this sub-
ject are referenced (Ref. 10 through 55) in this report. Many of these citations
were supplied by W. E. Haines of the Bureau of Mines, Laramie Energy Research
Center, Laramie, Wyoming, and J. E. Haebig, of the Physical Sciences Division,
Gulf Research and Development Company, Pittsburgh, Pennsylvania.
In the early studies of the composition of petroleum, there was little interest
in identification of fuel nitrogen compounds. Interest in these compounds began
to increase about 20 years ago with the advent of catalytic processing of crude
20
-------
oils (Ref. 13). Even in trace amounts, nitrogen compounds cause serious problems
in processing and in the stability of petroleum products. They cause catalyst
poisoning and are involved in the formation of gums, lacquers, and precipitates.
The nitrogen present in petroleum is usually divided into two broad types: basic
nitrogen, defined as that which is titratable with perchloric acid in acetic acid
solution, and nonbasic nitrogen, which is not titratable. The basic nitrogen com-
pounds are classified further according to their degree of basicity (Ref. 25 and
44). On the order of one-third of the nitrogen in petroleum is basic and the re-
mainder is nonbasic according to the above definition (Ref. 25 and47 ). The nitro-
gen compounds are concentrated in the higher-boiling crude oil fractions, with the
nonbasic nitrogen compounds becoming more abundant as the boiling point of the
fraction increases. This is demonstrated for Wilmington crude oil in Fig. 3 ,
which was reproduced from Ref. 25 .
.50
o
a;
o
LLl
.30
.25
z" .20
UJ
CJ
§ .15
H-
5 .10
.05
0
TOTAL
BASIC
150 200 250 300 350 AOO 1*50 500
BOILING POINT, C
Figure 3 . Distribution of Nitrogen in Wilmington Crude Oil
21
-------
The basic fuel nitrogen compounds have been studied more extensively because they
can be separated more easily from petroleum. The basic nitrogen compounds in pe-
troleum are mainly pyridines, pyridans, quinolines, isoquinolines, acridines,
phenanthridines, phenazines, pyrazines, and highly substituted pyrroles. The al-
kylanilines are important constituents of shale oil naptha (Ref. 16). The gen-
eral structures for these basic nitrogen compounds are shown in Fig. 4 . Also
included in Fig. 4 are the.structures of a pyridine and a pyridane that have
been identified in Wilmington Crude (Ref. 15 and 25).
The nonbasic nitrogen compounds found in petroleum were reviewed by Latham et al.
in 1965 (Ref. 36). As shown in Fig. 5 , these include most pyrroles, the indoles,
carbazoles, benzcarbazoles, amides such as quinolones, and benzonitriles. The
amides are of interest because they contain nitrogen and oxygen in the same part
of the molecule and, therefore, have some potential for forming NO directly dur-
ing pyrolysis.
It can be seen from the structures in Fig. 4 and 5 that, in general, the six-
membered ring heterocyclic nitrogen compounds containing a C=N double bond in the
ring (so that there is no hydrogen atom attached to the nitrogen) are basic. In
the five-membered heterocyclic rings, the nitrogen has a hydrogen atom attached
and is nonbasic. Latham et al. (Ref. 36) consider the phenazines (Fig. 4) to
be nonbasic, but Okuno et al. (Ref. 44) classify phenazines as strongly basic
and point out that only one of the two-ring nitrogens is titrated quantitatively
with perchloric acid. Nenner and Schulz have recently measured the electron
affinities of the N-heterocyclics including phenazines (Ref. 56) but this property
does not appear to relate to basicity.
Although a great deal of experimental work has been done to determine the physi-
cal and chemical characteristics of coal and its byproducts, very little is known
about the manner in which nitrogen is chemically bound in solid coal. From the
structure of nitrogen-containing organic compounds found in the products of
distillation of coal, there appears to be agreement that the major portion of
22
-------
O-1
PYRIDINES
PYRIDANS
r^V^
*-€jQN
0_U INCLINES
ISO-QUINCLINES
ACRIDINES
PHENANTHRIDINES
PHENAZINES
ANILINES
Figure 4. Representative Types of Basic Nitrogen Compounds
Found in Petroleum
23
-------
PYRROLES
INDOLES
CARBAZOLES
BENZCARBAZOLES
AMIDES
QUINOLONES
CN
BENZONITRILES
Figure 5. Representative Types of Nonbasic Nitrogen Compounds
Found in Petroleum
24
-------
the nitrogen in coal is present as heterocyclic linkages in large molecules. The
more abundant nitrogen compounds in low-temperature coal tars and light oil con-
densates include pyridines, picolines (methyl pyridines), lutidines (dimethyl
pyridines), quinolines, acridines, and carbazoles (e.g., Ref. 57).
GENERAL DISCUSSION OF FUEL NO FORMATION
A
Before discussing in more detail the chemistry involved in the formation of fuel
NO , the general features of the process will be outlined. Laboratory burner
J\
and combustor experiments have shown that in the case of fuel NO: (1) the NO
forms very rapidly even at moderate temperatures, (2) the NO concentration in-
creases continually early in the flame front until a concentration maximum is
achieved, and (3) the fraction of fuel nitrogen converted to NO rather than N»
is strongly dependent on the oxygen content of the mixture and decreases as the
nitrogen content of the fuel increases.
Near quantitative conversion of nitrogen compounds to NO is observed in flat flame
burner studies conducted under fuel-lean conditions. In combustor experiments,
with fossil fuels, typical maximum extents of conversion of fuel nitrogen to NO
(at equivalence ratios up to 1) are on the order of 50 percent for added model
compounds, such as pyridine, and somewhat less for naturally occurring fuel
nitrogen. All of the studies that have involved model fuel nitrogen compounds
indicate that, for the additives tested, the extent of conversion to NO is rela-
tively independent of the type of compound employed, but decreases as the addi-
tive concentration is increased. Data obtained in larger combustors and fur-
naces have demonstrated that fuel NO forms at much lower temperatures than does
J\.
thermal NO and, therefore, control techniques based on temperature reduction
A.
that are effective in reducing thermal NO are not generally effective for fuel
NO . However, Martin and Berkau (Ref. 58), Turner et al. (Ref. 4) and Heap et a'l.
Jt
(Ref. 59 and 60) have shown that fuel NO formation can be reduced by staging
J\
and burner design.
Shaw and Thomas (Ref. 61) were apparently the first to demonstrate that NO can
form from fuel nitrogen. In 1968, they studied the reactions of several nitro-
gen compounds, including pyridine, in low temperature CO flames and obtained up
25
-------
to 50-percent conversion to NO. The next laboratory studies of fuel NOX were those
of Martin and Berkau (Ref. 3 ), Turner et al. (Ref. 4 ), and Bartok et al. (Ref. 5 )
Martin and Berkau added pyridine, piperidine, and quinoline to a low nitrogen dis-
tillate oil and measured conversion to NO in a high-pressure atomizing laboratory
furnace. They found that the fraction of fuel nitrogen converted increases with
the air-fuel stoichiometric ratio and decreases with fuel nitrogen concentration,
but the total amount of NO formed increases with both SR and nitrogen content. The
conversion of fuel nitrogen to NO was from 20 to 70 percent and relatively indepen-
dent of the nitrogen compound type.
Turner, Andrews, and Siegmund added a variety of nitrogen compounds to a distillate
oil and obtained results similar to those of Martin and Berkau. In a fire-tube-
boiler domestic furnace, the extent of conversion to NO was independent of nitrogen
compound type (for all additives with boiling points above about 75 C), being about
85 percent at the 0.25-percent N level and about 65 percent at 0.5-percent fuel N.
They also studied residual oils containing various concentrations of (natural)
fuel-N. About 60 percent of the fuel-N was converted to NO for fuels containing
about 0.25-percent N and 40 percent was converted at 0.4- to 0.8-percent N.
Bartok, Engleman, Goldstein, and Del Valle added NO, N02, NH3, C2N2, and CH3NH2
to a methane-air flame in a jet-stirred combustor and measured the NO concentra-
tions in the combustion gas. The fraction of conversion to, or retention of, NO
decreased as the mixture was made fuel rich. Under fuel-lean conditions, all of
the added NO and N02 were recovered as NO and most of the NH3, (CN)2, and CH3NH2
were converted to NO, but the conversion was not complete.
A number of laboratory studies of fuel NOX have since been reported. These in-
clude a shock tube study (Ref. 62'), premixed-flames (Ref. 63 through 73 ), diffu-
sion flames (Ref. 68, 72, and 73), and a Rankine combustor (Ref. 74). Studies of
fuel NOV formation have been conducted in larger burners and furnaces with both
A
neat fossil fuels and doped fuel oils (Ref. 1, 58 and 75 through 88 ). In addition
to these experimental studies of fuel NOX formation, publications dealing with
the theoretical aspects have appeared (Ref. 2 and 89 ).
26
-------
Pershing, Martin, and Berkau studied the influence of design variables on the produc-
tion of thermal and fuel NO from residual oil and coal combustion (Ref. 79). They
concluded that the conversion of fuel-bound nitrogen to NO during residual oil combus-
tion is: (1) responsible for more than 50 percent of total NO emissions under all con-
ditions with the fraction being greater with no air preheat, (2) relatively insensi-
tive to flame zone temperature, (3) increased by increasing excess air at constant
throat velocity, and (4) relatively unaffected by the addition of flue gas recircula-
tion or by increased burner throat velocity. They also conclude that the conversion
of fuel-N during coal combustion is: (1) responsible for 80 to 90 percent of the total
NO emissions, (2) increased markedly by increasing excess air at constant velocity,
and (3) unaffected by flue gas recirculation or changes in burner throat velocity.
The effects on fuel NO formation of staged combustion, flue gas recycle and other
design and operating variables have been investigated by Martin and Berkan (Ref. 58),
Turner et al (Ref. 4), Heap et al (Ref. 59 and 60), and Armento (Ref. 90).
The types of physical and chemical processes that may be involved in the formation
of NO in combustion are diagrammed in Fig. 6. The hydrocarbon species will burn
J\
with the air and can form thermal NO as shown at the top of the figure. Because
nitrogen is present only at low concentrations, it may be assumed that most of the
properties of the flame are independent of the reactions of the nitrogen species.
These properties include temperature, flame speed, and the concentrations of most
species that do not contain nitrogen.
The nitrogen compounds present in the fuel may: (1) form volatile nitrogen species
by undergoing vaporization, pyrolysis, and/or oxidative pryolysis, or (2) remain in
the solid residue of the particle or droplet. The volatile nitrogen species can
then react in the vapor phase to form low molecular weight single N-atom compounds
or radicals, such as HCN, CN, or NH , or a nitrogen-containing soot particle.
These intermediates can be oxidized to NO or form N_ via reaction mechanisms that
will be discussed later in the report. Part of the fuel NO may also form via het-
erogenous reactions involving the combustion of the particle residue or any
nitrogen-containing soot particles that form. Also shown at the bottom of Fig. 6
are some of the chemical reactions that produce N~. However, only those that in-
volve NO appear likely to occur. These N2-forming reactions can also lead to in-
teractions between the mechanisms for the formation of thermal and fuel NO.
27
-------
ro
oo
CO, H20, C0£, 02, OH, H
AIR
HYDROCARBON
SPECIES
NITROGEN
SPECIES
FUEL PARTICLE
OR DROPLET
^
SOLID
PARTICLE
w
CO + OH - C02 + H
H + 0. - OH + 0
2
CO + 02 - C02 + 0
0-ATOM FORMATION
PYROLYSIS OR \
0X1 DATIVE PYROLYSIS \
\
\
2 w HFTf
(C02, H2<>r FUEL
\
\
\
\
\
\
\
[ROGENEOUS
. NO (OR N2)
N >'NO - N + 0
NH + NO - N2 + OH
CN + NO - N2 + CO
NH + N - N2 + H
b
k.
^
0 + N2 - NO + N
N + 02 - NO + 0
N2 * °2 " 2 N°
THERMAL NO FORMATION
VIA ZELDOVICH MECHANISM
HCN + OH - CN -f HoO
CN + 02 - NCO + 0
NCO + 0 = NO + CO
2 NCO = N£ + 2 CO
FUEL NO FROM HCN
NH + OH « N + HO
N + OH - NO + H
FUEL NO FROM NH
INTERACTION OF
FUEL NO AND THERMAL NO
MECHANISMS
Figure 6 . Potential Paths for NO Formation in Fossil Fuel Combustion
-------
KINETICS AND MECHANISMS OF PYROLYSIS OF ORGANIC
NITROGEN COMPOUNDS
Only a limited number of studies have been conducted of the reactions of organic
nitrogen compounds that may occur under "preflame" conditions. The investigation
in 1962 by C. D. Kurd and coworkers of the pyrolysis mechanisms of pyridine and
the picolines (Ref. 91 and 92) is the most relevant to this program. Johns et al.
(Ref. 93) have compared the thermal stabilities, under a variety of conditions, of
many organic compounds including some N-heterocyclics. Most of these data were ob-
tained at only moderate temperatures and rather long heating times. Hirsch and
Lilyquist (Ref. 94) compared the thermal stabilities of organic compounds using a
pyrolysis-gas chromatography technique similar to that employed in the model com-
pound pyrolysis studies of this program (Phase IB). Their measurements were made
at lower temperatures than in the present study; 750 C for the less stable com-
pounds and 870 C for the more stable. Both of these papers indicate that the het-
erocyclic nitrogen compounds are among the most stable of the organic compounds.
Hirsch and Lilyquist found that, whereas diphenylamine decomposed 54 percent at
870 C and naphthalene decomposed 29 percent, pyridine decomposed only 6 percent
and quinoline underwent less than 1 percent decomposition at this temperature.
Benzene, however, was found to have about the same stability as pyridine (5-percent
decomposed) in agreement with benzene being considered as a very stable organic
molecule.
Hurd and coworkers pyrolyzed pyridine and the three picolines in a flow reactor
at 700 to 850 C under inert conditions to study the mechanism of the pyrolytic
formation of arenes such as benzene. They claim to have shown that the picolines,
which are methyl substituted pyridines, are much less thermally stable than is
pyridine (Ref. 91 and 92'). This situation is relevant to the mechanism of fuel
NOX formation because most heterocyclic fuel nitrogen compounds are substituted.
29
-------
However, the following kinetic calculations on the data of Hurd and Simon (Ref. 92),
made under this program, demonstrate that this conclusion is not valid at the
highest temperature of their study.
The values listed in Hurd and Simon's Table 1 for contact time and percent pico-
line recovered were converted to first-order rate constants and plotted in the
Arrhenius form (Fig. 7). Although there is considerable scatter in the data,
the rate expressions listed on the figure were fitted to each of the picolines.
Hurd and Simon give only two decomposition rates for pyridine in their apparatus:
57-percent undecomposed after 5 seconds at 850 C*and 64-percent undecomposed**
after 9 seconds at 825 C. The rate constants calculated for these points (0.112
and 0.050 sec'1, respectively), and plotted in Fig. 7, give an activation energy
for pyridine pyrolysis on the order of 80 kcal/mole. Although rate expressions
obtained over a 25-degree temperature range cannot be accurate, the two pyridine
rates reported by Hurd and Simon give the rate expression k = 10 ' x
exp (-80200/RT), sec'1, which is plotted in Fig. 7. Also plotted is the pyri-
12 58
dine pyrolysis rate constant obtained in Phase IB of this program, k = 10 '
exp (-70000/RT). The rates obtained from these two expressions agree within 25
percent at these two temperatures. This is excellent agreement considering the
differences in the experimental techniques employed. The sample size used by
Hurd and Simon was larger by at least a factor of 10^, and they allowed the resi-
due to build up in their reactor (a problem that is discussed under Phase IB).
It can be seen from Fig. 7 that the picolines may be less stable than pyridine at
the lower temperature studied by Hurd and Simon, but at 850 C pyridine and the pico-
lines decompose at essentially the same rate. The differences are less than 10 per-
cent, which is well within the expected experimental error. At the lower tempera-
tures, the picolines probably decompose (as does toluene) by loss of hydrogen from a
methyl group followed by decomposition of the pyridyl radical. If the trends in
*The pyridine data point at 850 C is not shown in Fig. 7 because it coincides
with the 2-methylpyridine point.
**There is an apparent error on p. 4520 of Ref. 92 that is corrected on p. 4521.
It was stated that "at 825 C and 9 seconds contact time, one-third of the
pyridine was recovered but on the next page, "64% was recovered under these
conditions."
30
-------
850
825
800
TEMPERATURE, C
775 750
725
700
o
LLl
CO
CO
z
o
o
a:
o
i
l-
co
OC
C3
O
-2.0
—), 2-METHYLPYRIDINE
—), 3-METHYLPYRIDINE
—), ^-METHYLPYRIDINE
—), PYRIDINE
EXP(-4l,500/RT), SEC"
EXPH8,000/RT), SEC'
EXP(52,AOO/RT), SEC"
•k = 10 ""JV EXP(-70,000/RT)
(FROM THIS STUDY)
\
»
0.90
0.95
.00
1000/T
K
Figure 7. Arrhenius Plots of Pyrolysis Rates for Picolines and
Pyridine Obtained by Hurd and Simon (Ref. 92)
31
-------
Fig. 7 were to continue at higher temperatures, the picolines would become more
stable than pyridine.* Since the rates become equal at 850 C, however, it is
more likely that they will remain equal above 850 C. That is, the mechanism for
the decomposition of the picolines may become the same as for pyridine involving
either direct ring rupture or loss of a ring hydrogen followed by scission of the
pyridyl ring. If this is the case, the rate of decomposition of heterocyclic ni-
trogen compounds will be nearly independent of the substituted groups at combus-
tion temperature. This forms part of the rationale for studying low molecular
weight model compounds in some of the Phase IB pyrolysis experiments, and for us-
ing the kinetic parameters for the decomposition of these model compounds to rep-
resent fuel N-compounds in the kinetic-diffusion mathematical combustion model.
Hurd and Simon report the following products from the pyrolysis of pyridine and
2-picoline at 825 C:
Product
Benzene
Acetoni tri le
Acryloni tr i le
Benzoni tr i le
Quinol ine
Pyridine
3-Picol ine
4-P5col ine
Residue (weight percent)
Mole Percent
Pyridine
0.18
0.2
0.1*
1.1
1.3
--
• --
--
*
2-Picol ine
0.39
1.3
0.5
1.1
0.3
25
1.2
0.7
^7
»A considerable but unmeasured amount.
*This is one of the subtleties of high-temperature chemistry. Not only do endo-
thermic reactions become more favored thermodynamically as the temperature in-
creases, but the rates of reactions having large activation energies increase
faster than those of reactions having smaller activation energies. This point
will come up again in the Phase IB discussion where it is suggested that thermal
decomposition of fuel nitrogen compounds may be faster at combustion temperatures
than their oxidative decomposition.
32
-------
They explain the formation of the various products by cleavage of the pyridine
ring at a C-N bond to give the radical:
N-CH-CH-CH-CH-CH
6 y 3 a
which can cleave at the a, ft, y> or 6 positions with the last two being favored
by resonance. According to their proposed mechanisms, some of the observed prod-
ucts form as follows:
2 CH-CH-CH -> benzene
N-CH-CH -> acetonitrile
pyridine + CH-CH-CH-CH -»• quinoline
Sternling and Wendt have pointed out that rupture at the 6 position leads to the
formation of HCN. It will be seen that in the present study, the amount of HCN
formed from pyridine increases as the temperature is increased from 950 to 1100 C.
The quantitative conversion of pyridine to HCN occurs at 1100 C. In terms of the
above mechanism, this would suggest increased scission of the 6 bond at the higher
temperatures. However, similar trends were observed in the amount of HCN formed
from the ir,odel compound benzonitrile (the maximum conversion to HCN being about
80 percent at 1100 C). Thus, it may be necessary to consider other paths for HCN
formation. Of course, the direct formation of a CN radical is likely in the case
of benzonitrile.
A few other studies have been carried out that relate to the pyrolysis of nitro-
gen compounds of the type that may be present in fossil fuels (Ref. 95 through 99").
However, these do not appear to contribute information of importance to the high-
temperature chemistry of interest in fuel NOX formation. However, a current study
of the mechanism of the pyrolysis of ethylamine (Ref.100) is of interest with re-
spect to the later stages of combustion.
33
-------
ANALYSIS OF N-COMPOUND COMBUSTION
More data are available relating to the fate of small nitrogen molecules, such as
HCN and NH,, in the combustion process, and the results observed in Phase II of
this program have added significantly to this body of data. Such data have given
some indication of how the fuel nitrogen intermediates that form in the preflame
reactions are finally converted to NO or Nn in the flamefront, but the detailed
x 2
mechanisms involved are not yet firmly established. The background and theoreti-
cal aspects of N-compound combustion will be discussed here.* This subject will
be discussed further in the Phase II section of this report after the Phase II
results, obtained with a premixed flat-flame burner, have been presented.
Burner Studies of NOX Formation From N-Compounds
Fenimore (Ref. 66) added pyridine, methacrylonitrile', methylamine, and NH, to pre-
mixed ethylene-0 -N? and ethylene-CL-inert gas flames at atmospheric pressure and
measured the yield of NO. He worked under conditions where the formation of ther-
mal NO was not significant. Johnson (Ref. 71) measured NO formation from pre-
A
mixed, laminar CH -air-NH flames at a pressure slightly greater than atmospheric.
Slater (Ref. '72) measured the recovery of added NO and the conversions of NH,,
CH_NH0, and pyridine to NO in both laminar premixed and diffusion CH.-air flames
o 2. Q
at atmospheric pressure. In all three of these studies, the NO was measured suf-
ficiently far downstream that the NO had reached its final (exhaust) value.
Bartok et al. (Ref. 5 ) used a premixed jet-stirred combustor to study the effects
of adding NO, NO , NH , cyanogen, and CH NH to a methane-air flame.
Using premixed ethylene-O^-Ar flames at atmospheric pressure and cooled probes
to sample within the flame, De Soete investigated the conversion to NO of added
NH,, ethanediamine, dibutylamine, and triethylamine (Ref. 63); cyanogen (Ref. 64);
and the recovery of added NO (Ref. 65). In De Soete's experiments, as was also
observed by Fenimore and in this study, nearly quantitative conversion to NO occurred
''Sternling and Wendt (Ref. 2 ) reviewed the information on N-compound combustion
available in 1972.
34
-------
in fuel-lean mixtures, but the yield fell off rapidly as the equivalence ratio
was increased beyond stoichiometric. Interestingly, all of the amine additives
as well as added NO formed substantial amounts of HCN early in the flame under
fuel-rich conditions (Ref. 63 and 65). With cyanogen as the additive (Ref. 64),
some of the NO formed from an HCN intermediate, but De Soete concluded that the
oxidation of (CN). is the major path for NO formation. With cyanogen as the addi-
tive, HCN builds up as an intermediate reaching its maximum considerably before
NO reaches its maximum. Analysis of the lower temperature experiments (Fig. 1
of Ref. 64) indicates that either another nitrogen-containing intermediate is
formed or nearly all of the conversion to N9 occurs before about 20 percent of
the NO has been formed.* This can be seen from the following table (cyanogen was
initially added to a concentration of 265 ppm which is equivalent to 530 ppm N):
Milliseconds
0.12
0.35
0.55
1.4
ppm N as :
Cyanogen
210
75
32
0
HCN
18
81
105
12
NO
50
150
210
297
Total
278
306
347
309
Unrecovered
252
224
183
221
If no other species are involved, nearly 50 percent of the N was converted to N_
during the first 0.12 millisecond, after which nearly quantitative conversion of
HCN and (CN)2 to NO occurred.** Another more likely explanation would be that
early in the flame the unrecovered nitrogen was in the form of a radical, such as
NCO, that somehow converts to N~ in the probe. Later in the flame, the unrecovered
nitrogen would exist as N . When the additive was NO (Ref.'65), all of it was
recovered under fuel-lean conditions, but the fraction of NO recovered fell off
in fuel-rich mixtures. In fuel-lean ethylene flames, about one-fourth of the
NO is apparently consumed early in the flame and then reforms. No such effect
*This analysis ignores diffusional effects. The flame was slightly fuel-rich
(4)=1.07).
^De Soete did not measure NHj or N2 in his experiments. Therefore, NHs could
have formed and then been converted to N2 but this does not seem likely in
light of the results reported in Phase II of this report.
35
-------
was noted in \\2~®2 flames with added NO. When De Soete added 265 ppm NO to a
fuel-rich ethylene-02-Ar flame, the NO decreased to its final concentration of
65 ppm in about 0.16 msec, at which time the HCN reached its maximum measured
concentration of 180 ppm (Fig. 5 of Ref. 65). The HCN then decayed rapidly, being
completely consumed at 0.4 msec, while the NO concentration remained relatively
constant. The formation of HCN in the flame front was so rapid that its maximum
could have occurred between the measurement points and, therefore, have been
greater than 180 ppm.
In addition to the flame-probing studies reported in Phase II of this report, which
were conducted at a pressure of 0.1 atmosphere to spread out the flame zone, Merry-
man and Levy (Ref.101) have probed methane-air flames(most at 1 atm) containing
methylamine, pyridine, and piperidine. They observe that both in the presence and
absence of added fuel nitrogen, significant amounts of NO,, form in the flame zone.
The N0_ peaks just beyond the end of the visible flame and the maximum amount of
NO,., is equivalent to one-half or more of the NO concentration in the exhaust gases.
The formation and decay of N0» is attributed to the reactions:
NO + H02 = N02 + OH (1) •
N02 + 0 = NO + 02 (2)
Merryman and Levy added small amounts of NO to methane-air flames and found that
80 percent of the NO is removed in the preflame region and the remainder is removed
in the visible flame. Most of the NO is converted to N0» in this region and some
"organic" nitrogen is formed that could be either NH or HCN. In the post-flame
gases, most of the .N0_ is converted to NO under fuel-lean conditions, and all is
converted under fuel-rich conditions.
Mechanism of Thermal NO Formation
Except under very fuel-rich conditions, the amounts of thermal NO generated in
homogeneous combustion can be predicted approximately (Ref. 62) by integrating the
rates of the following three reactions (the first two of which were proposed by
Zeldovich in 1946, Ref. 102 and 103, to explain the results of his classic experi-
mental investigation of the formation of NO in combustion and explosions).
36
-------
AH (298 K)
kcal/mole
0 + N2 = NO + N +75.0 (3)
N + 02 = NO + 0 -31.7 (4)
N + OH = NO + H +18.6 (5)
Reaction 3 is rate-determining when NO is well below its equilibrium concentra-
tion and the rate of NO formation from N» depends mainly on the oxygen atom con-
centration in the flame and the local temperature. The rate of conversion of N
to NO has a very large temperature dependence because: (1) reaction 3 being en-
dothermic by 75 kcal/mole has an activation energy of 75 kcal/mole, and (2) the
concentration of 0 atoms in the flame is strongly temperature dependent. The
global activation energy for the formation of thermal NO is about 135 kcal/mole*
(Ref.104). For this reason, thermal NO forms near the maximum-temperature zone of
the flame and some methods of reducing thermal NO involve the reduction of peak
flame temperature, e.g., by exhaust gas recycle.
Mechanisms of Fuel NO Formation
The only detailed mechanisms for fuel NO formation that have been proposed are
ones in which it is assumed that the original fuel nitrogen compounds decompose
quantitatively, by an unspecified mechanism, to a simple nitrogen intermediate.
Sternling and Wendt used a flame-sheet model to predict the extent to which fuel-N.
would be converted to NO in the combustion of PL-air mixtures (Ref. 2 y. They con-
'cluded that the fraction converted should increase with increasing temperature and
decreasing fuel nitrogen concentration. Sternling and Wendt also concluded that
among the gaseous products of pyrolysis, HCN is expected to be the most important.
However, they assume in their calculations that the HCN is completely converted to
N atoms (via a reaction such as CN + 0 = CO + N) which then react according to the
Zeldovich mechanism—reactions 3 and 4.
*75 kcal/mole, from reaction 3 , plus one-half the bond dissociation energy
of 02.
37
-------
Fenimore (Ref. 66) found that the amount of NO he measured, [NO], was related to
the maximum possible NO, [NO] , which would form if all the additive were con-
in 3.x
verted to NO, by the following expression:
rxTni f [NO] + [NO] "I
[NO] _ . pYn max ffi,
•" ' ' — J. — CAU l~ u—•—M_- ,.,.__.._,_,. „ | \VJ
x L ^x J
where x is a parameter characteristic of the flame conditions (mixture ratio and
temperature) but independent of the amount of additive.
Fenimore was able to derive Eq. 6 theoretically by assuming that every addi-
tive molecule forms the same intermediate, I, which reacts via the two competing
paths:
k7
I + R ^*-NO + . . . (71
k t7J
I + NO —^"~N2 + . . . (8)
thus,
_ dj1] _ x + [NO]
where
x = k7(R)/kg
Equation 6 is obtained by integrating Eq. 9 to complete reaction of the additive.
Species R in Eq. 7 is an oxygen-containing radical. Fenimore showed that the
measured values of x are in agreement with R = OH and k7/kg = 2 to 3. The nature
of the species I cannot be definitely established, but Fenimore concludes that I
is most likely either NhL or an N atom. Fenimore's strongest argument against
CN intermediates in fuel NO formation was that his general equation (Eq. 6 ) fits
X
data on NO formation in NH,-0 flames where CN species cannot form.
38
-------
De Soete analyzed his results in terms of competing global reactions. In Ref. 63 ,
he concluded that NO is formed at a rate proportional to the concentrations of 0^
and of the nitrogen intermediates, N*, and is destroyed at a rate proportional to
the concentrations of NO and N*. This is similar to Fenimore's mechanism with
R = CL. The mechanism proposed by De Soete in Ref. 64 is also similar except that
1/2
the rate of NO formation is proportional to (0_) , suggesting that R = 0 atom.
De Soete proposed a similar overall mechanism in Ref. 65 and concluded that CN
and HCN may be important intermediates in hydrocarbon flames, but NH species are
J\
not. He predicted that further treatment of his data will demonstrate that CN and
HCN form N atoms as a key intermediate in the formation of fuel NO.
Constrained Equilibrium Models. Bowman has proposed (Ref. 62 ) that a constrained
equilibrium model* will qualitatively predict the effects of mixture ratio and
temperature on the yields of fuel NO. Flagan et al. (Ref. 1 and 89) have made
quantitative calculations with two rate-constrained equilibrium models and com-
pared the results with experimental data. These models will be described briefly.
Summary of Single-Constraint Equilibrium Model. This model assumes that
.early in the flame, the single-N species reach a constrained equilibrium with
each other and all C-H-0 species, the constraint being that N is not allowed
to form at this point in the flame. At equivalence ratios up to about 1.6, this
assumption results in nearly all the fuel-N being converted to NO very early in
the flame (i.e., at the constrained equilibrium condition). Some of the NO is
then converted to N ; the only two important reactions being N •*- NO = N_ + 0 and
NH + NO = NO + H. This single-constraint model predicts that NO reaches its max-
imum very early in the flame and then decreases gradually to its final value with
a characteristic decay time on the order of 1 to 100 milliseconds (see Fig. 2 and
3 of Ref. 89 in which T was taken as 2 to 14 milliseconds).
*The partial equilibrium approximations that have been proposed for modeling
thermal NO formation have been reviewed by Leonard and Mellor (Ref.105).
A
39
-------
Summary of Three-Constraint Equilibrium Model. This model assumes that early
in the flame, the single-N species reach a constrained equilibrium, the constraints
being: (1) N is not formed, (2) N-atoms are not formed, and (3) oxidized N-species
(NO, HNO, NO , etc.) are not formed. Starting with the constrained equilibrium
composition, N and NO are allowed to form at the kinetic rates determined by the
equations:
(10)
NH + 0 = N + OH
NH + OH = N + HO
NH -H 0 = NO + H (12)
N + 0 = NO + 0 (4)
N + OH = NO + H (5)
As in the single-constraint model, the N_ forms via the reactions:
N + NO = N2 +0 (-3)
NH + NO = NO + H (13)
with the N + NO reaction being of primary importance. (Note that this reaction
scheme is equivalent to that proposed by Fenimore.) Integration of this mechanism
predicts that the oxidation of fuel nitrogen occurs with a half-time on the order
of 10 microseconds and that significant amounts of NO are reduced to N? via reac-
tion with N and NH.
Comments on Constrained Equilibrium Models. These models predict generally
the trends in NO yields observed by Fenimore (Ref. 66 ) and Johnson (Ref. 71 ). In
both of these studies, the NO was measured at a fixed point some distance down-
stream from the flame front. The triple-constraint model is also in approximate
agreement with most of the NO profiles reported from the flame-probing experiments
40
-------
of De Soete (Ref. 63 and 64). That is, NO increases steadily with distance from
the burner and the reactions are very fast with much of the NO forming in less
than 0.1 millisecond. It is not apparent if this model can predict the results
obtained by De Soete in his experiments with added NO (Ref. 65).
The single-constraint model, which is apparently the same as that proposed by
Bowman, predicts that all of the fuel nitrogen is converted to NO and then, under
fuel-rich conditions, part of the NO is converted to N? (by reaction with N or NH)
to give the values measured downstream in experiments such as those of Fenimore.
However, the detailed probing results of De Soete (and Phase II of this study)
indicate that this model is not realistic because the NO concentration actually
increases with distance from the burner.
The three-constraint equilibrium model appears to be more realistic but it is
difficult to see its advantage over a direct integration of the kinetic system
if sufficient forward (and reverse) reactions are included to permit the actual
steady-state concentrations of each species to be estimated.
It can be seen that none of the proposed models for fuel NO formation give an
indication of the actual chemical mechanisms involved because they are general
enough to accept any principal intermediate that might form from the parent fuel
nitrogen compound.
Thermochemical Considerations.. Complete chemical equilibrium is not approached
in hydrocarbon-air flames except for combustion systems that operate at unusually
high pressures and/or temperatures.* In particular, thermal NO does not reach
*e g., those that use 02 enriched air, high preheat temperature, or initial
adiabatic compression of the fuel-air mixture (as in the Otto cycle engine)
41
-------
its equilibrium concentration (Ref. 106) until the gas begins to cool, and the
equilibrium concentration decreases. The initial stages in hydrocarbon oxidation
involve the formation of CO, rather than CO and this can lead to CO concentra-
tions much greater than the equilibrium values. The overshoot of CO can cause
oxygen atom overshoot via the following two reactions:
CO + OH = C02 + H
H + 02 = OH +0
CO + 02 = C02 + 0 ^16^
It can be shown that if reactions 14 and 15 are at constrained equilibrium,
(0)/(0)eq > (CO)/(CO)eq.
Although equilibrium is not attained, the consideration of the thermochemistry
involved does give some indication of the relative concentrations at which species
will be present in the flame under various conditions and, therefore, the likeli-
hood of their involvement in NO formation. However, such information may be more
directly applicable to thermal NO formation, which occurs mainly in the post-flame
gases, than to fuel NO formation, which is believed to occur earlier in the flame
and at lower temperatures. Shown in Fig. 8 and 9 are the mole fractions of the
equilibrium combustion products as a function of fuel-air equivalence ratio for a
hydrocarbon-air flame at 1 atmosphere and 2100 and 1600 K, respectively. The H/C
molar ratio of the fuel was taken as 1.6 to approximate a heavy fuel oil. The H/C
ratios in No. 6 fuel oils range from 1.4 to 1.8.
The species that contain oxygen that is available for NO formations are 0 , 0, OH,
C0?, and HO. The energy required to obtain an 0 atom from each of these species
is as follows: AH (2gg ^
kcal/mole
02 =0+0 119.2 (18)
OH = 0 + H 169.5 (19)
C02 = 0 + CO 127.2 (20)
H20 = 0 + H2 119.4 (21)
42
-------
Figure 8. Equilibrium Products as a Function of Equivalence Ratio
for CH, 6-Air Flame at 2100 K and 1 Atmosphere
43
-------
Figure 9. Equilibrium Products as a Function of Equivalence Ratio for
CH. .-Air Flame at 1600Kand 1 Atmosphere
1. o
44
-------
Zeldovich chose reaction 4 , N + CL = NO + 0, as the second step in his thermal
NO mechanism and reaction 5 , N + OH = NO + H, has since been added to the extended
Zeldovich mechanism. Because of the larger concentrations of C02, the reaction
N + C02 = NO + CO will compete for N atoms unless its rate constant is much slower.
It has been estimated (Ref. 107) this reaction has an activation energy of 25 kcal/
mole, making its rate constant at 2000 K, more than 200 times smaller than those
of reactions 4 and 5. If this estimate is correct, the N + C02 reaction will not
be important. Even if this reaction turns out to be fast, its inclusion in the
extended Zeldovich mechanism would not affect the calculated rate of thermal NO
formation because reaction 3 is rate-determining.
The oxygen species that should be considered in the fuel NO mechanisms are 0, 02,
OH, and C0_. The extraction of an 0 atom from H^O is not likely to be a fast pro-
cess, and the reactions of N* with (XL may also have sizable activation energies.
It can be seen from Fig. 8 and 9 that the 0-atom concentration is always less
than that of OH by a factor of at least 10. Therefore, 0-atom reactions will not
be important in fuel NO formation unless (1) there is no equivalent reaction with
OH, (2) the equivalent reaction with OH has a much lower rate constant, or (3) the
0/OH ratio exceeds the equilibrium value.
Although the fuel-N* species will be at much greater steady-state concentrations
than the equilibrium concentrations shown in Fig. 8 and 9 , these figures indi-
cated which N* species are favored thermodynamically. At 2100 K, nitrogen atom is
the favored single-N species that does not contain oxygen up to about tj) = 1.1,
above which NH3 has the lowest free energy. The equilibrium concentration of .
NH3 at 1600 K exceeds that of N atom above = 0.8. Of the NHX species, NH3
predominates at all equivalence ratios at 1600 K and above = 1 at 2100 K.
Below (j) = 1 at 2100 K, (NH2) > (NH3) . The equilibrium concentrations in Fig. 8
indicate that NO is the favored single-N species up to very high mixture ratios.
This is in agreement with the results obtained by Flagan et al. on the single-
constraint constrained equilibrium model.
NH is favored thermodynamically over HCN (and CN) at all temperatures and equiv-
*J
alence ratios. It should be noted that under these combustion conditions, solid
carbon (soot) is not a favored combustion product. Therefore, HCN is favored over
45
-------
NH in the presence of a solid carbon surface (see discussion under Phase IB),
but NH is favored in a homogeneous flame. The HCN that was observed to form
from NH in premixed flames (De Soete and Phase II of this study) must therefore
have formed from: (1) the heterogeneous reaction of NH with soot particles, or
o
(2) the reaction of NH with hydrocarbon species early in the flame before the
«J
fuel is consumed. The thermodynamic tendency for HCN to form NH in the later
O
stages of combustion (which can be shown to be independent of the concentration
of HCN) leads to the very significant conclusion that HCN formed in fuel pyrolysis
or in the early stages of combustion could either be oxidized to fuel NO or be con-
verted to NH_ which then forms fuel NO. NH_ has not been observed to form in
flames with added HCN, but De Soete's results (discussed earlier) are not incompa-
tible with such a hypothesis.
Elementary Reactions in Fuel NO Formation. To eventually model in detail the for-
mation of fuel NO, the important elementary reactions that nitrogen species undergo
in combustion will have to be identified and their rate parameters established.
Experimental pyrolysis data obtained under Phase IB of this program indicate that
the nitrogen in heterocyclic fuel-N compounds may be converted mainly to HCN in the
early stages of combustion. These data also provide preliminary rate parameters
that permit estimates to be made of the rate of this conversion process as a func-
tion of temperature. It can be seen from this and the previous discussion that
the elementary reactions that must be investigated are those involved in: (1) con-
version of HCN and CN to NO, (2) conversion of HCN and CN to NH species and vice
J^
versa (3) oxidation of NH to NO, (4) conversion of NO and other N species to N ,
X ™
and (5) conversion of NO to HCN. The burner studies conducted in Phase II were
oriented toward the investigation of these types of processes and indicate that
HCN, once formed, should survive deep into the flame front.
Many of the previous discussions of fuel NO formation mechanisms have been in
terms of global or general mechanisms (e.g., the mechanisms of De Soete, Bowman,
and Fenimore). However, Sternling and Wendt (Ref.2 ), Flagan (Ref.1 ), and
Slater (Ref. 72) have discussed the types of elementary reactions that may be
involved. The types of reactions they have considered are listed in Table 2 .
46
-------
TABLE 2. TYPES OF ELEMENTARY REACTIONS THAT MAY BE INVOLVED
IN FUEL NOX FORMATION
1. Conversion of HCN and CN
HCN + H = CN + H2
HCN + 0 = CN + OH
HCN + OH = CN + H20
II. Conversion of HCN and CN
HCN + 0 = CHO + N
CN + 0 = CO + N
HCN + 0 = CO + NH
III: Oxidation of NHX to NO:
NH3 + OH = NH2 + H20
NH3 + 0 = NH2 + OH
NH3 + H = NH2 + H2
NH3 + N = NH2 + NH
NH2 + 02 = NH + H02
NH2 + H « NH + H2
NH2 + 0 = NH + OH
NH2 + OH « NH + H20
NH2 + NH2 = NH3 + NH
NH + OH = N + H20
NH + H = N + \\2
NH + 0 = N + OH
NH + H2 = NH2 + H
IV. Formation of N2:
HCN + N = N2 + CH
CN + N = N2 + C
CN + CN = N2 + C2
NCO + NCO = N2 + 2CO
to NO:
CN + 02 =
CN + 0 =
to NHX:
NH + 02 =
NH + 0 =
NH + OH =
N + 02 =
N + OH =
NH2 + NO
NH2 + NO
NH + NO =
N + NO -
N + NH =
NH + NH =
NH + NH =
NO + CO
NO + C
NO + OH
NO + H
NO + H
NO + 0
NO + H
= N2 + . . .
= N2 + H20
N2 + OH
N2 + 0
No + H
N2 + 2H
N2 + H2
HCN + 0 =
CN + 02 =
CN + OH =
NCO + 0 =
NCO + 02 =
NH2 + 02 =
HNO + M =
HNO + OH =
HNO + H =
HNO + NH2
HNO + NH =
HNO + N =
HNO + 0 =
HNO + 0 =
HNO + 0 •
HNO + H =
NO + NO =
HNO + HNO
NCO + H
NCO + 0
NCO + H
NO + CO
CO + NO + 0
HNO + OH
NO + H + M
NO + H20
NO + H2
= NO + NH3
NO + NH2
NO + NH
NO + OH
N + H02
NH2 + 02
NH + OH
N20 + 0
= N20 + H20
47
-------
Not all these processes are likely to occur, and a screening of the most probable
processes is required before detailed reaction schemes can be postulated and tested.
Some of this screening has been done by Sternling and Wendt (Ref. 2), Flagan
(Ref. 1), Bartok and Engleman (Ref. 107) and, particularly, Slater (Ref. 72).
A few of the many possible elementary processes that may be involved in fuel NOX
formation have been incorporated in the combustion models that will be described
in the next section to demonstrate that the models are capable of testing various
detailed fuel NOX mechanisms. As specific reaction rates become established as
the most likely paths for fuel NO formation, these can be added to the mathemat-
ical model. A better mechanism for fuel NOX formation is proposed near the end
of this report based on the experimental burner results obtained under Phase II.
48
-------
COMBUSTION MODELS
To understand or predict the resultant effect of fuel nitrog-en, it is necessary
to describe in detail the heterogeneous combustion of fuel oil sprays or coal
particles. That is, determine droplet (particle) temperature and composition
"life histories," as well as to delineate the temperature and gas composition
profiles to which nitrogen-bearing compounds are exposed as they diffuse away
from the droplet (particle) and into the flame zones of bulk combustion gas
stream.
Until fairly recently, the spray combustion models, and the individual droplet
models upon which they are based, assumed that when combustion reactions occur,
they proceed abruptly to local thermodynamic equilibrium. As a result, they are
adequate for analyzing gross combustion field behavior, but are inadequate for
calculating the occurrence of species whose concentrations vary widely with depart-
ures from equilibrium. Combustion-generated pollutant species NO , smoke, and
A.
unburned hydrocarbons are prime examples of such species; their presence cannot
be predicted with any confidence by the previous models.
Chemical kinetics of fuel vapor combustion within the drop-surface-to-flame bound-
ary layer and the flame-to-free stream region have been employed only in single
droplet models restricted to near-stagnant flow conditions (i.e., large flame/
droplet radius ratios), quasi-steady gasification (i.e., constant droplet temper-
ature), and single component droplets. Under such conditions, Fendell et al.
(Ref. 108J, using matched inner and outer expansions across the flame region,
showed that earlier thin-flame and nonreacting vaporiziation models actually
represent limiting cases of the more general treatment which includes finite-rate
chemistry. Under the same restrictions, Bracco (Ref. 109) attacked the NO forma-
tion problem, but selected a simpler solution method. He solved the finite-rate
hydrocarbon chemistry problem to get droplet-to-free-stream profiles of fuel,
oxidant, and diluent species concentrations and temperature, and then used those
data as input for a kinetic analysis of NO formation via the well known Zeldovich
mechanism.
49
-------
The principal objective of this task was to develop droplet/particle combustion
models that could predict multicomponent droplet (particle) "life histories" and,
also, the production rate of pollutants in the film surrounding the droplets
(particles). One of the most important aspects of the pollutant formulation
process was to describe the fate of fuel nitrogen compounds.
Four separate computer models were developed: (1) a droplet vaporization model,
(2) a droplet flame front model, (3) a coal combustion model, and (4) a kinetic/
diffusion model. These models were programmed so that they could be used as sub-
programs in a decoupled combustor flow program.
Inherent in all four models are the continuity, diffusion, and energy equations,
but each differs in the assumptions and handling of the equations for its partic-
ular solution.
The two droplet models are used to calculate fuel droplet temperature and compo-
sition "life histories" and an approximation to the film temperature and composi-
tion profiles. The vaporization model neglects chemical reactions and the flame
front model employs infinite rate kinetics for the reaction of the vaporizing species
with oxygen and zero reaction rates for the other compounds present in the film
surrounding the droplet.
The coal combustion model is used to calculate the coal particle temperature and
composition "life history" including devolatization and heterogeneous reactions
at the particle surface. Chemical reactants in the film surrounding the particle
are neglected.
50
-------
The kinetic/diffusion model is used to calculate detailed kinetic and diffusion
processes occurring in the film surrounding the fuel droplet or coal particle.
This model is the main computer program for calculating pollutant generation rates
including NO produced from fuel nitrogen compounds. Initial input profiles for
A
the kinetic diffusion program are obtained from model calculations using the pre-
ceding fuel spray and coal combustion models. (See Appendix I for a description
of the use of each model as well as criteria for the selection of the appropriate
droplet program.)
Droplet Vaporization Model
Description of Model. The model for the heating and vaporization of multicompo-
nent fuel droplets surrounded by a specified gas flowfield is depicted schemati-
cally in Fig. 10. It should be noted that the model does not include any reac-
tions of the fuel vapor components with the surrounding gas.
The basic equations solved are the species diffusion equation:*
N.
(molar flowrate
with respect to
fixed axes)
species continuity equation:
dN.
dr
(spatial variation
in molar flowrate)
-47TT (c,
dX.
dr
X.
J
nr
(molar flowrate
resulting from
diffusion)
nt
£
m=l
(molar flowrate
resulting from
bulk flow)
4irr R. = 0,
(production of j by
homogeneous chemical
reaction)
(22)
(23.)
and the energy equation
_J_ d J.47rr2k
4Trr2 dr (
(rate of energy output
by conduction)
(rate of energy output
by convection including
chemical energy)
"'Symbols are defined in the Nomenclature List on page 345.
(24)
51
-------
Figure 10. Droplet Vaporization Model
52
-------
During the development and solution of the basic equations, the following assump-
tions are made:
1. The gas flow in the film surrounding the droplet is spherically symmetric
and the film thickness is related empirically to the droplet Nusselt num-
ber or the ratio of convective to conductive heat transfer rate.
2. Pick's law is presumed valid throughout, implying equal binary diffusion
coefficients which are evaluated at mean film conditions.
3. The thermal conductivity and gaseous specific heats are constant and
evaluated at mean film conditions.
4. Temperature and concentration gradients within the droplet are ignored.
5. Fuel vapors are assumed to diffuse through the boundary layer (film)
surrounding the droplet without undergoing pyrolysis or combustion
reactions.
6. The fuel is specified as a multicomponent mixture of compunds having
different properties,e.g., vapor pressure, molecular weight, etc.
7. Droplet temperature and composition varies with time.
8. The static pressure in the radial direction is constant.
Summing the diffusion equation over all vaporizing species yields:
_ dX
N + (4irr ) (c^] -r-^- = N X (25)
v ^ dr vv ^ J
where
vaporizing
species
N = 2- N. (total vaporization rate) (26
vaporizing
species
X = 4- X. (total' vaporizing mole fraction) (27)
v -i i
53
-------
Noting that the specie molar flowrates are constant (application of the species
continuity equation), the preceding total vaporizing species diffusion equation
is integrated between the droplet surface and the gas stream to yield the total
vaporization rate:
(28)
where the subscripts g and d represent the bulk gas and the gas at the droplet
surface,and the outer boundary of the film is determined from the mass-transfer
Nusselt number empirical relationship (Ref. 110):
rd
r6 = ~ C29)
m 1 - ^—
m
The mole fraction of the vaporizing species in the gas stream (Xv ) is determined
by the gas stream composition. The mole fraction of the vaporizing species at
the droplet surface (Xv ) is determined by:
vaporizing
species P
X 2 Y. -=i- (30)
vd i 1 P
where Y. is the volume fraction of the i species inside the liquid droplet and
Pv. is the vapor pressure, evaluated at droplet temperature, of that species.
Each droplet component vaporization rate is determined by integrating the com-
ponent's diffusion equation between the droplet surface and the gas stream and
is given by:
N.
— [X - X ) = X. (1 - X \ - X. (1 - X \ (31)
NV [ vd V xd l V xg l v
54
-------
The liquid droplet radius is therefore determined by:
[f
dt
and its composition is determined by:
& [f *t *i \]
The droplet heating rate (heat delivered to the droplet surface) is determined by
integrating the energy equation between the droplet surface and free stream. This
is accomplished by integrating the energy equation between the droplet surface
and a general radial location which yields:
-4-rrr2 k ^ + N C T = N C" T, + Qn - Q, - N AH (34)
dr v pT v pT d XR xd v vapT
where the. average heat capacity of the gaseous mixture (C ) is:
. .
vaporizing
species
PT
N C" = N. (T (35)
v PT i 1 Pi
and the average heat of vaporization is:
vaporizing
species
N AH = 2 N. AH (36)
v vap_ i i vap.
and
Q = radiation heat transfer rate
R
4ir 3 dTd
Q, = -=- r. p. C —r-- - droplet heating rate f?7l
d 3 d d p, dt r ^'j
55
-------
Integrating the preceding modified energy equation between the droplet surface
and the gas stream yields the droplet heating rate:
Qd = QR
"(T
i-V
AH
vapT
|y - 1) ST .
(38)
where the outer boundary of the thermal film is determined from the heat transfer
Nusselt number by the empirical relationship (Ref. 110 ):
(39)
H
1 -
Nu
H
and the "heat transfer blockage term" is:
\ ^~
Z = —
(47rr
I- /-
*
(40)
Quasi-steady-state specie concentration profiles on the gas side of the droplet
are determined from the integrated species diffusion equation and are given by:
Nu
X. - N./N
i i v
X, - N./N,
X. - N./N
Xi
/ r,
'-T
(41)
and the film gas temperature profile is determined from the integrated energy
equation and is given by:
T - T,
T - T,
g
(e - 1)
exp
(42)
The model described above accounts for changes in droplet density, latent heat
of vaporization, vapor pressure, and average vapor thermal and transport proper-
ties that arise due to the more rapid gasification of the more volatile droplet
components during the course of vaporization.
56
-------
These equations are solved by a second-order explicit algorithm for the algebraic
and differential equations describing the vaporization rate/droplet composition
process and a second-order implicit algorithm for the equations that describe the
heating process. As part of the computer program output, descriptions are given
for the droplet vaporization rate, the average film (boundary layer) surrounding
the droplet, the droplet temperature, liquid composition, and droplet diameter
from ignition to final burnout. Particular attention has also been given to the
calculation of the average film thickness, fuel vapor diffusion rates through
the film, typical vapor residence times within the film, and film temperature
and composition profiles.
The main limitations of the droplet vaporization model in predicting droplet
vaporization and heating rates are: (1) spherical symmetry, i.e., the model equa-
tions were developed for nonconvective conditions, and convection is accounted
for by employing empirical Nusselt numbers; (2) the exclusion of chemical reac-
tions; and (3) the nonapplicability of the model at supercritical pressures. The
model should be used in the analyses of burners with high convective loading,
i.e., burners whose gas velocity is much greater than droplet velocities so that
burning does not occur in a flamefront surrounding the droplet.
Model Calculations. Calculations using the droplet vaporization computer program
were made with parametric variations in the initial fuel composition, initial
droplet size, surrounding bulk gas temperature and composition, and degree of
convection. In part, these calculations were made to check the operation of the
model and to determine the effect of using average properties for determining
droplet histories, rather than multicomponent properties. (Using average prop-
erties, if appropriate, would simplify the calculations.) An example of the
effects of average versus variable droplet compositions is presented below. These
calculations.were run at the following conditions:
1. Relative gas to droplet velocity: 1 ft/sec
2. Initial droplet diameter: 40 microns
57
-------
3. Gas temperature: 1000 R
4. Droplet composition:
a. Variable composition with two.species which have different
normal boiling points. Species selected were the upper and
lower boiling point fractions for kerosene.
b. Fixed composition with an average normal boiling point based
on the species used in 4a.
The droplet life histories calculated by assuming constant and variable droplet
composition are presented in Fig. 11 (upper two curves). Note that both approaches
(average and variable compositions) provide essentially the same results with re-
spect to the overall burning rate. This is somewhat surprising since the actual
composition of the droplet varies considerably, as shown by the lower two dashed
curves. This result, however, is not expected to be generally true for all spe-
cies since wider variations in properties can occur. Inspection of the lower
curves, however, shows that the composition of the vapors leaving the droplet
varies substantially with time. This illustrates the necessity of using a multi-
component model to determine specie histories. Droplet temperature-time histories
for the two cases are shown in Fig. 12. Note that the initial heatup rates are
essentially equal for both cases; however, the maximum temperatures attained differ
by as much as 75 R. This difference could be significant, depending on the sensi-
tivity of the kinetics employed in the kinetic/diffusion model. (The kinetic/
diffusion model requires the droplet surface temperature calculated by the droplet
model for a boundary condition.)
