Skip to main content

Full text of "Spectroscopic investigations on methane air flames containing coal dust"

See other formats








In a work such as this there are many people who are partially 
responsible for its creation due to the assistance and advice rendered 
by them to its author. This acknowledgement can only begin to thank 
them for all their efforts. 

First and foremost, I must thank Dr. Alex E. S. Green for his 
constant support, through varied sources, throughout the course of 
these studies. His constant encouragement and belief in this work helped 
sustain my motivation throughout periods of thick and thin. 

I am also deeply indebted to Dr. Willis B. Person for his untiring 
efforts on my behalf and the many stimulating and informative discussions 
on the topic of applied molecular spectroscopy. His constant optimism 
and friendship turned the darkest night into day. 

I am also grateful to the rest of my committee. Dr. Martin Vala, 
Dr. John Dorsey and Dr. Richard Yost for their advice and attention to 
details in order to bring this study to a timely conclusion. 

My appreciation goes to my typist Joan Raudenbush for a most excellent 
job of typing under the most adverse circumstances and remaining cheerful 
throughout . 

I wish to thank Dr. D. B. Vaidya for assistance in early studies and 
on our publication and for the fine Indian dinners I enjoyed at his 


The constant stimulation and discussions with Dr. Krishna 
Pamidimukkala provided impetus for several experiments and provided 
an excellent sounding board for many of the ideas presented in this 
study and also for the use of his temperatiore data. 

I would like to thank Jeffer^' Samuels for his assistance in 
programming the calculations used in some of these studies and his 
valuable help in the laboratory. 

Enough can not be said about the assistance provided by Millie 
Neal of the Chemistry Library who always cheerfully renewed books 
over extended periods of time. 

Dorothy Lisca of the Physics Library was also most helpful which 
is greatly appreciated. 

I would like to thank Art Grant, Chester Eastman, and Daily Burch 
of the Chemistry Department Machine Shop for teaching me about Machine 
Shop practices and drawing figures and Harvey Nachtrieb and the Physics 
Department Machine Shop for turning these drawings into useable pieces 
of apparatus. 

Rudy Strohschein of the Chemistry Department Glass Shop made many 
valuable contributions by constructing all of the glass apparatus used 
in these studies. Also thanks to Bruce Green for his suggestions on 
coal dust feeders. 

I would also like to thank Dr. James D. Winefordner for loaning some 
of the equipment used in this study. Finally the independent gas-coal 
studies of Dr. Alex E. S. Green and Bruce Green provided access to supplies 
and materials otherwise unattainable. 









Structure and Mechanisms 4 

General Description 4 

Structure of Pre-Mixed Laminar Methane-Air Flames — 6 
Calculation of Physical Properties of Pre-Mixed Laminar 

Methane Air Flames 13 

Thermal and Diffusion Theories of Flame Propagation- 14 

Comprehensive Theories of Flame Structure 25 

Basic equations 25 

One-dimensional flame equations 30 

Boxmdary conditions for flame equations 33 

Modem theoretical treatments 37 

Chemical Reactions Occ\irring in Methane-Air Flames 43 

Elementary Reaction Mechanism 43 

Production of Excited State Molecules 47 

OH 48 

CH 49 

C2 51 

CO 55 



Chemical Analysis of Coal 56 

Coal Petrography 66 

Molecular Structxire of Coal 74 

Combustion of Pulverized Coal Particles in Laminar 

Pre-Mixed Flames 85 


Generation and Combustion of Volatile Matter 87 

Combustion of Solid Devolatilized Coal Particles 

(Char) 101 

Chemical Reaction Mechanisms for Coal Char 

Combustion 114 

Mathematical Descriptions of Coal Particle 

Combustion 124 


Gas Metering and Flow Control System 136 

Burner 139 

Laboratory Coal Delivery Systems 140 

Spectroscopic Apparatus 154 

General Considerations 160 

V. DATA 162 

Development of a Temperature Measurement Technique and 
Observations of the Effects of Varying Amoiants of 
Coal on OH Emission 162 

Survey Spectra of Methane-Air an.d Methane-Air-Coal 
Dust Flames 170 

Measurement of Excited State Populations in Methane- 
Air and Methane-Air-Coal Dust Flames 183 

Spatial Temperature Measurements of Methane-Air and 
Methane-Air Coal Dust Flames 212 

Investigations for Sulfur and Sulfur Compoxmds in 
Methane- Air-Coal Dust Flames 212 

Effects of Addition of Excess O2 on Methane-Air-Coal 
Dust Flames 218 

Fluorescence Experiments 219 

Lateral OH Emission Profiles of Methane-Air and 
Methane-Air-Coal Dust Flames Using an Inversion 
Technique 220 

OH, CH and C2 Emissions from a Single Cone Using an 
Inversion Technique 222 

Calculation of Ground State Concentrations Assuming a 
Thermal Distribution 238 


Methane -Air Flames 243 

Effect of ;j! Upon Excited State Concentrations 243 

Spatial Temperature Profiles and Lateral OH 

Emission Profiles 249 

Single Cone Data and the Formation of Excited States 253 


Methane-Air-Coal Dust Flames 257 

Stoichiometric ((j) = 1) Flame 258 

Lean ((}) = 0.77) and Rich ((() = 1.1) Flames 270 

Chemical and Physical Processes in Methane-Air- 
Coal Dust Flames 274 

Comparison with Computer Kinetic Code Simulation 

Data 282 

Conclusions 285 

Future Work 287 






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



John J. Horvath 
April, 1983 

Chairman: Alex E. S. Green 
Cochairman: Willis B. Person 
Major Department: Chemistry 

Recently the possibility has been proposed that a natural gas-coal 
mixture can be used to replace oil in an existing oil boiler. In order 
to study the physical and chemical effects of combining methane and 
coal dust in a flame a spectroscopic investigation of methane-air and 
methane-air-coal dust flames was initiated. A laboratory burner with 
suitable flow controls, was adapted for the simultaneous combustion of 
methane-air-coal dust mixtures. Pulverized bituminous coal dust 
('\>200 mesh) was introduced into this burner by means of a f luidized bed 
device which allowed for a steady and reproducible amount of coal dust 
to be input to the flame. 

Spectral snissions from radical species OH, CH, and C^ were measured 
along with CO emissions which were then used to calculate excited state 
species number densities. Methane-air flames of various equivalence 
ratios were studied along with stoichiometric, lean and rich methane- 
air-coal dust flames. Emission intensities were obtained as a function 

vi i 


of height in the various flames and differences between the methane- 
air and methane-air-coal dust flames are discussed. Spatial temperature 
measurements of these flames were also obtained. 

Spectral emission profiles of the primary reaction zone for the 
radicals OH, CH and were obtained and results are discussed in 
relation to excited state formation mechanisms. Current ideas on the 
chemical and physical processes occurring in methane-air flames are 
reviewed along with current knowledge of coal and its combustion. Data 
from the present study are compared to literature data and chemical and 
physical mechanisms are proposed to account for these observations. 



Recently the possibility has been proposed that a natural gas-coal 
mixture can be used to replace oil in an exciting oil boiler (1) . This 
study pointed out that great monetary savings could be yielded as well 
as decreasing the country's dependence on foreign sources for a steady 
and ever increasingly costly supply of oil. This study also pointed 
out that this idea, the burning of coal in the presence of natural gas, 
had not been discussed in the literature up to this time. An extensive 
literature survey revealed that ample literature could be found on the 
combustion of methane gas and solid coal, but not a single study 
involving the combined combustion of the two. In advance of the above 
study, an experimental program had been initiated by Professor Green. 
The work presented in this thesis represents the first spectroscopic 
investigation into the physical and chemical properties of combined 
combustion of methane-coal dust in a laboratory flame. 

The chemical and physical process occurring during combustion of 
methane flames have been well studied, and some basic understandings 
have been realized (2-5). From spectroscopic observations of methane 
flames an abundant amount of information can be obtained. For example, 
chemical species in the flames can be identified by their characteristic 
spectral emissions and their growth and decay can be followed by 



monitoring intensity as a function of position in the flame. The 
rates of change and the appearance and disappearance of chemical 
species yield valuable information on the chemical reactions occurring 
during the combustion process. Spectroscopic measurements are also 
non-perturbing upon the process being measured which allows measurements 
to be made in places inaccessible by other means of analysis. For 
these reasons and the availability of excellent spectroscopic facilities 
it was decided to study the spectroscopic enission of methane-air and 
methane-air-coal dust flames to try to understand the processes 
occurring in a methane-air-coal dust flame. 

Questions to be addressed in this study include 

1. Dependence of spectral emissions upon flame stoichiometry and 
their significance on flame chemistry. 

2. Do methane spectral emissions change upon addition of coal 
dust to the flame and are any new emissions produced? 

3. How does coal affect the chenistry of the flame? 

In Chapter II the general physical and chemical properties of a 
methane-air flame will be presented. A discussion of flame propagation 
theory follows and the methods used to theoretically calculate various 
properties of methane-air flames. Elementary reactions occurring in 
methane-air flames will also be presented along with a section 
discussing current ideas of excited state production through chemical 
reactions for the species investigated in this study. 

Chapter III will present chenical and physical properties of coal 
relevant to this study. Basic chemical composition of coal is discussed 
as well as techniques used to obtain these values. Different processes 
occurring during the combustion of coal, i.e. devolatilization. 


char combustion, etc., will be presented and discussed along with theoretical 
approaches to modeling coal combustion systems. 

Chapter IV will present the equipment and apparatus used in these 
studies as well as the experimental methodology used to obtain data. 
Chapter V will present the results of these scientific investigations and 
Chapter VI will discuss their significance and conclusions that can be 
drawn from these studies. The appendix contains the Fortran computer 
programs used to process data and calculate some of the results presented 
in this study. 


Structures and Mechanisms 

General Description 

A flame can be said to be a collection of gaseous matter at 
high temperature, produced by highly exothermic chemical reactions, 
which results in the emission of light. Two major classes of flames 
exist, propagating flames characterized by explosions and stationary 
flames represented by the common Bunsen burner. Both flame types 
require a fuel (ccmbustible gas or liquid) and an oxidant (usually 
air or oxygen) . Stationary flames can be considered to be of two 
basic types. In the first case the fuel is burned as it comes into 
contact with the air. In small flames of this type the combustion 
processes are mainly determined by the rate of diffusion of the air 
into the fuel and these types are called diffusion flames. In larger 
flames of these types, mixing due to turbulence becomes larger than 
mixing by diffusion, and flame stability and size are mainly governed 
by aerodynamic effects. These large turbulent flames have been 
studied in depth by many authors (6) . 

The second type of stationary flame is that in which the fuel and 
air or oxygen arepre*mixed before combustion occurs. These are known 
as pre-mixed flames, and in these flames the combustion processes are 
mainly determined by the rates of the chemical reactions occurring 



in the flame. A Bunsen burner is the simplest type of burner that 

burns a pre-mixed flame. In this bxirner the fuel enters from a nozzle 

shaped orifice about 1 mm in diameter at the base of the burner. The 

mixing air is then entrained through two adjustable air holes. Even 

if the air holes are fully open, the amount of air entrained is 

usually much less than the amount needed for complete ccmtustion 

(i.e. less than the stoichiometric amount), yielding a highly luminous 

and unsteady flame. The amount of air entrainment occurring is dependent 

upon the size of the gas orifice, gas pressure and size of the air 

holes. Relationships between air entrainment and flow conditions in 

Bunsen burners have been thoroughly discussed by Lewis and Grumer (7) 

and Lewis and von Elbe (8, 9). The chief limitations of the Bunsen 

bvurner are its inadequate entrainment of air, resulting in a fuel rich 

flame, the tendency to flash back with large diameter burners, putting 

an immediate halt to any observations currently in progress and being 

too unsteady to be suitable for detailed flame studies. One way to 

designate the fuel to air ratio is by the mixing ratio which is defined 

as the mole ratio of the oxidant gas (air, oxygen, etc.) to the fuel 

gas. The stoichicmetric mixing ratio will be a function of the fuel 

and oxidant used in a flame. In the hydrogen-oxygen flame where the 

overall chemical reaction is 2H„ + 0„->2H O the mixing ratio for a 

2 2 2 

stoichicanetric flame is 0.5 while for a hydrogen-air flame the 
stoichiometric mixing ratio is 2.5 (air = 20% O^) . One problem with 
this type of designation is not being able to specify flame conditions 
(lean, rich, stoichiometric) without knowing the stoichicxnetric . . : 


mixing ratio , which is dependent on overall flame chemistry. A more 
convenient and informative term to use is the equivalence ratio vdiich 
is defined as the actual mixing ratio divided by the stoichiometric 
mixing ratio. Therefore a stoichicanetric flame will have an equivalence 
ratio of 1.0, a rich flame >1.0 and a lean flame <1.0. Equivalence 
ratios will be used throughout this dissertation when discussing fuel/ 
oxidant ratios. 

A very steady, homogeneous and laminar flame is obtained when using 
a grid, or Meker burner with pre-mixed gases. The grid consists of 
a large number of finall circular holes, as opposed to a single large 
one in a Bansen burner. The gases are premixed before entering the 
burner so precise fuel-air ratios can be obtained, frc«n lean though rich 
conditions, whereas a Bunsen burner can only run rich flames. This is 
the type burner used in these studies, which is fully described in 
Chapter IV, using a pre-mixed volume of natural gas and air. Natural 
gas is predominantly methane (CH^) with approximately 3 percent occurring 
as higher hydrocarbon gases. Gas analysis of the natural gas used in 
these experiments is reported in Chapter VI and in this Chapter methane 
and natural gas will be used interchangeably, but methane will be used 
when discussing the chemistry of the flame. 
Structure of Pre-Mixed Laminar Methane-Air Flames 

In the flame used in the present study five distinct regions can 
be distinguished. 

1. The primary reaction zone consisting of cones, a few mm in 
height, above the holes, a very luminous region where initial 
combustion occurs. 

2. The dark zone, a preheating zone underneath the cones. 


3. A secondary reaction zone fed by oxygen from the surrounding 
room air. 

4. An equilibrium zone, which is the most homogeneous part of 
the flame both in concentrations of species and temperature. 

5. A diffusive mixing region where the flame mixes with the room 
air (mainly N^) and serves as a working flame boundary. 

The primary combustion zone is also called the inner cone since it 

has a conical form vAien the burner port is round, as in the present 

case. This combustion zone is where the bulk of the reaction products 

are formed and is about .05 to 1.5 mm thick at atmospheric pressure 

with laminar flow and the gas takes about 1-20 ms to pass through this 

region. The primary reaction zone occurs on a surface of a cone due to 

the variation of gas velocity through the input ports. The flame gas 

flow in the input ports assumes a parabolic velocity profile as defined 

by the Poiseville equation for laminar flow (10) . A typical flow 

velocity pattern inside a burner port is shown in Figure 1. Figure 2 

shows an idealized flame cone, which is assumed to be conical. The 

dotted line represents the surface inside the cone where the gas 

tanperature rises just above the initial temperature due to conduction 

frcxn the reaction zone. The solid line is the luminous surface of the 

flame of area a^, which can be visually observed. Assuming the burning 

velocity is constant over the entire surface and there are no heat losses 

to the burner, burning velocity can be given by the following equations. 

Total mass is conserved, yielding 

m = DVa, = pVa 
o o b o b o 


V a^ 

V = Ji— = -2^ 

b p a a 
o o o 








400 — 




40 , 


o o 




I I I — r 

I — I — I — I I I — r— r--r — I — r-^ 
6 4 2 0 2 4 6*^ 

Radius (mm) 

Figure 1. Experimental velocity distribution in a laminar stream 
of air at the exit of a cylindrical tube of 12.98 mm 
diameter with a flow of 241 cm /s (Adapted from B. Lewis, 
G. Von Elbe, J. Chem. Phys. 11, 75 (1943)). 


Figure 2. Ideal Flame Cone 


where ifi = rate of mass flow per unit time, = velocity of unburned 

flame gases, = unburned gas density at initial conditions, a^ = area 

of surface where gas temperature begins to increase, = cross-sectional 

area of burner tube and V = burning velocity. In terms of the cone 


apex angle, a^, 

V = V sin a 
bo o 

Burning velocity can then be simply calculated by measuring the cone 
apex angle emd measuring the input flow rates and calculating the input 
flow velocity. The simplicity of these equations in visualizing cone 
formation and burning velocity is unfortunately negated by the fact 
that the cone is not perfectly conical and the burning velocity is not 
constant over the whole cone. Distortion of a perfect cone shape is 
caused by the pressure inside the cone being greater than ambient, 
as a consequence of the pressure difference generated by the flame 
(11, 12). Particle tracking experiments (9, 11) have shown that burning 
velocity appears to increase at the flame tip and to decrease at the 
base. The latter is due to the quenching and stabilizing effects (heat 
loss to burner) of the burner port rim. In the former case, the 
unburned gas approaching the flame tip is surrounded by hot gas and 
the burning velocity is increased by heat conduction. 

As the unburned flame gases rise, their temperature will increase 
slowly by conduction until it reaches a point, T^ the ignition point, 
where exothermic reactions just begin to be significant. From this 
point on the reaction proceeds extremely rapidly with an exponential 
growth in temperature until final maximum flame temperature is reached. 


This rapid temperature rise is due to the rapid rate of the chemical 

chain reactions occurring in the primary reaction zone. In the 

reaction zone many molecules and radicals are formed with a large 

percentage in excited states, accounting for the highly luminous 

nature of the reaction , zone. Due to the large release of energy in 

the reaction zone, the states of the gases are far from equilibrium since 

the liberated energy has not had a sufficient amount of time to 

partition equally among all the energy modes of the reaction products. 

The radiation and composition of the reaction zone have been subjects 

of intensive study (4, 5). 

Above the primary reaction zone is a transition zone which leads 

to the zone of equilibriiim. Many experiments (5) have indicated a 

narrow zone above the cones where high energy distributions decay to 

a distribution representative of a thermal distribution at the flame 

temperature. This decay rate depends on the number of effective 

collisions per second and how soon the different forms of energy will 

reach a state of equilibrium. Almost instantly the translational energy 

is equally distributed amont the species present. The populations of 

the rotational, electronic and vibrational levels follow in that order. 

Once out of the reaction zone these forms of energy reach equilibrium 

rapidly and are able to adjust their populations quickly to reflect 

changes occurring in the flame as other species emit or absorb energy 

yielding temperature changes. In this way, height dependent phenomena, 

such as reduction in molecular emission at greater height (lower 

temperatures) can be explained. Due to the high collisional rates in 

9 -1 

flames (approximately 10 sec ) and the relatively small number of 





collisiona (13) needed to redistribute the nonthermal population 
into a thermal one, equilibrium conditions are reached soon after 
leaving the primary combustion zone (14) . In this equilibrium region 
the temperatures determined by translational , electronic, vibrational 
and rotational population distributions are all the same, as has 
been demonstrated by many authors (14, 15). 

It is possible that some of the flame reactants will not be 
fully combusted upon exiting the primary combustion zone. This effect 
is enhanced when the initial gas mixture is deficient in oxygen (fuel 
rich) . Then the burned gases will contain a non-negligible concentra- 
tion of molecules such as CO which can be further oxidized by the 
oxygen contained in the surrounding room air. This results in a 
secondary zone with reactions occurring which are similar to those in 
the primciry reaction. The degree to which this occurs is determined 
by the initial mixing ratio of fuel gas and oxygen and the availability 
of oxyten in the surrounding atmosphere. In extreme cases (greatly 
fuel rich) the outer secondary reaction zone can be hotter than the 
primary reaction due to this effect. In the fuel lean and stoichiometric 
flames this effect will be negligible or non existent. 

As the burned flame gases rise their temperature decreasea and the 
upper parts of the flame are less stable than the lower parts. There 
are several reasons why this occurs. The maximum heat is produced in 
the primary reaction zone where all of the chemical reactions occur 
under normal conditions. As these gases rise from the primary reaction 
zone they are continuously losing heat through radiation, mostly from 
the infrared bands of HO and CO . The flame gases also mix diffusively 


with the surrounding cold room air, mainly nitrogen which cools the 

flame by absorbing a great amount of heat, and because of the 

horizontal velocity gradient present turbulence occurs after a short 

time. These effects cause a flame to "disappear" after a fixed time 

period, resulting in a finite flame length. 

Calculation of Physical Properties of Pre-Mixed Laminar Methane 

Air Flames 

During the past years a considerable amount of work has been 
devoted to measurements of the internal structure of pre-mixed flames. 
There are many reasons why this has occurred. These investigations 
help to increase our under steinding of the mechanism of flame propagation 
and the physical processes occurring during combustion. A flame also 
represents a high temperature , flowing reaction system in which rapid 
gas phase chemical reactions may be studied. In a flame, the time 
scale of normal reaction rate studies is replaced by a distance coordinate 
and flow velocity. 

In a pre-mixed flame a cold combustible mixture, upon passing 
through the flame front (primary reaction zone) , is converted into hot 
combustion products. The analysis of flame structure consists of 
measuring or calculating 

1. The burning velocity, or rate of consumption of gas per unit, 
are of flame front and the thickness of this zone. 

2. "The variation of the flame temperature as a function of height 
in the flame. 

3. The composition and concentration of all flame species. 
Numerous theories have been developed in an attempt to calculate the 
flame parameters in question, and these can be divided into three major 
categories : 


1. Thermal theories. 

2. Diffusion theories. 

3. Comprehensive theories, involving hydrodynamic and chemical 
kinetic equations. 

Thermal and Diffusion Theories of Flame Propagation 

The earliest theoretical approach to flame analysis was the thermal 

theory of Mallard and Le Chatelier (16) in 1883, who proposed that it 

is propagation of heat back through layers of gas that is the controlling 

mechanism in flame propagation. It was assumed that normal heat transfer 

processes raised the reactants to the spontaneous ignition temperature 

and that this propagated the flame. They assumed that the flame was 

divided into two regions* a region of conduction and a region of burning, 

separated by an ignition point with temperature T^. Initial room 

temperature gases at T^ are burned to obtain reaction products at a 

final temperature T^. They assumed that the slope of the temperature 

curve was linear and that it could be approximated by the expression 

[ (T^-T^)/£^] where 6^ is the thickness of the reaction zone. The enthalpy 

balance equation is then given by 

mC (T.-T ) = X(T^-T.)/6 (II-l-l) 
p 1 o f 1 r 

where X is the thermal conductivity and ih is the mass flow rate, and C 


is the heat capacity. . The mass flow rate is given by 

m = pAV = pV, A (II-1-2) 

o b 

where p = density, A = cross sectional area which is set equal to unity, V 

velocity of unburned gas, and Vj^ = the laminar flame speed. Because 

unburned gases enter normal to the reaction zone, by definition V = V. . 

o b 


The above equation (II-l-l) can be written as 

pV C (T.-T ) = X(T -T.)/6^ (II-1-3) 
bpio fi r 


(T -T.) . 

V _ ^ f 1 1 (II-1-4) 

b pC (T.-T ) 6 

p ID a 

which is the expression for the flame speed obtained by Mallard and 

Le Chatelier. In this equation the reaction zone thickness, 6^, is 

not Icnown and would have to be estimated to obtain the burning velocity, 

This problem was overcane by the work of Nusselt (17) who introduced 

the velocity of the chemical reaction into the flame velocity equation. 

Nusselt used a term e to represent the concentration of the products, 

so that the following conditions apply: 

At x=x.,T=T.,e=0 
1 1 

At X = x^, T = T^, e = 

where x is linear distance in the flame in the upward direction and x^ 
is the height of the ignition point. Assuming that the degree to 
which the reaction has proceeded is linear in x we obtain 


de 1 _ 

dx ~ (x .-X . ) 6 
f 1 r 

so that 

^ ^(T^T^^ 

dx 6 f i'^ 1^ dx 


In this way the unknown thickness of the reaction zone, 6^, is 
eliminated but it now becomes necessary to find an expression for the 
extent of the reaction with respect to x. Nusselt assumed that the 
reaction rate, 03, is constant and by earlier assumptions the 


velocity of the gases in the reaction zone, V^, is also constant since 

— = (de/dt) (dt/dx) = u)/V^ (II-1-7) 
dx b 

His next assumption was that if the number of moles does not change, the 

burning velocity can be expressed as a function of temperature as 

follows- V = TV /T which contradicts his previous assumption which 
b o o 

states that V is constant. Combining equations he obtained his final 

equation for flame velocity: 
(T -T ) T 

V 2 = J: f i _o JiL (II-1-8) 

b C p (T.-T ) T e. 

P 1 o f 1 

This is similar to expressions derived by Jouguet and Crussard (18) 

and Daniell (19) also containing false assimptions. 

Damkflhler (20) derived an equation for flame burning velocity 

which resembles the equations of Nusselt, Jouguet and Crussard, and 

Daniell, but modified the assumptions so that the theory is more 

realistic in representing physical processes. He makes the approximation 

that the temperature gradient at the ignition point is related to the 

average temperature gradient across the reaction zone by a constant 

F. Then in place of Equation II-1-6, he wrote 

(dT/dx) = (T -T.)F/6 (I 1-1-9) 

' x=x. fir 

Instead of assigning a constcuit reaction velocity, oi, as Nusselt, he 
defines a mean reaction rate o) by the equation 


n V = / ^ wdx = u6 (II-l-lO) 
r b r 

where n = number of reactant molecules per unit volvime initially 

present. This eliminates the contradiction of constant burning velocity 



impose by Nusselt's constant mole assumption. Damkohler's final equation 

is then written in the form: 

X(T -T.)F/6 = C pV, (T.-T ) (II- 1-11) 

fir p b 1 o 


F^uCT -T.) 

\ = C pn (T.-T ) ^^^-^-^2) 
p r 1 o 

This equation shows the general feature that the square of the biorning 
velocity will be proportional to the average reaction rate in the reaction 
zone, bi. Equation II-l-ll can be solved for 6^ which shows that the 
thickness is inversely proportional to the density (i.e. pressure), 
turning velocity, and the thermal conductivity. 

Later attempts to put the thermal theory of flame propagation 
into a more quantitative form were made by Zeldovich and Frank- 
Kamemetsky (21, 22), Zeldovich and Semenoff (23, 24), Semenoff (25) and 
Zeldovich (26) who assumed that the reaction only started at a point near 
the final flame temperature. In this approach it was assumed that there 
was a temperature T^ very near to T^ below which there is practically 
no reaction; T^ is basically an ignition point but is much higher than 
in earlier theories, and is not used as a physically significant constant 
but as a mathematical device for approximate calculations. T. does not 
appear in the final equations. This theory also assumes a reaction 
model of form nA = B+C for which the reaction order may be zero,, 

first or second with respect to A. Intermediate reactions or species 
are not considered. It is also assvmied that the rates of reaction 

could be described using classical kinetic theory, taking the form 

— E/RT 2 — E/RT 

03 = ka e for a first order reaction or to = ka e for a second 



order reaction where a = molecules of reactant/cm ,k = frequency 


factor, E is the energy of activation, and R = gas constant. With 
a few additional assumptions they were able to derive equations for 
the burning velocity for first order reactions. 

k2X^C . T 

V i-^ ^ - 

b ,2 T. _f 




and for second order reactions 
^ 2 


- k2X (C ) a T 
2 f P r _o 

PfL f 

2 2 

'^f ' 








where X^, C and are the thermal conductivity, heat capacity and 
f p f 

density at the final flame temperature, a^ is the concentration of the 
reactants, L is the heat of cctnbustion of flame gases, D is the diffusion 
coefficient at the final temperature and n^ and n^ are the number of 
molecules of the reactants and products. Semenov (25) considered the 
conditions for which these equations were valid by calculating the errors 
incurred by the theories assumptions. He concluded that for i.ibi-nolecular 
reactions the solution was valid only for RT^/E <_ 0.1. Thus for a 

bimolecular reaction the application of Equation II-1-14 is only valid 

, o 
for values of E greater than 40 kcal , at = 2000 . For unrmolecular 

reactions he found a wider range of applicability. 

Belayeu (27) applied the above theory to the calculation of the 

speed of ccxnbustion of the vapor of nitroglycol. The heat of combustion 

of the vapor phase is conducted to the surface of the liquid and causes 

it to vaporize. The mass rate of vaporization is measured directly by 

the rate of lowering of the meniscus of the liquid and is assumed 

equal to the mass rate of combustion of the vapor. It is assumed 


that the activated molecule of nitroglycol decomposes by itself and 

the reaction occurs on collision with any other molecule. The 

equation for unimolecular reactions (Eq. II-1-13) was used, except 

the the frequency factor k, instead of having the first order value of 

about 10 , was replaced by Zn^, where Z is a collision parameter 

yielding the number of collisions/sec cm"^. Belayeu found that the 

calculated value was in quite good agreement with the experimental 

-2 2 

value of about 4 x 10 gra/sec cm . These equations were found not to 
hold for hydrogen and hydrocarbon flames but Zeldovich (26) found 
reasonably good agreement for carbon monoxide flames. 

Bartholom* (28, 29) investigated the combustion of hydrogen, hydro- 
carbons, alcohols, ethers, nitroparaf f ins and alkyl nitrates with air, 
oxygen, nitrous oxide, and their mixtures in order to determine 
experimentally the variables that have the greatest effect upon flame 
velocity. Out of these experiments he developed a thermal theory of 
flame propagation along similar lines and attempted to apply it to 
hydrocarbon flames with limited success. A later development of the 
thermal theory by Boys and Corner (30) did not make as many assumptions 
as previous workers but involved many various constants such as 
activation energies which were not usually available for use with 
their explicit equations and made the theory difficult to test. Murray 
and Hall (31) measured the burning velocity of the decomposition flame 
of hydrazine (N2H^) and found a value that agreed with the burning 
velocity predicted by the thermal theory. These experiments showed 
that the thermal theory could predict burning velocity in a few limited 
cases and also implied that thermal conduction plays and important role 


in flame propagation. Direct experimental evidence is given by the 
effect of preheating the flame gas mixture, which increases the burning 
velocity, and the observations by Spalding (3 2) who used a porous plate 
burner to abstract heat from the flame and obtain a reduction in burning 
velocity. However the ignition point is predicted to be at much higher 
temperatures than those measured by experiments. 

During the development of the thermal theories a contrasting 
school of thought started wtih the basic idea that the propagation 
rate depended on the speed of diffusion and on the concentration of 
active radicals (H, OH, etc.). In a series of papers Tanford and Pease 
developed the proposition that for certain flame reactions the rate 
of diffusion of active centers (radicals) into the unburned gas determines 
the flame velocity. The first paper (33) calculated equilibrium atom 
and free radical concentrations in moist carbon monoxide flames. The 
results of this paper indicated that the equilibrium concentration of 
hydrogen atoms was an important factor in determining the flame velocity. 
Tanford, in a second paper (34), presented calculations to establish 
the relative importance of diffusion and heat conduction in creating 
the hydrogen atoms in the flame zone. The conclusion reached was that 
diffusion was the controlling process. The third paper (35) developed 
an equation for flame velocity based on the conclusions of the first 
two papers. Assumptions used in this development are that the pressure, 
heat capacity and thermal conductivity are constant. The chemical 
reaction term is of form de/dx, with the result that T is the only depen- 
dent variable in the development and consists of two limiting cases: : 


1. All heat is released at the flame front, x = O, the flame 
front being defined as the point where the combustion has 
reached equilibrium, so that de/dx = O except at x = O. 

2. The chemical reaction proceeds evenly across the zone, so 
that de/dx is a constant. 

In order to determine the value of de/dx under assumption one, it 

becomes necessary to assume a flame zone thickness for calculational 

purposes. Thicknesses of 0.01 cm for the carbon monoxide flame and 

of 0.005 cm for the hydrogen flame were used. The combustion zone 

is assumed to be at a constant mean temperature and the diffusion 


coefficient has a constant value D = D 

m o 



where D is the 

diffusion coefficient for hydrogen atoms in unburned gas at roan temper- 
ature. It is also assumed that the rate of formation of the product 
at any point can be written as a sum of terms, one for each effective 
radical or atom, each one being of the first order with respect to the 
radical and the reactant, and chain branching reactions cure assumed not 
to occur. The general formula derived using these assumptions is 

V k.n P.D 

2^ L 1 r 1 (II-1-15) 
b i g^B.. 

where k^ = rate constant appropriate to the ith reaction, = equilibrium 
partial pressure at flame front of ith component, q^ = mole fraction of 
potential product and B^' is a constant, containing reaction rates, which 
allows for loss of the active species due to reaction. When flame 
velocities of carbon monoxide and hydrogen flames were calculated using 
the above equation, the calculated value of was never in error by 
more than 25 percent and in general the error was much smaller. For 
carbon monoxide flame mixtures the predicted flame velocities varied from 
25 to 106 cm/sec which is in good agreement with the stoichicanetric burning 
velocity of approximately 55 cm/sec (15) . 


Van Tiggelen (36, 37) proposed a theory of flame propagation 
which emphasized chain branching as the chief reaction mechanism, 
and derived an expression for the flame velocity by assuming that the 
velocity is limited by the rate of chemical reaction rather than by 
heat conduction. Van Tiggelen followed Semenov (38) in stating that 
the active centers, which propagate the chain reaction, undergo- in . . 
the course of a molecular collision either 

a. A branching reaction with probability a. 

b. A chain- breaking reaction with probability 8. 

c. A simple elastic collision or a reaction of simple propagation 
of the chain. 

Burning is established when the probability of branching exceeds that 

of breaking i.e. a >_ 6. The total pressure and partial pressure of 

the reactants are assumed to be known and constant. The temperature 

in the flame zone T is assumed constant with a mean value between T 

m o 

and T^. All molecules are assumed to be equal in size and mass so that 
D = a v/3 where a is the mean free path and v is the mean velocity of 
the molecules. If d is the mean linear distance which a molecule can 
move in a region then the number of collisions undergone during the 

displacement is given by (3 9) 

2 2 

Z = 3Trd /4a (II-1-16) 

In order for a chain reaction to be maintained the following condition 
must be realized 

3iTd^ (a-B/4a^) = 1 (II-1-17) 

the displacement d effected by diffusion in a time t is given by the 
approximate expression 

d^ = 2Dt = 2avt/3 (II-1-18) 


where the mean velocity of the molecules, v, is given by diffusion 

of the active molecules. 

Van Tiggelen then obtains an equation for the velocity of the 

flame front in a gas at a temperature T which is given by 


4T /2R (a-B) 

= — ° (II-1-19) 


For a = 3 equation 17 gives zero for the flame velocity. Thus the 
expression anticipates the existence of concentration mixture limits, 
outside of which combustion is not possible. If one neglects 3 for 
mixtures well within the combustion limits, a calculation of Vj^ is 
possible from a knowledge of and a. Assuming that the branching 
reaction involves the radical CH^ with and the energy of activation 
of this reaction is taken to be 40 kcal, the probability ct is written 

a = [O^Jexp (-40,000/RT^/p) (II-1-20) 
so that the expression for flame velocity for this reaction becomes 

4T /2R[0^]exp (-40,000/RT ) 
o 2 m 

= — ZZ (II-1-21) 

tt/ SmpT^. 

Using this equation, flame velocities in mixtures of methane and air 
have been calculated and the results are in good agreament with the 
experimental values obtained by Coward and Hartwell (40). Ideas similar 
to these of Van Tiggelen were also stated in a qualitative form by 
Spalding (41), who concluded that in branched-chain reactions the 
reaction rate should be determined by the rate of the branching process, 


because of its leading role, and that the reaction mechanism in the 
simplest case would be identical with that of a simple one-stage reaction. 

Similar diffusion theories were presented by Gaydon and Wolfhard (5) , 
Gaydon (42) , and Manson (43) with slight variations on previous theories. 

These studies proved that there is good evidence in many flames 
that diffusion of active species, such as free atoms or radicals, is 
important, but it also became obvious that flame propagation cannot be 
entirely explained by radical diffusion without also considering heat 
transfer. If the presence of a free radical alone was sufficient to 
initiate a reaction, then the fuel/oxidant mixture would be self- 
igniting. Normally, combustion occurs by a chain mechanism which involves 
a highly endothermic chain-initiation step, then slightly less endothermic 
chain-branching reactions and then exothermic chain-propagation reactions. 
Thermal ignition will normally be limited l?y the chain-initiation step, 
and will only occur, for short times, at fairly high temperatures. The 
role of diffusion may be to overcome this initiation step so that 
heating to a lower temperature may Suffice to provide enough energy for 
the branching step and cause the reaction to proceed. In dealing with 
diffusion and thermal theories of flame propagation, it must be remembered 
that molecular diffusion and heat transfer are governed by similar laws. 
If we have a propagation mechanism which depends on reactions being 
started by radical diffusion, but these reactions also require an acti- 
vation energy, then the flame propagation will depend on both molecular 
diffusion and heat transfer. In such a case, the less efficient process 
would tend to be the rate-determining step. For example , if there is 


an ample supply of radicals, then the heat transfer required to supply 
the activation energy will be limiting and factors affecting the heat 
transfer will have a more important influence than those affecting the 
supply of radicals. But the fact that heat transfer appears to be 
important should not mean that radical diffusion does not occur or is 
not important. 