Droplet Flame-Front Model <
Description of Model. The model for the heating and combustion of multicomponent
fuel droplet surrounded by a flame front is depicted schematically in Fig. 13.
58
-------
in
«£>
100
80
-TOTAL MASS
to
CO
on
CO
U-
o
60'
HIGHER NORMAL
BOILING POINT \
SPECIE
\
\
\
\
\
\
\
\
20
\
•LOWER NORMAL
.BOILING POINT
SPECIE
\
\
1.0
VARIABLE
COMPOSITION
FIXED COMPOSITION
2.0
TIME, MILLISECONDS
Figure 11. Droplet Mass and Composition
-------
1000
<
(£.
P 700 -
600
500
1.0
VARIABLE COMPOSITION
FIXED COMPOSITION
1 1 1
1 11
2.0
TIME, MILLISECONDS
3.0
Figure 12. Droplet Temperature
-------
Oxidizer
Vapor
Product
Vapor
Figure 13. Schematic of Reactant Diffusion and Reaction in a Flame Zone Surrounding
a Fuel Droplet
-------
Included in the figure is a conceptual graph of the radial variations of the gas
temperature and mole fractions of selected species through the vapor film, flame
zone, and external mantle. Radial diffusion of the droplet vapors carries them
into progressively higher temperatures until they reach the flame zone (flame
front), where they are oxidized and burned.
Again, the basic equations solved are the species diffusion equation, •
2 dX. nt
N. + (4-rrr ) (ex/.) -r^- = X- S N (43)
j *• "V dr A] „£. m
species continuity equation,
1
= nR., n = 0 if r ^ r, ' . (44)
4irr dr 1 t
and the energy equation,
1 .jd_
4,r2 dr
-4irr k
2 , • dT
v • /
+ L N H \ = 0 (45)
dr ni - i m m
During the development and solution of the basic equations, the following assump-
tions were made:
1. The gas flow in the film surrounding the droplet is spherically sym-
metric and the film thickness is related empirically to the droplet
Nusselt number.
2. Pick's law is presumed valid throughout, implying equal binary diffusion
coefficients which are evaluated at mean film conditions.
3. The thermal conductivity and specific heats are constant and evaluated
at mean film conditions.
4. Temperature and concentration gradients within the droplet are ignored.
5. Fuel combustion reactions are assumed to occur in an infinitely thin
flame front at stoichiometric conditions with the products of combus-
tion determined by equilibrium calculations at the stoichiometric con-
ditions which include dissociation effects.
62
-------
6. The fuel is specified as a multicomponent mixture of compounds having
different properties, e.g., vapor pressure, molecular weight, etc.
7. Droplet temperature and composition varies with time'.
8. The static pressure is constant.
The key equations solved by the computer program are summarized below without
detailing their derivations. The derivations are similar to the vaporization
model deriviations presented in the preceding section except the integration has
been performed in two steps. First the equations are integrated between the drop-
let surface and the flame front and then they are integrated between the flame
front and the gas stream.
The total combustion rate, derived from the summation of the species diffusion
equations, is given by:
$
1 - X '
Vf
1 - X
vd
(46)
where vaporizing species
X = 2, X. (total vaporizing mole fraction) (47)
3 = (2/Num)/(l - ^-) (48)
.r_ = radial location of the flame front, and the subscript I refers to conditions
inside the flame front. (For most conditions, the mole fraction of the vaporizing
species at the flame front (Xv ) is zero.) The combustion rate of each of the
droplet component species is determined from their diffusion equations and is
given by:
N.
—- /X - X \ = X. /I - X \- X. (l - X \
N I Vd Vfj *d \ VfJ ^ \ V
63
-------
where the mole fractions of the vaporizing species (X^,.) at the flame front are
also normally zero. It should be noted that the molar flowrate inside the flame
front of species other than vaporizing species is identically equal.to zero, i.e.,
the boundary condition at the droplet surface for species other than vaporizing
species is that their flowrates are zero and the species continuity requires that
the flowrates remain zero inside the flame front.
The molar flowrate outside the flame front for each species is determined by the
chemical reactions:
F
m
- |V 1 CL ->• 5 (v , ) P.
I oxl 2 j ^ \ prod./ j
\ /m products \ ^ j'm J
3
(Note: v is negative)
\jj\
m
where F is the M vaporizing fuel species and P. is the j product of combustion
and is given by:
The total molar flowrate outside the flame front is given by:
N
and the oxygen flowrate is
N = v N_ . (53)
ox, ox TT ^ J
where
/1N i- \
(54)
VT = I I », M,
1 ^ -: J 4 \ M >
V1
i j "i
.N. x
V / 1I\ (55)
v = 2, ' - '
ox ^ ox. \ •
i i \N (
I
64
-------
The radial location of the flame front is determined by employing the preceding
expressions and film diffusion equations inside and outside the flame front and
is given by:
1
1
Vf\ (CA^X
) - M In
vd/ 3I
'\ - V/VT
— —
X°2£ " V°X ^
(6 - 1)
Where the mole fraction of oxygen at the flame front is assumed to be zero.
(56)
The droplet heating rate (heat delivered to the droplet surface) derived from the
film energy equation inside the flame front is:
(57)
/Nu \
d R a 1 v 2 I
r(Tf - Td> ^ap;
ai c"
(e - 1) %
I J
where
NT C
I I / 2
(58)
ZINUH
m
* ± I \ cp
H, i XI %
(59)
(60)
AH
vapn
y N. AH
^ iT vap.
i I *i
(61)
Similar expressions define the parameters a± and Z, outside the flame front.
65
-------
The temperature of the flame front is determined by equating the heat of combus-
tion to the change in the slope of the temperature (dT/dr) across the flame front,
i.e.,
where
= 47rr
2
dj\
dr/
- 4irr_ k. (£i-
f d> \dr
-V
(heat release rate
due to combustion)
E — v /N. - N. } H.
lL i[ ^ V ^
(heat transfer rate away from the
flame front due to conduction)
(62)
(63)
H.
1f
- /
C dT + AH°
298 C i 298i
(64)
Employing these relationships and film energy equations inside and outside the
flame front yields the flame temperature:
al \ \ , «*
e \ .
-------
components during the course of combustion. Again, a second-order explicit algo-
rithm is used to solve the albebraic and differential equations that describe the
combustion rate/droplet composition process, and a second-order explicit algorithm
is used to solve the equations that describe the heating process. As part of the
computer program output, descriptions are given for the droplet combustion rate,
the average film (boundary layer) surrounding the droplet, the droplet temperature,
liquid composition, droplet diameter, and the radial location and temperature of
the flame front from ignition to final burnout. Particular attention has also
been given to the calculation of the average film thickness, diffusion rates
through the film, typical vapor residence times within the film, and film temper-
ature and composition profiles.
The main limitations of the droplet flame front model in predicting droplet com-
bustion and heating rates are spherical symmetry, infinite fuel kinetics (flame
front), exclusion of nonfuel reactions, and the nonapplicability of the model at
supercritical pressures. The model should be used in the analysis of burners
with low convective loading, i.e., burners whose gas and droplet velocities are
approximately equal so that a flame can surround the droplet.
Model Calculations. Calculations using the droplet flame front model at the con-
ditions listed below have been made with parametric variations in the fuel type,
initial droplet size, surrounding bulk gas temperature and composition, and de-
gree of convection. These calculations indicate that the maximum fuel droplet
heating rate is approximately 10 to 80 R per microsecond for a 40-micron droplet.
The higher heating rates are associated with high initial gas temperature and
convective conditions. Radiation was found to have considerably less influence
than these parameters. Under most flow conditions, the maximum heating rate is
located at a position which corresponds to low droplet temperatures and low
vaporization rates.
67
-------
An example of droplet calculations using the flame front model are presented in
Fig. 14 and 15. These calculations were run at the following conditions using
fuel grade No. 2:
1. Relative gas to droplet velocity: 10 ft/sec
2. Initial droplet diameter: 40 microns
3. Gas temperature: 2000 R
4. Oxygen mass fraction: 0.15
5. Radiation heat transfer: zero
Since fuel oils are a distillate or residue of petroleum hydrocarbons, the fuel
oils are actually composed of many different species which have widely different
properties. Composition and physical properties of the species as functions of
temperature are not, however, directly available and, therefore, properties
required by the combustion model have been approximated. In any case, distilla-
tion curves and physical properties of hydrocarbons are available and the hydro-
carbon properties are generally "smooth" functions of the number of carbon atoms
in the compounds.
Known values (Ref. Ill) of critical pressure and normal boiling temperature for
hydrocarbons are plotted in Fig. 16 and 17 as functions of the number of carbon
atoms (n) in the compounds. Boundaries were drawn through the data and extended
up through n = 40 to estimate the required properties.
Employing known distillation curves (Ref. 112), fuel oil compositions and proper-
ties were determined using the following technique:
1. Fuel composition: the 0-, 10-, 50-, 90-, and 100-percent distillation
temperatures were used to define the normal boiling points of five
"species."
2. Species carbon content: determined using the normal boiling point
and Fig. 17.
68
-------
ID
1100
1000 —
DISTILLATION CURVE
FLAME FRONT MODEL
VAPORIZATION MODEL
20
kQ 60
PERCENT BURNED
80
100
Figure 14. Droplet Temperature
-------
1.0
TIME, MSEC
1.5
2.0
Figure 15, Vaporization Rate, Flame-Front Model
-------
8 10 15 20
Number of Carbon Atoms
30
40
Figure 16. Critical Pressure for Hydrocarbons (Ref. Ill)
71
-------
1000
800
600 —
o
a.
o
CO
O
z
400
200
8 10 15 20
Number of Carbon Atoms
40
Figure 17. Normal Boiling Point for Hydrocarbons (Ref. Ill)
72
-------
3. Molecular weights defined by:
MW. = (m)(MW ) + (2m + 2) MWR (66)
4. Vapor pressures defined by the normal boiling point and the techniques
outlined in Ref. 113, curve-fitted to:
£n P = A + B/T (67)
i
5. Critical pressures defined using the number of carbon atoms and Fig. 16 .
6. Heat of vaporization, defined by the techniques outlined in Ref. 114,
curve-fitted to:
f r T 1 °'38
AH = AH L " !— (68)
V VNBP [ " NBP J
Shown in Table 3 is the final compositions and properties that have been selected
to simulate fuel grade No. 2.
Examination of Fig. 14 and 15 shows that upon injection of the droplet into the
gas stream, the combustion rate is very low and the heatup rate is very high.
Also, the flame front is located close to the droplet surface and the "flame tem-
perature" is initially very low due to the low incident rate of fuel vaporization.
As the droplet temperature increases, the combustion rate of the lower normal
boiling temperature species increases, the flame front moves away from the drop-
let surface and the flame temperature increases. This increase in flame tempera-
ture causes the heating rate to increase until the droplet temperature approaches
the "wet bulb" temperature. Upon approaching the "wet bulb" temperature, the
droplet heating rate decreases rapidly. As the combustion processes proceed from
initial conditions to burnout, the droplet composition gradually changes to higher
normal boiling point species and the droplet temperature continues to increase.
73
-------
TABLE 3. ESTIMATED PROPERTIES FOR FUEL GRADE NO. 2
Specie
1
2
3
k
5
Mass
Fraction
0.15
0.20
0.30
0.20
0.15
Molecular
Weight
102
123
163
19**
266
Critical
Pressure,
atm
33
28
23
20
15
Normal
Boi 1 ing
Point, F
212
292
405
478
625
A*
11.012
11.503
11.855
12.401
13.765
B*. R
-7,400
-8,650
-10,255
-11,632
. -14,935
J-J.
AHvNBp""'
Btu/lb
132
124
112
106
91
«t. j.
C"", F
525
599
. 716
777
891
*ln Pv = A + B/T
**AHV = AHVNBp
10.38
NBP
-------
Coal Particle Model
Description of Model. The model for analyzing heterogenous combustion and devol-
atilization of coal particles is depicted schematically in Fig. 18. It should be
noted that the model does not include reactions in the film surrounding the
particle.
The basic equations solved are the species continuity equation in the film sur-
rounding the particle:
1 dNi
2 "dF = Ri = °> <69)
4irr
species diffusion in the film;
7 - dXi ^ •
4irr (c<#) -jfir = *i Z N., (70)
the energy equation in the film:
_J_ * Lur2 k ^ + I N. H.) = 0, (71)
2 dr l J
4irr
the particle continuity equation:
T rs 6c) = -"H (47T rs^ = -*c' and (72)
the devolatilization rate for the jth volatile component:
16
d [f 4ir 3! D /. 4ir 3\ A | Vj) ,?,
zr K. Trsr -Rv. K. Trs -NC \r7 = -Nv. (73)
L J J J \ 3 / V c . J
(Total rate change (Volatile material (Volatile ma-
of volatile mass) rate loss due to terial rate
devolatilization loss due to
reactions) heterogeneous
combustion)
75
-------
FILM
BOUNDARY
(FILM) \ (COMBUSTION GASES)
\
\
\
M PRODUCTS DUE TO HETEROGENEOUS COMBUSTION
I
I
I
I
I
M
OXYGEN
M VOLATILE MATERIAL AND/OR PRODUCTS
OF HOMOGENEOUS COMBUSTION
Figure 18. Schematic of Particle Burning Process
76
-------
where 6. is the density of the ith particle component based on total particle
volume, and the -particle energy equation:
dT
p (74)
(Heat trans- (Rate change
fer rate to of particle
the particle) temperature)
During the development and solution of the basic equations, the following assump-
tions were made:
1. The gas flow in the film surrounding the particle is spherically symmetric
and the film thickness is related empirically to the particle Nusselt
number.
2. Pick's law is presumed valid throughout, implying equal binary diffusion
coefficients which are evaluated at mean film conditions.
3. The thermal conductivity and gaseous specific heats are constant and
evaluated at mean film conditions.
4. Temperature and concentration gradients within the particle are ignored.
5. The gases are assumed to diffuse through the boundary layer (film) sur-
rounding the particle without undergoing pyrolysis or combustion
reactions.
6. The coal is specified as a mixture of coal and volatile material having
different properties.
7. Particle temperature and composition varies with time.
8. The static pressure in the radial direction is constant.
9. Pore combustion is neglected but heterogeneous combustion of the coal at
the particle surface is included. (This assumption has been modified
under IR§D funding and the resulting coal model is presented in Appendix I.
77
-------
Noting that the species molar flowrates are constant (application of the species
continuity equation), the species diffusion equation is integrated5 between the
particle surface and a general radial location to yield the mole fraction profile:
~XA
[\\
- N.
- N.
/Nu
= B 1 '
S\2
(75)
where
6 =
47rr
Num
(76)
and the molar flowrate of compound i is:
N
N. = a. N + Z a. . N
1 1,C C 1,J
(77)
where a^ is the stoichrometric coefficient of compound i for the heterogeneous
i, c
reaction and a." is the stoichrometric coefficient of compound i for the products
i> s
of combustion of the jth volatile material. The N .'s include all products of
combustion and the volatile species. Therefore, the program has.the option of
reacting or not reacting the volatile material at the particle surface.
For a radius equal to the outer boundary of the film surrounding the particle,
i.e.,
r =
1 -
(78)
Nu
m
the mole fraction of compound i at the particle surface is determined and is:
N.
N.
(79)
78
-------
The particle heating rate is determined by integrating the film energy equation
between the droplet surface and the free stream and is:
nt *t Cpt (Tg - V
Qs = QR + QC + .\ V + TIT— (80)
i=l i (e -1)
where
Q = radiation heat transfer rate
K
Q = heat of combustion due to heterogeneous reactions
0 = heat of combustion due to volatile combustion
(82)
2irr k NuH
and the outer boundary of the thermal film is determined by the empirical
relationship:
(83)
JH
Quasi-steady-state film gas temperature profile is determined from the integrated
energy equation and is given by:
• •
[• (?) (• - ?)] -j
To apply the preceding expressions in a computer model, the heterogenous surface
reaction and devolatilization rates are required. The von Fredersdorff and
Elliott (Ref. 115) approach to heterogeneous combustion at the particle surface
is presently being used in the model. This involves a complex multistep
79
-------
[ki
4
mechanism of gaseous adsorption, reaction, and desorption on the solid surface.
In the process, different regions are assumed to exist simultaneously on the sur-
face. Partial surface oxidation occurs at active sites (C-.) to form oxygen-
occupied sites (C ) according to the surface complex formation reactions:
kl
Cf + °2 — - Co + ° (85)
k2
Cf + 0 — t- CQ (86)
Cf + C02 T^- CO ••- C (87)
k4
Cf + H20 TT-^- CQ + H2 (88)
k6
a gasification reaction:
CQ -L*. CO + Cf (89)
describes desorption of CO from the surface.
For steady state conditions:
(C0) - (Cf) - 0 C90)
and the reaction rate of carbon (n ) is:
= (Ct) (02) + k2 (0) + k3 (C02) + k5 (H0) (91)
|
80
-------
where
(Ct) = (CQ) + (Cf) (92)
Here, as in earlier models, the temperature-dependent surface reaction rates must
be determined from experimental data. Equilibrium among oxidized carbon species
and H- is assumed to take place in the gas phase, well away from the fuel surface.
The Howard and Essenhigh (Ref. 116) approach to devolatilization is employed in
the combustion model. Pyrolysis rates for coal have been analyzed by several
investigators (Ref. 116 and 117) and can be represented by:
(93)
.
dt o. i
where
k = kinetic reaction rate of ith component
The k0.'s are temperature-dependent reaction rates which can be determined from
experimental data.
The model described above accounts for heterogeneous combustion at the particle
surface, devolatilization, particle temperature, and composition. A second-order
implicit algorithm is used to solve the algebraic and differential equations that
describe the heterogeneous reaction/devolatilization/particle composition process
and the particle temperature. A.s part of the computer program output, descrip-
tions are given for the heterogeneous reaction rate, devolatilization rates, the
average film surrounding the particle, the particle temperature, composition, and
radius from ignition to final burnout.
The main limitations of the model in predicting coal combustion are spherical
symmetry, the exclusion of pore combustion, and the exclusion of chemical reac-
tions in the film surrounding the particle.
81
-------
Model Calculations
Life history calculations for burning coal particles have been made with this com-
bustion model to compare its predictions with some available laboratory experi-
mental data. Results of some of these calculations are presented in Fig. 19
through 21. Calculated particle composition histories are compared in these fig-
ures with experimental data from Ref. 116 (mass median particle radius for the ex-
periments was 15 microns). Inspection of the initial time region of the volatile
matter and fixed carbon content plots, (Fig. 19 and 20) suggest that during the
first 0.05 second, volatile matter (Fig. 19 ) is evolved without heterogeneous com-
bustion. This is suggested by the fact that no fixed carbon content is lost in
the initial 0.05 second, while the volatile matter content drops about 5 percent
during this time period. The heterogeneous combustion then begins and, thereaf-
ter, devolatilization proceeds simultaneously with surface combustion. Inspection
of the slopes of the calculated curves in Fig. 21 suggests that volatile matter
loss due to heterogeneous combustion contributes strongly to overall particle de-
volatilization. The increase in slope suggests that once the temperature of the
particles becomes sufficiently high, both the principal devolatilization reactions
and the heterogeneous combustion reactions are very fast and consequently, the
loss of volatile matter by gaseous evolution is relatively faster than the fixed
carbon loss due to heterogeneous combustion.
Inspection of Fig. 19 and 21 shows that after about 80 to 90 percent of the vola-
tile matter is driven off (0.2 second), devolatilization becomes very slow and the
remainder of the volatile material is simply burned off with fixed carbon by het-
erogeneous combustion. Examination of Fig. 20 reveals that, when compared with
the data of Ref.116, the reaction rate given by Ref. 118 appears to be too high
and the reaction rate given by Ref. 119 is probably too low. However, since
these calculations were performed using a monodisperse particle distribution, it
is impossible at this time to determine from them the validity of: (1) the reac-
tion rate, or (2) the fixed carbon model. The results do, however, lend confi-
dence to the overall model approach in that they agree fairly well with the trends
indicated by the coal particle combustion experiments.
82
-------
100
CD
80
LJ
° - X
<-> z 60
§ £
•> ^n
20
0.0
•EXPERIMENTAL
POINTS (REF. 116)
•PARTICLE SIZE
15 RADIUS:
0 RADIUS
CALCULATED.
_L
0.4
TIME, SECONDS
Figure 19. Volatile Matter Content
0.8
•EXPERIMENTAL
POINTS (REF. 116)
REF. 112 RATES
REF. 113 RATES
CALCULATED
0.0
0.4
TIME, SECONDS
Figure 20. Fixed Carbon Content
83
-------
z
o
CQ
CC
O
o
UJ
X
100
80
60
20
o
o
• EXPERIMENTAL
POINTS (REFP 116)
PARTICLE SIZE
RAD I US
100
80
60
20
FIXED CARBON CONTENT, WEIGHT PERCENT
OF ORIGINAL
Figure 21. Variation of Composition of Solid
Matter With Degree of Burnout
84
-------
Kinetic/Diffusion Model
Description of Model. The model for the calculation of molar production rates
and molar concentrations of all compounds in the film surrounding a droplet (par-
ticle) is depicted schematically in Fig. 22. Basic assumptions in the derivation
of the model equations are: (1) spherical symmetry, and (2) the outer boundary of
the film is determined by empirical Nusselt number correlation.
The basic equations solved are the species continuity equation:
dN.
2- =
A 2 dr
4-rrr
= R. (94)
and the Stefan-Maxwell (diffusion) equation:
dX. nt
Z
K~ A
where
N.-Xj Z
j,k/
This is similar to the diffusion equation used in the droplet vaporization and
flame front models except nonequal binary diffusivities are employed.
The energy equation:
85
-------
OO
Product
Vapor
Oxidizer
Vapor
Figure 22. Kinetic/Diffusion Model
-------
where
H = f C_dT+AH°no (98)
«A}o
k k
For the general chemical reaction:
nt k nt
t f *£ t +*
I v. . M. „ *" E v. . M. (99)
j-1 J'1 1^~ M Jjl J
.* •»*•
where v. . and v. . represent, respectively, the stoichiometric coefficients of
the reactants and product:
forward reaction rate is:
the reactants and products for the ith reaction and the chemical species M., the
a. -AE. I
k, = B. T X exp -==M (100)
±i i L RI J
and the reverse reaction rate is:
kbi = kfi/Kc. (101)
where B. is the frequency factor, a. is the temperature exponent, AE- is the acti-
vation energy, and KC. is the equilibrium constant. The equilibrium constant is:
AH° AS°
Hn Kc = -(taRT) An..--^+ -^- (102)
i
An. = E (v'' - v' .) (103)
= -1' -1'
nt .
AH° = Z (V." i - v. i) H° (104)
1 j=l J» ^ mj
87
-------
., ,,.
j j
where Hm_ and Sm. are, respectively, the enthalpy and entropy of the chemical spe-
cies m. at 1 atmosphere.
Employing the preceding expressions, the total molar reaction rate for the jth
species is given by:
(106)
vk
k£ (C) TT
IT
ck a * uk £
where v, is the third-body stoichiometric coefficient for the kth reaction.
At this time, it is convenient to express the preceding equations in a transformed
coordinate system. This is done to aid in the numerical computational procedure.
The transformed coordinate system is:
= -- (1 - -) (107)
since the outer boundary of the film is given by:
T, = —-5— (108)
0 1-2/Nu
The transformed coordinate £ varies from 0 to 1 for the domain of interest. Ap
plying the preceding transformation to the basic equations yields the following
set of transformed equations:
1. Species continuity equation:
dN. R-
,,rt^
(109)
[
88
-------
2. Species diffusion equation:
dX.
(4TOS) (c.^^) -^- = X^ E Nk - N ' (no)
3. Energy equation:
(111)
E N,
i
k
dN,
-EH, -r- = 0
k
To numerically solve the preceding equations, the species continuity and diffu-
sion equations are combined to yield the modified continuity equation:
• * N.
k kJ
-v lifE^lM^m-J!!—r-o (112)
E-
The solution procedure presently employed by the kinetic/diffusion model to solve
the preceding nonlinear equations consists of two second-order implicit algorithms.
The first algorithm couples the modified.continuity equations for the fuel com-
pounds and the oxygen together with the energy equation to calculate the oxygen
and fuel mole fraction profiles and the film temperature profile. For the other
species, the second algorithm is used in a decoupled mode to calculate the mole
fraction profiles employing the modified continuity equation. After all specie
mole fraction profiles have been updated, the program calculates molar flowrates
based on the species continuity equation.
89
-------
The kinetic/diffusion model includes the kinetics and diffusion of all compounds
and intermediate species in the film surrounding a droplet or particle. Specifi-
cally, it accounts for finite-rate combustion of mixtures of hydrocarbon compounds,
some of which may contain fuel nitrogen, and includes detailed kinetics and the pro-
duction rates of fuel NO , thermal NO , and other pollutants and combustion prod-
X X
ucts. The model input is structured such that quasi-global kinetics for hydrocarbons
can be used if detailed rates are unavailable. The model solves simultaneously the
energy equation, species continuity, and species diffusion equations using boundary
conditions at the droplet (particle) surface, and the temperature and molar concen-
trations in the bulk gas stream.
Boundary conditions at the droplet (particle) surface can be obtained by execut-
ing the droplet vaporization, droplet flame front, or coal combustion models.
The main limitations of the kinetic/diffusion model is spherical symmetry and
large computer execution times (i.e., ~ 2-3 min./calculation/number of axial
locations). It is the main program for predicting pollutant formation rates in
the film surrounding the droplet (particle).
Model Calculations. A limited number of preliminary calculations have been made
using flame front model output data, selected fuel nitrogen compounds, and reaction
rate data to input the kinetic/diffusion model. Using the reactions presented in
Table 4, typical model calculations for a 40-micron droplet of No. 2 fuel oil indi-
cated that the maximum production rate of thermal NO occurred on the oxygen-rich
side close to the flame zone but that a substantial amount of NO was also produced
on the fuel-rich side due to atomic oxygen and hydroxyl radical overshoot in addi-
tion to penetration of those species through the flame zone (Fig. 23). The cal-
culated production rate of thermal NO under the conditions of Fig. 23 was only
about 0.7 gram NO per kilogram fuel. This low value of thermal NO production rate
was mainly controlled by the flame zone temperature which was 2080 K. The flame-
zone temperature was lower than the equilibrium flame temperature due to conductive
heat loss from the flame zone to the free stream. This heat loss was high because
the flame zone is very close to the outer boundary of the gas film (boundary layer).
It should be noted, again, that this model does not include continued production
of thermal NO in the gas phase.
90
-------
TABLE 4. HYDROCARBON AND THERMAL NO REACTIONS
1. CO + OH = C02 + H
2. 02 + H2 = 2 OH
3. OH + H£ = H20 + H
k. 02 + H = OH + 0
5. 0 + H2 = OH + H
6. 0 + H0 = 2 OH
7. 2H+M=H2+M
8. 20+M=02+M
9. 0+H+M=OH+M
10. H + OH + M = H20 + M
11. NO+M=N+0+M
12. N2 + M = 2N + M
13. N2 + 0 = NO + N
Ht. N + 02 = NO + 0
15. N + OH = NO + H
16. CJL + f£ + £)o2-*.n CO +f H20
91
-------
1.0
.8
.4
.2
.1
.08
.04
.o:
.01
.008
o
H
U
oi
u.
U-i
I .004
.002
.001
.008
.004
.002
.001
NO. 2 FUEL OIL
A VELOCITY = 10 FPS
1(1) NITROGEN-CONTAINED
Cn"m
1
2 3
RADIUS/DROPLET. RADIUS
Figure 23. Preliminary Kinetic Diffusion Model Results
With Flame-Front Model Input
92
-------
Fuel NO production rates were examined briefly by adding to the reactions in Table
4 the reactions listed in Table 5. The HCN path to fuel NO was included because
nitrogen in some compounds can be converted almost completely to HCN during high-
temperature, nonoxidative pyrolysis.
TABLE 5. REACTIONS FOR FUEL NITROGEN*
1.
2.
3.
A.
5-
6.
7.
8.
HCN
HCN
CN
CN
CN
HCN
C ,
n
C ,
n1
+ M
+ 0
+ o2
+ NO
+ CN
+ H
Hm'Ni
m
H ,N,
m1 1
= CN
= OH
= CO
= CO
p
~ L2
= H2
= c
<
/i
< \
+
+
+
•f
+
+
(
I ^
2
H + M
CN
NO
N2
N2
CN
. vH, , . \ + k(HCN)
"" K/ \nl ~ K/
+ m')o =n'CO + — HO+-N
H / 2 2222
With the Table 4 and 5 reactions and a vaporizing fuel mixture containing 1-percent
(by mass) nitrogen, the calculated fuel NO production rate was approximately 7.4
grams NO per kilogram fuel. That corresponded to a fuel nitrogen to NO conversion
efficiency of approximately 34 percent for the fuel grade No. 2 droplet.
These preliminary analysis results are indicative of the capabilities which the
kinetic/diffusion model offers to a more complete pollutant production analysis.
*As discussed previously, these fuel-N reactions were used to test the operability
of the model. Other reaction schemes will be included as they are shown to be
important. A more plausible mechanism for fuel NO formation is proposed in the
Phase II discussion of the burner results.
93
-------
PHASE IB - FUEL AND MODEL COMPOUND DECOMPOSITION STUDIES
INTRODUCTION
Laboratory and field studies have been conducted to investigate conditions for the
conversion of nitrogen compounds to NO during combustion. Prior to this contract,
JC
only a limited amount of effort had been expended, however, in investigating the
pyrolysis of fossil fuels and nitrogen compounds under inert or oxidative condi-
tions representative of the initial stages of combustion. The previous pyrolysis
studies (e.g., Ref. 92)did not attempt to determine the major nitrogen-containing
species that were formed.
The objective of this task was to conduct thermal decomposition studies at ele-
vated temperatures to investigate the fate of nitrogen bound in fossil fuels
(fuel-N) under conditions relevant to those existing in the initial phases of the
combustion process. Pyrolysis experiments were conducted with model fuel-nitrogen
compounds to measure the kinetic parameters that determine under what conditions
(i.e., at what point in the flame) typical fuel nitrogen structures will decompose,
and to identify the nitrogen-containing species that are formed. Later in the pro-
gram, fuel oils and coals were pyrolyzed under conditions similar to those employed
with the model compounds. The rationale for employing model compounds in part of
the study was that the nitrogen exists in so many forms and molecular weight dis-
tributions in fossil fuels that the identification and direct study of actual fuel
nitrogen compounds would not be feasible. However, it has been established that
most fuel nitrogen exists in heterocyclic ring and nitrile structures.
As representative of compounds having these structures, the model compounds pyri-
dine, quinoline, pyrrole, and benzonitrile were chosen for study. These lower
95
-------
molecular weight analogs of the actual fuel nitrogen compounds have sufficient
volatility to allow them to be pyrolyzed in the vapor phase. The side groups
present in the natural fuel nitrogen compounds may decompose at much lower tem-
peratures than the very thermally stable heterocyclic ring structures. However,
it is this final step in the pyrolysis process (i.e., the scission of the hetero-
cyclic ring) that should determine the fate of the fuel nitrogen as to whether
the intermediate species that are formed will eventually result in the formation
of CN, NHX, or N2. The distribution of the fuel nitrogen among species of this
type could determine the fraction of fuel nitrogen that is converted to NOX in
combustion. Burner studies (e.g., Phase IIA of this report) have indicated that
HCN, (CN)2, and NHj all form NO readily in hydrocarbon combustion. Therefore, .it
was of particular interest whether conditions could be found that would promote
the direct formation of N2 in the initial pyrolysis process.
After the major features of model compound pyrolysis had been established, fuel
oils and coals were pyrolyzed under similar inert pyrolysis conditions, and the
nitrogen-containing inorganic products were measured and compared with those
formed from the model compounds. In a fuel matrix consisting of mainly high boil-
ing hydrocarbons, a nitrogen compound could undergo a different mode of pyrolysis
because of chemical interactions and/or changes in its temperature-time profile.
Also, whether the compound decomposes in the vapor or liquid phase will depend on
its volatility and thermal stability.
Even though the N-heterocyclic compounds, in general, have good thermal stability,
the thermal stability of individual compounds will vary over a wide range. Com-
pounds which have long side chains will usually start to decompose at lower tem-
peratures than unsubstituted heterocyclics. Depending on the point of bond
cleavage, volatile or nonvolatile compounds may be formed which may or may not
contain nitrogen. Thus, a study of the pyrolysis of fuel compounds can be quite
complex.
Because of the host of possible volatile organic compounds that might be evolved from
fuel pyrolysis, emphasis in the fuel experiments was placed on the measurement of
96
-------
the inorganic nitrogen-containing decomposition products under conditions where
most of the organic species would have decomposed. The organic pyrolysis pro-
ducts were determined, however, for the model compounds. The .material which
emerged as the principal volatile nitrogen-containing degradation product from
model compounds, fuel oils, and coals was HCN.
Inert pyrolysis was emphasized in this initial pyrolysis study for two reasons:
(1) in the case of heterogeneous droplet and particle combustion, the nitrogen
species could undergo decomposition before they have approached close enough to
the flame front for significant oxygen to be present; and (2) secondary processes
may occur in the presence of oxygen that could mask the nature of the initial
nitrogen-containing pyrolysis products that are formed.
97
-------
PHASE IB: EXPERIMENTAL
A schematic diagram is shown in Fig.24 of the experimental system used in the
pyrolysis studies. Flow techniques were employed because this permitted the re-
action times of volatile species to be kept quite small and facilitated the
quenching and analysis of the pyrolysis products (e.g., by flowing through a bub-
bler to remove HCN or NHj or into a gas chromatograph to measure N2 or various
organic pyrolysis products). Therefore, the solid arrows in Fig. 24 represent
the flow of helium carrier gas streams. During oxidative pyrolysis experiments
(conducted with model compounds only) and reactor burn out, the carrier gas con-
tained up to 20 percent oxygen. A description of the various test procedures and
analytical methods follow.
Pyrolysis Samples
All of the model compounds were purchased from commercial sources, and were used
as received. Petroleum samples were supplied to the program by the EPA, Gulf Re-
search and Development Co., Bureau of Mines-Laramie Energy Research Center,
CONOCO Oil Co., and Ultrasystems Inc. Coal and coke samples were supplied by the
EPA, the International Flame Research Foundation, Bureau of Mines, Bruceton, Pa.,
and Babcock and Wilcox Co. The coal samples were ground before use. The fuel
samples used are described in Table 6 and their reported analyses listed.
TABLE 6 . FUEL SAMPLES
Fuel Type
No. 6 Fuel Oil
No. 6 Fuel Oi 1
No. 6 Fuel Oi 1
No. 6 Fuel Oil
No. 6 Fuel Oil
No. 6 Fuel Oil
Wi Imington Crude
Coal
Coal
Suppl ier
Gulf
Gulf
Gulf
CONOCO
EPA
Ultrasystems
Bureau of Mines,
Laramie, Wyoming
EPA
IFRF
Crude Source
Venezuela
Various Crudes
Mainly California Crudes
--
--
--
—
Analyses,
percent
N
O.M
O.lilt
l.<4l
0.3
0.22
0.38
0.63
1.17
1.8
S
2.3
0.73
1.63
0.66
0.9
0.33
1.59
3.0
--
98
-------
GC COLUMN
FOR N2
(MOL. SIEVE)
THERMAL
CONDUCTIVITY
DETECTOR
He OR i
He/0, i
FUEL
PYROLYSIS
REACTOR
(QUARTZ)
O
FURNACE
to
to
He Oft
He/02|
He OR
He/0,
He OR i
He/0, |
H^
CATALYTIC
AMMONIA
CONVERTOR
FURNACE
INTEGRATOR
RECORDER
PRINTER
MONOPOLE
MASS
SPECTROMETER
I
FLAME
IONIZATION
DETECTOR
COLORIMETRIC
ANALYSES
FOR
HCN, NH,, NO
HELIUM
SEPARATOR
FLOW
SPLITTER
SATURATOR
bt
k.
VAPOR
INJECTOR
VALVE
f
I
1
MODEL
_CO_MPOUN_D
REACTOR
(QUARTZ)
GC
COLUMN
(CHROMASORB
103)
VENT
LIQUID
INJECTION
PORT
FURNACE
Figure 24. Schematic of Apparatus
-------
Sample Measurement
The model compounds that were injected as liquids* were measured by. means of 1-
microliter Hamilton microsyringe (Model N7100). Calibration experiments estab-
lished that 0.2-yl samples (the size usually employed) were reproducible to bet-
ter than ±2 percent (usually ±1 percent). The syringe (with tip up) was filled
with sample contained in an inverted serum bottle. The plunger was pumped several
times to expel any air and then drawn back to the 0.9-yl mark. After removing the
syringe from the serum bottle (still in a vertical position), the plunger was
moved to the 0.2-yl mark. The syringe was then inverted and the excess liquid re-
moved from the exterior of the needle by wiping it against clean glass wool.
There were indications that the use of a more absorbent material would draw some
of the sample from the tip of the needle. The sample was immediately injected
into the heated sample port. In a continuous series of experiments, the syringe
was kept filled with sample between experiments and stored with the tip up and in-
serted through the septum of the inverted serum bottle. This prevented air bub-
bles from becoming trapped in the syringe.
The fuel oils, being more viscous than the model compounds, and the coal samples
were weighed directly using a Mettler microbalance. The technique used for fuel
oils was to dip a nichrome wire into the oil, weigh the wire and the drop of oil
to the nearest yg, and transfer the oil sample (1 to 2 mg) to the fuel reactor's
sample holder (a quartz boat). The wire was reweighed and the sample's weight
determined by difference. The EPA "in-house" No. 6 fuel oil was of such low vis-
cosity that a larger diameter wire had to be used to hold a 1-mg drop.
In the case of the coal samples, an aluminum weighing pan that contained a V-shape
piece of aluminum foil (holding several milligrams of the coal sample) and a ni-
chrome wire was tared. The coal sample holder and wire were handled with forceps
and a sample of the coal was transferred into the quartz boat with the boat posi-
tioned over the pan. This transfer was accomplished by scraping the coal into
^Measurement of samples introduced as vapor will be described later.
100
-------
the boat with the nichrome wire. The quartz rod attached to the boat was gently
vibrated by means of a file to remove any coal adhering to the outside of the
boat. The pan, containing the foil, wire, and excess coal, was then reweighed
and the amount of coal transferred was determined by the weight difference.
In general, coal samples ranging from 1.5 to 2 mg (weighed to the nearest yg) were
used for the pyrolysis studies. To obtain a constant weight for the pan, it was
necessary to wait about 15 minutes after transfer of the coal sample to the quartz
boat. Since constant weight was immediately obtained when the aluminum pan and
coal sample were intially weighed, the time required to obtain constant weight
after transfer of the coal is attributed to static electricity and not, for
example, to absorption of moisture from the atmosphere.
Apparatus and Procedures-Model Compound Pyrolysis
The model compound apparatus is shown schematically at the bottom of Fig. 24 and
in two photographs (Fig. 25 and 26). The liquid model compound sample was vapor-
ized in a helium carrier gas stream at an inlet system temperature of 160 C. The
carrier gas then flowed through a 15-inch quartz reactor, with a nominal ID of
2 mm, that was heated at temperatures up to 1100 C in a temperature-controlled
electric furnace. After passing through the pyrolysis reactor, the effluent gases
were then analyzed for organic products by means of a gas chromatograph using a
flame ionization detector. The analytical techniques used for analyses of the
nitrogen-containing inorganic products are described in a subsequent section.
The furnace was 12 inches in length, with a quartz-to-metal graded seal attached
at each end of the quartz reactor tube. Pt-Pt/Rh thermocouples were attached to
the exterior of the reactor to measure the temperature and as sensors for the
temperature controller.
The entire model compound flow system, from the sample injectors to the GC column,
was constructed of 1/16-inch OD., stainless-steel tubing so that sharp GC peaks
could be obtained. The number of T-connectors and valves was kept to a minimum
101
-------
5AA21-4/7/75-C1D
Figure 25. Model Compound Apparatus
-------
o
W
5AA21-4/7/75-C1A
Figure
26. Model Compound Apparatus (top view with furnace open and
top removed from GC column oven)
-------
for this same reason. The T's were made from drilled rods for minimum volume.
The entire system was silver-soldered and small-volume bellows valves were used,
giving a leak-free system. The system upstream of the reactor, including the
sample injection devices, was contained in a small oven heated to 160 C. The
downstream tubing and valves were heated to 180 C until the GC or chemical analy-
sis train was reached.
Two-stage pressure regulators, set at 60 psig, were used to give constant carrier
gas flowrates. The flowrates were held constant by the use of flow regulators,
upstream of the sample injection ports, of the type that give flowrates indepen-
dent of the downstream pressure. The pressure in the reactor, from 15 to 20 psig,
was determined by the temperature of the GC column during sample injection (nor-
mally 60 C) and the carrier gas flowrate which was held at 43 cc/min, measured at
room temperature and pressure, except in the first experimental series in which
it was 50 cc/min. Calibrated gases containing 5- and 20-mole percent C"2 in helium
were obtained from Matheson for use in the oxidative pyrolysis experiments and for
reactor burnout between experiments.
Two methods were employed for vaporizing the model compound samples and introduc-
ing them into the, carrier gas stream. The larger samples were introduced by the
direct injection of a 0.2-yl liquid sample into the heated carrier gas stream
through a silicone rubber septum. This heated liquid sample injection port con-
sisted of two concentric metal tubes with the inner tube ending about 1/8 inch
from the septum. The septum seated against the outer tube which was threaded to
hold the septum cap. The carrier gas flowed toward the septum in the outer tube
and then down the center tube. During sample injection, the syringe needle was
inserted about 2 inches into the center tube and the sample expelled. The center
tube was packed with glass wool to wipe the sample drop from the needle and to
ensure rapid vaporization of the sample. Sharp GC peaks were obtained even for
the high-boiling liquid qiiinoline.
In some of the experimental series, pyridine was prevaporized and then introduced
into the reactor carrier gas stream by use of a Caryle valve. In this "vapor
104
-------
injection" technique, two loops of coiled stainless-steel tubing were attached to
the Caryle valve, each having a nominal volume of 1 cc. Carrier gas (with or
without oxygen) saturated with pyridine vapor at 0 C flowed through one loop (and
was vented in a hood) and the reactor carrier gas stream flowed through the other.
When the loops were suddenly interchanged by turning the Caryle valve, in which a
spring-loaded teflon plate slides against a polished metal plate, a 1-cc slug of
carrier gas containing pyridine was introduced into the reactor carrier gas stream.
In the model compound pyrolysis experiments, the carrier gas flowed continually
either through the GC column, which was held at 60 C, or to the analysis train
through a needle valve to give the normal backpressure in the reactor of 17 psig.
In the experiments in which the GC was used, all or part of the carrier gas stream
would flow to a H2-02 flame ionization detector. A measured fraction of the
stream could be directed to a monopole mass spectrometer for peak identification
or to an analysis train (bubbler). One minute after sample injection, tempera-
ture programming of the GC column was initiated. The peaks were plotted on a re-
corder and the peak areas were measured and printed by use of an automatic inte-
grator. The reactor was "burned out" after each experiment with an CL-helium
mixture that was vented and the GC column was back flushed.
In the oxidative pyrolysis experiments, two helium carrier gas streams were used.
With liquid sample injection, each stream flowed at 20 cc/min and they merged at
a point between the sample introduction port and the inlet to the reactor. This
was to ensure intimate mixing of 02 with the sample. The stream introduced down-
stream of the port contained either 5- or 20-percent 0 . Thus, the 0_ concentra-
tion at the reactor entrance was either 2.5- or 10-mole percent. In the vapor
sample oxidative pyrolysis experiments, the stream saturated with the pyridine
vapor and the carrier gas stream each contained 5-percent oxygen. In the oxida-
tive pyrolysis experiments, the carrier gas was switched back to helium 15 sec-
onds after sample injection so that the GC column would not be exposed to C>2 dur-
ing temperature programming.
105
-------
Apparatus and Procedures-Fuels Pyrolysis
The main component of the fuel pyrolysis reactor (Fig. 27) consisted of a tempera-
ture controlled, electrically heated, capillary quartz tube reactor into which
a sample of the fuel to be pyrolyzed was inserted. Pt-Pt/Rh thermocouples were
used as a sensor for the controller and measurement of the temperature in the
pyrolysis zone. The reported temperature was measured at the exterior of the
tube at the position of the quartz boat during pyrolysis. As this is about 2
inches from the center of the furnace, the pyrolysis temperature is nearly the
same as the maximum temperature at the center of the furnace (see Fig.28 for the
temperature profile in this type of furnace). The sample insertion device con-
sisted of a quartz boat attached to a quartz rod which, in turn, was attached, by
means of a loop, to a stainless-steel rod. The overall length of the insertion
device allowed the insertion of the boat into a 2-cc bulb blown into the quartz
tube reactor. The lightly greased metal rod of the insertion device ran through
a rubber septum which was placed inside an AN fitting. This provided as gas-
tight connection with the quartz tube reactor.
After a sample of the fuel had been weighed into the quartz boat, the insertion
device was attached to the quartz tube reactor with helium flowing. The boat was
only partially inserted into the quartz tube, outside the heated zone, and the
exterior of the tube was cooled with liquid nitrogen to prevent any of the sample
from being heated by radiation. The system was flushed with helium for about 10
minutes to purge any oxygen remaining in the system before pyrolysis of the sam-
ple was initiated.**
The fuel oil sample was then inserted very rapidly into the heated zone of the
pyrolysis reactor to the stop position. The pyrolysis products were entrained
in the helium stream and removed from the reactor. Analyses of the products from
the pyrolyses are discussed in a subsequent section.
*This bulb can be seen as a dark zone in the photo in Fig. 27 because a coal
residue had not yet been burned out.
**No oxidative pyrolysis experiments were conducted in the fuel reactor.
106
-------
o
-I
5AA21-4/7/75-C1B
Figure 27. Fuel Pyrolysis Reactor
-------
The carrier gas flowrate in the fuel reactor was 20 cc/min and the pressure was
ambient in the experiments that were designed to measure the amount of HCN
formed.* In the experiments in which N~ + NH_ were measured, the GC required a
backpressure of 12 psig. The flowrate was 40 cc/min in these experiments, giving
similar residence times.
A glass wool plug was inserted in a metal fitting at the downstream end of the
quartz reactor to trap any carbon particles that were present in the reactor prod-
uct stream. After each fuel experiment, the reactor was burned out and the down-
stream tubing was rinsed with benzene to remove any tars that had condensed.**
The quartz boat was cleaned in concentrated HC1 and HNO, to remove traces of
organic matter and metallic ash. In the coal experiments, the reactor and boat
had to be lightly etched with HF solution after every few experiments to remove
the residue and ash.
Temperature-programmed gas chromatography was used to identify and measure the
major organic products from model compound pyrolysis. A number of special ana-
lytical methods were used to investigate the amounts of inorganic nitrogen pro-
ducts that were formed during the inert and oxidative pyrolysis of model com-
pounds, fuel oils, and coal. These are described in Appendix A.
*It was demonstrated that the amount of HCN formed was not a strong function of
flowrate.
**The first benzene passed through the tubing after each experiment came out
slightly discolored.
108
-------
RESULTS AND DISCUSSION - MODEL COMPOUND PYROLYSIS
The model compounds pyridine, quinoline, benzonitrile, and pyrrole were selected
as representative of the types of nitrogen present in fuels. These nitrogen com-
pounds were pyrolyzed in helium over the temperature range of 850 to 1100 C. The
decomposition rates, the nitrogen-containing inorganic products, and the major or-
ganic products were measured. The oxidative pyrolysis of pyridine and benzonitrile
was also investigated in mixtures of helium and oxygen.
The pyrolysis reactions were studied in a quartz reactor* with an ID of about 2
mm, a volume of 1 cc, and a nominal residence time of 1/2 second. The details of
the apparatus and the experimental procedures are given in the Experimental sec-
tion. The volatile pyrolysis products were removed from the reactor by a carrier
gas which was helium. In the oxidative pyrolysis studies, a helium-oxygen mix-
ture was used as the carrier gas. In most of the experiments, a liquid sample of
the model compound (0.2 microliter) was injected into the carrier gas stream at
16,0 C, and vaporized before the gas entered the high-temperature zone of the re-
actor. These will be referred to as liquid samples, but it should be noted that
the reactant is in the vapor state and well mixed with carrier gas before decom-
position is initiated. A vapor injector was used in some of the pyridine exper-
iments in which approximately 1 cc of carrier gas saturated with pyridine was sud-
denly introduced (by use of a Carle valve) into the carrier gas stream that flowed
through the reactor. The only real experimental difference between the two sample
injection techniques was that the liquid sample injection technique gave a greater
model compound concentration (mole fraction of 0.03 versus ~0.003).
*A stainless-steel reactor was tested initially but it was found to decompose
both HCN and NH3, preventing identification and measurement of these potential
products.
109
-------
The organic products from model compound pyrolysis were measured by temperature-
programmed gas chromatography (GC) employing flame ionization detection. The
pressure in the reactor, 16 to 20 psig, was controlled by the pressure drop in
the GC column and the flowrate. The inorganic products NH-, N2, and HCN were mea-
sured by a combination of wet chemical and gas chromatographic techniques. Be-
tween experiments, the reactor was burned out to remove any carbonaceous residue
that formed on the quartz wall and the GC column was backflushed to prevent high-
molecular weight products from eluting during later experiments.
Decomposition Rates as a Function of Temperature
Pyrolysis experiments were conducted at temperature intervals of about 25 degrees
and the GC peak areas were used to determine the fraction of the reactant that
passed through the reactor undecomposed.
Measurements indicated that the reactor temperature was constant over the center
4 inches of the 12-inch furnace and then fell gradually to about 90 percent of the
center temperature (expressed in degrees C) at a distance of 1 inch from.each end
of the furnace. Therefore, the assigned reactor wall temperature profile listed
in Table 7 and plotted in Fig. 28 (for 1000 C) was used in calculating kinetic
parameters from the model compound pyrolysis data.
TABLE 7. ASSIGNED WALL TEMPERATURE PROFILE FOR MODEL COMPOUND REACTOR
(Furnace Length = 12 Inches)
Distance From Center, inches
0 to 2
2.5
3-0
3.5
k.Q
*».5
5-0
5 to 7
Percent of Tmax, C
100.0
99.8
99.5
99.0
98.0
96.0
90.0
*
'-At the ends of the furnace, the wall temperature was assumed to
decrease linearly from 90 percent of Tmax at a point 1 inch in-
side the furnace (5 inches from the center) to 160 C at a
point 1 inch outside the furnace. Virtually no reaction occurs
in this portion of the reactor.
110
-------
1000 -
RESIDENCE TIME
WALL TEMPERATUREj
GAS TEMPERATURE,
ASSIGNED
CALCULATED
600 -
in
o
o
<_)
CO
LU
O
O
CO
LU
0£
500
1
-7-6-5-4-3-2-1 0 1 2 3 4 5 6
DISTANCE FROM CENTER OF REACTOR, INCHES
Figure 28. Model Compound Reactor, Temperature Profile, and Residence Time
-------
The linear flow velocity (at 1000 C) was 23 inches/sec in the series A and B ex-
periments and 15 inches/sec under the conditions in series C. At these flowrates,
the flow is laminar, with Nu = 4, and the rate of change of the gas temperature
is given by:
dT (MW)0'5 (T )°'6
-£ = 9.8 x 10"° x —± x (Tw - TG) (113)
where MW is the molecular weight (which was taken as 4 since the gas is nearly all
helium), F is the mass flowrate in Ib/sec, T is the gas temperature in degrees R,
and TW the wall temperature. Equation 113 was integrated, along with the reaction
rate, with the computer program that was written to reduce the pyrolysis data.
The calculated gas temperature follows the wall temperature within a fraction of
a degree (as shown in Fig. 28), indicating that gas temperature lag is not a prob-
lem in the small -diameter reactors used in these experiments.
This can be shown directly from Eq. 113 as follows. The mass flowrate (in the
series C experiments) is 3 x 10" Ib/sec. At temperatures near 1000 C, Eq. 113
becomes :
dAT . 1 .-l (U4)
dx (Tw - TG) dx AT
Integration of Eq. 114 gives a characteristic distance, DI/O* for the reduction
of AT to one-half its initial value:
D1/2 = (An 2)/79 = 0.009 inch (115)
Also shown in Fig. 28 is the cumulative residence time (at 1000 C) calculated by
integrating the gas flowrate through the reactor.
112
-------
Liquid Sample Injection. The experimental decomposition curves obtained for the
model compounds (in helium) are compared in Fig. 1 (presented in the report
summary). Pyridine and pyrrole gave similarly shaped curves with slopes that
remained steep until beyond 95-percent decomposition. The pyrrole is less stable
than pyridine, the curves being separated by about 60 degrees. Quinoline gave
a decomposition curve that was nearly linear with temperature. Quinoline is less
stable than all of the other compounds below 910 C, but is more stable than pyrrole
above that temperature. Benzonitrile is unusual in that its decomposition curve
remains steep up to about 960 C and then tails out at high temperatures. In fact,
3 percent remained undecomposed even at a temperature of 1100 C. Thus, below 1010 C,
pyridine is more stable under these conditions than benzonitrile while, above 1010 C,
the reverse is true.
Calculation of Pyridine Kinetic Parameters. The experiments in which the
extent of pyridine decomposition (in helium) was measured using liquid sample
injection, are listed in Table 8 and these data are plotted in Fig. 29. In the
series A and B liquid injection experiments, the helium flowrate was 50 cc/min
(measured at room temperature and pressure), the reactor ID was 1.8 mm, and the
pressure was 20 psig. In the series C liquid injection experiments, the residence
time was longer because the reactor ID was 2.2 mm and the flowrate was 42.9 cc/
min. The longer residence led to larger extents of decomposition in the series
C experiments (Fig. 29), but it will be seen that the same rate expression is ob-
tained. The pressure was 16 psig at this lower flowrate. As described in a
footnote in Table 8, the decomposition rate would finally increase if sufficient
carbonaceous residue were allowed to build up on the wall of the reactor.
113
-------
TABLE 8. EXPERIMENTAL DATA ON RATE OF PYRIDINE DECOMPOSITION
IN HELIUM (LIQUID SAMPLE* INJECTION)
Series A (20 psig;
Experiment
Number**
9767-4-1
9767-^-2
9767-4-3
9767-4-4
9767-4-5
9767-4-6
50 cc/min; 1
Temperature,
C
959
966
968
998
999
1009
Series B (20 psig;
.8 mm ID; X = 0.030):
Percent
Undecomposed
60
54
50
28
27
21
50 cc/min; 1
Experiment
Number
9767-5-1
9767-5-2
9767-5-3
9767-5-4
9767-5-5
Experiment
Number
9767-4-7
9767-4-8
9767-4-9
9767-4-10
9767-4-11
9767-4-12
Temperature,
C
1023
1024
1025
1046
1050
1050
Percent
Undecomposed
12
11
12
3.7
(2.1)***
(1.4)***
.8 mm ID; X = 0.030):
Temperature,
C
965
963
900
904
852
Percent
Undecomposed
59
56
90
87
101
Series C (16 psig; 43 cc/min; 2.2 mm ID; X^ = 0.035):
Experiment
Number
9767-27-1
9767-27-2
9767-48-3
9767-48-4
9767-49-5
9768-8-4
9768-10-4
9768-10-5
9768-10-6
Temperature,
C
970
974
970
1114
1124
1053
1049
1053
1056
Percent
Undecomposed
32
30
33
<0
<0
0
1
1
0
.2
.6
.01
.01
.87
.54
.10
.97
Experiment
Number
9768-11-1
9768-12-2
9768-15-3
9768-15-5
9768-20-6
9768-21-7
9768-21-8
9768-21-9
9768-23-2
Temperature,
C
1027
1024
1000
1001
975
974
950
950
950
Percent
Undecomposed
3.9
3-7
10.6
7.8
44.7
41.9
(37.8)
53.8
50.8
*Liquid sample size: 0.2 microliters
**Experiments are numbered according to notebook, notebook page,
and number in series.
jV*Res idue was not burned out between experiments in Series A. This
finally caused the Pyridine decomposition rate to increase markedly
after 10 samples had been pyrolyzed. Starting with Series B, reactor
was burned out after every experiment.
114
-------
t/1
100
VAPOR SAMPLE INJECTION
EXP(-70,000/RT)
LIQUID SAMPLE INJECTION
k = 3.8 X 10 EXP(-70,000/RT)
• SERIES A
ASERIES B
X SERIES C
900
950
1050
1000
TEMPERATURE, C
Figure 29. Rate Data for Pyridine Pyrolysis in Helium
1100
-------
A simple rate expression of the form,
-d(MC)/dt = A (MC)N exp (-E/RT) (116)
where (MC) denotes the concentration of the model compound at any point in the re-
actor, was fitted to the model compound data where possible. Because a GC method
was used to measure the reaction rate, the experimental error is large at small
extents of decomposition (i.e., the rate is determined by the decrease in the area
of the reactant peak and thus cannot be measured accurately with little reaction).
Therefore, the kinetic parameters are determined mainly by the results obtained at
large extents of reaction, and logarithmic plots such as that in Fig. 30 are use-
ful in fitting the rate data.
Because the concentration, gas velocity, and reaction rate are temperature-depen-
dent, it was necessary to integrate the rate expression 116 through the reactor
using the assigned temperature profile. The velocity (and residence time) also
depends to some extent on the mole change that occurs during pyrolysis. The mole
change, An, was assumed to be plus one for the model compounds* and it turned out
that, because of the low mole fractions involved, varying An by a factor of 2 had
no significant effect on the calculated residence times and rate parameters.
The pyridine data obtained using liquid sample injection was fitted to a first-
order rate expression (N = 1). The theoretical curvestfor E = 65,000, 70,000,
and 80,000 kcal/mole are plotted in Fig. 30. The "A" factors for these curves
were selected so that the percent undecomposed at 1000 C was equal to the experi-
mental value of 28.2. It can be seen from Fig. 30 that the series A and B pyri-
dine experiments fit quite well a first-order rate expression with the following
constant:
k = 3.80 x 1012 exp(-70,000/RT), sec"1 (117)
However, the uncertainty in the activation energy is on the order of 5 kcal/mole,
partly because the fraction reacted is changing quite rapidly with temperature
*For pyridine, e.g., the overall stoichiometry is approximately:
CCH_N = HCN + 1/2 CH + 7/2 Cs + H9, representing a An of 1.5.
,00 4- /
Obtained by integrating the rate expression through the reactor.
116
-------
CO
o
a.
z:
o
o
LU
Q.
C3
O
O SERIES A
A SERIES B
k = 5.20 X 1011 EXP(-65,000/RT)
k = 3.80 X 1012 EXP(-70,000/RT)
k = 2.78 X 1013 EXP(-75,000/RT)
900
Figure 30.
950
1000
1050
TEMPERATURE, C
Pyridine Pyrolysis in Helium, Comparison of Experimental Results
With First-Order Rate Expressions (log scale)
-------
(on a log scale) at the larger extents of reaction. The A-factor in Eq.H7 is
larger by a factor of 1.5 than was originally reported from these data (Ref.
and 121) because the exact integration of the reactor temperature profile resulted
in an effectively shorter reactor.
It can be seen from Fig. 29 that this first-order rate expression for pyridine
decomposition also fits the series C liquid injection experiments, although the
fit is not quite as good (the vapor sample curve on the figure will be discussed
later).
Discussion of Pyridine Decomposition Rate. The assignment of a rate expres-
sion for pyridine decomposition permits the reaction profiles through the reactor
to be calculated using the computer program that was written to reduce the experi-
mental data. These profiles are plotted in Fig. 31. At moderate temperatures
from 960 to 1000 C, the reaction occurs gradually over the length of the reactor.
At 1100 C, however, most of the reaction occurs in a zone about 2-1/2 inches in
length. These calculations are in agreement with the positions of the carbona-
ceous deposits observed in those experiments where the furnace was opened before
the residue was burned out.
The purpose of assigning kinetic parameters to the model compound pyrolysis data
is to permit the decomposition rate to be extrapolated to the higher temperatures
which the nitrogen-containing compound would encounter if it survived until quite
close to the flame front in a combustion process. Shown in Fig. 32 is an Arrhen-
ius plot of the half-life of pyridine as a function of temperature. The solid
curve represents the rate constant in Eq.117 and the dotted curve the other two
rate expressions tested in Fig. 30. It can be seen that the (extrapolated) half-
life for pyridine is about 0.05 millisecond at 1800 K, and will be in error by
about 15 percent for each 1 kcal/mole that the activation energy is in error.
In their premixed burner studies with added pyridine, Merryman and Levy (Ref. 101
and 122) were able to predict the observed rate of pyridine disappearance by use
of the inert pyrolysis rate expression previously reported from this program
(Ref. 115 and 116). Why the presence of oxygen did not affect the decomposition
rate of pyridine in their burner will be considered later (in the discussion of
oxidative pyrolysis).
118
-------
SERIES A AND B - 1.8 MM ID; 50 CC/MIN; 20 PSIG
•SERIES C - 2.2 MM ID; k2.S CC/MIN; 16 PSIG
k = 3.8 X 1012 EXP(-70,000/RT), SEC"1
(FURNACE)
-7 -6 -5
-A -3 -2-10 1 23
DISTANCE FROM CENTER OF REACTOR, INCHES
Figure 31. Pyridine Pyrolysis in Helium, Calculated Percent Decomposed as a
Function of Distance and Temperature
-------
o
o
CO
o
o.
o
2000 1800 1600
1000
TEMPERATURE, K
1400 1200
100
1000
100
10
0.1
0.01
10
-3
10
10
-5
10
-6
HURD AND SIMON
T = 825 AND 850 C
1/2
AND 6 SEC
THIS STUDY
T ^ 1000 C
T]/2~0.2 SEC
HYPOTHETICAL
OX I DATIVE
PYROLYSIS CURVE
k = 5.20 X 1011 EXP(-65,000/RT)
k = 3.80 X 1012 EXP(-70,000/RT)
k = 2.78 X 1012 EXP(-75,000/RT)
0.5
0.6
0.7
0.8
1000/T, K
0.9
1.0
-1
1.1
Figure 32. Pyridine Pyrolysis in Helium, Arrheriius Plot of
Calculated Decomposition Half-Life
120
-------
The only other major study of the inert pyrolysis of pyridine was that of Hurd
and Simon (Ref. 92), which was discussed under Phase IA. The half-lives calcu-
lated at 825 and 850 C from the rate constants obtained from their data are 14
and 6 seconds, respectively (Fig. 32). The half^lives calculated from Eq.117 are
15.6 and 7.6 seconds at these two temperatures, respectively, indicating good
agreement between the decomposition rates obtained in the two studies (see Fig. 7 ).
Rate Parameters for Other Model Compounds. The pyrrole and quinoline decom-
position rate data, listed in Table 9 and plotted in Fig. 33, were fitted to first-
order rate expressions, using the detailed integration program, giving the kinetic
parameters listed in Table 10 and the theoretical (solid) curves shown in Fig. 33
and 1. The rate parameters obtained for pyridine and pyrrole suggest that the
rate-determining step is a unimolecular reaction. Such reactions typically have
pre-exponential factors in the range of 10 to 10 sec (Ref. 123). The sur-
prising feature is that pyrrole, which is less aromatic than pyridine, has a higher
activation energy for decomposition. The quinoline kinetic parameters suggest a
heterogeneous reaction.
The rate data obtained with benzonitrile, Table 9 and Fig. 33 do not appear to
fit any simple rate expression of the form of Eq.H6. The "tail" on the benzoni-
trile curve (i.e., 3-percent undecomposed at 1100 C) was further investigated be-
cause it was possible that a thermally stable product formed that had the same
retention time as benzonitrile. This possibility was investigated using the mass
spectrometer to analyze the eluted peak and there was no indication of the pre-
sence of any species other than benzonitrile.
Vapor Sample Injection. When the vapor sample injection system, described in the
Experimental section, was used to introduce pyridine into the model compound re-
actor,, the data listed in Table 11 and plotted in Fig. 29 were obtained. It is
apparent that the half-life for pyridine decomposition is much less under these
conditions. Interestingly, the vapor injector data fit fairly well a first-order
rate expression again with a 70 kcal/mole activation energy. However, an A-factor
12
of only 1.25 x 10 must be used to give the fit shown in Fig. 29. This is smaller
by a factor of 3 than the A-factor for the liquid injection data for pyridine.
121
-------
TABLE 9. EXPERIMENTAL DATA* ON RATE OF MODEL COMPOUND
DECOMPOSITION IN HELIUM (LIQUID SAMPLE** INJECTION)
Quinol ine
Experiment
Number
9768-6-16
9768-47-4
9768-47-5
13158-6-3
13158-7-4
13158-7-5
13158-7-6
Temperature,
C
1116
852
852
900
900
925
925
Percent
Undecomposed
0
74.7
77.2
47.6
49-5
42.0
40.9
Experiment
Number
13158-7-7
13158-8-8
13158-9-1
13158-9-2
13158-9-3
13158-10-4
Temperature,
C
950
950
975
975
1000
1000
Percent
Undecomposed
35.7
29-2
17.0
14.6
9.6
6.1
Benzonitrile
9768-6-15
9768-7-2
9768-7-3
9768-12-3
9768-12-4
9768-16-6
9768-16-7
1115
1051
1051
1027
1027
998
996
2.4
7.9
6.1
7.0
6.9
10.2
9-6
9768-17-8
9768-20-4
9768-20-5
9768-42-12
9768-42-13
9768-44-1
9768-44-2
(975)
975
975
940
940
900
900
13.5
14.6
14.3
32.5
37.7
70.5
65.3
Pyrrole
9768-19-3
13158-1-8
13158-1-9
13158-1-10
13158-1-11
13158-1-12
975
852
852
875
875
900
xl.2
79.9
79.9
77-3
86.5
67.2
13158-2-13
13158-2-1
13158-3-2
13158-3-3
13158-3-4
13158-3-5
900
925
925
950
950
975
57.8
31.0
27.2
9.1
6.0
1.3
"Conditions were same as for Series C Pyridine experiments listed in
Table
**Liquid sample size: 0.2 microliters
122
-------
CO
o
O.
O
LU
O
o
(£.
100
90
80
70
60
50
30
20
10
A
(• ) BENZONITRILE
8
(«) QUINOLINE, k = 2.k X 10 EXP (-^5,000/RT)
(A)JPYRROLE, k = 7.5 X 1015 EXP (-85,000/RT)
850
900
950 1000
TEMPERATURE, C
1050
1100
Figure 33. Rate Data for Model Compound Pyrolysis in Helium, Liquid Sample Injection
(See Fig. 29 for pyridine data)
-------
TABLE 10. KINETIC PARAMETERS FOR PYROLYSIS OF MODEL
COMPOUNDS IN HELIUM (Liquid Sample Injection)
-dC/dt = k(C)N - A • (C)N • exp(-E/RT)
Pyrid ine
Pyrrole
Quinol ine
N
1
1
1
A, sec'1
12
3.8 x 10
7.5 x 1015
2.1» x 108
E, kcal/mole
70
85
45
TABLE 11. EXPERIMENTAL DATA* ON RATE OF PYRIDINE DECOMPOSITION
IN HELIUM (Vapor Sample Injection)
Exper iment
Number
9768-34-1
9768-35-3
9768-35-4
9768-36-5
9768-36-6
9768-36-7
9768-36-8
9768-37-11
Temperature,
C
967
970
970
968
971
973
970
1027
Percent
Undecomposed
79.0
76.3
83.1
78.1
77.5
81.7
83.5
44.3
Experiment
Number
9768-37-12
9768-39-1
9768-39-2
9768-39-3
9768-39-/»
9768-40-5
9768-1*0-6
Temperature,
C
1027
970
970
1050
1050
1100
1100
Percent
Undecomposed
45.4
86.1
82.7
24.4
24.4
1.9
3.5
-15 psig, 40.9 cc/min, 1.9 mm ID, X0 S 0.003
124
-------
The vapor injector experiments were conducted in a new quartz reactor with an ID
of 1.9 mm at a pressure of 15 psig and a flowrate of 40.9 cc/min. The only major
difference between the liquid and vapor sample injection experiments was the mole
fraction of the pyridine vapor in the gas entering the reactor. To obtain an es-
timate of the pyridine mole fraction in the gas stream after injection of a liquid
sample, the half-width of the methane peak (which is not held up on the GC column)
was measured to determine how rapidly the pyridine vaporized in the injection sys-
tem. The time required for the pyridine sample to vaporize is about 2 seconds.
From this, it was calculated that the pyridine mole fraction was about 0.03 com-
pared with 0.003 in the vapor injector experiments (calculated from the vapor
pressure). This difference of a factor of 10 in pyridine concentration is the
only apparent explanation for the longer half-life in the vapor injection experi-
ments. That is, if the reaction order is greater than 1, the half-life would in-
crease as the initial concentration decreases. Thus, the decomposition of pyri-
dine fits a first-order rate expression when the initial concentration is held
constant and the extent of decomposition is varied by increasing the temperature,
but a reaction order of higher than 1 is indicated when the initial concentration
is varied.
An attempt was made to fit the pyridine data to a higher-order simple rate expres-
sion, but no satisfactory fit was found. An order greater than 1.5 would be re-
quired to account for the concentration effect, but then a much greater activation
energy is needed to give the proper temperature effect. Taking this approach, no
simple rate expression was found that would fit all the pyridine data. The vapor
injection results indicate that the inert pyrolysis of pyridine is not a simple
unimolecular process.
Oxidative Pyrolysis. Oxidative pyrolysis experiments were conducted in the same
manner as the inert pyrolysis experiments except that the carrier gas contained
up to 10-percent oxygen. After sample injection, the flow was switched to helium
carrier gas before temperature programming of the GC column was initiated to pro-
tect the column packing.
125
-------
The vapor injector system was built specifically for the study of oxidative pyroly-
sis so that oxygen could be intimately mixed with the model compound vapor before
it entered the reactor. When 5-percent oxygen was present in the carrier gas
stream, the temperature for half decomposition was decreased from about 1020 to
780 C, as shown in Fig. 34.
To obtain larger amounts of products, the oxidative pyrolysis of model compounds
was investigated further using liquid sample injection. To ensure good mixing
with 62, one-half of the carrier gas stream (helium + oxygen) was brought in
through the sample injection port and the remainder was mixed in, in a small-
volume "T" (1/16-inch OD), after the sample had been vaporized but before the
sample reached the reactor. The effect of oxygen at concentrations of 2.6 and
10.7 mole percent on the rate of decomposition of pyridine and benzonitrile, re-
spectively, under these conditions, is shown in Fig. 35 and .36. These results
are cross-plotted as a function of oxygen concentration in Fig. 37 and 38. With
pyridine, the 50-percent decomposition temperature was reduced from about 950 to
925 C with 2.6-percent 02, and to 710 C with 10.7-percent 02. The benzonitrile
50-percent decomposition temperature was decreased from 925 C to 860 and 735 C
with 2.6- and 10.7-percent 63, respectively. With 10.7-percent oxygen, the
amount of the model compound reacted increases rapidly with temperature (Fig. 35
and 36) and then suddenly levels off. At 2.6-percent 0 , the reaction rate levels
off at lower extents of reaction with both compounds, and then increases again as
the temperature region is reached in which appreciable decomposition occurs in the
absence of CL.
The most likely explanation of these changes in rate with extent of decomposition
is that the oxygen becomes depleted as the extent of reaction increases. The oxy-
gen/pyridine molar ratio was about 3.5 in the liquid injection experiments with
10.7 percent CL. This is somewhat less than the stoichiometric ratio of 4.25 for
conversion to CO, NO, and HLO, and considerably less than the ratio of 6.75 re-
quired for conversion to CO , NO, and HO. The oxygen/pyridine ratio in the vapor
injection experiments (Fig. 34) was about 20. This could account for the contin-
ued rapid increase in extent of decomposition beyond 85-percent decomposition in
Fig. 34 with only 5 percent 07, while the rate of pyridine decomposition slowed
126
-------
to
CO
c
Q_
100
90
80
70
60
| 50
S 'tO
LU
°- 30
20
1C
0
CARRIER GAS:
HELIUM + 5% OXYGEN
800
900
TEMPERATURE, C
IOOC
1 100
Figure 34. Pyridine Pyrolysis Results Using Vapor Injector
-------
K)
oo
0)
tn
o
I
•s
O!
O
PH
Helium
i—i—i—' ' ' • '
700
800
1000
Temperature, °C
Figure 35. Effect of Oxygen on Decomposition Rate of Pyridine (Liquid Injection)
-------
N)
to
-a
a>
o
o
a
s
O
100
90 -
700
800
900
1000
1100
Temperature, °G
Figure 36. Effect of Oxygen on Decomposition Rate of Benzonitrile (Liquid Injection)
-------
O4
o
8 9 10 11 12
Mole Percent Oxygen
Figure 37. Effect of Oxygen Concentration on Rate of Pyridine Decomposition
-------
-a
o
o
o
a
CJ .
o
fc
o
p.
8 9 10 11 12
Mole Percent Oxygen
Figure 38. Effect of Oxygen Concentration on Rate of Benzonitrile Decomposition
-------
at 65 percent reaction in the liquid sample experiments with 10.7 percent 0 ,
Fig. 35 (because most of the oxygen may have been consumed). It can be seen from
Fig. 36 that 2.6-percent oxygen has little effect on the decomposition of benzoni-
trile until a temperature of 750 C is reached at which about 80 percent decomposi-
tion occurs when 10.7-percent oxygen is present. This suggests that once the oxi-
dative pyrolysis is initiated it tends to go rapidly to completion (perhaps via a
chain reaction).
The rate of reaction of the model compounds increases greatly when CL is added to
the carrier gas stream. To determine if this would be the case at the higher tem-
peratures that would be encountered in or near a flame front, an exact rate expres-
sion for oxidative pyrolysis, with an accurate activation energy,, is required. It
is apparent that such a rate expression cannot be obtained from the present data,
except, possibly in the case of the vapor injection experiments, because the con-
centration of oxygen is decreasing in an unknown manner. Data obtained at small
extents of reaction, where the CL concentration remains near its initial value,
would permit an accurate rate expression to be derived. Because the technique
employed in this study of oxidative pyrolysis involved measurement of the amount
of unreacted compound, accurate decomposition rate data cannot be obtained at small
extents of reaction.
It can be shown from an Arrhenius plot that if oxidative pyrolysis were faster by
a factor of ten at 1000 C than inert pyrolysis, but the activation energy was only
30 kcal/mole for the oxidative pyrolysis, the half-life for oxidative pyrolysis
would be longer by about an order of magnitude at 1800 K than that for inert pyroly-
sis (dotted line in Fig. 32). Therefore, inert pyrolysis could dominate at high
temperatures even in the presence of oxygen if the heating rate were sufficiently
fast. This could account for the fact that Merryman and Levy found pyridine to
decay in their premixed burner studies at a rate predicted by our inert pyrolysis
rate constant (Ref. 122). It is not presently possible, based on available data,
to assess exactly the role of oxidative pyrolysis in the formation of fuel NO in
combustion.
132
-------
Decomposition Products
Using the techniques described in the Experimental section, the major organic and
inorganic nitrogen-containing pyrolysis products were measured for the model
compounds.
Organic Products of Inert Pyrolysis. The individual organic products (volatile
below about 200 C) that were identified in the model compound decomposition prod-
ucts are shown in Fig. 39 through 42. Each curve denotes the weight percent that
a given product represents of the amount of compound that has decomposed. Be-
cause the major interest was in the nitrogen-containing products, the amounts of
these species that are formed are plotted as solid curves. The organic products
obtained from pyridine at the lower temperatures (Fig. 39) are the same as those
reported by Hurd and Simon, but are recovered at much higher concentrations.
This probably resulted from the shorter residence time and the improved experi-
mental procedure used during this study.* At higher temperatures, the less ther-
mally stable products decrease in concentration and only the stable products are
observed. The major organic products from quinoline (Fig. 40) are similar to
those from pyridine (benzene, benzonitrile, and methane). It will be shown that
quinoline apparently forms mostly residue.
The major organic products from benzonitrile (Fig. 41) are benzene and biphenyl,
indicating that the first step is a C-C bond rupture followed by hydrogen abstrac-
tion, to form benzene, and some association of phenyl radicals. This organic
product distribution is that which could be expected (Ref. 124).
*Considerable residue built up in the reactor of Hurd and Simon. This may have
catalyzed the formation of residue as was found during this study.
Note that considerable benzonitrile is found in the products of pyridine decom-
position even at 1100 C. This is in agreement with its high-temperature stabil-
ity as shown in Fig. 1.
133
-------
100
80
60
ItO
30
20
Q
1 10
z 8
Ul
^ 6
13
£ *
£ 3
LU
O
o:
C9
ui
1
0.8
0.6
0.4
0.3
0.2
METHANE
CN
950
1000
TEMPERATURE °C
1050
1100
Figure 39. Organic Products of Inert Pyrolysis or Pyridine
-------
100
80
60
50
AO
30
20
en
O
oc
a.
o
o
O
a:
ui
10
8
6
5
A
1
0.8
0.6
0.5
0.3
0.2
I
850
900
TEMPERATURE, C
950
1000
Figure 40. Organic Products of Inert Pyrolysis of Quinoline
-------
100
80
60
30
*- 20
O
3
O
cc
°- 10
I 8
C9 6
3
Ul
1
UJ
O
cc
1
0.8
0.6
0.4
0.3
0.2
(TiTTtD
METHANE
I
950
1000 1050
- TEMPERATURE, °C
1100
Figure 41. Organic Products of Inert Pyrolysis of Benzonitrile
-------
100
80
60
50
kO
30
20
10
8
6
5
UJ
o
QC
i 1
0.8
0.6
0.5
0.4
0.3
0.2
_ UNKNOWN (CC-C,)
7 O
METHANE
\
UNKNOWN (~C2)
\
\
\
I
I
850
900 950
TEMPERATURE, C
Figure 42. Organic Products of Inert Pyrolysis of Pyrrole
1000
-------
The experiments with pyrrole are interesting in that all of the carbon is recov-
ered at low temperatures, i.e., little residue is formed. The unknown that accounts
for 48 percent of the carbon at 875 C (Fig. 42) has a retention time on the GC
column about the same as pyridine and the picolines, but the mass spectrometer was
not in operation at the time these experiments were run. It would be quite unex-
pected if the first step in the pyrolysis of pyrrole turns out to be the formation
of the aromatic pyridyl ring. If this product is pyridine, most of the pyrrole-N
was recovered in the organic products at 875 C.
Inorganic Products of Inert PYrolysis. Since little of the model compound nitro-
gen was recovered in the organic pyrolysis products (except possibly in the case
of pyrrole), the search for nitrogen in the inorganic products became of even
more importance. It will be seen that only minor amounts of NH_ and virtually
no N9 is formed in the pyrolysis of the model compounds. HCN, however, was found
to be a major decomposition product, particularly at higher temperatures.
HCN From Mode1 Compounds. As shown in Table 12, the amount of model compound
nitrogen that is converted to HCN under inert pyrolysis conditions increases from
50 percent at 950 C to 80 percent at 1100 C for benzonitrile and from 40 at 950 C
to 100 percent for pyridine at 1100 C. These high conversions to HCN from two
structurally.different fuel nitrogen compounds suggest that CN species may be the
key intermediates in the formation of fuel NO in combustion where even higher tem-
peratures and heating rates are involved. The common formation of HCN in pyrolysis
would thus account for the observation that most nitrogen compounds are converted
to NO to the same extent under a given combustion condition. No HCN measurements
were made at elevated temperatures with pyrrole. HCN was found in some early ex-
periments with quinoline at 1100 C (at about two-thirds the level found from ben-
zonitrile), but the method for determining HCN used in those experiments was found
not to be reliable (see Appendix A.). These results suggest, however, that quin-
oline would have formed appreciable HCN at 1100 C if quantitative measurements had
been made.
Hurd and Simon (Ref. 92) have reviewed the previous literature on the pyrolysis
of pyridine. They note that HCN had been reported as a product, but no estimates
138
-------
TABLE 12. HCN FORMATION IN INERT PYROLYSIS OF MODEL COMPOUNDS
Pyr idine
Pyr id ine
Benzoni tr i ]e
Benzoni tr i le
Temperature, C
956
960
959
1102
1106
1107
1107
957
955
1107
1106
Percent
Decomposed
(62)
65
65
100
100
100
100
(71)
69
98.0
98.0
Percent -N Converted to
HCN*
35.6
*»5.7
36.8
106
98
98
105
*5
53
82
81
"Based on N content of decomposed portion of sample.
of the amount were given. Ruhemann (Ref. 125), for example, reported in 1929 that,
on increasing the temperature of pyridine decomposition fr.om 600 to 900 C, nuclear
scission into HCN becomes important. Kurd and Simon do not mention if they observed
HCN from pyridine, but state that with the picolines "HCN was formed in every run
above 775 C, but no attempt was made to estimate it quantitatively."
After the quartz model compound reactor had been used in many experiments (probably
several hundred hours at 950 to 1100 C), it became catalytic toward HCN decomposi-
tion, causing most of the HCN to be converted to N at 1100 C, and lesser amounts
at lower temperatures. The interior of the tube was frosted in appearance sug-
gesting that it may have gradually reacted with the carbonaceous residues that
were deposited and then burned off after each experiment. When the quartz tube
had reached this condition, the decomposition rate of pyridine and benzonitrile
did not change but, of course, much less HCN was recovered in the decomposition
products. Installation of a new quartz reactor in the model compound apparatus
restored the previous conditions. HCN gas was used to calibrate the HCN measure-
ments before and after each series of experiments and to establish that hetero-
geneous decomposition of HCN was not occurring. Another direct indication of
139
-------
the presence of large concentrations of HCN in the pyrolysis products was a base-
line disruption of the gas chromatogram. The HCN came through effectively as a
broad peak with a peak half-width as long as 10 minutes, depending on the amount
of HCN present (with 7-1/2 degrees per minute temperature programming).
NHs From Model Compounds. Only small amounts of NH_ were found in the model
compound pyrolysis products, although considerable effort was expended on this
measurement. Using both the NH converter method (in which NH is decomposed to
•J O
N? and }\ and the N« subsequently measured by GC) and the phenate method, the con-
version of model compound nitrogen to NH in pyrolysis at 960 C ranged from 0.8 to
4 percent for pyridine and 1.5 to 2.6 percent for benzonitrile. Calibration of the
model compound reactor system with microgram quantities of NH, demonstrated that
slightly more than one-half of the NH_ that was passed through the system was re-
covered even with these small quantities (either as NH_ in the phenate method or
O
as N? by the converter method). The remainder of the NH_ was probably adsorbed
in the stainless-steel line between the reactor and the analysis train. These
results do permit upper limits of 8 and 5 percent, respectively, to be placed on
the amount of nitrogen that is converted to NH_ in the pyrolysis of pyridine and
benzonitrile.
The possibility was tested that NH3 could react with the carbonaceous residue and
form HCN and H2« However, when known amounts of NHg were passed over the hot resi-
due from the pyrolysis of pyridine at 1050 C, no detectable HCN was formed.
The equilibrium constant for the reaction:
NH3 + Cs = HCN + H2 ; (118)
is given by Kp = PH~ * PHCN/PNH3- Therefore, if equilibrium is attained, PHCN
will equal PNH-T wnen pHo = Kp- T^e values of Kp, obtained from the JANAF Thermo-
chemical Tables, are 0.38 atm at 925 C and 5.7 atm at 1125 C. The partial pres-
sure of H2 should remain less than about 0.2 atm in the model compound experiments
indicating that reaction 118 is favored thermodynamically. Therefore, the failure
of HCN to form in the above experiment must be the result of kinetic limitations.
140
-------
HCN has been produced from the reaction of NH with coal (Ref. 126) at 1250 C.
However, this is attributed to reactions of the type:
NH, + CH = HCN + 3H • (119)
o Q £
rather than to reaction 118. The conversion of NH to HCN was observed by DeSoete
(Ref. 63) in his premixed ethylene-0_ flames. It appears that the possible role
of NH_ in the formation of HCN from nitrogen compounds during pyrolysis warrants
further investigation.
N_2_ From Model Compounds. It was demonstrated conclusively several times dur-
ing the program that neither pyridine nor benzonitrile formed any detectable N«
when .pyrolyzed at temperatures between 950 and 1100 C. Calibrations with a dilute
N_-helium gas mixture demonstrated that the molecular sieve GC used for the N«
determination was capable of measuring a few tenths of a microgram of N,. There-
fore, if any of the model compound nitrogen formed N« directly during pyrolysis
it was much less than 1 percent.
The observation that N2 does not form during model compound pyrolysis is not sur-
prising considering the limited number of radical reactions that can lead to No
formation in the absence of oxygen (Ref. 2 ). The absence of N2 formation is
also in agreement with the observation that nitrogen compounds at low concentra-
tions can undergo nearly complete conversion to NO under the proper (fuel-lean)
conditions.
Although the model compounds do not form N2 when pyrolyzed in a quartz reactor,
a majority of the N~ in pyridine (the only compound tested) is converted to N- in
nickel or stainless-steel reactors at 1050 to 1100 C. This was observed in the
nickel reactors used at the start of the program (63 percent ^ from pyridine at
1000 C) and in the ammonia converter used near the end of the pyrolysis studies
(90 percent N2 from pyridine at 1050 C). Calibration experiments showed that in
such metal reactors, NH3 decomposes almost quantitatively to N2 and more than one-
half of HCN-N is converted to ^. Therefore, the formation of N2 from pyridine
in the metal reactors probably resulted from the decomposition on the metal sur-
face of the HCN that forms from the pyridine, although the quantities of N2
formed are somewhat larger than this hypothesis would predict. It will be seen
141
-------
that this heterogeneous conversion of fuel-N to N2 represents the only direct
method that was found for promoting N2 formation under pyrolysis conditions.*
The heterogeneous decomposition of HCN (or NHj) probably accounts for the low
yields of HCN that were obtained from the reaction of coal and NH3 when metal
reactors were employed (Ref. 126).
Residue From Inert Pyrolysis. Hurd and Simon found that most of their pyridine
was converted to residue under inert pyrolysis conditions. Black carbonaceous
residues were observed to form on the reactor wall under inert pyrolysis condi-
tions from the model compounds employed in this study, but the amount of model
compound carbon that was converted to residue was not determined directly. Of
more concern for the purposes of this program was the amount of nitrogen that was
contained in the residue.
After one inert pyrolysis experiment of pyridine at 960 C, the quartz reactor was
removed from the apparatus without burnout. The tube was cut to a length that
permitted it to be used directly as the combustion tube in an automatic Dumas
nitrogen analyzer. The amount of nitrogen that was determined in the residue
by this method represented 49 percent of the nitrogen present in the quantity of
pyridine that had decomposed in this experiment. The accuracy of this result
was limited by a rather high blank that was encountered in the Dumas measurement.
In another type of residue experiment, the residue from the pyrolysis of pyridine
at 950 C was heated (slowly) to 1110 C. No detectable HCN was produced during
this process. This suggests that the HCN formed during model compound pyrolysis
at 1100 C does not form from pyrolysis of the residue after it has formed on the
reactor wall. This possibility cannot be ruled out completely on the basis of
this experiment, however, because the rate at which the furnace temperature could
be increased was necessarily slow at these high temperatures.
*Robertus et al. (Ref.127) attempted to prevaporize the fuel-N compounds in coal
so that they could be converted to N2 before reaching the flame (presumably by
a process such as this) but were not successful.
A residue could have formed in the oxidative pyrolysis experiments, but it would
have burned off immediately.
142
-------
Product Mass Balances-Inert Pyrolysis. A complete mass balance for the inert py-
rolysis products cannot be made because the quantities and compositions of the
residues were not determined. However, the mass balance for nitrogen can be cal-
culated for pyridine because the amount of nitrogen in the residue was measured
at 960 C and all of the pyridine nitrogen is recovered as HCN at 1100 C. The ni-
trogen balance for the inert pyrolysis of pyridine is summarized in Table 13.
TABLE 13. NITROGEN BALANCE IN INERT PYROLYSIS OF PYRIDINE
Product
Residue
HCN
Benzoni tr i le
Qu i no 1 i ne
Acrylonitri le
NH.
N2
Percent N in Product*
960 C
*»9
kO-
5
3
3
<8
<0.1
«100
,1100 C
NMf
102
0.5
<0.1
<0.1
NM
NM
102
"Based on N in decomposed portion of sample
(65 percent at 960 C and 99.9 at 1100 C)
•h
NM
not measured
A summary of the carbon and nitrogen mass balances for the inert pyrolysis of the
model compounds is shown in Table 14. Based on the assumption that the carbon
and nitrogen not found in the measured products was contained in the residue,
the following amounts of material were calculated as being in the residue
(Table 15).
143
-------
TABLE 14. PRODUCT MASS BALANCES FROM INERT PYROLYSIS
Reactant
Pyr id ine
Qu i nol ine
Benzoni tr i le
Pyrrole
Temperature,
C
950
1000
1050
1100
900
950
1000
1100
950
1000
1050
1100
900
950
975
Carbon, Percent Found In:
CH/j
12
10
10
9
1
2
4
4
0.2
0.7
1.2
1.4
26
27
26
Other
Organic
Products
18
13
8
4
2
6
12
7
20
19
22
12
34
26
18
HCN
8
12*
16*
20
3 A *
/(**
5**
6**
7
9*
10*
12
NM
NM
NM
Unknown
Peaks
10
2
0
0
1
3
2
0
1
0
0
0
40
3
0
Residue by
Di f ference
62
63
66
67
93
85
77
83
72
71
67
75
0
44
56
Nitrogen, Percent Found In:
Organ ic
Products
12
6
2
1
2
3
4
1
0
0
0
0
68+
26
11
Residue
49
NM
NM
NM
NM
NM
NM
NM
NM -
NM
NM
NM
NM
NM
NM
HCN
4o
60*
80*
101
25**
35**
45**
55**
48
59*
70*
82
NM
NM
NM
Total1"
I0l
66
82
I 02
27
38
49
56
48
59
70
82
68
26
II
*lnterpolated
'c*Estimated based on qualitative measurements
"i"The remainder of the nitrogen is believed to be in the residue
"""Assume that unknown is Pyr id ine
NM = not measured
-------
TABLE 15. RESIDUE CALCULATIONS
Pyr idi ne
Quinol ine
Benzoni tr i le
Pyrrole
Carbon, Percent in Residue
900
93
0*
950
32
85
72
kk
975
56
1000
63
77
71
32*
1050
66
67
1100
67
83
75
Nitrogen, Percent
900
73
32*
950
k9
62
52
7^
975
89
1000
3*»
51
itl
n Residue**
1050
18
30
1100
0
kk
18
*This suggests that NHo or HCN may have formed from pyrrole at 900 C; these
measurements were not made.
**ln residue or inorganic products for pyrrole.
Products of Oxidative Pyrolysis. The products of oxidative pyrolysis were not in-
vestigated in any detail. The organic products from pyridine and benzonitrile
were quite similar in type and amounts to those obtained from inert pyrolysis. No
significant new products were formed (i.e., no additional GC peaks appeared that
would indicate the formation of partially oxidized species). These observations
suggest that, as was the case in inert pyrolysis, once reaction of the molecule
has been initiated, the reaction proceeds readily until mainly inorganic products
are formed.
Two experiments were conducted to determine how much pyridine nitrogen is con-
verted to NO in oxidative pyrolysis. Duplicate experiments were conducted on the
oxidative pyrolysis of pyridine in 5-percent oxygen at 860 C (the total pressure
was 15 psig as usual). The helium stream from the reactor was collected in an
evacuated flask for 2 minutes after the sample was injected and the NO determined
by the Saltzman method as described in the Experimental section. In a third ex-
periment under these conditions, the extent of pyridine decomposition was measured.
The pyridine decomposed to the extent of 52 percent in agreement with Fig. 37.
The first two experiments gave identical amounts of NOX, 3.8 micrograms (calcu-
lated as NO). Correcting this for unreacted N0*gives 4.6 micrograms of NO. Since
0.2 microliter of pyridine contains 35 micrograms of nitrogen, the decomposed
"Correction calculations for the Saltzman method are discussed in Appendix A.
145
-------
pyridine (52 percent) would yield 39 micrograms of NO if all nitrogen were con-
verted to NO in the oxidative pyrolysis process. Thus, only 12 percent of the
nitrogen from the decomposed pyridine was recovered as NOX. Calibration tests
were not run to ensure that no NO was lost by reaction or absorption in the line
from the reactor to the sampling flask. The results obtained in these experi-
ments suggest that at least pyridyl nitrogen is not readily converted to NO under
these (fuel-rich) conditions of oxidative pyrolysis, but calibration runs are re-
quired to confirm this.
Conclusions From the Pyrolysis of Model Compounds
1. The heterocyclic nitrogen compounds and nitriles are quite thermally sta-
ble in an inert atmosphere with 50-percent decomposition temperatures in
the range of 910 to 960 C (for a nominal residence time of 0.5 second in
a quartz reactor).
2. Pyrrole is much less thermally stable than pyridine under these inert
pyrolysis conditions. Quinoline and benzonitrile are less stable than
pyridine below 1000 C, but benzonitrile is more stable above 1000 C.
3. From 30 to 90 percent of the carbon that was present in the pyrolyzed
model compounds is apparently contained in the residue that forms on
the wall of the reactor. Much of the model compound nitrogen remains
in the residue at lower pyrolysis temperatures.
4. The major volatile organic products from the inert pyrolysis of pyridine
are methane, benzene, and benzonitrile. Quinoline forms mainly benzoni-
trile and some methane and, at higher temperatures, benzene. Benzoni-
trile forms mainly benzene and biphenyl; pyrrole forms methane, acryloni-
i
trile, acetonitrile, benzene and benzonitrile. In addition, pyrrole forms
two organic products that were not identified, one of which is a C$ or Cg
compound that has the same GC retention time as pyridine. This compound
is the major product from the low-temperature pyrolysis of pyrrole.
146
-------
5. The distribution of organic products from inert pyrolysis changes as the
reactor temperature is increased with the more thermally stable products
being favored at the higher temperatures. This suggests that the initial
product distribution may be temperature independent, i.e., that a two-
stage process is occurring at the higher temperatures. The possibility
cannot be ruled out, however, that the primary product distribution is a
function of temperature.
6. The presence of 2- to 10-percent oxygen in the carrier gas stream reduces
the temperature required for a given extent of model compound decomposi-
tion by up to 250 C.
7. The major volatile nitrogen-containing product from the inert pyrolysis
of the model compounds is HCN. The amount that forms increases with
temperature and the quantitative conversion of pyridine-N to HCN occurs
at 1100 C. This is probably the most important result that was obtained
in the model compound experiments.
8. Only a few percent NHj and virtually no N2 forms in the inert pyrolysis
of pyridine or benzonitrile.
9. The pyrolysis results obtained with model fuel nitrogen compounds indi-
cate that HCN and CN species are more likely precursors to fuel NO for-
mation in combustion than are NH3 and NHX radicals.
147
-------
RESULTS AND DISCUSSION—INERT PYROLYSIS OF FUELS
The pyrolysis of fuel samples (six No. 6 fuel oils, one crude oil, and two coals)
was conducted under inert conditions in a small-volume quartz flow reactor similar
to that used for the model compounds. The details of the reactor and the analysis
procedures have been discussed. The sample sizes ranged from 1 to 2 milligrams
and the residence time in the reactor was on the order of 2 seconds. The quartz
reactor was 2 mm ID and 12 inches in length. The ID of the reactor increased to
6 mm for a distance of 2 cm at the point (about 4 inches into the furnace) at which
the fuel samples were pyrolyzed. The purpose of this volume was to accommodate
the larger volumes of gaseous products that form from the pyrolysis of the fuel
samples which were 5 to 10 times larger than the model compound samples. Other-
wise, the volatile species that are released from the fuel as it is heated would
have a much shorter residence time in the reactor than did the model compoundvapors.
For example, 1 mg of No. 6 fuel oil would form about 5 cc of hydrogen (at 1100 C)
if the pyrolysis products were simply carbon and H (assuming a carbon/hydrogen
mass ratio of 8).
The fuel sample was weighed into a small quartz boat and the boat .was placed in
the reactor tube in a cold zone external to the furnace. After any air had been
purged from the system by the helium flow, the boat was rapidly moved into the
furnace by means of a rod attached to the quartz boat that extended to the exter-
ior of the apparatus through a rubber septum. The carrier helium flowing from
the reactor was analyzed for HCN, N , and NH . The HCN was determined by a color-
imetric technique after scrubbing it from the gas stream in a bubbler containing
aqueous Na CO with CdCO suspended in the solution.
£ *J O
HCN From Fuel Oil Pyrolysis (
The evidence obtained from the pyrolysis of the model compounds that HCN was a
major decomposition product prompted an investigation of HCN formation in the
decomposition of fuel oils. Initially, the pyrolysis gases were analyzed for
HCN using an Orion cyanide electrode. It was found that the sulfide ion present
148
-------
as H S in the pyrolysis gases could interfere in HCN analyses by this technique
and, subsequently, an ASTM colorimetric method for HCN determination was modified
and developed for this analysis. A detailed description of this analytical method
is given in the Experimental section.
Two nominal pyrolysis temperatures, 950 and 1100 C, were used in these studies.
Inert pyrolysis of six No. 6 fuel oils and one crude oil gave the results summar-
ized in Tables 16 and 17. These results indicate that HCN is a major product of
pyrolysis of fuel oils and that the amount of HCN produced at 1100 C is substan-
tially greater than at 950 C. However, the extent of nitrogen conversion to HCN
is less with fuel oils than from the inert pyrolysis of model compounds. The
difference in the behavior of the model compounds and fuel oils may be due to a
number of factors which could include: (1) part of the nitrogen present in fuel
oils may be in chemical structures different from those of the model compounds,
(2) the presence of the other components of the fuel oil may modify the pyrolysis
reactions, (3) part of the fuel nitrogen may remain in the oil residue (Ref. 2 ),
because of a higher boiling point or (4) the heating rate may be slower in the
oil pyrolysis experiments. In any case, these results indicate HCN to be an im-
portant intermediate in fuel NO formation in combustion processes. At the higher
heating rates and temperatures encountered in combustion, HCN could be the prin-
cipal intermediate.
NH3 and N2 From Fuel Oil Pyrolysis
The presence of appreciable quantities of HCN in the pyrolysis products of fuel
oils is very significant inasmuch as the oxidation of NH_ and/or NH radicals has
•J J\.
usually been considered as the principal path for fuel NO formation. To deter-
X
mine if NH was also a major product of fuel oil pyrolysis, analyses of the pyroly-
o
sis gases were performed for this inorganic nitrogen species.
A number of analytical techniques were evaluated for determining the ammonia
formed. Most of these methods lacked the sensitivity required to detect the
small amounts of ammonia that could form from the pyrolysis of milligram samples
of fuel oils, one of which contained only 0.22 percent N. Interference from ^S
149
-------
TABLE 16. SUMMARY OF EXPERIMENTAL DATA ON HCN FORMATION
FROM INERT PYROLYSIS OF OILS
Sample
Gulf No. 6 Fuel Oil
(Venezuelan Crude)
0.1*3% N, 2.3% S
Gulf No. 6 Fuel Oil
(Various Crudes)
0.44% N, 0.73% S
Gulf No. 6 Fuel Oil
(Mainly California Crudes)
1.41* N, 1.6% S
Conoco No. 6 Fuel Oi 1
0.3% N, 0.7% S
"EPA" No. 6 Fuel Oil
(In-House)
0.22% N, 0.9% S
No. 6 Fuel Oi 1 (from
Ul trasystems)
0.38% N, 0.3% S
Wi Imington Crude Oi 1
0.63% N, 1.6% S
Temperature, C
1106
953
1103
950
1108
950
1108
952
1103
953
1104
950
1107
950
% Fuel -Nitrogen Converted to HCN
% HCN
33.5, 32.3
19.7, 19.0
38.0, 39.6
23.1, 23.3
23.3, 23.2
14.2, 14.7
43.2, 40.9
19-8, 21.0
36.8, 37.3
24.3, 24.8
35.3, 35.0
18.5, 17.2
51.5, 48.1
29.4, 31.3
Average
32.9
19.4 .
38.8
23.2
23.2
14.4
42.0
20.4
37.0
24.6
35.2
17.8
49.8
30.3
150
-------
TABLE 17. COMPARISON OF HCN FORMATION FROM VARIOUS OILS UNDER
INERT PYROLYSIS CONDITIONS
Sample
Gulf No. 6 Fuel Oil
(Venezuelan Crude)
0.43% N, 2.3% S
Gulf No. 6 Fuel Oil
(Various Crudes)
0.44% N, 0.73% S
Gulf No. 6 Fuel Oil
(Mainly California Crudes)
1.41% N, 1.6% S
Conoco No. 6 Fuel Oil
0.3% N, 0.7% S
"EPA" No. 6 Fuel Oil
( In-House)
0.22% N, 0.9% S
No. 6 Fuel Oil
(from Ul trasystems)
0.38% N, 0.3% S
Wi Imington Crude Oi 1
0.63% N, 1.6% S
% N to HCN
(1100 C)
32.9
38.8
23.2
42.0
37.0
35.2
49.8
(950 C)
19-4
23.2
14.5
20.4
24.6
17.8
30.3
HCN (950 0/HCN (1100 C)
0.59
0.60
0.63
0.49
0.67
0.51
0.61
151
-------
and other sulfur species also had to be considered. Two analytical techniques
were finally developed. The first involved the catalytic decomposition of ammonia
to N- and measurement of the N? by gas chromatography. The second method utilized
the sodium phenate colorimetric analysis. Both of these methods are discussed in
detail in the experimental section.
Two major experimental difficulties were encountered during the examination of
the pyrolysis gases for NH, content. First, continual use of the quartz reactor
tube for fuel oil and coal pyrolysis results in changing what was intially an
"inert" surface toward NH decomposition to one that is catalytically active.
Thus, it was found that 90 percent or more of any NH passed through the used
fuel reactor was converted to N at temperatures of 950 C or above. It was post-
ulated that this catalytic behavior of the tube walls was the consequence of
deposits from trace quantities of metals present in the fuel oils and particularly
from the ash in the coal samples. This problem was not encountered in the model
compound reactor.
The second difficulty associated with the determination of NH_ was the fact that
O
a major part of any HCN formed was also catalytically decomposed to N- in the am-
monia converter. Decomposition of HCN in the reactor was not a problem except
in some of the coal experiments discussed in the next section. Once it had been
established that most of any NH, formed was decomposed immediately in the quartz
reactor, it was decided to decompose any remaining NH, to N? by an auxiliary cata-
lytic converter. By this means, the sum of the elemental N- and NH formed during
pyrolysis was determined. To accomplish this, it was necessary to remove HCN from
the pyrolysis gases to prevent its catalytic decomposition in the converter. Attempts
to find converter conditions that were specific to NH, conversion were not success-
ful. The removal of HCN from the pyrolysis gases was accomplished by means of an
Ascarite (NaOH on asbestos) trap. The trap was heated to 150 C to avoid potential
problems with moisture. Tests performed with HCN passed into the Ascarite trap
demonstrated its quantitative removal.
The results listed in Table 18 gave the N + NH values found at pyrolysis temper-
atures of 950 and 1100 C for four of the oils that had been tested previously for
152
-------
TABLE 18. RESULTS OF MEASUREMENTS OF N_ AND NH (AS
FROM INERT PYROLYSIS OF OILS
Run
18*
19*
20*
31
37
42
43
46
47
21*
22*
33
34
38
39
48
•49
23
25
35
36
40
41
50**
52**
51**
44
45
Fuel Oil,
% N
EPA No. 6 (0.22)
1
EPA No. 6 (0.22)
EPA No. 6 (0.22)
t
Wilmington Crude (0.63)
Wilmington Crude (0.63)
Wilmington Crude (0.63)
Gulf No. 6 (1.41)
Gulf No. 6 (1.41)
1
Gulf No. 6 (1.41)
Gulf No. 6 (1.41)
Ultrasystems No. 6
(0.38)
Ultrasystems No. 6
(0.38)
Reactor
Temperature,
C
954
953
954
960
954
1102
1102
1105
1106
954
950
959
957
1103
1102
1103
1102
953
956
957
959
1102
1101
1107
1106
1104
1099
1099
Through
Converter?
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Yes
Yes
Yes
Yes
Yes
No
No
Yes
Yes
. No
Weight
of Oil,
mg
1.457
1.360
1.341
1.344
1.328 .
1.453
1.319
1.323
1.399
1.136
1.446
1.281
1.366
1.246
1.386
1.227
1.091
1.478
1.076
1.193
1.458
1.389
1.193
1.377
1.186
1.213
1.415
1.462
Ugm N per
mg of Oil
(calculated)
2.20
2.20
2.20
2.20
2.20
2.20
2.20
2.20
2.20
6.30
6.30
6.30
6.30
6.30
6.30
6.?0
6.30
14.1
14.1
14.1
14.1
14.1
14.1
14.1
14.1
14.1
3.8
3.8
ligm N2 per
mg of Oil
(measured)
0.55
0.52
(0.60)
0.52
0.51
0.52 ±0.01
(average)
0.60
0.64
0.62 ±0.02
(average)
0.54
0.43
0.66
0.66
0.76
0.57
0.66 ±0.05
(average)
0.74
0.76
0.75 ±0.01
(average)
0.80
0.87
0.94
0.73
0.73
0.52
0.73 ±0.10
(average)
0.78
0.89
0.83 ±0.05
(average)
0.72
0.92
3.48
0.61
0.50
% -N as N2
25.0
23.6
(27.3)
23.6
23.2
23.8 ±0.5
(average)
27.3
29.1
28.2 ±1 .0
(average)
24.6
19.6-
10.5
10.5
12.1
9.0
10.5 ±0.8
(average)
11.7
12.1
11.9 ±0.2
(average)
12.7
13.8
6.7
5.2
5.2
3.7
5.2 ±0.8
(average)
5.5
6.3
5.9 ±0.4
(average)
5.1
6.5
24.7
16.1
13.2
*Runs 18 through 25 were with Ascarite trap at 25 C; in all others, the trap was at ISO C
**The Ascarite trap was removed from the system in runs 50, 51, and 52
153
-------
HCN formation. These results indicate the small quantities of nitrogen that were
measured as well as the precision of the analyses. The average results from
Table 18 are summarized in Tables 19 and 20. The EPA No. 6 oil gave the highest
percentage conversion to N + NH in inert pyrolysis and its nitrogen content is
the lowest of the oils tested. The amount of NL measured as the consequence of
pyrolysis of the EPA oil represented about one-fourth of the nitrogen contained .
in the sample.
Interestingly, the absolute amount of N_ that is formed (representing NH_ + N )
<- o Z
was found to be only weakly dependent on the bound nitrogen content of the oil
and the temperature of the reactor (Tables 18 and 20). It is conceivable that
part of the N_ present in the pyrolysis products was due to dissolved N-, but
this hypothesis is weakened by the fact that the inert pyrolysis of model com-
pounds in a quartz reactor yielded no detectable N? if the converter was bypassed
(see above). However, the solubility of N~ (at 1 atmosphere) is 0.14 mole per-
cent in turbine oil (Ref. 128). If the molecular weight of the oil is 140, the
solubility of N is 0.3 microgram per milligram of oil, which is about one-half
the value shown in Table 20.
The results given in Tables 19 and 20 indicate that the amount of NL formed changed
only slightly when the converter was bypassed. This strongly suggests that nearly
all of the N,, was, therefore, formed in the quartz reactor. Thus, it is not possi-
ble to give an accurate value for NH_ content in the pyrolysis gases. Since this
converter was shown in calibration experiments to convert pyridine to NL (probably
via HCN formation and decomposition), the limited effect of the converter also
indicates that volatile fuel nitrogen compounds are not present in the reactor
products and being converted to N in the converter.
During studies on the catalytic decomposition of HCN, it was found that about
60-percent conversion to N7 occurred at a temperature of 1050 C. In run 51
(Table 18) the Ascarite trap was removed and all of the gases, including any HCN,
were passed through the converter. The increase in N~ formed was 2.65 ygm per mg
154
-------
TABLE 19. SUMMARY OF PERCENT FUEL-N CONVERTED
IN OIL PYROLYSIS
(Measured as N ~)
TO NH3 +
Oi ] Sample
EPA No. 6
Wilmington Crude
Gulf No. 6
Ultrasystem No. 6
% N
0.22
0.63
1.41
0.38
Reactor Products
Through Ascarite
Trap and Converter
950 C
23.8 ±0.5
10.5 ±0.8
5.2 ±0.8
--
1100 C
28.2 ±1.0
10.9 ±0.2
5.9 ±0.4
16.1
Reactor Products
Bypass Converter
1100 C
24.6, 19.6
12.7, 13.8
5.1, 6.5
13.1
TABLE 20. SUMMARY OF MICROGRAMS N PER mg OIL
FROM OIL PYROLYSIS
0 i 1 Sample
EPA No. 6
Wi Imington Crude
Gulf No. 6
Ultrasystem No. 6
% N
0.22
0.63
1.41
0.38
Reactor Products
Through Ascarite
Trap and Converter
950 C
0.52 ±0.01
0.66 ±0.05
0.73 ±0.10
—
1100 C
0.62 ±0.02
0.75 ±0.01
0.83 ±0.05
0.61
Reactor Products
Bypass Converter
1100 C
0.54, 0.43
0.80, 0.87
0.72, 0.92
0.50
155
-------
of oil (3.48-0.83). Since this oil had previously been found to form 23.2 percent
HCN (defined as percent fuel-N converted to HCN) at 1100 C (Table 17), the increased
amount of N2 represented about 80-percent conversion of the HCN postulated to be
present. Thus, this test served to demonstrate once again the effectiveness of the
Ascarite trap for HCN removal. Calibration samples ot HCN were passed through the
quartz reactor periodically at 1100 C to ensure that no significant amount of HCN
decomposition occurred in the reactor.