It is evident that any exact solution of laminar flame propagation 
must take into account both thermal and diffusion theories as well as 
basic fluid dynamics equations modified to account for the liberation 
and conduction of heat and for changes of chemical species within the 
reaction zone due to chemical reactions. 
Comprehensive Theories of Flame Structure 

There seems to be general agreement of workers in the combustion 
field that the detailed structure of a flame can be determined by 
solving the equation of state, the equations of conservation of mass, 
momentum, energy and the hydrodynamical equations of change together 
with the boundary conditions which are experimentally imposed. The 
equations of change ccanprise (44) : 

1. The equation of continuity of each chemical species. 

2. The equation of motion. 

3. The equation of energy conservation. 

In addition to these principle relations there are the auxiliary 

4. The complete set of reaction rate equations in terms of the 
chemical reaction rate constants. 

5. The equations for the diffusion velocities in terms of the 
usual binary diffusion coefficients and the multiccsnponent 
thermal diffusion coefficients. 


6. The equation for the pressure tensor in terms of the 
coefficients of shecir and bulk viscosity. 

7. The equation for the energy flux in terms of the coefficient 
of thermal conductivity, 

8. The equation for the radiation energy flux in terms of the 
absorption cind emission spectra of the molecules and the 
radiation emitted or absorbed as a result of the chemical 
reactions . 

These are also the subsidiary relations which determine 

9. The coefficient of viscosity for the multi-ccmponent mixture. 

10. The coefficient of heat conductivity for the multicomponent 
mixture disregarding the effects due to chemical reactions 
(which are accovinted for in the energy flux equation) . 

The boundary conditions should include heat transfer from the flame 

to the flame holder and to the surrounding medium. 

The hydrodynamic equations of change, representing the overall 
conservation of momentum and energy in molecular collisions and the 
rate of change in molecular species due to chemical reactions and 
diffusion, can be rigorously derived using kinetic theory beginning with 
the Liouville equation (45) . The same result can also be obtained in 
a simpler manner from a physical derivation of the Boltzmann equation 
(46) , followed by the identification of the hydrodynamical variables 
and the development of the equations of change. 

Basic equations . The state of a gas mixture may be described by 

the temeprature T, the velocity V and the molar density of each species 

n^ at each point r. In addition, it is convenient to introduce the 
overall density p 

P " i "i™i (II-2-11) 


where is the molecular weight of species i. The equations of change 
are then 

1. The equations of continuity of each chemical species, 


3 -> -»- 

•T=p\ -n. V + V. = K. (II-2-2) 
or/ 1 1 1 

where V. is the diffusion velocity and K. is the net rate of formation 
1 1 

of i molecules due to chemical reactions. Since chemical reactions 
neither create or destroy matter as a whole, 

^ m.K. = O 

2. The overall equation of continuity (sum of individual equations, 
each weighted with its corresponding molecular weight) is given by 

9 \ -> 

p+ [-^j -pv = O (II-2-3) 

3. The equation of motion, 

(ir) ^ ^' {^] ^ = - d/p) (S/3?)-p + (l/p)J n^X^ (II-2-4) 

where p is the pressure tensor (including the effects of viscous forces) 
and is the external force on a mole of particles of species i 

4. The energy balance equation, 

1^ U + v.|y U = - d/p) 0/8?)- (q + q^^) - 

^ ^ T (II-2-5) 

(l/p)p: 0/8r)v +(l/p); n.V.-X. 



where U is the thermodynamic internal energy of the mixture per unit mass, 
q is the energy flux due to molecular motions and q is the radiation 
contribution to the energy flux. 

These equations are inconplete until expressions for the and 
the various fluxes are given. To reduce complexity of calculation these 
expressions will be given for mixtures of ideal gases. The chemical 


kinetics of a reacting mixture may be described in terms of a set of 
chemical reactions, which may be written symbolically in the form 
3, . [l]+3- . [2]+...;> n, . [i]+n^. [2]+... (11-2-6) 

where the 3 . . and the r\ . . are integers and the [il indicate the 
molecular species. Let the rate constant for the j-th forward reaction 

by k. and that for the reverse reaction be k.'. Then in a mixture of 
3 D 

ideal gases the rate of the forward reaction is given by 

3,. 3,. 
k.n, ^'n, 2D, 
D 1 2 


and a similar expression gives the rate of the reverse reaction. The 
total rate of formation of i due to chemical reactions is then 

k. = ] (n,j-3..)[k.n^ ... -k.'n^ ...] (II-2-8) 

This is the equation for the which is used in the equation of 
continuity of species i, Equation (II-2-2). 

In a mixture of ideal gases, the diffusion velocities are 

V. = {n2/n.p)E .D. .d.-(D'^/n.m. )8 (lnT)/3r 



d. = + 
2 9r n 

n. n.m. 

n p 

81np/8r - 

(n^mVp p) 

p ->■ Z 

— X. - , n X, 

m_. 3 k k k 


and p = nRT (II-2-11) 

Hers n is the total number of moles per unit volume and R is the gas 

constant. The D^_. are the multiccmponent diffusion coefficients and 

the D are the multiccmponent thermal diffusion coefficients. Both 

the D^_. and are complicated functions of the temperature and the 


conposition. In flames it is usually more convenient to use the 
implicit expressions for the diffusion velocities given by the 
diffusion equations obtained by combining Equation (II-2-10) with 
the equation 

„ n.n. 

'^i" j 2 B 
n D . . 

: 1 



1 1 

2 B 
n D. . 


(o'^/n.m.- o'^/n.m. ) (II-2-12) 
3 3 3 111 

where D. . are the binary diffusion coefficients associated with the 

interdif fusion of species i and j. The binary diffusion coefficients 
are to a good approximation independent of the composition. The 
diffusion velocities are defined relative to the mass average velocity 
V so that 

. n.m .V . = O 
1 111 


In any isotropic, Newtonian fluid, the pressure tensor is 

p = p U - 2ns 
where S is the rate of shear tensor 


" 1 

9 _ 








the t indicates the transpose of the vector gradient. U is the unit 
tensor, r\- and K'' are the coefficients of shear and bulk viscosity, 
respectively, and £ is the static pressure, which for an ideal gas is 
given by Equation (II-2-11) . 

In a mixture of ideal gases, the contribution of the molecular 
motions to the energy flux is 


q = + . n.H.V. + 

^ 9r 1111 

— ^. (n.oVm.D^. (V.-V.) (II-2-16) 
n i: 3 1 1 13 1 3 

where X is the coefficient of thermal conductivity and is the 

enthalpy per mole of species i. The energy flux due to radiation 

processes, q , depends in an important way on the exact chemical 

ccmposition of the mixture and can normally be neglected in laminar 

prefixed flames (45) . 

When the expressions for the aiv3 tte various fluxes are used 

in the equations of change. Equations (11-2-2,4,5)., one obtains a set 

of nonseparable nonlinear partial differential equations which describe 

the time and space variation of the variables, n^, v, and T. These 

equations involve the rate constants (T) and k_. (T) for the chemical 

B T 

reactions and the transport coefficients D^^, D^, r\, K', and X. 

Equations (11-2-2,3,4 and 5) are the fundamental relations 
governing the dynamics of any reacting mixture and as written are, for 
all practical purposes, completely rigorous. Thus they provide a 
starting point for flame structure calculations but, as is often the 
case, complete rigor is also accompanied by a complete lack of practical 
utility. In order to make use of these equations it is necessary to 
simplify them in several ways, and these simplifications will now be 

One-dimensional flame equations . Normally only one-dimensional 
flow is considered in conbustion models and a reasonable experimental 
approximation to this can be acccanplished by the "flat flame" (41) . 


The equations of change, described above, can be considered for the 
special case in vAiich the flow is only in the z direction (up) and 
in which all variables are independent of the x and y coordinates. 
At this time it is convenient to define several additional quantities: 

1. M is the total mass flux, M = pv. 

2. is the mole fraction of component i, = n^/n. 

3. G . is the fraction of the mass flux due to component i, 

G'^ = n.m. (v + V. )M. 
Ill 1 

The are often called the chemical progress variables. It is 

convenient to consider the G^'s as dependent variables and to use the 

diffusion equation (II-2-12) , as part of the hydrodynamic equations. 

The expressions for the various fluxes and the above definitions 

can now be used in the equations reduced to one-dimensional form. 

With some rearrangement the following equations are obtained: 

1. The equation of continuity of each chemical species, 

{3:/3t) (n.X.)= - (d/dz) (MG./m.) + K. (II-2-17) 
11 111 

2. The overall equation of continuity, 

0/3t)p = -{9/3z)M (II-2-18) 

3. The equation of motion. 

St az 

^^Mv-[| , + k.] |lj + ^ n.x.X. 

4. The energy balance equation, 

9 r E ^ 1 2I 9 r, 9T Z " 

— n.x.m.H.- p + - pv = — IX— -M.G.H. - 
9t|_iiii*- 2 J 92L9z 111 


m X . 

-LJ. G 

m .X . 1 

G_. - 

' + (II-2-20) 


5. The diffusion equations, 

ax. „ ^ (x.m.G .- X .m .G . ) /l-nin.\ _ 

az n j m .m .D . . i I p / 3z 

3 3 ID (II-2-21) 

T T 
(x .m .D . = X .m .D . ) X . n 

3lnT Z 3- J- 3 111 ^ _i_ 

3z ran .m .D . . p 

mil . 

X. ^ ^ x.X. 

1 P D D D 

These equations would be used in the study of time dependent phenomena, 
such as traveling wave canbustion, stability questions and the possible 
transformation of a flame into an explosion. 

One-dimensional steady state flames and explosions travel with 
a constant velocity into the unburned gas. If the unburned gas is 
stationary, the flame front moves in space with a constant velocity. On 
the other hand , if the unburned gas moves with the proper constant 
velocity, the flame front may be maintained at a constant position in 
space. This occurs when a flame is stabilized on a burner such as the 
one used in these studies. In this case it is convenient to use a 
coordinate system fixed with respect to the flame front and consider 
the stationary state solutions of the equations. This is possible 
because the hydrodynamic equations apply in any system of coordinates 
moving with constant velocity. 

To obtain the time-independent equations we require that all time 
derivatives in the hydrodynamic equations, (II-2-17 - 21) be zero. 
Also we will assume that no external forces act upon the system, so 
that all the = O. Equations (II-2-18, 19 and 20) may then be 
integrated with respect to z, with the result taht the mass rate of flow, 
M, is a constant. The equations are then as follows: 


1. The equation of continuity of each chemical species. 

dG . m . K . 
1 11 

dz M 

2. The equation of motion, 

( — Ti + k-*] — = p + Mv + L. 
\3 I dz ^ 1 

3. The energy balance equation. 

) dz 111 

— Mv + q + L - — — . . — 

2 2 2 1] m .m .D . . 

n 1 D ID 

G . - 


m X . 
m .X . 
1 1 



-, (II-2-24) 

where and are the constants of integration. The diffusion equations 

are given by Equation (II-2-21) are unchanged (except for omission of 

the terms due to the external forces, X^). 

Boundary conditions for flame equations . In the laminar pre-mixed 

flame the burnt gases are free to flow undisturbed ifinitely far in the 

positive z direction. The various properties of the burned gases then 

asymptotically approach limiting values as z approaches infinity. These 

limiting values are called the "hot boundary values and these quantities 

are indicated by a subscript The constants L. and can then be 

expressed in terms of the hot boundary values of the various quantities 

by noting that in this limit dv/dz and dT/dz are equal to zero. In 

most flame problems one may neglect radiation effects (q^^ = 0), neglect 


thermal diffusion and the Dufovtr effect (D . = 0) and neglect pressure 
diffusion. When these terms are dropped fron the hydrodynamic equations 
and the constants of integration evaluated in terms of the hot boundary 


conditions, one obtains the following "flame equations"; 
1. The equation of continuity, 

dG . m . K . 
1 11 

dz M 

2. The equation of motion. 


(t^ + k' l = (£_ - p )+ M (^^ - V ) (II-2-26) 

^3 / dz — " 

3. The energy balance equation, 
Mdz 111 111 p^-^ 

CO 00 ^ 00 

(1/2) (VI - V )^ (II-2-27) 

4. The diffusion equations. 

dx . „ ^ (x .m ,G . X ,m .G , ) 
i_ _ M E 1 1 j - ] ] 1 

dz n j m ,m .D . . 

1 3 1] 


The problem is then one of solving these equations for specified 
boundary conditions. 

At the hot boundary limit, z " , the various quantities approach 
finite limits corresponding to complete chemical and thermal equilibrium. 
Accordingly, the derivatives of all of the quantities approach zero as 
z -> «>. 

The cold boundary conditions are more difficult to specify. First 
consider the case where all of the are equal to zero at the cold 
boundary layer. In this case we can consider an "unattached" flame and 
let z ->■-«> . The boundary conditions at the cold boundary are then 
completely analogous to those at the hot boundary; the various quantities 
approach finite limits (properties of unburned gas mixture) as z ->■ 
and the derivatives approach zero. Equations relating the values at 
the two boundaries can be obtained frctn equations (II-2-26, 27 and 28) 


by setting the derivatives equal to zero. Unfortunately this is not 
a reasonable approximation to actual physical processes occurring. 
In usual practice Arrhennius type expressions are used for the 
temperature dependence of the the chemical reaction rate constants. 
This leads to a small but finite value for the reaction rate even at 
the cold boundary temperature, T^. Other problans involved with the 
cold boundary condition include back-diffusion of the product gases 
beneath the burner head, causing the chemical composition of the 
molecules entering the flame to be ill-defined. Also, unless there 
is a small amount of heat transferred fron the flame to the flame 
holder, the position of the flame with reference to the flame holder is 

Hirschf elder and Curtiss (45, 47) eliminated this cold boundary 
problem by introducing the concept of a flame holder at the cold 
boundary. This was originally developed as a mathematical concept 
but it is also physically realistic because all stationary flames use 
a flame holder (burner) for support and it does heat up due to the 
flame. Their flame holder had the following two properties: 

1. It serves as a porous plug permitting the reactant gases 

to pass through it freely but preventing the back-diffusion 
of the product gases. 

2. It serves as a heat sink with a small amount of heat transfer 
taking place from the flame to tlie flame holder. 

The cold boundary conditions at the flame holder, which is taken 

to be the origin of the z coordinate system, are' 

1. The are the mass fractions of the various components in 
the unburned fuel-air mixture. 

2. The heat transfer to the heat sink or the value of energy 
flux at z = 0 is 

q = -X (dT/dz) 
^o o ' o 


The concept of the flame holder is apparently necessary if the 

rate constants are taken to be finite at T . The concept is, however, 


useful even if the are equal to zero at T^ since the heat transfer 

to the flame holder determines the quenching distance. If the initial 

temperature, pressure and composition of the fuel-air mixture are 

specified along with the strength of the heat sink, q^, a solution of 

the equations exists (and satisfies all of the boundary conditions) 

only for a single value of the mass rate of flow, M. This value of M 

is the product of the flame velocity (the value of v at the flame 

holder, Z = 0 and p the gas density at the flame holder. The problem 


then is an eigenvalue problem for obtaining the proper value of M. 

Hirschf elder et al. (48) studied flames involving the first-order 

deccmposition of azomethane (C^H^N^ -> C + N^) and the bimolecular 

2 6 2 2 6 2 

decomposition of nitric oxide {2N0 ;^ + 0^) in detail. Four cases 

for the azomethane deccmposition were considered: (a) no heat loss to 

the flame holder, so that q^ = 0; (b) a reasonable amount of heat loss 


to the holder q = 10 cal/cm s) ; (c) 50 percent by volume of inert gases 

introduced with the fuel, and (d) the deccmposition assumed to take 

place by the autocatalytic mechanism (C^H^N^ + C 2C^H^ + N_). 

262 25 26 2 

For cases a, b and d at one atmos;phere pressure the decomposing 
azomethane has a large flame velocity and a narrow flame zone. For 
case c, the flame velocity was calculated to be 15cm/sec with flame 
thickness of 0.019 cm, in good agreement with the measured .valoes. 


Modern theoretical treatments . Various other approaches to solve 
the flame equations have been proposed by several authors. It was shovm 
by Spalding (41) that the steady flame speed could be found conveniently 
by solving the equations for the unsteady-flame situation. In this 
paper he calculated the flame speed of a hydrazine decomposition flame 
using a chain reaction mechanism instead of a \inimolecular decomposition. 
The chain reaction scheme used was proposed by Adams and Stocks (49) , 
as follows: A 2B, radical production, B+A B+2C, chain propagation, 
B+B+G 2C+G, chain breaking. In this scheme no difference is made 
between the radicals NH^fH or other radicals which may be present. 
A denotes hydrazine and C end products. Spalding used for the 
calculation of the radical concentration and for the burning velocity 
a numerical integration method which starts from an arbitrary temperature 
profile through the flame zone and tried to follow the transient process 
which converges to the stationary state. He used the assumption that 
the sum of the thermal and chemical enthalpies remains constant through- 
out the flame so that the mass fraction of the hydrazine is a linear 
function of the temperature. In this paper he introduced a function 
defined by the equation 

d^ii = pdz 

If there were uniform flow in the z direction, ^ could be referred to 
as a stream function. Its definition is such that equal masses of 
gas are contained between planes where the ^ values differ by equal 
amounts. As the denisty of the gas changes, however, the distances 
separating these planes vary. The origin of ij; was chosen in such a way 
that there is no net mass flow across planes of equal ;(/, although 


diffusion may occur. The flame equations were rewritten using this 
variable and then solved by a finite-difference method (41). Using 
this method Spalding obtaind burning velocities that were in good 
agreement with published values (49) . 

In later papers, Spalding and Stephenson (50) and Spalding et al. 
(51) , presented a new procedure for the solution of the differential 
equations of transient one-dimensional flame propagation employing a 
coordinate system which moved with the flame. The chief contribution 
of this paper was the presentation of a series of modified flame 
equations which reduced the time and storage needed for computation, 
and which, by suitable implicit formulation of the finite-difference 
equations, allowed a standard computer program to be used. In earlier 
work Spalding (41) used the coordinates of time, t, and 'stream function', 
ip , defined as - = p and D'Tj/Zdt = -pv. In this work, time remainedas 
the first coordinate, while the second became the dimensionless stream 
function, w, defined as 

w = ( l!)_ij)jj)/{ij)^ - ; 

where and ijj^ are the instantaneous values xSf \|j at the hot and cold 
boundaries of the flame, respectively. As the flame develops and spreads 
into the unburned gas, tj/^^ and ij;^ change their values; at large times, 
both ip^ and Tp ^ increase linearly with time yielding 

d't'H _ _ 
dt~ ~ dt~ ~ ^v^u 

where S is the speed of propagation of the flame relative to the unburned 

gas, and p_^ is the density of that gas. Values may be ascribed 
arbitrarily to the ty, (t) and \p (t) , but the saving of computer storage 


and ccmputational time results from choosing these functions so the 
hot and cold boundaries lie at the locations at which the gradients 
of temperature or concentration have just significant values. Therefore 
the boundary conditions- imposed are 0<_ w <_ 1. 

This choice of coordinate system enables the flame equations to be 
solved by adaptation of a computer program of Patankar and Spalding 
(52) who developed it to solve the parabolic equations which arise in 
the theory of the steady two-dimensional boundary layer problem of hydro- 
dynamics. Modifications to this work were presented in Spalding and 
Stephenson's paper to allow use on the flame equations. Using this 
program they were able to calculate burning velocity for laminar hydrogen 
and bromine flames, both for flame propagation into an unlimited expanse 
of pre-mixed reactants, and for a flame stabilized on a cooled porous 
plug burner. They were able to obtain quite good agreement with 
experimental results for both cases, which was attributed to use of a 
complete chemical kinetic reaction scheme using four reactions. Of course 
this is still a very simple case. 

A more complex case was studied by Dixon-Lewis (53, 54) who studied 
the hydrogen-oxygen flame both theoretically and experimentally. In 
these papers hydrogen-oxygen burning velocity and concentrations were 
calculated using different reaction mechanisms and the procedure proposed 
by Spalding and Stephenson (50), using finite difference methods. 
Detailed specification of thermodynamic and transport properties were 
also used in this approach. 

The two reaction mechanisms used were as follows: 

1. (i) OH + ;J H^O + H 
(ii) H + 0_ -V OH + O 


(iii) O + OH + 0 

(iv) H + H + M-^H^+M 

Assuming reaction (ii) to be the rate controlling step, reactions (i) , 
(ii) and (iii) were ccmbiried to give (iia) H + O2 (+ BH^) ->- 2H2O + 3H. 
Reactions (iia) and (iv) provided the complete flame reaction mechanism. 

2. (i) OH + H H O + H 

2 ^ 2 

(iib) H + 02(+ H^) 20H + H 

(iv) H + H+ M->-H2+M 

(v) H + OH + M ^ H^O + M 

This is a slightly more complex mechanism than before and reaction (v) 
has been included to observe the effect of the radical recombination 
reaction on the equilibrium of H and OH by the rapid reaction (i) . 

In these papers the trcinsport fluxes in the multicomponent flame 
system due to diffusion and thermal conduction, and to thermal diffusion 
and its reciprocal effect (dufour effect) , are considered from the 
standpoint of an extension of the Chapman-Enskog kinetic theory 
of bimolecular gases to polyatomic gases by Wang Chang , Uhlenbeck and 
de Boer (55) and the subsequent development by Mason, Monchick and co- 
workers (56-60) to polar-polar mclecular interactions. Dixon-Lewis (54) 
gives equations for the various diffusional, thermal diffusional and 
thermal fluxes which it is necessary to derive, in order to obtain reaction 
rates from experimental temperature and composition profiles in flame. 
Also the organization of computer programs for calculation of the 
multicomponent diffusion and thermal diffusion coefficients aind the 
thermal conductivity are described. The expressions for the transport 
fluxes are then used to derive equations for the mole fraction and 
temperature gradients in the flame where transport processes and 


chemical reactions occur side by side. From the mole fraction and 
temperature at one point in the flame it is then possible to solve 
the flame equations by a numerical integration method such as the 
Runge-Kutta procedure to conpute the complete composition and temperature 
profiles. The calculated values for composition profiles were checked 
experimentally by sampling from the flame in different positions with 
a quartz microprobe and subsequently analyzed mass spectrometrically 
using a continuous flow system. Good agreement between theoretical 
and experimental profile was found except for the hydrogen profile 
wliere the computed profile was slightly displaced with respect to the 
experimental one. This observation was explained by possible diffusion 
effects in the pressure gradient at the probe tip. 

Dixon-Lewis's treatment of flame transport pheonmena is the most 
comprehensive and rigorous discussion currently available that can 
be used in a computer program for flame calculations ^54) . This treatment 
should be included in any flame model that porports to represent actual 
physical-chemical processes occurring in flames. A ccxnplete flame 
theory would also have to represent the actual chemical reactions occurring 
and not just a simplified set. 

This was partially accomplished in the work of Smoot et al. (61) 
who studied the kinetics and propagation of laminar methane-air flames 
using a one-dimensional flame propagation model based upon the equations 
presented by Spalding et al. (51) . This model is based on a numerical, 
unsteady-state solution of transformed species and energy conservation 
equations using explicit techniques for diffusion terms (binary diffusion) 
and linearized, implicit techniques for kinetic terms. A methane-oxygen 


flame kinetic mechanism consisting of 28 elementary reactions and 15 
species was postulated and used in this model. Flame velocity, flame 
thickness, temperature profile and concentration profiles of 13 species 
were predicted for a series of methane-air flames. In this study the 
effects of pressure, methane concentration, initial temperature, rate 
constants and transport coefficients, were investigated. The predictions 
of the model were compared with experimental data, and agreement was 
generally quite good. The concentrations of the radicals H, OH and O 
were found to be of major importance in the propagation of methane-air 

The most complete model presented at this time is by Tsatsaronis 
(62) who solved the transformed species and energy conservation equations 
using the numerical computer solution proposed by Spalding et al. (51) . 
The diffusion model used was the rigorous and ccmplete one used by 
Dixon -Lewis (54) based on the modified Chapman-Enskog kinetic theory of 
Manson, Monchick and co-workers (56-58) for polyatomic gas mixtures contain- 
ing polar components. He also used a methane-oxygen reaction mechanism 
containing 29 elementary reactions and 13 species. The effects of 
equivalence ratio, pressure, initial temperature and transport coefficients 
on the flame velocity, flame thickness temperature profile and concentra- 
tion profiles were investigated for a series of methane-air flames. 
Model predictions were compared with experimental results (63, 64) and 
were found to be in quite good agreement. It was found that flame 
characteristics were very sensitive to changes of certain multicomponent 

diffusion coefficients (eg. D D ) such that an increase of 

H,H20, ^2'^2° 

50% in D was found to result in a decreasing of the flame thickness 

H,H O 


by about 40% and an increase of the flame velocity by more than 100% 

for a stoichiometric methane-air flame at atmospheric pressure. Neglecting 


the thermal diffusion (D ) resulted m a decrease of the flame velocity 
of about 2% and in slight changes in the profiles. It was also found 
that the concentrations of the radicals H and OH were major factors in 
determination of flame characteristics and the validity of some chemical 
rate constants must be questioned. 

. Chemical Reactions Occurring in Methane-Air Flames 

In the most simplistic view, the overall chemical reaction occur- 
ring in a methane-air flame can be written as 

CH^ +20^ ■> CO^ + 2H2O 
which in fact does describe the reaction, although rather in black box 
terms, i.e. it accurately describes initial reactants and final products, 
but gives us no idea of how this overall reaction actually takes place. 
Fortunately this overall reaction can be broken down into more basic 
elementary steps which can be studied in greater detail. In fact, the 
combustion of methane with air has been found to occur through a complex 
chain reaction scheme with a multitude of individual reactions. 
Elementary Reactions Mechanism 

As is typical in a chain reaction process the methane reaction 
scheme consists of a chain initiation reaction followed by chain branching 
and propagation reactions, finally ended by chain termination reactions. 

The chain initiation step is now thought to proceed through the 
following reactions: 

CH^ + OH ^ CH^ + H^O (II-3-1) 

CH^ + H CH^ + (II-3-2) 


CH + 0 ->- CH^ + OH (II-3-3) 
4 3 

CH^ + M CH^ + H + M (II-3-4) 
Reactions 1 and 3 are most important in lean and stoichicaiietric flames 
while reaction 2 becomes important in rich flames. Reaction 4 becomes 
enhanced with increasing temperature and may become significant at 
tanperatures approaching or greater than 2000 K. In a stoichiometric 
flame the rate of destruction of CH^ by H is about equal to that by OH 
(65) but H'.s low concentration precludes it from. playing a. major role. 

The propagation steps involve the consumption of the methyl radicals 
(CH^) formed in the initiation reactions. It was first thought (66) 
that methyl radicals reacted directly with O^ 

CH^ + O^ HCO + H^O (II-3-5) 
Later mechanisms have replaced the above overall reaction by (67) 

CH^ + 0^ ^ CH^O + OH (11-3-6) 

In addition, methyl radicals can directly oxidize through the following 
CH^ + O ^ CH^O + H (II-3-7) 
CH^ + 0^ ^ CH^O + O (II-3-8) 
CH^ + OH ^ CH^O + (II-3-9) 

while at very high temperatures CH^ may also dissociate, 

CH^ + M ^ CH^ + H + M (II-3-10) 

It has also recently been shown that methyl radical recombination (68) 

CH^ + CH^ ^ C^Hg (II-3-11) 

is also responsible for much of the methyl consumption. C H can be 

2 6 

reacted in the following manner (69) : 

C^Hg + H -> C2H5 + H^ (11-3-12) 


C^Hg + 0 -y C^H^ + OH (II-3-13) 

CH +0H->- CH + H^O (II-3-14) 
2 6 2 5 2 

C^H^ + H ^ C^H^ + (II-3-15) 

C^H^ + H -s- C^H^ + (II-3-16) 

C + O CH^ + HCO (II-3-17) 
2 4 3 

C H + OH ^ C^H, + HO (II-3-18) 
2 4 2 3 2 

*^2"3 ^ ^2^2 ^ (II-3-19) 

^2^2 " ^2" "2 (II-3-20) 

C H + OH C H + HO (11-3-21) 
2 2 2 2 

C^H + ©2 ^° (II-3-22) 

C H + O CO + CH (II-3-23) 

The CH^O radicals react by means of 

CH^O + M -> CH^O + H + M (II-3-24) 

CH^O + CH^O + HO^ (II-3-25) 

with the deccsnposition reaction providing the major reaction. CH^O 

can be remove! via the following 

CH^O + OH ^ CHO + H^O (II-3-26) 

CH^O + O -v CHO + OH (II-3-27) 

CH^O + H ->- CHO + (II-3-28) 

with the resultant removal of CHO 

CHO + CO^ + OH (11-3-29) 

CHO + OH CO + H^O (II-3-30) 

CHO + O CO^ + H (II-3-31) 
Experimental evidence has indicated a large increase in CO as the 

concentration of CH^ decreases (70) . It is evident that the CO must 

be produced from the methyl radicals formed in the initiation step 

and a possible first step is 


H^CO + OH HCO + H^O (II-3-33) 
followed by 

HCO + OH ^ CO + H^O (II-3-34) 

the CO will build up as the CH^ is destroyed until the production is 
overccsne by the destruction step 

CO + OH ^ CO^ + H (II-3-35) 
The reactions discussed up to now, which account for the appearance 
or disappearance of the main species discussed, are all basically of the 
simple chain type. In addition to these reactions there are the chain 
branching reactions which lead to the production of the radicals 
essential for all the other reactions. These branching reactions include 
HHO^^ OH + OH (II-3-36) 
H + O^ OH + 0 (II-3-37) 
H + O OH + H (II-3-38) 
O + H^O -v OH + OH (TI-3-39) 
The termination steps of the methane-air flame system would include 
the following radical recombinations which reduce the contraction of 
radicals present thereby reducing and ending the reaction. 

H+H + M ^H^+M (II-3-40) 
0 + 0+ M O^+M (II-3-41) 

H + O + M OH + M (II-3-42) 

H + OH + M -i- H^O + M (II-3-43) 
All of these recombinations are preV:alent in methane-air flames with 
the exception of the H atom recombination which would be much less 
important in lean flames. 


The reactions given in the above section are probably the most 

important reactions in methane-air flames, with the exception of the 

C H sequence which would only become important in rich flames. 
2 6 

Of course this is not a definitive set of equations, as many more 
combinations of reactions can be thought of, and reaction schenes of 
up to 100 reactions have been postulated. However it is considered 
by this author that some intuitive observations can be made on a 
small reaction set whereas a set of a hundred reactions only tend to 
cloud the picture unless used in a massive computer code where individual 
reactions can be varied and the effects observed. Ccmprehensive 
compilations of reactions have been obtained for computer calculations 
and the work of Tsatsaronis (62), Jachimowski and Wilson (69), Smootet al. 
(61) and Olson and Gardiner (67) give excellent treatments of large 
reaction schmes . 

Production of Excited State Molecules 

In emission spectroscopy one observes the light anitted when an 
excited atom or molecule undergoes a radiative transition to the ground 
state with the resultant emission of a photon. In a flame there are 
two basic ways that an excited state molecule can be produced. A 
molecule can become excited through collisions with other flame gas 
molecules whereupon the colliding molecules internal molecular energy 
and translational energy are converted into electronic energy in the 
other molecule (71, 72) . In this case the number of molecules in the 
excited state is governed by the temperature of the flame and the energy 
of the excited level involved and is given by Boltzmann's equation. 


In the second case an excited state can be produced during a 

chemical reaction. In most cases the heat of combustion is distribated 

over the many degrees of freedom of the flame molecules which can then 

be collisionally transferred to other molecules by inelastic collisions 

with these molecules resulting in a thermal population. But in 

certain other cases, a part of the energy released in a particular 

exothermic chemical reaction is directly converted into excitation energy. 

In this case the resultant emissions are known as chemiliaminescence. 

This choniluminescence can be expected to occur in flames because the 

energy released in many reactions in the flame is of the same order of 

magnitude as the electronic excitation energy of the molecules (about 

2 to 8 eV) . In flames both types of excitation mechanisms are present 

with chemiluminescence processes being more prominent in the primary 

combustion zone in which the oxidation of the fuel gas takes place. 

Excellent chemiluminescence studies of non-flame systems have been 

performed by Polanyi and co-workers (73-7 5) . 

OH . In hydrocarbon-air flames the main process for the formation 

of OH (excited OH) in the primary reaction zone with high rotational 

excitation temperature is (5) 


CH + 0^ -> CO + OH 

This reaction is exothermic to the extent of about 159 kcal/mole which 
gives sufficient energy for both electronic and rotational excitation. 
This reaction was suggested (76) by the observation that high rotational 
temperatures for OH were only obtained from flames in which CH emission 
was also strong. It has also been supported by deuterium tracer work 
(77) and the observation that atcmic flames using pure 0 by titration 


did not give strong OH anission but addition of to these same flames 

gave strong OH emission (78). The rate constant of this reaction has 

-3 3 

been measured (7 9) to be approximately 1 x 10 cm /molecule sec in low 

pressure C^H^- air and CH - air flames. 
^ 2 2 4 

In flames containing hydrogen the inverse predissociation can occur 

4 - 

O + OH OH ( E )-^ OH* 
The strength of this emission depends on the square of the concentration 
of free atoms (which are in pseudo-equilibrium with each other and ground 
state OH) and which leads to selective excitation of the V = 2 and 3 
levels. In low tonperature flames containing hydrogen the recombination 
reaction occurs 


H + OH + OH ->- HO + OH* 

which results in a normal distribution of rotational and vibrational energy 

in OH. The rate constant for this reaction has been measured by Belles 

5 2 

and Lauver (80) who obtained an average rate constant of 2 x 10 liter / 

mole s for a mixture of 5% H - 95% air in a shock-tube experiment. 


Recently (81) the rate constant for the formation of OH* by 

H + O + M ^ OH* + M 
in a H^/O^/Ar mixture was measured in a shock-tube experiment and was 
found to be 1.2 x lO''''^ exp (-6940/RT) cm^mol ^s 

CH. For the production of excited CH in flames Gaydon (4) has 

suggested the following reaction 

C^ + OH CO + CH* 

while Hand and Kistiakowsky (82) have given the following reactions 
C^H + O CO + CH* 
C^H + 02"*' 


The first reaction is exothermic by 92 Kcal/mole and the last two 

are exothermic by 62 and 7 0 Kcal/mole respectively. In a study of C^, 

CH and OH radicals in a low pressure ^2^2^^2 ^^^^ Bulewicz, Padley 

and Smith (83) found evidence for the reaction and obtained an 

-12 3 

estimated rate constant for this reaction of 8 +_ 4 x 10 cm /molecules s 
at a temperature of 2200 K. 

Kinbara and Noda (84) shudied the time evolution of emission and 
absorption spectra following flash ignition of a mixture of C ^/O'*' /m ^ 
and discussed as possible reactions 

C H + -> CO^ + CH* 

2 2 2 

C^H^ + OH H^ + CO + CH 
2 2 2 

CH + OH + H ^ CH* + HO 


Another mechanism for formation of ground state CH was given by Becker 

and Kley (85) 

CO + H CH + CO 

fron observations made in a flow-tube system for generation of atomic 

flames. They suggested that the C^O was formed in the reaction 

CHCO + H -> CO + H^ 
2 2 

with the HCO formed by hydrogen atcm reactions with C O and CH CO or 

3 2 2 

by oxygen atoms with C^H^- 

A more complicated mechanism for the formation of CH* was proposed 
by Quickert (86) using a two step process 

C^H^ + O -> CH„* + CO 
2 2 2 


CH^* + CH^* -> CH* + CH, 
2 2 3 


He then used this mechanism to explain the experimental results 

obtained in the room temperature reaction of oxygen atoms with C^H^ 

at low pressure with good correlation to the data. 