HCN From Coal Pyrolysis
Inert pyrolysis of a limited number of coal samples gave results similar to those
obtained from the pyrolysis of fuel oils: considerable HCN and smaller amounts
of N2 + NH3 with much of the fuel nitrogen apparently remaining in the nonvolatile
residue.
A particular experimental problem associated with these tests was the decomposi-
tion of HCN in the presence of deposits from the coal pyrolysis runs. Simple
burnout of the residues did not restore the inert property of the reactor walls.
Chemical etching of the quartz reactor tube and boat, after removal from the
ancillary equipment, with 30- to 60-percent HF solutions, was necessary after
every two or three runs with coal. To ensure the validity of the reported HCN
measurements, HCN calibrations were run before and after each coal pyrolysis
experiment. The calibration experiments indicated that: (1) at most, only a
couple percent of HCN decomposed in a freshly chemically etched reactor at
1100 "C, and (2) after two or three coal samples had been pyrolyzed in an etched
tube, the decomposition of HCN at 1100 C was 5 to 40 percent (i.e., the residue
effect is not predictable and may be a function of coal type and pyrolysis
temperature).
Samples of EPA and IFRF-N coals (1.5 to 2 milligrams) were pyrolyzed in helium and
the HCN formed was measured by the barbituric acid colorimetric method. The results
of the two coals were as follows:
Coal Sample
EPA
IFRF-N
% N
1.17
1.8
% Fuel-N Converted to HCN
957 C
20.3, 18.0
1106 C
29.8, 30.2
26.2, 22.7
156
-------
g and N2 From Coal Pyrolysis
Chemical etching of the quartz tube reactor after coal samples had been pyrolyzed
in it restored its compatibility with HCN at elevated temperatures, but did not
prevent catalytic decomposition of NH . Thus, it was necessary to measure total
N2 (N_ + NH ) by the GC method, as was done for the fuel oil pyrolysis studies.
The total N2 formed from inert pyrolysis of the EPA coal in two experiments at
954 C were 8.3 and 10.5 percent conversion of fuel-N, or 0.97 and 1.23 ug of N_
per mg of coal, respectively. This quantity is somewhat higher than the 0.5 to
0.7 ug of N- per mg that was obtained from the fuel oils at this temperature.
Some of the measured N2 could have been produced by the catalytic decomposition
of HCN in the quartz reactor.*
Discussion of Fuel Results
The inorganic nitrogen product distribution for coal is surprisingly similar to
that obtained with the fuel oils. The fuel oils underwent from 23- to 42-percent
conversion to HCN at 1100 C and the two coals studied gave 24- and 30-percent HCN,
respectively, at this temperature. The oils formed 15- to 25-percent HCN at 950 C,
and the EPA coal formed 19-percent HCN at this temperature and about 10-percent
NH3 + N2> The ratio HCN C950 O/HCN (1100 C) varied from 0.50 to 0.67 for the
oils, while this ratio was 0.67 for the EPA coal. The conversion of fuel nitro-
gen in the oils to N? + NH varied from 0.5 to 0.7 ygm (as N«) per mg of oil at
950 C. The EPA coal produced nearly twice this amount of N? + NH,, but these
products still only accounted for about 10 percent of the nitrogen in the coal.
This similarity of product distributions suggests that the percolation of the
pyrolysis products of coal through a hot solid matrix before entering the vapor
phase .has little effect. Under combustion conditions where the flame front has
not collapsed onto the surface of the particle, the mechanisms for fuel NO forma-
tion may be similar, therefore, for coal, and fuel oils. The results obtained under
*This was not a problem in the fuel oil experiments.
157
-------
this program strongly suggest that this mechanism for homogeneous fuel NO forma-
tion involves CN species rather than NH species or NH,.
J\ *J
Although the maximum amount of fuel-N that was converted to HCN was from 25 to
35 percent, this could represent most of the "volatile-N" in the fuels, especially
in the case of the coals. Sternling and Wendt (Ref. 2) estimated that 60 percent
of the fuel-N would remain in the initial residue from the vaporization of a resid-
ual oil and 80 percent on the average, from the vaporization of coal. They point
out that some additional nitrogen will become volatile during the pyrolysis of the
char, but the amount is unknown. The reverse process must also be considered,
i.e., the formation of residue during pyrolysis from initially volatile nitrogen
compounds. In the fuel pyrolysis experiments, a carbonaceous residue formed on
the reactor well downstream of the sample boat. In any event, the quantities of
HCN that were measured may well contain most of the volatile fuel nitrogen and,
therefore, HCN may be the principal intermediate in the formation of "homogeneous"
fuel NO.
Information on how fuel nitrogen compounds are distributed between the volatile
fractions and the char is quite limited (Ref. 2). Data reviewed in Ref. 71 indi-
cate that, for a number of types of coal, only about one-third of the fuel nitro-
gen is contained in the volatile fraction (even at pyrolysis temperatures up to
1300 K). Cursory experiments were conducted to investigate the volatility of the
nitrogen compounds in the coals used in this study. Unfortunately, the results
obtained were clouded by the fact that analyses conducted 4 months apart by the
same commercial laboratory gave quite different total nitrogen contents for these
coals. These results are presented in Appendix B.
i
In most previous studies of the rapid pyrolysis of'coals, the nitrogen-containing
products such as HCN and NH were not collected and measured. However, Ramachandran
et al., in India (Ref. 120), rapidly pyrolyzed an "N-enriched" coal (coal fertilizer
which contains about 20 percent nitrogen and is prepared by reacting coal with NH,)
and measured quantities of HCN that appear to be in good agreement with the amounts
obtained in this study. Both slow and rapid heating were employed and substantial
158
-------
yields of NH, and HCN were obtained. Increasing the pyrolysis temperature from
900 to 1200 C increases both N and HCN. AT a heating rate of only 5 degrees/
minute, the HCN yield was 60 pounds HCN per ton of coal. By thermal shock heat-
ing at high temperatures, the HCN yield was increased to 266 pounds HCN per ton.
This represents 35 percent of the total N in the coal. The high yields of N. and
NH, obtained indicate that the type of nitrogen compounds present in the enriched
coal are quite different from natural fuel nitrogen compounds.
159
-------
PHASE II - BURNER STUDIES OF FUEL NO FORMATION
x
The phase I pyrolysis experiments, which investigated the chemical processes that
occur in the prefame stages of combustion, demonstrated that HCN is probably the
major initial nitrogen-containing species formed from fossil fuels. Some NH, may
form as well. The objective of the Phase II burner study was to investigate in
detail the kinetics and mechanisms of fuel NO formation in flames from the poten-
tial combustion intermediates HCN and NH,.
HCN and NH, were added to premixed, flat CH^-O^-Ar flames and the temperature and
species concentrations measured at various distances above the burner. NO was
also employed as an additive in some experiments to determine its fate once formed.
Argon was used as the diluent rather than N~ to preclude the possible formation of
thermal NO . The experiments were conducted at a pressure of 0.1 atm to spread the
flame zone.
The measured species profiles, when corrected for downstream diffustion of react-
ants and upstream diffusion of products, establish where in the flame the added
nitrogen compounds react and the NO forms and the rates at which these processes
occur. Therefore, such measurements determine, in principle, the global rates of
fuel NO formation as a function of distance (time), temperature and species
concentration.
PHASE II: EXPERIMENTAL
A photograph of the low-pressure flat-flame burner system used in these experiments
is shown in Fig. 43. The apparatus consists of a water-cooled burner port with a
micrometer positioner, a low-pressure chamber with a large vacuum pump, a quartz
microprobe for gas sampling, a thermocouple probe, a gas-sampling manifold with a
small oil diffusion pump and fore pump, a chemiluminescent analyzer, and a gas
metering system. The pressure in the chamber is normally held at 76 torr. The
gases are metered by means of critical flow orifices for CH., 0?, and Ar feed to
the burner and Brook's flowmeters for Ar purge gas and the N-compound additives.
These items will be described at greater length in the following pages. .
161
-------
ro
5AA21-4/7/75-C1C
Figure 43. Low-Pressure Flat-Flame Burner Apparatus
-------
Burner
The flat-flame burner port was machined from 2-inch nominal schedule 40 stainless-
steel pipe and screws into an aluminum base where an 0-ring seal provides a
pressure-tight fit. A drawing of the burner is shown in Fig. 44. Cemented in a
recess at the burner exit is a 1/8-inch-thick sintered porous plate which furnishes
a flat velocity profile across the ports. The porous plate was cut from a sheet
of 1/8-inch-thick stock composed of sintered-steel (type 316L) particles and hav-
ing a mean pore size of 65 microns (Pall Trinity Micro Corporation). An 8-mesh
stainless steel screen was spot welded to the top of the burner port about 1/8
inch above the porous plate to act as a flame holder. Immediately below the
porous plate, and in contact with it, is a 1/8-inch copper tubing coil for water
cooling the burner; the copper tubing enters and exits the burner through flexible
lead fittings located in the aluminum base. The burner interior, below the copper
coil, is filled with Pyrex balls (Fig. 44) of various sizes to promote gas mixing
and to negate chances of combustion occurring inside the burner. Gas is fed to
the burner via a length of 3/8-inch stainless-steel tubing which also serves as
a pedestal, or support, and is used to position the burner with respect to the
fixed gas sampling and thermocouple probes.
The 3/8-inch stainless-steel tubing "pedestal" enters the low-pressure chamber
enclosing the flat-flame burner via a rotary vacuum connector seal (CEC type No.
SR-37) fitted to the steel base of the chamber. Outside the chamber, the pedestal
is firmly clamped to a micrometer-driven translation stage which, in turn, is
anchored to the base of the low-pressure chamber with a right-angle bracket. The
micrometer-driven translation stage has a 50-mm movement and can easily be posi-
tioned with a precision of less than 0.01 mm and locked in place. Spring loading
of the translation stage was inadequate to move the burner up when the micrometer
was so.adjusted. This required the translation stage to be manually moved up to
the new micrometer position. Below the translation stage, the 3/8-inch-stainless
steel tube is connected to 1/4-inch copper tubing containing a large single turn
of tubing to facilitate the vertical movement of the burner. Moving the burner
163
-------
QUARTZ
MICROPROBE
POROUS
PLATE
2-INCH (NOM)
SCHEDULE 1»0 PIPE
GLASS
BALLS
0-RING
SEAL
ALUMINA TUBING
10-MIL SUPPORT WIRE
3 MIL COATED
THERMOCOUPLE WIRE
8-GAGE
STAINLESS-STEEL SCREEN
COOLING
WATER
OUT
CHV 02, Ar
FUEL-N
Figure 44. Flat-Flame Burner (Cut-Away View)
164
-------
over a distance greater than 50 mm can be done by loosening the clamp holding the
burner pedestal to the translation stage, manually positioning the burner, and
then reclamping the pedestal. The distance moved by the burner in such a case
must then be determined by using a cathetometer.
The flame was ignited with a high-voltage spark fed into the low-pressure chamber
to the burner with an insulated igniter rod. The spark was generated by touching
the igniter wire outside the chamber with a Tesla coil. Ignition was carried out
at 130 to 150 torr, after which the pressure was reduced to the value desired.
Fuel-rich mixtures were more difficult to ignite and, as a result, the flame would
be ignited closer to stoichiometric condition and then the CH. feed would be in-
creased to obtain the desired equivalence ratio.
Low-Pressure Chamber
The low-pressure chamber consists of three pieces of 4-inch ID Pyrex pipe
connected with glass pipe joints and gaskets (Fig. 43): a lower section,
a cross with 1-inch ID side arms, and a reducer section to go from 4-inch ID to
1-inch ID. The two side arms for the cross are capped with aluminum plates con-
taining Swagelok fittings with Teflon inserts to hold the gas sampling microprobe
and the Pt-Pt/10%Rh thermocouple probe. The reducer forms the top of the chamber
and is connected to 3/4-inch stainless-steel tubing leading to a pressure control
valve and a large vacuum pump. As long as the pump can maintain a pressure ratio
greater than 2 across the valve, the pressure control valve is, in effect, a
critical flow orifice with a variable area. With sonic flow at the valve, pres-
sure fluctuation from the pump will not propagate back to the low-pressure cham-
ber. The base of the low-pressure chamber is fabricated of 1/2-inch aluminum
plate and contains fittings for the burner pedestal (burner gas supply), cooling
water entrance and exit, pressure tap, an insulated igniter feedthrough, and an
argon gas purge line. There are, in addition, four screw holes for mounting the
bracket for the micrometer-positioned translation stage.
165
-------
The chamber pressure was monitored with a Heise gage marked off in 1 torr divisions
from 0 to 1000 torr. The gage was adjusted frequently for changes in barometric
pressure.
Gas Flow Metering
Methane, oxygen, and argon for the burner were supplied from cylinders with pres-
sure regulators to drop the pressure to about 50 psig. Each gas line is equipped
with line filters (Matheson Gas Products) to deep large particles out of the flow
metering system. The gases were metered with critical flow orifices consisting
of accurately sized sapphire orifice jewels (Aurele Gatti, Inc.). The small
jewels were mounted by force fitting them into Teflon holders which were then in-
serted by a force fit into Swagelok fittings. Upstream pressure on the orifices
was accurately regulated by Brook's model 8601 line pressure regulators. Mercury
manometers, approximately 2 meters long, were used to measure the upstream pres-
sure at the orifices. As long as the pressure ratio is greater than the critical
ratio,
/ 9 YY/Y-1
h4-T7 (120)
the orifice flowrate will be independent of the downstream pressure. The flow
system is quite similar to that reported by Andersen and Friedman (Ref. 130).
Flowrate calibration of the critical-flow orifices and flowmeters was obtained by
measuring the rate of pressure rise in a known volume being filled with the
metered gas. A Heise gage equipped with a potentiometer provided a means of re-
cording P versus time in the known volume.
The methane used was 99.97 percent pure (ultrahigh purity-Matheson), the argon
was 99.99 percent pure and contained a maximum of 15 ppm N» and 5 ppm 0 , while
the oxygen was 99.96 percent pure and contained a maximum of 10 ppm N_ and 3000
ppm Ar. The lecture bottle of anhydrous ammonia had a minimum purity of 99.99
percent and the commercial hydrogen cyanide had a specified minimum purity of
99.5 percent with 0.5 percent maximum water.
166
-------
Gas-Sampling Microprobe
Gas samples were taken at various distances through the flat flame (along the cen-
ter line) by a quartz microprobe. The probe was uncooled; the reacting gases were
quenched aerodynamically by expanding them from the pressure in the reaction cham-
ber (usually 76 torr) to about 1 torr. The probes were constructed from 9-mm
quartz tubing which was pulled to taper down to a very small-diameter tip contain-
ing the sonic sampling orifice. About 25 mm from the tip, the sampling probe was
bent 90 degrees so that the reacting gases in the flame could be sampled parallel
to the gas flow when the microprobe was mounted in one of the 1-inch-diameter side
arms of the low-pressure chamber. The necessary design considerations for micro-
probes that will result in a minimum flame disturbance have been described in the
literature (Ref. 131). The diameter of the sonic orifice at the tip of the tapered
probes was always between the recommended values of 10 and 100 microns to ensure
that the flow disturbance occurring upstream of the probe as a result of the samp-
ling rate would be minimized.
The flames examined in this experimental program were about at the upper temperature
limit that could be handled with the uncooled quartz probes./ Radiation cooling
permits the probes to be used to sample gases at temperatures somewhat higher than
the softening point of the quartz. After each detailed probing experiment, the
microprobe sampling orifice was visually inspected for damage with a binocular
microscope. It was then checked by samping an N2-NO standard gas mixture at 76
torr using the chemiluminescent analyzer (CA) to (in effect) determine if changes
had occurred in the sampling rate which would indicate changes in the orifice
diameter. The gas sampling rate was important to this experimental program be-
cause the capillary flowmeters in the CA were bypassed (the reasons for this will
be discussed later in this report).
Temperature Measurement
Temperature measurements in the flame were performed with Pt-Pt/10%Rh thermocouples.
The thermocouple junction was formed from either 1- or 3-mil wire by flame welding
the wires using a very small flame under a binocular microscope. (The use of 1-mil
167
-------
wire was discontinued after the NH_ screening tests; thereafter, 3-mil wire was
used) . As much as possible, excess metal at the thermocouple junction was trimmed
away. The 3-mil thermocouple was then given a 2- to 3-micron-thick coating of
A1203 using a sputtering technique available in one of our laboratories. Coating
the platinum wire prevents the catalysis of reactions on its surface that would
heat the wire and result in erroneous temperature measurements.
Thermocouple Coatings. Al^O- coatings were employed because of a lack of success
with the widely used SiCL coatings. Before using the A10,-coated wires, the
^ Z o
silica-coated thermocouples were used for NH_ additive screening tests where they
were made well above the flame front. When an attempt was made to determine a
temperature profile completely through a flame at = 0.8 or 1.5 for the detailed
probing experiments, the noncatalytic Si02 coating would fail at gas temperatures
of 1860 to 1900 K (corrected for radiation cooling of the thermocouple). The
wire temperature at failure would be just below 1700 K. An examination of the
wire surface revealed the SiO- had beaded up and exposed the metal surface.
The SiO? was deposited on the thermocouple wires by vaporizing dimethylpolysiloxane
(Dow Corning 200 Fluid) into the gas fed to a Meker burner and passing the thermo-
couple wire through the flame gases (Ref. 131 and 132). When trouble was exper-
ienced with the SiO_ coating, the thermocouple wires were carefully cleaned by
immersion in acetone and then in concentrated nitric acid before they were coated.
A recorder was connected to the thermocouple to be coated and the SiO- coatings
were applied at 1800 K (Ref. 132) . When the coatings continued to fail, the ther-
mocouple wire diameter was increased from 1 to 3 mils to obtain more radiation
cooling. The maximum temperature measured before Si02 coating failure in any of
these series of tests was about 1910 K.
Several methods of applying Al-0 coatings were examined after the use of SiO_
was abandoned. An alumina base ceramic, used in the electronics industry as a
dielectric material in thick-film circuitry, was applied to the wire and fired
on at 1530 K. A check of the thermocouple after firing the ceramic indicated
that-it was no longer reliable. Perhaps some of the glassy binding materials
168
-------
(silicates) in the ceramic attacked the thermocouple junction or, possibly, the
spot-welded joints, where the thermocouple wire was attached to the 10-mil support
wire, were loosened by the ceramic during the firing process.
A plasma-spray gun technique for applying an Al-0 coating was tested briefly. A
small length of 3-mil platinum wire was coated with AIJD, particles. The coating
did not adhere well to the metal, but formed a protective casing around the wire.
With practice on the part of the spray gun operator, it might have been possible
to lay down an even, thin coating of Al_0_ particles on the wire, but the initial
attempt produced a thick coating. In addition to its thickness, the coating, com-
posed of A1..0 particles partially fused together, appeared to be porous. In a
thinner coating, this porosity might result in less protection of the
platinum surface.
The third approach to applying an Al-O- coating, a high-vacuum sputtering tech-
nique, proved successful. The coating is glassy in appearance and very adherent.
The coated wire was stiff and springy and when bent into a tight loop, would tend
to kink or bend sharply at several points rather than bend smoothly. Experiments
indicated the Al CL coatings applied were good up to a thermocouple wire tempera-
ture of about 1900 K which, for the 3-mil wire and flow conditions used in these
experiments, would mean a gas temperature of about 2100 K because of radiation
cooling. Al 0, coating failure appears to result from differences in the thermal
^ O
expansion of the metal and the aluminum oxide which leads to crazing of the oxide
coating. To avoid problems of coating deterioration, the thermocouple was re-
placed after each detailed probing experiment.
Thermocouple Probe. The thermocouple probe itself consisted of 10-mil thermo-
couple wire support arms of Pt and Pt/10%Rh strung through a two-hole aluminia
tube and cemented in place with Sauereisen cement (No. 29). The support arms ex-
tend out about 35 mm from the end of the cement-sealed two-hole aluminia tube
(Fig. 44). The alumina tube was, in turn, cemented into a short length of 1/4-
inch stainless-steel tubing. A 1/4-inch Swagelok fitting, drilled through to pass
the 1/4-inch tubing of the thermocouple probe and fitted with Teflon inserts,
169
-------
severed as a feed-through for the thermocouple probe to enter the low-pressure
chamber. The 3-mil thermocouple wire was strung through small eyelets formed at
the ends of the 10-mil support wires, wound around the support wire, and then spot
welded to it. Normally, an 18- to 20-mm length of 3-mil thermocouple wire was sus-
pended between the support arms. The length of the suspended wire is dictated by
the 1-inch diameter of the side arm in the low-pressure chamber through which the
thermocouple must slide to be set in place over the burner. The support arms are
bent 90 degrees from the probe so that the arms hang downward when the thermocouple
probe is inserted over the burner. This is done to minimize sagging of the support
arms that might result from long exposure to high temperature. Sagging of the 3-
mil wire is lessened by the glassy, very adherent Al^O, coating which makes the
3-mil wire stiffer and springy, and by reliance on the 10-mil support wire to keep
the 3-mil wire under slight tension.
The thermocouple readings must be corrected for the effect of radiation cooling to
ascertain the true gas temperature. The procedure used to obtain the temperature
correction is described in Appendix F.
Chemiluminescent Analyzer
The gas sample withdrawn from the flame was analyzed for NO, NC^, NH,, and HCN
with a Chemiluminescent gas analyzer (CA). The CA used was a model 10A self-
contained unit manufactured by the Thermo-Electron Corporation. This instrument
utilizes the Chemiluminescent reaction between NO and 03 as a means of monitoring
NO concentration, i.e.,
NO + 03 »-N02* + 02 (121)
N02* »-N02 + hv (122)
The ozone is produced from an 0- supply by an ozonator in the instrument and mixed
with the sample gas in a flow reactor. A photomultiplier and optical filter
mounted at one end of the reactor monitors photons emitted by the electronically
excited N09. The number of photons detected per unit time is primarily a function
170
-------
of both NO concentration and the sample flowrate through the reactor. The resi-
dence time of gas in the CA reactor is quite large compared to the reaction time
so that the photon emission rate is directly proportional to the feed rate. Since
the excited NO- can also lose energy via collisions with other species, the rate
of photon emission can be a function of pressure, at sufficiently high pressures,
and alter the direct relationship between the sample flowrate and the photon em-
ission rate.
The CA contains a capillary flow metering system designed to operate with a re-
actor pressure between 3 and 10 torr and a gas sample pressure of 1 atmosphere.
Since the gas samples from the flat-flame burner was to be quenched by expansion
to about 1 torr and the gas was sampled through a sonic orifice with a diameter
less than 100 microns, the flow metering system in the CA was superfluous and was
removed. The pressure in the reaction chamber was adjusted by turning a control
valve installed in the ozone feed line until the reactor pressure was 0.8 torr
with no sample feed. With sample feed the pressure increased to about 0.9 torr
and the use of the calibrated air leak (described later) increased this another
0.1 or 0.2 torr.
Calibration for NO. The CA was calibrated for use with the sonic sampling orifice
of the quartz microprobe by inserting the microprobe in a small calibration cham-
ber (Fig. 45) containing a standardized mixture of 92 ppm NO in N2 at 76 torr
(the pressure at which the detailed probing experiments were performed). A cali-
bration adjustment knob on the CA was used to adjust the dial reading of the CA
until it agreed with the NO concentration in the standard gas. The molar flow-
rate, N , of the standard gas to the CA (as limited by the flow through the probe
orifice) is given by:*
gY
(123)
*The variables in the following equations are defined in the Nomenclature List
on page 345.
171
-------
t-0
CHEMILUMINESCENT
ANALYZER
TO VACUUM PUMP
1.7 LITER
RESERVOIR
HEISE
GAUGE
TO VACUUM PUMP
Figure 45 . Schematic of Chemiluminescent
Analyzer Calibration System
-------
For a standardization mixture that is mainly N_:
= CA P
o s 59.7 RTs .
Normally C, A , g, R, and Pg will be constant throughout a calibration of the in-
strument and its subsequent use to measure ppm NO in an unknown. When using the
microprobe to measure NO concentration in an unknown (i.e., the flame) the ppm NO
reading of the CA must be corrected for the change in the molar feed rate through
the orifice between the measurement of the calibration standard and the unknown.
The equation for this correction is:
T i \ V-'UTJ'-'/ VIu~
\'\^rr- \(TX\
\ u u yu I
T is the temperature of the standard gas at the critical flow orifice and TU is
the temperature of the "unknown" gas at the critical flow orifice. Since a lower
unknown flowrate, N , would produce a lower ppm NO reading on the CA, the correc-
tion to the ppm NO reading on the CA dial is given by:
NS
NO corrected = NO read x — (126)
N
u
The linearity of the response of the CA to the sample flowrate was checked (because
of the effect of collision rate on the probability of photon emission) by noting
the CA reading of NO concentration in the standard gas as the pressure on the
sonic sampling orifice (p in Eq. 123) was changed. For room-temperature calibra-
tions with the orifice size and CA reaction chamber pressure used, the response
did become nonlinear at sample pressures much above 76 torr. Therefore, Eq. 126
would not be accurate for higher pressure room-temperature calibrations. Because
of the lower gas densities involved, the CA response was linear with respect to
sample flowrate when probing flames at pressures as high as 0.4 atm.
Because the CA will drift slightly over a period of time, it is convenient to have
a method for checking its calibration during an experiment without having to shut
down, remove the microprobe from the low-pressure chamber, and install it in the
173
-------
calibration chamber setup. A means of doing this was devised and plumbed into
the CA sample feed line (Fig. 45). A 4-mil (approximately 100 microns) sapphire
jewel, mounted similarly tothe orifice jewels used for metering CH., Ar, and 0-
to the burner, was setup to meter the NO standard gas at the desired upstream
pressure (76 torr usually) in a parallel feed line to the CA. With the microprobe
installed in the calibration setup, when the CA is properly adjusted for the micro-
probe flow, the response of the CA to the calibration gas fed through the 4-mil
orifice at the same upstream pressure is noted. One can then use the NO reading
obtained with the 4-mil orifice as a calibration standard for the CA-microprobe
combination. The upstream pressure on the 4-mil orifice and the calibration cham-
ber for the microprobe is measured with a Heise gage.
Conversion of N-Species to NO. Concentration measurements of nitrogen compounds
other than NO are quite possible with a CA if they can be reliably converted to
NO. Supplied with the Thermo-Electron Corporation CA is a heated (400 C) catalytic
converter (and a temperature regulator) for converting NO- to NO. If operated
at higher temperatures, this catalyst (molybdenum) can also be used to convert NH,
and HCN to NO. Calibration curves were obtained for the percent conversion of
both NH, and HCN to NO. In the absence of sufficient 0- concentrations in the
gas sample, the percent conversion of NH, and HCN to NO on the catalyst dropped
appreciably. Thus, a controlled air leak was installed in the CA sample line (Fig.
45). The air leak is a small hole in a piece of quartz tubing set in a side arm
to the sample line. Under these conditions, the conversion of NH, to NO in the
molybdenum converter ranged from 45 to 60 percent, depending on the NH, concentra-
tion, and the range for HCN conversion was 60 to 87 percent.
Typical calibration curves are shown in Fig. 46 for HCN, with and without air
added to the sample, and for NH, with added air (a! complete calibration was not
made for NH, without the air leak). With added air, the conversion of HCN to NO
was nearly quantitative up to 750 ppm. The HCN curve then broke sharply with the
additional HCN (above 750 ppm) being converted only to the extent of about 50 per-
cent. The NH, calibration curve was nearly linear giving about 50 percent conver-
sion at all concentrations. The calibrations were conducted by microprobe sampling
of metered mixtures of the additive in unburned Ar, Q~ an^ CH. as the gas mixture
174
-------
1500
1000
a.
o.
o
500
(NO ADDED AIR)
500 100 1500
HCN OR NH3> PPM
2000
2500
Figure 46. Typical Calibration Curves for Molybdenum Converter at
800 C (Solid Curves Denote Range of Measurements)
issued from the burner. When using the metered air leak, the conversion efficiency
was shown to be independent of the 0- content of the sample entering the probe
(from 0 to 30 percent). No calibration curve was obtained for cyanogen but brief
tests indicated that it also was converted to NO by the molybdenum catalyst at
800 C.
The percent conversion to NO was a function of the actual amount of NH, or HCN
passing over the catalyst per unit time. Consequently, when using the catalytic
converter, the CA meter reading of NO ppm was first converted to the equivalent
amount of NH, or HCN via the calibration curve, then the correction to the ppm
NH, or HCN was made for the change in the sampling rate at the sonic orifice of
the microprobe according to Eq. 125 and 126.
The molybdenum catalyst that is supplied by the CA manufacturer for the conversion
of N02 to NO is not a true catalyst but rather a regenerable reactant. The
"catalyst" was not intended for use at temperatures over 500 C because an oxidation
175
-------
process, which can apparently be reversed if it occurs below 500 C (by passing
a stream of H through the converter for several hours), becomes irreversible
at higher temperatures. When used at 800 C, the catalyst may change in efficacy
for conversion of NH3 and HCN to NO as an oxide coating develops. As a result,
the catalyst calibration curve must be checked often.
Gas Sample Analyses
Batch gas samples taken during an experiment were analyzed by an outside commer-
cial laboratory using mass spectroscopy. The gas samples were taken and stored
in 1-liter sample bottles until they were analyzed. The sample bottles had two
feed lines, one with a break seal and the other open but with a slight constric-
tion. After the sample bottle was mounted on the sample manifold, it was evacu-
ated to several microtorr with an oil diffusion pump and heated. The bottle
was then filled to a pressure of several torr with gas sampled from the flame,
evacuated to several microtorr again, and heated with a torch (the glass around
the feed line constriction was heated until a slight softening was perceived).
After the second filling with sample gas to a pressure of 1 to 2 torr a torch
was used to seal the inlet line at the constriction. The break seal in the re-
maining line was opened when the bottle was mounted on the inlet system of the
mass spectrometer for analysis.
The analyses reported the mole fractions of H2, CH4, NH3, H20, C2H2, CO, N2, 02>
Ar, C02, NOX (including NO and N02), H2CO, C2H6, C2H4, and HCN. During the course
of experimentation, it soon became clear that the mole fractions reported for
NH-, HCN, N2, and usually N0x were quite inaccurate and that some H20 was always
missing from the analyses. Therefore, NH3, HCN, and NO analyses were obtained by
the use of the CA operating "in line," i.e., in a continuous stream of freshly
sampled gas. An alternative means of analysis for N2 in the low-pressure gas
samples was not developed.
The mass spectroscopy (MS) results were corrected for the missed H20 as well as
possible by use of a material balance for H and 0. There are difficulties asso-
ciated with using a material balance on a rapidly reacting gas mixture in which
176
-------
diffusion plays a role. Strictly speaking, one should utilize all of the gas
samples and obtain flux profiles through the flame for each of the species and
from the flux data determine the deficient components. This, however, can be
long and tedium process even when the diffusion coefficients for the species are
available. The procedure used here to roughly determine the amount of missing
KLO consisted of using the argon analysis as a base since it does not change ap-
preciably in mole fraction through the flame and, hence, requires no diffusion
correction. The Ar was used to calculate the total H and 0 from the metered
QL/02/Ar feed rates; the amount of H and 0 missing were found by subtracting the
H and 0 material balance obtained by the MS analysis from the total H and 0 cal-
culated from the Ar. Usually, the ratio of missing H to missing 0 was about 2.
In all circumstances, the value for the missing H was divided by 2 and used as
the value for the missing H-O. All the mole fractions were then normalized to
a sum of 1.00.
RESULTS AND DISCUSSION - SCREENING EXPERIMENTS
A series of screening tests was conducted first to select the proper conditions
for the detailed probing experiments. These tests involved the determination of
the effects of a number of variables on the N(D yields from NH, and HCN additives
X <3
in premixed QL-C^-Ar flames. The input concentration of NH3 or HCN and the con-
centration of NO in the completely burned gas, measured about 80 mm above the
burner in most experiments, were compared (the luminous zone of the flame was
about 3 mm above the flame holder). The flame temperature was also measured at
this sampling point. These screening experiments, which have been designated as
input-output experiments, can be carried out quickly permitting the effects of a
large number of variables to be screened.
The variables tested in the experiments reported here included: fuel-air equiva-
lence ratio (), argon dilution, additive concentration and type, pressure, and
burner feed rate. It will be seen that the general effects of changing these
variables were studied but the input-output experiments could not be analyzed in
as much detail as the flame probing experiments because the temperature-time pro-
files between the burner and the sampling point were not measured. The detailed
177
-------
probing experiments, conducted later, showed that the temperature reaches a maxi-
mum about 8 mm above the flame holder and then decays gradually. At = 0.8 and
1.5, the temperature was found to have dropped by about 250 K from its maximum
value at a point 80 mm above the burner.
Typically, after the CH4, 02, and Ar feed rates were established and the burner
had been lighted and achieved thermal equilibrium with the system, the axial
temperature profile was determined with a coated Pt-Pt/10%Rh thermocouple around
the point from which the quartz microprobe would withdraw gas samples. The
thermocouple was then withdrawn from the flame and the microprobe inserted to ob-
tain a measure of the NO concentration using the chemiluminescent analyzer (CA).
J\
Most of the screening experiments were run with the CA in the NO mode (molybdenum
A
converter at 400 C) but runs without the converter (NO mode), gave the same results
within about 1 percent. As discussed in the previous Experimental Section, the gas
temperature at the sampling point must be known to correct the CA reading for
changes in the sampling rate through the critical-flow sampling orifice. The mole-
cular weight and y of the sampled gas, which also are required to determine the
sample gas flowrate through the orifice, were taken to be the values computed at
thermodynamic equilibrium at the adiabatic flame temperature for the system. To
obtain an axial position reference, the distance from the tip of the burner to
the location of the thermocouple and microprobe (at one axial position point) was
measured with a cathetometer. Thereafter, the micrometer on the translation stage
used to position the burner was employed to measure burner location relative to
that initial reference measurement.
A diluent ratio, denoted as DR, is defined as the molar ratio, Ar/02, in the
burner feed divided by the N2/02 ratio in air. Therefore, DR was defined as 1
when Ar/02 =3.76 because this is the (N2+Ar)/02 ratio in air. Conducting exper-
iments at a DR of 1 results in the mole fractions of reactants being the same as
in CH4-air flames of the same equivalance ratio. However, the lower heat capacity
of argon results in much higher adiabatic flame temperatures. Duplication of the
heat capacity of air requires a DR of 1.6. The maximum flame temperatures measured
at DR = 1 in the detailed probing experiments (reported in the next section) were
only slightly higher than the adiabatic temperatures for CH4-air flames because of
heat loss to the burner face.
178
-------
Ammonia Addition
Input-output experiments were conducted with ammonia added to the premixed feed
gas. In most experiments, the nominal ammonia concentration was 2500 ppm (molar)
based on the total inlet gas flow, including argon and oxygen. At an equivalence
ratio of 1 and a DR of 1 (argon/oxygen molar ratio of 3.76), an ammonia concentra-
tion of 2500 ppm defined in this manner is equivalent to the addition of 2.3 weight
percent nitrogen to the methane. This high additive concentration was selected to
permit the nitrogen species to be measured by mass spectrometry although this method
of analysis for nitrogen species was later abandoned in favor of the CA.
Effect of Equivalence Ratio and Feed Rate on NOX Yield. If the NO yield was to
•^™ ••^fc^^^^fc™ j^
be determined from = 0.5 to 1.5, without damaging the thermocouple or microprobe,
it was imperative that the measurements be performed far enough downstream to en-
sure that temperatures less than 2000 to 2100 K would be encountered. After a
number of tests, it was decided that a distance of 80 mm above the burner would be
satisfactory for these experiments. The results of all ammonia addition screening
experiments at 76 torr are summarized in Table 21. The initial measurements of
NOX yield were made with DR = 1 and a total burner feed rate (except for the NH^
additive) of 6014 cc/min at <}> = 0.5, with an additional 60 cc/min of CH. for each
0.1 increase in <}>. These experiments are listed as Series A at the top of Table
21 and are plotted on the lower curve in Fig. 47. In the second series of input-
output experiments with added NH3 (Series B of Table 21), the diluent ratio was
held at 1.0 and the total feed rate to the burner was increased to 7520 cc/min
(plus the NH3 additive). It can be seen from Fig. 47 that increasing the feed
rate increases the NO yield with the maximum effect being at about <(> = 0.8.
Jv
The temperatures listed in Table 21 for these two experimental series (A and B)
indicate that increasing the feed rate increases the temperature by as much as
300 K with the maximum increase being obtained near stoichiometric. These obser-
vations suggest that the increase in NO yield with feed rate results from an in-
«
crease in the flame temperature brought about by the flame front being moved far-
ther away from the flame holder allowing less heat loss to the water-cooled burner.
At 4> = 1.4 and 1.5, the feed rate of 7520 cc/min caused the flame to blow off.
179
-------
TABLE 21. SUMMARY OF NO MEASUREMENTS IN SCREENING EXPERIMENTS
WITH AMMONIA ADDITIVE (p = 76 torr)
Experiment
No.
A-l
A- 2
A-3
A-4
A- 5
A-6
A-7
A-8
B-l
B-2
B-3
B-4
B-5
B-6
B-7
B-8
C-l
C-2
C-3
D-l
D-2
D-3
D-4
D-5
D-6
D-7
0.5
0.8
0.9
1.0
1 . 1
1.2
1.3
1.4
0.5
0.7
0.8
1.0
1.0
1 . 1
1.2
1.3
0.5
0.6
0.6
0.9
0.9
0.9
0.9
0.9
0.9
1 .0
Di luent
Ratio
(OR)
1.0
1.0
1.0
1.0
1 .0
1.0
1.0
1 .0
1.0
1.0
1.0
1.0
1.0
1 .0
1.0
1 .0
1 .0
1.0
1 .0
1 .2
1.2
1.4
1.4
1.6
1.6
1 .2
Symbol
Used in
Figures
O
O
O
O
0
0
O
O
X
X
X
X
X
X
X
e
e
e
A
•
•
Distance
From Burner,
mm
81.50
77.45
77.74
81.40
75.95
75-96
76 . 25
80.70
81.30
81.30
80.70
80.70
80.08
81 .00
80.75
80.70
41.65
40.15
19.98
75.75
75.75
75.75
75-75
77.77
77 .77
80.60
T, K*
1446
1488
1512
1542
1661
1618
1599
1622
1456
1676
1731
,1846
1856
1816
1772
1735
1570
1656
1776
1756
1756
1821
1821
1762
1765
1664
Feedrate,
cc/min
at STP
6014
6194
6254
6314
6374
6434
6494
6554
7520
7520
7520
7520
7520
7520
7520
7520
6014
6074
6074
7157
7157
8060
8060
8963
8963
7217
NOX,**
ppm
2011
1846
1777
1623
1424
1 185
1258
932
1964
2086
2098
1976
1823
1571
1339
1179
2354
2246
2365
1773
1929
1572
1989
1466
2012
1607
NH3, ***
ppm
2692
2560
2503
2433
2408
2358
2315
2314
2564
2547
2563
2519
2519
2432
2484
2404
2660
2655
2655
2234
2468
1994
2476
1783
2482
2197
Percent
NHj
Converted
74.4
73.3
71.0
66.9
59.1
50.2
44.2
40.3
76.6
81 .9
81.9
78 . 4-* **
72.4
64.6
53-9
4'9.0
88.5
84.6
89.1
75.0
76.2
78.8
80.4
82.2
81.1
73.2
180
-------
TABLE 21. (Concluded)
Exper iment
No.
D-8
D-9
D-IO
D-ll
0-12
E-]
E-2
E-3
E-4
E-5
E-6
F-l
F-2
F-3
G-l
G-2
G-3
G-4
H-l
H-2
H-3
H-4
H-5
H-6
1-1
1.0
1.0
1.0
1.0
l-.O
0.8
0.8
0.8
1 . i
1.1
1 . 1
0.8
0.8
0.8
0.7
0.7
0.7
0.7
0.8
0.8
0.8
0.8
0.8
0.8
1.5
Di luent
Ratio
(OR)
1.2
I.**
1.4
1.6
1.6
1.2
1 .4
1.6
1.2
1.4
1.6
1.6
1.6
1.6
1.0
1.0
1.0
1.0
1.6
1.6
1.6
1.6
1.6
1.6
1.0
Symbol
Used in
Figures
A
•
•
A
•
•
A
•
•
e
9
9
9
9
9
9
9
9
Distance
From Burner,
mm
80.60
. 80.60
80.60
80.50
80.50
80.40
79.50
79.55
81 .00
81.35
81 .00
19.87
19.91
19.91
47.60
42.59
37.56
32.62
22.75
26.90
28.88
17.49 .
17.95
1^.95
49.35
T, K*
1664
1717
1722
1732
1736
1651
1644
1527
1797
1773
1764
1734
1729
1729
1935
1962
1968
2016
1732
1726
1723
1732
1723
1723
1776
Feedrate,
cc/min
at STP
7217
8120
8120
9023
9023
7520
7520
7520
7520
7520
7520
7419
7419
7419
8763
8763
8763
8763
7419
7419
7419
7419
7419
7419
6247
NO ,**
ppm
1796
1523
1920
1372
1921
1935
1949
1957
1629
1676
1694
2015
'2640
3436
2158
2236
2286
2308
1995
2001
2125
2014
2008
1766
757
NH ***
ppm
2399
1967
2473
1775
2488
2487
2523
2528
2488
2476
2552
2517
3390
4909
2579
2579
2579
2579
' 2517
2517
2517
2517
2517
2314
2349
Percent
NH3
Converted
74.9
77.4
77.6
77.3
77.2
77.8
77.2
77.4
65.5
67.7
66.4
80.1
77.9
70.0
83.7
86.7
88.7
89.5
79.5
79.5
80.0
80.2
79.8
76.3****
32.2
•••Temperature has been corrected for radiation as described in Appendix F.
••'"ppm (molar) of NOX in the gas sampled.
••'•••••"ppm NHo that would be in gas if the NH3 did not react and the other species reacted to chemical
equilibrium at the adiabatic flame temperature (i.e., corrected for reaction mole change).
**-"*Resul t not plotted in figures because apparently spurious.
181
-------
100
BO
o J£fije5 A (LOW FLQWWTE,
& Seff££S 3 (HIGHER FLOWMTE
C (LOW FLOWRATE, d
L i i i i I iii i
RICH)
Figure 47. Conversion of NHs to NOX as a Function of Equivalence
Ratio, $, and Flowrate in Screening Experiments at a
Pressure of 76 torr
182
-------
In sufficiently lean and rich mixtures, the lower flame velocity should result in
a flame reaction zone far enough from the burner to reduce heat losses and the low
and high feed rate correlation curves in Fig. 47 should approach one another at
mixture ratios far removed from stoichiometric. As an illustration of this, at
<|> = 0.5 (experiment B-l), the 7520 cc/min feed rate caused the flame to distort
from its normally flat appearance and lift some distance above the burner. With
a 20 percent lower feed rate (A-l), which would place the experiment in the lower
feed rate category, the percent conversion of NH, to NO was essentially the same.
O "A
Regardless of the complications introduced by energy losses from the flame to the
burner, it is readily apparent in Fig. 47 that for $ < 1.0 (fuel-lean flames) the
yield of NO is quite high while for $ > 1 the yield decreases rapidly as is in-
Jt
creases. Similar effects of mixture ratio on NO yield have been reported by other
A,
investigators (Ref. 66, 71 and 72). Based on the results of the other investiga-
tions even higher NO yields would be expected at lower NH, concentrations.
X o
The data reported here at the higher flowrate show a decline in N0x yield, measured
at a fixed distance from the burner, as is decreased from 0.8 to 0.5. However,
higher yields of NO were obtained when the measurements were made closer to the
J\
luminous zone of the flames at 4> = 0.5 and 0.6 (Series C in Table 21) as shown in
Fig. 47, suggesting that some of the NO is decomposing before reaching the probe
JC
when the measurements are taken far above the flame front. The rate of a small
shift toward equilibrium in the post-flame gases would be enhanced by the excess
oxygen present under fuel-lean conditions.
Effect of Argon Dilution. Screening experiments were performed with diluent ratios
of 1.2, 1.4, and 1.6 at = 0.8, 0.9, 1.0, and 1.1. At $ = 0.9 and 1.0 (Series D
in Table 21), the conditions were the same as in the Series A low flowrate experi-
ments and the argon feed rate to the burner was increased while leaving the CH.
and 0» feed rates unchanged. Instead of causing a decrease in temperature because
of the added heat capacity, the additional argon produced a higher combustion gas
temperature at the gas sampling point (Fig. 48) and a sizable increase in the N0x
yield (Fig. 49). The increased feed rate apparently served to lift the flame
183
-------
reaction zone away from the water-cooled burner increasing the maximum flame temp-
eratures. This change in reaction temperature tends to obscure the direct effects
of argon dilution of NO yield.
X
At = 0.8 and 1.1 (Series E in Table 21), the Ar/02 ratio was increased while the
total feed rate to the burner was kept constant at 7520 cc/min plus NH, additive
(as in the Series B experiments). Although the temperature at the sampling point
did decrease with argon dilution (Fig. 48), the change was not as large as would
be predicted from the effect of the added heat capacity on the adiabatic flame
temperature (at = 0.8, the adiabatic temeprature decreases from 2300 to 2097 K
as DR is increased from 1.0 to 1.4). Argon dilution again altered the proximity
of the flame front to the burner and, hence, changed the rate of heat loss to the
burner. At = 0.8, the measured temperature decreased sharply with argon addi-
tion and was accompanied by a decline in conversion to NO . The temperature de-
JC
creased only slightly at = 1.1 and the addition of argon had little effect on
the amount of NO formed.
x
The argon dilution data are replotted in Fig. 50 to demonstrate the approximate
direct relationship between gas temperature and NO yield at = 0.8, 0.9 and 1.0.
J\.
At = 1.1, the addition of argon under these conditions had little effect on
temperature or NO yield.
X
s
In contrast to the indications of a direct relationship between T and NO yield
yC
illustrated in Fig. 48 and 49, is the absence of an effect on NO yield of the
yt
lower reactant concentrations accompanying argon dilution. This is best illus-
trated by reference to Table 22 which lists calculated oxygen concentrations be-
fore and after combustion, assuming equilibrium combustion products at the adia-
batic flame temperature. At = 1.1, for instance,, the 02 concentration varies
widely with argon dilution; however, Fig. 49 indicates only a very small change
in NO yield. The interpretation of the argon addition results at <)> = 0.9 and
Jt
1.0 are again clouded by the increase in temperature that accompanied the increase
in flowrate. Argon addition decreases all reactant concentrations by the same
ratio (except NH_) and hence, decreases the rates of all reactions of the same
order by a constant factor. These results appear to indicate that as long as the
184
-------
•t»oo
*-0.9
- CONSTANT FLOWRATE, 7520 CC/MIN
1 1
'.(? /. 2.
l-LUWKAIt INCKtAbhU Ab AK
WAS ADDED
i ' I
SON
1 '
1, 6
DILUENT RATIO
Figure 48. Effect of Argon Dilution on Temperature at
Sampling Point (p = 76 torr, d = 80 mm)
185
-------
x
o
— CONSTANT FLOWRATE, 7520 CC/MIN
FLOWRATE INCREASED AS ARGON
WAS ADDED
2 80 \~
tz.
LU
O
o
o
OC
LU
a.
DILUENT RATIO
Figure 49. Effect of Argon Dilution on NO Yield
(p = 76 torr, d = 80 mm)
186
-------
oo
80
o
z
o
to
QC
LU
O
70
60
1500
1600
1700
= 1.1
X
1800
TEMPERATURE AT SAMPLING POINT, K (d = 80 mm)
Figure SO. Comparison of NO Yield from NH and Temperature (measured approximately 80 mm
above burner) For Argon Dilution Screening Experiments
-------
sampling point is far enough behind the flame front, the decrease in reaction
rates accompanying argon addition is of no importance to the NO yield because the
A.
competing reactions for the formation of NO and N~ are affected equally (i.e.,
J\. &
are of the same order--probably second).
TABLE 22.
EFFECT OF ARGON DILUTION ON OXYGEN CONCENTRATION OF
REACTANT MIXTURE AND COMBUSTION PRODUCTS
"oRX^
1.0
1.2
1.4
1.6
Percent Oxygen
Reactant Mixture
0.8
19.3
16.9
15.0
13.5
0.9
19.2
16.8
14.9
13.4
1.0
19.0
16.6
14.8
13.3
1.1
18.8
16.5
14.7
13.2
Combustion Products*
0.8
4.06
3.38
2.88
2.52
0.9
2.71
2.13
1.69
1.36
1.0
1.66
1.18
0.81
0.52
1 .1
0.89
0.53
0.27
0.11
'^Assuming adiabatic combustion and equilibrium products
Effect of NHs Concentration. There is ample information in the literature to
establish that the percent conversion of nitrogen additives to NO decreases as
additive concentration is increased (e.g., Ref. 59, 60, and 62). Limited evidence
from the screening tests made in this study, however, are ambigous with respect
to the effect of additive concentration. Most of the screening tests were conducted
using about 2500-ppm NH addition. This is equivalent to adding 1.6 weight percent
.j
nitrogen to the CH4 (as NH3) at <$> = 1.6 and 2.8 weight percent at = 0.8. At cf> =
0.9 and 1.0, during experiments where the argon feed rate to the burner was in-
creased (Series D in Table 21), tests were made with the NH3 feed rate adjusted
to 2500 ppm of the total burner feed including the additional argon, and then to
2500 ppm of the feed rate without the additional argon. Of the six sets of data
taken in this manner (three sets at <)> = 0.9 and three at <(> = 1.0), four sets show
an increase in NO yield at the higher NH, concentration. For <)> = 0.8, experiments
X -3
were made at 2517, 3390, and 4909 ppm NH3, with a DR of 1.6 and the probe set 20 mm
from the burner (Series F in Table 22). The NOX yields were 80.1, 77.9, and 70.0
percent, respectively. The results of this series (Series F) are in agreement
with the expected decrease in NOX yield at higher additive concentrations.
188
-------
Effect of Sampling Position. It was noted previously that the Series C NH3 exper-
iments demonstrated that in fuel-lean flames the NO concentration decreases with
j\
distance above the burner as the post-flame gases flow from the flame front to the
normal sampling point (80 mm above the burner). This effect, which is attributed
to thermal decomposition of NO, is shown in Fig. 51. A similar effect is demon-
strated by the Series G experiments (Table 21 and Fig. 51) which were conducted
at <(> = 0.7 at a rather high feed rate of 8763 cc/min. Since the thermal decompo-
sition of NO has a moderately large activation energy, the rapid decrease in NO
A
yield with distance in the Series G experiments must result from the higher temper-
atures attained at these high flowrates (because of lower heat loss rates to the
burner). The NO yields obtained in the Series H experiments with added Argon at
, -?V
4> = 0.8 (Fig. 51) were essentially independent of sampling distance. This indi-
cates that NO decomposition is slow at the low temperatures obtained under these
conditions. The NO yield in Experiment 1-1 ( = 1.5) would lie on the extension
X
of the solid curve if plotted in Fig. 47. Thus the yield from the fuel-rich flames
is apparently not strongly dependent on sampling position.
Effect of Pressure. The effect of pressure on NO yield from NH^ was measured for
4> = 0.9 and 1.2 and a diluent ratio of 1.0 using burner feed rates, of 7520 and
15,040 cc/min (plus NH,) primarily, and 10,450 cc/min in one experiment. Above
a certain pressure, the flame, which is normally stabilized above the 8-mesh screen
located about 1/8 inch above the porous plate, flashed back through the screen and
stabilized on the porous plate. Because of the possibility of damage to the porous
plate by overheating, input-output experiments were not performed under these condi-
tions. Thus, the pressure at which flashback occurred, listed in the following
table, determined the maximum pressure for the flowrate and value of $ being used:
*
0.9
0.9
1.2
1.2
Flowrate,
cc/min
7,520
1 5,0*10
7,520
15,0^0
Flashback Pressure,
atm
0.5+
«0.35
0.4+
«0.38
189
-------
90
85
to
o
80
75
-l)
(c-:
(G-l)
(H-5)
=0.7
DR=1.0
w=8?63
' <|>=0.5
'DR-1.0
'' W-601A
(A-l)
10
20
30
d , mm
50
60
70
80
90
Figure 51. Effect of Sampling Distance on N0x Yield From NH3 at a Pressure of 76 Torr
-------
While a higher burner feed rate might be expected to keep the flame from flashing
back to the porous plate until a higher pressure was achieved, just the opposite
was found to be true. This result suggests flashback is dependent on the temper-
ature of the burner top. The higher feed rate (and, hence, greater energy release
rate) could heat the burner face to a higher temperature than did the lower feed
rate, resulting in flashback at a lower pressure.
The pressure effect data are given in Table 23 and plotted in Fig. 52. It can be
seen from Fig. 52 that the NO yield was much smaller at (f> = 1.2 than at <)> = 0.9,
J^
and the difference persisted over the entire pressure range evaluated. It can be
seen also that at 7520 cc/min, the NO yield at both <)> = 0.9 and 1.2 declined as
A
pressure was increased (dashed lines). However, this decline cannot necessarily
be attributed to a pressure effect, since the measured temperature also decreased
(Table 23). To reduce the heat loss to the burner, the feed rate was doubled.
Doubling the feed rate, to 15,040 cc/min, resulted in a smaller variation of NO
J^
yield with pressure (Fig. 52): the measured temperatures were higher but still
decreased with increasing pressure. In general, the higher NO yield at the
A.
15,040 cc/min feed rate is believed to be due to a lower percentage heat loss to
the burner and a resultant higher reaction zone temperature.