C . The formation of excited molecules in a flame is one of 

the great unsolved problems in combustion chemistry today. Simple 

reactions can be written for the formation of but most are very 

endothermic and would have difficulty in forming even unexcited C^, 

not to mention the high level of vibrational, rotational, and electronic 

excitation observed in experiments. Also, isotope studies (87, 88) 

in C H /O flames indicate that the two carbon atcms come not from one 
2 2 2 

acetylene molecule but rather from separate carbon fragments. This 
rules out simple stripping reactions such as 

S«2 + « " S« ^ "2 
C^H + H + 

or similar stripping reactions by O atom or OH. At this point in time 
there is still no generally accepted consensus of how C^* is formed, 
but there are many proposed mechanisms. 

One obvious simple reaction which has been proposed (87) 

CH + CH ^ C^* + H^ 

is indeed sufficiently exothermic (84 kcal/mole) to produce in an 

excited state. However quantitative measurements of CH contractions 

by Bleekrode and Nieuwpoort (89) and Bulewicz, Padley and Smith (83) 

show such low concentrations of CH that an unconceivably high rate 
-9 3 

constant (2 x 10 cm /molecules s) is required to explain the observed 
C^* concentrations. 

Another proposed reaction 

C + CH ^ C^* + H 


is also fairly exothermic (61 but also depends on CH. 
The atomic carbon is most likely to be formed by 
CH + H C + 


so that once again the C^* emission would again tend to depend upon 

the square of the CH concentration. Fairbairn (88) has discussed the 

type of reactions forming atomic carbon and proposed for the formation 

of C * the reaction 

C + C+ M> C*+M 

which is a very endothermic reaction. 

A similar type reaction for the formation of C^* has been proposed 

by Peeters et al. (90) 

CH^ + C C^* + H„ 
2 2 2 

which although exothermic (38 kcal/mole) is probably not energetic enough 
to form C^ in an excited state. 

Gaydon and Wolfhard (76) have proposed 

C^O + C ->- C^* + CO 
to explain the formation of excited C^ with the C^O being formed in a 
methane-air flame by the following reaction scheme 

O 4- C^H^ (C^H^O) C^O + H^ 

where (C^H^O) is a short lived transition state complex or 

0 + H C O -> HO + CO 
2 2 2 2 

Both of the above reactions involve the atomic carbon species whose 
formation would not be considered likely on thermochemical grounds, 
but concentrations of atomic carbon have not yet been measured in flames 
so these mechanisms remain feasible. These mechanisms, ho-wever, have 
the advantage that the C atoms in the C„ are drawn from two different 


species and would therefore show the randomization of carbon isotopes 
observed in isotope tracer experiments (87, 88) . 

Studies of emission in low pressure hydrocarbon flames have 
been made by Gaydon and Wolfhard (76) who measured the effect of 
dilution and also replacement of part of the hydrocarbon fuel by 
hydrogen or carbon monoxide. In these studies dilution with nitrogen 
caused a minor decrease in emission whereas excess oxygen had a 
strong quenching effect on the emission. When part of the hydrocarbon 
fuel was replaced with hydrogen or carbon monoxide in the correct 
amount to ranain stoichiometric, it was found that there was a suprisingly 
rapid fall in the and CH emissions. This effect was strongest with 
saturated hydrocarbons, such as CH^, and not so strong with unsaturated 
fuels such as acetylene. Thus in mixed ethane-hydrogen flames the 
intensity varied roughly as the square of the ethane concentration whereas 
in a mixed methane-hydrogen system the intensity varied with the 
sixth power of the hydrocarbon concentration. This behavior indicated 
that the C^* was formed from several hydrocarbon molecules or fragments 
produced from them. 

These observations led Gaydon and Wolfhard to suggest that C^* 

formation occurred via a polymerization or condensation step in which 

large polyacetylenes were formed. These large molecules would then 

decompose exothermically forming C *, a possible mechanism for C H 

2 8 2 

may be 

or other reactions might occur such as 

^4»2 °2 " ^4«2°2 " "2 ^ S* 


Evidence of these large unsaturated molecules occurring in flames has 

been obtained using mass spectrometry but it still is not a popular 

mechanism with many researchers in the field . 

Savadatti and Broida (91) studied the gas-phase reactions of 

carbon vapor with a variety of simple gases and their discharge products 

produced by an electrodeless microwave discharge. The carbon vapor was 

produced using both arc and resistance heating which provided sufficient 

carbon vapor to produce bright visible flames. Flames with atomic 

nitrogen contained only emissions from the CN red and violet systems. 

Reaction with atomic oxygen gave a brilliant green flame which filled the 

reaction chamber, indicating a much faster reaction than the case of 

atomic nitrogen. Spectra of this flame showed very strong Swan and 

0^(405 nm) bands with weaker CO emissions. When molecular oxygen was 

used/a green flame, weaker by a factor of 20, was observed which showed 

all the features of the atomic flame, except bands. It was proposed 

that the observed emission of could be explained by the following 


C- + O CO + C^* 
3 2 

C3 -H 0^ -> CO^ f c^* 
both of these reactions are exothermic with a value of 80.7 Kcal/mole 
for the first and 87.6 Kcal/mole for the second. 

Recently Mann (92) has measured the rate constant for the second 

reaction in a flow through furnace and obtained a value for the lower 
-12 3 

bound of K> 2 X 10 cm /molecules. 


CO. The far ultra-violet bands of the fourth positive bands were 
observed by Kydd and Foss (93), who observed strong emissions down to 
141 nm from low pressure flames of several hydrocarbons including methane. 
Based upon these observations the reactions 

CH* + O -> CO* + H 


CH* + OH -> CO* + 
were considered possible. 

Becker and Bayes (94) studied CO emission frcaa flames of carbon 
suboxide (^^02^ suggested the reaction 

C 0 + 0 CO* + CO 

to account for the fourth positive emissions in both the C^O^ and hydro- 
carbon flames. It was pointed out in this paper that 0 atom could react 
with C^O^ in the following ways 

O + C^O^ ^ SCO 
3 2 

O + C 0^ CO + C^O 
3 2 2 2 

which results in generation of more than enough C^O to account for the 
observed CO emission. The C^O atomic oxygen mechanism is usually 
accepted to be the major reaction responsible for the fourth positive 
emission (4) . 

Other reactions that have been suggested but not proven include 

C^ + O ^ CO* + CO 
2 2 

• 20 + C^H^ ^ 2C0* + H^ 

and the three body recombination 
O + C + M -> CO* + M 


Chemical Analysis of Coal 

There is general agreement that coa] originated through the 
accumulation of plant debris that was later covered, compacted, and 
changed into the organic rock that we find today. Due to the 
differences in plant formations and geological conditions no two coals 
are identical in nature, cctnposition or origin. The formation of 
coal from large plant masses via biochemical and geochemical processes 
is called coalif ication. The extent of coalif ication determines the 
degree to which the original plant material approaches the structure 
of pure carbon. Most bituminous coal seams were deposited in swamps 
that were regularly flooded with nutrient-containing water that supported 
abundant peat forming vegetation. The lower levels of the swamp water 
were anaerobic and acidic; this environment promoted certain structural 
and biochanical decanpositions of the plant remnants. This microbial 
and chemical alteration of the cellulose, lignin, and other plant 
substances, and later depth of burial, resulted in a decrease in the 
percentage of moisture and a gradual increase in the percentage of 
carbon content of the mass. This change from peat through the stages 
of lignite, bituminous, and ultimately to anthracite is characterized 
physically by decreasing porosity and increasing gelification and 



vitrification. Chemically there is a decrease in "volatile matter" 
content, as well as an increase in the percentage of carbon, a gradual 
decrease in the percentage of oxygen and as the anthracite stage is 
approached a marked decrease in the percentage of hydrogen. The 
progressive changes involved in the coalif ication process are called 
an "advance in rank" of coal, and the rank of a coal designates the 
degree to which the metamorphosis from plant debris to coal has occurred. 
The highest rank would be pure carbon and the lowest lignite. Table 1 
shows the classification of coals by rank (95) showing characteristic 
properties of each group. 

In order to determine the rank of a coal some knowledge of its 
chemical and physical properties must be known. For practical purposes 
the chemical composition of coal is always defined in terms of its 
proximate and ultimate analysis. Neither offers significant information 
about the molecular structure of coal, but both furnish data that can be 
correlated with most facets of coal behavior learned from long experience 

Proximate analysis determines 1) moisture contents, 2) volatile 
matter, (VM) , content, 3) ash, inorganic material left behind when all 
comtustible substances have been burned off, and 4) indirectly, the fixed 
carbon content (FC) which are defined by % FC = 100 - (% H^O + % VM + % a 
Comprehensive reviews of the experimental methods for analysis of coal 
have been published (96, 97) but in most coal-producing countries 
national standard techniques are used. In the United States, proximate 
analysis methods are based on standard methods formulated by the American 
Society for Testing Materials (ASTM) , which are under periodic review. 




H O 
XI -H 

ns >M 







■H ^ 










OJ 00 H 



V 1 1 


C'l 00 




o g 
>1 ^ 

Q) M-l 


>! e 
o — • 


ni g 

u -a 





o o o 
o o o 
o o o 
^ ^ ^ 

^ ^ ^ 

rH H H 


V V V 






















CM (N 













I O I O 

o o o o 

o o o o 

o in - 

•. ^ •• ro 

ro rH O rH 










































































■rl W 

-H M 







■P n 

■P 3 







(d 0 

Id 0 








rH C 

rH C 








O -rl 

0 ■H 







> 6 



^ ^ 




CP +J 

tn -P 





•H -H 

■rl -H 




K X5 






















I O 

O O 1 

o in o 
in - o 
H in 

O rH - 


o I o 

in o o 

•« o in 

o ro •• 

rH - C5^ 

,— 1 















































































I o o 
o o o 

O ro ro 
ro * • 

- 00 vO 





0 o 
in in 

1 I 
o o 


















Moisture content (AS1M D 3173-73) is determined by measuring the per- 
centage weight loss of a ground (<60 mesh) 1 gram sample after heating 
for one hour at 107 + 4°C in vacuo or in an inert atmosphere (purified 
nitrogen) . The forms of moisture measured by this method are . 

1. Bulk water which is present in large cracks and capillaries 
and which possesses the normal vapor pressure of water. 

2. Physically adsorbed water, which is held in small pores and 
has a vapor pressure corresponding to its adsorbed state. 

Moisture content is determined upon samples that have been allowed 


to equilibrate under normal laboratory conditions (20 hh 5 C and 30 - 60% 
relative humidity), and this is reported as air-dried moisture content. 
This differs frcm natural bed moisture, or capacity moisture, which is 
the amount of water that a coal will hold when fully saturated at 100% 
relative humidity (as in an undisturbed coal seam) and thus reflects 
the total pore volume of the coal accessible to moisture. The usual 
moisture quantity given is air-dried moisture. 

Volatile matter, which varies qualitatively and quantitatively with 
the rank and cctnposition of the coal, consists of a wide range of low 
molecular weight hydrocarbons, carbon monoxide, carbon dioxide and 
chemically reacted water. Except for small amounts of methane and 
carbon monoxide which may be chemisorbed on the coal, all of these 
substances form by thermal deconposition of the coal. Table 2 shows 
typical volatiles released by coal upon heating. Volatile matter is 
determined (ASOM D 317 5-77) by measuring the weight lost when a 1 gram 
sample (<60 mesh) of coal is heated for 7 minutes at 950 +_ 20°C. 

The ash content is determined by measuring the residues left behind 
when weighed (1-2 gm) test samples are completely incinerated in air 


Table 2 

Volatile Yields from Bituminous Coal Under Various Conditions 

Yield (wt 

% of coal , 



1 atm He 

69 atm He 

69 atm H^ 

























Other HC gases 




Light HC liquids 












Heating rate, 1000°C/s; 

average particle 



J. B. Howard in "Chem. of Coal Utilization, 2nd Suppl. Volume, 
M. A. Elliot, Ed. (John Wiley and Sons, New York, 1981). 


at 725 + 25°C (AS1M D 3174-77). Ash consists of the minera:! material 
in coal which is non-ccxnbustible. Table 3 lists minerals found in coal. 
Typically, mineral matter is randomly distributed in coal as approximately 
2 ym inclusions. In addition to mineral matter, 20-30 trace metals are 
distributed through the coal. Some metals (boron) are organically 
bound to the coal molecules; others (zirconium, manganese) form inorganic 
bonds with mineral matter, while others (copper) occur in both the organic 
and inorganic forms. Concentrations of 5-500 ppn are typical for trace 
metals, however some appear at higher levels. Table 4 lists trace 
metals found in coal and typical concentrations. 

The calorific value Q, which is a complex function of coal elemental 
composition and rank is usually obtained as part of the proximate analysis 
of the coal. This is obtained by combusting a weighed sample, usually 
1-2 gm, under oxygen (at '^,25 MPa) in a bomb calorimeter. For adiabatic 
measurement (ASTM D 2015) the temperature of the calorimeter jacket is 
continuously adjusted to approximate that of the calorimeter, while in 
the isothermal procedure (ASIM D 3 286) the jacket temperature is held 
constant and a correction for heat transfer fron the calorimeter is 
applied o In either form of measurement the recorded value of Q is the 
gross heat of combustion, since all water generated during the test 
renains in liquid form. The net value, which is more useful for practical 
purposes, because water would normally be allowed to evaporate, is 
obtained by subtracting 1030 Btu/lb of water. 


Table 3 
Minerals Found in Coal 



Clay minerals (aluminosilicates) 



KAl^CAlSi^O^p) (OH) 2 

Sulfide and sulfate minerals 








Carbonate minerals 
Calc ite 



Silicate minerals 




ZrSiO , 

Na(Mg,Fe),Al^ (BO,),Si^O„ 
J 6 3 3 D 8 

Oxide and hydroxide minerals 






Table 4 
Trace Metals in Coal 

Element ppm (Avg) 

Boron 600 

Germanium 500 

Bismuth 20 

Cobalt 300 

Nickel 700 

Zinc 200 

Lead 100 

Silver 2 

Tin 200 

Molybdenum 200 

Cadmium 5 

Beryllium 300 


Ultimate or elemental analysis is a quantitative determination of 
carbon, hydrogen nitrogen, sulfur (organic) , and oxygen, vAiich make up 
the coal substance, and is usually performed using oxidation, decomposition, 
and/or reduction methods (96, 97). Carbon and hydrogen are usually 
determined by variations of the conventional Liebig method (98) , i.e. 
by burning a 0.2-0.5 gm sample of coal in pure dry oxygen at 800 to 900°C 
and completely converting the combustion products to carbon dioxide and 
water by passing them through heated cupric oxide. The exit stream is 
then led over hot lead chromate and silver gauze, which removes all 
oxides of sulfur, and chlorine, and finally sent through absorbers which 
collect CO^ and H^O. Carbon and hydrogen contents are calculated frcrni 
the weight increases of the absorbents. Alternative sani-micro methods 
have been developed which allow for accurate analysis of 50 mg samples 
in a shorter period of time (99) . 

Nitrogen is determined using the Kjeldal-Gunning jJrocfedure 
in which the coal (1 gm) is digested with concentrated sulfuric acid 
and potassium sulfate in the presence of a suitable catalyst (e.g. 
mercury, a mercury or selenium salt, cobaltic oxide, or perchloric acid). 
The cooled solution is then made alkaline, and the liberated ammonia 
distilled into a standard boric or sulfuric acid solution fron which 
nitrogen (as NH^) is obtained by back-titration , (100) . 

Methods for quantitative determination of total (organic and inorganic) 
sulfur are based on conbustion of sulfur containing compounds to sulfate 
ions, which can be determined gravimetrically or volumetrically. The 
simplest and most useful (101) is the Eschka procedure (ASTM D 3177) 
which entails burning the coal sample with a 1:2 mixture of sodium 


carbonate and calcined magnesium carbonate in air at 800 +_ 25 C. The 
resultant sodium sulfate is then extracted with an acid or alkaline 
solution and precipitated as barium sulfate. In the alternative bomb 
combustion method (ASIM D 3177) the coal sample is decomposed with 
sodium peroxide or burned in oxygen at 2-3 MPa. In both cases the 
proportion of organically bound sulfur in the coal is obtained by 
separately determining inorganic (sulfate and pyritic) sulfur by the 
Powell and Parr method and subtracting this from the total sulfur 
content (102). 

Oxygen, the only other element in the organic coal substance is 

usually reported by difference as % oxygen = 100 - (%C+%H+%N+ 

% S^j-g) tiut this value reflects all the errors in previous measurements. 

Several procedures for direct oxygen analysis have been developed mainly 

using oxidation or reduction of the coal (103, 104). The most widely used 

direct method for oxygen is due to Schutze (104) and Unterzaucher (105) 

and is based on pyrolysis of the coal Ln nitrogen at 1100 °C. Volatile 

material released is then passed over carbon at 1100 to 1200°C, and the 

resultant CO is estimated. Recently it has become practical to determine 

oxygen content by neutron activation analysis using the reaction 

16 16 

O + n ^ N + p 

Oxygen concentrations in the coal are then obtained by counting gory 


radiation from the N in the sample and comparing the result with the 
count for a specimen of known oxygen content (106, 107). This technique 
has the advantage of being fast and very accurate, but if only organic 
oxygen is to be determined the sample must be carefully demineralized 
before being tested. 


Since moisture and ash are extraneous to the actual coal substance, 
analytical data can be expressed on several different bases in order to 

a) the ccrnposition of coal as received, air dried, or fully water- 
saturated coal (at capacity moisture) or 

b) the composition of dry coal: dry, ash-free (daf ) ; or dry, mineral- 
matter free (dmmf). 

To convert data for moist coal to the dry basis, it is necessary to 
multiply each measured quantity other than moisture by 100/(100-% H^O) e.g. 

% carbon, dry = 100 x % C/{100-% H^O) 
Likewise, to convert data for moist coal to the dry, ash-free basis both 
moisture and ash must be taken into account, such that 

% carbon, daf = 100 x %C/[(100-(% ash)] 
However, conversion to the dry, mineral-matter free basis is only possible 
if the composition of the mineral matter and the actual weight relation- 
ship between the original mineral matter and the ash produced frcm it 
are known. 

Coal Petrography 

Coal is a sedimentary organic rock and is a very complex, heterogeneous 
mixture of organic compounds and minerals. Rank is thus an over- 
simplification of the highly heterogeneous nature of the coal substance 
but is usefiiL to describe gross features. The study of coal as a rock 
is called coal petrology and the description or classification of coal 
as a rock is referred to as coal petrography. Coal is not a uniform 
mixture of carbon, hydrogen, oxygen, sulfur and other elements nor is 
it a simple uniform polyarcmatic polymeric substance. One of the features 
of coal is that it may be composed of alternating bands of material with 


different appearances. It may consist of shiny bands separated by dull 
bands or layers that are rich in mineral matter (predominantly clays) . 
The nature of these different type bands is dependent upon the various 
components in the plant material, such as wood, bark, sap leaves, etc., 
which were used in creation of the coal and the conditions of the coal 

The basic system of coal petrography used today was developed by 
M. S. Stopes, a British coal scientist. The nomenclature of Stopes 
began with a description of the different components of coal visible 
to the unaided eye, called lithotypes or rock types (108). Vi train 
is the most common component of coal and originates from wood or bark. 
This material forms the uniform, shiny blackbands commonly found in 
coal and is the main component of bright coals. Clarain is laminated, 
being composed of shiny and dull bands and has a silky luster. Its 
origin is variable. Durain is dull, nonref lecting , poorly laminated 
and of \'ariable composition and origin. It is the hardest component 
of coal and is a large component of dull coal. Fusain is mineral 
charcoal, the softest component of coal and is believed to originate 
from the same material as vitrain. The fate of this material before 
incorporation into the sediment which became coal is in doubt. Two 
theories predominate: one that it is the charred remains of wood from 
forest fires, and the other is that the material was oxidized by bacterial 
attack near the water line in peat swamps. Fusain is generally found 
in small amounts in association with other components or lithotypes. 


Later Stopes revised her system of nomenclature to include 
components visible on microscopic examination of polished sections 
of coal in reflected light (109) . The term macerals was suggested 
for these microscopic components of coal, in analogy to minerals in 
rocks. Coal could now be thought of as an aggregate of microscopically 
distinguishable, physically distinctive and chemically different 
substances called macerals and minerals, with the organic material 
contained in the macerals and the inorganic in the minerals. Macerals 
are the smallest subdivision commonly used and are often referred to 
by their group which can be identified by their -inite endings. Since 
macerals are derived from different types of substances and often are 
subjected to different conditions before burial, they were affected to 
different degrees by the coalif ication processes that occurred. As 
rank increases, differences in the properties of macerals become 
minimized particularly in the anthracitic range. 

In bituminous codls there are 3 types of maceral groups which car 
be broken up into individual types. The vitrinite group originates from 
plant structural elements, such as wood and bark, which are composed of 
lignin and cellulose. It is gray in reflected light and the reflectance 
varies significantly with the rank of the coal. Because most coals 
contain more than 50% (commonly 75-80%) vitrinite, the gross properties 
of coals reflect the properties of vitrinite. Three varieties of 
vitrinite have been recognized to be sufficiently different to warrant 
their differentiation in a petrographic analysis. The three varieties 
are vitrinite occurring as bands or lenses (also called telinite. 


banded or vitrinite A) , vitrinite occurring as a ground mass of 
attrial coal (collinate, matrix vitrinite, vitrinite B) and pseudo 
vitrinite. Vitrinite occurring as bands or lenses typically has 
higher reflectance than associated attrital vitrinite. This band form 
of vitrinite has less hydrogen and yields less volatile matter than 
the attrital vitrinite of the same coal (110) . 

Pseudo vitrinite may occur as either the band or the attrital 
form. It is recognized in the microscope by its slightly higher 
reflectance, slitted structure, remnant cell structure, and the presence 
of serrated edges. First noticed as unfused particles in coke, it was 
then sought and found in coal (111) . It ccmmonly comprises 15-25% 
of vitrinite in coal and may be as much as 50%. Vitrinite of bituminous 
rank coals is the fusible component that yields porous coke on pyrolysis, 
whereas pseudo vitrinite, ranges from fusible to nonfusible (inert). 
Although it has been shown that some pseudo vitrinite is inert in the 
coking process (111) , it is apparently reactive (converts to liquids) 
in liquefaction processes (112) . 

In bituminous coals the exinite or lipinite group are low-reflectance 
substances of high hydrogen content. Four major categories of exinite 
have been recognized, and their differentiation is based upon the materials 
of their origin rather than differences in chemical and physical properties. 
Sporinite is derived from the exines of spore and pollen grains eind is 
seen under the microscope as thin lenses typically measxiring about 2-3 
by 5-25 ym. Cutinite is derived from chemically similar material (cuticle) 
that coats leaves, needles, and some fruiting bodies; it commonly appears 
in microscopic preparations as elongate strips with serrated edges. 


Both sporinite and cutinite have highly polymerized, alicyclic structure, 
with a significantly higher proportion of . hydroaromatic hydrogen than 
is found in associated vitrinite (113). Resinite is derived from 
resinous plant materials. Resinites are materials of relatively low 
reflectance that occur as spherical or ovoid bodies, or as irregularly 
shaped masses. Resinites originate from hydrogen-rich substances, .... 
retain high hydrogen/carbon ratios in coals through the bituminous rank 
and conmonly have a higher hydrogen/carbon ratio than associated 
sporinites (114) . Alginite originates from oil-rich algae. It is 
relatively rare, being only found in coal layers formed under very wet 
conditions. The infrared (IR) spectra of alginites are similar to 
those of resinite, with strong nonaromatic carbon- hydrogen bonds and 
little evidence of aromaticity (115). Alginite exhibits low reflectance 
and is distinguished from other exinite macerals by its form, occurring 
as ovoid bodies with a crenulated border . 

The inertinite group, named for its relatively inert behavior 
on coking, comes from basically the same type of plant components as 
vitrinite, although this material was strongly affected by oxygen during 
the early stages of coalif ication. The reflectance of the inertinite 
groups is greater than that of the vitrinite group, with a reflectance 
of about 1-2.5%. The inertinite occurs ccsnmonly as discrete particles 
that except for a small proportion with reflectance near that of the 
vitrinite, contrast strikingly with the vitrinite. The vitrinite is 
readily distinguished frcm the inertinite due to its much larger abundance, 
its relatively gray appearnce, and its; form in bands. There are five 


classes of inert inite macerals commonly recognized. Sclerotinite is 
formed from the hard chitinous remains of fungi which were resistant 
to decay, and relatively reflectant in their natural state. This type 
maceral is usually found in younger coals. In its most common form 
sclerotinite occurs as round or ovoid bodies, commonly more than 2 ym 
in diameter, and more reflectant than associated vitrinite, but seldom 
exceeding a reflectance of about 1.5%. Often the bodies exhibit an 
internal cellular structure. Little is known about the properties of 
sclerotinite, except that its reflectance would indicate that it has 
a higher carbon content relative to associated vitrinite. 

Fusinite or fossil charcoal is believed to have been formed from 
woody material that was either burned or attacked by bacteria. Semi- 
fusinite is similar but differs in the degree of oxidation. Fusinite 
shows a distinct cellular structure and has a higher reflectivity than 
semifusinite and the cellular structure is not well defined. Fusinite 
and semifusinite both have lower hydrogen and higher carbon contents 
than vitrinite picked from the vitrain bands in the same coal. 

Particulate materials with reflectance greater than that of associated 
vitrinite and which do not exhibit cellular structure have been referred 
to by a number of different names. According to definitions found in 
ASTM D 2796 (95) if the maximum dimension of these materials is greater 
than 10 lim, they are called macr inite; and if less than 10 ym micr inite. 
Macrinite is a sediment of inert detritus, while micrinite is believed 
to be a degradation product of protoplasm. 

Microlithotypes were introduced to denote typical associations of 
macerals 50 ym or larger. This has been found to be operationally 


useful to recognize certain associations of macerals that occur in 
particles of crushed coal. Microlithotypes usually contain one to 
three macerals in varying proportions. The microlithotype concept is 
predicated on the observations that few coal particles are monomaceralic 
and that the reactions of individual coal particles in technological 
situations are often influenced by the interactions of their constituent 
macerals and minerals. Where whole particles or portions of particles 
more than 50 ym thick are uniform in their compositional characteristics 
the particle or zone can be classified as a specific type of coal. The 
classifications of microlithotypes according to the relative proportions 
of the macerals is given in Table 5. 

This identification of the maceral (and mineral) associations in 
coal particles is potentially useful. Especially in the areas of 
ccanbustion and liquefaction, the proximity of certain macerals and/or 
minerals could be as important as the overall maceral composition . 

Maceral carbon content (% carbon by weight) increases and the 
atonic hydrogen to carbon ratio (H/C) decreases in the order: exinite, 
vitrinite, inertinite (116) . Maceral behavior during devolatilization 
indicates the strong influence of petrography (117). Vitrinite is 
the plasticizing coke forming portion of the coal structure. Exinite 
fluidizes and decomposes to tars and gases while inertinite neither 
plasticizes nor devolatilizes . It has been observed that vitrinite 
enclosed within relatively inert materials (e.g. mineral matter, 
inertinite) is not readily liquified because of poor contact with the 
potential hydrogen-donating solvent. Recently, combustion efficiency 
has been found to be inversely related to inertinite content. 


Table 5 
Common Microlithotypes 

Microlithotype Maceral Groups 

Vi trite 



Vitrinite + inertinite 


Inertinite , mostly micrinite 


Inertinite, without micrinite 


Ex in it e 


Vitrinite + exinite 


Inertinite + exinite 


Vitrinite, exinite, inertinite 


Inertinite, exinite, vitrinite 


inertinite apparently being responsible for the carbon content of 
ash particulates (118) . 

Molecular Structure of Coal 
In order to understand the complex chemical behavior of coal 
many researchers have developed various methods (119) . Experiments 
designed to determine the structure and chemistry of coal have been 
carried on for decades. Much early work consisted of chemical decomposition 
of coal and solvent extraction to analyze deccanposition products in an 
attempt to understand the basic building blocks of coal structure. A major 
problem of these techniques was that the coal structure was destroyed 
and intermediate reactions may have taken place which could modify 
final products yielding erroneous structures. Since the reaction paths 
of coal oxidation and reduction were (and for the most part still are) 
unknown, relationships between reaction products and their precursors 
were usually speculative. In these circumstances, real progress in 
understcjiding the chemical structure of coal became possible with the 
development of better instr\imental techniques, especially non-destructive 
and modifying techniques such as X-ray diffraction, nuclear magnetic 
resonance spectrsocopy (NMR) , reflectance infrared spectroscopy (RIR) 
and other spectroscopic techniques. These new methods for investigating 
coal chemistry did not simplify matters, but they offered a means for 
studying coal without chemically altering it, providing much previously 
unobtainable information, and for the first time allowed verification of 
experimental data by cross-checking against data obained by other 


independent methods. Although current ideas of coal molecular 
structure are still rudimentary and more illustrative than definitive 
it is at least possible to outline the main features of molecular 
structure in coal and to identify some significant details at this 

Much of the work reported in the literature on coal chemistry 
relates specifically to vitrinite which, while the most abundant 
canponent of bitiaminous coal, is also the most homogeneous maceral. 
The structural chemistry of exinites and inertinites has been far 
less studied, and statements about molecular structure in coal must 
therefore be taken as referring only to a particular constitutent. 
However an analysis of evidence from chemical, spectroscopic and 
statistical studies leads to the conclusion that coal can be thought 
of as a natural mixed polymer similar to a complex synthetic copolymer, 
and that a statistically average coal molecule can be postulated which 
contains within itself all essential mclecular configurations 
characteristic of the rank of the coal. Individual coal molecules will 
differ in size and internal organization, so that their diverse 
cOTiponents (aromatic, aliphatic, etc.) will not always be present in 
the same proportions or arranged in the same pattern. And like an 
incompletely cross-linked three-dimensional copolymer, coal can be 
separated into fractions that differ from each other in physical 
properties and, depending on peripheral configurations surrounding the 
molecular core, may differ in chemical properties also. But in each 
case the statistically average molecule describes the entire collection, 


and to be consistent with the concept of progressive metamorphic 
development of coal (from lignite to anthracite) , the core structure 
of the average molecule, sub units linked by molecular bridges, changes 
relatively little upon increasing rank. 

Experimental evidence for a polymeric structure for an average 
coal molecule was given by X-ray diffraction spectra. Coal structures 
are much too small for single-crystal studies^ therefore, all X-ray 
diffraction interpretation is based upon powder techniques and procedures. 
The X-ray patterns of coal show diffuse peaks at the positions of the 
most prominent graphite bands, and the peaks become less diffuse with 
increasing rank. Although most coals produce patterns with a few diffuse 
lines, high rank anthracites and meta-anthracites show additional lines 
identical to those of graphite (120) . Exinite and some higher rank 
coals produce an Additional line, which may be quite strong, known as 
the gamma band. The diffuseness of tha X-ray pattern of coal has been 
attributed to particles in which the arrangement of carbon is that of 
a graphite crystal, but with extremely small size of the elemental 
crystallites (121) . The similarities to graphite have led to the 
interpretation that coal consists of very small, graphite like layers 
packed turbostratically (120) . This means that coal contains stacked 
aromatic layers which are roughly parallel and equi-distant, but with 
each layer having a completely random orientation in plane and about 
the layer normals. The gamma band is not related to the graphite structure, 
and its origin is not understood. Possible explanations are alicyclic 
structures (122) with parallel stacking or the irregular packing of .\ 
aromatic layers (123) . The gamma band is the most prominent feature 


of the exinite pattern. Further X-ray diffraction work (124-126) 
associated the diffraction maxima with partly ordered aromatic nuclei, 
the average diameter of these structures in coals with 80-90% carbon 
is now considered to be no larger than 7-7.5 S (equivalent to 15-18 
carbon atoms or 2-3 condensed benzene rings). On average, only 2-4 
nuclei appear to be vertically stacked into crystallites. The generalized 
average structure then can be deduced from X-ray data and the ultimate 
elemental composition of coal was one of units of small aromatic nuclei, 
containing 2-4 rings, linked by hydroaromatic, alicyclic and aliphatic 
structures and containing various peripheral function groups. Due to 
theirnature X-ray diffraction measurements cannot be used to fill in the 
remaining chemical structures obtained in a "typical" coal molecule - - 
only its backbone. 

In Table 6 are band assignments in the infrared spectrm of coal 
showing chemical structures responsible. The following features should 
be note-^d: 

1. A broad band with maximum near 33 00 cm is due to the OH 

groups appearing in the organic material since this absorption persists 


after drying the coal at 105 C for 24 hours (127) or even at these 
conditions in vacuum. That it consists essentially of phenolic OH 
seems to be unequivocally proven by the strong decrease in the intensity 
of this band and by bands which appear at other characteristic positions 
when the OH groups are acetylated. The position of this band indicates 
hydrogen bonded OH and must be regarded as an essential feature of the 
coal structure (128) . Strong intramolecular hj^drogen bonding cannot 
be excluded but, as seen from the spectra of model compounds, the 


Table 6 

Band Assignments in the IR Spectrum of Coal 
(adapted from ref. 119) 

Band Position 




2978 sh. 




phenolic 0H....0 
-OH str. -OH (Hydroperoxide) 
-NH str. >NH....N 

ar. CH str. 

CH^ str. 

CH str., CH str., al. CH str. 





ar. C=C str. 

C=0. . . .Ho- 
ar. C=C str. 



CH^ assym. def. 

CH^ scissor 

ar. C=C str. 










CH^ sym. def. 

C-0 str. (phenols) 
OH def. 

C -0-C str. 
ar ar 

C-0 str. (alcohols) 

C -0-C , str. 
ar al 

C ,-0-C , str. 
al al 

"aromatic bands" 



Table 5' 










ar. HCC rocking (single and 



/•'(^n rion cio/l Tirrc; i 



subst. benzene ring with isolated H 



subst. benzene ring with 2 neigh. H 



0 -subst. benzene ring 

ar. = aromatic; al. = aliphatic; str. = stretching; def. = deformation; 
sh. = shoulder; subst. = substituted. 


small intensity and extraordinary band width of the OH absorption in 
such structures make them generally difficult to detect (129) . Hydrogen 
bonded NH groups may also contribute to the intensity of the band at 
3300 cm but this can be at most a small amount since the total 
amount of nitrogen in coals is not large. 

2. There is no : doubt about the assignment of the specific bands 
occurring in the 3 000 to 2800 on ^ region. The very small intensity of 
the 3030 cm band suggests, on the assumption of high aromaticity in 
coals, that the aromatic systems present are highly substituted or 
highly cross linked or both (130) . The fact that the band near 2980 cm 
is only detected as a shoulder led to the conclusion that the CH^ content 
in coals may be small, but CH^ groups on aromatic systems absorb mainly 
at 2925 cm and 2860 cm Aliphatic CH^, CH^ and CH groups all 
contribute to the intensity of the two bands at 2925 and 2860 cm ^ but 
the dominant contribution may come from CH^ groups. At this time it is 
not possible to determine whether the CH^ groups are present in aliphatic 
chains, in hy.droaromatic structures or in cycloparaf f ins. 

3. There is a large very strong band centered at 1600 cm whose 

assignment ranains in doubt. The alternatives for this band are either 

that the band is due almost solely to the arcsnatic carbon skeleton (131) 

or that a significant conbribution arises from oxygen containing species 

in the coal structure (132, 133). The view that this band is due only 

to arcanatic C - C bonds and hence that a high concentration of arcxnatic 

carbon is present in coal is based on the following considerations: 

a) A very intense band at 1600 cm occurs also in high molecular 
weight aromatic petroleum fractions and even in those containing 
less than 1% oxygen. 


b) The 1600 cm" band is not influenced either in intensity or 
position by acetylation or oxidation of coals which manifests 
itself in a band near 1700 cm~l. 

c) The discovery that ether oxygen (as in diphenyl ether) or the 
oxygen in phenols may produce a marked increase in the intensity 
of absorption of the aromatic C-C bond (129) . 

d) The spectrum of a coal (89%C) has been compared with that of 

(1) a condensed pitch 

(2) a high molecular-weight pitch fraction 

(3) a pitch 

(4) a mixture of 25 model aromatic substances, and 

(5) a number of highly condensed aromatic hydrocarbons. 

It was found that the optical density of the band at 1600 cm in all 
of these spectra were about the same with the exception of the pitch. 
The view that this band is also caused by oxygen containing species is 
based on the following considerations: 

a) All compounds with strong intramolecular hydrogen bonding in 
the form of six-membered chelate rings (e.g. hydroxyquinones) 
or f ive-^nembered chelate rings possess very strong absorption 
at 1600 cm"l, and the presence of quinone oxygen in coals seems 
to be proved (134) . 

b) Huckel molecular orbital calculations have been used with a bond 
order concept to show that free carbonyls cannot contribute 

to the intensity of the band, but show that strongly hydrogen- 
bonded carbonyls with bond orders of 0.81-0.85 can be expected 
to absorb in the 1600 cm"-'- region (135) . 

4. The absorption at 1500 cm typical of benzene rings has 
been clearly established in lignites and subbituminous coals, but could 
not be observed in bituminous coals (136) . This may be explained 
by the fact that increased substitution on the benzene ring is know 
generally to weaken this band and the fact that as the degree of conden- 
sation increases this band is displaced to smaller wavenumbers and 
finally disappears in the strong absorption at 1450 cm (137). 


5. The strong absorption at 1450 cm may be due in part to CH^ 
groups, but contributions may originate from CH^ groups, aromatic 
C-C bonds, and strongly hydrogen-bonded OH groups. 

6. From the rather weak absorption at 1380 cm which can be 
assigned to CH groups, it was concluded that the CH content in 
coals should be relatively small, but investigations on model compounds 
have shown that the intensity of this band can vary a great deal. A 
contribution to this absorption may arise from CH^ groups in cyclic 

7. The broad and poorly resolved absorption between 1300 and 
1000 cm""'" has its origin in the C-0 stretching vibration of various 
ethers but may also contain some aliphatic and aromatic HCC wagging 

8. The three bands between 900 and 700 cm "'' which are characteristic 
of coals of more than 81% carlon are mainly due to aromatic HCC rocking 
vibrations in benzene and condensed aromatic ring systems. Contributions 
are also to be expected from the aliphatic HCC rocking motion and may 
occur from C=C-C in plane deformation vibrations of the carbon skeleton. 
The bands between 900 and 700 cm ^ have also been interpreted as 
resulting only frcxn individual benzene rings substituted in various ways 
(129) as indicated in Table 6. 

9. Frcm the absence of distinct absorption bands near 700 cm ^, 
and the fact that a thin section of a coal studied between 700 and 900 cm 
showed no infrared active bands , it has been concluded that no long 
aliphatic chains are present in coal (129) . 

These observations have been deduced from experimental evidence 
obtained from many investigations into the infrared spectrum of coals 


and an abundance of these spectra have been published in the literature 
(129, 131-133, 135). Other spectroscopic investigations (138) and 
wet chemical methods including chemical and electrochemical reduction, 
catalytic dehydroge nation, reductive alkylation, oxidation, acid-catalyzed 
depolymerization , alkylation and acylation (139) have been used to 
obtain a model average molecule of coal. In Figure 3 the basic type 
molecules and functional groups found in coal are shown. The fundamental 
carbon backbone structure consists of polynuclear and condensed aromatics 
such as biphenyl, nanthalene, triphenylene and phenanthrene. A 
considerable proportion of the potential hydrogen in the condensed 
aromatic units is substituted. The degree of aromatic substitution 
varies frcsn about 40% to 70% (140, 141) . The substituting hydrocarbon 
groups are primarily hydroaromatic ring junctions and to a varying 
extent aliphatic groups. The aliphatic groups include methyl and other 
alkyl groups as well as methylene, ehtylene, and other alkyl bridge 

Major oxygen functional groups are the hydroxyl (-0H) carboxyl 
(-COOH) and carbonyl (-C=0) with lower rank coals containing ether, 
quinone, methoxyl and heterocyclic oxygen structures. The nitrogen in 
coal is found in substituted aromatic structures such as ring substituted 
azines and pyridines also as amine groups attached to aromatic ring 


clusters. Organic sulfur appears in heterocyclic rings such as thio- j 
phenes and occurs as -SH as a substituent on aromatic rings. Sulfide ' 
can also appear as a linking group to connect arcxnatic clusters. 


Oxygen containing 

— OH — C — OH 

OH ^^3^° 



Nitrogen containing 

or -c ^ N 

Sulfur containing 



Figure 3. Basic Molecular Types in Coal. 


A typical bittirainous coal (Figure 4) is believed to consist of a 

series of aromatic/hydroaromatic clusters containing an average of 2-5 

rings per cluster and joined by aliphatic structures, such as methylene, 

ehtylene, etc., and also ether and sulfide linkages (142). This 

arrangement promotes a complex interlocking molecular structure similar 

to organic polymers. Since the clusters are only loosely connected 

by aliphatic linkages, clusters will appear on various planes and thus 

cross linking and development of an extensive pore structure are favored. 

Pore sizes in coals can range from 200 to 20 % in diameter. Aliphatic, 

hydroaromatic and heterocyclic bonds are most susceptible to bond 

breakage (142): therefore, during heating these structures bear the 

heaviest responsibility for devolatilization. During carbonization, 

aromatic carbon is primarily responsible for char formation, while tar 

comes mainly from hydroaromatic carbon with aliphatic carbon producing 

CH^, CO, CO^, , and higher straight chain hydrocarbons. Chars, the 

devolatilized portion of the coal, are characterized by highly carbon-rich, 

polynuclear aromatic structures. In such structures, edge carbons are 

at least an order of magnitude more reactive than basal carbon due to 

the availability of unsaturated chemical bonds (143) . 

Comtxistion of Pulverized Coal Particles in Laminar Pre-mixed 


The burning of pulverized coal particles in a flame is a conplex 
mixture of chemical and physical processes. Upon entering the flame a 
coal particle will heat up until reaching a point where it will undergo 
devolatilization, the generation of volatile material from the re- 
arrangement of chemical bonds in the coal due to the influence of heating, . 


Figure 4. 

Proposed coal molecule. 


Upon forming, the volatiles will leave the coal particle, which is now 
in a plastic state, causing structural changes to the coal particles. 
One change is the formation of pores with increased diameters and the 
formation of cenospheres, hollow spheres formed when coal particles 
are rapidly heated. Their mass is concentrated in an outer shell, 
rather than being uniformly distributed throughout the particle, so 
that internal combustions reactions can occur along with external surface 
reactions. Once the volatiles have left the coal pcirticle they react 
homogeneously in the atmosphere surrounding the coal particle to form 
hot combustion products. Once the volatiles are reacted, or during 
this process, the remaining solid coal particle, char, undergoes hetero- 
geneous canbustion reactions until the carbon is totally consumed and 
only the non-combustible ash remains. 

The nature of these steps and the speed at which they occur are 
greatly dependent upon the size of the coal particles and its general 
classification according to rank. 
Generation and Ccmbustion of Volatile Matter 

When coal is heated at a conventional rate (0.05°C/s) , it appears 
to decompose, emitting vapors and liquid starting at a temperature 
around 350-400°C and the products consist of a carbon-rich residue (char) 
and ahydrogen-rich volatile fraction. This decomposition continues 
until a temperature of around 950°C is reached which if maintained for 
an extended time results in a residue of nearly pure carbon, with a 
structure resembling that of graphite. The accmulated volatiles are 
composed of various gases and liquids whose proportions depend on the 
coal type and the manner of heating. 


Given (142) postulated that the volatiles would be created during 
pyrolysis in the following steps: 

1) a low temperature (400-500°C) loss of hydroxyl groups 

2) dehydrogenation of some of the hydroaromatic structures 

3) scission of the coal molecules at the methylene bridges, and 

4) rupture of the aclicyclic rings. 

Wiser et al. (143) postulated a sequence of events initiated by the 

formation of free radical species from a thermal cracking of the linkages 

forming the aromatic clusters. These free radical species would then 

stabilize through a .rearrangement of atoms within a fragment or by 

colliding and reacting with another species. The resultant structure, 

depending on its vapor pressure, would either evolve as volatile material 

or remain as part of the remaining char. Typical volatiles evolved upon 

heating coal are listed in Table 2 (119) . 

In the early 1960's various researchers found that rapid heating 

(1000-10 C/s) techniques for coal permitted substantially more volatiles 
to be generated than the traditional slow heating methods used in proximate 
analysis (144-147) and that the liquid and tar fractions were the most 
strongly influenced. The effective volatile content of a coal must be 
distinguished frcm volatile matter (VM) , determined by proximate analysis, 
when the conditions of combustion differ significantly from those of the 
proximate test. It is incorrect to associate VM with the potential 
yield of volatiles from pulverized coal particles widely dispersed in 
a hot gas. The volatiles escaping from particles in the packed bed 
used in proximate analysis undergo secondary reactions including cracking 
and carbon deposition on solid surfaces, and the extent of such reactions 
within the bed is not well established. 


Badzoich and Hawksley (147) observed that the yield of volatiles 


frcm pulverized coal particles entrained and rapidly heated to 950 C 

in a preheated inert gas (N^) exceeded the loss of proximate volatile 

matter found frcsn the difference between the proximate volatile matter 

of coal and that of char. The results for several different coals 

were correlated by the empirical relation' 

V* = Q (1-VM )VM 

where V*, expressed as weight fraction of the original coal, is the 

ultimate yield of volatiles achieved by heating the coal for an infinite 

time (t = «>) , approximated by measurements at long residence times; 

Q is the ratio of total weight loss to loss of proximate volatile 

matter; and VM is the fraction of the initial proximate volatile matter 

remaining in the char at t = <». Experimental values of Q varied from 
1.3 to 1.8 depending on coal type, which led to values of V*/VM 
greater than 1. Kimber and Gray (148) extended the measurements to 
2000°C and found Q values from 1.3 to 1.95. The observation of volatile 
yields exceeding proximate volatile matter is consistent with the results 
of other investigators who rapidly heated pulverized coal particles 
dispersed in gas or held in monolayers or thin beds (145, 149) . 

Earlier work (143, 150) seemed to indicate that the yield of 
pyrolysis products, for reasonably long experimental times, beccxnes 
independent of tanperature above about 1000°C. However recent data 
(148,151, 152) obtained at higher tanperatures suggest that additional 
devolatilization in relatively short times may be observed by heating 
the coal to temperatures well beyond 1000°C. This conclusion seens 
reasonable because the absolute time required for a given degree of 


ccxnpletion of devolatilization is much less for the higher temperatures. 

o 0 

For example, the heating of a coal frctn 1000 to 1900 C might produce 


in fractions of a second an additional weight loss that at 1000 C could 
only be obtained in times too long to be measured in laboratory experiments. 
Also the slower rate of volatiles production frcm the coal particles at 
the lower temperature might present a larger opportunity for secondary 
cracking reactions and carbon deposition, which could eliminate the 
possibility of ever attaining the weight loss achievable at a higher 

The ccmplex molecular decomposition and transport phenomena 
involved in coal devolatilization and the relatively meager experimental 
data available on these processes have led to rather inexact theoretical 
descriptions of the devolatilization mechanism with many various simplify- 
ing assumptions called upon. Many authors have approximated the overall 
process as a first-order decomposition occurring uniformly throughout 
the particle. The rate of devolatilization is given by: 

g- = k (V* - V) (III-l-l) 


where V is the cumulative amount of volatiles produced up to time t, 

expressed as weight fraction of the initial coal, k is the rate constant, 

and V approaches V* as time becomes infinite. V* represents the 

effective volatile content of the coal which should not be confused 

with the VM content found by proximate analysis. The rate constant 

is correlated with temperature by an Arrhenius type expression.. 
— E/RT 

k = Ae (III-1-2) 
where A is the pre-exponential factor, E is the apparent activation 
energy, R is -the gas constant and T is the absolute temperature. 


Most kinetic studies have focused upon the determination of k and V*, 
finding little agreement on the observed rates of pyro lysis, with several 
orders of magnitude difference at a given temperature. 

Other authors have contended that a simple first order model is 
inadequate to describe experimental results. Stone et al. (150) improved 
the utility of Equation (II-l-l) by describing pyrolysis as a series 
of several first-order processes occurring in different time intervals, 
vAiich are dependent upon coal type and temperature of volatilization. 
Wiser et al. (143) used an nth-order expression to describe devolatilization: 

and found that n=2 gave the best fit to their data over the first 3500 s 
of weight loss, whereas n=l was better for longer times. Skylar et al. 
(153) used the equation of Wiser et al. (143) to fit nonisothermal 
pyrolysis data for different coals and found that values of n frcm 2 to 8 
were needed for good correlation. Pitt (154) used an empirical relation 

where A and B are constants, to correlate his pyrolysis data on different 
type coals. 

One of the serious problems with the above first order approximation 
is that it fails to describe the experimental result that the apparent 
value of V*, is a function of temperature, a fact that is neithfer 
mechanistically correct or mathematically treatable by these equations. 
As is reflected in the time zone theory of Stone et al. (150) the 
relatively slow rate of weight loss observed after extended times at a 
given temperature requires a set of parameter values that differ greatly 
from those that fit the behavior at short times and higher temperatures. 



V* - V 

= A-B log (t) 



Since coal devolatilization is not a single step process but rather is a 
multiplicity of overlapping decompositions, yielding volatiles in 
different time intervals for isothermal pyrolysis, or in different 
time and temperature intervals for the usual case of pyrolysis during 
heat up, any one set of parameter values for these equations cannot be 
expected to represent data accurately over a wide range of conditions. 

It had become evident that a multiple-reaction model was necessary 
to model the actual conditions occurring during devolatilization and 
that it should also be applicable to nonisothermal conditions which 
are representative of actual physical conditions of pulverized coal 
combustion. Neither wiser et al. (143) or Badzioch and Hawksley (147) 
considered the difficult integration involved in nonisothermal cases 
where heat up is not instantaneous, claiming little decern position 
had occurred during their rapid heat up and neglecting any that might 
have occurred. The problem of how to include, in kinetic studies, 
the weight lost during heat up had baffled many researchers attempting 
to evaluate coal devolatilization on an isothermal basis . 

Pitt (154) and later Rennhack (155) adapted Vand's (156) treatment 
of a large nimiber of independent, parallel rate processes to describe 
the pyrolysis of coal particles in a flame. The thermal decomposition 
of coal was assumed to consist of a large number of independent chanical 
reactions representing the rupture of various bonds, forming volatiles, 
within the coal particle. Differences in the strengths of chemical 
bonds throughout the coal molecule accounted for the occurrence of 
different reactions in different temperature intervals. Since the 
thermal decomposition of a single organic species is usually described 
as an irreversible reaction that is first-order in the amount of 


unreacted material ranaining, the rate of volatiles evolution 
originating from a particular reaction within the coal can be described 
in a similar manner to Equation (III-l-l) with i denoting one particular 
reaction out of many such that: 

— i=k. (V* - V.) (III-1-5) 
dt 1 1 1 

if k. is of the form of Equation (III-1-2) integration of Equation 
(III-1-5) for isothermal conditions yields. 

V* - = V* exp (-k t exp -E/RT)^ (III-1-6) 

for the amount of volatiles ranaining to be released. Values of k^, 
E. and V* cannot be predicted and must be estimated from experimental 
data, which become more ccxnplex as more reactions are used. The problem 
can be simplified if it is assumed that the k^'s differ only in activation 
energy and that the number of reactions is large enough to permit E 
to be expressed as a continuous distribution function f (E) with f(E)dE 
representing the fraction of the potential volatile loss V* that has an 
activation energy between E and E+dE. Then V| becomes a differential 
part of the total and may be written as 

dv* =V*f(E)dE (III-1-7) 



J f (E)dE = 1 (111-1-8) 


The total amount of volatile matter remaining to be released is obtained 
by summing the contribution from each reaction or by integrating 
Equation (111-1-6) over all Values of E using Ecjuation (III-1-7) 

^*V* ^ ? ^""P [■''o ^ (il)] ^^^^"^^ (III-1-9) 


Equations (III-1-6) and (III-1-9) can be generalized to allow for 

nonisothermal processes: 


V* - V. = V* - f k.dt (III-l-lO) 
111 1 

= J exp /- J kdt] f (E)dE (III-l-ll) 

V* - V ^ 

o * o 

Also, f (E) can be approximated as a Gaussian distribution with mean 
activation energy E^ and standard deviation a obtaining 

- (E-E ) 

f(E) =|o(2tt)""'| exp — (III-1-12) 


Assuming that k. = k . exp (-E./RT) with k , = k for all i, 
1 Ol 1 oi o 

Equations (III-l-ll) and (III-1-12) give 


V* - V 1 f 

CT(2Tr) ^ o 

(E-E ) 

-k I exp I -r- I dT - ° 

J exp(^)^ 

o J \ RT I" _ 2 

0 2a 



Equation (III-1-13) allows correlation of coal devolatilization data 

using four parameters (V*, E , a, k ) which is only one more than 

o o 

required by the single step model represented by equations (III-l-l) 
and (III-1-2), However the use of this slightly more complicated 
equation eliminates the temperature dependence of V* and allows data 
on a given coal under different sets of experimental conditions to be 
compared with one set of parameter values. The simple model required 
a different set of parameter values for each set of experimental conditions, 
hindering analysis of different sets of data on different coal types. 

An Alternative approach to modeling coal devolatilization has been 
to assime the existence of competitive reactions by which the decomposing 
coal may follow any number of reaction paths depending on the 


temperature - time pattern. One mcxiel of the devolatilization process 
developed by Kobayashi et al. (151) considers the coal to degrade into 
a residue (char) and a volatile component via two competing reactions 
forming char A and volatile A along with char B and volatile B via the 
following mechanism: 

Volatile A + char A 

Coal (C) ^^^^ ^A^ ^^A^ 

'k^ '^l ^'-"l^ 
^Volatile B + char B 

^B^ <^B^ 

The and are mass stoichiometric coefficients and the pseudo rate 
constants are given by 

^1 ^ ^1 (-E^/RT) 

k^ = exp (-E^/RT) 

where B^ and B^ are pseudo frequency factors and and E^ are pseudo 

activation energies. The A reaction is assumed to dominate at relatively 

low temperatures and leads to a certain volatiles yield. Reaction 

pathway B, which has a higher activation energy than path A (60.0 kcal/mole 

vs 17.6 kcal/mole), becomes more important at higher temperatures 

{> 1200°k) and also results in more volatiles than reaction A. The 

overall weight loss is given by 

V. + 

m__ . „ r" *» 

t,Vl = -^ ^= J (a k + a k )le 

m^ _ 1 1 2 2 "^^ 

(k +k^)dt 

dt (III-1-14) 

where m^ is the d.a.f mass of the raw coal, and , m the mass of 

A ^B 

the generated volatiles. Thus this model accounts for the Observed 
increases in volatile yields with increasing temperature , and has been 
shown to accurately represent actual experimental results (151) . 


The use of multiple independent parallel first-order reactions 
to describe the evolution of a volatile component was presented by 
Suuberg et al. (157) who started with equation (III-1-5) and used a 
series of reactions i. Assuming that the rate constant for reaction i 
was given by k. = k . exp (-E./RT), the contribution of reaction 

1 Ol 1 

i to the rate of evolution of the product evolved up to time t for 

isothermal conditions is given by: 


dV./dt = V* k . exp I — ^ -k . t exp 
1 1 Ol \ RT Ol 

I k , t exp I 

I RT Ol ^ RT / 

. = V* [^1-exp (-k^. t exp ^) J 


Here V . is the amount of volatile product evolved from reaction i at 
the time t, and V* is the value of V . at t = «>. The total yield of 

volatiles, the yield measured experimentally, is given by 

V„ T = ^ 'V. 
Total J. 


If the coal is heated at a constant rate dT/dt = m to a temperature T 
and maintained for a holding time t at this temperature, the evolution 
rate and amount of volatiles evolved becomes: 

dV . 

— ' — = V* k , exp 

dt i oi ^ 



— - k . + T I exp -— - 

T oilmE. I RT 

V. = V* « 
1 1 


j_ Ol \ mE^ / 




where m>o. 

It should be noted that the predicted behavior occurring during 

heating at a constant rate is equivalent to that of an isothermal process 

of duration RT /mE at the peak temperature attained. Since RT/E<<1, 
the equivalent isothermal time at the peak tenperature is much less 


than the time to heat up, which is of the order of T/m. Also the 

extent to which most of the reaction occurs at or near the peak 


temperature increases with decreasing RT /mE. 

Derivation of kinetic parameters for the evolution of individual 
products of coal not only is an effective modeling technique, but 
offers the potential of helping to identify the deccanposition reactions 
responsible for each product. However progress towards understanding 
the complete chemical and physical processes occurring is hindered by 
the large disagreements in parameters determined by different authors 
(150-157) . 

One reason for the discrepancy in the kinetic parameters calculated 

may be caused by secondary reactions occurring during volatile formation 

(158). Anthony et al. (146) derived a selectivity expression describing 

assumed competition between diffusional escape and secondairy reactions 

in developing an approximate model for the effect of secondary reactions 

on the production of volatiles. In this model they defined two types 

of volatiles, reactive V and nonreactive V . In an inert atmosphere 

r nr 

the material balance on the reactive volatile species present in a coal 
particle is 

^ - k (C-C )-k.C = 1^ (III-1-20) 

dt c " 1 dt 

where dv'^/dt, the foinnation rate of reactive volatiles expressed as 
mass fraction per unit time, represents ein irreversible decomposition 
reaction dependent on temperature and unaffected by the atmosphere 
containing the decomposing coal particle. C and C^ are vapor phase 
concentrations of reactive volatiles in the void volume of the coal 


particle and in the ambient gas, respectively, expressed as mass per unit 

mass of original coal, K is the overall mass transfer coefficient (s ^) 


and k. is the overall rate constant (s ''') of the deposition reaction, 


assumed to be first order. In the above equation the coal particle 
is assumed to possess a fixed internal void space, caused by pores and 
cracks, of uniform composition and temperature. Reactive species released 
by thermal decomposition of the coal particle enter this void volume and 
leave either by mass transfer to the surrounding gas or by solid deposition. 
Since the mass of the reactive species residing in the vapor phase within 
the particle is relatively negligible a pseudo-steady state condition 
exists so that dC/dt =0. If C ^ O, the above equation becomes 

-rr- - K C - k.C = O (III-1-21) 
dt c 1 

Since the loss of reactive volatiles from the particle is V\ then 

dV /dt = K C and from equation (III-1-21) 
r c 

dV = dV*^ 

' (l^k VK ) (III-1-22) 
i c 

Assxaning that K^/K^ is independent of the temperature- time: history, 
integration of equation {III-1-22) fran t = O to t = « gives 
V* = V** 

r _r (III-1-23) 

(l+k./K ) 
X c 

where V** is the potential ultimate yield of reactive volatiles with no 

mass deposition at infinitely long reaction time, likewise V^* is the 

actual ultimate yield of reactive volatiles at infinitely long reaction 

time. The mass transfer coefficient is assumed to be proportional to the 

diffusion coefficient of the volatiles, so that K = K /P, where p is 

c c 


total pressxire. Since V* and V^^* are the ultimate yields of total 

volatiles and .nonreactive volatiles and V* = V* + V *, using 

r nr 

equation (III-1-23) one obtains: 

V* = V * + V** (1 + k. p/K )"■'" (III-1-24) 
nr r i c 

this relation predicts the approach of V* asymptotically to (V^^* + V**) 

and V^^* at low and high pressures, respectively. The variation with 

pressure of the observed yields of total volatiles obtained experimentally 

(146) is consistent with this asymptotic behavior. 

Although equation (III-1-24) is consistent with the observed effects 

of pressure, the underlying description of the complex mass transfer 

and secondary reactions is highly simplified and the mechanistic details 

are concealed in the ratio k,/K . The escape of volatiles frcm a 

1 c 

particle is not a simple diffusion process but a complex process involving 
a time-dependent pore structure and hydrodynamic flow during the period 
of rapid devolatilization. To account for this behavior, Lewellen (159) 
developed a bubble transport model for volatiles flow in coal particles. 
The model describes the nucleation, growth, coalescence, and escape of 
volatiles-f illed bubbles under the influence of viscous forces, pressure, 
and surface tension. Mass is added to the bubbles by volatile formation 
and lost through mass deposition accompanying secondary reactions at 
the bubble surface. 

General observations tliat can be made from devolatilization studies 
are (143-153) . 

1) A high heating rate will promote the creation of volatiles in 
excess of the amount found by proximate analysis. 


2) A high final temperature (> 1000 K) will also increase the 
volatile production as well as increase the speed of formation 
with greater yields found at higher temperatures (>1500 K) . 

3) Pyrolysis of coal particles in the presence of hydrogen results 
in a higher yield of volatiles at a much faster rate than 
pyrolysis in inert atmospheres (160) . The species that react 
with hydrogen during pyrolysis are not clearly identified and 
are referred to in the literature as rapid-rate carbon. 
Evidence suggests that hydrogen interferes with the char forming 
secondary reactions, presumably by hydrogenating reactive 
fragments to form light hydrocarbons sufficiently stable to 
avoid the secondary char-forming reactions. 

4) The size of the coal particle determines the amount of volatiles 
produced with small particles (<_200 ym) yielding more % by 
weight loss than large particles (> 200 ym) . Ubhayaker et al. 
(152) measured volatiles produced by coal particles of 20 ym 
and 100 ym in a hot gas at a temperature of 1700°C and found 
that the yield of volatiles from the small particles exceeded 
that from the large by 14 weight % of the coal after 7 ms. 
Greater differences have been observed for larger particles (161) 
and this behavior apparently reflects an increased extent of 
secondary reactions and carbon deposition for larger particles, 
which is consistent with the idea that larger particles offer 
more resistance to the escape of volatiles giving more time for 
secondary reactions to occur. 

Once the volatiles have been formed and leave the coal particle in 

a normal flame environment, they will start to react with the oxidizing 

atmosphere surrounding the particle and begin to burn. The combustion 

reaction scheme that occurs is very similar to the reactions present in 

the methane-air flame presented in Chapter II. Due to the fact that the 

majority of the volatiles are CH^, H^, and CO, no modifications of this 

reaction schme are deoned necessary by this researcher or others 

(160-163). One consequence of the devDlatJlizatioh process is that it can 

not always be considered separately from the heterogeneous reaction of 

the solid coal particles. The flux of volatiles leaving a coal particle 

can be large enough to reduce the rate of carbon consumption at the surface. 


If the rate of volatiles evolution is slow, then oxygen will readily 
reach the particle surface to react. On the other.hand at high rates 
of devolatilization, the gaseous volatiles can effectively screen 
the coal particle from oxygen attack. The volatiles leave the particle 
surface and subsequently burn in a flame layer which surrounds the 
particle. As the devolatilization decreases, the flame front recedes 
towards the surface of the particle and eventually stabilizes there, 
allowing both heterogeneous and homogeneous reactions to occur 
simultaneously. For any given type of coal, the rate of volatile 
production has been found to vary with particle size. For bituminous 
coal particles, Howard and Essenhigh (164) found that particles larger 
than about 65 ym did not react heterogeneous ly during the period of 
rapid devolatilization. 

Combustion of Solid Devolatilized Coal particles (Char) 

The combustion of a devolatilized coal particle (char) can be 
considered to occur in the following steps: 

1) Diffusion of the oxidant molecules to the surface of the coal 

2) Transfer of the oxidant molecules from the gaseous phase to the 
surface of the particle. 

3) Chemical reaction of the oxidant with the char surface. 

4) Transfer of the products back to the bulk of the gas. 

At any time in the combustion of coal particle the mass rate of 
consumption of carbon per unit area of external surface of the particle, 
R^, is given by the expression. 

R + 1/ ( - + i ) 

T R ^ R V. 

mt chera 

where R is the rate which would be obseirved if the oxidant molecules 

were to react to form a gaseous product instantly on reaching the 


particle surface and R ^ is the corresponding rate if the oxidant 


molecules were to encounter negligible resistance in their transfer 
to the surface but take a finite time to react. R^^ and ^^^^ thus 
represent the limiting mass-transfer-controlled and chaaically-con trolled 
rates respectively. 

Early workers such as Nusselt (165), proposed that the oxidation 
of a coal particle was controlled by diffusion of oxygen through a 
stationary film to the surface of the particle where it reacts to form 
CO and/ or CO^. A double film model was proposed by Burke and Schuman 
(166) who postulated that carbon reacts at the surface of the particle 
with CO^ producing CO which is burned in a thin flame front in the 
boundary layer. No oxygen will reach the particle surface and no CO 
will reach the boundary layer in this theory and the mass flows are the 
rate limiting step. These theories also assumed that the coal particle 
had a solid smooth surface and internal reactions were neglected. 

Today, there is now general agreement that reaction control by 
diffusion is limited to coal particles above about 100 ym in diameter 
and that the reaction is partially or totally internal (167) . Opinions 
are divided whether the particles burn at constant diameter or with 
reducing diameter but all agree that the temperature coefficient of 
the chemically-dominated reaction is high, with activation energies 
typically quoted in the region of 20 to 60 Kcal. While it has been 
shown that the rate determining step is a chemical one, the question 
becomes whether it is the absorption or desorption step that limits the 
rate of reaction. 


An understanding of the overall kinetics of char-gas reactions 
requires an appreciation for the detailed interactions between gas and 
solid. These interactions will be discussed in a normal pulverized coal 
flame where the oxidizing molecules are obtained from air. In early 
models of gas surface reactions it was common to assume that the surface 
is more or less uniform, which is far from being the case for combustion 
of a char particle which contains nearly 25-50% of its volume in pores, 
dependent upon the coal type (168) . 

The simple concept of a gas absorbing at a solid surface to form 
an adsorbed layer which then decomposes after a finite time to yield 
reaction products has been conceptually elaborated in a number of different 
ways in the past (169, 170). There are however, only two alternatives 
for initial adsorption step on coal carbon: 

1) That the ambient raolcules (O^, CO^, H^O, H^, CO, etc.) present 
or formed by reactions) adsorb as molecules or, 

2) That these molecules dissociate as th^ adsorb. 
Investigations during the 1970' s have led to the consensus that molecular 
chemisorption does not occur and that all chonisorption is dissociative, 
even if mobile. Adsorption of these species is not uniform over the 
surface of the particle with preferential adsorption occurring on the 
edge carbons of the coal molecules with basal carbon atoms showing little 
or no activity for adsorption. When measuring surface area of coal 
particles it is necessary to distinguish between active and inactive 
areas. The total surface area (TSA) can be divided into three regions: 

1) The basal planes, where there is little or no reaction occurring, 
and no chemisorption film forming. 

2) The reacting chemically active surface (RASA) where there is 
again little or no chemisorption film because reaction is so 
fast that the residence time of the oxygen atoms is negligible. 


3) The unreacting chemically active surface (UASA) , which is 

covered by adsorbed oxide film, where the reaction initially 
takes place. 

The evidence for the absence of molecular chenisorption is partly 

direct and partly inferential. Evidence frcxn isotropic tracer studies 

14 13 
using C (171-173) and C (174) show that CO is not chemisorbed 

to any important extent while oxygen isotope studies (175-177) likewise 

have shown that neither nor CO^ chemisorbs as molecules. Walker 

et al. (177) studied nine possible reactions for the formation of CO^ 

using mixtures of oxygen containing 0-0, 0-0 and O- 0 in 

16 18 

different proportions reacting with carbon. Oxygen with 2.6% 0- O 

16 18 

was found to produce 50-52% of C 0 O, matching the percentages of 
16 18 

the O and 0 percentages in the original mixture. If molecular 
chanisorption and reaction had occurred the percentage in the mixed 
isotope CO^ should be 2.6%. Bonnetain et al. (178) found that the 
volumes of CO and CO^ produced varied linearly with each other and also 
concluded that groups of molecules desorbed together on account of 
electronic interaction in the graphite surface. The dissociative chemi- 
sorption of has not been explicitly dononstrated; rather it is implied from 
the agreement with dissociative adsorption isotherms (179). 

The finding that molecular chemisorption does not occur allows 
one to reduce the large set of multiple eqiaations considered by past 
reviewers (e.g Walker et al. (180): 14 reactions) . Another simplifying 
assumption that can be made is that the oxide film formed by dissociative 
adsorption has the same structure whether the O atoms are supplied by 
O^/ CO^j H^O and possible OH. The film formed can thus be represented 
in each case as C (O) with no labeling of where the oxygen originated. 


as some writers have done in the past. No matter what its origin the 
C (0) should decompose in the same way. The net reaction steps for the 
adsorption and desorption process in oxygen transfer are now considered 

to be: 

2C + O ^ 2C (0) (III-2-1) 
F 2 



C + CO ^ C(0) + CO (III-2-2) 

F 2 





S ^2° i C{0) + (III-2-3) 


2Cp + 2C(H) (III-2-4) 




C (O) ^ CO + C_ (III-2-5) 

20(0) -> CO + C„ (III-2-6) 

C(0) CCO)-*' (III-2-7) 

where C indicates a free active site, C (O) represents chemisorbed 

oxygen atoms and the k's are rate constants. Reaction (III-2-1) 
represents dual site adsorption, reaction (111-2-6) represents dual 
site desorption, and reaction (III-2-7) represents a surface migration 
of an adsorbed oxygen atom to another site. Equation (III-2-1) also 


represents irreversible chemisorption of oxygen, with no film 
decomposition to recover free oxygen. 

Although the above reaction schane is a reasonable explanation 
of the overall reaction process, reactions such as (III-2-2) and 

(III-2-3) may be ccxnposed of several steps with the reverse action 
occurring by a different route than the foward reaction. In dis- 
sociative adsorption the gas may dissociate as it adsorbs, or before 
it adsorbs as an intermediate independent step. Thus, for equation 

(III-2-2) we may write; 

C + CO ^ C + O + CO ^ C (O) + CO i (III-2-2f) 
F 2 E 

the reverse reaction would have great difficulty retracing the same 

intermediate path and a more likely reverse path may be given by; 

C(0) + CO C{OOC) C_ + CO (III-2-2r) 

r 2 

which implies that equation (III-2-2) should be written as two independent, 
irreversible reactions. For equation (III-2-3) it is very likely that 
the H^O will dissociate into OH, as suggested by Long and Sykes (179) 
who also postulated that molecular hydrogen would be formed on the surface 
as a result. In order to describe these occurrences equation (III-2-3) 
can be written as a reversible reaction sequence: 
C + H^O ^ C„ + OH + H -> C (OH) (H) 

F 2 -f- F -tr -(r 

C(0) + 2H J C(0) + H^ -(111-2-36) 
Molecular dissociation leading to dual site adsorption and single 
site desorption are probably the most important process occurring in 

the combustion of coal particles. In active site (C ) theory one 


usually presumes the following: 


1) Localized adsorption occurs via collisions with vacant active 
sites (Cp) ; 

2) One adsorbed molecule or atom per site due to strong valence bonds; 

3) A constant surface (chemisorptional migration/desorption) 

4) Surface coverage less than a complete moncmolecular layer. 
Compared to physical adsorption, chemisorption is characterized by a 
higher heat of adsorption, a slower rate on account of having an 
activation energy typical of any chemical reaction, and occurrence 

at much higher tanperatures than physical adsorption, which should only 
occur at temperatures close to the boiling point of the adsorbate due 
to the weaker forces involved (van der Waal's forces) . 

The mathematical model used to model the high tonperature gas-solid 
reactions occurring during char ccxnbustion was developed by Langmuir 
(for single site mechanisms) , and later extended by Hinshelwood to 
include dual site mechanisms. The following three assumptions were 
made in Langmuir" s derivation (169, 'l70). 

1) The adsorbed entities are attached on the surface of the 
aisorbent at definite, localized sites, through collision of 

gas molecules with vacant sites; 

2) Each site can accotmodate one and only one adsorbed entity; 

3) The energy of the adsorbed entity is the same at all sites 

on the surface, and is independent of the presence or absence 
of other adsorbed entities at neighboring sites. 

Langmuir assumed a dynamic equilibrium between adsorbate molecules 

in the gas phase at a pressure p, and the absorbed entities in the 

surface layerj the fraction of sites, covered equalisto- 0. The nvimber 

of molecules colliding with the adsorbent surface per unit area; per 

unit time is proportional to the pressure. The rate of adsorption 


is proportional to p (1-9) , and the rate of desorption is proportional 
to e only. At equilibrium these two rates are equal, such that 

p (l-e) = k_^e (111-2-8) 

where k is the rate constant for adsorption and k the rate of 

desorption. Hence, 

= _i_ p = bp (III-2-9) 
1-0 k_j^ ^ ^ 


e = (III-2-10) 


Equation (III-2-10) is commonly known as the Langmuir isotherm and 
is the most ccromon form used, although it conceals the significance 
of the constant b. 