Interpretation of the effect of pressure on NO formation reactions is complicated
Jv
by the fact that temperature, gas density (and therefore gas velocities), and heat
loss to the burner also vary as the pressure is changed. The final NO concentra-
J\.
tion is affected by the temperature-time history of the gas in the zone where the
temperature is high enough to result in significant reaction rates. Obtaining the
data to evaluate this temperature-time history was beyond the objectives of the
input-output screening tests. The measured temperatures listed in Table 23 were
those at the location of the microprobe. (about 80 mm above the flame holder in
most experiments). It may be that the temperatures in the reaction zone did not
vary to the same extent with pressure as did these measured temperatures.
Because of these complicating factors, a more detailed interpretation of the pres-
sure effect data does not appear to be warranted. It may be that the direct ef-
fect of pressure is small, as was apparently the case with argon dilution, because
191
-------
TABLE 23. SUMMARY OF PRESSURE EFFECT SCREENING EXPERIMENTS
WITH NH3 ADDITIVE (DR =1.0)
to
NJ
Experiment
No.
J-]
J-2
J-3
J-4
J-5
J-6
J-7
J-8
J-9
B-7
J-10
J-ll
4>
0.9
0.9
0.9
1.2
1.2
1.2
0.9
0.9
0.9
1.2
1.2
1.2
Symbol
in Fig. 52
O
O
O
A
A
A
•
•
•
.- A
A
A
Distance
From Burner,
mm
101.3
78.3
78.15
80 . 05
80.15
80.30
81.65
80.70
80.15
80.75
80.75
52. 4
Temperature ,*
K
2006
I960
1816
1876
1726
1624
1792
1785
1248
1772
1401
1483
Burner
Feed Rate,
cc/mi n
15,040
15,040
15,040
15,040
15,040
15,040
7,520
10,450
7,520
7,520
7,520
7.520
NOX,
ppm,
1927
2141
2093
1374
1358
1265
1792
1825
1402
1339
1030
816
NH3,
ppm
2522
2458
2490
2421
2402
2421
2535
2473
2523
2484
2402
2434
Percent
NH3
Converted
76.4
87.1
84.0
56.7
56.5
52.2
72.5
73.8
55.5
53-9
42.9
33.5
P, atm
0.1
0.2
0.3
0.2
0.3
0.368
0.1
0.2
0.5
0.1
0.3
0.4
-Corrected for radiation
-------
o
z
° 7=1.2, w=7520
PRESSURE, ATM
Figure 52. Effect of Pressure on the NO Yield
From NH (DR = 1)
193
-------
the rates of formation of NO and N2 are equally pressure dependent, and the effects
shown in Fig. 52 result mainly from the effect of pressure on the flame temperature.
In any event, these pressure screening experiments have shown that the NO yield
is not strongly dependent on pressure and indicate that conducting the detailed
probing experiments at 0.1 atm (to spread the flame zone) and a feed rate of 7520
cc/min should give NO yields that are comparable to (although somewhat higher than))
those that would be obtained at 1 atm.
Hydrogen Cyanide Addition
Similar types of screening (input-output) experiments were conducted with HCN as
the additive. The experience gained with NH, permitted conditions to be selected
that would give the most meaningful results.
Effect of Equivalence Ratio on NOX Yield. The percent NO yield from HCN additive
was determined as a function of the equivalence ratio, , from 0.5 to 1.5. The
data are given in Table 24 and the results are plotted in Fig. 53. These experi-
ments were all conducted at 76 torr. The HCN concentration was approximately 2500
ppm and the diluent ratio was 1.0 (i.e., Ar/0- = 3.76). Earlier experience with
TABLE 24. SUMMARY OF NOX MEASUREMENTS IN SCREENING
EXPERIMENTS WITH HCN ADDITIVE*
Experiment
No.
K-1
K-2
K-3
K-4
K-5
K-6
*
0.5
0.7
0.9
1.1
1.3
1.5
Distance
From Burner,
mm
79.70
79.70
80.85
80.85
80.25
80.00
Measured
Temperature**,
K
1538
1722
1843
1876
1678
1685
NO ,
x'
ppm
2126
2115
2002
1564
1017
652
HCN,
ppm
2530
2515
2490
2442
2377
2314
Percent
HCN
Converted
83.8
84.1
80.3
64.1
45.3
28.2
*p = 76 torr, DR = 1.0, w
**Corrected for radiation
= 7520 cc/min (7539 cc/min including HCN)
194
-------
o
o
o
<£.
Al>T>lTloti
0.4 0.7
-------
NH, addition indicated that the heat losses to the water-cooled burner varied con-
siderably with the burner feed rate employed since a lower feed rate apparently
produced a flame that hovered closer to the burner face. Therefore, a constant
burner feed rate was used in the HCN addition screening experiments for the determ-
ination of N0x yield versus curve for NH, and HCN (i.e., a high
A O
NO yield for < 1 and a rapid decline in NO yield as is increased beyond 1)
A Jv.
closely resembles results obtained in many studies using a wide variety of nitrogen-
containing organic compounds as additives (e.g., Ref. 3, 63 through 66). Similar
results have been reported for pulverized coal burned in an 0--Ar atmosphere, i.e.,
0,,-rich conditions produced a higher NO yield (Ref. 133). The similarity has
£• A,
been cited as providing strong evidence that the rate-controlling steps in the con-
version of fuel-N to NO and N9 are independent of the molecular structure of the
A £*
fuel-N compound and, hence, involve some common active nitrogen intermediate. The
data obtained from these screening experiments indicate that this commonality
applies even to the simple nitrogen-containing species, NH, and HCN.
Argon dilution was not run as part of the HCN screening experiments. From the
screening experiments performed with NH, addition, it was determined that the prim-
ary effect of argon dilution on the NO yield at constant , was probably mani-
A
fested through its effect on the flame temperature. The reason for this appears
to be that dilution with argon lowers the concentration of all reactants propor-
tionately and, hence, may lower the rates of formation of NO and N- in the same
proportion. The data from argon dilution tests with the NH3 additive were judged
to be adequate to select conditions for the detailed probing tests with HCN as
the additive.
196
-------
Effect of Pressure. The experimental data obtained on the effect of pressure on
N0v yield from HCN are given in Table 25 and Fig. 54. The pressure effect on NO
x ^
yield from HCN was measured primarily at a burner feed rate of 15,040 cc/min (plus
2500 ppm HCN) with one experiment (No. K-3, reproduced from Table 24) being con-
ducted at a feed rate of 7520 cc/min. The maximum pressure used was 185 torr,
or 0.243 atm. Higher pressures were unattainable without apparatus modification
because the HCN supply line pressure was limited to 264.4 torr, the vapor pressure
of HCN at 0 C. If pressure were found to have a large effect on N0x yield, the
HCN supply system would have been modified to provide the higher pressure required.
TABLE 25.. SUMMARY OF PRESSURE EFFECT SCREENING EXPERIMENTS,
NO YIELD WITH HCN ADDITIVE ( = 0.9, DR = 1.0)
Exper iment
No.
K-3
L-l
L-2
L-3
L-*4
1-5
Pressure,
torr/ (atm)
76(.l)
76(.I)
11M.15)
152 (.2)
185(.2*43)
185(.2*43)
Distance
From Burner,
mm
80.85
106.70
106.70
79.90
65.05
79.90
Burner
Feed rate,
cc/min
7,520
15,0*40
15,0*40
15,0*40
15,0140
15,0*40
Measured
Temperature, *
K
18*43
1965
1856
1376
1876
1808
NOX, ppm
2002
1916
2069
2106
2102
2095
HCN, ppm
2*490
2*458
2*458
2*458
2*458
2*458
Percent
HCN
Converted
80.3
78.0
8*4.2
85.7
85.5
85.2
"Corrected for radiation
Screening experiments conducted with NH, had indicated that at constant 4> the
flame reaction zone temperature had a sizable effect on NO yield. If the rate
A.
of heat loss from the flame changes as the pressure is varied, the changing of
flame reaction zone temperature could mask the true effect of pressure on the N0x
yield. Experience from similar screening experiments with NH3, led to the use of
a burner feed rate of 15,040 cc/min to minimize the effect of heat loss to the
burner on the N0v yield. The data for HCN plotted in Fig. 54 show a slight increase
A.
in NO yield with increasing pressure.* The dashed line in Fig. 54 is the correla-
A
tion obtained previously for NH.,, also at $ = 0.9. The results for HCN and NH3
are seen to be quite similar.
*The solid curve in Fig. 54 is drawn through experiment 66 rather than 70 because
it was obtained at the same residence time.
197
-------
tOOr-
x
o
z
o
X
z
o
10
>
o
o
oc
60
10 -
O
A
V- //C/V
PRESSURE, ATM
Figure 54. Effect of Pressure on NO Yield from HCN
(4)= 0.9, DR =1.0) X
198
-------
CONCLUSIONS - SCREENING EXPERIMENTS
The following conclusions have been drawn from the screening (input-output) exper-
iments with NH, and HCN addition:
1. NH, and HCN give nearly identical yields of NO under all conditions
o X
tested.
2. Less than 1 percent of the NO is N09 at the downstream probe position.
A £
3. The NO yield is high (> 80 percent) at equivalence ratios of less than
0.9 but falls off rapidly under fuel-rich conditions.
4. Increasing the burner feed rate increases the NO yield slightly apparently
by reducing the heat loss to the burner face.
5. Increasing the argon feed rate by 60 percent has only a small effect on
the NO yield from NH3.
6. The results were inconclusive on the effect of additive concentration on
NO yield but this varialbe was not studied in any detail.
7. At the higher temperature conditions, some of the fuel NO that forms de-
composes thermally before reaching the probe in these input-output exper-
iments .
8. The effect of pressure on NO yield is relatively small over the pressure
J\
range of 0.1 to 0.4 atm. However, at the feed rates selected to be used
in the detailed probing expeirments (6247 and 7520 cc/min), the NO yield
J^
at 0.1 atm will be somewhat higher than it would be at 1 atm (assuming
that the trend in NO yield versus pressure does not reverse above 0.4 atm)
J{
The screening experiment results were quite useful in planning the detailed prob-
ing experiments. In particular, information was provided on the effects of flow-
rate, equivalence ratio, diluent ratio, additive concentration, and pressure that
permitted contiions to be selected for the conducting of meaningful probing exper-
iments. In addition, the input-output results form part of the data base required
for an understanding of the mechanism of fuel NO formation. The similarity in
199
-------
the NO yields from HCN and NH3% suggests that it may be difficult to establish
whether CN or NH mechanisms predominate in the formation of fuel NO from fossil
A X
fuels. For a fuel N0x mechanism to be definitive, it will have to account for the
similarity in N0x yields from HCN and NH3 under various conditions. '
RESULTS AND DISCUSSION - DETAILED PROBING EXPERIMENTS
The screening experiments were quite informative but detailed probing experiments
were required to determine where in the flame front the NO forms and the NH, and
HCN react under various conditions, and to measure .the rates of these processes.
The probing experiments were carried out to obtain information on the following:
1. Does fuel NO form rapidly (in the CH. combustion zone)?
2. Do the similar NO yields from NH- and HCN result from the formation of a
common intermediate in the combustion zone?
3. Do the additives react in the same zone that the NO forms, suggesting
direct conversion to NO, or react well ahead of the formation of NO indi-
cating the formation of large concentrations of reaction intermediates?
Based on the results of the screening experiments, it was decided to conduct de-
tailed probing experiments at fuel-air equivalence ratios, <{>, of 0.8 and 1.5 with
the additive concentration at 2500 ppm molar based on total feed to burner. Equiv-
alence ratios closer to stoichiometric would have produced maximum flame temperatures
that could not be tolerated by the apparatus used (coated thermocouple and uncooled
quartz probe) unless nitrogen were used as diluent rather than argon.
Eight premixed CH4-0--Ar flames were studied at the conditions listed in Table 26.
The experimental data are tabulated in Appendix G. Both NH_ and HCN were used at
each equivalence ratio in the first four flame experiments. Argon dilution (DR =
1.4) was used in one HCN addition experiment at = 0.8 to lower the flame zone
temperature (flame 5). The next two detailed probing experiments (flames 6 and 7)
were conducted at = 0.8 and 1.5, respectively, with 625-ppm NO being added init-
ially along with the HCN. The final detailed probing experiment was performed at
200
-------
TABLE 26. CONDITIONS FOR PROBING EXPERIMENTS
Flame
No.
1
2
3
4
5
6
7
8
Nomi nal
Addi ti ve
Concentration,
1
ppm
NH3 (2500)
HCN (2500)
HCN (2500)
NH3 (2500)
HCN (1793)
HCN (2500)
+NO (625)
HCN (2500)
+NO (625)
NO (675)
Equivalence
Ratio,
*
0.8
0.8
1.5
1.5
0.8
0.8
1.5
1.5
Di 1 uent
2
Ratio
1 .0
1.0
1 .0
1 .0
1.4
1.0
1 .0
1 .0
Burner
Feed Rate ,
cc/mi n
7520
7520
6247
6247
5851
7520
6247
6247
Calculated
Adiabatic
Fl ame
Temperature,
2295
2306
2231
2227
2097
~2306
~2231
~2230
Measured
Temperature
Maximum ,
K
2078, 19785
2050
2012
2080
1890
1992
2030
1976
Distance From
Burner at
max'
mm
6.0
6.5
8.3
8.0
6.7
6.3
8.3
7.3
Based on total feed gas to burner
Ar/0. molar ratio divided by 3-76
i
Excluding additive flowrate
[j
Corrected for thermocouple radiation
Rerun - Temperature difference attributed to higher flowrate of coolant caused by increased
water pressure.
-------
{() = 1.5 with NO addition only. The experiments were conducted at 76 torr with a
total burner feed rate of 6247 cc/min, excluding the nitrogen additive, at =
1.5 and of 7520 cc/min at <|> = 0.8 (except for the dilution experiment in which the
feed rate was 5851 cc/min).
The analyses by mass spectrometry of 1-liter batch samples of the collected gas,
using the procedures described in the Experimental section, gave the mole fractions
of CH4, 02, CO, C02, Ar, C2H6, C^, H2CO, and H2 in the flames at various distances
from the burner. However, this technique did not give reliable measurements for
the concentrations of the species NH,, HCN, and N2 at the low concentrations in-
volved in these experiments.
Except for N2, the nitrogen species were measured with a chemiluminescent analyzer
(CA) employing a molybdenum converter operated at 800 C. At this temperature, and
in the presence of oxygen, the species HCN, NH,, (CN)2, and N02 are all converted
to NO to some extent. Calibration experiments (Fig. 46) have shown that if suffic-
ient oxygen is present, this converter is quite efficient for the conversion of
N02 and HCN to NO, but only about 50 percent of the NH- is converted*. It was
also demonstrated that (CN)2 will form NO in this converter at 800 C, but cali-
bration experiments were not conducted with (CN)2> If the converter temperature
is reduced to 400 C, only N0_ is converted to NO permitting NO- to be distinguished
from oxidizable nitrogen species. NO- measurements were made in the flame 2 and
4 experiments only. The amounts formed were quite small and N0» was detectable
only in the early stages of the flame. The concentration of molecular nitrogen
in the flames was not determined because of high nitrogen blanks that were encount-
ered in the mass spectrometric results.
Data Interpretation ;
In reviewing the data obtained in these flame-probing experiments, it should be
noted that when a species is reported as "HCN" or "NH-," its mole fraction was
obtained by dividing the measured concentration of NO formed in the converter by
"Perhaps the remainder decomposes to N» before it can be oxidized.
202
-------
the converter calibration factor for the denoted species. It will be seen that,
despite the limitations that this inability to distinguish directly between HCN
and NH, puts on the interpretation of the data, considerable new information was
obtained in this study regarding the chemical processes involved in the conversion
of nitrogen species to NO in flames. In probing atmospheric flames, DeSoete was
able to measure HCN and (CN)_ by specific chemical methods but did not measure
NH, or N-. Merryman and Levy measured NO- and convertible nitrogen, which would
include both NH, and HCN, but did not measure N2. Modifications are planned for
the chemical analysis system of the apparatus used in the present study that will
permit NH, and HCN to be measured independently and may allow N- to be measured
directly.
It will be seen that all of the nitrogen introduced into a flame via the additive
is not recovered in the measured products once the reaction temperature has been
reached. Most of the missing nitrogen is undoubtedly inthe form of N? in the later
stages of the flame but it will be seen that under certain conditions (Flame 1:
NH,,
-------
then be converted to NO in the converter. This leads to the interesting conclusion
that even if the small quantities of N^ present in the gas samples had been measured,
there is no assurance that all of it existed as N2 in the flame front. This is,
of course, the classic problem associated with flame-sampling experiments.
The temperature and mole-fraction profiles obtained in these probing experiments
were used to calculate flux profiles for the various species of interest through
the flame, and the flux profiles were used, in turn, to calculate the rates of
production and depletion of these species at various points in the flame. It will
be seen that to do this properly, the diffusion of the species must be taken into
account. If this is not done, erroneous conclusions can be drawn. For example,
NO may appear to form earlier in the flame that it actually does because it dif-
fuses upstream from the zone in which it first starts to form from chemical reac-
tion. Also, the N-additive would appear to start reacting earlier in the flame
than it actually does because it diffuses downstream.
The rate of reaction of a species i in a 1-dimensional flow is given by the
expression:
which is derived as shown in Appendix H. R. is the rate of reaction (loss) of
*7
species i in moles/cm -sec, v is the gas velocity in cm/sec, and *f. . is the
2 1>-'
diffusion coefficient for i in excess j in cm /sec. If the diffusion velocity
of species i is defined as:
Then, Eq. 127 becomes:
R. = d [c.(v+v.)l (129)
i dx L 1 * J
The quantity (L (v+V.) is the molar flux of species i in moles/cm -sec. Because of
diffusion, the velocity of species i is not the same as that of the bulk gas.
204
-------
It can be seen that a plot of the flux for each species is very informative because
the flux only changes as the result of chemical reaction whereas species mole frac-
tions change as the result of both diffustion and chemical reaction. The slope
of the flux (on a distance plot) gives the reaction rate for each species, as shown
in Eq. 129.
The diffusion velocity was calculated from the slopes of plots of measured species
mole fraction, f., versus distance using the following equation which is a modi-
fication of Eq. 128 (the change in C^ with T was neglected in making the transor-
mation, Ref. 131):
df.
i_
dx
(130)
Thermal diffusion was ignored because, generally, it is of negligible importance
for the components considered in this study (only the low mo.lecular weight species
such as H or H_ might require its use). As is usual for these conditions, the
diffusion coefficients were calculated using Lennard-Jones potential parameters
obtained from viscosity measurements and an empirically fitted expression for the
Lennard-Jones collision integral (Ref. 131). The diffusion coefficients were
calculated for the minor species NO, HCN, NH3, 02> CH4, and C02 diffusing through
the major species, Ar. The final expressions for the diffusion coefficients re-
sulting from the above considerations are shown below at P = 0.1 atm:
j# AT1 .67 2,
^. . = AT cm /sec
1 >J
Species (i)
NO
NH
HCN
C00
2
°2
Species (j)
Ar
Ar
Ar
Ar
Ar
Ar
A
0.1463 x 10"3
0.1854 x 10"3
0.1282 x 10"3
0.1282 x 10"3
0.1631 x 10"3
0.1467 x 10~3
T*. , K
mm
308
763
771
472
388
*Minimum Temperature for which calculated/^ . are recommended.
^-» 3
205
-------
The gas velocity as a function of distance through the flame was determined from
the known mass feed rate to the burner, w, by the equation
where T is the measured temperature, P the pressure, A the burner area*, and M the
molecular weight of the gas. Gas molecular weight was determined from the mass
spectrometric analyses of the composition of the gas samples.
2
Values of the species flux, C. (v+V.), moles/cm -sec, were calculated as a function
of distance through the flame. The slope of the flux versus distance correlation,
or the reaction rate, was measured as a function of distance to obtain the species
reaction rate profiles through the flame. The average gas residence time was cal-
culated as a function of distance by integrating Eq. 131.
The importance of the diffusion calculations to the interpretation of the detailed
probing results is illustrated in Fig. 55 which presents some of the Flame 1 results
that will be discussed further in the next section. In Flame 1, 2500 ppm NH_ was
added to a CH.-CL-Ar flame at an equivalence ratio of 0.8. The CH. mole fraction
curve in Fig. 55 shows that by the time the gases had reached the first CH, sampling
point, which was only 0.5 mm above the top of the flame holder (screen), the CH,
mole fraction had dropped to about one-half of its initial value. However, the
calculated CH. flux curve demonstrates that the decrease in mole fraction up to
that point resulted only from the downstream diffusion of CH, into the luminous
zone. Virtually none of the CH. reacted below the bottom of the luminous zone.
The NO mole fraction is appreciable and increasing below the bottom of the luminous
zone and has reached about three-quarters of its final value at the top of the
luminous zone. However, the calculated NO flux curve, which was obtained from the
mole fraction curve by correcting for diffusion, indicates that nearly all of the
NO present in the post-flame gases forms just above the top of the luminous zone.
The rather large concentrations of NO that are present in the luminous zone result
*Sufficient purge argon flows around the burner to minimize divergence of the
flame
206
-------
too -—
N>
O
, .MOLE FRACTION
3
Figure SS. Comparison of Mole Fraction and Flux Curves,
Flame 1, NH3 Addition With = 0.8
-------
from the upstream diffusion of NO. A net consumption of NO occurs in the luminous.
Most of the NO consumed was formed at the top of the luminous zone but part was
formed below the luminous zone. Of course, additional NO could be simultaneously
formed and consumed at points in and below the luminous zone giving no net effect
on the flux and mole fraction curves.
Comparison of NOX Flux Profiles
Before presenting the detialed results that were obtained under each condition,
the NO flux profiles will be compared. The NO yields are plotted in Fig. 2 (which
may be found in the Report Summary) relative to the top of the luminous zone (be-
cause the luminous zone is thinner in fuel-lean flames, the time scales were ad-
justed for this comparison so that the tops of the luminous zone coincide). Be-
cause these NO yields were calculated from the NO flux curves (to be presented)
the correction for diffusion has been taken into account. It can be seen that
relatively little NO is produced in or below the luminous zone. At <|> = 0.8, where
NO forms in high yield, the NO forms rapidly just above the top of luminous zone
(particularly with NH- as the additive). Under fuel-rich conditions, 4> = 1-5,
much of the NO forms slowly in the post-flame gases far above the luminous zone.
Temperature Profiles
Compared in Table 27 are the temperature profiles obtained under the conditions
employed in the detailed probing experiments to be discussed. The listed temper-
atures were taken from Tables 21 and 26. The maximum measured temperatures, which
occurred about 6 to 8 mm above the burner, were 200 to 300 degrees lower than the
adiabatic flame temperatures. The primary reason was heat loss to the water-
cooled burner with heat losses to the rest of the surroundings also taking a toll.
The maximum temperature was reached closer to the burner at = 0.8 than at <)> =
1.5 even though the burner feed rate was larger. It will be seen that reaction
occurs over shorter distances in the presence of excess oxygen. Increasing the
diluent ratio to 1.4 moved the position of maximum temperature further from the
burner even at a lower feed rate. This presumably results from lower reaction
rates, resulting from lower temperatures and species concentrations. The measured
208
-------
TABLE 27. TEMPERATURE PROFILES (P = 0.1 atm)
Equivalence Ratio, $
Di luent Ratio, DR
Burner Feed Rate, cc/min
Adiabatic Flame Temperature, K
Maximum Measured Temperature, K
Range
Average
Distance above burner, mm
Screening Experiment Temperature, K
d = kS.k mm
d = 80.7 mm
0.8
1.0
7520
2300
1978-2028
2025
6.3
--
1731
' 1.5
1.0
62k7
2229
1976-2080
2025
8.0
1776
--
0.8
l.J»
5851
2097
1885
1885
6.7
—
--
temperature decreased by about 250 degrees at distances far above the burner
giving temperatures 50 to 80 mm above the burner that were lower than the adiabatic
flame temperature by about 450 to 550 degrees.
In the following discussion of results, plots are presented as a function of time
through the flame for: (1) temperature, (2) species mole fraction, (3) species
flux, and (4) species reaction rate. Time zero was taken to be 1550 K for = 0.8
and 1250 K for = 1.5 because measurable reaction occurred at lower temperatures
in fuel-rich flames. The data from each flame study are accompanied by a discus-
sion of their interesting features and implications.
Fuel-Lean Flames ( = 0.8)-- Detailed Probing
The first two flames studied (Table 26) were with NH3 and HCN addition at 2500
ppm, DR=1, = 0.8, and feed rate = 7520 cc/min. The expected NO yields based on
the screening experiments (Fig. 53) were about 82 percent for NH3 and 83 percent
209
-------
for HCN.* The reactant gas composition under these conditions (in mole percent)
was: CH4, 7.73; 02, 19.33; Argon, 72.69; and additive 0.25. The nitrogen content
of the mixture was that which would be obtained if 2.83 percent nitrogen were
added to the CH4 (as NH3 or HCN).
Flame l--Ammonia Addition With 4> =0.8. The data acquired from the detailed prob-
ing of Flame 1 are given in Tables G-l, G-2, and G-12 in Appendix G and in Fig.
56 and 57. Temperature, and ppm NO profiles are given for the initial Flame 1 ex-
periment and a temperature profile and ppm profiles for all major species are
presented for a complete rerun. The two temperature profiles plotted in Fig. 56
show an almost constant difference of about 100 K on a plot of T versus distance
from the burner. The temperature difference is believed to have been caused by
a change in the flowrate of cooling water to the burner.
The ppm NO profiles from each run, plotted in Fig. 57, are quite similar. The
somewhat steeper ppm curve for the rerun caused some minor differences in the
calculated NO flux profiles. The NO flux from the rerun experiment (Fig. 58) goes
negative near the top of the luminous zone and then rises rapidly just past the
luminous zone. An NO flux profile calculated from the initial Flame 1 experiment
-8
(not shown) also had a value of 4 x 10 below the luminous zone but then remained
constant through the luminous zone increasing rapidly just above the luminous zone.
These differences do not, however, affect the major conclusion that under fuel-
lean conditions nearly all of the NO forms just above the luminous zone in the
region where most of the C02 forms. This demonstration of the sensitivity of
the flux profiles to the slopes of the mole fraction curves is one of several
that will be encountered.
The NO flux curve in Fig. 58 indicates that some NO, about 10 percent of the total
NO, forms below the first measurement point, i.e., at or below the flame holder.**
However, all of this NO plus the NO that forms just below the top of the luminous
zone react in the luminous zone giving no net NO flux at the top of the luminous
*Even higher yields might be expected because the screening experiments indicated
that at = 0.8 (excess oxygen) some NO decomposed before reaching the downstream
probe positions in the higher temperature screening experiments.
**The possibility that the flameholder screen catalyzes the oxidation of part of
the NH_ to NO cannot be ruled out.
210
-------
1900
ISM
(0
IMP
1500
II I II I
2 N
•a
I ~
§ . -
I I
l.u
z,o
Msec,
S
in
3.0
6.0
•7.6
J J
8.0
Figure 56, Flame Temperature vs Time, Flame 1, NH3 Addition With <|> = 0.8
211
-------
NJ
I—•
K)
/-*
IZ
10
•4
'o
•^
/.O
T/ME y MSEC.
/.S
-iff
Figure 57. Species Mole Fraction vs Time, Flame 1, NH, Addition With = 0.8
-------
TIME,
Figure 58. Species Flux vs Time, Flame 1,
NH Addition With (j) = 0.8
-------
zone. The upstream diffusion of NO reaches its maximum at about the position
(0.5 msec) where the inflection point in the NO mole fraction curve occurs (maxi-
mum slope). This gives rise to a negative NO flux in that region of the flame.
Species mole fractions change both from diffusion and chemical reaction whereas
the flux for each species is constant unless chemical reaction is occurring. For
this reason, the flux plots will be emphasized in the analysis of the data ob-
tained in these flame probing experiments. From Eq. 129, the slope of the flux
curve (on a distance plot) gives directly the reaction rate of a given species
(in units of moles/cm -sec). Reaction rates calculated in this manner for Flame 1
are plotted in Fig. 59. The maxima in the reaction rate plots indicate the region
in the flame at which each reactant and product is lost or formed most rapidly.
Because the rates are plotted as a function of time, the area under a given rate
curve represents the extent of reaction (in moles/cm ) that occurs in that region
of the flame. (The reaction rate is defined in Eq. 127 as positive when concentra-
tion decreases with increasing distance from the burner; this definition is retained
in the rate plots presented for each flame.)
Denoted on each flux plot (e.g., Fig. 58) for comparison, are the position of the
observed luminous zone and the points in the flame at which selected temperatures
are reached. These five temperatures are, respectively, the temperature at zero
on the time scale (1250 K for 4> = 1.5 and 1550 K for = 0.8), those at the bottom
and top of the luminous zone, the temperature that is 20 degrees below the maximum
temperature, and finally, the maximum temperature. Because the temperature profile
is quite flat near the maximum temperature (e.g., Fig. 56), the point where the
temperature is 20 degrees below the maximum is identified on the flux plots to in-
dicate where the temperature begins to approach the maximum value. Also included
on the flux plots, at the left edge, are the initial fluxes of CH4 and the nitro-
gen additive (s). These fluxes are calculated from the metered reactant flowrates
and the area of the burner.
214
-------
tn
10
ft
-z
REACTION RATE,
MOLES/CM3-SEC
= 0.8
IO
o,?
1.0
1.5
Figure 59. Species Reaction Rate vs Time, Flame 1,
NH3 Addition With = 0.8
-------
The NO flux in Flame 1 reached its maximum value in about 1.5 msec on the assigned'
time scale* and then remained constant (Fig. 2 and 58). The final NO flux was 77
percent of the meter.ed NH3 flux. This is slightly lower than the 82 percent (or
greater) yield that would have been predicted from Fig. 53. Between'0.5 msec and
0.85 msec, the point at which the temperature reached 1958 K or within 20 degrees
of the maximum, the NO flux increased from 10 percent of its final value to 76
percent of the final value. Thus, two-thirds of the NO formed just above the lum-
inous zone in a region 0.35 msec "thick" where the temperature range was 1925 to
1960 K. This is the same region or the flame in which most of the CO- is formed
(Fig. 58). The NO reached its maximum rate about 0.1 msec after the C02 (Fig. 59).
The nitrogen species that were oxidized to NO in the molybdenum converter at 800 C
were measured and plotted as "NH," in Fig. 57 and 58. This is probably NH, but
some of the NH, could conceivably have been converted to HCN (or cyanogen, but
this is less likely). The "NH3" mole fraction and flux curves were calculated
using the converter calibration factors and the diffusion coefficients for NH^.
The NH, reacted so early in the flame that the "NH," mole fraction had decreased
to one-third its initial value (by reaction and/or diffusion) by the time the
first sampling point had been reached despite the fact that this point was only
about 0.1 mm above the top of the flame holder (Table G-2). Because of the
limited number of data points taken and the steepness of the "NH," mole fraction
curve, the "NH.," fluxes calculated at the first three data points were erratic.
It was established, however, that the "NH," flux was less than 20 percent of its
initial value at the fourth sampling point (0.3 msec) and decreased to nearly zero
at the top of the luminous zone. Even if the "NH," measured in the luminous zone
were actually HCN formed from the added NH,, the HCN flux would also be near zero
at the top of the luminous zone. Because it was not possible to determine from
the data exactly where most of the NH- reacted and how much, if any, was converted
to HCN, the additive flux curve was dotted up to fourth sampling point and labeled
"NH3."
*It can be seen from the time-distance correlations in Fig. 56 and Table G-2 that
this point is at a distance of 6.3 mm above the burner. Since the distance zero
was taken at the bottom of the flame holder screen and this screen is 1.3 mm
thick, the NO reaches its maximum concentration about 5.0 mm above the flame-
holder.
216
-------
The observations that (1) all of the added NH, and any measurable nitrogen species
formed react before reaching the top of the luminous zone and (2) no net formation
of NO occurs below this point in the flame lead to the very interesting conclusion
that, at <|> = 0.8, all of the added NH, is converted to nitrogen species that are
not measurable by the converter-CA technique before the rapid formation of NO be-
gins. This results in the "N-BAL" flux in Fig. 58 being equal to the initial NH3
flux near the top of the luminous zone. The nitrogen balance flux curve, N-BAL,
was obtained at each point by subtracting the "NH," and NO fluxes from the initial
NH3 flux.
It can be seen from Fig. 58 that the nitrogen intermediate(s) that apparently
forms quantitatively from the NH, reacts rapidly just above the luminous zone to
form NO in high yield. It may react with 0 or OH radicals, which increase rapidly
in concentration in this region, or with 02 which is still present at a concentra-
tion of more than 6 percent in this fuel-lean flame.
Only small amounts of N02 were detected in and just above the luminous zone (Table
G-l). The maximum mole fraction of N02 was about 8 ppm or 0.3 percent of the
added NH,. N0~ was not detected in the later stages of the flame. Early N09 has
O 6* £
been detected by others (Ref. 101, 134, and 135). Whether it is present in the
flame, or the result of reactions occurring in the microprobe during quenching,
is apparently still a moot point (Ref. 135).
It will be seen that under both lean and rich conditions nearly all of the CH,
reacts in the visual luminous zone. The time required for the gases to travel
through the luminous zone is about 0.4 msec at = 0.8 and 0.9 msec at § = 1.5
(see Fig. 2). The oxygen reacts in the luminous zone with the CH. and in the near
post-flame gases where CO is converted to C02. The flux curves for 02 and CO were
not calculated. The C02 flux was used to determine the rate of CO oxidation. In
several of the flames, including Flame 1, the C02 flux curve goes negative in the
luminous zone. This is not believed to be a real effect because it does not occur
217
-------
in all flames of the same <)> and, should some C02 be consumed in the luminous zone,
the flux curve should attain a negative slope in that region as a result of this
reaction.
To check the carbon balance, the CO flux can be calculated readily at the point
where CO reaches its maximum mole fraction (about 0.58 msec in Fig. 57) because
the diffusion velocity of CO is zero at that point (Eq. 130) and the CO flux is
simply the product of its concentration and the bulk gas velocity. The concentra-
-8 3
tion of CO is 3.18 x 10 moles/cm and the velocity is 321 cm/sec giving a CO
flux of 1.02 x 10 moles/cm -sec. At this point, the CH4 flux is 0.25 x 10 and
the calculated CO- flux is about zero giving a total measured flux of carbon species
of 1.27 x 10"5 which is 83 percent of the initial CH4 flux of 1.54 x 10"5. If the
C02 flux curves in Fig. 58 were shifted upward so that the flux at 0.7 msec were
zero, the carbon balance would become 106 percent again indicating that the neg-
ative CO- fluxes are not real.
Flame 2--HCN Addition With (j> = 0.8. The data obtained from this experiment are
listed in Tables G-3 and G-13 in Appendix G. The measured temperatures and species
mole fractions are plotted in Fig. 60 and 61. The calculated fluxes and reaction
rates are plotted in Fig. 62 and 63.
The NO flux (Fig. 62) has a slight positive value at the first sampling point and
then remains constant through the luminous zone. NO formation occurs rapidly
above the luminous zone as was the case when NH_ was added to the fuel-lean flame.
It can be seen from Fig. 2, however, that the maximum rate of NO formation occurs
about 0.24 msec later when the additive is HCN. The NO measurements were carried
out to 1.7 msec (7 mm above the burner) but the NO flux had not yet reached its
maximum value (Fig. 2). The NO yield at 1.56 msec'was 63 percent and was increas-
ing at a rate that indicated that it would have reached a maximum value in the
range of 80 percent near the value of 83 percent that was obtained in the HCN
screening experiments (Fig. 53). As discussed, values greater than 83 percent
should have been obtained in the probing experiments because some NO apparently
decomposed (at
-------
two
Zaoo
1900
1800
nao
I
1C
JS
ft
1500
1300
TEMPERATURE
HCN, <(> = 0.8
*t
1.
-O 1 •»••
T/M&':,
J^ » w.
^ i S
ts
>e
N
I I
3
D/STANCE .
Figure 60. Flame Temperature vs Time, Flame 2, HCN Addition With = 0.8
219
-------
rtoar-
II
t ~\
-lo
Figure 61. Species Mole Fraction vs Time, Flame 2, HCN Addition With <}> = 0.8
-------
N)
NJ
-•OS
7/Al£,
Figure 62. Species Flux vs Time, Flame 2,
HCN Addition With = 0.8
-------
10
Is)
tsj
REACTION RATE,
MOLES/CM3-SEC
HCN, <(> = 0.8
16
I.O
TIMS , MSEC.
Figure 63. Species Reaction Rate vs Time, Flame 2,
HCN Addition With = 0.8
-------
Mole fractions and molar fluxes were calculated for the nitrogen species that
were oxidized to NO in the molybdenum converter using the calibration factors and
diffusion coefficients for HCN. It can be seen from the "HCN" flux curve in Fig.
62 that more than 60 percent of the added HCN survives the luminous zone. The HCN
reacts gradually below and through the luminous zone and then more rapidly above
the luminous zone. It is apparently the low reactivity of the HCN that causes
the NO to form farther above the luminous zone in Flame 2 than in Flame 1.
It is likely that the curves labelled "HCN" actually do represent HCN because if
any NH, were to form it would have reacted very rapidly as in the Flame 1 experi-
ments. The formation of (CN^ is not likely because any CN radicals formed should
abstract a hydrogen atom and reform HCN rather than combine to form cyanogen. In
addition, DeSoete (Ref. 64) has shown that cyanogen forms HCN in a hydrocarbon
flame.
The rate maxima in Fig. 63 indicate that NO forms from HCN in the same region of
the flame in which CO is oxidized to C02 (suggesting that OH radicals may be in-
volved as was proposed by Fenimore, Ref. 66). The rate of maximum NO formation
lags the maximum rate of HCN reaction by about 100 microseconds (Fig. 63) indicat-
ing that an intermediate species forms from the HCN. The formation of an inter-
mediate from HCN is also demonstrated by the flux profiles. The N-BAL flux reaches
a maximum that represents nearly 50 percent of the added HCN and then decreases.
Nitrogen species other than N~ must be present because the N-BAL flux must finally
_7 ^
decrease to 1.0 x 10 if an 80 percent yield of NO is obtained. An even stronger
related argument for the hypothesis that much of the NO does not form from the
direct oxidation of HCN is that from 0.8 to 1.6 msec on the assigned time scale,
the NO flux increases by 2.0 x 10" while the HCN flux decreases by only 1.3 x
10~ . In fact, if an 80 percent NO yield is obtained, the NO flux would increase
by 2.8 x 10" from its value at 0.84 msec and yet the HCN flux is only 1.2 x 10" .
The maximum rate of NO formation (Fig. 63) occurs at 0.85 msec and is larger than
the HCN decay rate at that point by a factor of 2.3
223
-------
Comparison of the Flame 1 and Flame 2 results demonstrates that although NH., and
HCN form similar amounts of NO at 0 = 0.8, their rates of reaction in the flame
are quite different. A significant fraction of the HCN remains unreacted up to
as much as 1 msec longer than does NH3> The results indicate that neither NH3
nor HCN form NO directly at an equivalence rate of 0.8 but rather the NO forms
from a moderately long-lived nitrogen intermediate. The intermediate cannot be
NH3, HCN or (CN)2 because these would be measured by the converter-CA technique
It is not likely to be CN or NH radicals, because these would probably form NO
A
in the probe-converter system; or N atom, because it would be too short lived;
or N2, because its (fractional) rate of conversion to NO is probably too slow.
We postulate that this intermediate is the NCO radical. To explain the results
obtained from Flames 1 and 2, the intermediate would have to (1) form quantitatively
fromNH- and in at least 50 percent yield from HCN, (2) be oxidized later in the
flame to form NO in high yield, and (3) recombine in the probe to form N_
so that it could not be converted to N- in the molybdenum converter.
Fuel-Rich Flames ((fr = 1.5)-- Detailed Probing
Flames 3 and 4 (Table 26) were with HCN and NH3 addition at 2500 ppm under fuel-
rich conditions (4> = 1.5, DR=1, and feed rate = 6247 cc/min. The reactant gas com-
position under these conditions (in mole percent) was: CH,, 13.58; 02» 18.10;
Argon, 68.07; and additive 0.25. The nitrogen content of the mixture was that
which would be obtained if 1.61 percent nitrogen were added to the CH. (as NH, or
HCN).
Flame 3--HCN Addition With 4> = 1.5 . The data obtained from the detailed probing
of Flame 3 (Tables G-4, G-5, and G-14) are plotted in Fig. 64 through 67. The NO
and HCN data were obtained in a partial rerun because experimental difficulties
were encountered with the CA measurements in the first Flame 3 experiment.
The NO flux curve in Fig. 66 is near zero at the top of the luminous zone, in-
creases slightly just above the luminous zone and then increases gradually with a
small inflection starting at 2.7 msec. The Flame 3 flux yield is plotted in Fig. 2
for comparison with the other conditions. It can be seen that for fuel-rich flames
224
-------
20VO
/800
1606
IZOO
1000
TEMPERATURE
HCN, <
(>= 1.5
Figure 64. Flame Temperature vs Time, Flame 3, HCN Addition With
<(> = 1.5
22S
-------
10
to
Figure 65. Species Mole Fraction vs Time, Flame 3, HCN Addition With $ = 1.5
-------
N)
K)
-vl
Figure 66. Species Flux vs Time, Flame 3, HCN Addition With $ = 1.5
-------
K)
to
00
Figure 67. Species Reaction Rate vs Time, Flame 3, HCN Addition With = 1.5
-------
most of the NO forms far above the luminous zone and well above the region where
CO is oxidized. The NO flux at the last measurement point in Flame 3 was 23.4
percent of the initial HCN flux and was increasing slowly. The screening results
suggest that the NO yield would have reached about 32 percent if additional data
points had been taken further above the burner. HCN experiments were not conducted
at this lower flowrate but NH, gave 32 percent NO yield (Run 1-1 in Table 21) and
HCN gave about the same yields under all conditions.
The "HCN" mole fraction in Fig. 65 decreased gradually and almost linearly through
and after the luminous zone except for a dip near the top of the luminous zone.
The "HCN" flux (Fig. 66) calculated beyond this dip in the mole fraction curve
remained nearly constant for about 0.4 msec at the top of the luminous zone and
then decayed very slowly. The "HCN" flux was only slightly less than the NO flux
at the last measurement point. About two-thirds of the added HCN apparently sur-
vives the luminous zone and then reacts very slowly with 20 percent of the initial
HCN remaining unreacted well above the flame front (at about 4 msec on the assigned
time scale). This suggests that considerable HCN may pass through the burner under
these fuel-rich conditions. The moly converter was not operated at 800 C in the
screening experiments in which the probe was 80 mm above the burner. Such "exhaust"
measurements of HCN will determine the amount of HCN that survives under these
conditions.
The "HCN" flux calculated in the luminous zone (Fig. 66) dropped slightly at the
bottom of the zone and then remained constant until just above the middle of the
zone where it increased to a sharp maximum before dropping abruptly to its lower
constant value below the top of the luminous zone. The shape of the "HCN" flux
curve in the region where it apparently increases to a maximum value depends on
the shape of the mole fraction curve near its dip. More data points in this region
are required to determine if this spike in the flux "HCN" curve is real. In any
event, it cannot be as large as shown in Fig. 66 because the top of the spike is
well above the initial HCN flux. These conditions were repeated in a later exper-
iment, Flame 7, except that some NO was added intially with the HCN. It will be
seen (Fig. 83) that the "HCN" flux curve again dropped abruptly near the center of
the luminous zone and then leveled off. No spike in the flux curve was obtained
in Flame 7.
229
-------
The CC>2 reaches its maximum rate of formation at 1.6 msec. At this point, only
one-third of the HCN has reacted and only 17 percent of the final NO has formed.
It is apparent, therefore, that under fuel-rich conditions, fuel NO forms from
HCN slowly over a considerable distance above the luminous zone. Because of the
low NO yields obtained under fuel-rich conditions, it is not possible to determine
if most of the NO forms directly from HCN or if an intermediate is involved. The
NO formation rate (Fig. 67) never exceeds the decay rate of HCN except just above
the top of the luminous zone at which point the NO must form from an intermediate.
Flame 4--Ammonia Addition With <(> _= 1.5. The data acquired on Flame 4 are listed
in Tables G-6, G-7 and G-15 in Appendix G. The initial experiment with the fuel-
rich Flame 4 gave the temperature profile shown in Fig. 68 and the species mole
fractions plotted in Fig. 69 for those species measurable by mass spectroscopy.
A partial rerun of Flame 4 was carried out (Table G-7) in which the NO concentra-
tion measurements were made and the NO formed in the molybdenum converter at 800 C
was measured. For reasons that will be discussed, the NO formed in the converter
was plotted in the mole fraction figure (Fig. 69) both as "NH," and as "HCN" by
using the converter calibration factors for NH- and HCN, respectively, and two
flux are presented for this flame (Fig. 70 and 71).
The NO flux profile for Flame 4 shown in Fig. 70 shows an apparent inflection at
a point 6 mm above the burner or at about 1.45 msec on the assigned time scale.
This inflection in the NO flux curve indicates the rapid formation of NO near the
top of the luminous zone followed by a region of little NO formation and then a
second stage of NO formation as the maximum temperature is approached. The ap-
parent two-stage nature of NO formation in this flame is indicated by the two
rate maxima in Fig. 72. It sould be noted, however, that the first peak results
from a small inflection in the NO concentration curve. The inflection in the
NO flux profiles cannot result from the formation of N0_ because only small amounts
of NO. were formed (Table G-7) and this was in the preflame and luminous zones.
The NO flux at the last measurement point represents 23 percent of the added NH3
(Fig. 2) but appears to be increasing. This is in agreement with the prediction
from the screening experiments that 32 percent of the NH, should form NO at $ = 1.5
230
-------
$.000
.*
v^ ts*0
X '
-v.
v: .
V
1000
TEMPERATURE
= 1.5
I/WI*""*
, /msec.
«
8" O
-------
to
CM
N)
10
tf
eS"
Figure 69. Species Mole Fraction vs Time, Flame 4, NH, Addition With = 1.5
-------
2.5-
2.0
I 2
TIME, MSEC
Figure 70. Species Flux vs Time, Flame 4, NH, Addition With
(Assuming Additive Remains as NHJ
= 1.5
-------
TLME,
Figure 71. Species Flux vs Time, Flame 4, NH_ Addition With = 1.5
(Assuming Additive Forms HCN)
--as
-------
REACTION RATE,
MOLES/CM^-SEC
NH3, 4)= 1.5
2.5
Sf
z
15
Figure 72. Species Reaction Rate vs Time, Flame 4,
NH3 Addition With $ = 1.5
(Assuming Additive Forms HCN)
235
-------
and this feed rate (Run 1-1, Table 21). Whether or not the NO actually forms in
two discrete stages, it is apparent that in this fuel-rich flame most of the NO
forms late in the combustion process, as was the case with added HCN in Flame 3.
Assuming that the final extent of conversion of the additive nitrogen to NO reached
30 percent (above the last measurement point), only 15 percent of the total NO was
formed in the luminous zone and only one-third was formed by the end of the first
reaction stage (1.7 msec).* Thus, about two-thirds of the NO was formed in the
post-flame gases in the approximate temperature range 2055 to 2075 K (actually,
the 25 percent of the NO estimated to form beyond 3.1 msec formed at somewhat
lower temperatures because the temperature begins to decrease at this distance
from the burner). The maximum rate for this second stage of NO formation occurred
1.2 msec above the top of the luminous zone and 1.0 msec above the point where CO-
formation reached its maximum rate. Only about one-half of the final NO had formed
even at this late point in the flame.
Consider next the "NH " concentration curve in Fig. 69. NH, was added at 2500 ppm
and the measured "NH," concentration was about constant at 1800 ppm until the top
of the luminous zone beyond which it decayed rapidly. It can be seen from Fig.
70, however, that if all of the nitrogen measured by the converter were NH,, the
_y *
sum of the fluxes for NH_ and NO at 1.6 msec would be 4.7 x 10 which is in ex-
-7
cess of the initial NH, flux of 4.1 x 10 (giving a negative value to the N-BAL
curve). Also, the calculated "NH," flux increases in the luminous zone reaching
a maximum at 1.6 msec. It is apparent, therefore, that all of the nitrogen species
that is oxidized to NO in the converter is not NH,.
DeSoete (Ref. 63) reported the formation of significant amounts of HCN "early in
the flame" when NH3 or amines were added to sufficiently fuel-rich ethylene flames.
The converter-formed NO measured from Flame 4 was recalculated using the converter
calibration factor for HCN and plotted as "HCN" in Fig. 69, 71 (using the diffusion
coefficient for HCN) and Fig. 72. The calculated "HCN" mole fractions and fluxes
are smaller than those for "NH3" because, with air addition to the sampled gas,
the extent of conversion of HCN to NO in the molybdenum catalyst is greater than
with NH3.
*The area under the second NO peak in Fig. 72 will be twice that of the first peak
when a 30 percent yield is attained.
236
-------
Considering first the situation below the luminous zone at the point where the
temperature is 1250 K (defined as zero on the time scale), it can be seen that
the "NH3" and "HCN" fluxes are not sufficient to account for the NH3 that was
added to the flame. Therefore, significant amounts of the added NH, must have
been converted to other forms of nitrogen very early in the flame before the
temperature exceeded 1250 K. If the NO formed in the converter at this point in
the flame is from unreacted NH_, about one-third of the NH, has been converted
to other species (N-BAL curve in Fig. 70). Assuming that all of the NO formed in
the converter were from HCN only at this early point in the flame, i.e., that no
NH3 remained unreacted, gives an "HCN" flux (Fig. 71) that represents 40 percent
of the added NH3>
As discussed previously, all of the nitrogen species measured with the converter
cannot be NH, at 1.6 msec where the calculated "NH " and "HCN" flux curves reach
6 :> _j
their maxima. The sum of the calculated NH3 flux (4.3 x 10 ) and the NO flux
(0.4 x 10"7) exceeds the initial NH_ flux of 4.15 x 10"7. Also, the NH., flux
-7
cannot increase from its value at zero on the assigned time scale (~2.7 x 10 )
unless it reforms from other nitrogen species and this does not seem likely. If
the measured nitrogen species were entirely HCN at this point in the flame (1.6
msec) its flux would be 2.06 x 10 and <
as N? or other nitrogen species (N-BAL).
msec) its flux would be 2.06 x 10 and one-half of the nitrogen would be present
It can be seen from the "NH3", "HCN" and N-BAL flux curves in Fig. 70 and 71 that
at least one-third of the NH, reacts in the preflame zone to form nitrogen species
that are not oxidized to NO in the probe-converter system (i.e., "N-BAL" species).
The conversion of any or all of the remaining NH, to HCN (and some N-BAL) in the
preflame zone would also be compatible with the results obtained. At 1.6 msec,
above the top of the luminous zone, about one-half of the added nitrogen is present
as HCN or some other nitrogen species that gives a much higher response in the
converter-CA analysis technique than does NH_. The conversion of NH, to HCN in
the luminous zone appears to be a reasonable process but the reaction of signifi-
cant amounts of NH3 in the preflame region was unexpected. One possibility would
237
-------
be that one-third of the NH3 decomposes to N2 and H2 on passing through the heated
screen at the top of the burner. This does not appear likely, however, since NH_
undergoes nearly complete conversion to NO at = 0.8 after passing through the
same flame holder and HCN and NH3 undergo similar extents of conversion to NO at
= 1.5. It was established (Flames 2 and 3) that HCN definitely survives passing
throught the screen at = 1.5 and 0.8. The nature and fate of the "N-BAL" species,
that are calculated to form on the basis of nitrogen flux balances, cannot be de-
termined from the measurements that were made. One-third of the NH,-N could have
formed N_ in the preflame zone or could have produced a species that does not form
NO in the probe-converter system but forms NO and N? later in the flame.
The similarities at = 1.5 of the "HCN" flux curves above the luminous zone for
added NH3 (Fig. 71) and for added HCN (Fig. 66) strongly indicate that most of the
added NH3 is converted to HCN in or below the luminous zone. Once formed, much of
the "HCN" survives the luminous zone as it did when HCN was added initially. As
discussed in the case of Flame 3, it is not possible to determine from the data if
most of the NO forms directly from "HCN" in fuel-rich flames or via a long-lived
intermediate because the rate of NO formation does not exceed the rate of "HCN"
decay except in a region just above the luminous zone.
It appears from Fig. 71 that some of the HCN will still be present in the exhaust
gas. How much will depend on the downstream decomposition rate of the HCN and the
rate of temperature quenching. Fenimore (Ref. 66) demonstrated that under fuel-rich
conditions, at sufficiently high additive concentration, some NH,-N survives the
flame in a form that can be converted to NO in another (fuel-lean) flame. The re-
sults obtained from Flames 3 and 4 of this study indicate that the surviving fuel-N
must have been in the form of HCN rather than NH,.
:
Flame 5--HCN Addition With (j) = 0.8 and Added Argon. In all of the other probing
experiments, the diluent ratio was 1 (Ar/02 = 3.76 molar ratio). For Flame 5,
however, the DR was increased to 1.4 (Ar/0- = 5.26 molar ratio). The purpose of
the additional argon was to lower the flame zone reaction temperature. Screening
experiments had indicated that added argon affected the N0x yield from NH^ and HCN
238
-------
via its effect on the flame temperature but was not very effective in reducing
temperature. In an attempt to obtain even lower temperatures, the burner feed
rate was reduced to 5851 cc/min, a rate lower than used in any of the screening
experiments.
The pertinent data obtained from the detailed probing of Flame 5 are plotted in
Fig. 73 through 76. A comparison of the flame temperature profiles given in
Table 27 indicates that the flame temperature was lowered about 140 K by argon
dilution, and the reaction zone shifted slightly further from the burner in spite
of the fact that the burner feed rate was lowered from 7520 cc/min for Flame 2 to
5851 cc/min for Flame 5. The temperature decrease should have been about 200 K
(based on adiabatic calculations), but the movement of the flame front away from
the water-cooled burner apparently lowered the heat loss to the burner.
The final NO flux was 84 percent of the metered HCN flux and had nearly leveled
off. This is about the same as the values obtained in the screening experiments
at <{> = 0.8 but they were not conducted at this feed rate and additive concentra-
tion. About 4 percent of the HCN remained unreacted at this point. The position
of the luminous zone was not measured but, based on the CH, flux, the top must
have been at about 0.9 msec. Comparison of Fig. 75 and 62 demonstrates the sim-
ilarities in the flux profiles obtained at DR of 1.0 and 1.4. The NO and C02 form
in the same region, the HCN flux decays rapidly just above the luminous zone, and
a peak appears in the N-BAL curve just above the luminous zone.