The full kinetic derivation, given by Langmuir (181), and which 
is much more enlightening, will now be considered. For the general case 
of chonisorption at the surface of a solid the velocity of chanisorption 
will depend upon 1) the pressure, which determines the number of collisions 
with the surface; 2) the activation energy of chemisorption E^, which 
determines the fraction of the colliding molecules which possess enough 
energy to be chemisorbed ; 3) the fractional coverage of the surface 
f (e) , which is equal to (1-0) only when single site adsorption occurs, 
and 4) a steric factor or condensation coefficient a which may be 
defined as the fraction of the total nxjmber of colliding molecules 
possessing the necessary activation energy, E , that results in adsorp- 
tion. The velocity of chemisorption is therefore given by: 



where yTT ' ot)tained from the kinetic theory of gases, is 


the number of molecules of mass m striking each unit area of surface 
in unit time. Likewise the velocity of desorption, V_^, can be written 


V = k f'O) exp (—J) (III-2-12) 

where k ^ is the specific rate constant for desorption, f ' (9) 
represents an appropriate function of the fractional number of sites 
available for desorption (f ' (e)=9 when the desorption occurs from 
sites which cire occupied by a single adsorbed molecule) , and is 
the activation energy of desorption. 
At equilibriiom V^^ = V ^ so that 


k_^ / "■^d'*"^a \ 
p = (2TimkT) / exp ( — j 

f (e) 

Since the difference between the activation energies of desorption and 
adsorption equals the heat of adsorption, Q, for the special case when 
one moelcule occupies a single site on the adsorbent surface 
(f'(e) = e, f(e) = (l-e) ) ■ we obtain 

p = (2TimkT)-^/^ "J^^^P ^if^ I?^ (1II-2-13) 
If assumption 3 is valid, that the heat of adsorption Q is constant 
for all sites, then equation (III-2-13) can be written as 


(2iimkT)"^/^ exp (^) (III-2-15) 



Equation (III-2-14) is synonymous with the Langmuir equation (III-2-10) 
previously derived. 

Using statistical thermodynamics, the value of the constant b in 
the Langmuir equation has been shown to be given by (182, 183): 

(2Trm)^/^kT)^/\(T) , 
i = 2 exp iP^) (III-2-16) 

^ h^f (T) 


where f (T) and f (T) are the internal partition functions for a molecule 
g a 

in the gaseous and adsorbed state, h is Planck's constant and q' is the 

energy required to tranfer a molecule from the lowest adsorbed state to 

the lowest gaseous state, otherwise known as the heat of adsorption at 

absolute zero. The ratio k ^/a of the desorption and adsorption constants 

can be calculated using equations (III-2-15) and 1II-2-16) , provided 

that the values of f (T) and f (T) can be .Calculated. Evaluation of f (T) 

g a g 

is relatively easy for simple molecules (184) and f (T) is also obtainable, 
although its value varies according to whether the adsorbed molecule 
is mobile or immobile (185) . 

If, on adsorption, each molecule dissociates into two entities 
and each one occupies one site, the predominent mechanism in coal particle 
combustion, and Is: free to move from site to site (mobile) , the 
Langmuir equation becomes 

^=-^^2^ (III-2-17) 
1+ (bp)^^^ 

2 2 

under these circumstances, f (©) = (1-©) and f (©) = 9 . The Langmuir 
isotherm for two gases adsorbed simultaneously, a condition of high 
probability during combustion of a coal particle, is given by 


b p 

Q = L± (III-2-18) 



e = — O (III-2-19) 

B 1 + b^p^ + b^ 


where 6 and e refer to the fraction of sites covered by molecules of 
A B 

type A and type B, respectively, and b^ and b^ are the Langmuir constants 
for molecules A aixi B and p and p are their respective partial pressures. 

Taking pressures as being proportional to concentration C , and 
using equations (1X1-2-11) and (III-2-12) we can write equations for 
the intrinsic rates of adsorption ard desorption as: 

R = k C f (6) (III-2-20) 

a a a 


R = k^f (9) (III-2-21) 
d d 


where k , the rate constant for adsorption, equals , a/ {2-nm. kT) , with 
a a 

m the mass of the adsorbed species and k,, the rate constant for 
a d 

— E /RT ^ 
desorption equals k e d . The adsorption rate, R , can be measured 

— 1 a 

by exposing a clean surface (vacuum heat treatment) to the gas of interest, 


and the desorption rate R , can also be measured by rapidly imposing 


a vacuum on the sample (169, 170). 

For single site adsorption and desorption the rates are given by; 

R = k C (l-e) (III-2-22) 
a a a 


R = k^ e (III-2-23) 

d d 


If on adsorption each molecule dissociates and occupies two sites, 

and the adsorbed layer is mobile the rates of adsorption and desorption 

are given by 

R = k C (1-e)^ (III-2-24) 
a a a 


R, = k,©^ (III-2-25) 
d d 

Under steady state isothermal conditions, R = R » and using 

Si a 

equations (III-2-22) and (III-2-23) one obtains 

5 — = R, C (III-2-26) 
1-e k a 

where R^ = k^A^« Solving equation (III-2-26) for 9 one obtains 

1 ^ R^C 


which is identical to equation (III-2-10) with b = R^^. Substituting 
equation (III-2-26a) into equation (III-2-23) we obtain for the overall 


intrinsic surface rate R 

A A A ^ J ^ ^ 

R = R = = 1 A% (III-2-27) 
a d 1 + I^C^ 


R = kC™ (g - carbon/m^-s) (III-2-28) 


is the overall global rate for the removal of carbon, k is the global 

intrinsic rate constant, C is the local adsorbant gas concentration and 


is the true overall reaction order. Conparing equations (111-2-27) and 
(111-2-28), it can be seen that R will become either first or zeroth 
order depending upon temperature and concentration (partial pressure) . 
The two limiting cases are given by 


1. RC«1 -»-R = kC,m=l 

K a a a' 

2. RC »1 ->R = k^, m = 0 

K a d 

Thus if a global rate expression is fit to a system controlled by 
Langmuir kinetics the true order will fall in the range 0<_in<_l depending 
upon the concentrations and temperatures of interest. In general, 
adsorption control (first order kinetics) is promoted by lower reactant 
concentrations and higher temperatures. It should also be noted at 
this time that either first or zeroth order kinetics also arise from 
dual site adsorption with dual site desorption and for the situation 
of dual site adsorption with single site desorption the possibility of 
half -order kinetics arises in addition to first and zeroth order kinetics. 

In the case of a multiple component mixture adsorbing on a surface, 
the isotherm is given in equations (III-2-18) and III-2-19) and the 
overall rate, for species A, can be found to be 

^ . < 

^ = A A TT (III-2-29) 

1 + Rf + Rf 

k a k a 

where the superscripts A . and B denote quantities relating to those species. 
Fran this eqx»tion we see that the second species, which may be an 
inert., reactant or product gas, inhibits the reaction by taking up active 


Chgnical Reaction Mechanisms for Coal Char Combustion 

The principle global reactions occurring during coal char combustion 
are shown in Table 7. The kinetic mechanism of the carbon-carbon 
dioxide reaction has been studied by many various authors (172, 177, 
186-190) . The work of Ergun (17 2, 188) presents convincing evidence 
for the following oxygen-exchange mechanism: 


CO^ + t CO + C (O) 


C(0) CO + c„ 

where the number of free sites is assvmied to remain constant with burn 
off. The global reaction rate for loss of carbon is given by: 

R = k,-C 

If CO 

— ^ (III-2-30) 

"2 =° "2 '=°2 

The oxygen exchange mechanism suggests that CO inhibition of the reaction 

occurs not by adsorption, but rather via reaction between carbon monoxide 

and chemisorbed oxygen. Previous investigators (186) have thought that 

inhibition occurs by a simple process of filling active sites with 

carbon monoxide based on the following reaction mechanism: 

CO^ + C„ ^ CO + C (O) 
2 F 

C(0) CO + C^ 
CO + C^ -> C(CO) 


Table 7 
Principal Char Reactions 

1) Cp+CO^ ^ 2C0 

2) Cp+H^O CO+H^ 

3) C +2H ■> CH^ 

F 2 4 

4) Cp+O^ CO^ 


This reaction mechanism also produces a global rate equation con- 
sistent with equation (III-2-30), which successfully correlates 
existing experimental data for the carbon-carbon dioxide reaction (189) . 
However oxygen isotope tracer experiments (171-173) have shown that 
the absorption of CO can be considered to be negligible, eliminating 
this mechanism. It is now accepted that the inhibition of the reaction 
by carbon monoxide is caused by CO + C(0) -> CO^ + C^, 

Inspection of equation (III-2-3 0) indicates that the Langmuir- 
Hinshelwood mechanism postulated by Ergun (188), involves single site 
adsorption and desorption. However it should be recognized that the 
chemisorption reaction may actually be a dual site mechanism composed 
of more fundamental steps, such as 
CO^ + 2C ->■ C (CO) + C(0) 

C (CO) -> CO + C„ 

If C (CO) is a short lived intermediate, application of the steady state 

approximation generates the single site adsorption reaction and thus 

the mechanism remains single site. 

The reaction of coal char with steam which can be expressed by the 

stoichiometric equation, C = H O H + CO, has been shown to be 

F 2 2 

similar to the char reaction with CO^, both in mechanism cind relative 
rate. The principle mechanistic studies of the carbon-steam reaction 
have been reported by Ergun aind Mentser (191) , Gadsby et al. (192) , 
Blakely and Overholser (193), Strickland-Constable (194), Long and 
Sykes (179), Johnstone et al. (195) and Ergun (188). 

There is general agreement based upon experimental data that the 
rate of gasification of char by steam is given by an equation of the form: 


R = k^C^ ^ (III-2-31) 


1 + k C + k^C ^ 
2 3 H^O 

where the k's are functions of one or more elementary rate constants. 
The form of this equation is identical to that for the carbon-carbon 
dioxide reaction and it also predicts that hydrogen will act as an 
inhibitor in the same way that carbon monoxide does in the char-carbon 
dioxide reaction. This has been found to be true (179, 192) but does 
not predict the rate retardation due to carbon monoxide which has been 
observed by several workers (188, 191, 193, 194) . 

Two equivalent mechanisms have been postulated for the char-steam 

(A) H^O + -V + C (O) 

C (O) ^ CO + Cp 
+ C. ^ CiU^) 


(B) H^O + Cp ^ + C(0) 

C (O) ^ CO + C„ 


Inhibition in the first mechanism is based on hydrogen adsorption (179, 
192, 195) and in the second mechanism, on adsorbed oxygen exchange by 
CO (188, 194). The detailed steps occurring during pathway A have been 
discussed by Long and Sykes (179) who proposed that the steam molecule 
decomposes at the carbon surface into a hydrogen atom and hydroxyl radical 
both of v^iich chemisorb rapidly on adjacent carbon sites. This is 
followed by the hydrogen atom on the chemisorbed hydroxyl radical joining 
the hydrogen atom on the adjacent carbon atom which then leaves as a 


hydrogen molecule. Therefore the steps in reaction scheme A may 
be written as: 

2C + H O C (H) + C (OH) 

F 2 

C (H) + C (OH) C (H^) + C (O) 

C(0) CO 
C(E^) - + H^ 

Evidence for dissociation of the hydroxyl radical has been obtained 
by Blackwood and McTaggart (196) . The inhibition by carbon monoxide 
was thought to occur in a manner similar to the mechanism in the char- 
carbon dioxide reaction; i.e. 

CO + C (O) ^ CO^ + C„ 
-<- 2 P 

However Gadsby et al. (192) postulated that CO did not by itself 
inhibit gasification, but by shifting the water-gas equilibrium, 
H^O + CO J CO^ + H^ 

more hydrogen is produced, which does inhibit the char-steam reaction. 
Ingles (197) postulated that a carbon surface accelerates the water- 
gas shift reaction by acting as a chain initiator for the following 

C (H) ^ Cp + H 

H + HO ^ OH + H^ 
2 ^ 2 

OH + CO CO + H 

<- 2 

Strickland-Constable (194) , observing that hydrogen is not only strongly 
but very rapidly adsorbed on carbon, supported the view that the 
hydrogen inhibition in the carbon steam reaction is caused by its 
chemisorption on active sites. 


The reaction of char with hydrogen can be represented by the 
overall reaction 

There are few kinetic studies on the carbon-hydrogen reaction with 
the main mechanistic studies being reported by Blackwood (198, 199), 
Zielke and Gorin (200), Moseley and Paterson (201) and Johnson (202). 
These workers all used high pressure reactors (1-100 atm) , but even 
under these conditions the rates were very slow. Most studies report 
that the global rate has a simple first order dependence on hydrogen 
concentration (198, 201-203); i.e. 

R = k c (III-2-32) 

^ "2 

However, Zielke and Gorin (200) (T=925°C, P 30 atm) find 
" _ 2 

^ - '^i^H (III-2-33) 

while Blackwood (199) (T = 650-87 0°C, p = 4-50 atm) reports at high CH^ 

= (III-2-34) 

where the k's are functions of one or more elementary rate constants. 
Under appropriate conditions, equation (III-2-34) reduces to either 
equation (III-2-32) or equation (III-2-33). The following mechanism 
was postulated to explain the rate behavior of equations (III-2-33 and 
(III -2-34) : 


cm) + C„ 2C(H) 
Z c. 

2C{H) + ^ 2C(}i^) 

C (H ) + H ■> CH + C 
2 2 4 E 

The existence of C (H) was postulated since the reaction between 

carbon and atomic hydrogen also produces methane (196, 203). zielke 

and Gorin (200) also postulate that surface methylene (-CH^-) could 

replace C (H) ^ as an active intermediate by the following mechanism: 

H H 

\C = C k, ,C - 

V / \ ^ ^ > / \ . 

C=C P = C, C=C C = C 

c = c c = c 






H C 
3 \ 

C = 

.c = 

H H 
/ \ - 

c = c c = c 

/ \ / ^ +2CH 

c = c 

Blackwood (199) has shown that the global rate R is proportional 
to the oxygen content of a char sample with purified carbons having 
R 1^ 0. The oxygen content of a char particle remains constant during 

+ 2H„ ^ 


gasification, an indication of oxygen-based active sites. Addition 
of steam also accelerates the rate of the char-hydrogen reaction 
significantly (202, 204), probably due to C (OH) and C (H) reactions. 
Despite the simplicity of the char-oxygen reaction, 

the reaction mechanism is not well understood and disagreements still 
exist among investigators on some aspects of the problon (205). The 
major question concerning the mechanism of the carbon-oxygen reaction 
has been whether carbon dioxide is a primary product of the reaction 
of carbon with oxygen or a secondary product resulting from the gas- 
phase oxidation of carbon monoxide. The CO/CO^ question has been 
studied by many researchers (180, 206-209) with the general consensus 
being that both carbon monoxide and carbon dioxide are primary products, 
with the CO/CO^ ratio increasing substantially at higher temperatures 
and lower pressures. Experiments using oxygen-activated carbon clearly 
show that carbon monoxide is favored at higher temperatures (210) . A 
possible explanation for this behavior is that CO is formed at carbon 
edges while CO^ is formed at inorganic sites (205, 209). Lower 
temperatures favor CO^ due to catalytic activity while higher tempera- 
tures promote utilization of carbon edges. The CO/CO^ ratio can be 
correlated by (208, 209) 

CO/CO^ = Ae"^/^^ 

2.5 3.5 

where A [v^ 10 * , E 6-9 kcal/mole for low pressure, and A^;^ 10 ' , 

E 2^12-19 kcal/mole at high pressures. Arthur (208) showed that the 
ratio could be expressed, for temperatures between 750 and 1690 K, and 
for two different carbons, as 


CO/CXD^ = 2500 exp (-240/T) 
while Rossberg (211) reported similar results for temperatures between 
7 90 and 1690°K for two different carbons. 

At ordinary combustion temperatures, the predominant oxidation 
product still appears to be carbon monoxide which further oxidizes 
to carbon dioxide in the gas phase. According to von Fredersdorff and 
Elliott (206) this reaction is very rapid and occurs via the following 



0 + C_ -> C (O) + 0 
2 F 

0 + C_ C(0) 
C (O) -> CO + 


where the last reaction represents loss of free active sites at high 
combustion tanperatures due to thermal annealing. This oxygen mechanism 
is consistent with the observed extremely rapid rate of reaction between 
carbon and oxygen atoms (196) and the blue glow, attributed to 
CO + O CO^ + hv, often observed around burning carbon particles. 
Neglecting the fourth reaction and assuming a steady state concentration 
of C(0) we obtain for the global rate: 


^ " '^l^O (III-2-35) 


•^3 °2 

which allows for reaction orders of 0 and 1. However the speed of the 
reaction cannot be explained unless the active carbon sites are 


drastically greater for 0 compared to CO and HO. A mechanism 

such as: 

0(0^) + 2C0 

could account for the speed of the reaction but experimental evidence 
favors a C (0) mechanism, not C (O^) • Experimental evidence has shown 
that, in the temperature range of 1100-1600 K, the reaction order is 
one-half (205, 212) and more recent work indicates that for even 
larger tanperature ranges (650-1800 K) a half -order reaction mechanism 
applies (213, 214). Based upon these results a kinetic mechanism 
must allow for reaction' orders of^O, 1, and 0.5, a mechanism that is 
indicative of dissociative chemisorption with surface migration. 
Blyholder and Eyring (212) have recommended: 

where the C ' (0) represents a mobile ionic bond while C (O) represents 
an immobile covalent (carbonyl) bond. This mechanism is also consistent 
with quantum mechanical calculations which indicate that mobile 
dissociative two-site chemisorption is more favorable than single site 
molecular adsorption (215). Unfortunately this mechanism does not 
account for the rapidity of the C/O^ reaction at lower tanperatures and 
the reverse of the first reaction cannot occur since experimental evidence 
has shown that O^ adsorption is irreversible for temperatures greater 
than -7 0°C (17 0, 210) . 

O + 2C -> 2C'^0) 

Z if 

C (O) c (0) 

C (O) 

CO + c 


A possible alternative mechanism is. 

(1) + 2C^ -> 2C* (O) 

(2) C* (O) ^ C (O) 

(3) C(0) ^ CO + 


(4) C''(0) ^ CO + 

(5) C* (0) + C* (0) ^ CO^ + C^ 

Reaction 4 is postulated to have a lower activation energy than 

reaction 3 therefore allowing for faster conversion to CO at lower 

temperatures than reaction 3. Using oxygen isotopes Walker et al. 

(177) showed that reaction 5 was the mechanism responsible for primary 

CO formation and was able to eliminate reactions suggested by others 

such as 

0(0^) -> CO^ + Cp 

CO + C (O) CO + C 

Reaction 5 has also been found to be consistent with experimental 
results which show that CO^ is produced during the reaction between 
carbon and atomic oxygen (196) . 

Mathematical Descriptions of Coal Particle Combustion 

The theoretical description of coal combustion had its origin 
in the work of Faraday and Lydell (216) , whose volatiles-ignition 
theory was generally accepted for most of the 1800' s. According to 
this theory, coals must first devolatilize, partially or completely, 
before ignition can start. It was assumed that ignition of the solid 

12 5 

residue was in some way conditioned by prior ignition of volatiles. 
According to the volatiles theory, the inflammability of coals should 
increase roughly in proportion to the volatile content of the coal. 
Early work (217-219) appeared to support this theory but by the end 
of the century there was clear experimental evidence (220, 221) which 
disputed the volatile theory. 

The first mathematical treatment was presented by Nusselt in 1924 
(165) who introduced the single film model. In this model it was 
assumed that the oxidation of carbon is controlled by diffusion of 
oxygen through a stationary film to the surface of the carbon particle 
where it reacts to form CO or CO^. No other chemical reactions were 
considered and the flame propagation mechanism was based upon radiation 
of the burning particles. 

Cassal et al. (222, 223) modified the gas-phase Mallard-Le Chatelier 
burning velocity equation (equation (II-1-4) to include the effects 
of radiation on coal part.'.cles . This approach did not allow for quantita- 
tive predictions but it was capable of indicating relative effects of 
the various parameters. The modified euqation is 

4 4 

X(T^-T.) 6z efak(T^ -T ) 
V = + E f 2_ 


(p + a C^) (T. - T ) 
g p p p 1 o 

where e = emissivity of the coal particle surface , a = correction factor 

> 1, which accounts for radiation of glowing ccmbustion products (gas 

and solid, F = geometrical view factor, = concentration of dust, 

r = mass reaction rate of particle, p , and P , are the 
p ^ ' "^P' P g' P 


density and heat capacity of the particle and gas respectively, k is 
theBoltzmann constant, and X is the thermal conductivity of the gas. 
No ccsnparisons of experimental results have been made with this theory. 

The radiation theory of coal dust flame propagation was extended 
further by Essenhigh and Csaba (224) who accounted for temperature 
differences between particles and gas and added a finite pre-ignition 
zone. Assumptions included negligible combustion in the pre-ignition 
zone, a fixed ignition temperature, equal gas and particle velocities 
and particulate radiation from a grey-body flame at an average tempera- 
ture, T^. 

In this model, based upon a thermal theory, the details of the 
combustion process were neglected . The mixture was assumed to leave 
the burner at some temperature, T^, and in the pre-ignition zone, the 
coal was heated only by radiation. Because of the temperature dif- 
ferences between the gas and solid phases, some of the heat is lost 
by conduction to the gas, HDwever conduction from the flame to the 
incoming particles was neglected. At some point in the system the 
particles were ignited and this ignition temperature was fixed at the 
coal deccwiposition temperature (y 600 K) . This allowed the pre-ignition 
zone to be decoupled from the flame zone. The radiation intensity 
decayed exponentially away from the flame front due to absorption 
by the particles, and emitted and scattered radiation by the coal 
particles were neglected. 

The flame velocity for the case where the gases and particles 
are in thermal equilibrium is given by 




mt . 

1^(1 - e ^) 

vdaere m 

3 Z V _/4r p , and is called the cloud attenuation co 

p Bps 


and t. is the ignition time. I is the intensity of 


radiation frcra the flame at the point of ignition, which is evaluated 
at the mean flame temperature. This model predicted burning velocities 
varying frcsn 50 to 120 cm/s for stoichiometric mixtures of 50 and 10 pm 

theory was compared with experimental work, determining coal flame 
velocities as a function of temperature, using a coal dust concentration 
of about 0,3 kg/m (225). Flame temperatures were varied by changing 
the oxygen percentage, and although the magnitude of the burning velocity 
was not in good agreanent with the theory, the trend of increasing velocity 
with increasing temperature was found in both cases. It was also 
concluded from the experimental measurements that thin laminar flames 
were not likely dominated by radiative heat transfer. 

Bhaduri and Bandyopadhyay (226) advanced the radiative approach by 
incorporating the heat generated due to chemical reactions for the case 
of equal gas-particle temperatures and for a single coal size. Heat 
conduction was neglected and only heterogeneous oxidation of carbon in 
low volatile coal to CO^ was considered. They obtained a differential 
equation that required n\jmerical solutions for flame velocity, tonpera- 
ture profile and heat release along the flame: 

where Y, the radiation coefficient, equals 3Z /4r p , Q is the volumetric 
' p p s 

heat release due to coal char combustion and y is the distance in the 

diameter particles, assuming an ignition distance of 10 cm. This 

a D dT 

Yl£ exp[-Y(y^-y)]-{C^ + {— )+ Q = O 


Using this equation, Bhaduri and Bandyopadhyay predicted turning 
velocities for suspensions of anthracite coal and air. For 20% excess 
air, which yields a flame temperature of 2098 K, the burning velocity 
was found to decrease from 37 cm/s for 20 ym particles, to 34.6 cm/s 
for 100 ym particles. For a stoichiometric mixture of 50 ym pcirticles, 
with a predicted temperature of 2373 K, the predicted burning velocity 
was 50.0 cm/s. Flame thickness predicted by this equation were about 
0.10-0.15 m, which is greatly in excess of the experimental values 
for laminar coal-air flames. 

Marshall et al. (225) developed a simple theory of coal flame 
propagation which included radiative and conductive effects as well as 
including an approximate treatment of coal devolatilization. They 
postulated that the rate limiting steps were diffusional escape of 
volatiles from the coal, coupled with particulate radiation and gaseous 
conduction. Their work also indicated that the burning velocity should 
vary inversely as the square root of the coal particle diameter. Calculated 
flame velocities were found to be approximately a factor of two larger 
than experimental values and also predicted a slight increase in flame 
speed with decreasing particle size and a large increase in speed upon 
increasing oxygen concentration. 

Smoot and co-workers (227-230) have recently described the 
development of a generalized model for predicting flame propagation 
in laminar, particle-laden flames. This model specifically takes into 
account devolatilization and subsequent reactions of gas-phase species. 
This model has been based upon the numerical technique of Spalding 
et al. (51), used for laminar methane-air flames, vAiere the steady state 


solution for the propagating flame is obtained from the convergence 
of the unsteady-state conservation equations. Much of this model 
has been patterned after earlier solutions for propagating methane-air 
flames (53, 61) but has been generalized to include particle effects. 

In this model, basic unsteady-state conservation equations for 
laminar, multi-component, ccxnpressible gas/particle mixtures (231) 
were used to formulate a one-dimensional, propagating flame model. 
The coal particles were assumed to be spherical and sufficiently small 
and numerous enough to be treated as a continuous medium, while particle 
volume, particle diffusion and collisions among particles were neglected. 
Pressure was assumed to be uniform while the particle velocities were 
taken to be equal to the local gas velocity in the flame. Effects of 
gravity, viscous dissipation, forced diffusion, thermal diffusion, and 
temperature gradients within particles were neglected. Using Spalding's 
stream line co-ordinates (50) and earlier work of Smoot et al. (61) 
the following six equations were derived to describe coal-air laminar 
flames (230): 

1. Gas-species conservation 

P (310. /3t) = P 0/cHIj1 (P P D. /3\f)) + 
g ig t g t la ^ ' 

r. + Z (r .) .-(E r .)to. 

ig pi : p: ig 


2. Gas phase thermal energy 

P (3h /at) = 0/a<(;) [ (K P^/C ) (3h /3i|)) ] + 
g g t g t pg g 

^(Q .)-Q + S[r . (h . -h ) + P^(9/8if/){ 
D PD sg P3 pjg g t 

^[P D. -(K /C )] h. • Oo). /^ii)}- 

1 g ig g pg ig ig 

(p/P ) r ./P . ) 

g : PD PDg 

3. particle thermal energy 

P . Oh ./St) = Q - Q .- Q - 
PD PD Pf PD ps 

r . [ (h . - h . ) - (p/P . ) ] 

PD p:g p: ^ pgD 

4. Particle mass 

9m ./3t = -r ./n . 

5. Particle species conservation 

P (3(i) .)3t) = (0) . ^ r .)-r . 

6. Particle number balance 

P 3 .)/3t ^ n . [3p /at - ^r .] 
g npD - PD g D PD 


^/it = -P^V, a^j/Sly = P^ 


P^ = P + ^ P . 
t g D PD 

for the gas particle mixture. These six equations are a set of i + 4j 

+ 1 equations describing the dependent variables W. , h , h . , w . and n . 

1 g PD PD PD 

as functions of time (t) and position ii {t,y), where i = the number of 


gas phase chemical species and j = the number of particle phases 

(discrete sizes or types), with remaining symbols defined in Table 8 

These equations reduce to those of Spalding et al. (51) for gas-phase 

systems such as a normal laminar methane-air flame. 

Like the model for the methane-air flame (61) the model equations 

for coal-air flames also includes variables that must be calculated 

(i.e. thermal conductivities and binary diffusion coefficients) using 

auxiliary equations. Gas phase properties were calculated using 

procedures used by Smoot et al. (61) and described in Chapter II. The 

heat losses from the gas and the particles to the surroundings (Q^g* 2ps^ 

were neglected. Coal particles were assumed to be composed of specified 

amounts of char (carbon) and volatile matter (hydrocarbons, and other 

gases), whose values were obtained from experimental values (231). 

The volatiles part of the coal were assumed to react at a rate proportional 

to the amount of volatile materials remaining: 

r . = P /[ (d ^6 K*^) + 1/k ] 
V3 V p] V V 

where both activated devolatilization of the coal volatiles k = A exp 

V V 

— E 

( v/RT .) ""arKi the .. diffusion of the volatiles products from the 

surface (k = 2M D C v/p ) determine the net devolatilization rate. 
V V vm p s 

The char part of the coal particle was assumed to react with oxygen as 

a first order heterogeneous process to produce carbon monoxide 

r . = ttM d . c n .A/ [1/k + 1/k ] with oxygen diffusion 
CD c P3 og p]^ ' o ' c ^ 

k = [ (D /d .) (2B./exp B.-l)] and activated surface reactions 
o om p3 D D 

k = A exp (E /RT .) determining the net char reaction rate, 
c c c pj 


Table 8 

Symbols for Smoot et al. Coal Combustion Model (230) 


A pre -exponential factor, cm/sec, sec 

B. particle transpiration parameter 



C molar concentration, gmol/cm 

C heat capacity, cal/g°K 


d particle diameter, cm, y 

P 2 
D diffusivity, cm /sec 

E activation energy, cal/gmol°K 

partial molar enthalpy, cal/g 

Ah heat of reaction, cal/g 

k surface reaction rate coefficient, cm/sec 

k mass transfer coefficient, cm/sec 


equilibrium constant 

volatiles diffusion parameter, cm^/sec 

m particle mas, g 


M molecular weight, g/gmol 


particle number density, cm 
Nu Nusselt Number 

p static pressure 


Q heat transfer rate, cal/cm sec 

r mass reaction rate, g/cm"^sec 

R gas constant 

r mass react 

t time , sec 


T temperature , K 

V velocity, cm/sec 

y distance along f oame , cm 

^ .th 

0) mass fraction of i specie 


^ streamline coordinate, g/cm 

0 surface area factor 


P density, g/cm 



Table 8-Continued 

c char 

f flame 

g gas 

he hydrocarbon 

X 1 specie 

j 3 particle 


k k volatile specie 

, th 

k k gas phase reaction 

m product 

n reactant 

o oxygen 

p particle 

s solid 

t total 

V volatiles 


Swelling of the coal particles was assumed to be proportional 

to the extent of devolatilization. Coal or char particles were heated 

or cooled by conduction from the gas: 

O . = irN k d . (T .-T ) n . where 
PD u g PD p: g PD 

N = 2B./ (exp B. - 1) and 
u D D 

B. = r . C /2T7dp.k n . 
D PD Pg D g PD 

Products of devolatilization were specified as an arbitrary hydrocarbon, 

C H , together with oxygen, nitrogen, etc., and the carbon/hydrogen 
n m 

ratio in the hydrocarbon was obtained from material balance considera- 
tions. The rate of oxidation of the hydrocarbon to carbon monoxide 
and hydrogen was taken from Edelman et al. (232) and is given by 

The coal particle enthalpy was taken to be a weighted average of 

the enthalpy of the char and volatiles phases, while the enthalpy of 

the products of particle reaction is: 

h . = [r . (h . + Ah ) + r . (h . + Ah ) ] / (r . + r . ) 
pg3 c] c] c V] V3 v c] vj 

where r . = r . + r . 

PD c: V] 

The model equations were transformed following the approach of Spalding 
et al. (51) and were solved simultaneously using a numerical finite 
difference scheme to iterate to the steady state solution. 

Predictions were made for a monodisperse suspension of Pittsburgh 
coal particles which were assumed to contain 50% volatiles initially. 
For 33 ym particles, the predicted burning velocities increased from 
14 to 17.5 cm/s as the coal concentration increased from 125 to 625 mg/1. 


The predicted flame thicknesses were about 1 cm which agreed well with 
their experimental values. From these calculations Smoot et al. (228) 
conclude that coal pyrolysis and subsequent diffusion of the volatiles 
are the most important mechanism for this type flame, and it should 
be noted that the radiation propagation theory of Bhaduri and 
Bandyopadhyay (226) predicted flame thickness on the order of 10 to 
20 centimeters. 


Gas Metering and Flow Control System 

One of the most important parameters to control in any flame 

diagnostic experiment is that of gas flow rates into the burner 

undergoing study. The requirements for a gas metering system to 

obtain accurate results have been discussed by Mavrodineanu and 

Bolteux (15). Major parameters are accurate control of flow and 

constant delivery pressure. In order to insure accurate gas flow 

monitoring and reproducibility the system shown in Figure 5 was 

designed and constructed. 

Pressure regulators used on supply gases were all of 2-stage 

type (Matheson Corp., Morrow, GA) and were chosen to be compatible 

with flow rates used during experiments. In order to insure constant 

delivery pressure, a 2-stage regulator must be used or delivery 

pressure will be dependent upon remaining tank pressure, which will 

constantly change as the experiment progresses. The output of the 

regulators were connected to the main control panel by polyethylene 

tubing (1/4" od) and Swagelok connectors (1/4"). The main control 

panel was a free standing aluminum plate (24" x 24" x 3/16") to which 

the flow control devices were fitted. The control panel was 

constructed for six different flows in the following arrangement. 




Gas Tank 

2 -Stage 




needle valve 

Figure 5. Gas Flow Control System. 


Rotameter jholders were constructed out of stainless steel and were 
attached to the plate. Gases entered into the rotameter holders 
at the bottOTi through a 1/8" MPT connector, which was screwed into the 
bottom holder through a hole in the plate. Tapered flute style 
rotameters (Matheson Corp., Morrow, GA) were used due to their larger 
linear working range vs a tapered bore style. The rotameters were 
mounted to maintain a constant inner gas pressure across the tube by 
placing the pressure regulator prior to the rotameter and the variable 
restricting orifice (needle valve) after the rotameter (233). Rotameter 
size (#601-605) as well as needle valve size, (for #601-603 low flow 
rotameters, fine metering valves (NUPRO, "Series M" Cleveland, OH) 
were used and for the higher flow 604 and 605 rotameters, forged body 
regulating valves (Whitey Co., Oakland, CA) were used, chosen so as 
to give about mid-scale readings of the stainless steel balls ('^50 mm) 
in order to assure a high degree of accuracy and reproducibility. 
Following the regulating valves were toggle operated forged body shut- 
off valves (Whitey Co., Oakland, CA) since regulating valves should 
never be used as a shut-off (may damage valve seat causing irreproducible 
flow settings) and also to speed flow shut -off in the advent of an 
accident. From the exits of the toggle valve all flows went directly 
to the burner. 

Delivery pressure gauges on the 2-stage regulators were calibrated 
against a calibrated absolute pressure gauge (Wallace aind Tiernan, 
Model #FA-233 , Belleville, NJ) . All delivery pressures were measured in 
psig (pounds per scjuare inch gauge) and the system was calibrated for 
10, 20 and 30 psig. Gas flow rates were calibrated using a linear gas 


flowmeter (Hastings-Teledyne , #ALK-50K Hampton, VA) . Flow rate vs 
ball height was found to be linear over the entire rotameter range 
(150 mm) as long as the tube pressure was held constaint with the 
exception of a jump in the #605 rotameter caused by the pyrex and 
stainless steel ball sticking together. This could be overcome by 
increasing flow so that the 2 balls separated and then reducing flow 
to desired rate. Calibration was acccmplished using both balls, with 
the stainless steel ball yielding higher flow rates. Rotameters were 
operated with both pyrex and stainless balls for a larger range tut most 
flows were referenced vs the stainless steel balls. 