The reaction rate curves, Fig. 76 and 63, indicate that the reaction rates reach
their maxima in the same sequence in Flames 5 and 2. Perhaps more data points
would be required to establish the exact maximum for each rate curve. As a con-
sequence of the lower temperature and somewhat diluted gas, the maximum reaction
rates for NO, HCN, CH^, and C02 for Flame 5 are all smaller than the corresponding
maximum rates in Flame -2. The reaction rate ratios are 0.70, 0.92, 0.58 and 0.71,
respectively. This is in approximately agreement with the observation that the
position of maximum temperature increased slightly even though the feed rate was
239
-------
rtti
1600-
MOO
130 y -
Ilei
3
J
*
, Msec.
TEMPERATURE
HCN,
DR =
= 0.8
l.fc
I
2
v>
N
0 "i
1 i 1
3
*
1 l
4
M
m
J> 01 ^
I '1 1
•«•
N
1
<
5 ^
1ft N.
f 7
N.
"^
1
s
D/STAMCi , MM
Figure 73. Flame Temperature vs Time, Flame 5,
HCN and Argon Addition With = 0.8
240
-------
2.5
12
2.0 -
10
-9
1.0 •
0.5
TIME , MSEC.
Figure 74. Species Mole Fraction vs Time, Flame 5,
HCN and Argon Addition With <)> = 0.8
241
-------
to
-p*.
to
3r
Figure 75. Species Flux vs Time, Flame 5, HCN and Argon Addition With = 0.8
-------
5
?*
'HCN
CH
REACTION RATE,
MOLES/CM3-SEC
HCN, <|> = 0.8
DR = }.k
v>
5
5 *
MSEC.
Figure 76. Species Reaction Rate vs Time, Flame 5,
HCN and Argon Addition With cf> = 0.8
243
-------
decreased by 22 percent in Flame 5. The effect of argon dilution on reaction
rates is analyzed in more detail later in the report. There are several factors
involved some of which are compensating.
Because of a change in the HCN flowmeter calibration, the HCN mole fraction in
the burner feed gas was 1793 ppm rather than 1936, the ppm that would have been
required to keep the HCN/0- ratio equivalent to 2500 ppm in an undiluted burner
feed stream (DR=1). The HCN flux correlation curve in Fig. 75 exceeds the metered
HCN flux in the region just before the HCN flux begins to decline. It is possible
that these high values of HCN flux are somehow related to the negative NO flux
values occurring in the same region since an experimental determination of the
HCN mole fraction depends on the measurement of two mole fractions, i.e., NO mole
fraction with and without the molybdenum catalyst. It should be noted that the
N-BAL curve is quite close to zero in this region, indicating that an error is the
measurement of NO could cause the increase in the "HCN" flux curve. A plausible
explanation, that does not involve experimental error, might be that in this region
of the flames, small amounts of NO- are formed by the reaction of HO- or OH with
NO. The presence of NO- would lead to higher apparent HCN readings on the CA be-
cause the conversion efficiency of the catalyst for N02 at 800 C is greater than
for HCN. NO- measurements were not carried out for the HCN flames but the amounts
of NO- formed in the NH, flames were much smaller than this explanation would
require.
244
-------
NO Addition With and Without HCN--Detailed Probing
Three flames were investigated using NO addition. The first two (flames 6 and 7)
were conducted with HCN and NO added simultaneously at = 0.8 and 1.5, respectively,
while the last detailed probing experiment (flame 8) involved NO addition only at
= 1.5. These experiments were carried out to determine whether there was an inter-
action between HCN and NO that would not be readily apparent unless NO were present
at a greater initial concentration.
Flame 6--HCN and NO Addition with (j) = 0.8. The data taken on flame 6 have been
plotted in Fig. 77 through 80. The HCN flux profile obtained for this flame has
some very unusual features that indicate that either (1) the added NO has a marked
effect on the rate of HCN reaction in the region of the flame holder or (2) some
unidentified experimental error is involved. When HCN alone was added at a con-
centration of 2500 ppm and = 0.8 (flame 2), the measured "HCN" concentration just
above the flame holder (Fig. 61) was about 1900 ppm dropping almost linearly to
about 900 ppm at the top of the luminous zone. The flux curve derived from that
ppm profile (Fig. 62) was nearly equal to the initial HCN flux (based on the added
HCN feed rate) at the first sampling point and then decreased gradually to abo.ut
63 percent of the initial flux at the top of the luminous zone. In flame 6, 2500
ppm HCN and 625 ppm NO were contained in the feed gas to the burner. The "HCN"
concentration at the first measurement point in this experiment was only 1235 ppm
dropping to 300 ppm at the top of the luminous zone. The "HCN" flux curve ob-
tained for this flame had an initial value of only 60 percent of added HCN flux
at the first .sampling point, and remained nearly constant until just below the
top of the luminous zone.
The lower initial measured flux in flame 6 could result from an error in metering
or measurement of the HCN. However, a calibration check of the molybdenum catalyst
for the determination of HCN was made the day after the experiment and the HCN
flowmeter had been calibrated the day before. In addition, incorrect metering of
HCN would not account for the difference in the shapes of the "HCN" flux curves.
Precluding experimental error, the initial low value for the "HCN" flux indicates
that the added NO somehow catalyzes the reaction of HCN in the region of the
245
-------
2otx>
19*
not
HtO
/50O •
TEMPERATURE
HCN + NO, <)> = 0.8
I3oo
X)
1
I I I
_l I
M
N
J I
Figure 77. Flame Temperature vs Time, Flame 6,
HCN and NO Addition With $ = 0.8
246
-------
K)
Figure 78. Species Mole Fraction vs Time, Flame 6, HCN and NO Addition With = 0.8
-------
K)
-P>
OO
Figure 79. Species Flux vs Time, Flame 6,
HCN and NO Addition With = 0.8
-------
6-
J
IV
I
-2
CH
REACTION RATE,
MOLES/CM3-SEC
HCN + NO, $ = 0.8J
•16
o.s
T/ME , MSEC.
1.0
<0
Q
,
-5
I.S
Figure 80. Species Reaction Rate vs Time, Flame 6,
HCN and NO Addition With = 0.8
249
-------
flandholder. If this turns out to be the case, the nitrogen species formed must
be later converted to NO in high yield (above the luminous zone) because it will
be shown that the NO yield curves for flames 2 and 6 are nearly identical and
the HCN consumed initially represents 32 percent o'f the added nitrogen compounds.
The NO flux begins to decrease below the center of the luminous zone while the
"HCN" flux does not start to decrease from its initially measured value until
well above this point (Fig. 79). These flux profiles as well as the rate pro-
files (Fig. 80) suggest that the removal of the added NO by chemical reaction
precedes the luminous-zone depletion reaction of HCN. When the depletion of
"HCN" begins, the destruction of NO has ceased. Because N0? measurements were
not made for this flame, the point at which HCN reaction began cannot be estab-
lished for certain. The constant portion of the "HCN" flux curve could result
from the depletion of HCN being compensated for by the conversion of NO to N0».
Further data are required to resolve this point and to determine if added NO does
catalyze the early reaction of HCN.
The point at which NO depletion reaches its maximum rate in flame 6 (about 0.3
msec, Fig. 80) is the same region of the luminous zone in which NO reached its
maximum rate of consumption in Flame 1 (0.4 msec, Fig. 59). Therefore, at = 0.8,
any NO that forms below or in the luminous zone, or is added initially, can react
near the center of the luminous zone. Because the added NO that is lost before
the NO flux begins to increase represents only 7 percent of the added nitrogen
species (HCN + NO), the consumed NO could form either N or a reactive nitrogen
intermediate and be compatible with the NO yields obtained.
Farther into the flame reaction zone, where the "HCN" flux rapidly declines, the
N-BAL flux becomes quite large before NO formation begins. As in the other flames
studied at = 0.8, NO and CO production occur in the same region of the flame.
-7
Based on the total metered additive flux of 6.3 x 10 , the NO flux in Fig. 79
gives the following percent yields:
t(msec) 0.69 0.76 0.92 1.07 1.22 1.37
%NO 28.2 35.7 47.1 53.8 57.0 59.0
250
-------
Comparison of these yields with the flame 2 yields plotted in Fig. 2 demonstrates
that the yields obtained at 1.37 msec (2.07 msec on the shifted time scale of
Fig. 2) were identical in the two experiments although the NO formed earlier in
Flame 6 and the yield appeared to be leveling off more rapidly. Assumtpion.,that
the HCN was inadvertently added initially at a flux of only 3.0 x 10 gives a
calculated yield at 1.37 msec that is 86.2 percent and still increasing rapidly.
This is a much higher yield than was obtained in Flame 2 indicating the metered
HCN flux in flame 6 was actually about 5 x 10 as planned.
Flame 7--HCN and NO Addition with = 1.5. The data for flame 7 are given in
Fig. 81 through 84. From the flux curves in Fig. 83, it can be seen that destruc-
tion of the added NO occurs near the top of the luminous zone but, under these
fuel-rich conditions, NO consumption does not occur until after much of the HCN
has reacted. Just above the luminous zone, after more than one-half of the added
NO has been consumed, NO reforms reaching its metered flux and then is consumed
again. This regeneration of the consumed NO occurs in the same region where the
rate of CO formation is at its maximum. Beyond the minimum at 1.8 msec, NO
forms gradually much as in flame 3 in which only HCN was added (at = 1.5). The
NO yield at the last data point was 16.3 percent, based on the added HCN + NO,
and was increasing. This is in agreement with the NO yield obtained in flame 3
at 2.7 msec (Fig. 2).
The "HCN" flux remains nearly constant over the region of the flame in which the
added NO is consumed, reforms and then is consumed again. Therefore, the N-BAL
flux curve is nearly a mirror image of the NO curve in this region. The NO is
apparently converted first to a nitrogen species that forms NO in high-yield just
above the luminous (much as did the intermediates formed from HCN and, particularly,
NH, in the fuel-lean flames) and then is converted to nitrogen species that may
form NO but in low yield. Much of the NO could be converted to N in the second
stage of NO consumption.
In flame 7, the "HCN" flux remains constant approximating the initial flux of
added HCN, until near the center of the luminous zone where about one-third of
the "HCN" is suddenly consumed just before the consumption of the added NO begins.
251
-------
/(.DO
k
TEMPERATURE
HCN + NO, <(> = 1.5
tCVC -
__L.
1?
i<%
i
TIME"
Figure 81. Flame Temperature vs Time, Flame 7,
HCN and NO Addition With = 1.5
"J
N
7
N
252
-------
-|»
to
in
£000
1006
I6t»
aoo
oat
600
ri**e , MSEC.
Figure 82. Species Mole Fraction vs Time, Flame 7, HCN and NO Addition With = 1.5
-------
t-o
tn
--OS
> MSEC
figure 83. Species Flux vs Time, Flame 7, HCN and NO Addition With 4> = 1.5
-------
to
en
in
REACTION RATE,
MOLES/CM3-SEC
HCN + NO, ° 1.5
T/#te , MSEC.
Figure 84.
Species Reaction Rate vs Time, Flame 7,
HCN and NO Addition With
-------
The "HCN" flux then remains nearly constant until well above the luminous zone
where it decays slowly at a rate that is more than twice the rate of NO formation.
At the last measurement point, 28 percent of the "HCN" still remained unreacted.
In flame 3, the "HCN" flux also dropped suddenly near the middle of the luminous
zone and then remained constant before beginning its gradual decay. The duration
of the constant "HCN" flux plateau at the top of the luminous zone was somewhat
longer in flame 7. It will be seen that the results of flame 8, in which only NO
was added at = 1.5, demonstrate that part of the added NO is converted to "HCN"
in the region of the "HCN" flux plateau. This could account for the longer plateau
in flame 6 than in flame 3, i.e., HCN is being consumed and formed simultaneously.
Flame 8--NO Addition With = 1.5. The data obtained from the detailed probing of
flame 8 are presented in Fig. 85 through 88. The sole nitrogen additive for this
fuel-rich flame was NO, The NO flux profile displayed in Fig. 87 is very similar
to that determined for flame 7 except that in flame 7, the final stage of NO forma-
tion is 4 times faster and, of course, yields greater amounts of NO since the nitro-
gen additive concentration was much higher. The similarities include the time, or
distance from the burner, at which (1) NO destruction begins, (2) NO reforms and
(3) NO is again consumed. In this flame, however, the NO consumed near the top of
the luminous zone is not completely reformed. The NO flux at the last measurement
point was 0.47 x 10 and appeared to be increasing slightly. This represents 45
percent of the metered NO flux. This higher NO "yield" might indicate that under
these conditions NO behaves differently than other nitrogen additives. However,
the higher "yield" of NO from NO probably resulted from the lower additive con-
centration employed in this experiment.
As the NO disappeared, the formation of small amounts of nitrogen compound
was detected by the use of the catalytic converter;and the CA. The data on this
measured nitrogen compound have been reduced using the assumption that it is HCN.
A nitrogen flux balance revealed that 40 percent of the reacted NO was converted
to "HCN" while the rest was converted to N-BAL (probably N ). It is not likely
that the "HCN" formed was N02> The "HCN" mole fraction was over 100 ppm at the
top of the luminous zone and less than 10 ppm NO was formed in flame 4 even
though the NO mole fractions were similar in this region of each flame. The
256
-------
200) •
1100
!2J?
i 5
i i i
7v*f£"j Msec.
\ \ \
i ]L T*
k
•L.
2.
DISTANCE, /MA/
Figure 85, Flame Temperature vs Time, Flame 8
NO Addition With = 1,5
/o
257
-------
K>
cn
00
2 4
120
I
80
1
zo
, Msec.
Figure 86. Species Mole Fraction vs Time, Flame 8, NO Addition With = 1.5
-------
to
in
~-o. s
Figure 87. Species Flux vs Time, Flame 8, NO Addition With = 1.5
-------
K)
O\
O
REACTION RATE,
MOLES/CM3-SEC
-6£ -
Figure 88. Species Reaction Rate vs Time, Flame 8,
NO Addition With <]> = 1.5
-------
small amount of NO formed just above the luminous zone does not cause a decrease
in the "HCN" flux curve indicating that this NO is not formed from HCN but rather
from an N-BAL species, or if the NO is formed from HCN, an equal amount of HCN is
formed from an N-BAL species.
After a further decline in NO flux, accompanied as before by rises in the "HCN"
and N-BAL flux curves, the accumulated HCN produced from the NO additive gradually
reacts reforming NO. Beyond 2.2 msec, the rate of NO formation is about equal to
the rate of HCN consumption (giving a nearly constant N-BAL flux curve) indicating
a 100 percent yield of NO from "HCN" in this region of the flame. In this region
of flames 3 and 7, the yield of NO from the reacting "HCN" was only on the order
of 50 percent. This demonstrates that at lower concentrations of "HCN" and NO,
the yield of NO from "HCN" in the post-flame gases is greater (in flames 3 and 7,
the NO and "HCN" were present at 450 to 650 ppm at 2.8 msec while in flame 8
they were each only at 240 ppm). However, even when the yield of NO from HCN
is near 100 percent, it is not possible to establish that the NO forms from the
HCN directly. The HCN could form an N-BAL species, (e.g., NCO) and the NO could
be forming from the N-BAL species still giving a constant N-BAL flux in this
region.
Comparison of Flames 3, 7 and 8. The NO flux profiles for the fuel-rich flames 3
(HCN additive) and 8 (NO only added) are compared in Fig. 89. In both flames, the
rapid consumption of NO occurs somewhat below the top of the luminous zone (at
about 1.13 msec on the assigned time scale). The depletion of NO is followed by
a region of rapid NO formation in both flames but this occurs later in flame 8
(above the luminous zone). At about 1.65 msec, NO is consumed rapidly in flame 8
while NO forms slowly in flame 3 at this point. It is possible, therefore, that
NO forms faster from flame 8 in this region of the flame than the observed rate
but part of the NO is consumed as it forms giving a lower net rate of NO formation.
The NO flux of flame 7, in which HCN and NO were both added (at $ = 1.5), is com-
pared in Fig. 90 with the sum of the NO fluxes measured in flames 3 and 8 to test
additivity. The flame 8 NO flux profile was decreased by 6.6 percent before adding
to the flame 3 profile (to make the curves in Fig. 90 coincide below the luminous
261
-------
to
&
to
10
3
8
oo 7
o
O
.-2
5
A
3
2
I
0
LUMINOUS ZONE
I
FLAME 8 (NO ONLY ADDED)
FLAME 3 (HCN ONLY ADDED)
TIME, MSEC
Figure 89. Comparison of NO Flux Profiles for the Fuel-Rich Flames 3 and 8
-------
K)
11
10
9
3
oo
o
X
X
3
2
1
0
|»-
LUMINOUS ZONE
,i;
FLAME 7 (HCN + NO ADDED)
8 (NO) + FLAME 3 (HCN)
TIME, MSEC
Figure 90. Comparison of Flame 7 NO Flux Profile With Sum of NO Fluxes From Flames 3 and 8
-------
zone) because the added NO flux was slightly higher in flame 8 than in flame 7.
The formation of NO around the center of the luminous zone in flame 3 is not
reflected in flame 7. One possibility is that NO is in quasi-equilibrium with
the other species in flame 7 and forms in this region of the flame in flame 3
because it is below this "equilibrium" concentration.
Above the luminous zone, both curves in Fig. 90 increase, decrease and then in-
crease again. The larger inflection in the solid curve indicates that the pre-
sence of "HCN" and its reaction products (N-BAL species) has an effect on the
amounts of NO that are consumed and formed above the luminous zone. As discussed
previously, the shapes of the calculated flux profiles are quite sensitive to the
slopes measured at various points on the mole fraction curves (because these de-
termine the calculated species diffusion velocities, Eq. 130). Therefore, the
differences between the Fig. 90 curves above the luminous zone might not be as
great as the measured results indicate.
Comparison of NO Addition Experiments With Those of DeSoete. DeSoete (Ref. 65)
added NO to premixed ethylene-0 -Ar flames at atmospheric pressure and measured
the NO concentration profile in the flame. He reported NO "yields" from added NO
of the same magnitude as from amines and cyanides. Under fuel-lean conditions
( = 0.7), the NO decreased rapidly with distance from the burner from its initial
concentration of 215 ppm to 160 ppm and then increased to 210 ppm and remained
constant (98 percent yield). This occurred at temperatures between 2300 and 2400
K in his experiments. The dip may be the same as was observed at 0.3 msec in the
flame 6 flux curve (Fig. 79)*.
Under fuel-rich conditions ( = 1.47), DeSoete reports the added NO decreased
from its initial concentration of 200 ppm to 50- ppm and then increased to its
final concentration of about 75 ppm (38 percent yield). In another experiment at
4> = 1.5, shown in Fig. 91, the NO decreased from 265 ppm to 59 ppm and then
*Because of diffusion in the low-pressure flames studied here, the mole fraction
of NO did not show a dip in flame 6 (Fig. 78) but the calculated flux curve did.
De Soete did not make diffusion corrections to his atmospheric flame results.
264
-------
NO, HCN
(PPM)
(PERCENT)
TIME, MSEC
Figure 91. Species Mole Fractions in C2H4 -02-Ar Flame at
Atmospheric Pressure, NO Added Initially at
265 PPM, = 1.5 (Ref. 65)
265
-------
increased slightly to a final value of 67 ppm (25 percent yield). HCN was measured
in this experiment and it reached a maximum concentration of about 188 ppm near the
point in the flame where the NO was at a minimum and then decayed rapidly. This
point was well beyond the point in the flame where most of the C-H had reacted.
The HCN at its maximum mole fraction represented 96 percent of the NO that had
been consumed. These concentration profiles obtained by De Soete are almost
identical to those from flame 8 (Fig. 86) except that his HCN yield was higher
and his flames were not as spread out. The higher yields of HCN from NO in
De Soete's experiments may result from the higher flame temperatures (Ar/0 molar
ratios of only 2.6 to 3.3) or from the use of ethylene as the fuel.
266
-------
ANALYSIS OF RESULTS OF FLAME-PROBING EXPERIMENTS
The results of the detailed flame-probing experiments will be analyzed and re-
viewed briefly before considering the possible chemical mechanisms involved in
the formation of NO from NH_ and HCN in these flames.
Fuel-Lean Flames
The flame 1 results definitely establish that with NH3 addition in the presence
of excess oxygen (1) NH, is converted to NO almost exclusively via a relatively
long-lived nitrogen intermediate and (2) most of the NO forms rapidly and in high
yield from the intermediate just above the luminous zone (Fig. 58). The fluxes of
all of the measurable nitrogen species (which include NO, N02, HCN and cyanogen)
are near zero at the top of the luminous zone indicating that in fuel-lean premixed
flames all of the NH_ is converted to an intermediate species before the produc-
tion of most of the NO begins? The flux of the nitrogen species that are not
measurable by the probe-converter-CA technique employed was calculated by sub-
tracting the fluxes of the measured nitrogen species from the initial additive
flux (i.e., by nitrogen mass-balancing). In this report, therefore, these are
referred to as "N-BAL" species. The N-BAL species include any N- that has formed
plus any radicals or other nitrogen intermediates present in the sample entering
the probe that form a species in the probe (probably N~) that will not convert
to NO over the molybdenum catalyst at 800 C (or is converted to a smaller extent
than the species NH3 or HCN).
The results obtained in flame 2 on the formation of NO from HCN under fuel-lean
conditions also establish that most of the NO is formed via a nitrogen inter-
mediate. This conclusion cannot be reached as directly as in the case of NH_
additipn because the much slower rate of reaction of HCN in the flame prevents
more than about one-half of the HCN-N from being present as N-BAL species at any
point in the flame (Fig. 62). The final flux of N2 in flame 2 was 1.0 x 10"7
(if an 80 percent yield of NO were obtained). Therefore, the maximum value of
the N-BAL flux, obtained at about 0.8 msec, establishes that at that point in
*This conclusion is based on a rather large correction for the upstream diffusion
of NO but is within the expected error for this type of correction.
267
-------
the post-flame gases at least one-third of the added HCN-N is present as an
(immeasurable) intermediate that later forms NO (it will be assumed throughout
this analysis of the results that N_, once formed, will not form NO in these
moderate temperature flames). At this point in the flame, which is slightly
above the point of maximum "HCN" consumption, the rate of formation of NO is at
its maximum. These and other arguments presented in the previous discussion of
flame 2 establish that much, and very possibly most, of the NO formed from HCN
in the fuel-rich flame forms from a nitrogen intermediate.
The candidates for the nitrogen intermediate(s) that is the precursor to NO for-
mation in fuel-lean flames are probably limited to CN, NH , N and NCO. The inter-
Jt
mediate, or whatever product it forms when quenched in the sampled gas, has been
shown not to be converted to NO by the molybdenum catalyst at 800 C. The very
reactive nitrogen atom is expected to be too short-lived to achieve the concen-
trations attained by the N-BAL species in flames 1 and 2 (the maximum N-BAL mole
fraction in flame 1 is about 2000 ppm* of which a maximum of 20 percent can be
N because 80 percent later forms NO). CN and NH intermediates would probably
£ A.
form HCN, cyanogen and NH_ in the probe and, therefore, be converted to NO over
the catalyst and measured as "HCN" or "NH," (however, NH radicals could conceivably
O
react in the probe to form N» + 2H). The most likely candidate for the precursor
to NO formation in fuel-lean flames is the NCO radical. The probe reaction for
this species might be 2 NCO = N2 + 2 CO.
Fuel-Rich Flames
When HCN was added to the fuel-rich flame (flame 3, Fig. 66), about two-thirds
of the HCN apparently survived the luminous zone and then reacted very slowly
in the post-flame gases. The NO formed slowly and! in low yield with most of
the NO forming far above the luminous zone.
"From Eq. 129, C = Flux/(V. + v). V. of N-BAL species is zero at maximum mole
fraction of N-BAL which will be somewhat upstream from the point of maximum
flux. Therefore at this point (about 0.3 msec), flux = 4 x 10-7 and v = 300
cm/sec giving C = 1.32 x 10-9 moles/cm3. At 0.1 atm and 1816 K, -this concen-
tration is 1970 ppm.
268
-------
The results also strongly indicate that when NH is added to the fuel-rich flame
O
(flame 4) most of the NH reacts in and/or below the luminous zone to form HCN,
N-BAL species and a small amount of NO. The measurement method employed could
not distinguish directly between NH, and HCN but comparison of the flux curves
O
for the fuel-rich flames 3 and 4 (Fig. 66 and 71), over the time range from 1.5
to 3 msec shows a remarkable agreement even though the additives were HCN and NH3,
respectively. The "HCN," NO and N-BAL curves are very similar in this region.
The "HCN" flux is somewhat higher at 1.5 msec in flame 3 than in flame 4 where it
represents only 50 percent of the added NH,-N. The "HCN" rate curves do not appear
O
as similar as the flux curves but this results from minor differences in the shapes
of the "HCN" flux curves in this region. The "HCN" rates for flames 3 and 4 are
actually quite comparable. Since De Soete also observed the conversion of NH_ to
HCN in fuel rich flames, it will be assumed in the following discussion that the
"HCN" flux plotted in Fig. 71 for the NH_ addition experiment is actually entirely
HCN above the top of the luminous zone. Experiments are planned that will permit
the conversion of NH, to HCN to be followed through the flame front in the appara-
tus used in this study.
In both flames 3 and 4, more than one-half of the NO is formed in the region of the
flame between 1.5 and 3.0 msec (slowly and in low yield). It is not possible to
determine for certain, however, if NO forms in these fuel-rich flames directly
from the reacting HCN or, as seems more likely, from a reactive intermediate and,
if an intermediate is involved, its relative lifetime. The amounts of NO formed
between 1.5 and 3 msec are sufficiently small that, in both flames, the HCN reacted
in this flame zone and the amount of N-BAL species present at 1.5 msec are each
sufficient to supply more than twice the nitrogen required for the NO that is
formed.
Maximum NO Formation Rates
The conditions that existed at the points of maximum NO formation rate in the
post-flame gases are compared in Table 28. The maximum NO formation rates in the
fuel-rich flames were lower by factors of 20 to 50 than the rates at 0 = 0.8.
269
-------
TABLE 28. CONDITIONS AT POINT OF MAXIMUM
NO FORMATION RATE IN POST-FLAME GASES
Equivalence Ratio
Add i t i ve
Time, msec
Temperature, K
NO Rate, moles/cm -sec
0 Concentration, moles/cm
NO Rate v (Oj
Flame 1
0.8
NH.
0.58
1956
9.5 x 10"6
4.3 x 10"8
219
Flame 2
0.8
HCN
0.82
2040
3.8 x 10"6
2.8 x 10"8
136
Flame 3
1.5
HCN
2.7
2010
2.2 x 10~7
<1 .5 x 10"9
>148
Flame 4
1.5
NH
2.5
2075
2.0 x 10"7
-Q
<1 .44 x 10 J
>139
270
-------
The maximum NO rates in the fuel-rich flames occurred beyond the last 0- measure-
ment points (at which the 02 mole fraction had decreased to about 0.3 percent).
It can be seen from the values listed in the last line in Table 28 that the rate
of NO formation is, as would be expected, strongly dependent on the 0- concentra-
tion. This does not indicate that molecular oxygen is necessarily involved in the
rate-determining step since the concentrations of other oxygen species such as 0
atom and OH will also be dependent on the 0_ concentration.
Effect of Argon Dilution
The flame 2 and flame 5 experiments were conducted under similar conditions, 0 =
0.8 and HCN addition, except that in flame 5 the diluent ratio was increased from
1.0 to 1.4 and the burner feed rate was decreased from 7520 to 5851 cc/min. The
additional argon and decreased flowrate lowered the measured maximum temperature
from 2050 K to 1890 K. It was noted in the discussion of the flame 5 results that
the measured maximum reaction rates of the species NO, HCN, CH and 0. were de-
creased by 30, 8, 42 and 29 percent, respectively, by argon dilution (and the re-
duced flowrate). Analysis of the data for flames 2 and 5, presented in Fig. 60
through 63 and 73 through 76, indicate that a number of factors determine the
effect of argon dilution on the maximum rates that are observed. It will be seen,
for example, that when a reaction rate is decreased by lowering the temperature
and species concentrations, the maximum rate will in general occur at higher
reactant concentrations partially offsetting the decrease in the maximum rate
relative to the decrease in the average reaction rate of a given species.
Considering first the points in flames 2 and 5 at which the HCN depletion rates
are at their maxima, the conditions at these points are, respectively: time (msec)
= 0.72 and 0.89; T = 2005 K and 1850 K; (HCN) = 2.37 x ID"10 and 3.77 x 10"10
moles/cm3; (0 ) = 3.65 x 10"8 and 3.62 x 10"8; and -d(HCN)/dt = 4.0 x 10"6 and
-6 3
and 3.67 x 10" moles/cm -sec. Because the HCN concentration at which the maxi-
mum rate of HCN depletion occurs is higher in the argon dilution flame, a decrease
of only 8 percent in the maximum HCN rate represents a larger decrease in the rate
271
-------
constant for the HCN consumption reaction. This can be shown as follows: If the
global rate expression for HCN consumption were RHCN = kHCN (HCN)X (02)y, then the
ratio of the rates under any two given conditions (selected here are the maximum
HCN rates in flames 5 and 2 defined as R5HCN and R2HCN) is given by:
R5HCN k5"^ T(HCN)5
R2
HCN
(132)
Since the 02 concentrations were nearly the same at these maxima, (ks /k )
= (R HCN/R HCN) ' (0.63)x = 0.92(0.63)X. If the consumption of HCN is first-
LJ^\T I I/^M
order in HCN (x=l), then kg A2 = 0,58 and the rate constant decreased by
42 percent even though the rate only decreased by 8 percent. A 42 percent de-
crease in a rate constant from 2005 K to 1850 K would require an activation energy
for HCN consumption of 26 kcal/mole. This activation energy was obtained assum-
ing that the reaction consuming HCN is first -order in HCN. If x=0.5, for example,
the calculated decrease in the rate constant is only 27 percent and the activa-
tion energy obtained is 15 kcal/mole.
Considering next the points in flames 2 and 5 at which the CH4 depletion rates
are at their maxima, the conditions are, respectively: time (msec) = 0.52 and
0.75; T = 1950 K and 1820 K; (CH4) = 3.44 x 10"9 and 3.62 x 10"9; (02) = 4.5 x
10"8 and 4.35 x 10"8; and RCH4 = 1.14 x 10"4 and 6.62 x 10"5. Because the con-
centrations of CH. and 0_ remained nearly unchanged, the decrease in CH. oxida-
tion rate of 42 percent must have resulted mainly from the decrease in tempera-
ture. This gives an activation energy for CH. oxidation of 30 kcal/mole.
At the points in flames 2 and 5 where CO oxidation is at its maximum: time (msec)
= 0.78 and 0.93; T = 2015 K and 1853 K; (CO) = 1.57 x 10"8 and 2.37 x 10"8; (OJ
o o '8 -8 CO
= 3.52 x 10"s and 3.49 x 10 ; (H?0) = 9.2 x 10~° and 6.6 x 10 ; and R = 1.22
A C.
x 10 and 8.65 x 10~ . The global rate expression for the oxidation of CO is
well established to be first-order in CO and one-half-order in H20. Since the 0^
concentration remained unchanged, the 29 percent decrease in the maximum rate of
272
-------
oxidation of CO can be calculated to represent a decrease of 45 percent in the
rate constant for CO oxidation. The accepted activation energy for CO oxidation
of 27 kcal/mole would predict a rate constant decrease of 45 percent from 2015 K
to 1853 K in exact agreement with the measured rates.
At the points of maximum NO formation rate in flames 2 and 5: time (msec) =0.82
and 1.16; T = 2020 K and 1875 K; (NO) = 7.45 x 10"10 and 7.47 x 10~10; (02) =
3.44 x 10"8 and 2.70 x 10"8; (HCN) = 1.57 x 10~10 and 1.98 x 10~10; N-BAL Flux
= 2.7 x 10"7 and 1.0 x 10~7 moles/cm2-sec; v = 336 and 239 cm/sec; and RNO = 3.40
x 10~ and 2.37 x 10" . The effect of argon dilution on the maximum rates of NO
formation cannot be analyzed directly because the rate-determining step involves
unidentified nitrogen and oxygen species whose concentrations are unknown.
Consider first the possibility that the rate of NO formation is first order in
oxygen and first order in N-BAL species. Because the N-BAL fluxes are at their
maxima in the region where the NO formation rates are at their maxima, the diffu-
sion velocity of the N-BAL species is near zero (assuming that N-BAL represents
a single species) permitting the N-BAL concentrations to be calculated. Thus
the N-BAL concentrations, obtained by dividing the fluxes by the gas velocities,
are 8.0 x 10 and 4.2 x 10 in flames 2 and 5, respectively. If it is assumed,
as a first approximation, that the rate of NO formation is proportional to the
concentrations of the N-BAL species and of 02, the rate should decrease by a factor
of 0.52 for the decrease in N-BAL concentration and a factor of 0.78 for the de-
crease in 0? concentration. Therefore the maximum rate of NO formation should have
decreased by 60 percent because of concentration factors alone, not taking into
account any effect of the lower temperature, but the maximum NO rate actually only
decreased by 30 percent between flames 2 and 5.
If the rate of NO formation were proportional to both the 02 concentration and the
HCN concentration, which is actually larger by 26 percent in flame 5 than in flame
2 at the point of maximum NO formation rate, the concentration effects cancel and
273
-------
the 30 percent decrease in rate can be attributed to the effect of temperature on
the rate constant. This would give an effective activation energy for NO forma-
tion of 19 kcal/mole. This analysis indicates, therefore, that the rate of maxi-
mum NO formation is flames 2 and 5 correlate more closely with the "HCN" concen-
tration than with the concentration of N-BAL species.
NO Addition
The fuel-lean flames 2 and 6 were conducted under similar conditions, 0 = 0.8
and HCN additive, except that some NO was also added initially in flame 6. It
can be seen from the reaction rate curves in Fig. 63 and 80 that the reaction of
CH4 reaches its maximum at about the same point in each flame. However, HCN,
CO- and NO reach their maximum rates much earlier in flame 6 than in flame 2. The
rate curves for these three species retain their same order and approximate
spacing in flame 6 but reach their maxima earlier than in flame 2. About 60 per-
cent of the added HCN has reacted by the top of the luminous zone in flame 6 while
more than 60 percent of the HCN survives the luminous zone in flame 2. It is not
possible to determine for certain if the added NO contributes to the earlier reac-
tion in this flame or if some unidentified experimental difficulty is involved.
The maximum C0_ formation rate occurs much closer to the luminous zone when NO is
added initially but the presence of NO should not affect the rate of oxidation of
CO. An experimental discrepancy is, therefore, suspected.
Some consumption of NO in the luminous zone was indicated in the fuel-lean flames
6 and 1. However, no such effect was observed in flame 2 even though the NO mole
fraction was about two-thirds of that of the other two flames in this region. In
each case, the calculated consumption of NO (decrease in the flux curve) occurs
in a region of the flame where, because of upstream diffusion of NO, the NO mole
fraction is increasing steadily. Therefore, the magnitude of NO consumption in
the luminous zone of the fuel-lean flames is not well established.* Also, the
addition of NO alone was not investigated under fuel-lean conditions.
*It should be noted that in flame 1, one of the two NO profiles that were measured
did not give a negative calculated flux or rate in the luminous zone.
274
-------
Under fuel-rrich conditions, NO was added alone (flame 8) and with HCN (flame 7).
In each case, about one-half of the added NO is suddenly consumed just below the
top of the luminous zone. The NO flux at one point in flame 8 decreases to about
one-fourth of its initial value. The consumption of NO in and above the luminous
zone is well established under fuel-rich conditions. In addition to a decrease
in the calculated flux, the measured mole fractions of NO actually decrease in both
flames (because the downstream NO concentration gradients are so small that little
NO diffuses upstream).
These results with added NO demonstrate that, in fuel-rich flames with HCN or NH,
added, more NO could form than the flux curves indicate because both NO formation
and consumption are occurring simultaneously. In flames 7 and 8, the maximum rates
of NO consumption are 10 x 10 and 7 x 10 , respectively, at NO mole fractions
of 305 and 250 ppm (t = 1.16 msec). In flame 3 (HCN addition), the maximum in
the NO consumption rate is 8 x 10 at an NO mole fraction of 100 ppm. In flame
4, however, the NO flux is increasing in this region of the flame even though the
NO mole fraction is 200 ppm. Apparently some NO is being formed in the luminous
zone as part of the added NH, is being converted to HCN. Most of this NO is being
consumed by the reactions that consumed added NO in flame 8. The net effect in
flame 4 is a positive NO formation rate (at 1.16 msec) of 2 x 10 .
Reaction Mechanism
The elementary reactions that may be involved in the formation of NO from NH,j and
HCN are discussed in this section and compared with the experimental results. More
data and theoretical analysis are required before a detailed mechanism can be
established. Computer calculations need to be carried out to determine if the
reactions that are considered can predict the reaction rates observed for each
of the measured species in various regions of the flame as well as the overall
yields of NO.
275
-------
The reactions under consideration are listed in Table 29.* The tabulation is
not all inclusive but contains most of the elementary reactions that"would appear
to be of importance in the production of NO and N2 from NH3 and HCN in hydrocarbon
flames. According to this mechanism, the major path for the formation of NO in-
volves the reaction of CN radical with 0 or OH to form NCO which is postulated
to be relatively stable and can eventually react with oxygen atom to form NO:
CN + 02 = NCO + 0 (25)
CN + OH = NCO + H (26)
NCO + 0 = NO + CO (28)
As discussed previously, the NCO radical is proposed as the principal intermediate
in the formation of NO from NH, and HCN because it could have the characteristics
required by the experimental results obtained in this study; namely, of not forming
NO in the probe-converter system and of being sufficiently non-reactive to have a
relatively long lifetime in the region of the luminous zone. It does not appear
that the NCO radical can undergo reactions other than Reaction 28, and possibly,
Reactions 29 and 30. Reaction 28 will not occur rapidly until above the luminous
zone where the oxygen atom concentration is high. Davies and Thrush (Ref. 136)
have proposed that NCO is an intermediate in the formation of NO from the reaction
of oxygen atom with HCN. More recently, Mulvihill and Phillips (Ref. 137) have
proposed NCO as the precursor to NO formation from cyanogen added to a low-tempera-
ture H2-N2-0 flame.
According to the proposed reaction scheme listed in Table 29, molecular nitrogen
can form in the following four reactions:
N + NO = N2 + 0 (16)
NH + NO = N2 + OH • ' (13)
NCO + NCO = N2 + 2 CO (29)
NCO -•- N = N2 + CO (30)
*Reactions 1 through 30, as listed in Table 29, are numbered here out of sequence
with those in the rest of the report for simplification of the discussion.
276
-------
TABLE 29. REACTION SCHEME FOR FUEL NO FORMATION
FROM AMMONIA AND HYDROGEN CYANIDE
AH2gg, KCAL/mole
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27-
28.
29.
30.
NH. + 0 + NH2 + OH
NH- + OH •»• NH2 + H20
NH + H •*• NH2 + H2
NH2 + 0 + NH + OH
NH- + OH + NH + H-0
2 2
NH2 + H •*• NH + H2
NH + 0 -> N + OH
NH + OH -»• N + H20
NH + H -»• N + H2
NH + 0 -»• NO + H
NH + OH ->• NO + H2
NH + 02 -»• NO + OH
NH + NO -»• N2 + OH
N + 02 -»• NO + 0
N + OH -»• NO + H
N + NO + N2 + 0
N + CH ->- HCN + 2 H
N + CH + CN + 2H
N + CHO -> NCO + H
N + RH -> NH + R
HCN + 0 -»• CN + OH
HCN + 0 + NCO + H
HCN + OH ->• CN + HO
HCN + H -> CN + H2
CN + 0 -*• NCO + 0
CN + OH -»• NCO + H
CN + 0 •*• CO + N
NCO + 0 -»• NO + CO
NCO + NCO -*• N + 2 CO
NCO + N = N0 + CO
+ 1.0
-16.2
-1.1
0.0
-17.2
-2.0
-27-3
-44.5
-29.3
-76.0
-78.1
-59.1
' -102.2
-31.8
-48.8
-75.0
-11.3
+2.9
-33.2
+21.6
-1.7
+4.4
+ 19.6
-6.3
-23.3
-77.0
-102.5
-129.0
177.5
277
-------
Reaction 16 is part of the reverse Zeldovich mechanism. It will be seen that the
amount of N2 that can form via Reaction 16 is limited by the competing Reaction
14 from the forward Zeldovich mechanism:
N + 02 = NO + 0 (14)
The experimental procedure did not permit the molecular nitrogen flux profile to
be obtained and N2 had to be included in the N-BAL species. The final measured
yields of NO establish, however, that N is formed in low yield in the fuel-lean
flames and in high yield in the fuel-rich flames.
A probable reaction mechanism for NH, would first involve the progressive stripping
of H atoms by reaction of NH3> NH2> and NH with H, OH, and 0 to produce N and NH
(Reactions 1 through 9). Reaction of N atom with hydrocarbon fragments such as
CH3 and CH2 via Reactions 17 and 18 is known to be fast (Ref. 138). About one-
half of the NH, forms HCN in fuel-rich flames (flame 4) while all of the NH, is
•* 6
apparently converted to NCO in fuel-lean flames (flame 1). This may be partially
due to the lower concentrations of hydrocarbon fragments in the fuel-lean flames.
Also, the presence of intermediate oxidation products in the lean flames such as
CH-O and CHO could lead to the direct formation of NCO by, for example, Reaction
19. Any CN radicals formed in reactions such as Reaction 18 could form NCO in
fuel-lean flames via Reaction 25 but would form HCN in fuel-rich flames by the
reverse of Reactions 21 and 24.
NO Formation From NH3 in Fuel-Lean Flames. According to the mechanism being con-
sidered, NHj reacts at 0 = 0.8 (flame 1) to form N and NH which react with hydro-
carbon fragments and partially oxidized species to form CN and possibly some NCO
directly. Nearly all of the CN formed then reacts with Q>2 to f°rm NCO. After the
methane reaction rate declines at the top of the luminous zone, the 0 atom concen-
tration rises and the NCO is converted to NO rapidly and in high yield via Reac-
tion 28.
NO Formation From HCN in Fuel-Lean Flames. About one-third of the HCN reacts in
the luminous zone under these conditions (flame 2) to form CN and then NCO (Reac-
tions 21 through 25). However, most of the HCN does not react until just above
the luminous zone where it reacts rapidly, probably through Reactions 22 and 23,
278
-------
forming CN and then NCO which forms NO in high yield (Reaction 28). All of the
reactions that involve the initial breakdown of the HCN molecule are either endo-
thermic or sterically hindered. The results indicate that the reactions of the
HCN molecule are much slower than those of NHj and, according to the proposed
mechanism, the HCN reacts with 0 atom above the luminous zone more slowly than
does NCO giving a somewhat slower rate of NO formation from HCN (Fig. 2). How-
ever, once the HCN molecule reacts, Reactions 25 and 28 predominate in the fuel-
lean flame giving high yields of NO.
Formation of N2 in Fuel-Lean Flames. In the fuel-lean flames (1 and 2), about
18 percent of the additive nitrogen ultimately forms N2 (based on an 82 percent
yield of NO) but the region in which most of the N- forms cannot be established
from the experimental data. Two sources of molecular N- are included in the pro-
posed mechanism. Reactions 13 and 16 form N2 from NO by reactions of the type
proposed by Fenimore (Ref. 66). In Reactions 29 and 30, N- forms from NCO in
competition to Reaction 28.
Consider first the situation if it is assumed that N forms solely from Reaction
16 in these fuel-lean flames where the 0 concentration is much greater than the
NO concentration even at complete reaction. The rate constant ratio kigAi^ is
not large and has the following values:
Temperature, K 1600 1800 2000 2200 2400
k^/k., 11.2 8.0 6.0 4.8 3.9
lo 14
The molar ratio (0 )/(NO) is about 50 just above the luminous zone dropping to 20
at complete reaction. Therefore the reaction rate ratio R.. ./R ,. is 7 above the
luminous zone and 3 at complete reaction. Thus if the 18 percent N2 formed is
from Reaction 16 only, nearly all of the NO would have to form via Reaction 14
rather than Reaction 28 as proposed.
If such a Fenimore mechanism were occurring in the fuel-lean flames instead of
the proposed "NCO" mechanism, the required maximum mole fraction of N atom can be
calculated as follows. The N-atom concentration required for Reaction 14 to
occur at a given rate, R^ , is (N)=R -=-[k (0 )] . It can be seen from Table 28
279
-------
that at the maximum rate of NO formation in flame 1, RMn/(0~)=219 and T=1956 K.
12 11
The value of k . at this temperature is 2.5 x 10 giving (N)=8.8 x 10 mole/
3
cm or 141 ppm N atom at 0.1 atm. This is a much larger steady-state N-atom con-
centration than can form from N2 in hydrocarbon-air flames (via the Zeldovich
mechanism). It is possible, however, that the major N-BAL species formed from
NH_ and HCN in fuel-lean flames is actually CN radical rather'than NCO and Reac-
tion 27 occurs so rapidly just above the luminous zone that the N atom concen-
tration reaches the high values required for the observed NO and N- to form via
reactions 14 and 16, respectively. Computer modeling will be required to deter-
mine if this mechanism for NO formation is reasonable. Assuming that large con-
centrations of CN radicals accumulate in the flame front rather than CNO radicals
does not appear to be in agreement with the flame 1 results, however, because CN
would be expected to form HCN or (CN)? in the probe and, therefore, be measurable
as "HCN" in the converter-CA system.
Consider next the favored mechanism in which all of the NO forms from NCO through
Reaction 28. The N- now cannot form from Reaction 16* but must form from NCO.
Reaction 29 then appears to be the most likely path for N2 formation from NCO
(see page 18 of Ref. 2). This mechanism predicts a lower NO yield at higher
additive concentrations because R00/R00 increases with increasing NCO concentra-
^y /.o
tion. At the point in Flame 1 where NO is forming at its maximum rate (0.58 msec),
R = 9.5 x 10"6 and R^2 is presumably about 2 x 10" . If the NCO mole fraction
is estimated to be 1000 ppm** at that point and 0 atom is assumed to be in equilib-
-8
rium with 07 which is present at a concentration of 4.3 x 10 (Table 28), then
(NCO) = 6 x 10"10 and (0) = 2.6 x 10"10 moles/cc***. Therefore, to be compatible
13
with this reaction scheme under these conditions, k00 would have to equal 6 x 10
12 3
and k_g would be 5.5 x 10 cm /mole-sec. An oxygen-atom overshoot of a factor of
6 at this point just above the luminous zone would reduce the required value of
*Similar arguments can probably be made against the other Fenimore-type reaction,
Reaction 13.
**The N-BAL flux at 0.58 msec in Fig. 58 is about 90 percent of NH3 which was
2500 ppm. However, the N-BAL mole fraction is about one-half of the initial
NHs mole fraction because of diffusion (compare with methane in Fig. 55 which
diffuses somewhat faster).
***Log Kp = -6.6 for 02 = 0 + 0. KC = Kp/RT = 1.5 x 10~12 moles/cc. (0) =
Kcl/2 (02)l/2 = 2.59 x lO'lO.
280
-------
k_Q to 1 x 10 making the proposed NCO mechanism quite reasonable (the values of
13 12
k,6 and k,, at this temperature are, for example, 1.6 x 10 and 2.6 x 10 ,
respectively).
Reactions in Fuel-Rich Flames. Under fuel-rich conditions (flames 3 and 4) most
of the NO forms slowly and in low yield far above the luminous zone although some
NO does form in the luminous zone and a surge of NO formation occurs just above
the luminous zone. When the additive is HCN (flame 3) about one-third of the HCN
reacts in the flame front forming mainly N-BAL species which could be either N2 or
NCO. With NH addition at 0 = 1.5 (flame 4), about one-half of the NHj appears
to form HCN in the flame front by the mechanism that has been discussed and the re-
mainder forms N-BAL species.
With either additive the HCN reacts very slowly above the luminous zone. Because
the final overall yield of NO is about 30 percent, the rate of NO formation in
the post-flame gases is about one-half of the rate of HCN consumption. The slower
rate of "HCN" consumption in fuel-rich flames compared with the fuel-lean condi-
tions could result from several factors: (1) the slower rate of Reaction 22, (2)
CN formed in Reaction 23 may be converted back to HCN by the reverse of Reaction
24, and (3) the reduced rate of Reaction 25 could permit the CN concentration to
build up (the CN radical is expected to be measured as "HCN" in the probe-converter-
CA system). The major path for NO formation in fuel-rich flames probably involves
Reactions 23, 26, and 28. The small amounts of NO formed in the luminous zone
could form from Reaction 14 particularly with NH» addition.
According to the proposed mechanism, the temporary increase in NO formation rate
just above the luminous zone could come about as follows. The NCO reaches a steady-
state concentration at the top of the luminous zone that is controlled by the
relative rates of Reactions 26 and 28. The 0-atbm concentration increases rapidly
just above the luminous zone as the methane nears complete reaction and the NCO
"reservoir" is consumed by Reaction 28 producing a momentary surge in NO produc-
tion. The rate of NO formation then falls off until NCO achieves a new lower
steady-state concentration.
281
-------
The increase in N2 yield over NO yield in fuel-rich flames results from an increase
in the average ratio R29/R2g. This ratio is proportional to (NCO)/(0) at a fixed
temperature. Since the post-flame oxygen-atom concentration should decrease by
two orders-of-magnitude (not including a reduction in overshoot) in going from
0=0.8 to 0=1.5 (see Fig. 8), the increased yield of N2 appears reasonable
in terms of the favored mechanism. Computer modeling will be required to test
this conclusion.
Flames With NO Addition. NO was added along with HCN in flames 6 (0 = 0.8) and
7 (0 = 1.5), and NO alone was added in flame 8 (0 = 1.5). The results of these
NO addition experiments were discussed in the Results and Discussion section and
were compared there with the NO addition experiments of De Soete. These results
were also analyzed further earlier in this section of the report. NO appears to
be consumed in the luminous zone under both fuel-rich and fuel-lean conditions
but NO consumption under the latter condition is not well established because the
effect is small and.NO was not added alone to a fuel-lean flame. The mechanism
of NO consumption in fuel-lean flames will be considered.
The only reactions that are listed in Table 29 that could consume NO are the
reverse of the NO formation reactions (i.e., Reactions 10R, 11R, 12R, 14R, 15R
and 28R) but all of these are too endothermic to be likely sinks for NO. Taking
Q
the reverse Zeldovich reactions, for example, k..D = 1.8 x 10 and k1Cn = 2.7 x
q _ 14K I5K
10 cm /mole-sec at 2000 K*. The maximum rate of NO consumption in flame 8 (at
1.16 msec) is 7 x 10~ moles/cm -sec at an NO concentration of 1.6 x 10" moles/
cm (250 ppm) and a temperature of 1880 K. The 0-atom concentration required
for this consumption rate of NO to be given by Reaction 14R is (0) = RNf) T
:14R
[k . (NO)] = 2.4 x 10" moles/cm or 3.7 atm**. The required H atom concentra-
tion for Reaction 15R is 1.6 x 10 or 0.25 atirt. These calculations demonstrate
conclusively that the reverse Zeldovich reactions cannot be important processes
*k14 = 6.4 x IQyT exp(-6300/RT) giving 2.65 x 1012 at 2000 K and the equilibrium
constant is 6.8 x 10"5. k^5 is about 4 x 1013 with an equilibrium constant at
2000 K of 6.5 x 10-5.
**The rate constant at 2000 K was used in these qualitative calculations. There-
fore, the required radical concentrations would be even higher at 1880 K.
282
-------
in the consumption of NO in fuel-rich flames. Similar conclusions can no doubt
be drawn for the other endothermic reverse reactions. The proposed reaction scheme
is, therefore, deficient in that it does not predict the consumption of added NO.
Other reactions will have to be added to account for the hydrocarbon-initiated
consumption of NO. These will probably involve the reaction of NO with hydro-
carbon fragments. Note that De Soete did not observe consumption of NO added
to a H2-02 flame (Ref. 69).
CONCLUSIONS FROM COMBUSTION EXPERIMENTS
Screening Experiments
1. The NO yields from NH, and HCN added to 0.1 atm premixed CH.-O^Ar flames
exhibit a similar dependence on equivalence ratio from 0 = 0.5 to 1.5.
The NO yield is high from 0 = 0.5 to 0.8 (about 80 percent at the rather
large additive concentrations employed) and declines steadily thereafter
as 0 increases.
2. The flame temperature is a function of the flowrate (burner feed rate)
because of heat losses to the water-cooled burner.
3. Increasing the mole fraction of argon in the reactants can either in-
crease or decrease the flame temperature (measured in the post-flame
gases) depending on whether or not the flowrate is increased simultane-
ously. There is a general correlation between the effect of argon dilu-
tion on temperature and its effect on NO yield. The NO yield increases
with increasing temperature and vice versa.
4. Increasing the total pressure from 0.1 to 0.4 atm causes the yield of
NO to decrease somewhat at low flowrates and to remain relatively un-
changed at high flowrates. There again appears to be a general correla-
tion between the effect of pressure on temperature and the effect on NO
yield.
5. Changing the species concentrations by argon dilution or by varying the
total pressure appears to have little effect on NO yield (except for the
attendant temperature effect). This suggests that NO and N- are formed
by reactions of the same order -- undoubtedly second.
283
-------
Detailed Probing Experiments
1. NH3 reacts rapidly in or below the luminous zone whereas HCN reacts
much more slowly with more than one-half of the HCN surviving the
luminous zone in either a rich or lean flame.
2. NH_ forms an unmeasurable nitrogen species (probably NCO radical) in
high yield in the fuel-lean flame. At 0 = 1.5, about one-half of the
NH_ apparently forms HCN.
3. In the lean flame, the surviving HCN reacts rapidly just above the
luminous zone. HCN (added or formed) reacts slowly in the rich flame
with much HCN remaining unreacted far above the luminous zone.
4. Nearly all of the NO that exits the reactor forms above the top of the
luminous zone. In the fuel-lean flame, NO forms in high yield just
above the luminous zone in the region of the flame where CO is oxidized
to C02. The rate of NO formation is somewhat slower from HCN than
from NH_. In the rich-flame, the NO forms slowly and in low yield with
much of it being formed far above the flame front as the remaining
HCN continues to react slowly.
5. Reduction of flame temperature and species concentrations by argon
dilution does not have a large effect on the NO yield but does reduce
the species reaction rates as would be expected.
6. Added NO is rapidly consumed in the luminous zone of the rich flame
(forming some HCN) and then partially reforms.
7. NO forms from NH_ or HCN in the lean flame via a reaction intermediate
that is probably NCO.
8. The similarity of the NO yields from the NH3 and HCN additives can be
explained in the rich flame by the observed conversion of the NH_ to
HCN before most of the NO forms. The similarity appears to be more
coincidental in the lean flame although a common intermediate, such as
NCO, is probably involved.
284
-------
APPENDIX A
CHEMICAL ANALYSIS TECHNIQUES
CHEMICAL ANALYSIS TECHNIQUES-ORGANIC SPECIES
The organic decomposition products formed from pyrolysis of model nitrogen-contain-
ing compounds were identified by temperature-programmed gas chromatography. The
2-foot by 1/8-inch stainless-steel column was packed with Chromasorb 103 that had
been screened, rinsed five times with Freon TF, and activated at 275 C for several
hours. Pertinent operating conditions for this gas chromatograph are given in
Table A-l.
TABLE A-l. INSTRUMENT CONDITIONS FOR ORGANIC PRODUCTS ANALYZER
Instrument: Hewlett-Packard Model 700 Gas Chromatograph
Column: 2-foot by 1/8-inch Chromasorb 103
Column Temperature: Programmed from 60 to 250 C
Heating Rate: 7-5 C/min
Detector: Flame lonization (H£ and 02)
Carrier Gas: Sample gas from pyrolysis tube: He at k3 cc/min
and an additional kO cc/min added at head of
column
Recorder: L and N Speedomax W
Integrator: Hewlett-Packard Model 33718
The procedure used for analyses of the decomposition products was as follows.
After injection of the sample into the pyrolysis apparatus, the pyrolysis products
and any undecomposed starting materials were entrained in the flowing helium and
passed into the gas chromatographic analysis system with the column at 60 C. One
minute after injection of the sample, temperature programming of the column from
60 to 250 C was initiated at a rate of 7.5 C/min. The effluent gases were usually
split into two streams: 30 percent of the gas was directed to the flame ioniza-
tion detector for identification of the organic products, and the remaining 70
285
-------
percent of the gas was passed through the appropriate scrubber solution for inor-
ganic product analyses or to the mass spectrometer for peak identification. In
this manner, complete analyses of the effluent gases were performed. Prior to
subsequent analyses, the column was backflushed at 250 C for about 5 minutes.
In most cases, the organic products were identified on the basis of their reten-
tion times. The retention times of the various materials pyrolyzed and their de-
composition products are given in Table A-2. Quantitative analyses for the or-
ganic species were obtained by automatic integration of the peak areas using a
Hewlett-Packard Integrator, Model 3371B.
TABLE A-2. RETENTION TIMES BY ORGANIC PRODUCTS*
Compound
Methane
Acetoni tr i le
Acryloni tr i le
Benzene
Pyr idine
Benzoni tr i le
Napthalene
Quinol ine
Biphenyl
Retention Time, minutes
0.5
5.7
7.8
10.8
12.8
19-2
23. k
2k. 5
26.6
"The retention times given were obtained under
the conditions given in Table A-l.
Gas Chromatograph Calibration
Careful calibration of the GC pyrolysis apparatus was important because under
some conditions large losses of pyridine and/or saturation of the flame inoization
detector were found to occur. Pyridine, being basic, was absorbed in various
parts of the system if certain column packing or materials of construction were
286
-------
used, A Poropak N column packing and brass and copper fittings caused significant
amounts of pyridine to be removed from the carrier gas stream. It was also found
that traces of solder flux were detrimental in this respect but they could be re-
moved by flowing warm water through the tubing for several days after assembly.
To ensure that pyridine was not being lost in a given series of experiments, spe-
cial calibrations were made. Mixtures containing equal amounts of pyridine and
benzene at various concentrations in benzonitrile were run. The pyridine peak
area would decrease with respect to the benzene peak under conditions where pyri-
dine loss was a problem. The pyridine concentration at which this occurred would
indicate if the problem was serious enough for corrective action to be taken.
Benzonitrile was chosen as the carrier because it has a longer retention time than
benzene and pyridine and, therefore, does not interfere with these earlier peaks.
Saturation of the flame ionization detector can occur if: (1) the sample size is
too large, (2) the 1^ and $2 flowrates to the detector are not in the proper
range, and (3) if the carrier gas port into the flame becomes partially plugged.
If saturation occurs, the peak area will not be proportional to the amount of the
material in the flame. Calibration mixtures containing a range of concentrations
of the compounds of interest were run periodically to ensure that flame satura-
tion was not occurring.
A standard calibration sample was prepared that was a mixture of the various model
compounds and their major reaction products. This was run before and after each
series of model compound pyrolysis experiments and the areas measured were inputed
to a computer program to calculate the organic product distribution from each py-
rolysis experiment.
CHEMICAL ANALYSIS TECHNIQUES-INORGANIC •
NITROGEN SPECIES
The inorganic species included elemental nitrogen, ammonia, hydrogen cyanide,*
and mixed oxides of nitrogen (N02 and NO), commonly referred to as NOX. All of
^Consideration also had to be given to possible interference by any (CN)2
that might form (discussed later under HCN analysis).
287
-------
these species have previously been determined at the microgram level by several
methods in a variety of matrixes, and standard analytical procedures are available
in the literature (Ref. 139 through 143). At the outset of this program, the
literature methods were critically examined to select the optimum method for each
species determination. A number of factors were considered in the selection: (1)
sensitivity: sample sizes as small as 0.2 to 1 milligram were to be pyrolyzed,
the fuel oil samples contain as little as 0.3-percent N and, therefore, some py-
rolysis samples would contain only a few micrograms of nitrogen; (2) interferences:
the method had to be capable of sampling and determining the species in a flowing
helium stream containing a wide variety of other compounds including unreacted
starting material; (3) specificity: the method had to be specific for the species
to be determined; and (4) speed of analysis.
As the program progressed, some of the selected methods proved to be satisfactory.
Others, however, were found to be deficient and proposed methods were abandoned
for alternate methods. In other cases, no adequate method was available and mod-
ification of existing procedures was required. A review of the analytical proced-
ures (proposed and used) on this program follow.
Elemental Nitrogen
Elemental nitrogen was determined by gas chromatography (GC) using a molecular
sieve 5A column and thermoconductivity detection. The instrument used was an
Aerograph 200 with an L§N Model Speedomax W recorder and a Hewlett-Packard
Integrator. The GC inlet system was modified so that the helium effluent from
either pyrolysis apparatus passed through a 1/4-inch, glass-bead-packed, stain-
less-steel U-Tube and then into the GC column for N2 analysis. The column used
was a 9-foot by 1/4-inch aluminum tube packed with 40 to 50 mesh Molecular Sieve
5A. This column will separate hydrogen, oxygen, nitrogen, methane, and carbon
monoxide. It was originally intended to use the U-tub.e to cold-trap HCN, NHj,
and NO for subsequent chemical analyses. However, its main use was to remove
A
from the helium stream condensible species, i.e., water, undecomposed model com-
pounds, and intermediate decomposition products that would eventually deactivate
the chromatographic column. When used as a removal trap, the U-tube was packed
288
-------
with 60 to 80 mesh glass beads. This minimized dead volume and prevented diffusion
of the nitrogen, thus providing a sharp chromatographic peak. Table A-3 lists
the instrument conditions for N_ determination and Table A-4 shows the absolute
sensitivity of this method and the sensitivities of the various analytical methods
(to be discussed) as percent of nitrogen present in a typical pyrolysis sample.
By increasing the bridge current of the detector, sensitivity of the N2 analysis
was better than originally estimated at the start of the program. Late in the
program, a calibration gas containing 0.102 mole percent N- in helium was used to
calibrate the GC under conditions appropriate to those of the actual fuel pyrolysis
experiments. The repeated injection of O.Sug samples of N2 gave integrated peak
areas that were reproducible within 10 percent. However, the quantity of N2 that
was measured was about 30 percent larger than the actual quantity. Since the GC
was normally calibrated with 442-yg N2 samples, this result indicates that the
calibration curve has a positive deviation from linearity for small sample sizes.
Thus, the N- values obtained from pyrolysis of coals and oils may be as much as
30 percent lower than reported in the Results and Discussion section.
Hydrogen Cyanide
Quantitative determination of hydrogen cyanide present in the pyrolysis products
was evaluated using two different methods of analysis: (1) the Orion Research
activity electrode that is reported to be specific for cyanide ion, and (2) an
ASTM spectrophotometric method that utilized barbituric acid as the color-develop-
ing reagent. With the specific electrode, inconsistent HCN analyses were obtained
for model compound and fossil fuel pyrolysis. In the case of the fuel pyrolysis
experiments, the inconsistent results were attributed to the presence of sulfide
ion from H2S in the pyrolysis gases. Some other type of interfering species was
apparently present in the model compound products but its identity could not be
established. Very reproducible results were obtained, however, with the spectro-
photometric method. The sample collection used for both types of analyses was
the same and is briefly described.
289
-------
TABLE A-3. INSTRUMENT CONDITIONS FOR N2 DETERMINATION
Instrument: Aerograph 200
Column: 9-foot by 1/4-inch Molecular Sieve 5A
to 50 mesh)
Column Temperature: 43 C
Detector: Thermoconduct i vi ty
Bridge Current: 175 ma
Detector Temperature: 45 C
Carrier Gas: Sample Gas from Pyrolysis Tube
Sample Gas From Pyrolysis Tube: He at 20 cc/min
Reference Gas: Independent He Supply at 20 cc/min
Recorder: L&N Speedomax W
TABLE A-4. SENSITIVITIES OF THE ANALYTICAL METHODS
Component
N2
HCN
HCN
NH3
NHj
NOX
Method
Gas Chromatography
Cyanide Specific
Electrode
Barbi tur ic Acid
Distil lat ion-
Ti tration
Nessler
Color imetric
Decomposition to N_
Sal tzman
Absol ute,
ug
0.5
0.7
0.2
34
1 .0
0.6
0.5
Sensitivity in Terms of
Percent of Sample-N
0.2-yl Pyridine
1.2
•0.5
0.14
80
2.4
1.4
0.7
Imgof 0.3^-N Oil
17
23
7
>100
33
20
17
290
-------
Sample Collection System. All methods investigated required that the hydrogen
cyanide be collected in an alkaline solution. The effluent helium stream was
bubbled directly into a tube containing 25 cc of NaOH solution with the specific
electrode method. The scrubber solution was later changed to sodium carbonate
solution for the spectrophotometric determinations along with some other modifica-
tions that will be discussed. Quantitative recovery of HCN by the scrubber solu-
tion was proved by direct injection of gaseous HCN, with a gas-tight syringe,
into the helium stream just before the bubbler. The HCN used for calibration
was obtained from Fumico, Inc., Amarillo, Texas.
Specific Ion Electrode Method. The Model 94-06 Cyanide Ion Activity Electrode
is one of a family of electrochemical sensors developed by Orion Research. De-
signed to be used like a conventional pH electrode, it has a solid-state sens-
ing element which is a solid mixture of inorganic silver compounds. This membrane
is an ionic conductor which allows silver ions to pass between a sample solution
and the internal reference solution, which is kept at a fixed silver ion level.
The distribution of silver between the two solutions develops a potential which
depends on the silver ion activity in the sample solution:
RT
E = Ea + 2.3 Z± log AAg+ (A-l)
where
E = measured total potential
, E& = potential due to reference electrodes and internal solution
2.3 RT/F = Nernst factor, R and F are constants, and T is the temperature
in degrees Kelvin
A. + = the silver ion activity in the sample solution
The operation of the electrode depends on the reaction of cyanide ion with spar-
ingly soluble silver salts from the membrane surface:
AgX + 2CN" •*• Ag(CN)2" + X" (A-2)
291
-------
and it can be shown that the potential developed between the two solutions varies
with the cyanide ion activity in the sample:
RT
E = Eb ~ 2.3 Tlog ACN- (A-3)
where E^ is a new constant. With frequent calibrations, measurements down to 1 x
10~6 M (0.027 yg/ml) cyanide ion were attainable. For a 25-ml scrubber solution,
the sensitivity was, therefore, 0.7 yg. At the 2.5 yg level, reproducibility was
found to be ±15 percent.
For the fuel oil and coal experiments, however, both sensitivity problems and in-
terferences were encountered when the specific ion electrode was used. For a 1-mg
fuel oil sample containing 0.4-percent nitrogen, only 4 yg of nitrogen is in the
sample. To see 5-percent conversion to HCN, a sensitivity of 0.14 yg HCN is
needed. This is below the sensitivity of the specific ion electrode method. In
addition, the electrode will malfunction if ions which form very insoluble salts
of silver are present at sufficiently high levels to form a layer of silver salt
on the membrane surface. The Orion instruction manual states that sulfide ion,
S=, must be absent. It was found that sulfide ion at approximately the same level
as cyanide ion (5 yg) caused a positive interference ranging from 25 to 50 percent.
All of our fuel oil samples contained sulfur at levels equal to or more than the
nitrogen level. Because of the problems encountered in analyses of the fuel oil
and model compounds pyrolysis products by the specific ion technique, it was nec-
essary to use the following analytical method for both types of materials.
Barbituric Acid Method, D2056-72. A spectroscopic cyanide method, ASTM D2036-72,
which uses barbituric acid to develop the color after oxidizing the cyanide with
Chloramine-T, was investigated. According to Mr. Bill Fitzgibbons of Sohio Re-
search in Cleveland, with this method calibration curves are repeatable, the re-
agent solutions are stable for up to 6 months, and a time-consuming distillation
step could be eliminated because the method is more interference-free than either
the cyanide electrode or the pyridine-pyrazalone procedure recommended in the
"Standard Methods for Water Examination," Method 207C. ASTM D2036-72 states only
292
-------
that sulfides adversely affect a recommended titration method. Good reproducibility
was obtained in preliminary calibration experiments at cyanide levels of 0.26 to
5.20 yg. However, at the 1-yg level, it was found that an equivalent amount of
sulfide ion caused a significant decrease in the color development (absorbance at
578 my decreased from 0.10 to 0.035), and a threefold concentration of sulfide com-
pletely prevented formation of color.
We learned subsequently, from personnel at Sohio Research, of a method for removal
of sulfide interference. It is a time-consuming procedure involving precipitation
with lead carbonate, filtration, and distillation from acid solution. The ASTM
method suggests precipitation with cadmium carbonate, followed by filtration and
distillation. It was obvious that additional development work was required to
provide a method that was adequate for our purposes.
Although not stated in either the Standard Methods or ASTM D2036-72, sulfide will
react with HCN (Ref . 144) . Therefore, the elimination of sulfide interference is
a matter of removal of sulfide ion before it can react with cyanide. In most
aqueous cyanide analyses, this is not possible because the sample is obtained with
the cyanide and sulfide already together. However, our experimental procedure was
the scrubbing of a carrier gas containing HCN and H2S by a Na2CO, solution. By
placing cadmium carbonate in the scrubber solution, sulfide may be removed before
reaction with cyanide can occur:
CdC03 + S= •*• CdS + C03= (A-4)
Both the Standard Methods and ASTM D2036-72 suggest filtration as the method of
removal of CdS and CdCOj. From Latimer's Oxidation Potentials, it can be shown
that:
°
02 + 2CN- + 2CNO" E= 1.36 (A-5)
This indicates that air oxidation of cyanide could be spontaneous and a filtra-
tion procedure which involves a significant mixing of air and the solution would
293
-------
be deleterious to accurate CN~ determinations. We found that a single, rapid fil-
tration of 1.0-yg cyanide in 30 ml of solution resulted in a loss of 25 percent of
the cyanide although the exact cause of this loss was not established. The use
of centrifugation to remove CdC03 and CdS precipitate avoids the oxidation. The
following table shows the recovery of 1.0-yg cyanide in the presence of varying
amounts of sulfide ion obtained when CdCO, was suspended in.the absorbing solution
J
prior to addition of cyanide or sulfide. The recovery was .90 to 100 percent.
RECOVERY OF 1.04 yg CN~ AND S= (CdCO ADDED BEFORE CN~ AND S=)
Mg S~
3.2
6.4
9.6
12.8
16.0
32.0
Percent Recovery of CN
102
97
95
88
96
9^
If the CdCO? is added 30 minutes after the cyanide and sulfide, recovery of dup-
licate 1-yg samples was only 1 and 10 percent. These results confirm the need
to remove sulfide immediately from a cyanide solution. Two other interfering
species were tested. It was found that on a mole for mole basis, thiocyanate
ion, CNS", gives the same color development as cyanide, and sulfite ion, S0j~,
interferes by preventing color formation.
The modified ASTM D2036-72 spectrophotometric procedure appears to be adequate
for the determination of 1-yg HCN in the presence of large excesses of sulfide
ion, and this technique was used to obtain the HCN. results reported. The tech-
nique is described in more detail in Appendix C. Preliminary checks on (CN)2
interference indicated that about one-half of any (CN)2 present in the sample
would be measured as HCN by this procedure.
Before going on to the next analytical procedure, we must report on the difficult
availability of the barbituric acid reagent. Under the comprehensive Drug Abuse
Prevention and Control Act of 1970, every person who manufactures, sells, or uses
294
-------
any substance controlled under this act must register with the United States De-
partment of Justice, Bureau of Narcotic and Dangerous Drugs. Barbituric acid is
controlled as a Schedule IV substance. Without the BNDD certificate, purchase of
barbituric acid is unlawful, and vendors will not supply the item without a BNDD
registration number. Although a license costs only $5.00, at least 10 weeks are
required for processing an application, including a visit to the laboratories by
U.S. Justice Department compliance investigators. In addition, because of the ad-
ditional controls, the number of vendors has become limited. Our first purchase
of 25 gm was a purified sample from Matheson, Coleman, and Bell. The order was
for 100 gm, but only 25gmwas available through the local supply house. Subse-
quently, we were informed that MCB no longer provided barbituric acid. A second
order of 100 gm required over 2 months to be filled by Eastman Organic Chemicals
and the material was practical grade. The latter material gives a significantly
higher blank (0.07 versus 0.01A) than the purified MCB material. As a result of
limited availability of the reagent, we could not test the method out as exten-
sively as desired. Since that time, the Rocketdyne purchasing department has pur-
chased Eastman reagent grade barbituric acid, Cat. No. 1090, from the Frese Divi-
sion of Scientific Chemical Co. in Los Angeles. One-hundred grams were received
within 7 weeks of ordering.
Ammonia
Ammonia determinations in the pyrolysis gases were evaluated by three different
chemical analysis methods: (1) distillation of a collected sample from alkaline
solution, followed by titration with standard acid solution, (2) direct Nessleriza-
tion of collected samples, and (3) a phenate method. The first two of these methods
suffered from some limitation and, therefore, the phenate method was used. In ad-
dition, an indirect method was developed based on the catalytic conversion of ammo-
nia to elemental nitrogen:
2 NH3 •* N2 + 3 H2 (A-6)
Both the phenate and conversion methods were used for the fuel oil, coal, and
model compound studies.
295
-------
Sample Collection. For the indirect method (conversion of NH_ to N, and H ),
tj £* £
there is no sample collection. The carrier stream is flowed continuously through
the converter and into the gas chromatograph. For the distillation method, the
sample was cold-trapped (-196 C) in a U-tube. At the end of the pyrolysis, the
tube was disconnected while cold; the NH was frozen (freezing point = 33.4 C);
and several ml of 0.1N HC1 added. The tube was then capped and allowed to warm
to room temperature. The NH,, now dissolved in the HC1 solution was transferred
O
quantitatively to the Kjeldahl distillation apparatus for subsequent determination.
In later experiments, when the Nessler and phenate colorimetric methods were used,
the system was modified so that the effluent helium stream was bubbled directly
into 25 cc of HC1 solution.
Kjeldahl Distillation and Titration. A Kjeldahl ammonia distillation apparatus
was assembled and checked out early in the program. At that time, pyrolysis of
model compounds was anticipated to be as large as 25 ul. The sensitivity of the
method, 34 yg (see Table A-4), is sufficient to detect less than 1-percent con-
version of nitrogen for a 25 yl-sample of pyridine. Ammonia was determined in the
pyrolysis products of 5-yl samples of pyridine and 2-picoline at temperatures
where complete decomposition of model compound occurred. The trap downstream of
the reactor was cooled to collect condensible products and ammonia in this trap
was determined by distillation from alkaline solution, collection in boric acid
solution, and titration with standard acid solution. Since unreacted starting
material (pyridine and picoline) interfere slightly, as well as other organic
amines, the end-point of the titration was not sharp and precision was not good.
As the program progressed, and sample sizes down to 0.2 yl were used at tempera-
tures where only partial decomposition occurred, it became apparent that this
method could not be used because of its insensitivity.
Nessler Method. The Nesslerization method is the standard colorimetric procedure
recommended in the Standard Methods and by ASTM (Ref. 139 and 140). A carefully
prepared Nessler reagent may respond, under optimum conditions, to as little as
1-yg ammonia nitrogen; however, reproducibility below 5 yg may be erratic. This
did, however, afford a tenfold increase in sensitivity over the Kjedahl method.
In several pyrolysis experiments, determination of ammonia was performed by the
296
-------
Nessler method. Several of the samples developed a greenish turbidity, preventing
the use of a photometer. Results were therefore obtained by visual comparison
with prepared standards. It is reported (Ref. 139) that a number of aliphatic
and aromatic amines, acetone, aldehydes, alcohols, and other undefined organic
compounds can cause trouble. Compounds of this type have been found to yield a
yellowish or greenish off-color or a turbidity. Sulfide has also been reported
to cause turbidity following Nesslerization. Since methods of removal are time
consuming and the interferences we encountered were not identified, the method
was not used for subsequent experiments.
Phenate Method. The phenate colorimetric method for NH, was found to be.quite
sensitive and satisfactory for this program. The method (from page 232 of Ref. 139)
is outlined in Appendix D.
Very small quantities of NHj, on the order of 1 to 2 yg, were formed during the
pyrolysis experiments. To determine the reliability of the phenate NH^ values
reported, calibrations were conducted utilizing a dilute NHj (1.68 mole percent)
in helium mixture. In duplicate experiments, 1.0 cc of this gas (9.26 yg NHj)
was introduced into a volumetric flask containing 0.1 N HC1 and the NHj deter-
mined by the phenate method. The NHj recoveries were 9.6 and 9.4 yg, indicating
that the phenate method is satisfactory for measuring small quantities of NH^.
Conversion to Nitrogen (An Indirect Method). After investigating the above anal-
ytical techniques for determining the small amount of ammonia formed during pyrol-
ysis of fossil fuel samples, a catalytic decomposition technique was developed
which, in combination with gas chromatography, was finally used to determine in-
directly the amount of ammonia formed. At temperatures of 900 C or higher, am-
monia decomposes completely to elemental nitrogen and hydrogen when passed through
a nickel tube or a tube packed with granular nickel metal (Table A-5):
2NH3 -»• N2 + 3 H2 (A-7)
When ammonia was run through the quartz tube, used for the pyrolysis studies of
model compounds, it was not decomposed up to temperatures of 1100 C (Table A-5).
297
-------
By placing a heated (1050 C) 1/8-inch nickel tube after the quartz pyrolysis tube,
ammonia can be determined by running duplicate experiments: one with the nickel
tube unheated (or bypassing it) and the second with the tube at 1050 C. Ammonia
is calculated from the increase in elemental nitrogen from the first experiment
to the second. The sensitivity of this method is better than 1 ug NH_ (Table A-4)
•5
and interference may be less than by other methods.
TABLE A-5. DECOMPOSITION OF AMMONIA
Nickel Tube
Temperature, C
520
630
750
890
1000
% Conversion to N2
1.1
15
64
100
100
Quartz Tube
Temperature, C
530
650
800
900
1000
1080
% Conversion to N2
0.0
0.0
0.0
0.0
0.05^
0.14
This method was used for all of the subsequent model compound and fossil fuel py-
rolysis experiments. Most experiments were run with the heated nickel tube in the
line, so that results were calculated as the sum of NHj and N2. In most cases,
this value was low (<5 percent conversion of organic N); therefore, there was no
need to determine only the ammonia.
To prevent any catalytic conversion of HCN to N2, a tube containing Ascarite
(NaOH on asbestos) was placed between the exit end of the pyrolysis reactor and
the catalytic converter. The Ascarite trapped the HCN, but not the NH3 which was
then catalytically converted to N2 and H2. After;the effluent gases were passed
through a liquid nitrogen-cooled trap, the remaining gases were gas chromatographed.
After a number of experiments were performed with fuel oil and coal samples, it was
found that the quartz tube was no longer "inert" toward ammonia. Ammonia, when
298
-------
passed through the tube at 1100 C was decomposed to N-. It was assumed that me-
tallic deposits (from the coal or fuel oils) on the quartz wall were now catalyt-
ically decomposing the ammonia. When this phenomenon occurs, the method is lim-
ited to the determination of the sum of NH, and N .
Oxides of Nitrogen, NOX
The Saltzman method was used to determine nitrogen dioxide, NO-, plus nitric oxide,
NO, from oxidative pyrolysis and residue experiments. In order to determine NO
A.
at levels which account for as little as 1-percent conversion of organic nitrogen,
a method was needed that was sensitive to less than 1 ug of nitrogen as NO . The
J\.
colorimetric method of Saltzman (Ref. 141 and 142) has this sensitivity. The TECO
Model 10A chemiluminescent analyzer, which was used in Phase II of this program,
was considered. However, with these small batch samples, considerable modifica-
tion of the apparatus would be required. The Saltzman method is one of the most
widely used for the determination of NO . The determination of both NO and NO
(NO ) necessitates oxidation of the NO. The Saltzman technique for batch NO
A J\
analysis from flames (about 95 percent of which is NO) was recently critically
evaluated using oxygen as the oxidizer (Ref. 145) and optimum operating conditions
were established. Halstead and co-workers (Ref. 146) also reported on the use of
oxygen to convert NO to NO . Using information from both-papers, a system for
collecting samples from the oxidative pyrolysis experiments for determination of
NO was designed and assembled.
A
In the oxidative pyrolysis experiments with model compounds, the oxides of nitro-
gen were analyzed spectrophotometrically using the Saltzman reagent. Samples
were taken, in evacuated 1-liter pyrex flasks that contained 30 ml of Saltzman
reagent. The key step in the analysis was the slow oxidation of NO to N02,
since the reagent is for the determination of N02- After the sample (about 0.2
atm) was introduced, pressure in the flask was brought up to 1 atm with oxygen.
In calibration experiments, a known standard of 130 ppm NO in N_ was used. In a
typical experiment, 55 torr of the standard was introduced into an evaluated 1-
liter flask, then pressurized to 650 torr with 0» (to bring the NO concentration
to 11 ppm), and allowed to stand 24 hours. In experiments of this type recoveries
299
-------
of 89, 99, and 92 percent were obtained by this method for samples containing 8,
13, and 14 micrograms of NO, respectively. These recoveries were measured against
a calibration curve using nitrite ion, NO-'. Because these calibration samples
contained 6 to 11 ppm of NO after being pressurized to 1 atm with oxygen, Fig. 2
of D. H. Fine's review of the Saltzman method (Ref. 145) suggests that much less
than 100-percent recovery would be expected. This figure is misleading, however,
because with 4:1 oxygen dilution, the curve labeled 5-ppm NO applies to the case
where the initial NO concentration is only 1 ppm, etc. Thus, a 10-ppm sample
under our conditions would correspond to the curve labeled 50 ppm in Fine's paper,
and a recovery of about 95 percent would be predicted. This is in good agreement
with the results obtained.
The following equation, which is equivalent to Fine's Eq. No. 3, but is more use-
ful in predicting the expected extent of NO oxidation under given conditions, is
derived in Appendix E.
% NO Unreacted = (0.089 p, . CA-8)
where
PO? = partial 0« pressure in torr (a constant)
t = time in hours
WNQ = initial NO in micrograms
V = flask volume in cc
Under our standard conditions of t = 24, PQ2 = 600 torr and V = 1000 cc, this
equation becomes:
% Unreacted = 100/(1.30 WNQ + 1) (A-9)
Thus,, for the calibration runs with 8, 13, and 14 yg of NO, the predicted yields
are 91, 94 and 95 percent, respectively.
300
-------
APPENDIX B
COAL AND RESIDUE ANALYSIS
Three coal samples were anlyzed for nitrogen in January 1973 by the Dumas method.
The results obtained are shown in the first column of Table B-l. These analyses
were repeated 4 months later, giving the much lower results shown. The reason
for these lower results is not known, but sample inhomogeneity must be suspected.
After the second series of analyses, the percent volatiles* were determined (ASTM
Method D-271) for these coals as well as the percent nitrogen in the residues.
It can be seen from the table that only about one-fourth of the nitrogen remained
in the residue (or less if the higher nitrogen values are correct). These are
smaller amounts of nonvolatile nitrogen than have been estimated previously
(Ref. 2 and 71).
TABLE B-lo COAL ANALYSES AND VOLATILE NITROGEN RESULTS
Coal
IFRF-A
IFRF-N
EPA
EPA
(Dupl icate)
N, weight percent
January 1973
1.16
1.47
1.17
May 1973
0.54, 0.64
0.91, 1.16
0.60
0.58
Percent
Volati les
27.3
41.5
38.0
38.9
N, weight percent
in Residue
0.16
0.18
0.27
0.25
Percent N
in Residue
21.5
11.5
27.8
26.5
Minutes of heating at 950 C.
301
-------
APPENDIX C
MODIFIED ASTM METHOD D2036-72, HYDROGEN CYANIDE.IN A
HELIUM STREAM CONTAINING HYDROGEN SULFIDE
SCOPE
This method covers the determination of hydrogen cyanide at the microgram level
in a helium carrier gas stream containing up to 30 times as much hydrogen sulfide
as hydrogen cyanide.
REAGENTS
Prepare the reagents exactly as described in the indicated paragraphs of ASTM
D2036-72:
Chloramine-T, para. 6.8
Hydrochloric Acid (1 + 9), para. 6.11
Phenolphthalein Indicator, para. 6.14
Potassium Cyanide, Stock Solution, para. 6.15
Potassium Cyanide, Standard Solution, para. 6.16
Pyridine - Barbituric Acid Reagent, para. 6.19*
Sodium Carbonate, Anhydrous, Baker No. 3602
Cadmium Carbonate, Powder, B § A No. 1483
EQUIPMENT
Test Tubes, 25 by 175 mm
Centrifuge Tubes, Graduated, Pear Shape, VW$R No. 21119-001
*When preparing pyridine-barbituric acid reagent, as described in para 6.19 of
ASTM D2036-72, the barbituric acid does not completely dissolve and it varies
from one bottle of material to another. Filter the solution several days after
preparation through Whatman No. 4 paper.
303
-------
Centrifuge, International, Model AE (with appropriate head, trunnion rings,
and centrifuge holder)
5.0-cm Matched Cells (visible region)
Gary 14 Recording Spectrophotometer (or equivalent)
Micro-stirring Bars (1/2-inch length by 1/8-inch diameter)
SAMPLING
Place 125 milligrams of anhydrous sodium carbonate, 40 mg of cadmium carbonate,
and a micro-stirring bar in a 25- by 175-mm test tube. Add 25 ml of distilled .
water and stir to dissolve the sodium carbonate. Collect the sample by bubbling
the helium effluent from the pyrolysis tube through the stirred solution, for
5 minutes, using a 20-gage stainless-steel needle. Analyze the collected sample
immediately.
SUMMARY OF METHOD
The sample is centrifuged to remove suspended cadmium carbonate-cadmium sulfide.
Aliquots of the supernatant liquid are then analyzed as described in ASTM D2036-
72, page 557, para. 12.3.1 and 12.3.2. To save barbituric acid reagent, one-half
the called-for recipes were used and, to increase sensitivity, 5.0-cm cells were
used instead of 1.0 cm.
PROCEDURE
Transfer the collected sample to a 100-ml, pear-shaped centrifuge tube. Stopper
lightly with a cork stopper and place in a centrifuge holder. Place 25 ml of
distilled water in another tube, place it in a holder, stopper, and bring its
weight to within 0.1 gram of the sample tube and holder on a double-beam trip
balance. Place the centrifuge holders in opposite trunnion rings in the centri-
fuge and centrifuge for 5 minutes at 1500 rpm.
304
-------
Immediately transfer an appropriate aliquot of the sample into a 25-ml volumetric
flask. (An aliquot up to 20 ml can be used.) Bring the volume to 10 ml with
sodium carbonate solution (125 mg/10 ml). Place a micro-stirring bar in the
flask. Add one drop of phenolphthalein indicator solution. Add HC£ (1 + 9) drop-
wise with constant stirring until the solution becomes colorless. Add 1 ml of
chloramine-T solution and mix; then immediately add 2.5 ml of pyridine-barbituric
acid solution and again mix. Dilute to the mark with water; mix well by inversion
and allow 8 minutes for color development. Measure the absorbance of the developed
color with the spectrophotometer at 578 mm in a 5.0-cm cell within 15 minutes.*
Prepare a calibration curve from 0.25 to1..5 yg "HCN" using the standard potassium
cyanide solution. The calibration curve is fairly reproducible. Therefore, only
one standard (1 yg) is needed every time samples are run, if the same pyridine-
barbituric acid solution is being used. Run a blank every time a sample is run,
but read the sample(s) against distilled water. Using the calibration curve and
the following equation, determine the total micrograms of HCN collected in the
original sample:
yg HCN = A x •
where
A = micrograms of cyanide read from calibration curve
B = total milliliters of absorbing solution used
*The solution, during and after color development, liberates gas and when it is
in the cell, bubble formation on the cell windows can cause false readings. If
this occurs, it can be minimized by tapping firmly on the wall of the cell with
your finger (Caution: cells are fragile.)
305
-------
APPENDIX D
PHENATE METHOD (TENTATIVE) FOR NH3 MEASUREMENT-STANDARD METHODS
FOR THE EXAMINATION OF WATER AND WASTEWATER*
1. General Discussion
a. Principle: An intensely blue compound, indophenol, is formed by the re-
action of ammonia, hypochlorite and phenol catalyzed by a manganous salt,
b. Interference: Over 500 mg/1 alkalinity, over 100 mg/1 acidity; color
and turbidity interfere. These interferences may be removed by pre-
liminary distillation.
2. Apparatus
a. Colorimetric Equipment--One of the following is required:
1. Spectrophotometer, for use at 630 my with a light path of approxi-
mately 1 cm.
2. Filter photometer, equipped with, a red-orange filter having a maxi-
mum transmittance near 630 my and providing a light path of approxi-
mately 1 cm.
b. Magnetic Stirrer.
3. Reagents
a. Ammonia-free water: Prepare as directed in Preliminary Distillation
Step, Section 132A.3a.
b. Hypochlorous acid reagent: To 40 ml distilled water add 10 ml of a
5% commercial bleach. Adjust pH to 6.5-7.0 with HC1. Prepare this
unstable reagent weekly.
*13th Edition, 1971, pp 232 and 233
307
-------
c. Manganous sulfate solution, 0.003 M: Dissolve 50 mg MnSCVH-O in 100 ml
distilled water.
d. Phenate reagent: Dissolve 2.5 g sodium hydroxide, NaOH, and'10 g phenol,
C^HrOH, in 100 ml ammonia-free water. Since this reagent darkens on
standing, prepare weekly.
e. Stock ammonium solution: Dissolve 381.9 mg anhydrous ammonium chloride,
NH^Cl, dried at 100 C, in ammonia-free water, and dilute to 1,000 ml;
1.00 ml = 100 yg N = 122 yg NHj.
f. Standard ammonium solution: Dilute 5.00^1 stock ammonium solution to
1,000 ml with ammonia-free water; 1.00 ml = 0.500 yg N = 0.610 yg Nil-.
• O
4. Procedure
To a 10.00-ml sample in a 50-ml beaker, add 1 drop (0.05 ml) manganous sulfate
solution. Place on a magnetic stirrer and add 0.5 ml hypochlorous acid solution,
followed immediately the the addition, a drop at a time, of 0.6 ml phenate reagent.
Avoid erratically low results by adding the phenate reagent without appreciable
delay and yet not too rapidly. Use bulb pipets for convenient delivery of the
reagents. Mark the pipet for hypochlorous acid at the 0.5-ml level and deliver
the phenate reagent from a pipet which has been calibrated by counting the number
of drops previously found to be equivalent to 0.6 ml. Stir vigorously during ad-
dition of the reagents. Since the color intensity is affected somewhat by the age
of the reagents, carry a blank and a standard through the procedure along with
each batch of unknows. Measure the color after zeroing the photometer on the
blank. (Color formation is complete in 10 min and is stable for at least 24 hr.
Although the blue color has a maximum absorbance at 630 my, satisfactory measure-
ments can be made in the 600-660 my region.) Prepare a calibration curve in the
ammonia nitrogen range of 0.1-5 yg, treating the standards exactly as the sample
throughout the procedure.
308
-------
5. Calculation
Beer's law governs. Calculate the ammonia concentration as follows:
... ... A x B D
mg/1 Ammonia N = ^ - =• x =•
U X o E
C = absorbance of standard, and S = ml unknown water sample used, D = total dis-
tillate collected, including the acid absorbent, and E = ml distillate used for
color development. The ratio D/E applies only to the diltilled samples, and should
be ignored when the color is developed on undistilled samples.
6. Precision and Accuracy
See Section 132B.6 and Table 132 (2).
BIBLIOGRAPHY
Rossum, J. R. § P. A. Villarruz. 1963. Determination of ammonia by the indophenol
method. JAWWA 55:657.
Weatherburn, M. W. 1967. Phenolhypochlorite reaction for determination of
ammonia. Anal. Chem. 39:971.
309
-------
APPENDIX E
DERIVATION OF NO OXIDATION EQUATION
-d(NO)/dt = k(NO)2(OO (E-l)
Integration of this equation gives::
(E-2)
k(02)t(NO)° = - - 1 (E-3)
(NO) _ _ 1 „
(NO) " k(02)t(NO)° -i- 1
Substituting the following into Eq. E-4:
k = 2.4 x 109 exp(+1046/RT)
= 1.41 x 1010 (m/cc)"2 sec"1 at 298 K
<°2>»/cc =
= PQ T (760 x 82.1 x 298)
= 5.35 x 10"8PQ
(NO)0 , = 3.3 x 10"8 (W..n) •=• V
^ m/cc v NO' yg cc
t = 3600 t,
sec hrs
gives Eq. E-5:
Percent NO Unreacted = (0.089 PQ^ (E'5)
311
-------
APPENDIX F
THERMOCOUPLE RADIATION CORRECTION
The determination of the flame temperature profiles with a thermocouple necessi-
tates that a correction be made for radiation cooling of the thermocouple wire.
The formulation used to correct the measurements made in this study is attributed
to Kaskan (Ref. 132), and considers heat transfer from the gas to the wire to oc-
cur by convective heat transfer due to flow at right angles to single cylinders.
The correlation for the heat transfer coefficient, h, under these flow conditions
is given as (Ref. 147):
h = 0.8 j (Re) 1/4 . (F-l)
w
Equating the rate of heat gain via convection to the radiation heat loss, one ob-
tains the following equation for the temperature correction:
1/4 d
AT = 1.250- e (Re)"i/4 ^ (T4-T4) (F-2)
K W • G
The emissivity of the platinum, e, was taken to be 0.2. This was judged to be
about the average total emittance for platinum given in Ref. 148 between 1400 and
1800 K. The thermal conductivity of the gas, k, was computed at the adiabatic
flame temperature by Rocketdyne's Thermochemical Computer Code and corrected to
the measured gas temperature by multiplying it by the ratio of the corresponding
thermal conductivities of pure argon. The gaseous viscosity term which appears
in the Reynolds number, Re, was determined from a plot of viscosity versus flame
temperature made up from data generated by Rocketdyne's Thermochemical Computer
Code.
313
-------
APPENDIX G
FLAME-PROBING DATA
The data obtained in the flame-probing experiments are listed in Tables G-l
through G-19. Mass spectroscopy data are presented in Tables G-12 through G-19.
Most of these results are plotted in Fig. 56 through 86 presented in the main
body of the report.
315
-------
TABLE G-l. FLAME 1,
-------
TABLE G-2. FLAME 1, COMPLETE RERUN, $ = 0.8, 2500 PPM NH,
Di stance,
mm
1.45
1.77
2.10
2.42
2.75
3.07
3.40
3-72
4.05
4.55
5.55
6.55
7.55
9-05
10.05
Time','
msec
-0.059
0.064
0.182
0.290
0.398
0.500
0.603
0.703
0.803
0.957
1.262
1.566
1.871
2.329
2.635
T.
K
1465
1628
1737
1816
1871
1907
1931
1945
1954
1965
1977
1977
1975
1969
1964
v,
cm/sec
243.4
270.5
288.6
301.7
310.8
316.8
320.8
323-1
324.6
326.5
328.4
328.4
328.2
327-1
326.3
NO,
ppm
394
464
558
684
863
1157
1514
1660
1699
1806
1856
1879
1855
1881
1850
NH3r
ppm
814
733
497
141
109
87
0
0
0
0
0
0
0
0
0
-Time = zero at T = 1550 K
»*NH, mole fraction measured with molybdenum catalyst at 800 C
317
-------
TABLE G-3. FLAME 2, = 0.8, 2500 PPM HCN
Di stance,
mm
7-04
6.5*
6.04
5.54
5.04
4.54
4.04
3-54
3.04
2.80
2.54
2.05
1.59
1.31
T i meV
msec
1-707
1 .561
1.414
1.267
1.12
0.972
0.824
0.674
0.521
0.447
0.364
0.201
0.0347
-0.0766
T,
K
2050
2050
2048
2047
2043
2035
2020
1996
1951
1919
1875
1748
1589
1470
v,
cm/sec
341
341
340
340
339
338
335
331
324
319
307
289
263
241
NO,
ppm
1594
1583
1567
1542
1470
1401
1238
1017
736
639
539
390
310
247
HCN,
ppm
60
122
175
260
498
900
1037
1270
1478
1670
1878
''-Time = zero at T = 1550 K
318
-------
TABLE G-4. FLAME 3, = 1.5, 2500 PPM HCN
Di stance,
mm
2.85
3-35
3.85
4.35
4.85
5-35
5.85
6.35
6.85
7-35
8.11
8.85
9-35
-r • -,'r
Time,
msec
0.176
0.420
0.641
0.846
1.039
1.223
1.401
1.574
1.744
1.913
2.166
2.410
2.574
T,
K
1360
1510
1635
1735
1820
1882
1928
1962
1985
2003
2012
2012
2011
v,
cm/sec
193-2
216
235
251.6
265.6
276.5
285
291.4
296.5
299-2
300.2
300.2
300.1
NO,
ppm
24.5
38
52
72
87
114
151
174
202
216
254
266
271
HCN7"
ppm
-
1200
1463
1700
1912
2008
*Time = zero at T = 1250 K
>»HCN measured using molybdenum catalyst at T = 800 C with no air addition.
319
-------
TABLE G-5. FLAME 3 RERUN = 1.5, 2500 PPM HCN
Di stance,
mm
3-0
3-5
4.0
4.5
5.25
5.75
6.25
5-0
6.75
7-25
7-75
8.25
8.75
9.25
10.0
10.75
11.5
12.25
13.25
T i me V
msec
0.25
0.485
0.700
0.901
1.183
1.361
1.536
1.706
1.875
2.042
2.208
2.373
2.537
2.787
3-038
3.288
3-54
3.88
J. -L
TV"
K
1407
1550
1670
1761
1870
1920
1957
1840
1982
2000
2009
2012
2011
2008
2004
2000
1996
1990
1984
v,
cm/sec
200
219
236
249
275
284
290
267
294
298
299
301
301
300
300
299
298
297-5
296.6
NO,
ppm
44.6
63.3
78.8
95.8
112.9
162.5
177-6
103
202
239
282
308
348
378
442
476
498
522
552
HCN***
ppm
2039
1780
1703
1622
1333
1267
1157
1314
1058
987
858
802
715
650
538
473
439
391
340
*Time = zero at T = 1250 K
"•'-Temperature profile is from original flame 4 experiment
-""HCN measured using molybdenum catalyst at T = 800 C with air addition to aid
efficiency of conversion of HCN to NO
320
-------
TABLE G-6. FLAME 4, (J) = 1.5, 2500 PPM NH,
Di stance,
mm
1.85
2.35
2.85
3-35
3.85
4.35
4.85
5-35
5-85
6.35
7-10
7.85
T • A
T i me ,
msec
-0.406
-0.076
0.199
0.441
0.661
0.863
1.051
1.228
1.398
1.564
1 .806
2.047
T,**
K
990 .
1190
1386
1524
1635
1732
1825
1904
1968
2017
2066
2080
v,
cm/ sec
137-1
166
195.2
218
237-9
2 57' 5
274
289
298.7
305.8
313-6
309-8
NO,
ppm.
37
35-6
50.4
77-7
113
152
207
249
284
317
395
416
1
j. j, .».
No2r"
ppm
0.6
3.0
3-3
4.8-5.5
5.6-6.3
7-9
5.8
0
0
0
0
0
*Time = zero at T = 1250 K
•"•Temperature measurements were made at different distances than those at which
product samples were drawn. The temperature measurements are plotted in
Fig. 68. .Temperatures listed here were read off that plot.
*A*NO mole fraction measured with molybdenum catalyst at T = 400 C
321
-------
TABLE G-7. FLAME 4 RERUN, =- 1.5, 2500 PPM NH.
Di stance,
mm
2.52
3-02
3.51
4.02
4.52
5.02
5.52
6.02
6.52
7-02
6.77
6.52
7-52
8.02
8.77
9-50
10.27
11.27
j.
T i me ',"
msec
0.0227
0.285
0.513
0.729
0.925
1.108
1.283
1.451
1.615
1.776
1.686
1.615
1-937
2.098
2.341
2.583
2.826
3-152
T***
K
1265
1433
1560
1668
1763
1856.
1930
1985
2030
2060
2047
2030
2078
2080
2078
2075
2064
2056
v,
cm/sec
177
204
226
246
264.5
280
293
302
308
312
310
308
309.5
309.8
309.5
309
307.3
306.2
NO,
ppm
58
78
93
127
155
183
220
281
273
320
311
298
353
402
447
518
560
609
NH3,
ppm
1818
1636
1850
1719
1754
1805
1818
1701
1612
1431
-
-
1261
1111
940
774
670
532
HCNf
ppm
986
916
1000
966
984
1012
1018
971
920
' 833
-
-
746
646
544
447
378
306
''-Time = zero at T = 1250 K
»"Data on NH^ additive reduced as if it were HCN
"••-•Temperature profile is from original flame 1 experiment (Fig. 68)
322
-------
TABLE G-8. FLAME 5, (j) = 0.8, 1793 PPM HCN + ADDITIONAL ARGON (DR =1.4)
Di stance,
mm
3.265
7-765
6.776
6.765
6.265
5-765
5.445
5-035
5.28
4.805
4.63
A. 385
4.245
4.065
3-845
3-645
3.435
3-015
T i meV
msec
2.755
2.127
1.715
1.715
1.503
1.294
1.160
0.998
1.091
0.890
0.816
0.753
0.692
0.614
0.516
0.425
0.328
0.1278
T,
K
1870
1888
1890
1890
1883
1882
1877
1861
1871
1848
1836
1814
1799
1780
1752
1720
1688
1605
v,
cm/sec
237-9
240.2
240.4
240.4
239-5
239-5
238.8
236.7
238
235-1
233.6
230.8
228.9
226.4
222.9
218.8
214.7
204.2
NO,
ppm
1523
1496
1436
1410
1334
.1245
1172
975
1078
871
783
642
586
496
428
386
344
254
1
HCN,
ppm
66.7
87
137
-
187
242
304
444
356
574
717.6
903
1017
1059
1167
1280
1370
1456
-Time = zero at T = 1550 K
323
-------
TABLE G-9. FLAME 6, = 0.8, 675 PPM NO + 1525 PPM HCN
Di stance,
mm
6.00
5-50
5.00
4.50
4.00
3.75
3-50
3.25
3.00
2.75
2.50
2.25
2.00
1.75
1.46
Time'/
msec
1.373
1.222
1.071
0.919
0.766
0.690
0.612
0.535
0.456
0.376
0.29^
0.210
0.121
0.0272
-0.0886
T,
K
1991
1989
1985
1979
1966
1957
19^7
1930
1904
1860
1808
1740
1661
1575
1466
v,
cm/sec
330.5
330.3
329-6
328.6
326.4
324.9
323.2
320.4
316.1
308.8
300.2
288
274.9
259-8
240.8
. NO,
ppm
1942
1909
1866
1789
1637
1528
1407
1241
1108
1013
901
838
791
751
726
HCN,
ppm
35-7
29.8
71.2
148
207
241
276
448
679
851
942
1045
1168
1227
''-Time = zero at T = 1550 K
324
-------
TABLE G-10. FLAME 7, = 1.5, 675 PPM NO + 2500 PPM HCN
Di stance
mm
1.46
2.00
2.50
3-00
3-50
4.00
4.50
5.00
5.50
6.00
6.50
7-00
7-50
8.00
8.50
9.00
9-75
Time,»
msec
-0.657
-0.257
0.006
0.276
0.5H
0.724
0.921
1. 107
1.284
1.457
.1.626
1.794
1.960
2.126
2.292
2.457
2.705
T,
K
890
1092
1277
1415
1558
1677
1787
1864
1924
1971
2000
2018
2025
2028
2029
2027
2025
v,
cm/sec
121.4
148.9
166.1
202.1
224.5
243.9
262.5
277-1
286
292
297-3
299-9
301
301.3
301.5
303-3
303
i ]
NO,
ppm
476
488
478
451
428
393
262
309
316
349
279
339
376
457
496
564
657
HCN,
ppm
1586
1549
1470
1440
1335
1282
1171
1149
1088
1035
984
852
824
683
632
561
527
*Time = zero at T = 1250 K
325
-------
TABLE G-ll. FLAME 8, (j) = 1.5, 675 PPM NO
Di stance,
mm
1.50
2.00
2.50
3.00
3-50
4.00
4.50
5.00
5-50
6.00
6.50
7.00
7.75
8.75
9-75
10.75
11.75
T i me?
msec
-0.483
-o. 138
0.149
0.398
0.622
0.825
1.011
1. 190
1.365
1.536
1.705
1.873
2.124
2.460
2.797
3-135
3.474
T,
K
965
1157
1339
1492
1623
1735
1821
1872
1927
1955
1970
1977
1976.6
1972
1965
1959
1953
v,
cm/sec
131.5
157.7
189-3
212.3
232.4
261.7
274.6
282.3
290.6
294.8
297
298
298.4
297.4
296.4
295.5
294.5
NO,
ppm
502
510
485
457
416
352
296
239
217
204
173
184
190
225
244
266
283
HCN,
ppm
8.2
35.8
31
37.4
50.7
66.3
96
115
118
121
126
110
86
50
33.2
13-9
2.8
'-Time = zero at T = 1250 K
326
-------
TABLE G-12. FLAME 1 RERUN, 4> = 0.8, 2500 PPM NH.
Compound
H2
CH,,
NH
H20
C2H2
CO
N2
°2
Ar
co2
NO
X
H2CO
C2H6
C2H4
HCN
Distance, mm
Time, msec
Sample No.
1*
4.55
0.956
2*
4.05
0.803
3
1.37**
0.022
0.09
12.97
0.023
4.49
0.916
6.14
68.7
4.58
0.24
-
0.014
-
0.024
3-72
0.703
4
1.38
0.157
0.069
12.56
-
4.98
1 .2
6.64
69.18
3.87
0.22
-
0.077
0.061
-
3.40
0.603
5
1.77
0.55
0.03
11.44
-
4.85
1 .02
8.11
69.0
2.89
0.16
0.037
0.013
0.091
-
3.07
0.500
6'
1.86
1.21
10.22
0.009
3.82
1.58
8.94
69.8
1.96
0.102
0.07
-
0.14
0.012
2.75
0.398
7*
2.42
0.2902
8
1.6
4.0
0.014
5.85
0.04
-
-
14.94
72.94
0.32
0.07
0.028
0.2
-
0.02
1.77
0.0642
"Samples 1, 2, and 7 were destroyed when the break seal was opened
inadvertantly by mass spectroscopists.
•'•Mole percent
327
-------
TABLE G-13. FLAME 2, = 0.8 2500 PPM HCN
Compound
H2
CH/t
NH
H20
C2H2
CO
N2
°2
Ar
co2
NO
X
H2CO
C2H6
C2H4
HCN
Distance, mm
Time, msec
Sample No.
4
0.75*
0.037
0.034
14.0
0.077
1.50
1.78
5.40
70.4
5.66
0.13
0.13
-
0.003
0.13
4.54
0.972
5
0.76
0.034
0.003
13-9
0.063
2.28
1.53
5-73
70.4
4.91
0.126
0.054
0.044
-
0.07
4.04
0.824
6
0.89
0.086
-
14.94
-
3-49
1.27
6.23
68.8
3.70
0.118
-
0.047
-
-
3-54
0.674
7
1.11
0.534
-
12.66
0.05
3-93
0.59
7.17
70.9
2.78
0.11
0.04
-
0.08
0.075
3.04
Q.521
8
1 .21
1.58
-
10.23
0,023
3.51
0.35
9.96
70.6
2.13
0.064
0.08
-
0.16
0.088
2.54
0.364
9
1.03
2.80
-
7-88
-
2.62
0.224
12.59
70.9
1.62
0.026
0.066
0.05
0.16
0.053
2.05
0.201
10
0.86
3.85
-
5.09
-
0.98
0.25
13.75
73.1
1.29
0.0403
-
0.25
0.58
-
1.59
0.0347
Samples 1 and 3 had large air leaks; this condition could not be corrected.
Sample 2 had an extremely poor carbon balance.
Mole percent
328
-------
TABLE G-14. FLAME 3, =1.5, 2500 PPM HCN
Compound
H2
CH4
NH3
H20
C2H2
CO
N2
°2
Ar
CO,
N0x
H2CO
C2H6
C2H4
HCN
Distance, mm
Time, msec
Sample No.
1
0.1 7*
12
0.036
-
0.10
-
4.8
18
65
-
0.026
0.12
-
0.025
0.27
2
8.62
0.061
0.012
14.96
0.058
9.984
1.089
0.22
60.81
3-99
0.024
0.0499
-
0.2
0.044
8.85
2.41
3
8.86
0.051
0.032
14.44
0.079
9.95
1.27
0.0615
59.69
4.52
0.081
0.050
-
0.071
0.127
7.35
1.913
4
9.01
0.069
0.042
14.50
0.054
9-91
1.08
0.34
61.27
3.78
0.077
-
0.066
0.234
-
6.85
1.744
5
7.35
0.108
0.048
16.1
0.134
9.86
0.986
0.565 .