The methane-air flame used in these studies was produced by a 
multi-hole circular burner using the described flow system. The burner 
design is similar to that used by Horvath et al. (234) in that it 
consists of an inner and outer flame. The outer flame with the same 
composition as the inner was used to prevent temperature gradients 
across the flame and the entrainment of ambient air. This laboratory 
constructed burner is entirely made of brass with the burner head 
being 1.25" in diameter. The burner contained 37 holes 0.125" in 
diameter arranged in three concentric rings around a central hole 
(r^ = 0.300" and 6 holes, r^ = 0.600" and 12 holes, r^ = 0.875" and 
18 holes). Under the burner head is the flame separator, which divides 
the flame into an inner and outer flame. The inner flame consists of 
the center hole and the first ring of 6 holes yielding an inner flame 
diameter of about 0.4375". The outer flame consists of the final 
2 rings yielding a total flame diameter of 1.000". The inner flame 


gases are introduced through a bottom opening and the outer flame gases 
are introduced through the upper side openings. The flame separator is 
seated upon 0-rings to make a gas tight seal between the two input 
flows. Normally a sample is introduced only to the center flame but 
can be spread throughout the whole flame by removal of the flame 

The burner head is screwed down by 8 1/4-20 bolts into the retain- 
ing collar, where it makes a gas tight seal on an 0-ring. This retain- 
ing collar is then screwed into the main body, sealing upon an 0-ring. 
The bottcKi plate is secured by spring loaded bolts to the main body 
to allow for expansion in the advent of a flashback. The input tube 
has two internal 0-rings to seal upon the inner feed connector which is 
usually made of glass. The total length of burner is approximately 
5 1/4". 

Laboratory Coal Delivery Systems 
In order to study the combined burning of methane and coal, 
some method had to be devised to introduce a solid sample (coal) into 
a standard laboratory burner. In normal spectrochemical methods the 
sample is usually dissolved in water or some solvent and then introduced 
using a pneumatic nebulizer. A pneumatic nebulizer uses a high pressure 
flow of air (approximately 30 psig) to produce a fine spray of droplets 
which are then easily introduced into the main stream of flame gases 
(15, 235). The problem in these studies was that coal powder 
(approximately 200 mesh, 75 \xm diameter) was the sample to be studied 
and is not soluble in water. It was desirable to introduce coal into 
the flame without modifying it so the possibility of finding a suitable 


solvent for coal was eliminated. The only preparation that the coal 
should undergo is to be ground into the desired mesh size. 

Several authors (236-238) have described methods for introducing 
solid samples to the flame but these were rejected due to their 
complexity and inaccuracies. An ideal system would deliver a constcint 
and uniform flow of coal at a reproducible rate using relatively simple 
apparatus. Due to the powdery nature of the coal, when ground to 200 
mesh size, it was thought that some type of fluidized bed system 
might be suitable. 

The first model shown in Figure 6 is made from a 350 ml fritted 
glass funnel. Air to create the fluidized bed is input through the 
bottom tube (1/4" od) which then travels through the glass plug 
(10-15 ym pores). Coal dust is loaded into the chamber through the 
1/2" id tube on the side from a large capacity resejrvior. Two 
tangential 1/4" od tubes serve as auxiliary air input and aerated coal 
output. Air is input through the lower tube tangent ially to the inside 
wall in an effort to get a swirling air motion inside the chamber, and 
the output tube is also located tangentially to remove coal dust in 
the direction of air flow. Both the aspirating air and the auxiliary 
air could be varied independently to optimize air flow conditions inside 
the chamber. 

This coal nebulization device was tested, as all others, by 
introducing the coal output into the flame and making visual observations 
of the coal input rate. This could be easily accomplished by use of 
a simple glass adapter at the bottom of the burner which was used to 
mix the main gas flows with that of the coal-air mixture. Input rates 
could also be easily observed because upon introduction into the flame 




Air input 
Figure 6. First Coal Feeder 


the coal particles burned with a bright orange glow and a considerable 
lightr.emisslbh. It was then quite easy to see if the flow was uniform 
and constant by just looking at it. At these early stages this is 
all that was required to judge success or failure of a system, but 
later on more precise methods were used to verify uniformity of flow. 
Upon testing this device the coal input to the burner was found to be 
very erratic with a pulsating effect being observed. Upon closer 
observation of the aspirating chamber it appeared that bubbles were 
bursting at different places on the surface of the coal, causing a 
small amount of coal to be dispersed throughout the volume of the chamber. 
Upon raising the flow of fluidizing air, larger bubbles were made hut 
the flow to the burner was still observed to be very erratic. Upon 
raising input air pressure from 20 psig to 3 0 psig, more coal was 
dispersed inside but flow was still very erratic. After approximately 
five minutes the coal flow dramatically decreased to a very slight 
level. Upon looking into the chamber, a few clear spots were found on 
the fritted disk and the coal renained in most other places. It appeared 
that once the coal was removed from one spot on the disk the air flow 
took the path of least resistance thereby leaving the rest of the coal 
unmoved. Reasons for failure of this system were thought to be: 

1. Large surface area of glass frit (3" diameter) resulting in a 
large pressure drop across the frit yielding a small aereating 
gas velocity. 

2. Glass frit pore size (10-15 ym) was too large to induce a 
fluidizing action. 

3. Volume of chamber was too large to maintain a coal-air 
dispersion with uniform particle density. 


Thinking that in principle it was a good idea, a new coal feeding 
device was constructed along the lines of the first (Figure 6) with 
a few modifications. A 60 ml fritted glass funnel was now used with 
a fine frit pore size (4-5.5 ym) , with all input and output connections 
as in the larger model. Total length of this model was approximately 
3 inches long and frit diameter was approximately 1 1/4". Unfortunately 
this one did not work any better than the first one by producing an 
even worse "percolating effect". Fluidizing and auxiliary air were 
adjusted in an attmpt to prevent this effect but to no avail. It 
was decided to pursue a different approach to the problem of asperating 
coal into the burner. 

The fine powdery nature of the coal still gave one hope that some 
type of air stream method could be used to propel it into the burner. 
Instead of blowing air up through the coal in an attanpt to disperse 
it another way of accomplishing this was developed. In Figure 7 is 
shown the second generation coal feeder that was developed. Chamber 
size was 1 7/8" in diameter and 3" high with a 1/4" id feed tube on top. 
The input aerating air came in through a nozzle at 45° angle to the 
bottom of the chamber and the output came out at a right angle to the 
wall at the top of the chamber. The nozzle ended inside the chamber 
about 1" from the bottcxn, with the nozzle hole diameter approximately 
1 mm. A magnetic stir bar was placed in the bottom and the apparatus 
was placed upon a magnet stirrer. A fixed volume of coal was placed 
inside, enough to fill the chamber approximately 3/4" high. To test 
this device the stir bar was set on high speed and air was input 
through the nozzle. A slight amount of air caused a fair amount of 


Coal in 


Figure 7. Second Coal Feeder 


coal to be introduced into the flame, and upon increasing the air a 
large amount could be input to the flame. One problem with the glass 
frit types was the large amount of air that was needed to asperate 
any coal (about 1 Z/min) , while with this method it was found that 
a very small amount of air (< 100 ml/min) would put a visible amount 
of coal into the flame. Upon increasing the air flow the amount of 
coal reaching the flame could be varied over a verfy wide range. On 
testing for stability over a time period (10 min) , it was found that 
the amount of coal would remain constant for about 2 min, whereupon 
it started to decrease until hardly any coal was being asperated into 
the flame after 10 min. This was caused by the sticking nature of the 
powdered coal. As the air was sent in, it moved all the coal away 
from one spot forming a crater in the coal contained in the chamber. 
Even though the stir bar was spinning vigorously, it could not knock 
down the coal that was sticking to the sides of the chamber. Filling 
the chamber with coal so it coA'ered the nozzle only added a few more 
minutes of constant operation to this version of coal feeder. 

At the same time two different approaches were being developed for 
delivering coal into the flame. The first of these, suggested by results 
obtained in the large cranbuster development program, is shown in Figure 8. 
Whereas all the previously discussed coal feeders worked by blowing 
iar into a coal reservior, this new design worked by sucking the 
coal out by using an induced partial vacuum. In this design a large 
amovmt of coal was placed in a 4" diameter 60° angle funnel which 
filled the funnel and the straight tube leading from the bottcm. At 
right angles to the bottan tube was another tube which had a room air 
inlet and a stopcock valve attached. At the end of this tube a 






Coal dust output 

Air input 

Figure 8. Funnel Feeder 


T-tube was connected which contained a nozzle type device. Air was 
input into the end indicated in Figure 8 and by exiting through the 
inner nozzle a partial vacuum was sent up through the other tubes 
due to the Venturi effect (15) . This induced vacuum would cause the 
coal in the funnel to be pulled down through the tubing and exited 
at the end of the nozzle which was then sent to the burner. The room 
air inlet had a very narrow opening and served to keep the funnel 
tube junction agitated in order to obtain a constant flow of coal 
down from the funnel. The stopcock opening could be used to vary the 
amount of vacuum produced along with the flow rate of air sent through 
the nozzle. The position of the nozzle was varied in the T-tube and 
it was found that the vacuum effect became pronounced at the T junction 
and was at its greatest at the position shown in Figure 8, at the very 
end. In this position the coal would be drawn out rapidly causing a 
crater to be formed in the feed funnel. After the crater was formed 
no further coal was sucked down due to its sticking properties. 
This could be prevented by tapping on the funnel causing the sides of 
the crater to fall down the tube and then into the flame. To prevent 
this cratering effect from occurring, a 4-vane stir rod was constructed 
and rotated with a motor as shown in Figure 8. This stirrer was built 
so that it fit completely inside the funnel and would scrape along the 
inside walls causing the coal to be moved lower into the funnel. 
When the coal volume in the funnel became low more coal was dumped in 
from a supply container (large beaker) , this system wDuld allow for a 
reasonably constant flow rate of coal to be delivered. 


The major problem with this system was that low coal feed rates 
could not be attained. There was a threshold level for the attainment 
of a vacuum sufficient enough to draw the coal into the system. A 
small flow rate of air (approximately 200-600 ml/min) would have no 
effect at all, but once a limiting value was reached (approximately 
900 ml/min) the coal would be propelled at a great rate. Further 
increase of air would only increase the coal feed rate. When full the 
funnel contained approximately 500 g of coal, when running at a 
threshold rate of air with regulating stopcock closed the funnel 
emptied in less than 2 min resulting in a flow rate about250 g/min. 
When this large amount entered the burner, it caused the flame to 
extend into the hood (5 ft above burner) and a large amount of coal 
was not burned resulting in coal depositing on everything on the 
laboratory table. This was obviously too much coal to inject into 
the burner. Upon adjusting the stopcock to vary the amount of vacuum 
the amount of coal entering the flame decreased (producing only a 4 ft 
high flame) but there still was a large amount of unburned coal. The 
only way the coal flow could be reduced was by moving the position 
of the nozzle behind the T-junction and by opening the stopcock valve. 
This then allowed a relatively small amount of coal to enter the flame 
but at very erratic intervals. After much fiddling, not much improve-', 
ment was realized so this method was abandoned. This would probably 
work on a much larger burner system which could handle the large 
output rates of coal that it generated. 

After the dismal failures of the previously described methods it 
was thought that a completely different approach was required. The 


earlier methods used air streams to both agitate the coal and deliver 
it to the burner, and it was found that delivery to the burner was no 
problem but obtainii^ a constant flow was. To overcome the constant 
delivery problon the device shown schematically in Figure 9 was 
devised and constructed. This method broke the system down into two 
parts, a constant coal feed part and a burner delivery part. It had 
been found from earlier experiments that coal could be sent into the 
flame relatively easily by blowing air over it and sending the air to 
the burner. The coal particles were light enough so that they were 
suspended in the air stream and could be transported through plastic 
tubing without much loss. Coal would be deposited on the insides of 
the tubing but once a layer was deposited the rest flowed rather easily 
through the tubing. In the design shown schematically in Figure 9 
the coal is contained in a reservoir made from an aluminum funnel 
and the lower end of it feeds into a chamber containing an auger like 
rod. This rod had a spiral grove cut into it approximately 1/8" wide 
and a 1/16" deep. This rod was driven by a variable speed drill 
through a multi-ratio gearbox. The auger rod was contained in a tube 
like channel with chambers at both ends. The coal was deposited upon 
the auger at one end and was then transported by the twisting motion 
of the auger to another chamber where an air stream was made to pass 
over the end of the auger. This air, supposedly carrying coal, was 
then passed to the burner. 

Upon testing this device it was found that no coal was transferred 
into the flame. Upon further observation it was found that the coal 
dust was sliding in the slots and not being moved down the tube at all. 







The coal particles would stick together and the auger would just 
twist through the coal without any being deposited in the slots. It 
was thought that pressure upon the coal would cause it to be transported, 
but, when pressure was applied by a push plug the only thing accomplished 
was to compact the powder into a hard mass with no movement of coal 
down the channel by the auger. Further modifications of the auger 
(slot width and depth, turning speed, etc.) proved to show little or 
no improvement in coal delivery. 

Based upon the previously stated results it was decided to use an 
air flow method to asperate coal into the flame. Shown in Figure 10 
is the final coal feeder used in the following studies. Several 
improved features can be seen in this version which lead to improved 
performance and a useable and reliable coal feeder. The modifications 
are as follows: 

1. Use of a 250 ml round-bottcm flask for the coal reservior. 

2. Angling of the input air nozzle so that it pointed down into 
center and at a tangent to the inside surface to promote a 
swirling motion inside the flask. 

3. Exhaust tube for the coal-air mixture which opened directly 
straight up and went directly to the burner without any bends 
in the tubing. 

4. Use of a octagonal stirring bar (1/2" x 5/16") in bottom of 
round bottom flask and positioning of the magnetic stirrer so 
that the flask was not at the center causing a more vigorous 
motion to the stir bar. 

5. Extremely narrow opening of the nozzle ('^'0. 5 mm) causing the 
air to enter at a higher velocity. 

6. Introduction of an auxiliary air inlet at top of flask to 
dilute the aireated coal if necessary. 


Coal dust output 

Coal feed 

Air input 

Auxiliary air 

Figure 10. Final Coal Feeder 


Upon testing this new coal feeder it was found to give a 
controllable and steady output of coal into the flame. It was 
also possible to use a very small amount of air (about 100 ml/min) 
to get a substantial amount of coal into the flame. It was found 
that the amount of coal fed into the flame could be varied easily 
by varying the amount of air input to the flask. Coal input rate 
was found by weighing the flask containing the coal before and after 
a test run, timing the length of a run, and then calculating the 
rate loss of coal per unit time. 

One major problan with all the previous coal feeders was in 
obtaining a constant delivery rate over an extended period of time. 
This problen could be eliminated with this feeder by having a sufficient 
amount of coal contained in the flask, approximately 1/3 full. To 
test reproducibility and steadiness of flow a series of tests were 
run at the same air flow rate for various^ times. Results are shown 
in TabAe 9, as can be seen the reproducibility of the new system is 
quite good and the flow appeared constant and uniform to the eye. 

Spectroscopic Apparatus 
A schsnatic diagram of the gas-coal spectroscopic experimental 
set up is shown in Figure 11. With this arrangement it was possible 
to do emission, absorption and fluorescence measurements by only 
varying the electronic detection components. Table 10 lists the 
components used in the gas-coal diagnostic system. All of the equip- 
ment with the exception of the gas control systen and electronic 
components, was mounted on a 4' x 8' x 12" vibration isolation table 
(Newport Corp., Fountain Valley, CA) to provide a steady and stable 
experimental environment. 


Table 9 

Reproducibility of Coal Feeder 

Air Flow (mlAiin) Time (s) Wt of Coal (mg) (Feed Rate (mg/min) 

30 220 58 15.8 

498 132 15.9 

1003 250 15.0 

80 200 126 37.8 

574 368 38.5 

636 398 37.5 

110 216 243 67.5 

420 483 69.0 

600 724 72.4 






Figure 11. Block diagram of experimental arrangement. 


1.26 m, monochromator 




diode array 






strip chart recorder 


PMT high voltage supply 


PDP 11/34 


floppy disk drive 


digital plotter 


0.5 m monochromator 








Table 10 

Experimental Components of Gas-Coal Spectrometric Diagnostic System 


Model No. 



Photomultipl ier 
Power Supply 





Intensified Diode Array/ TN 1223-21 
Multichannel analyzer TN 1710-21 

TN 1710-30 
TN 1710 

Floppy Disk Drive System TN 1710-24B 



Standard Lamps 

Lamp Power Supply 

150 A 

Servo/Riter II 





Spex Industries, Inc. 
Metuchen, New Jersey 08840 

Jarrell-Ash Division 
Fisher Scientific Co. 
Waltham, MA 02152 

EMI Gencom, Inc. 
Plainview, New York 11803 

Harrison Division 


Berkeley Hts., New Jersey 


Tracor Northern 
Middleton, WI 53562 

Tracor Northern 
Middleton, WI 53562 

Keithley Instruments, Inc. 
Cleveland, Ohio 44139 

Texas Instruments Co. 
Houston, Texas 77006 

Eppley Laboratory, Inc. 
Newport, RI 02840 

Laboratory constructed 

Digital Equipment Co. 
Jacksonville, Florida 32205 

Digital Plotter 

Hi-Plot DMP 7 

Houston Instrument Division 
Bausch & Lombe 
Austin, Texas 78753 


The burner was mounted on a laboratory constructed height 
riser which was then bolted on a X-Y milling table (Mastercraft 
tools, #500, Riverside, CA) allowing for accurate positioning of 
the burner with ability to perform carefully controlled spatial 
measurements, the Z axis (up-down) had a total travel distance of 
10 cm and a variable speed motor (INSCO Corp., Groton, MA) was 
connected to the X-axis rotation arm allowing for precise scanning 
of the burner across the entrance slit of the monochromator . Lateral 
movement scan speed could be varied from l"/min to 0.02"/min with 
the X-axis travel being calibrated in .001" steps. 

In order to afford a convenient amount of work space under the 
burner and to allow for observations high in the flame the optical 
axis for all observations was set at a height of 23" above the top 
of the table. Supports were designed and constructed to place the 
optical axis of all spectroscopic instrxaments and collimating optics 
at this level. Light emitted from the flame Was focused by a quartz 
lens (dia = 1". focal length = 4") on the slit of a 1.26 m grating 
spectrometer (ruled area 136 x 116 mm) blazed at 3 00 nm, with a 
dispersion of 0.65 nm/mm. . Two exit apparatus allow the use of two 
different detectors with a rotation of one mirror. Detectors used in 
these experiments were a side-on photcxnultiplier and an intensified 
diode array detector (IDARSS-512 Channels). The photomultiplier was 
operated at an applied voltage of -700 V, and the IDARSS was operated 
with an intensifier gain of 10 and variable exposure time and number 
of exposures depending upon the luminous intensity flux incident upon 
the IDARSS. For emission measurenents at high resolution, the 


photomultiplier output was sent to a current to voltage preamplifier 
and then to a lock-in volt meter for measurement, output was then dis- 
played on the chart recorder. The sources for both absorption and 
fluorescence measurenents wtare chopped to distinguish them frcm the back- 
ground emission and obtain improved signal to noise ratios (239, 240) . 

Moderate-resolution spectra were obtained using a 0.5 m grating 
spectroneter (ruled area 52 x 52 mm, dispersion of 1.6 nm/mm) in 
conjunction with the IDARSS system. Broad band low-resolution spectra 
were obtained with a 0.2 m holographic grating spectrograph (ruled area 
70 mm dia, dispersion of 24 nm/mm) in conjunction with the IDARSS system. 
Signals from the IDARSS detector are processed by its own signal- 
processing module contained in a main-frame with other signal -processing 
devices (i.e. scale calibrator, data smoother, etc.). Data can be 
stored on a floppy disk for later analysis or transferred to a PDP 11/34 
for more processing (i.e., correcting for spectral response). Hard 
copies of data stored can be plotted by the digital plotter interfaced 
to the 11/34. 

The 1.26 meter spectrophotcraetric system spectral response 
characteristics were determined using a calibration standard lamp 
according to reccxunended procedures (241) . For absorption measurements 
a secondary light standard source was used and its light was collimated 
and passed through a quartz diffuser to obtain a uniform beam whose 
size was determined by blackened aperatures . Visual inspection insured 
that the slit was evenly illuminated. The same light source was used 
for line reversal temperature measurements (15) by removing the diffuser 
and focusing the filament of the light source into the flame by a quartz 
lens. Procedures recommended by Snellenan (242) were used to accomplish 


line reversal temperature measurements. For these measurements a 
1000 ppm Na solution (from NaCl) was asperated into the flame by a 
pneumatic nebulizer which used 30 psi air to drive it. Temperatures 
were also determined using a thermocouple (platinum vs platinum - 10% 
Rhodium, Omega Engineering Inc., Stamford, CT) oriented vertically so 
as to provide less perterbation of the flame flow characteristics. 

Spatial resolution of +_ 1 mm could be attained and upon movement 
from one position to another, equilibration of the thermocouple was 
allowed to take place ('v- 1 min) before reading the next measurement. 
The fluorescence excitation source was an electrodeless discharge 
lamp (Hg, Tl, Ga) powered by a microwave generator (Model MPG-4, 
Opthos Instrimients, Ind. Rockville, MD) . Electrodeless discharge lamps 
were run in a 1/4 - wave Evenson type cavity and were externally heated 
according to specifications given by Ball (243). Two quartz lenses 
were used to focus the light into the flame and a chopper between the 
two caused an AC fluorescence signal to be generated which could be 
detected by AC electronics. An He-Cd laser (Model 4050 Liconix, 
Mountain View, CA) (13 mW) was used as an excitation source for a few 

General Considerations 
The following general procedures were followed during all experiments 
performed during this study. 

1. All gas flow rates and delivery pressures were checked at 

the beginning and end of the experiment to insure that desired 
flame conditions (lean, rich, stoichiometric) remained the 
same during the entire run. 

2. The coal feeder was checked to insure that sufficient coal was 
present to insure constant delivery rate during the entire 
experiment; if it was found to be low coal was added. Coal 
was allowed to enter the flame and run for about 1-2 min to 


check if all conditions were right for constant coal 
delivery (stir motor speed, position of feeder, etc.). 

3. In cases where it was needed to know precise coal rate 
(ex. effect on spectra of increasing coal flow) each and 
every run was weighed before and after and duration of a 
run was timed by a stop watch to obtain accurate flow rates. 
Any run with a visually observable erratic flow was discarded 
frcm flow rate studies. On general spectroscopic studies, 

a known air flow rate was set and periodic weighing and 
timings were made to check reproducibility of flow settings. 

4. For spatial dependent studies all movement occurred in one 
direction only. To reset for another measurement the height 
adjuster or length adjuster was returned past reference mark 
and moved to it from whence measurements began. 

5. For all measurements using a light source, the source was 
allowed to warm up for at least one-h&lf hour at which time 
it was monitored for stability for a period of 10 minutes. 
If fluctuations occurred another period of warm-up occurred. 


The ubiquitous nature of the OH radical in hydrocarbon flames 
provided a good stcirting point for these studies. Initial observa- 
tions indicated that the introduction of small amounts of coal dust 

(< 1 mg/min) into a CH^-air flame caused the flame to become highly 
luminous. As the OH emissions were a useful diagnostic species 

(i.e. temperature measurements), it was desired to know how the coal 
dust influenced the OH emissions in the ultraviolet region. Spectro- 
scopic observations of the OH 0-0 band showed variances in the over- 
all beind shape, which was further investiaged both theoretically and 
experimental ly . 

Developnent of a Temperature Measurement Technique and Observations 
of the Effects of Varying Amounts of Coal on OH Emissions 

If the 0-0 vibrational band of OH is viewed in emission under low 

resolution (AX'S' 2 nm) it appears as a wide hump starting at 306 nm and 

extending to around 320 nm with two broad peaks at 307 and 309 nm. 

The peak at 307 is caused by the heads of the and R^ rotational 

branches and the peak at 309 is caused by the and branches. 

Mavrodineanu and Boileux (15) have given fortrat parabolas, showing 

the six main branches of the OH 0-0 band near 305 nm. The intensities 

of ninety-five lines in 5 branches were calculated by Vaidya et al. (244 

who show how the intensity ratio of the two peaks at 307 and 309 can 

be used to determine the temperature in flames. 



The line intensity for emission is given by (4) 

I = CAv exp (-E^/kT) 

where C is an instrumental constant, v is the wave number of the 

line. E is the rotational energy of the initial level, k is 

Boltzmann's constant, T is the temperature and A is the transition 

probability for the line. From the above equation, intensity I can 

be calculated at different wavelengths for various temperatures 

provided the values of A and E^ are known. The numerical values 

2 2 

for all the transition probabilities and energy levels of the Z- 11 
transition system in OH have been tabulated by Dieke and Crosswhite 
(245) and these values were used for the intensity calculations. The 
line intensities for five branches, R^, R^* ^1 "^^^ 

calculated for the temperature range of 600-3500 K. Figure 12 shows 
intensity vs wavelength data for these branches at temperatures 1500, 
2000, and 2500 K. It is seen from Figure 12 that, whereas the shape 
of the Q^, and branches remains practically unchanged, there is 
a marked change in the shape of R^ and R^ branches with tonperature. 

For a given resolution AX , the convolution of line intensities 
with the triangular slit-function approximation (246, 247) gives the 
intensity vs wavelength curves which are shown in Figure 13. The 
equation used for the convolution is 

KX) = 


where the slit function 


S(X,X^) = 

for|x-X . |<_AX 










• ■ 

• * ■■ 












■O*-- A^? °°o. 

T=2000 K 






o □ 


o □ 

• ■ ■ 
• ■ 
• ■ 




■■■ A 


• ■ 
« ■ 
• ■ 
• ■ 

o □ 


T=1500 K 



o □ 


o □ 

o □ 

J L 




Figure 12. 


310 312 
Wavelength (nm) 



Calculated intensities of 5 branches of OH 0-0 band at 1500, 
2000 and 2500 °K. 



Here AX is the resolution, is the intensity of the line, is 
the wavelength of the line, X is the wavelength of the bandpass 
center, and I(X) is the convoluted intensity. The instrumental profile 
necessary to determine the shape of the slit function was obtained with 
a He-Cd laser at 325 nm. 

The following features can be observed in Figure 13. 1. All 
the curves show two prominent peaks, one near 307 nm and the other near 309 
nm. 2. The intensity ratio of these two peaks changes with the temper- 
ature. 3. The minor feature observed near 3 08 nm diminishes with 
an increase in temperature. 4. The features observed at wavelength 
above 310 nm become prominent at higher temperatures. Plots of the 
ratio I (309)/ I (307) vs temperature are shown in Figure 14. Thus, at a 
given resolution AX, for a known ratio of 1 (309)/ 1(307) the temperature 
in a flame can be determined from the corresponding curve. 

In the 600 <T<2000 K temperature range the ratio I (309) /I (307) may 
be fit with very good precision ('\'1%) with the empirical equation (244) 

R(t,AX) - ^^^Q^^ - B + (T/T 


where T = 1000 K, A = 2.825 and P - 1.145 exp (-0.5076 AX), B = 1.927 AX 


- 0.1216 - 1.378 X . By inverting the above equation a convenient 

interpolation formula can be obtained 

T = T [ (A/R) -b] 

for calculating the temperature from an experimental ratio measurement. 
Figure 14 illustrates the temperature vs intensity ratio relationship 
for three values of AX. The points represent the theoretical data and 
the lines represent the analytic expression. As is seen, the fits are 


Figure 14. Intensity ratios vs. temperature for spectral resolutions 
AX=0.2 ran, AX=0.6 nm and AX=1.0 nm. 


very good. Since slit functions usually depart frccn triangular shape, 
it would be wise in actual applications to use test experimental data 
to determine the parameters A, P, and B. 

At this point in time many OH spectra were obtained frcan a methane- 
air flame, but the effects of coal upon these emissions were still 
unknown. To investigate this question an experiment was conducted 
where various amounts of coal (200 mesh, 5-400 mg/min) were added to 
a CH^/air flame, and the OH emissions were observed. The total air 
flow was held constant and the coal dust was only introduced into the 
center section of the barner. All spectroscopic observations were made 
at a height of 4 cm (measured from the burner head) and the IDARSS system 
was used with both 0.5 m and 1.26 m grating monochromators. 

The coal used (Virginia bituminous) had a calorific value of 
13,000 Btu/lb. Assuming complete ccxnbustion, the percentage of energy 
produced by the coal was calculated as a function of coal flow rate. 
These percentages are shown in Table 11. As increasing amounts of 
coal are added, the total luminous output of the flame increased 
dramatically. At low flows an orangish glow is seen, and the particles 
of coal leave Iviminous trails as they rise in the flame. The larger 
coal particles continue their incandescence for longer periods of time 
than smaller ones, which stop before reaching very high in the flame. 
As the particles first enter the flame they appear to emit a burst of 
luminescence which rapidly decays to a lower level of light emission 
which continues until the particle is completely combusted. Larger 
particles seemed to emit larger bursts for a slightly longer period 


Table 11 

Coal Flow Rates Used in Experiment and Heat Production Ratios 

Spectrum # 

Coal Flow 

Released Heat 
from Coal 

AH^ (kcal/min) 







c c 



20.73, 3.41 































r\ r\ n o 

u • X J J** 





















Calculated assuming complete combustion of 13,000 Btu/lb coal. 
Ratio of heat energy produced by coal combustion to that produced 
by methane/air flame. 

Above ratio for inner flame region where coal is injected and 
observations are made. 

'Heat released by gas flame only, total flow 1.56 1/m (1.11 g/min) 
inner flow only 0.36 1/min (0.257 g/min). 


of time than small particles. High-resolution spectra (AX = 0.06 nra) 
showed no gross structural changes in relative rotational line structure 
intensities, but a slight increase was seen in total overall intensity. 
Figure 15 shows a high-resolution spectra of the laboratory methane-air 
flame, and Figure 16 shows the same flame with the addition of 'V'305 mg/min 
of coal dust. 

The intensity ratio (I (309) /I (3 07) temperature method was used to 
determine the temperature of the laboratory methane-air flame, as well 
as for the gas flame of a large experimental burner designed in connection 
with a gas coal burning project by other workers in this laboratory. 
Figure 17 shows a moderate resolution spectrum (AX =0.6 nm) obtained 
by a 0.5 m monochromator of the laboratory gas flame in the 27 9-3 24 nm 
spectral region. From this spectrum the intensity ratio of the two peaks 
is found to be 1.206 which gives a temperature of 2025 K from the inter- 
polation formula given previously. This value of the temperature is 
close to the 2050 K temperature obtained with the sodium line-reversed 
method. Temperature determined from the rotational structure of the 
branch of the 0,0 band (245) was found to be 2020 K. 

Figure 18 shows the spectrun of an after- burning flame from the 
large experimental combustor taken with this laboratory's mobile 
spectroscopic facility (248, 249) . The temperature derived from the 
spectrum is '^'1500 K, which was expected due to the nature of the after- 
burning flame and is in good agreement with thermocouple temperature 
measuranents of 1550 K (250). 

Survey Spectra of Methane-Air and Methane-Air-Coal Dust Flames 

A series of survey spectra were obtained with the 1.26 m spectro- 
meter in conjunction with a high gain photomultiplier with enhanced 



280 290 300 310 320 

Wavelength (nm) 

Figure 17. 0.5 m spectrum of CH /Air flame 0-0 band. 

Figure 18. Spectrum of afterburning flame taken with mobile laboratory 
using 0.5 m spectrograph through periscope. . 

17 5 

ultravLol'et . response (EM1t9781vBFL) The spectral region covered was 
from 180 nm to 600 nm and a 100 ym slit, 1 cm high was used to obtain 
reasonable resolution and adequate light gathering ability. All spectra 
were run with room lights off to avoid mercury lines emitted by the 
fluorescent overhead lights, which were observed during preliminary 
observations. Spectral emitters positively identified (251) and 
observed in a stoichiometric CH^/air flame in this region were: 

1. CO - fourth positive system: A''"n-X'''Z^ from ^^192 to 23 5 nm, 
many different vibrational bands. 

2. OH - a. B^E^-A^e''" (1,9) vibrational band at 254-260 nm 

b. A^E^-X^n^^ vibrational bands, (3,0) at 242-252 nm; (2,0) 
at 262-272 nm; (1,0) at 281-300 nm; (0,0) at 305-325 nm 
listed in order of increasing intensity. 

2 - 2 

3. CH - a. B E -X n (0,0) and (1,1) vibrational bands from 
'\.387-404 nm. b. A2A-x2n(0,0) and (1,1) and (2,2) vibrational 
bands from 417-440 nm. 

3 3 

4. C2 Swan bands: d Ilg-a 11^ vibrational bands: (5,4), (6,5), 
(4,3), (3,2), (2,1) and (1,0) frcxn 466-474 nm; (3,3), (2,2), 
(1,1) and (0,0) from 504-516.5 nm; (2,3), (1,2) and (0,1) 
from 552-564 nm. 

Molecular energy level symbols were given as upper level first to 
be consistent with the usual molecular spectroscopic conventions as 
discussed by Herzberg (252). These spectra are also shown in Figures 19 
to.. 24. Spectra of CD, CH and C^ were only observed low in the flame 
(0-10 mm) while OH emissions (0,0) were visible higher up in the flame 
('V'7cm) but with diminishing intensity with increasing height. 

Upon introducing coal into the flame slight changes were seen in 
the spectra of CO and OH with larger changes in the spectra of CH and 
C^j mainly intensity changes. New emissions were observed in the 
region between 402 and 412 nm which can be assigned to the C radical 




n \ 

280 290 300 

Wavelength (nm) 

Figure 21. OH (1,0) band (slit lOOy , 1cm high chart l"/min mono 
5x10"^ nm/s) . 




Figure 25. Spectra of CH Region 

a) CH^/Air Flame 

b) CH^/Air +100 mg/min Coal 

(slit=100y 1 cm high, 
chart l"min 2xl0~3nm/s) 


■''n -■'"E transition. Figure 25 shows the region of 386 to 440 nm 

of a methane-air flame and methane-air flame with approximately 

100 mg/min coal dust showing the decrease in the CH emissions and the 

appearance of emissions in the coal dust flame. 

Measurement of Excited State Populations in Methane-Air and 
Methane-Air-Coal Dust Flames 

In this set of experiments the spectral emissions of OH, CH, C^ 
and CO were measured as a function of height in the flame. In a 
methane-air flame fuel/air ratios (({>) of 0.66, 0.77, 0.90, 1.00, 1.10, 
1.30, 1.50, 1.70, 1.80, 1.90 and a diffusion flame were measured while 
in a methane-air-coal dust flame ratios o.f 0.77, 1.00 and 1.10 were 
measured using a coal, flow rate of approximately 30 mg/min. Total 
gas flow rates were held constant, with fuel and air being adjusted to 
obtain desired (j) . In this experiment the flame separator was not 
used and the coal was dispersed through the total flame volume. For 
the (|) = 1.00 flame methane flow was 1.614 Z/min with an air flow of 
16.9 Jl/min for a total flow rate of 18.514 2,/min, which was held ■■ 
constant for all gas flames and flames with coal. Table 12 gives 
volume flow rates and mass flow rates for the flames studied. 

All measurements were made using the 1.26 m monochromator with 
the same photomultiplier detector as was used for the survey spectra. 
Two millimeter high slits were used for all measurements while 100 ym 
wide slits were used for OH, CH and C^ while a 350 y wide slit was 
used for CO measurements due to the low intensity of these bands. 

The intensity of a spectral line in emission,I , is defined as the 
energy emitted by the source per second (252). In an optically thin 
gas this is equal to l/4Tr times the product of four factors: 


Table 12 

Volume and Mass Flow Rates for Experimental Flames 





CH /air 





























1. The number of molecules in the upper level per unit area 
in the line of sight. If n2 is the number per unit volume 
of these molecules, and 2. is the thickness of the emitting 
layer then the number of molecules is given by n^il. 

2. The transition probability A^^ of the line being observed. 

3. The energy hv of the line, where h is Planck's constant and 
V is the frequency (s"-'-) of the line. 

4. Q. the solid angle of collection observed. 

This yields for the intensity of a line, the equation 

I = — f^iln„A^,hv 
4-n 2 21 

which can be solved for n^ yielding 

!_ 4tt 
^2 ~ I A^^hvn 

The intensity of a line was measured experimentally in the following 
manner. First the survey spectra were studied to find lines that were 
relatively intense and free from overlap by other lines. Unfortunately 
no lines could be found of sufficient intensity which were completely 
isolated frcm other spectral lines. A comprcmise was made between the 
need for an individual line and the need for adequate signal strength to 
perform the measurements. 