60.95
3-32
0.067
0.0036
-
0.206
0.035
6.35
1.574
6
7.36
0.227
0.036
14.95
0.058
10.0
0.80
1.18
61.79
2.82
0.044
-
0.309
0.418
• -
5.85
1.401
7
7.125
'0.84
0.029
14.22
0.146
9.135
0.713
2.83
62.12
2.10
0.040
-
0.192
0.484
-
5.35
1 .223
8
6.76
1.76
0.007
12.48
0.111
8.057
0.676
4.816
62.97
1.574
0.011
-
0.042
0.491
-
4.85
1.039
9
5.83
3-33
0.009
10.37
0.063
5.65
1.204
7.50
62.97
0.87
-
0.020
0.398
0.472
-
4.35
0.846
10
4.78
5.34
-
8.27
0.057
4.50
2.44
10.31
62.78
0.581
0.008
0.052
0.178
0.319
0.032
3.85
0.651
OJ
to
vo
Sample 1 was a sample of unburned mixed combustible fed to the burner
"Mole percent
-------
TABLE G-15. FLAME 4, $=1.5, 2500 PPM NH,
Compound
H2
CH4
NH,
H20
C2H2
CO
N2
°2
Ar
co2
NO
X
H2CO
C2H6
C2H4
HCN
Sample No.
1**
-
13-0
-
0.23
-
0.26
0.60
18.0
68.0
0.006
_
0.002
0.016
0.004
-
Distance, mm
Time, msec
2
2.4**-
12.0
0.006
0.046
-
1.1
-
16.0
68.0
0-15...
0.004
-
0.11
0.003
-
1.85
3
3.2
12.0
0.002
0.092
-
1 .2
0.032
16.0
68.0
0.18
-
-
0.19
-
-
2.35
4
7.88
0.0042
0.005
16.28
0.073
8.38
1.84
0.302
61.21
3.44
0.054
-
0.101
-
-
7.85
2.047
5
10.39
0.029
0.0008
13-50
0.029
8.66
1.56
0.364
62.35
3.20
0.048
-
0.156
0.182
-
7.10
1 .806
6
10.43
0.122
0.0008
13-2
0.174
9.56
1.48
0.756
61.74
2.78
0.056
0.063
-
0.235
-
6.35
1.564
7*
5.85
1 .40
8
8.20
0.837
0.005
13.8
0.319
7.34
1 .12
2.68
63.01
1.73
0.0259
0.015
0.121
0.155
-
5-35
1.228
9
8.62
2.54
0.0027
9.42
0.327
6.17
1.81
5.81
62.60
1.36
0.013
0.033
0.029
0.154
0.009
4.85
1 .051
10
8.49
4.96
0.0037
8.72
0.164
5.11
0.666
8.67
62.07
0.904
0.009
0.091
0.10
0.301
-
4.35
0.862
11
6.96
7.16
0.001
3.45
0.097
3.48
0.599
11 .60
65.75
0.58
0.003
0.023
0.242
0.184
-
3.85
0.661
12
5-90
8.85
0.0039
1.72
0.0984
2.656
0.384
13.77
65-9
0.413
0.0069
0.038
0.0619
0.061
0.045
3.35
0.441
13
4.9
0.5
0.005
0.11
0.074
1.7
0.39
15-0
67.0
0.23
0.006
0.026
--
0.063
2.85
0.199
"Sample 7 developed an air leak when MS analysis was performed
""Sample 1 was a sample of unburned mixed combustible fed to the burner
:"*Mo1e percent
-------
TABLE G-16. FLAME 5,4) =0.8, 1793 PPM HCN + ADDITIONAL ARGON (DR = 1.4)
Compound
H2
CM,,
NH,
H20
C2H2
CO
N2
°2
Ar
co2
N0x
H2CO
C2H6
C2H*4
HCN
Distance, mm
Time, msec
Sample No.
1
0.56*
0.01
0.0*43
10.7
-
2.*»9
0.58
3.97
76.57
*4.3*»
0.157
-
0.6*4
0.09
-
6.265
1.503
2
0.51
0.02
0.032
10.9
-
2.78
0.575
3.90
77.03
3-90
0.139
0.186
-
0.087
0.002
5.765
1 .29**
3
0.58
0.02
0.069
10.79
0.0*43
2.78
0.77
*».085
77.05
3.81
0.121
-
0.0*43
-
0.05
5.*»*»5
1.16
4
0.815
0.111
0.035
10.26
0.088
3-52
0.898
*4.91
75.9*»
3.06
0.111
0.033
-
-
0.13
5.035
0.988
S
*».63
0.816
6
0.808
0.7*4
0.03*4
8.5*t
-
3.10
0.855
7.2*.
77.08
1.50
0.0*t8
-
0.263
0.018
0.006
*4.2*45
0.692
7
0.73
1.39
0.021
7.18
-
2.*4l
0.85*4
8.*45
76.11
0.928
0.033
0.0195
0.213
0.088
O.OC*t
3.8*45
0.516
8
0.66
2.09
0.007
6.*tO
0.02*4
1 .80
1.23
10. H
76.89
0.49*4
0.036
0.10*4
-
0.123
0.123
3.*f35
0.328
9
0.**38
0.022
0.0*43
11.6*4
0.006
2.33
0.596
3-63
76.35
5.**0
0.1*49
0.037
-
0.1*»9
-
6.765
1.715
10
0.078
5-3
-
-
0.022
. -
2.61
16
76
-
0.02
0.09
-
0.12
0.16
-
-
Sample 10 was a sample of unburned premixed combustible fed to the burner
*Mole percent
-------
TABLE G-17. FLAME 6, = 0.8, 675 PPM NO + 1525 PPM HCN
Compound
H2
CM,,
NH
H20
C2H2
.CO
N2
°2
Ar
co2
NO
X
H2CO
C2H6
C2H4
HCN
Distance, mm
Time, msec
Sample No.
1
1.022*
0.0158
0.0827
13-63
0.011
2.97
• -
4.785
70.62
6.60
0.297
0.0158
-
0.0901
0.0344
5.00
1.071
2
1.11
0.0167
0.056
13.52
0.0037
3.426
-
5.004
70.37
5.93
0.194
0.086
-
0.0565
0.047
4.50
0.919
3
1.11
0.025
0.0926
13-31
-
4.258
-
5-37
70.35
5.09
0.222
0.036
-
0.157
-
4.00
0.766
4
1.503
0. 103
0.131
12.45
-
4.70
-
6.182
70.47
4.134
0.225
-
0.15
0.0921
-
3-50
0.612
5
1.71
0.627
0.0427
11.43
0.053
4.844
-
8.025
70.28
2.75
0.152
0.0066
0.1235
-
0.105
3.00
0,456
6
2.14
1.85
0.0526
8.34
-
3.216
-
11.168
71.14
1.56
_
0.29
-
0.244
0.166
2.50
0.295
7 '
1 .504
3.309
0.059
5.631
-
0.582
-
14.84
73.19
0.531
0.0762
-
0.572
-
0.0191
2.00
0.121
8
1.11
4.66
-
4.877
0.071
-
16.21
72.93
-
0.142
0.0304
-
-
0.192
1.46
-0.0886
*Mo1e percent
332
-------
TABLE G-18. FLAME 7, <)> = 1.5, 675 PPM NO + 2500 PPM HCN
Compound
H2
Ch\
NH
H2°
C2H2
CO
N2
°2
Ar
co2
NO
X
H2CO
C2H6
C2H4
HCN
Distance, mm
Time, msec
Sample No.
1
6.20*
7-99
-
4.27
0.069
-
-
13-13
68.31
0.0189
0.147
-
0.252
0.147
3.00
0.276
2
7-21
4.504
0.07
8.798
0.16
3.60
-
7-524
66.06
0.731
0.023
0.118
0.094
0.35
0.13
4.00
0.724
3
8.24
0.947
0.069
13.10
0.104
9.278
-
2.82
62.49
2.18
0.036
0.069
-
0.70
-
5.00
1. 107
4
12.59
0.126
0.126
10.86
0.0639
10.65
-
0.255
61.96
3.58
0.0513
0.019
-
0.368
-
6.00
1.457
5
8.13
0.032
0.12
15.32
0.029
10.158
-
0.12
61.00
4.34
0.0545
-
0.102
0.139
0.027
7.00
1.794
6'
8.107
0.022
0.079
15-707
0.0286
10.13
-
0.043
60.80
4.70
0.057
0.035
-
0.0857
0.068
8.00
2.126
7
8.39
0.0308
0.0811
15.35
0.045
10.254
0.028
60.60
4.848
0.149
-
-
0.077
0.078
9.00
2.457
8
7-72
0.0327
0.069
16.229
0.0472
9.994
-
0.0136
60.87
4.815
0.118
0.0254
-
0.0436
0.10
10.50
2.952
"Mole percent
333
-------
TABLE G-19. FLAME 8, = 1.5, 675 PPM NO
Compound
H2
NH
C2H2
CO
2
°2
Ar
co2
NO
X
C2H6
HCN
Distance, mm
Time, msec
Sample No.
1
5.21*
9.03
0.0021
3.295
0.017
13.916
68.02
0.0106
0.106
0.213
-
2.5
0.149
2
6.38
6.28
0.0142
6.58
0.094
1.113
11.216
67.82
0.182
0.0081
0.098
0.263
0.263
-
3-5
0.622
3
13.04
2.207
0.0692
6.09
0.291
7.424
4.465
63.21
1.706
0.0361
0.2307
0.492
0.0873
4.5
1.011
4
10.576
0.26
0.026
12.004
0.0423
10.576
1.096
61.54
3.173
0.0298
0.125
0.49
.
5.5
1.365
5
11.48
0.058
0.096
12.07
0.0076
10.52
0.316
61.23
4.018
0.041
0.032
0.277
-
6.5
l.;705
6
11.445
0.034
0.077
12.45
0.015
10.49
0.105
61.04
4.58
0.035
0.0315
0.105
0.0019
7.75
2.124
7 '
13-73
0.077
0.0176
9-38
0.147
10.79
0.055
59.82
4.806
0.074
0.186
0.05
0.255
8.75
2.46
8
8.68
0.027
0.018
15.41
0.0397
10.16
0.031
60.95
4.80
0.066
0.019
0.023
0.048
9.75
2.797
•Mole percent
334
-------
APPENDIX H
DERIVATION OF ONE-DIMENSIONAL FLOW EQUATION INCLUDING
DIFFUSION AND CHEMICAL REACTION
Considering two plane sectors, 1 cm in area and perpendicular to the direction
of flow, located a small distance apart at axial positions x, and x_, dx = x_-x...
The molar flowrates of species i through the first and second sectors, respectively,
is given by the expressions:
vC.
(H-l)
and
vC.
(H-2)
where
v
= gas velocity, cm/sec
C. = concentration of i, moles/cc
]3.. = diffusion coefficient for i in excess j, crc /sec
Since the volume between the plane sectors is dx cm , the moles of i lost by
chemical reaction per second in this volume is R. dx, where R. is the reaction
3
rate of i in moles/cm -sec. Therefore,
..
ij \ dx
= vC. , - &. .
- R. dx
R. =
R. =
ft- 1
i,j dx
d2C.
h)
L\dx/2
dC.
i.J j 2 dx
M 1
Vdx/J
(H-3)
(H-4)
(H-5)
335
-------
APPENDIX I
APPLICATION OF MODELS TO PREDICT
COMBUSTION BEHAVIOR
SINGLE DROPLET APPLICATION
Although the droplet/particle, KDM, and GKAP* models described in the body of
this report are presently uncoupled, they may be combined to analyze the overall
behavior of the droplet, vapor reactions, and fate of all species as they diffuse
to the bulk gas stream. This sequence is schematically illustrated in Fig. 1-1
where the areas of interest are the droplet surface, diffusion zone (which could
include a flame front if the appropriate model were used, and the bulk gas flow-
field. Several input parameters must be specified in order to select the ap-
propriate droplet model (i.e., vaporization or flame-front) and to initialize the
input parameters required to start the solution procedure. The first step is to
specify the convective conditions around the droplet. Based on the relative
velocity between the gas and droplet (for an oil droplet) either the vaporization
or flame-front model is selected. If the relative velocity is greater than about
10 to 20 ft/sec, the vaporization model should be selected. For relative veloci-
ties of less than 10 ft/sec, the flame-front model is used. This velocity cri-
terion is based on the fact that as the relative velocity increases, the flame
surrounding a droplet will be swept into the wake of the droplet and, therefore,
at high relative velocities, the droplet vaporizes without a flame front sur-
rounding the droplet. The exact extinction velocity can only be determined by a
detailed kinetic/diffusion analysis around the droplet, and is dependent upon the
fuel, environmental oxygen concentration, and temperature. (Note: This detailed
study is performed later in the analysis so that the analyzer will know if he
selected the correct model.)
*Ultrasystems program developed for the EPA describing the bulk gas species and
flow characteristics.
337
-------
DROPLET
VAPORIZATION
MODEL
DROPLET
FLAME-FRONT
MODEL
PARTICLE
COMBUSTION
MODEL
INITIAL PREDICTION OF TEMPERATURE
AND SPECIES PROFILES
SELECTION OF
APPROPRIATE
DROPLET/PARTICLE
MODEL
INITIAL DATA
INPUT
KINETIC/DIFFUSION
MODEL
DROPLET
OR
PARTICLE
FINAL DETERMINATION OF
TEMPERATURE AND SPECIES
PROFILES
BULK
FLOWFIELD
DIFFUSION
ZONE
GKAP
MODEL
Figure 1-1. Simplified Schematic Showing the Required
Sequence of the Models
338
-------
The droplet/particle ensemble models calculate: (1) the droplet/particle com-
position, temperature, and diameter "life-histories;" (2) combustion rate of
each droplet particle species; (3) droplet/particle heating rate; and (4) esti-
mated film phenomena, including film thickness and composition/temperature pro-
files through the film. It should be noted that depending on which droplet/
particle model is used, the assumption of infinite or zero kinetic rates for
reactions between the droplet surface and the bulk stream is assumed. This assump-
tion, however, does not in the end affect the results, since the various profiles
as output from the particle ensemble model are used only as initial input to the
average kinetic/diffusion model (KDM) where the actual kinetics are considered.
The KDM model is only executed for the droplets/particles at selected times
during the droplet lifetime in order to reduce overall computing time. The dro-
plet temperature calculated by the particle ensemble model (previously) and com-
position are now used as initial droplet surface boundary conditions in the KDM
model. In addition, the film thickness calculated by the ensemble model is used
to define the outer boundary radius of the film. Lastly, gas phase composition
is used to define the composition at the outer boundary. It should again be
emphasized that the estimated film composition and temperature profiles calculated
by the ensemble model are used only as initial input profiles in the KDM model.
Based on the boundary conditions and detailed kinetic reactions, the KDM program
solves the continuity, diffusion, and energy equations for each droplet/particle
in order to calculate the exact specie mole fraction, specie molar flowrate, and
temperature profiles in the film surrounding the droplet/particle. These cal-
culations include the production rate of pollutant species, such as NO, CO, and
unburned hydrocarbons in the film. This model provides corrected values for the
temperature and species concentration profiles in the diffusion zone as depicted
in Fig. 1-1.
Since the KDM program includes the effects of convection (through the film thick-
ness) and fuel kinetics, the model is capable of determining if a flame surrounds
the droplet/particle. If the KDM calculations are not in agreement with the
model used as the particle ensemble model in relationship to flame structure, then
339
-------
the correct particle ensemble model must be executed in order to determine the
correct input data for the KDM model.
Lastly, as shown in Fig. 1-1, the KDM output, specifically the molar flowrates at
the outer boundary of the film, are then used as mass addition input terms for the
General Kinetic Addition Program (GKAP). The GKAP program calculates the gas flow-
rate, composition, temperature, velocity, and/or pressure along the streamtube.
These calculations include emission levels generated by gas phase kinetics and
pollutants produced in the film surrounding the droplets/particles (through the
mass addition terms).
COMBUSTOR APPLICATION
Although the droplet and kinetic/diffusion submodels described in this report are
written for a single droplet, it is possible to extend them from the analysis of
a single droplet to a complete fuel spray by assuming that the initial distribu-
tion of drops within any streamtube is identical to the overall distribution of
drop sizes initially injected into the combustor, and then summing the individual
droplet contributions. While this type of model assumes uniform mixing in the
radial direction along any given streamtube, it does not restrict streamtubes
from having differing equivalence ratios. Lastly, the model in its present form
will not consider mass addition from other streamtubes, nor does it allow for
droplet-droplet interactions within the same streamtube. However, even with
these restrictions, the models used in this manner can be useful to the research
scientist or design engineer.
The general method for analyzing the streamtube flow is somewhat different than
that described for a single droplet because of.the variation of flow parameters
as a function of axial length. First, gas flow conditions (such as temperature,
velocity, and oxygen concentration) along the streamtube are estimated, based on
experimental data and/or past experience. These estimated gas flow conditions
are then used as input data to the particle ensemble combustion model (Fig.1-2).
The particular particle ensemble model to be used is selected as previously
discussed.
340
-------
• COMPOSITION. TEMPERATURE. AND DIAMETER
"LIFE-HISTORIES"
• COMBUSTION RATE Of EACH DROPLET/PARTICLE
SPECIE
• DROPLET/PARTICLE HEATING RATE
• ESTIMATED FILM PHENOMENA
• COMPOSITION AND TEMPERATURE PROFILES
• RESIDENCE TIMES WITHIN THE FILM
• DIFFUSION RATES THROUGH THE FILM
• FILM THICKNESS
PARAMETERS:
• FUEL TYPE
• DROPLET OR PARTICLE SIZE
• SURROUNDING ENVIRONMENT
• DEGREE OF CONVECTION
• MODIFIED FILM COMPOSITION AND
TEMPERATURE PROFILES BASED ON
DETAIL KINETICS AND DIFFUSION
• THERMAL AND FUEL NO, FILM
PRODUCTION RATES
• FILM PRODUCTION RATES FOR CO. CO}.
H2O. HC. ETC.
PARAMETERS:
• DROPLET OR PARTICLE SIZE
• SURROUNDING ENVIRONMENT
• FILM THICKNESS
• REACTION RATE DATA
• DIFFUSION COEFFICIENTS
ONE-DIMENSIONAL GENERAL
KINETIC ADDITION PROGRAM
(GKAPI
• GAS FLOWHATE. COMPOSITION. TEMPERATURE.
VELOCITY. AND PRESSURE ALONG STREAMTUBE
• GAS PHASE PRODUCTION RATES OF CO. CO2.
H2O. HC, ETC.. INCLUDING POLLUTANTS
PARAMETERS:
• FLOW GEOMETRY
• MASS AND ENERGY ADDITION TERMS
CALCULATED BY KDM INCLUDING POLLUTANTS
DOES GAS FLOW CONDITIONS
AGREE WITH ESTIMATED
VALUES?
Figure 1-2. Flow Diagram for Determination of
Combustor Emission Levels
341
-------
As before, the output of the droplet/particle model is input to the KDM model.
For the streamtube case, the KDM model is only executed at selected locations
along the streamtube to reduce overall computing time. The calculated results
are the input into the GKAP model. If the gas temperature and composition along
the streamtube that are calculated by GKAP are different than the temperatures
and compositions used as boundary conditions in the KDM calculations, the KDM
and GKAP calculations must be rerun with the new gas temperatures and composi-
tions. If the GKAP gas temperatures, velocities, and oxygen concentrations are
different than the conditions used in the particle ensemble model calculations,
the particle ensemble, KDM, and GKAP programs must be re-executed with the new
estimated gas flow conditions. The gas temperature, velocities, and oxygen con-
centrations are only checked for the possibility of re-executing the particle
ensemble model because the ensemble models are most sensitive to these gas
variables.
After agreement has been obtained between gas flow variables calculated by GKAP
and the estimated values used in the particle ensemble and KDM models, GKAP yields
the final emission levels, including both gas-phase- and two-phase-generated
pollutants.
Since the particle ensemble models are insensitive to gas variables, except tem-
perature, velocity, and oxygen concentration, overall computation time would be
reduced if these variables could be determined without executing the KDM program.
The main reason for this time reduction is that the KDM program requires at least
an order of magnitude more computation time than does the particle ensemble models
or the GKAP program. If experimental temperature, velocity, and oxygen composition
data are available, then the particle ensemble model can be executed only once to
i
obtain the input data required by KDM. If no experimental data are available,
then the user must estimate the gas flow variables before executing the particle
ensemble model. The initial guess of the variables will probably not be correct.
Therefore, the following procedure should be used to converge on the gas flow
variables.
342
-------
First, the flow variables are estimated, based on past experience (Fig. 1-3).
Then the particle ensemble model is executed to determine both the fuel combus-
tion and oxygen consumption rates. These mass addition terms are then used as
input to the GKAP program, which calculates the gas temperature, velocity, and
oxygen concentration. These variables are then compared with the estimated values
and the procedure repeated until convergence has been obtained. After convergence
has been obtained, the KDM program is executed, and the procedure is the same as
discussed in the previous paragraphs, It is recommended that this procedure be
applied to a selected set of data where little or no recirculation occurred to
determine the validity of the model approach.
343
-------
GAS
ESTIMATED
FLOW CONDITIONS
PARTICLE ENSEMBLE
MODEL
COMBUSTION RATE
OF EACH SPECIE
GKAP
MODEL
GAS TEMPERATURE,
VELOCITY, AND
COMPOSITION
NO
AGREEMENT WITH
ESTIMATED
VALUES •
YES
I TO
KDM
Figure 1-3. Estimation Procedure for Gas Flow Conditions
344
-------
NOMENCLATURE
A Pre-exponential factor in rate constant or burner exit port area
A Area of orifice
B Pre-exponential coefficient in forward Arrhenius reaction rate
equation
c Molar density
C Orifice constant or centigrade
CA Chemiluminescent analyzer
C. Concentration of species i
C Specific heat at constant pressure
dw Diameter of thermocouple wire
J& Diffusivity
fir. . Diffusion coefficient of species i into species j
DR Dilution ratio: (Ar/02)/3.76
E Activation energy or electrical potential
AE Forward activation energy in Arrhenius reaction rate equation
f^ Mole fraction of species i
F Mass flowrate
g Acceleration of gravity
h Heat transfer coefficient
H Enthalpy
AH° Q Heat of formation
AHVAp Latent heat of vaporization
i Summation indices or species number
IFRF International Flame Research Foundation
j Summation indices or species number
k Thermal conductivity
or
Summation indices or species number
or
Reaction rate constant
345
-------
k" Thermal conductivity
K Kelvin
K Equilibrium constant
£ Summation indices or species number
m Summation indices or species number
M Molecular weight
MC Model Compound
MW Molecular weight
n Summation indices or species number
N Reaction order
•
N Molar flowrate
NuH,Nu,. Nusselt numbers for heat and mass transfer
P Pressure
ppm Parts per million (volumetric basis)
q Heat of combustion
Q Heat transfer rate
Q Heat of Combustion
C*
r Radius or radial coordinate
R Gas constant or reaction rate
R Reaction rate
Re Reynold's Number
S .Entropy
SS Stainless Steel
t time
T Temperature
T Time integrated average temperature of combustion gas from the
burner face to a distance y
T Temperature of surroundings '.
6
Tr Gas temperature (model compound reactor)
T Temperature of the NO calibration standard gas at the critical flow
sampling orifice
T Temperature of the unknown gas at the critical flow sampling
orifice
346
-------
T Temperature of thermocouple wire or wall temperature
W
v Velocity
V Volume
V Flowrate of gas to the CA from the sonic sampling orifice
V^ Diffusion velocity of species i
V Flowrate of NO calibration standard gas to the CA from the sonic
sampling orifice
V Flowrate of unknown gas through a sonic sampling orifice to the CA
W Weight
W Total volumetric feedrate to the burner
x Distance
X Mole fraction
y Distance from the burner
Y Volume fraction
Z Heat blockage parameter
a Heat transfer parameter, stoichiometric coefficient, or
temperature exponent
3 Flame radius parameter
Y Ratio of specific heats
6 Density based on total volume
e Emissivity
v Stoichrometric coefficient
£ Dimensionless coordinate
p Density
a . Stefan-Boltzmann constant
Equivalence ratio (fuel-air)
347
-------
Subscripts
d - droplet
£ - flame
g - gas
H - heterogenous
I - inside
I - liquid
NBP - normal boiling point
0 - outside
ox - oxidizer
p - particle
R - radiation
S - surface
t, T - total
v, VAP - vapor, vaporization, or volatile
6.,, 6H - outer boundary of the mass and heat transfer film
348
-------
REFERENCES
1. Flagan, R. C., "NO Formation from Fuel Bound Nitrogen in.an Oil Fired Burner,"
MIT Thesis, May 1973.
2. Sternling, C. V. and J.O.L. Wendt, "Kinetic Mechanisms Governing the Fate of
Chemically Bound Sulfur and Nitrogen in Combustion," Final Report S-14129,
Shell Development Co., Emeryville, Calif., EPA Contract EHS-U-71-45, Task 14,
August 1972.
3. Martin, G. B. and E. E. Berkau, "An Investigation of the Conversion of Various
Fuel Nitrogen Compounds to Nitrogen Oxides in Oil Combustion," AIChE Symposium
Series, Vol. 68, No. 126, 45(1972).
4. Turner, D. W., R. L. Andrews, and C. W. Siegmund, "Influence of Combustion
Modification and Fuel Nitrogen Content on Nitrogen Oxides Emissions from Fuel
Oil Combustion," Ibid, p. 55 (1972).
5. Bartok, W., V. S. Engleman, R. Goldstein, and E. G. del Valle, "Basic Kinetic
Studies and Modeling of Nitric Oxide Formation in Combustion Processes," Ibid,
p. 30 (1972).
6. American Flame Days Meeting, Chicago, Illinois, July 1972.
7. EPA-65012-73-021, "Proceedings, EPA Coal Combustion Seminar," Research Tri-
angle Park, N. C., June 1973.
8. Symposium on "Control of NO Emissions from Stationary Sources," AIChE 66th
A
Annual Meeting, Philadelphia, Nov. 1973.
9. EPA Contractor's Meeting on NO Kinetics, University of Arizona, Tucson,
Arizona, March 1974.
10. Ball, J. S. et al., "Nitrogen Content of Crude Petroleums," Ind. § Eng. Chem.,
43_, 2577 (1951).
11. Albert, D. K., "Determination of Nitrogen Compound Types and Distribution
in Petroleum by Gas Chromatography with a Coulometric Detector," Anal. Chem.,
39, 1113 (1967).
349
-------
12. Ball, J. S., et al., "Composition of Colorado Shale-Oil Naphtha," Ind 5 Eng.
Chem., 41_, 581 (1949).
13. Ball, J. S. and W. J. Wenger, "How Much Nitrogen in Crudes from'your Area?"
Petroleum Refiner^ April 1958.
14. Ball, J. S. and H. T. Rail, "Nonhydrocarbon Components of a California
Petroleum," Amer. Petr. Inst., 42, Sect. Ill, 128 (1962).
15. Bradenburg, C. F. and D. R. Latham, "Spectroscopic Identification of Basic
Nitrogen Compounds in Wilmingston Petroleum," J. Chem. and Eng. Data, 13,
391 (1968).
16. Brown, D., et al., "Gas-Liquid Chromatographic Separation and Spectrometric
Identification of Nitrogen Bases in Hydrocracked Shale Oil Naptha," Anal.
Chem., 42_, 146 (1970).
17. Cady, W. E. and H. S. Seelig, "Composition of Shale Oil," Ind. and Eng. Chem.,
44_, 2636 (1952).
18. Cook, G. L. and F. M. Church, "Determination of Pyridines by Infrared Spec-
troscopy," Anal .Chem. , 38_, 993 (1956).
19. Dinneen, G. U., et al., "Composition of Crude Shale Oils," Ind. and Eng. Chem.,
44, 2632 (1952).
20. Dinneen, G. U., et al., "Application of Separation Techniques to a High-
Boiling Shale-Oil Distillate,". Anal. Chem., 2_7, 185 (1955).
21. Dinneen, G. U., et al., "Estimation of Types of Nitrogen Compounds in Shale-
Oil Gas Oil," Anal. Chem. , 30_, 2026 (1958).
22. Drushel, H. V., and A. L. Sommers, "Isolation and Identification of Nitrogen
Compounds in Petroleum," Anal. Chem., 58^ 19 :(1966).
23. Fabbro, L. A., et al., "Extension of the Microcoulometric Determination of
Total Bound Nitrogen in Hydrocarbons and Water," Anal. Chem., 43, 1671 (1971).
24. Haines, W. E., "Some Chemical and Physical Techniques for Separation and
Identification of Nitrogen Compounds in Petroleum," Div. of Sci. and Tech.
(Summer Symp), Vol. 42 (VIII), 1962.
350
-------
25. Haines, W. E., et al., "Techniques for Separating and Identifying Nitrogen
Compounds in Petroleum and Shale Oil," Seventh World Petroleum Congress.
26. Haines, W. E., et al., "Development of Analytical Techniques for Heavy Hydro-
carbons and Related Compounds," 26th Mid-Year Meeting American Petroleum
Institute, Division of Refining, San Francisco, May 1971.
27. Heistand, R. N., "Automated Kjeldahl Determination of Nitrogen in Petroleum,"
Anal. Chem., 42_, 903 (1970).
28. Helm, R. V. et al., "Distribution of Nitrogen Compounds in Wilmingston, Calif.,
Petroleum," Ind. and Eng. Chem. and Eng. Data Series, Vol. 2, No. 1, 95 (1957).
29. Helm, R. U., et al., "Identification of Carbazole in Wilmington Petroleum
through Use of Gas-Liquid Chromatography and Spectrescopy," Anal. Chem., 32,
1975 (1960).
30. Hunt, R. H., and M. J. O'Neal, Jr., "The Composition of Petroleum," Advances
in Petroleum Chemistry and Refining, Vol. 10, J. J. McKetta, Editor, John
Wiley § Sons, Inc., New York, 1965.
31. Janssen, A. G., et al., "Isolation and Identification of Pyrrole and 2-Methyl-
pyrrole from Shale Oil," J. Amer. Chem. Soc., 7.5, 4040 (1951).
32. Jewell, D. M. and R. E. Snyder, "Selective Separations of Nonbasic Nitrogen
Compounds from Petroleum by Anion Exchange of Ferric Chloride Complexes,"
- J. Chromatog., 38^, 351 (1968).
33. Jewell, D. M., et al., "Ion-Exchange, Coordination, and Adsorption Chromato-
graphic Separation of Heavy-End Petroleum Distillates," Anal. Chem., 44,
1391 (1972).
34. Jewell, D. M., et al., "Combination of Techniques forthe Characterization of
Residuals," Division Petroleum Chemistry, American Chemical Society Annual
Meeting, New York, August 1972.
35. Kirk, R. E., and D. E. Othmer (Editor), Encyclopedia of Chemical Technology
Interscience Encyclopedia, Inc., New York (1953).
36. Latham, D. R., et al., "Nonbasic Nitrogen Compounds in Petroleum," Hydrocarbon
Analysis, ASTM STP 589, Am. Soc. Testing Materials, p. 385 (1965).
351
-------
37. Lochte, H. L., "Petroleum Acids and Bases," Ind. and Eng. Chem., 44_, 2597
(1952).
38. McDonald, F. R., et al., "Qualitative Applications of the NMR Spectra of
Alkylpyridines: Part 1. Chemical Shifts of the Ring Protons," Appl.
Spectroscopy, 22_, 325 (1968).
39. Ibid: "Part 2. Chemical Shifts of the Protons of the Alkyl Substituents,"
2_2, 329 (1968).
40. McKay, J. F., et al., "The Separation of Acidic Compound Types Isolated
from High-Boiling Petroleum Distillates," Separation Science, 1_, 361
(1972).
41. Morandi, J. R. and H. B. Jensen, "Comparison of Porphyrins from Shale Oil,
Oil Shale, and Petroleum by Absorption and Mass Spectroscopy," J. Chem.
and Eng. Data, 11, 81 (1966).
42. Morris, J. C., et al., "Review Article on Nonmetal Elements and Compounds
in Oils," Anal. Chem., 43., No. 5, 187R (1971).
43. Ode, W. H., "Coal Analysis and Mineral Matter," Chemistry of Coal Utiliza-
tion, Supplementary Volume, H. H. Lowry, Editor, NAS-NRC Committee on
Chemistry of Coal, John Wiley § Sons, INc., New York, 1963.
44. Okuno, I., et al., "Type Analysis of Nitrogen in Petroleum Using Nonaqueous
Potentiometric Titration and Lithium Aluminum Hydride Reaction," Anal. Chem.,
57_, 54 (1965).
45. Petersen, J. C., et al., "Molecular Interactions of Asphalt, Tentative
Identification of 2-Quinolones in Asphalt and their Interaction with
Carbonxylic Acids Present," Anal. Chem., 43, 1491 (1971).
46. Poulson, R. E., "Stationary Phases for Separation of Basic and Nonbasic
Nitrogen Compounds or Hydrocarbons by Gas-Liquid Chromatography,"
J. Chromatographic Science, 7_, 152 (1969).
47. Richter, F. P., et al., "Distribution of Nitrogen in Petroleum According to
Basicity," Ind. and Eng. Chem., 44, 2601 (1952).
48. Sauer, R. W., et al., "Nitrogen Compounds in Domestic Heating Oil Distillates,"
Ind. and Eng. Chem., 44, 2606 (1952).
352
-------
49. Smith, J. R., et al., "Separation of Nitrogen Compounds by Adsorption from
Shale Oil," Anal. Chem., 22, 867 (1950).
50. Snyder, L. R. and B. E. Buell, "Characterization and Routine Determination
of Certain Non-Basic Nitrogen Types in High Boiling Petroleum Distillates
by Means of Linear Elution Adsorption Chromatography," Anal. Chim. Acta,
33, 285 (1965).
51. Snyder, L. R., B. E. Buell, and H. E. Howard, "Nitrogen and Oxygen Compound
Types in Petroleum - Total Analysis of a 700-850 F Distillate from a
California Crude," Anal. Chem., 40_, p. 1303 (1968).
52. Snyder, L. R., "Nitrogen and Oxygen Compound Types in Petroleum-Total
Analysis of a 400-700 F Distillate from a California Crude Oil," Anal.
Chem.. 41_, 314 (1969).
53. Van Meter, R. A.,.et al., "Oxygen and Nitrogen Compounds in Shale-Oil Naptha,"
Anal. Chem.. 214, 1758 (1952).
54. Ward, L. F., et al., "Nitrogen Compounds in Distillate Fuels," Anal. Chem.,
25_, 1070 (1953).
55. Wenger, W. J. and J. S. Ball, "Characteristics of Crude Oils from Utah,"
Paper No. 42, Bulletin 54 - Oil and Gas Possibilities of Utah, Re-evaluated,
Utah Geological and Mineralogical Survey, 1963.
56. Nenner, I and G. S. Schule,"Temporary Negative Ions and Election Affinities
of Benzene and N-Heterocyclic Molecules: Pyridine, Pyredazine, Pyrimidine,
Pyrazine, and S-Triazine," J. Chem. Phys., 62, 1747 (1975).
57. Karr, E. Jr.,"chemistry of Coal Utilization: Supplementary Volume," W. H. Ode,
editor, Wiley, New York, pp. 545-554.
58. Martin, G. B. and E. E. Berkau, "Evaluation of Various Combustion Modifica-
tion Techniques for Control of Thermal and Fuel-Related Nitrogen Oxide
Emissions," 14th Combustion Symposium International, 1972.
59. Heap, M. P., T. M. Lowes and R. Walmsley, "Emission of NO from Large Turbulent
Diffusion Flames" 14th Combustion Symposium (International), 1972, p. 883.
353
-------
60. Heap, M. P., T. M. Lowes and R. Walmsley, "NO Formation in Pulverized Coal
Flames," European Combustion Symposium, 1973, p. 493.
61. Shaw, J. T. and A. C. Thomas, "Oxides of Nitrogen in Relation to Combustion
in Coal" 7th Inter Conf on Coal Science, Prague, 1968.
62. Bowman, C. T., "Kinetics of NO Formation in Combustion Processes" 14th
Combustion Symposium (International), 1972, p. 729.
63. DeSoete, G., "The Mechanism of Formation of NO from HNj and Amines in
Hydrocarbon Flames," Rev. Inst. Fr. du Petr. Ann. Combustion Ljquide, 28 (_!_)_,
96(1973).
64. UeSoete, G., "III-Formation of NO in Flames, Formation of NO from Fuel
Nitrogen, B-Cyanogen," Preprint-B., I.P.P.-21.309.
65. DeSoete, G. and A. Querand, "Formation of NO from Fuel Nitrogen C-Added NO,"
Preprint, Inst. Fr. du Petrole, 1973.
66. Fenimore, C. P., "Formation of NO from Fuel Nitrogen in Ethylene Flames,"
Comb.and Flame, 19, 289 (1972).
67. Sawyer, R. F., W. J. McLean, and C. M. Maes, "The Production of NO in NHS
Oxidation Flames Ammonia Oxidation Flames," Comb. Sci. and Tech, 6_, 47 (1972),
68. Fine, D. H., S. M. Slater, A. F. Sarofim, and G. C. Williams, "Nitrogen in
Coal as a Source of NO Emission from Furnaces, Fuel, 53, 120 (1974).
69. Kaskan, W. E. and D. E. Hughes, "Mechanism of Decay of Ammonia in Flame Gases
From an NH3/20 Flame," Comb, and Flame, 20, 381 (1973).
70. Merryman, E. L. and A. Levy, "N02 Formation," Proc Third Inter Clean Air
Congress, Dusseldorf, West Germany, Oct. 1973.
71. Johnson, C. T., Jr., "The Conversion of Fuel Bound Nitrogen to NO in Low
Volatile Pulverized Coal Flames," MIT Thesis, 1973.
72. Slater, S. M., "NO Formation in the Combustion of Nitrogen-Containing Hydro-
carbons," MIT Thesis, 1973.
73. Sarofim, A. F., G. C. Williams, M. Modell, and S. M. Slater, "Conversion of
Fuel Nitrogen to NO in Premixed and Diffusion Flames," AICHE 66th Annual
Meeting, Philadelphia November 1973.
354
-------
74. Hazard, H. R., "Conversion of Fuel Nitrogen to NO in a Compact Combustor,"
A.
J. Eng. Power, 185, July 1974.
75. Martin, G. B. and E. E. Berkau, "Preliminary Evaluation of Flue Gas Recircu-
lation as a Control Method for Thermal and Fuel Related Nitric Oxide
Emissions," Western States Sect Mtg, The Combustion Institute, Oct. 1971.
76. Bartok, A. R. Crawford,E. H. Manny, and G. J. Piegari, "Reduction of Nitrogen
Oxide Emissions from Electric Utility Boilers by Modified Combustion
Operation," American Flame Days Mtg., 1972.
77. Turner, E. W. and C. W. Siegmund, "Staged Combustion and Flue Gas Recycle:
Potential for Minimizing NOX from Fuel Oil Combustion," American Flame Days
Mtg., 1972.
78. Flagan, R. C., F. Pompel and J. P. Appleton, "Atmospheric Pressure Burner
Studies II: NO Formation by Combustion of Nitrogen Containing Fuels,"
Central States Sect Mtg., The Combustion Institute, March 1972.
79. Pershing, D. W., G. B. Martin and E. E. Berkau, "Influence of Design
Variables on the Production of Thermal and Fuel NO from Residual Oil and
Coal Combustion," AICHE 66th Annual Meeting, Philadelphia, November 1973.
80. Armento, W. J. and W. L. Sage, "The Effect of Design and Operation Variables
NOX Formation in Coal Fired Furnaces: Status Report,1' AICHE 66th Annual
Meeting, Philadelphia, November 1973.
81. Pai, R. H., R. E. Sommerlad and R. P. Welden, "NO Emission, and Evaluation
of Test Data and Design," AICHE 66th Annual Meeting, Philadelphia, 'November
1973.
82. Flagan, R. C. and J. P. Appleton, "NO Formation from Fuel Bound Nitrogen
in an Oil Fired Burner," AICHE 66th Annual Meeting, Philadelphia, November
1973.
83. Henry, J. M., C. R. McCann, J. J. Demeter and D. Bienstock, "NOX Control by
Flue Gas Recirculation in Pulverized Coal Combustion," AICHE 66th Annual
Meeting, Philadelphia, November 1973.
355
-------
84. Pershing, D. W., J. W. Brown and E. E. Berkau, "Relationship of Burner Design
to the Control of NOX Emissions through Combustion Modification," EPA Coal
Combustion Seminar, June 1973.
85. Heap, M. P., T. M. Lowes, R. Walmsley and H. Bartelds, "Burner Design
Principles for Minimum NOX Emissions," EPA Coal Combustion Seminar, June
1973.
86. England, C.and J. Houseman, "NOX Reduction Techniques in Pulverized Coal
Combustion," EPA Coal Combustion Seminar, June 1973.
87. McCann, C. R., J. J. Demeter and D. Bienstock, "Preliminary Evaluation of
Combustion Modifications for Control of Pollutant Emissions from Multi-
Burner Coal-Fired Combustion Systems," EPA Coal Combustion Seminar, June
1973.
88. Crawford, A. R., E. H. Manny, and W. Bartok, "NOX Emission Control for Coal-
Fired Utility Boilers," EPA Coal Combustion Seminar, June 1973.
89. Flagan, R. C., S. Galant and J. P. Appleton, "Rate Constrained Partial
Equilibrium Models for the Formation of Nitric Oxide from Organic Fuel Ni-
trogen," C^mb._and_Jlamje_22_, 299-311 (1974).
90. Armento, J. W., "Effects of Design and Operating Variables on NO from
Jx
Coal-Fired Furnaces--Phase II," EPA Report EPA-650/2-74-002b, February 1975.
91. Hurd, C. D., et al., "Pyrolytic Formation of Arenes. I. Survey of General
Principles and Findings," J. Amer. Chem. Soc, 84, 4509 (1962)
92. Hurd, C. D. and J. I. Simon, "Pyrolytic Formation of Arenes. III. Pyrolysis
of Pyridine, Picolines and Methylpyrazine," Ibid, 4519 (1962).
93. Johns, I. B., et al., "Thermal Stability of Spme Organic Compounds,"
J. Chem and Eng. Data,- ]_, 277 (1962). i
94. Hirsch, S. S. and M. R. Lilyquist, "Thermal Stability of Model Compounds
Related to Thermostable Polymers by Pyrolysis-Gas Chromatography,"
J. Appl. Polymer, Sci., 11, 305 (1967).
356
-------
95. Jennings- E. C. and K. P. Dimick, "Gas Chromatography of Pyrolytic Products
of Purines and Pyrimidines," Anal. Chem., 34, 1543 (1962).
96. Jacobson, I. A., Jr. et al., "Thermal Reactions of Organic Nitrogen Compounds
I. 1-Methylpyrrole," J. Rhys.. Chem, 6^, 1563 (1958).
97. Jacobson, I. A. Jr. and H. B. Jensen, "Thermal Reactions of Organic Nitrogen
Compounds. II. 1-N-Butylpyrrole," J. Phys. Chem, 66, (1962).
98. ibid, "III. 1-Isopropylpyrrole," J. Phys. Chem, 68^ 3068 (1964).
99. Smith, E. B. and H. B. Jensen, "Autoxidation of Three 1-Alkylpyrroles,"
J. Org.. Chem. 32_, 3330 (1967)
100. Lovas, F. J. and F. 0. Clark, "Pyrolysis of Ethylamine I. Microwave Spec-
trum and Molecular Constants of Vinylamine," J. Chem. Phys., 62, 1925 (1975).
101. Merryman, E. L. and A. Levy, "Nitrogen Oxide Formation in Flames: The Roles
of N02 and Fuel Nitrogen," 15th Combustion Symposium (International), 1974.
102. Zeldovich, 51. B., "The Oxidation of Nitrogen in Combustion and Explosion,"
ACTA Physicochimica, USSR, XXI, No. 4 577 (1946).
103. Zeldovich, flf B., P. fl. Sadovnikov, and D. A. Frank-Kamenetskii,
"Oxidation of Nitrogen in Combustion," 209 pages, Publishing House of the
Academy of Sciences of USSR, Moscow-Leninsrad, 1947 (Translated by M. Shelef
of the Ford Motor Company).
104. Singer, J. M., et al., "Flame Characteristics Causing Air Pollution: Pro-
duction of Oxides of Nitrogen and CO," Bureau of Mines,' Report of
Investigation 6958. 1966.
105. Leonard, P. A. and A. M. Mellor, "Partial Equilibrium Approximations for
Computing NO Formation Rates," 1974 Spring Meeting, Western States Section,
The Combustion Institute, Pullman, Washington.
106. Takagi, T., et al., "Evaluation of Various Factors Affecting NO Formation
and Effectiveness of Some Emission Control Systems," Fuel, 53, 114 (1974).
107. Bartok, W. and V. S. Engleman, "Definition of the Mechanisms and Kinetics
of the Formation of NOX and Other Pollutants Under Normal and Combustion
Modification Conditions," Progress Report No. 1, EPA Contract NO. 68-02-0224,
1973.
357
-------
108. Fendell, F. E., P. C. Coates and E, B. Smith, AIM Journal, Vol. 6, 1968,
p. 1953.
109. Bracco, F. V., Fourteenth Symposium (International) on Combustion, The
Combustion Institute, Pittsburgh, 1973, pp. 831-42.
110. Ranz, W. E. and W. R. Marshall, Jr., Chemical Engineering Progress, Vol. 48,
1952, pp. 141-46 and 173-180.
HI- Rossini, F. D., K. S. Pitzer, R. L. Arnett, R. M. Braun, and G. C. Pimental,
Selected Values of Physical and Thermodynainic Properties of Hydrocarbons
and Related Compounds, Carnegie Press (1953).
112. Field Study of Emissions from Oil Burners, 1971.
113. Othmer, D. F., P. W. Maurer, C. J. Molinary, and R. C. Kowalski, "Correlat-
ing Vapor Pressures and Other Physical Properties," Industrial and Engineer-
ing Chemistry, Vol. 49, No. 1, January 1957, pp 125-137.
114. Othmer, D. F., and D. Zudkevitch, "Correlating Latent Heats and Entropies
of Vaporization with Temperature," Industrial and Engineering Chemistry,
Vol. 51, No. 6, June 1959, pp. 791-796.
115. von Fredersdorff, C. G. and M. A. Elliot, "Coal Gasification," Chapter 20
of Chemistry of Coal Utilization, Supplementary Volume, (H. H. Lowry,
Editor), Wiley, New York, 1963.
116. Howard, J. B. and R. H. Essenhigh, "Pyrolysis of Coal Particles in Pulverized
Fuel Flames," I£EC Process Design and Development, Vol. 6, No. 1, January
1967, pp 74-84.
117. Dainton, A. D., and Elliot, D. E.: "Researches into Combustion of Coal,"
Seventh World Power Conference, Transactions, 1968, pp 525-546.
118. Essenhigh, R. H., R. Froberg and J. B. Howard, "Combustion Behavior of
Small Particles," Industrial and Engineering Chemistry, Vol. 57, No. 9,
September 1965, pp 33-43 and 65.
119. Smith, I. W., "Kinetics of Combustion of Size-Graded Pulverized Fuels in
the Temperature Range 1200-2270 K," Combustion and Flame, Vol. 17, (1971),
pp. 303-314.
358
-------
120. Axworthy, A. E., Jr, "Chemistry and Kinetics of Fuel Nitrogen Conversion to
Nitric Oxide," 66th Annual AICHE Mts, Philedelphia, Pennsylvania, November
1973.
121. Axworthy, A. E. and M. Schuman, "Investigation of the Mechanism and Chemistry
of Fuel Nitrogen Conversion to Nitrogen Oxides in Combustion," EPA Coal
Combustion Seminar, Triangle Park, North Carolina, June 1973.
122. Levy, A., EPA Contractors Mtg on NOX Kinetics, University of Arizona, Tucson,
March 1974.
123. Benson, S. W., "Thermochemical Kinetics," John Wiley and Sons, New York, 1968.
124. Janz, G. J., "The direct Cyanogenation of Benzene," J. Am. Chem. Soc. 74,
4529 (1952).
125. Ruhemann, S., Braunkohle, 28, 749 (1929).
126. Johnson, G. E., et al., "Hydrogen Cyanide Produced from Coal and Ammonia,"
I§EC, Process Design and Development, 7_, 137 (1968).
127. Robertus, R. J., et al., "Suppression of Thermal and Fuel Nitric Oxide
Production in Pulverized Cool-Fired Furnaces," 1974 Spring Meeting of
the Western States Section, The Combustion Institute, Pullman, Washington.
128. Snyder, R. E. Gulf Research and Development Co., Pittsburg, Pennsylvania
(Personal Communication).
129. Ramachandran, L. V., et al., "Formation of HCN by the Pyrolysis of Nitrogen-
Enriched Cool," Proc. Symp. Chem. Oil Coal, 1969 (Publ. 1972) 499-505
(C.A. 81:66022n).
130. Andersen, J. W. and R. Friedman, "An Accurate Gas Metering System for Laminar
Flow Studies," Rev. Sci. Instr. 2£, 61 (1949).
131. Fristrom, R. M. and A. A. Westenberg, Flame Structure, McGraw-Hill, New York,
1965.
132. Kaskan, W. E. "The Dependence of Flame Temperature on Mass Burning Velocity,"
Sixth Symposium (International) on Combustion, pp. 134-143, New York,
Reinhold Publication Corporation, 1957.
359
-------
133. Diehl, R. C. and R. J. Haren, 1st Quarter 1972 Progress Report, Bureau of
Mines.
134. Merryman, E. L. and A. Levy, "NO Formation in Combustion: Anomalous Early
Jv
NO- Formation," Proceeding of the Third International Clean Air Congress,
p. C117, Dusseldorf, Germany (1973).
135. Allen, J. D., "Probe Samplings of Oxides of Nitrogen from Flames," Comb.
and Flame, 24_, 133 (1975).
136. Davies, P. B. and B. A. Thrush, Trans. Far. Soc. 64, 1836 (1968).
137. Mulvihill, J. N. and L. F. Phillips, 15th Combustion Symposium (International),
Tokyo, 1974. .
138. Brocklehurst, B. and K. R. Jennings, "Reactions of Nitrogen Atoms in the Gas
Phase," Progress In Reaction Kinetics, Vol. 4, G. Porter, Editor, pp 3-36,
Pergamon Press, London, 1967.
139. Standard Methods for the Examination of Water and Wastewater, 13th Edition,
1971.
140. Annual Book of ASTM Standards, "Water; Atmospheric Analysis," Part 23, 1973.
141. Saltzman, B. E., Anal. Chem. 32, 135-137 (1960).
142. Saltzman, B. E., Anal. Chem. 26_, 1949 (1954).
143. Orion Research Instruction Manual, Cyanide Ion Activity Electrode, Form
IM94-06/271, 1972.
144. Hiiro, K. and Y. Yamamoto, "A New Spectrophotometric Determination of Cyanide
by Solvent Extraction," Anal. Letters, 6, 761 (1973).
145. Fine, D., Environmental Science and Technology, 6, 348-350 (1972).
146. Halstead, C. J. et. al., Analyst 97, 55-63 (1972).
147. McAdams, W. H., Heat Transmission, p. 259, McGraw-Hill, New York 1954.
148. Touloukian, Y. S. and D. P. DeWitt, Thermo Physical Properties of Matter,
Vol. 7, "Thermal Radiative Properties-Metallic Elements and Alloys,"
pp. 525-4, IFI/Plenum, New York, New York, 1970.
360
-------
TECHNICAL REPORT DATA
(Please read Instructions on the reverse before completing)
1. REPORT NO.
EPA-600/2-76-039
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
Chemistry of Fuel Nitrogen Conversion to Nitrogen
Oxides in Combustion
5. REPORT DATE
February 1976
6. PERFORMING ORGANIZATION CODE
7. AUTHOR(S)
A.E. Axworthy, G.R. Schneider,
M.D. Shuman, andV.H. Dayan
8. PERFORMING ORGANIZATION REPORT NO
R-9698
9. PERFORMING OPflANIZATION NAME AND ADDRESS
Rocketdyne Division
Rockwell International
6633 Canoga Avenue
Canoga Park, CA 91304
10. PROGRAM ELEMENT NO. 1AB014'
ROAPs 21ADG08/21BCC12'
11. CONTRACT/GRANT NO.
68-02-0635
12. SPONSORING AGENCY NAME AND ADDRESS
EPA, Office of Research and Development
Industrial Environmental Research Laboratory
Research Triangle Park, NC 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final; 6/72-5/75
14. SPONSORING AGENCY CODE
EPA-ORD
15. SUPPLEMENTARY NOTES
EPA project officer for this report is B. Martin, 919/549-8411, Ext 2235.
16. ABSTRACTThe repOrt gjves results of an experimental and analytical investigation of
chemical mechanisms involved in the conversion of fuel nitrogen to NOx in combus-
tion. The pyrolysis of fossil fuels and model fuel nitrogen compounds was investi-
gated, droplet and particle combustion models were developed, and premixed flat-
flame burner experiments were conducted to study the conversion of HCN and NH3 to
NOx in low-pressure CH4-O2-Ar flames. Decomposition rates and products were
measured in helium from 850 to 1100C for pyridine, benzonitrile, quinoline, and
pyrrole; products were measured for six No. 6 fuel oils, one crude oil, and two
coals. HCN was the major nitrogen-containing pyrolysis product: the amount formed
increased with temperature. NH3 was a minor product and little if any N2 was formed
The burner experiments demonstrated that fuel NO forms relatively slowly above the
luminous zone in the same region where CO is oxidized to CO2 or later. Although
HCN and NH3 gave similar yields of NO, the NH3 reacted very early in the flame
front; most of the HCN survived the luminous zone and then reacted slowly. A mech-
anism was proposed in which fuel NO forms via the reaction: O + NCO = NO + CO.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lDENTIFIERS/OPEN ENDED TERMS
c. COSATi Field/Group
Air Pollution
Chemical Reactions
Combustion
Nitrogen
Conversion
Nitrogen Oxides
Pyrolysis
Fossil Fuels
Hydrogen Cyanide
Ammonia
Air Pollution Control
Fuel Nitrogen
13B
07D
2 IB
07B
2 ID
B. DISTRIBUTION STATEMENT
Unlimited
19. SECURITY CLASS (This Report)
Unclassified
21. NO. OF PAGES
373
20. SECURITY CLASS (This page)
Unclassified
22. PRICE
EPA Form 2220-1 (9-73)
361
------- |