The absolute spectral response of the photometric system was determined 

using a calibrated standard of spectral irradiance. By also measuring 

the bandpass of the monochromator , by scanning across a suitable narrow 

line source (He-Cd laser) , the photcxnetric system was calibrated for 

absolute intensity measurements. In order to measure the intensity of 


a line the monochromator was slowly scanned (1 x 10 nm/s) across the 
line of interest and the photcmultipler output was displayed on a strip 
chart recorder (2"/min) . The lines were selected such that a baseline 


could be established after the line was scanned to obtain a value for 
the background signal. The intensity of this line was taken to be the 
peak value as the line was scanned. The bandwidth of the monochromator 
(0.065 nm) insured that the line was totally within the bandpass of 
the monochrcmator due to its small line width (<0.005 nm) (245) , there- 
fore yielding an integrated value for the intensity of this line. 

This peak intensity, measured in nanoamperes , could then be converted 


into W/cm Sr using the calibration scale at that wavelength. 

The line observed was identified by ccmparison to reference spectra 
and by the wavelength dial of the monochromator which was sufficiently 
accurate (+ 0.1 X) so that the reading could be used for line identifica- 
tion. An example would be the Q.9 line of OH (0-0) which is given by 
Dieke and Crosswhite (245) to be 306.372 nm was found to be 306.383 by 
the monochromator dial. 

For all species except CO, where an entire vibrational band was 
maasured, there was more than one line contained in the monochromator ' s 
bandpass when centered on the primary line. To calculate a number density, 
the transition probabilities of the lines within the bandpass of the 
monochromator were summed together to obtain cin effective transition 
probability for the observed intensity. This summation of transition 
probabilities induces a systematic error into the number density 
calculations by yielding an effective value larger than the "true!' value. 
This error would cause the calculated number densities to appear smaller 
than the "true" value. In order to accurately estimate this error one 
must know the fraction of the total measured intensity caused by each 
individual line within the observed bandpass. The actual error might be 


calculated by convoluting the slit function with the expected 
line profile which can be determined using the Voigt profile 
equation assxaihing the same broadening mechanism for all lines 
observed. This systematic error would at most cause a factor of 
two difference between the measured values and the "true" absolute 
number densities, but the relative numbers should be as accurate 
as the transition probabilities '^-+20%. 

For the calculations involving OH the rotational line 
transition probabilities were obtained fron the work of Dimpel and 
Kinsey (253) who observed the radiative lifetimes of individual 
rotational states which were then used to calculate Einstein A and 
B coefficients for the (0-0) band of the A-X transition. Rotational 
lines observed within the spectral bandpass (0.065 nm) were 
identified using the tabulations of Dieke cind Crosswhite (245) and 
the spectropho tome trie atlas of the A-X transition by Bass and 
Broida (254) . The transition probabilities of the lines found to 
lie within the bandpass were summed to obtain an effective transition 
probability (255). The frequency used for calculations was the 
center frequency of the bandpass. 

For the calculations involving CH and no compilations of 
individual rotational transition probabilities were found to 
exist in the literature so rotational transition probabilities were 
calculated in the following manner. This discussion follows the 
nomenclature and notation of Tatum (256) and the recommended conventions 
for defining transition moments and intensity factors in diatomic 


molecular spectra by Whiting et al. (255) as well as the conventional 
notations of molecular spectroscopy (i.e. ' = upper state, " = lower 
state, etc.) (252) . 

The strength of a rotational line (S ,) can be given by the 
expression (256) 

S (N* J' ,N"J") I R^'^" (n'v' ,n"v") | ^ 

^J'j" " {2^ J (2S+1) (2J'+1) 


where] R^'^" (n'v* ,n"v")| ^is the band strength, which is the square of 

the transition moment of the band and is a function of the electronic 

and vibrational levels involved in the transition. The n' ,n" define 

the two molecular states and the v' ,v" the vibrational levels. The 

S(N'J',N"J") are the line strength factors, also known as the Honl- 

London factors, which determine the relative intensities of the branches 

within a band cind the relative intensities of a line in a branch. 

The Honl-London factors are dimensionless while the line strength has 


the same dimensions as the band strength which is [dipole moment] 
for electric dipole radiation. The original formula of Honl and London 
(257) only applied to transitions between singlet states, however 
formulae which apply to transitions between states of higher multi- 
plicity have been given by many other workers such as Mulliken (258) , 
Kovacs (259) and Schadee (260). 


For calculation of rotational line strengths the Honl-London 
factors must obey the following sum rule. The sum of the Honl- 
London factors for all those transitions that have a common upper level 
is 2J'+1. This implies that the sum of the Honl-London factors for 
all the branches in a band is (2-6 ,) {2S+1) (2J'+1) when the factors 
are expressed in terms of J. This accounts for the denominator in 
the line strength equation where (2-6 .) is the electronic statistical 

O J A. 

weight which equals 2S+1 for E states and 2 (2S+1) for other electronic 
states and 2S+1 is the spin multiplicity which is given as a left hand 
superscript in the molecular term symbol. J' refers to the rotational 
level of the upper state. 

2 2 3 3 

Honl-London factors for CH A A-X n and d Il-a n were calculated 

using the formula of Schadee (260) . For the rotational levels involved 

in CH (258) and (260) both molecular levels involved in thetransitions 

were considered to belong to Hunds case b. Once individual rotational 

line strengths (S , „) wero calculated the transition probabilities 
J J 

were obtained by using the following equation (256) ,: 

3he C 

where is the permittivity of vacuxjm and the numerical term converts 

line strength in atomic units to SI units . 

Band strengths for CH were obtained from the work of Hinze, Lie 

and Liu (261) who obtained a value of 0.3785 a.u. for the (0,0) band 
2 2 

of the A - n transition and a value of 0.33 61 a.u. for the 1-1 band. 


These values were obtained from an ab initio calculation of the dipole 

transition matrix elements using extended configuration interaction (CI) 

electronic wavefunctions as reported by Lie, Hinze and Liu (262). 

Frcm the calculated transition matrix elements, band strength and band 

oscillator strength were calculated. The authors reported an accuracy 

of + 15 percent for their calculated results of band strengths. Calculated 

radiative lifetimes were in good agreement with experimental values. 

Rotational lines observed within the spectral bandpass were identified 

using the tabulations of Moore and Broida (263) and the spectrophotonetric 

atlas of CH from 300 to 500 nm by Bass and Broida (264). 

The band strength for C^ was obtained from the work of Cooper (265) 

3 3 

who obtained a value of 4.10 a.u. for the (0,0) band of the 11- 11 

transition which is in good agreement with the value of 4.12 a.u. 

obtained by Arnold and Langhoff (266) . Cooper calculated potential 

energy curves and spectroscopic constants for the triplet states of 

C^ using self-consistent-field plus configuration interaction techniques. 

The variation of the electronic transition moment with internuclear 

separation was calculated for the Swan, Ballik -Ramsay and Fox-Herzberg 

band systems and was found to be in good agreement with existing 

experimental and theoretical data. Rotational lines observed within 

the bandpass were identified using the tabulations of Phillips and 

Davis (267j) on the Swan system of the C^ molecule. 

For the CO calculations the vibration band at 208c9nm(5 , 12) was 

used and the transition probability of this band was given as 
7 -1 

0.221 X 10 s by Mumma, Stone and Zipf (268). They used electron 


impact on CO to generate emission from the fourth positive bands 
cmd used relative intensity measurements on 28 bands to determine 
the dependence of (electronic transition moment) on the r centre id. 
which is R = 1.9(1.0-0.6 x , „) . Absolute transition probabilities 

Q V V 

were ccmputed using the above functional form for R^ and normalizing 
the total transition probability of the v'=2 level to published 
experimental results. The vibrational level observed was identified 
using the tabulations of Krupenie (269) and the published CO fourth 
positive spectra of Hornbeck and Herman (270) and Headrich and Fox (271). 

Transitions observed and the transition probabilities used are 
given in Table 13 and the calculated concentrations as a function of 
height for all the different flames are shown in Figure 26 to 39. 
The solid angle of collection (Q) was the same for all measurements 
witha value of 0.0123 Sr. Because the flame is not a perfect cylinder 
eind various fuel/air ratios led to different size flames the individual 
flame widths were measured as a function of height and these values 
are used for the path length for the respective flames. Table 14 
gives the values of path length used for the different flames at 
different heights. 

In Figures 40, 41 and 42 the differences between the 3 flames with 
coal dust and the identical flames without coal dust are shown. A 
nxamber less than one indicates a decrease in excited ;state population 
upon the introduction of coal while a number greater than one indicates 
an increase in excited state population upon the introduction of coal dust. 
These ratios also indicate present changes which will be used when dis- 
cussing the data. For example, a ratio of 1.10 means a ten percent increase. 


Table 13 

Lines Observed-Transition Probabilities vs Excited State Calculations 
Species and lines Wavelength (nm) A(s ) 

OH (0-0 

308.9734 6.403 E5 


0"^ 308.9861 5.448 E5 

0^ 308.9861 4.964 E5 


Q^, 431.1306 4.357 E4 


431.1306 3.870 E4 


0^ 431.0908 3.871 E4 


431.1110 3.412 E4 

q"' 431.1425 3.869 E4 


o"' 431.1425 3.436 E4 





p^^ 516.5026 6.879 E4 

516.5026 5.536 E4 

p^^ 516.5026 5.925 E4 

p^^ 516.5026 5.194 E4 

p^^ 516.5026 6.361 E4 



516.5026 5.916 E4 



Table 13 -Continued 

Species and lines Wavelength (nm) A(s ^) 


^2 516.5026 5.528 E4 

p^^ 516.5242 6.382 E4 

p^^ 516.5242 6.905 E4 

p^'* 516.5242 5.932 E4 

p^-"- 516.5026 7.511 E4 

p-"-^ 516.5026 5.542 E4 


CO{5,12) 208.99 2.21 E6 



T 1 . T 

T 1 r r 

-1 r r 


O O 

A A 

O O 
A O 




O A 

+ + 


X X 


X X 

' I 1 L. 

I I 1 L. 

5 10 
Height (mm) 


_1 1 1 L. 




X = CO(xlO) 

Figure 26. Excited state concentrations for 0 = 0.66. 


10 : 

-T 1 1 r 

-I 1 1 r 

O O 


A A 

A A 




- O 


+ + 






+ X 



_l I 1 L. 

I I 1 l_ 



Height (mm) 





c8 (xlO) 

Figure 27. Excited state concentrations for 0 = 0.77 


10 : 

-I 1 1 r 

n r 1 r 

n r 











J I L 

o o 

A O 

_I I L. 


o o o A 

J I I 1- 


Height (iraa) 



X = CO(xlO) 

Figure 28. Excited state concentrations for 0 = 0.90 


Height (mm) 

o = OH 

A = CH 

+ = C. 


X = CO(xlO) 

Figure 29. Excited state concentrations for 0 = 1.00 



-I 1 r 


^ O 


-I 1 1 r 




O O o O ^ 






_l 1 1 1_ 

J I I 1_ 


Height (mm) 






Figure 30. Calculated Excited State Concentrations for (j) = 1.1 


10 t 

T 1 T 












A A 


O * 

J I 1 L- 




-1 r 1 r 


o o 


o 6 

_l 1 1 L 


Height (mm) 



X = CO(xlO) 

Figure 31. Excited state concentrations for 0 = 1.3. 



1 1 1 r- 

T 1 r 

-| 1 1 r 




+ + 



O O 




A A 

O O 








-1 I I l_ 

_1 1 1_ 

_J I I L. 

Height (mm) 






CO (xlO) 

Figure 32. Excited state concentrations for 0 = 1.5. 



"1 r— — I r 

"T 1 r 

"] I 1 1 r 












o o o 




X X 


_ X 


-J 1 I 1 I 


Height (mm) 


o = OH 
A = CH 

X = CO(xlO) 

Figure 33. Excited state concentrations for 0 = 1.70. 



"1 1 1 r 

"I 1 r 

"1 I 1 r 



A A 

+ + 


o o 


o o 





-I 1 1 L. 


-1 I 1_ 


Height (mm) 



= C_ 

Figure 34. Excited state concentrations for 0 = 1.8 


Height (mm) 

o = 


A = 


+ = 

Figure 35. Excited state concentrations for 0 = 1.90. 






-1 1 r 

-I 1 1 r 

A A A A ^ ^ 

A A 


O + 

* J s S S 4 s 

10 _ 


I ' L. 

9 9 © i 

' I ' 1- 

' I I L. 



Height (mm) 



X = CO(xlO) 

Figure 36. Excited state concentrations for 0 = <» 



-I 1 1 r 

-T 1 r 

T ; 1 r 


A ^ O 





o H 

+ + + 


A + 


_1 I I 1_ 


_1 1 1 L. 


Height (mm) 

o = 

A = 

+ = 

X = 



Figure 37 . Excited state concentrations for 0 = 

0.77 + 30 mg/min coal. 



"T T 





"T r 












X X X X 


r X 


I • I 

-J 1 t 

Height (mm) 


-J ' 1 L. 

o = OH 
A = CH 
+ = C_ 


X = CO{xlO) 

Figure 38. Excited state concentrations for 0 = I.OO + 

30 mg/min coal. 


10 C ^ 1 1 r- 

T 1 1 1 1 r 






A ° O O O 





5 I ' I ■ i_ 

_i I I i_ 



Height (mm) 



Figure 39. Excited state concentrations for 0 = 1.10 + 30 mg/min coal. 


Table 14 

Flame Widths (cm) vs Height (itim) 

2 4 6 8 10 12 14 16 18 20 






























































































































































II 1 1 1 






□ o 




A A ° 




° O 

□ □ 

- A 


~ 0 




1 1 1 1 1 1 1 1 1 1 1 i 1 1 I 

5 10 15 

Height (mm) 

Figure 40. Ratio of change for flame 0=0.77. 






□ A 







t-t. ^ 8 o o o , 5 1 o o o o 

-A ■ ■ ■ ■ 

I I I I I I I I I I I I 1 ' I 

5 10 15 
Height (mm) 

O = OH 

A = CH 

■ = CO 

Figure 41. Ratio of change for 0=1.0. 



Figure 42. Ratio of change for flame 0=1.1. 


Spatial Temperature Measurements of Methane-Air and Methane-Air- 
Coal Dust Flames 

Spatial temperature profiles were obtained by Dr. K. M. Pamidimukkala 
in concurrence with another gas-coal study (27 2) who graciously made 
these data available to aid in the present study. Illustrative tempera- 
tures for two flames studied in emission are given in Tables 15 and 16. 
These temperatures have been measured and calculated to include heat losses 
due to radiation according to equations given by Fristrom and Westenberg 
(273). This heat loss was calculated to be 400 K for the thermocouple 
used. Due to the radial symmetry present in the round burner all tempera- 
ture measurements were made in one horizontal plane outward from the 
center in one direction parallel to the burner head. Measurements within 
1 cm of the center are accurate to jf 10 K while measurements further out 
from the center would be accurate to + 50 K due to fluctuations and 
disturbances in the outer edges due to room air mixing. Table 17 shows 
percentage differences between the 3 flames with coal dust and the 
identical flames without coal dust for the temperatures at 0 radius. A 
negative percentage shows a cooler flame upon addition of coal dust while 
a positive percentage indicates a hotter flame with coal dust. 

Investigations for Sulfur and Sulfur Compounds in Methane-Air- 
Coal Dust Flames 

A series of experiments were undertaken to observe any sulfur 

containing species that might be present in a methane-air-coal dust 

flame. The coal flows in this series were varied from 30 mg/min to 

150 mg/mxn in a stoichiometric flame. Species searched for were (251) 
3 - 3 - 

1. S„ - B E -X E from 470-503 nm with most intense bands at 
479.06 (§,18)? 484.21 (4,19), 489.36 (2,18), 493.70 (3 ,19), 
498.95 (4,20) 503.62 (5,21), 509.02 (6,22) 519.42 (5,22), and 
524.97 (6.23). Bands also investigated between 370 and 420 nm 


Table 15 

Experimental Spatial Flame Temperatures for a 0=1.00 Flame 

Height (mm) 


Radial axis (mm) 
6 8 10 















Table 16 

Experimental Spatial Flame Temperatures for a 0=1.70 Flame 

Height (mm) 

Radial axis (m) 
3 6 8 













































Table 17 

Percent Change in Temperature Upon Addition of Coal (30 mg/min) 
Height (mm) 0=1.0 0=1.1 0-.77 























O A 




















were 373.98 (0,7), 383.71 (0,8), 393.89 (0,9), 404.56 
(0,10) and 419.36 (1,12). 

2. SH - A E -X n the most intense band was investigated (251) 
which is the (0,0) with heads at 323.66 (R^) , 324.07 (Q^) arid 
327.91 (Q^). 

3. SO - B-^z"-X-^E" strong bands at 327.10 (0,10), 316.48 (0,9), 
306.41 (0,8) and 282.74 (1,6) due to interference of strong 
lines of OH the following region was also investigated. 

A^n-X^E~ strong bands at 254.17 (3,0 ^21^' ^55. 85 
(2,0 R^2^' 256.80 (2,0 R^^) and 257.81 (2,0 R^^) . 

4. SO - -^A observed in the region of 280-309 nm with band 
heads at"'"283.23, 285.20, 286.89, 288.77, 290.65, 292.31, 
292.48, 293.77, 294.38, 296.12, 298.00, 300.10, and 302.21. 

■^B -"'"A^ system was looked for in absorption between 
220 and 235 nm. 

5. CS - A"'"II-X"''E observations were made in the region of 250-270 nm 
where the following bands were located, 252.32 (2,1), 253.87 
(3,2), 257.56 (0,0), 260.59 (2,2), 262.16 (3,3), 266.26 (0.1) 
and 269.32 (2,3). 

6. S - atomic lines at 469.625, 469,545 and 469.413 nm. 

All the above regions yielded negative results when observed in 
emission with low and high amounts of coal at the heights of 4,20 and 
50 mm. The only changes that occurred in these spectral regions when 
coal was added to the flame were 3 lines found in the wavelength region 
of 325-332 nm. These lines are shown in Figure 43. It was first 
thought that these lines might belong to SH but the narrowness of these 
lines seemed more indicative of atomic rather than molecular spectra. 
Higher resolution spectra with a slower scan rate were made of these 
lines but no visible structure could be seen. Upon checking atomic 
wavelength tables the lines at 324.8 and 327.5 nm were assigned to Cu 
and the line at 33 0 was assigned to Na. Verification was accanplished 

\ r 

324 326 328 

Wavelength (nm) 

Figure 43. Ultraviolet line spectra (slit=50vi, 2inin high, chart l"/min 
5xlO"3 nm/s). 


by asperating a 1000 ppm solution of CuSO^ and observing the emission 
which appeared at the position of the first two lines tentatively 
assigned to Cu. The line at 3 30 was confirmed by asperating a 1000 
ppm solution of NaCl and observing a line at the same place. 

Upon review of the literature it was found that SO^ is a better 
absorber than emitter so an attempt was made to observe SO^ in 
absorption with resultant negative results. To check the technique 
used for absorption measurements an attempt was made to observe OH 
in absorption. The absorption of OH was seen in the first attempt so 
technique was assumed to be correct. The absorbance was not large 
(<1%) , but amplification scale was expanded to enable positive 
identification of OH. 

As the coal used in these experiments contained a low percentage 
of sulfur <2% it might not be surprising that no sulfur containing 
compounds were seen. Samples of coal were then made which contained 
1, 2, 4, 8, 15, and 30% by weight sulfur, which was made from powdered 
flowers of sulfur. Once again all the studies proved to be negative. 
Effects of Addition of Excess O^ on Me thane- Air-Coal Dust Flames 

In this experiment the coal was introduced to only the center 
section of the flame and the oxygen was only introduced into this 
region also. The flame was set up as stoichiometric and a coal flow 
rate of 30 mg/min was used. Oxygen was introduced into the inner flame 
gas lines and its flow was varied from 0.075 to 0.25 J,/min. Upon 
introduction of oxygen to the methane-air-coal dust flame the visible 
character of the flame changed from an orange color to a whitish color 


with the white emission coming fran a round area approximately 3-5 mm 
abDve the burner head. Spectra were obtained from this region but no 
new emissions were observed but a slight increase was seen in the intensity 
of the OH, CH, and emissions. 

Fluorescence Experiments 

Two excitation sources were used for these experiments, a He-Cd 
laser and a mercury electrodeless discharge lamp. The He-Cd laser 
was a continuous type which gave lines at 325 and 442 nm with powers 
of 13 and 35 mW respectively. The Hg discharge emitted all the normal 
mercury lines with the majority of power in the 254 nm line. Both 
soxirces were chopped by a mechanical chopper and the resultant 
fluorescence, if any, was detected by the 1.26 m monochromator/photo- 
multiplier combination in conjunction with a lock- in voltmeter. 

Two experimental runs were made using each source, first a run 
on just a methane-air flame to observe any background fluorescence 
that might occur and then a run with the introduction of coal into 
the flame. The spectral region scanned was from 180-600 nm at a height 
of 2 cm and the coal flow rate was '^50 mg/min with a stoichicmetric 
flame used for all cases. For both sources no background fluorescence 
was seen in the methane-air flame and no fluorescence was seen when 
coal was added also. In order to check optical alignment, a critical 
variable in fluorescence measurements, the flame was run without being 
ignited and water was aspirated into the flame gases using a pneumatic 
nebulizer. The water droplets in the gas flow scattered the excitation 
light source, which was clearly seen by the detection apparatus. 


Lateral OH Emission Profiles of Methane-Air and Methane-Air- 
Coal Dust Flames Using an Inversion Technique 

In this set of experiments OH (0,0) emission profiles were taken 
of a series of flames at different heights. The methane air-flames 
had (t)'s of 0.66, 0.77, 1.00, 1.10, 1.30, 1.70 and <» while the methane 
air-coal dust flames had ^'s, of 0.66, 1.00, 1.10. The burner was 
moved across the slit of the monochromator by a motor driven X-Y 
milling table at a rate of 0.5"/min, the missions were monitored 
with the 1.26 m monochromator/PMT system and the resultant signal was 
recorded on a strip chart recorder run at a speed of 2"/roin. This 
resulted in 4 mm of chart paper for every millimeter in the flame for 
an approximate flame spatial resolution of 0.25 mm. The slit used was 
100 vim wide and 2 mm high. These spatial scans of OH emission intensity 
were digitized and stored on a PDP 11/34 computer for later calculations. 

In the spectroscopic study of cylindrically symmetric emitters, 
such as the flame in the present study, measurements are usually carried 
out on a small cross-section of the light source in a direction per- 
pendicular to its axis. A "side-on" measurement such as this includes 
contributions from the outer regions of the flame along with emissions 
fron the center of the flame giving an integrated average of a small 
strip in the flame. In order to separate the contributions of the 
different radial zones and to obtain the true radial characteristics 
of the source, one of the many graphical or numerical procedures for 
solving Abel'.s integral equation must be used (274). Upon solving Abel's 
equation one obtains the true radial intensity distribution. Computer 
programs were written and are given in the Appendix for data manipulation. 


The following programs were used: 

1. SETJEF. This program accepted the raw data and was used to 
convert chart paper units into intensities in nanoamperes and 
flame distance in millimeters. As some scans were made in both 
directions this program could also flip the data so all 
measurements wouM be from the same reference direction. 

2. SM005J. This program accepted data sets from the SETJEF 
program and performed a five point least squares smoothing 

of the data. This program was taken from the work of Savitzky 
and Golay (27 5) with modification necessary to function on our 
computer system. 

3. INVERA. This program accepted data sets from the SM005 program 
and performed an Abel inversion upon the data to obtain the 
distribution of emitters from the observed projected intensity 
profile. The procedure used to calculate the inversion is that 
of Barr (276) who used an analytic form for the observed 
emission which could then be solved numerically to obtain the 
inversion. In this method, if the horizontal slice is taken 
at a distance y from the center of the flame axis and is of 
height 6y (slit height) and width H (slit width) then the total 
light flux falling on the detector is given by 


I (y)8.6y=2.6y^ f (r)dx. 

where f (r)ildx6y is that part of the flux which originates in 
the volume element S,6ydx located at radial distance r from the 
axis. In this expression the integral is taken along a strip 
at constant y, x^ + y2 = r2, x2 = - y^, and R is the radius 
beyond which f (r) is negligible. Therefore the equation can also 

be written as 


C f(r)rd 
*Jv (r -v 

f (r)rdr 


y (r -y 

which is one form of Abel's equation with I (y) the measured 

If f (r) is zero for all r>R then the above equation 
inverts analytically (277) into 

f(r) . _^ (dl/dy)dy 
^ ' - ^ 2 2,1/2 

(y -r ) 


In most experiments I (y) is obtained as a set of numerical 
data rather than as an analytic function. Several different 
methods (278, 279) have been developed for performing the 
above integral numerically, but they all necessarily suffer 
from the fact that when the derivative of I (y) is evaluated, 
noise on the data is amplified. In the method of Barr 
the integration is performed before the final differentiation 
with a smoothing technique applied between the two. This 
method showed good response to noisy data and was of suitable 
form for canputer calculation. Once the inversion was 
obtained it was formatted to allowing direct plotting by a 
digital plotter. 

Figure 44 to 54 show the observed intensity and the inverted 
profile for a typical height in the flames studied. 

OH, CH and Emissions from a Single Cone Using an Inversion 


In these experiments the spectral omissions from a single cone 
were studied. A single cone was obtained by plugging all the cones 
except the center one with clay from underneath the burner head. The 
clay was scraped off the top of the burner to insure a flat surface to 
avoid any disturbing aerodynamic effects due to the clay. In these 
experiments the burner was moved at the rate of 0.2"/min with a chart 
speed of 2"/min yielding 10 mm of chart for every 1 ram in the flame 
for an approximate resolution of 0.05 mm in the flame. The 1.26 m 
monochromator/PMT system was used with a slit width of 100 ym and a 
height of 2mm. Measurements were taken at 1 mm intervals up to a 
height where the cone no longer existed ["^S mm) . 

In Figure 55 to 57 are emission profiles and inversion profiles 

for OH, CH and C^ at a height of 3 mm. In Figure 58 is a diagram of 

an idealized cone defining cone width, C , which is measured from one 


intensity maximum on one side of the cone to the other intensity 
maximum on the other side of the cone, and reaction zone thickness, R , 



9 18 27 

Radial distance (mm) 


Figure 44. OH emission flame profile (bottom) and inversion profile 
(top) for a flame with 0 = 0.66 at a height of 10 mm. 

224 - 


0.60 — 


^ 0.20 







3.61 - 

2.39 - 

1.16 - 

1 f 

9 18 27 36 

Radical distance (mm) 

Figure 45. OH emission flame profile (bottom) and inversion profile 
(top) for a flame with 0 = 0.77 at a height of 10 mm. 


9 18 27 36 

Radial distance (mm) 

Figure 46. OH emission flame profile (bottom) and inversion profile 
(top) for a flame with 0 = 1.00 at a height of 10 mm. 


9 18 27 36 

Radial distance (mm) 

Figure 47. OH emission flame profile (bottom) and inversion profile 
(top) for a flame with 0 = 1.10 at a height of 10 mm. 


9 18 27 36 

Radial distance (mm) 

Figure 48. OH emission flame profile (bottcm) and inversion profile 
(top) for a flame with 0 = 1.30 at a height of 10 mm. 


Figure 49. OH emission flame profile (bottom) and inversion profile 
(top) for a flame with 0=1.70 at a height of 10 ram. 


Figure 50. OH emission flame profile (botton) and inversion profile 
(top) for a flame with 0 = 1.70 at a height of 18 mm. 



Figure 52. OH emission flame profile (bottom) and inversion profile 
(top) for a flame with 0 = 0.77+Coal (30 rag/min) 


Figure 53, OH emission flame profile (bottom) and inversion profile 
(top) for a flame with 0 = 1.00 + 30 mg/min coal at a 
height of 10 mm. 


Radial distance (mm) 

Figure 54. OH emission flame profile (bottcm) and inversion profile 
(top) for a flame with 0 = 1.10 + 30 mg/min coal at a 
height of 10 mm. 




5. 00-, 




Radial distance (nun) 

Figure 55. OH emission flame profile (bottom) and inversion profile 

(top) fran a single cone with 0 = 1.00 at a height of 3 mm. 



0.375 — 

0.25 -z- 

'Jj 5.00 -t 



3.75 - 

2.50 - 

1.25 - 


Radial distance (mm) 

Figure 56. CH emission flame profile (bottcm) and inversion profile 

(top) from a single cone with 0 = 1.00 at a height of 3 mm. 






Radial distance (mm) 

Figure 57. C emission flame profile (bottom) and inversion profile 

(top) from a single cone with 0 = 1.00 at a height of 3 mm. 


region of 
maximum emission 

Figure 58. Ideal Cone Defining C^, 


which is the width of the cone wall as determined by spectroscopic 

emissions. C can be measured from both raw emission data and 

inverted data while R is best determined fron inversion data. Table 


18 shows C values obtained from both emission and inversion data 

while Table 19 shows R values obtained from inversion data. 


Calculation of Ground State Gbncentrations Assuming a Thermal 


The population in an upper state, n^, thermally populated from 

a lower ground population, n^, can be expressed as (252). 

n -AE/kT 
u _ e 


where AE is the energy difference between the upper and lower levels 
and Qint the internal partition function which is a product of the 
internal rotational, vibrational and electronic partition functions. 
The internal partition function is both a function of temperature 
and molecular properties and has been discussed in great detail by 
Tatum (280). 

Assiiming the flame to be in thermal equilibrium, which is not 
valid in the primary reaction zone, we can calculate a ground state 
population for the molecules. 

Table 20 shows ground state concentrations calculated using 
temperatures measured using a thermocouple (272) and excited state 
concentrations calculated earlier. Values of partition functions 
were obtained from the calculations of Tatum (280) . These calculated 
ground state values are significantly greater than experimentally measured 
values (79, 83) indicating a non-thermal excited state population due to 
chemiluminescent reactions. The maximum temperatures measured in these 
flames are given in Table 21. 


Table 18 

C (mm) Values from Emission and Kbel Inverted Data 

H (mm) emission inversion emission inversion emission 

1 3.90 3.82 3.50 3.59 3.10 3.11 

2 3.55 2.91 3.20 3.21 2.90 2.86 

3 2.85 2.68 2.50 2.42 2.10 2.14 

4 2.40 2.00 2.00 2.00 1.80 1.88 

5 1.50 1.67 1.10 1.08 0.65 0.69 

6 1.20 1.19 - 1.07 - 0.15 

7 - 0.97 . - 0.78 


Table 19 

R (mm) Values for Single Cone Emissions 

H : (mm) OH CH C^^ 

1 1.14 1.18 1.31 

2 1.25 0.95 0.97 

3 1.69 1.15 1.48 

4 1.80 1.63 1.15 

5 2.15 0.87 1.52 

6 2.06 

Avg 1.68 1.15 1.36 

a 0.41 0.29 0.23 


Tabl| 20 

Ground State Concentrations (#/cm ) Assuming a Thermal Distribution 
Height (mm) 0=1.0 0=1.O+Coal 0=0.77 0=O.77+Coal 0=1.10 0=1.1O+Coal 































Table 21 

Maximum Temperatures vs 0 for CH^/Air Flames 

0 T (K) 


0.66 1694 

0.77 1738 

1.00 2000 

1.10 1956 

1.30 1931 

1.40 1856 

1.70 1789 

1.90 1797 



Methane-Air Flames 

Effect of 4) Upon Excited State Concentrations 

Figure 59 shows the maximum population of OH*, CH* , C* and CO* 

as a function of (j) and Figures 26 to 36 show concentrations as a 

function of height in the flame for the different (f> values studied. 

From Figure 59 we can see that OH* has a maximum population at ^ = 1.1 

with a relatively flat distribution around this value from (J) = 0.90 

to 1.3. CH* is found to have a maximum value at <j) = 1.3 along with 

CO* while C* is maximum at * = 1,5. 

8 3 

Number densities at these ^ values are OH* = 5 x 10 /cm , 
CH* = 1.6 X 10^/cm^, C* = 8.0 x 10^/cm'^ and 6.9 x 10^/cm"^ for CO*. 
These can be compared with the values obtained by Porter et al. (79) 
who obtained concentration values of OH, CH and in both ground and 
electronically excited states in low pressure C^H^ and CH^ flames. 

Concentration values obtained for excited species in a fuel rich flame 

7 3 7 3 

{(}) = 1.58, p = 18 torr) were OH* = 1.8 x 10 /cm , CH* = 1.9 x 10 /cm 

* 7 3 

and C = 5 x 10 /cm . 


The values of Porter et al. can be converted to atmospheric 
pressure values assuming ideal gas behavior, in order to compare values. 
In the flame of Porter et al. a pressure of 18 torr was used and a 

















A A 




0.66 0.77 0.9 1.0 1.1 1.3 1.5 1.7 1.8" 

0 = OH 
A = CH 

■ = COxlO 

Figure 59. Maximum populations vs 0 


temperature of 1600 K was obtained for this flame. Using PV = nRT 
and converting torr into pascals we can solve for total number density 
at this pressure, 

n 2394 ^ ^^gQ ^ ^Q-1 j„Qies/i^3 

V (8.3144) (1600) 
Likewise, for an atmospheric pressure flame (101325 Pa) at 1900 K we 


obtain a number density of 6.41 moles/m . Taking the atmospheric 

value and dividing by the low pressure value we obtain 35.6 so to 

convert number densities at low pressure to atmospheric values we 

8 3 

must multiply by 35.6. We therefore obtain values of 6.4 x 10 /cm 

8 3 9 3 

for OH*, 6.8 X 10 /cm for CH* and 1.8 x 10 /cm for C^, from Porter's 

values which are in quite good agreement with the experimentally 

obtained atmospheric pressure flame values obtained in the present study. 

Ground state populations measured by this group (76) were OH = 

2.8 X 10"'"Vcm^» CH = 4.4 x 10"'"^/cm"^ and = 7.6 x 10''"Vcm^ in the 

14 3 

rich flame and in the lean flame (()> = 0.43) OH = 3.21 x 10 /cm , 

12 3 

and CH = 1.5 x 10 /cm with undetectable. Bulewicz, Padley and 

14 3 

Smith (83) obtained ground state values in 665 Pa flame of 1.1 x 10 /cm 

12 3 11 3 

for OH, 1 X 10 /cm for CH and 3 x 10 /cm for C^ in a flame with 

(j) = 1.0. As the flame was made rich ((j> = 1.5) the OH concentration 

13 3 13 3 
dropped to 6 x 10 /cm with both CH and C^ increasing to 1.5 x 10 /cm 

13 3 

and 1.2 x 10 /cm respectively. Maximum concentrations were obtained 
at equivalence ratios of 1.0 for OH, 1.4 for CH and 1.5 for C^. Only 
relative excited state populations were measured in this study with 
arbitrary units of 90 for OH*, 6.5 for CH* and 0.8 for C^ at (j) = 1.0. 
Cattolica (281) has obtained similar ground state concentration values 


in an atmospheric methane-air flame while Heffington et al. (282) have 
obtained excited state concentrations in good agreement with the present 
study in a shock-tube study of CH^/O^/Ar mixtures. No previous 
studies of excited state absolute concentrations in atmospheric 
methane-air flames were found in the literature to obtain a direct 
comparison with the present study. 

On comparing Figure 59 with Table 21 we see that the maximum 
temperature of 2000 K is found at ^ = 1.0, while maximum OH concen- 
tration is found at (|> = 1.1 with a relatively small change frcxn ()> = 0.9 
to 1.3. In Figure 29 for (j) = 1.00 we see that initially the OH* con- 
centration is greater than the CH* but after a height of 2 mm the CH* 
is greater until after 7 mm, v*ien the OH* is once again greater. The 
CH* emissions are no longer visible after a height of 9 mm whereas the 
OH* is visible to a much greater height. This extended OH emission 
is caused by radical reccmbination reactions of the type 

H + OH + OH ^ OH* + H^O 


H + O + M OH* + M 

Experimental evidence (281) has shown that atomic oxygen has its 

peak concentration in lean flames with a maximum value at ()> = 0.9, 

while hydrogen atom reaches a maximum in rich flames at ^ = 1.2. 

Therefore in lean and stoichiometric flames the primeiry CH^ destruction 

step is 

CH^ + O ^ CH + OH 
4 3 

which accounts for the high concentrations of OH radicals in these flames. 


As the amount of methane in the flame is reduced (increasingly lean) 
it is obvious that less carbon is available for molecule formation so 
that as (|> approaches zero, carbon containing molecules should decrease. 
This is the effect seen in Figure 59 with the molecular species con- 
taining more carbon, c* , declining much faster than CH*. The 
production of C* involves at least 2 or more carbon fragments coming 
together which is less likely as <i> decreases. This same effect causes 
a drop in CH* and CO* emissions which are both dependent on multi- 
carbon containing molecules for their production. 

Also in this figure one can see evidence for the reaction 

CH + CO + OH* 

because the decrease in OH* appears to speed up once the CH* population 
begins to decline rapidly at ())<_ 0.90. Another interesting feature in 
the figure is the way the CO decreases as the OH and O atom concentra- 
tions become smaller, possibly indicating that reactions 

CH* + 0 ->■ CO* + H 

CH* + OH CO* + H 


C^O + O CO* + CO 
may play a role in CO* formation. 

As the flame is made more fuel rich we see a substantial increase 

in carbon containing species and a slight increase in OH* at (j) = 1.1 

with a constant reduction at larger ({>'s. The CO*, CH*, and C* reach 

maxima at successively larger equivalence ratios. In these rich flames 

we now have more hydrogen atoms than oxygen atoms and the predcciinant 

methane destruction reaction is 

CH. + H ^ CH, + H„ 
4 3 2 



CH^ + OH -> CH^ + H^O 

also occurring in the region near (|) = 1.00. At large values of 

(j) (> 1.3) the reactions 

CH + M ^ CH, + H + M 
4 3 

CH3 + CH3 - C^H^ 

become more probable. Warnatz (283) has calculated that for lean or 
stoichiometric conditions about 3 0% of the methyl radicals are 
consumed by recanbination, with the percentage increasing to 80% 
for rich flames. 

As the equivalence ratio is increased the maximum temperature 
drops from a value of 1956 K at ({> = 1.1 to 1789 K at (}> - 1.7. This 
decrease in temperature is due to incomplete combustion occurring due 
to a lack of oxygen. This is also the reason for the decrease in CO* 
because all the discussed mechanisms partially depended upon the 0 
or O^ concentration with the exception of 

CH* + OH ^ CO* + H^ 
which depends upon OH which is also decreasing rapidly in these 
rich flames. 

A possible explanation for the behavior of CH* is that even though 
the oxygen concentration is decreasing, the OH concentration stays at 
a sufficient level, due to the slow nature of the radical recombination 
reactions, in order to ccanbine with the C^ which is increasing until 
(j) = 1.5 is reached. The C^ concentration should increase with larger 
({)*s if Gaydon and Wolfhald's polymerization mechanism is responsible 


for the formation of C*. Formation of ground state C is also 

favored by higher ()) by possibly being formed from C H as discussed 

2 6 

previously. As the equivalence ratio increases further the C* and CH* 
concentrations drop possibly due to the decrease in temperature or due 
to larger polymeric molecules forming in the excess hydrocarbon 
environment, which are less likely to decompose into C* and CH*. 

Methane-air flames with large equivalence ratios appear very 
luminous to the eye due to the formation of soot particles in the 
flame (284, 285). These soot particles are thought to be made out of 
polycyclic arcxnatic hydrocarbons and long chain hydrocarbons which have 
condensed into larger particles. Soot in flames is what appears as 
smoke and can be observed in most any fuel rich flame. When the soot 
is formed it may possibly attract small hydrocarbons which may be 
precursors to C* and CH* therefore lowering the amounts of these species 
produced. Flames observed in these studies evidenced signs of soot 
formation (luminous character) at a (t) of 1.5 with increasing 
Iximinousity at higher (j> values. 

Spatial Temperature Profiles and Lateral OH Emission Profiles 

A flame is not a uniform hot collection of gases but has different 
temperatures at different spatial positions depending upon the equivalence 
ratio (9) . Tables 15 and 16 show spatial temperatures vs height for 
two methane-air flames with different if values. Figures 44 to 54 
show integrated lateral intensities of OH emission along with the Abel 
inverted plots to show spatial onission intensities. Frcm the above 
figures it becomes evident that the temperature distribution and 
emission profiles are highly dependent upon the value of (j). 


In these studies the total flow volume was kept constant so we 
would only observe the effects of varying (j) without worrying about 
a change due to aerodynamic effects. In the normal stoichicmetric 
case the cones were well defined with a height of approximately 4 mm. 
The flame ran without any visible fluctuations and appeared to have a 
well defined shape. The overall color of the flame at this ^ value 
was a bluish-green color in the cones with less green color higher 
up in the flame when observed in the dark. When room lights were 
on the cones still retained their color but the remainder of the flame 
appeared colorless. Also at this equivalence ratio the flame had 
a total height of about 10 cm. Above the cones at a height of 10 mm 
the flame appeared to be uniform and homogeneous to the eye. 

As the flame was made lean 1.0) several noticeable changes 

occurred with the first observable one being a change in the color of 
the cones to a much bluer color with a slight touch of green at a (j) 
of 0.90, Also the flame decreased in height to about 8 cm at this 
equivalence ratio. The cones also appeared to change shape now having 
a height of about 3 mm. and appearing much more squat. The flame still 
appeared uniform but seemed to be more transparent than the stoichio- 
metric flame. At an equivalence ratio of 0.77 there was a visible 
change in the appearance of the flame. First there was a great reduction 
in height down to approximately 5 cm with an extremely transparent 
bluish color. The outer cones on the burner consisted of only the inner 
edge and the flame gave the appearance of being blown upward. There 
were some small gaps between the cones and the flame consisted more of 
streamers than of a uniform flame. These inhomogeneities can be seen 



in the emission profile. Figure 45, and are clearly evident in the 
inverted profile above the emission. 

At a <() Of 0.66 no cones were visible with the flame assuming a 
very distorted shape. The flame had decreased greatly in height down 
to approximately 3 cm and seemed to consist of disjointed pieces. In 
this flame there was a great amount of movement and turbulence and the 
flame had the shape of a large V. In this flame there was only a faint 
trace of bluish color, and the falme was very transparent. Temperature 
profiles of this flame show a low temperature in the center which 
increases outward which is in direct contrast with the stoichiometric 
flame which is hotter at the center. 

A completely different series of changes occurs when the flame is 
made slightly rich (({i = 1.10). The major observable change is in the 
color of the cones. Overall flame shape and size is identical to the 
stoichiometric flame but the cones now have more of a greenish color as 
opposed to a blue-green color in a stoichiometric flame. The cone 
height and shape are identical and the OH emission profile is basically 
identical to the stoichiometric flame with the inversion showing a 
slight increase in emissions at the flame center. As the equivalence 
ratio becomes larger the cones become greener and gradually increase in 
height until at (j) = 1.3 they all coalesce into one large cone with an 
approximate height of 2 cm. As the fuel is increased the cone increases 
in height until at (j) = 1.5 the cone is no longer visible and the flcime 
becomes luminous, giving off an orangeish-yellow light similar in 
appearance to that emitted by a candle. At this point the total flame 
length has increased, reaching a height of 15 cm with the upper 5 cm of 
the flame not well defined due to its turbulent, flickering appearance. 


At (j) = 1.7 the flame has the same dimensions as at <|) = 1.5 
with the exception of more Ixaminousity and a little more of a flickering 
upper area. An OH emission profile at a height of 10 mm (Figure 4i9) 
shows a maximum intensity at the edges with less in the middle while 
the inversion shows that the edge emission comes from what appears to 
be 2 zones at the edge. This effect is more pronounced at a height 
of 18 mm {Figure 50) where these two zones are seen clearly in both 
the emission and inversion plots. Referring to the temperature profile 
for this flame (Table 16) we also see that the hottest region of the 
flame is out at the edges where the maximum emission occurs. 

The reason for these two emission zones is as follows. In such 
a rich flame as this there is not enough oxygen present to completely 
combust all the fuel. Even though it is not visible in this luminous 
flame the primary reaction zone is where the combustion occurs and in 
this case it is near the edge of the flame which we observe as the 
inner peak. The outer anission zone is another region similar to the 
primary reaction zone but in this case the room air diffusing inward 
towards the flame allows combustion to occur in this region at an 
effective equivalence ratio smaller than the inner combustion zone, 
therefore resulting in a greater intensity of OH missions which would 
be indicated by Figure 53. 

In a diffusion flame, which has no premixed air, the inner 
primary reaction zone should not exist and all combustion will only 
take place at the outer edges. As can be seen in Figure 51 in the 
diffusion flame only one region of combustion occurs on the boundary - 
layer between the fuel and the room air. The maximum temperature in 


a diffusion flame is also lower due to the inefficient transfer of 

oxygen into the flame as opposed to a pre-mixed flame. 

Single Cone Data and the Formation of Excited States 

The spectral emissions from a single cone were studied in order 

to gain some understanding of the sequence of processes occurring in 

the primary reaction zone. Figure 58 shows an ideal cone and defines 

the cone width, C , and the reaction zone thickness, R . The emissions 
w w 

of OH*, CH*, and C* were measured spatially and the cone width and 
reaction zone thickness were obtained from emission plots and inversion 
plots. The direction of the burning velocity is also indicated in 
Figure 55 which can also be used as a reaction coordinate with events 
in the inner edge of the cone occurring before events on the outer 
edge. Measured values for C and R are given in Tables 18 and 19 


and emission and inversion profiles at a height of 3 mm are shown 
in Figures 55 to 57. 

In these series of measurements the single cone appeared to be 
a slight bit larger ('^' 5 mm high) and wider than the cones in a full 
flame which can be attributed to a neighbor effect in the full flame, 
whereby the cones were not able to spread out due to the influence of 
a nearby cone. In a full flame the cone base was about the size of 
a burner hole (0.125") while with a single cone the base extended 
approximately 1.5 mm out from the edge of the hole. The fuel and air 
flow rates were adjusted to obtain a steady and well shaped cone and 
the ratio of the two flows were adjusted to obtain a stoichiometric flame. 

From Table 18 it can be seen that C* reaches a maximum value first 
followed by CH* and then the OH*. The average Separation between the 


C* maximum and CH* is 0.39 mm with a standard deviation of 0.06 and 

the average separation between the CH* maximvun and the OH* maximum is 
0,38 mm with a standard deviation of 0.03. Using a value for the 
burning velocity of a stoichicmetric methane-air flame of 45 cm/s, 
obtained frcm Andrews and Bradley (286) , we convert distances in the 
flame into time. Upon making this conversion we find that the CH* 
emission peaks 0.87 ms after the C* maximum with the OH* maximum 
occurring 0.84 ms after tje CH* maximum and 1.71 ms after the C* 

The reaction zone width, R , of each species is such that there 


is a considerable overlap between the 3 zones, with C* emission reach- 
ing past the center of the OH* reaction zone. From the first appearance 
of C* it is 0.15 mm or 0.33 ms to the first appearance of CH* and 
0.4 mm or 0.88 ms to the first appearance of OH* emission, while the 
first appearance of OH* occurs 0.28 mm or 0.56 ms after the first 
appearance of CH*, as observed at a height of 1 mm. The region of C* 
emission persists until 0.25 mm (0.55 ms) before the end of CH* 
emission and until 0.44 mm (0.97 ms) before the end of OH* emission 
yielding a total reaction zone width of 1.52 mm, as measured by optical 
emissions from C*, CH* and OH*. 

These data shows that in the primary ccxnbustion zone the excited 
species are formed in the following order C*, CH* and OH*. Porter 
et al. (79) in a study on low pressure '^2^2'''^2'^^2 ^•'■^^^ found that 
in a rich flame ground state C^ and CH reached a maximum before excited 

C and CH with both ground state maxima at a height of 12 mm with C* 
^ 2 

at 16 mm and CH* at 17 mm. OH* reached a maximum value at 11 mm vAiile 


OH ground state was maxiraiim at 20 mm in this flame. However in a lean 
flame all three excited species, OH*, CH* and C* reached a maximum 
value at 5 mm with maxima for both ground state CH and OH at 9mm, 
with ground state undetectable in this flame. These results indicate 
different pathways for the production of ground states and excited 

In the present study the pattern of emissions can be explained 

by the following mechanism. Assviming for now that the C* is formed 

in some unknown step, we observe that the onset of CH* emission occurs 

0.33 ms after the beginning of C* emission. Due to the short radiative 

lifetime of C*, 160 ns (287), a ground state population of will 

build up via the reaction 

C* ->■ C +hv 
2 2 

which has a rate of 6.27 x 10^ transitions s ^. In a stoichiometric 

flame the predominant initiation mechanisms are 

CH^ + OH -J- CH + EO 
4 3 2 

CH^ + O -> CH^ + OH 
which imply a reasonable concentration of OH radicals in the primary 
combustion zone under stoichiometric conditions. Therefore CH* can be 
produced by the following reaction 

C^ + OH CO + CH* 
with the C^ being produced by the radiative decay of C*. 

Once the CH* is formed it too will undergo radiative transitions, 

CH* ■* CH + hv 

resulting in a buildup of ground state molecules. In the case of CH* 
the radiative lifetime is longer than the lifetime of C* with a value 


of 508 ns (286), resulting in a decay rate of 1.96 x 10 transitions 
s ^, thereby resulting in a slower buildup of ground state molecules 
than the case. We observe that the onset of OH* emission occurs 
0.56 ms after the beginning of CH* emission. This indicates the 
production of OH* is due to 

CH + 0^ OH* + CO 
where the CH is produced through radiative decay of CH*. Due to the 
longer lifetime of CH* one would expect the production of OH* to take 
longer than the production of CH* from C^ which is observed in this 

As for the formation of C*, this study indicateis that it should 

be a relatively fast process and precurser for the radicals OH* and 

CH* in a stoichiometric methane-air flame. A possible mechanism 

would involve a scheme such as 

CH, + O CH^ + OH 
4 3 

CH, + H ^ CH^ + H^ 
4 3 2 

CH^ + OH ^ CH^ + H^O 

followed by formation of CH^ by 

CH, + OH CH^ + H^O 
3 2 2 

CH^ + H CH^ + H^ 
CH + 0 ->- CH + OH 


CH3 + CH^ - C^H^ 

which forms a molecule containing two carbons which can be reduced to 
C by hydrogen atom abstraction. 


Another possible reaction scheme is 

CH^ +0 ^ CH + OH 
CH^ + OH ^ CH + H^O 
CH^ + H ->• CH + H^ 

followed by 

CH^ + CH -> C^H^ 

^2«3 ^ °2 " S« 
C^H + OH C* + H^O 

C^H + 0 C* + OH 
2 2 

Although it may result in a faster overall reaction process , it 
is not very likely to form C* due to the endothermic nature of the 
final two reactions with values of +32.79 and +18.00 kcal/mole 

The only other possible mechanisms for the formation of C* would have 

to involve C^. Possible endothermic reactions for the production of C* 

from C include 

+ 0 CO + (-80.8 kcal/mole) 

and + OH C^ + COH (-3.41 kcal/mole) 
with the former more likely than the latter to form C*. 

Me thane -Air -Coal Dust Flames 

To determine the chemical and physical processes occurring when 
an amount of coal dust is added to a methane air flame we must examine 
the nature of the changes which occur upon the addition of coal. 
Figures 37, 38 and 39 show excited state concentrations as a function 
of height for a methane-air-coal dust flame with equivalence ratios 
of 1.0, 1.1 and 0.77 with a coal flow rate of 30 mg/min. Figures 27, 
29 eind 30 show the respective flames without coal. Figures 40, 41, and 42 


show the change in emissions when coal is added to a flame. 

This is defined as 

= ratio of change 

where LAJ 4 

CH, + coal 

is the niimber density of a species in the methane 

air-coal dust flame and [a] 4 is the number density of the same species 
in a ihfethane-air flame of the same equivalence ratio. Therefore a 
number less than one indicates less of a certain species in a methane- 
air-coal dust flame than in a methane-air flame while a number greater 
than one indicates an increase upon the addition of coal. 

Table 17 shows the percentage change in temperature upon the 
addition of coal, which is calculated for the flame center in a similar 
way as the concentration percentages. Table 22 shows flame center 
temperatures for two methane-air-coal dust flames and four methane-air 
flames showing measured temperatures from 2 mm above the burner head 
to 4 mm below the head. These beneath the burner head temperatures 
were obtained by inserting the thermocouple down the center hole. 
Temperatures obtained at negative values were not corrected for radiation 
losses as was the case for the above head measurements. 
Stoichiometric (<{) = 1.0) Flame 

One change that occurs in a flame with =1.0 which can be seen 
from Figures 29 and 38 is that the C* and CH* emissions are visible 
at greater heights in a flame containing coal than one without coal. 
In the methane air flame the signals from C* and CH* became undectable 
at heights of 8 and 9 mm respectively while in a flame containing coal 
dust they were detectable to heights of 9 and 13 mm respectively. 


Table 22 

Center Line Thermocouple Temperatures (K) for Methane Air and 
Methane Air Coal Dust Flames (30 mg/min)^ 

Height(mm)/0 1.0 1.0+Coal 1.1 1.1+Coal 0.77 0.77+Coal 

2 1987 1790 1939 1755 1589 1270 

1 1940 1686 1865 1703 1433 1139 

0 1838 1658 1763 1659 

-1 1266 1222 1270 1087 

-2 1078 1097 1196 946 

-3 950 889 950 840 

-4 890 810 870 750 

Zero height = burner head surface 


In the observations in the methane-air flame the limiting noise, 
which determined the lower limit of detectability , was detector noise, 
for all the species, while in the methane -air-coal dust flames, the 
limiting noise source was flame background emission due to particle 
incandescence for the two species in the visible region C* and CH* 
with detector noise as the limiting noise for the 2 ultraviolet emitters , 
OH* and CO*. This is why in the flames with coal the same detection 
limits could not be obtained for C* and CH* as in the normal methane-air 

Fran the temperature profiles. Tables 17 and 21 it becomes apparent 
that the addition of coal to a flame reduces the temperature. Before 
the coal can begin to burn it must reach an ignition temperature and 
this heat is provided by the flame, which is cooled in the process. 
Upon the addition of coal to a stoichictnetric flame the temperature drops 
by a large amount initially until the coal particle can begin to heat 
up whereupon the flame temperature is within 1% of its gas flame value 
at a height of 6 mm. 

From Figure 41, change in concentrations upon addition of coal, we 
can see several interesting features. Initially at a height of 1 mm all 
species in the coal-gas-air flame are decreased relative to a methane air 
flame with the exception of C* which shows an increase of 280% over that 
of a normal methane air flame. As we progress to a height of 2mm 
we see a decrease in C* over the previous value but still much 
greater than a methane-air flame by 238%. As this is occurring OH* 
increases to a +12% value and both CH* and CO* show increases from 
their 1 mm values with gains of 11 cind 26% respectively although they 
are still below normal methane-air values. At a flame height of 3 mm 


the C* drops to its lowest value but still retains a positive value 
and at this height OH* experiences a local minimum and CO* a local 
maximum. Here also CH* has increased to a +30% value, a 42% increase 
from the 2mm concentration. 

After 3 mm the concentrations of C* and CH* rapidly increase obtain- 
ing values of +1780% at 8 mm for C* and +1840% at 9 mm for CH*. 
CO* declines after 3 mm and then rises to obtain a + 5% value at 8 mm 
which is also where OH* obtains its maximum value of +15%. The CO* 
then dips down to -3% at 9 mm returning to -6% at 10 mm while the OH* 
drops to 0% at 9 mm and oscillates around this value. 

These observations can be explained with the aid of Figure 60. 

As the coal particles enter the burner through the bottom they first 

pass through a preheating and devolatilization zone where the coal 

particle undergoes heating at large heating rates as shown in Table 23. 

When the peirticle exists the burner it enters a region of high temperature 


(>1600 K) which also possess a high heating rate (1.20 x 10 K/s) 
although not as large as under the burner where the velocities are 
greater. This preheating zone and its high heating rate promotes the 
production of volatiles from the coal particle. As a result when the 
coal particle enters into the flame it is surrounded by a cloud of 
volatiles produced in the pre-heating zone which causes the original 
stoichiometric equivalence ratio to become rich. 

At a burner height of 0 mm the coal particle and its accompanying 
volatiles enter into the flame where the volatile can immediately 
start to burn. Once this volatile cloud starts to burn the coal 
particle will be open to attack from flame gas species or ccmbusted 






u c 

(d -H 

Xi +> 

u Id 










distance time 

\ . 



V • 






mm ms 

Figure 60. Proposed Coal Burning Model for 0=1.0 Flame. 


Table 23: 

Devolatization, Char Heating and Char Ignition Times for 
Coal Particles (30 mg/min) in Methane Air Flames 

0=0.77 0=1 . 00 0=1.10 

Devolatization (ms) 8.0 5.5 8.6 

Char Heating 2.9 3.1 6.7 

Char Ignition 





Table 24 

Heating Rates for Imm Travel in Methane Air Coal Dust Flames (K/s) 

Height interval (mm) 





0, +1 

1.20 X 


1.98 X 


-1, 0 

8.72 X 


1.14 X 


-2, -1 

2.50 X 


2.82 X 


-3, -2 

4.16 X 


2.12 X 


-4, -3 

1.58 X 


1.80 X 


Zero height = burner head surface 


Table 25 

Analysis of Coal and Natural Gas Used for Study 

Coal - Proximate Analysis 

% Fixed carbon 
% Volatile matter 
% Ash 

% Moisture 





Natural Gas Analysis 


Mole % 


Hexane + 



volatile species to further promote devolatization . As the volatiles 
are produced the flame gases surrounding the particles will becane 
extremely fuel rich and possess characteristics of a fuel rich flame. 
After all the possible volatiles are produced the coal char patticle 
must heat up in order to begin combustion. This region is designated 
as the char heating zone in Figure 60 and during this time period the 
released volatiles should burn and be destroyed. 

Once the particle leaves this char heating zone, which is defined 
as when the particle starts burning, it will once again produce local 
fuel rich conditions as combustion takes place with attack of the coal 
particle by the various species present in the flame. This char 
combustion zone is where the char is finally consumed and has the 
largest length of the 3 zones discussed. The length of time the particle 
spends in these zones will be dependent upon flame conditions such as 
temperature and equivalence ratio, which governs the species available 
to attack the coal, as well as coal characteristics such as particle 
size and rank. 

Looking at Figure 41 and Figure 60 in more detail we can explain 
the observations in this flame as follows. Before the coal particles 
enter the flame zone they have undergone high rates of heating for a 
period of 2 ms, with a resultant production of a substantial amount 
of volatiles. As the pcirticle volatile mixture enters the flame the 
volatiles immediately begin to burn, resulting in a fuel rich mixture, 
which produces an increased amount of C* emission, similar to the case 
of a fuel rich methane air flame. Therefore we observe greatly 
increased C* emissions at a height of 1 mm which is 4.2 ms after the 


beginning of heating in the pre-heating and devolatilization zone. 
At a height of 2 mm we still see very elevated levels of C* indicating 
that more volatiles were produced in the flame from the coal particle 
during the 2 ms spent in the flame gases at very high temperatures 
(>1900 K) . At this point it should be remembered that the observations 
of the emissions from the flame will have a slight time lag from the 
actual physical and chemical processes actually occurring to the coal. 

At this point in the flame (2 mm) we also observe an increase in 
OH*, CH* and CO* concentrations froti the initially observed values. 
The large excess of C* will start to decay, forming ground state C^, 
which will promote the reaction 

+ OH -> CH* + CO 
which will enhance the likelihood of 

CH + O^ -> CO + OH* 
occurring . The increase in CO* could be caused by the reactions 

CH* + O -> CO* + H (-104 kcal/mole) 

CH* + OH CO* + H^ (-106 kcal/mole) 

C^ + O^ ->■ CO* + CO (-117 kcalAiole) 

Due to the larger amount of C* produced in this flame the following 
reactions may become possible 

C* + OH CO* + CH (-12 kcal/mole) 

C| + 0 -> CO* + C (-34 kcal/mole) 

C* + 0^ ->- CO* + CO (-172 kcal/mole) 

Heats of reaction for reactions involving excited states were cal- 
culated using the JANAF (289) tables to obtain heats of formation 
for the ground state species at a typical combustion temperature (1900 K) 


in a stoichiometric flame. To Obtain heats of formation for excited 
states the energy of the respective transitions (erg/molecule) were 
calculated using molecular constants obtained from Huber and Herzberg 
(290) and then converted to a value in kcal/mole using relations given 
by Fischbeck and Fischbeck (291) . The JANAF value for the ground state 
molecule was then added to this value to obtain the heat of formation 
for an excited species. The total heat of reaction was then calculated 
by summing the heats of formation of the products and subtracting the 
sum of the heats of formation of the reactants. These values calculated 
for over all heats of reaction should be considered as an upper limit 
as it does not account for adiabatic reactions. 

A possible explanation for the increase in C* other than just the 
effect of a locally rich flame can be given due to interaction of the 
volatiles under the burner head. It may be possible that, when the 
volatiles are released underneath the bxirner head, they will be constrained 
from diffusing great distances, and a high density vapor will be formed. 
In the hot region under the head, the volatiles may react and condense 
with themselves to produce molecules containing more than one carbon 
and possibly large molecules such as CgH^. These large molecules 
would be highly susceptible to attack by the radical species in the 
flame with the possibility of forming or even C* due to the strongly 
exothermic nature of radical reactions. If is produced in the 
combustion of these large molecules, the most likely process to occur 
would be 

C +0 ^ C* + CO (-22 kcal/mole) 


or possibly 

C* + OH ->- C* + CHO (-17 kcal/mole) 

FrcMii its minimum value of + 4% at a height of 3 nin,the C* 
concentration shows a rapid increase to values of + 28%, + 152% 
and + 433% at heights of 4, 5 and 6 mm. A rapid increase in the 
concentration of CH* is also seen to occur, but at a slightly later 
time than the C* rise, which is to be expected if CH* is formed from 
C^. This very large increase in the emissions of C* and CH* indicate 
the ignition of the coal char particle and its continued ccmbustion. 
As the coal char particle burns, the reaction products will be rich in 
carbon type molecules indicating a rich combustion zone around the 
particle which favors the production of C* and CH*. The coal particle 
begins to burn 8.6 ms from the time Of initial heating under the burner 
and 3.2 ms fran the end of volatile evolution. Possible chemical reaction 
mechanisms for coal char combustion will be discussed in a later section. 

Figure 60 shows a diagram for the processes occurring to a coal 

particle in a stoichiometric flame, along with a time scale for the 

various zones and heating rates present in these zones. It should be 

noticed that gas flow rates are faster before ccmbustion takes place 

with a velocity of 200 cm/s before ignition and a velocity of 45 cm/s 

in the flame. This variance in gas flow rates has been studied in great 

depth in the work of Lewis and Von Elbe (9) and the two velocities are 

related by the equation = sin a as given in Chapter II, where 

S is the velocity in a burning flame and V is the velocity of the 
u o 

input gases before ignition and a is the cone half-angle. 


Table 22 gives times for the various processes of devolatilization 
and char heating in flames of different <}> values and the percentage 
changes for flames with ^ = 0.77 and 1.1 upon the addition of coal 
are shovm in Tables 15 and 17. Table 34 compares the heating rates 
between two methane-air-coal dust flames with different values. 
Lean (({> = 0.77) and Rich (<{) = 1.1) Flames 

Figures 40 and 42 shows changes for a lean and rich flame upon 
the addition of coal dust. The general shape of these curves is 
similar in form to the one for stoichiometric flame, i.e. initial 
increase in C* emissions followed by a reduction in emission with large 
increase in both C* and CH* at later times. However there are several 
distinct differences in each flame which could be accounted for by the 
physical and chemical ccmposition of these flames. Table 17 shows 
percentage temperature change upon addition of coal while temperatures 
beneath the burner head are given in Table 21 for a methane-air flame 
and methane-air -coal dust flame with equivalence ratios of 1.0 and 1.1. 

In the case of the lean methane-air-coal dust flame we see once 
again a large increase (+54%) in C* early in the flame with a reduction 
to +22% at 4 mm which later increases to +202% and +288% at heights 
of 5 and 7 mm respectively. These increases in C* show the same general 
trend as was seen in the ij" = 1.0 methane-air-coal dust flame but with 
an overall reduction in percentage increases when compared to a methane- 
air flame with cj) = 1.0. Once again we observe that the concentrations 
of all other species are decreased relative to the gas flame. In this 
flame the coal char particle takes a longer time to start burning 
as ccmpared to the stoichiometric flame. Coal char ignition times 


are given in Table 33 along with pre-heat devolatilization and char 
heating times for the three coal flames studied. Fran the Table we 
observe that the coal particle takes a longer time to devolatilize 
in a lean flame (8 ms vs 5.5 ms) which results in a later start for 
char combustion (10.8 ms vs 8.6 ms) after a char heating time of 2.9 ms, 
very close to the sbaichicxnetric value of 3.1 ms. 

Comparing temperature values for a (j) = 1.0 flame with coal and 
a <j) = 0.77 flame with coal we find that the lean flame containing 
coal is greater than 500 K cooler than the stoichiometric flame at 1 mm 
above the burner head and approximately 400 K cooler at a height of 2 mm. 
Comparing temperature values for flames of (j) = 1.0 and 1.1 with and 
without coal we see that temperature difference between 1 mm below the 
burner head and just above it is approximately 520 K. Therefore at 
1 mm below the burner in a lean coal dust flame, the temperature is 
probably "^620 K, whereas in a (ji = 1.0 coal dust flame the temperature 
at 1 mm below the burner head was 1266 K, and even at -4 mm a temperature 
of 890 K was present. These low temperatures under the head in a lean 
coal dust flame will result in much lower heating rates, resulting in 
a smaller amount of volatiles being produced when compared to a 
stoichiometric coal dust flame. 

The smaller amount of volatiles produced in the pre-heating 
devolatilization zone would therefore yield a smaller percentage increase 
in C* emissions than a large amount of volatiles. Another reason for 
the small observed increase in C* emissions, when compared to the 
increase in the (j) = 1.0 flame, is that because the lean flame has an 


excess of oxygen it may be able to absorb more excess volatiles than 
the stoichiometric flame before forming a fuel rich region surrounding 
the coal particle. But the low level in C* is most likely due to the 
low heating rates and low temperatures found in the methane-air coal 
dust flame. 

An interesting feature in this flame is although it takes longer 
to generate volatiles in this flame the heating time for a char particle 
is about the same as that in the stoichiometric coal duest flame. 
Also in the other two flames with coal dust ((j) = 1.0, 1.1) there 
is not much of a change in the OH* concentrations with the largest 
increase being +23% in the rich flame and the largest decrease being -6% 
in the stoichiometric flame, but in the lean flame however we observe 
a 55% decrease initially at 1 mm which slowly rises to a 0% increase 
value at 7 mm which then rapidly rises to a value of +397% at 12 mm. 
This large rise in OH* occurs after the char has started to burn, which 
may indicate a different char combustion mechanism in a lean flame 
compared to rich and stoichicxnetric flames. 

When coal is introduced into a rich flame we observe the lowest 
increase in the initial C* concentration of all the methane-air-coal 
dust flames. Figure 42 showsa small increase (37%) at 1 mm with a 
slight rise to 51% at 2 mm with higher values declining until a low of 
20% is reached at 5 mm whereupon the coal char ignites and rises to a 
value of +44% at a height of 8 mm. In this flame the initial value 
for OH* and CH* shows small increases of +6 and +3% respectively 
at a height of 1 mm. The CH* concentration then drops to a -40% 


value and gradually returns to 0 by the height of 4 mm. Between 

heights of 5 and 6 mm the CH* concentration shows a greater value than 

the C* concentration but C* easily passes the CH* gain when the char 
2 2 

particle starts to combust and C* concentrations increase to a +109% 
value at 7 mm and a -443% value at 8 mm. The concentration of OH* 
stays near a value of +10% except for a region between the heights of 
7 and 9 mm where percentage increases greater than 20% are located. 

In this flame the coal char takes 15.3 ms to reach the ignition 
point, the longest time of any in this study. Volatile production is 
found to end after 8,6 ms while the char heating zone is 6.7 ms long. 
The calculation of values for the end of devolatilization and beginning 
of char heating time was rather difficult due to the lack of a clear 
separation between the devolatilization zone and the char heating zone 
as is present in the stoichiometric flame. The devolatilization was 
thought to have ended at a height of 3 mm in the flame due to the drop 
in percentage of C* when conpared to the value at 2 mm (32% vs 52%) 
in analogy with the stoichiometric C* concentrations at 1 and 2 mm used 
to determine the end of devolatilization in that flame. 

As can be seen from Tables 17 and 21, the temperature in this flame 
is less than the coal flame with (j> = 1.0 both above and below the 
burner head. Even though the heating rate (Table 23) is comparable 
to the heating rate in a (j) = 1.0 flame, the difference in final 
temperatures (200 K less 1 mm below burner head, 100 K less 1 mm above 


burner head) causes a resultant decrease in the devolatilization rate. 
The char heating time until ignition is also affected by this difference 
and also possibly by the oxygen deficient atmosphere of this flame. 
Chanical and Physical Processes in Methane-Air-Coal Dust Flames 

The methane-air-coal dust flames observed in this study can be 
broken down into 3 distinct zones in which different chemical and physical 
processes are occurring. In the first of these zones, the pre-heating 
and devolatilization region the coal particle encounters two completely 
different environments. The first zone extends from beneath the burner 
head to a certain distance into the flame depending upon flame (J) value. 
In this zone the particle undergoes at first just a heating effect under 
the burner which promotes the generation of volatiles. Upon entering 
the flame the particles are still affected by the temperature but are 
now also surrounded by a highly reactive combusting environment. The 
previously generated volatiles will immediately begin to turn upon 
entering the flame and these reaction products may further affect the 
coal particle subsequent volatile production. 

After the coal particle has lost all of its available volatiles 
it enters into the char heating zone where the particle heats up 
to a temperature sufficiently high for combustion to ccramence, upon 
which point it enters into the char ccxnbustion zone. Both of these 
zones are totally surrounded by burning or burnt flame gases at high 
temperatures. In these zones the rates of heating are relatively small 
due to the relatively uniform temperature of the hot reaction products 
and their constant drop in temperature as they rise in the flame. 


The cOTibustion of the whole char particle will take the longest time 

of these three processes and the end of this zone was not observable 

in the flames in these studies. 

In the stoichiometric methane-air-coal dust flame a devolatilization 

time of 5.5 ms was calculated from the experimental data for bituminous 

coal particles of 75 ym diameter with 32.8% volatile matter as detemined 

by proximate analysis. The complete proximate analysis is given in 


Table 24. The hfeating rate in this flame varied between 1.2 x 10 and 

8.7 X 10 K/s with 2 ms of total time spent beneath the burner and the 
rest in the flame (3.5 ms) . The temperature of the particles environ- 
ment at the beginning of devolatilization was 890 K and at the end was 
about 1960 K. 

This value for devolatilization time can be conpcired to the values 
obtained by other researchers to check the correctness of our cal- 
culations. Kobayashi et al. (151) studied the devolatilization of a 
ignite and a bituminous coal at high temperatures and rapid heating 

conditions in the presence of an inert atmosphere (Argon) in a laminar 

4 5 

flow furnace. Heating rates of 10 - 2 x 10 K/s were used with final 
temperatures between 1000 and 2100 K with time resolution down to a 
few milliseconds. The proximate analysis of the coal types used show 
the bituminous coal had a volatile matter content of 40.7% with the 
lignite containing 36.2%. Volatile yields of both coals was found to 
increase significantly with temperature and these yields were signific- 
antly in excess of the 46% value given by proximate analysis with a 
maximum value of 63% VM obtained on rapid heating to a high temperature 
(2100 K) . Lower VM values were obtained at lower temperatures with 


a value of 60% at 1940 K and 45% at 1740 K. At a temperature of 2100 K, 
after 10 ms 71% of the maximum total volatiles was released, with 100% 
volatile production after 20 ms. Coal particles in this study were 
size graded through a +400-325 Tyler Mesh for an average particle sizes 
between 45-55 ym. 

Ubhayakar et al. (152) studied the rapid devolatilization of 
pulverized bituminous coal particles in the size region around 75 ym. 
In this study the coal particles were injected into a constant cross- 
section gasifier, which was heated by hot combustion products produced 
by an oil burner using #2 fuel oil having a hydrogen to carbon atc«n 
ratio (H/C) of 1.72, burned in stoichiometry with oxygen-enriched air. 
The coal in the study, received from a supplier as 7 0% passage through 
a 200 mesh screen, was divided into two groups, the fine fraction 
which passed through a 200 mesh screen and the coarse fraction which 
remained on the screen. When these coal particles were input to the 
gasifier at a temperature of 1970 K it was found that most of the 
devolatilization of the fine fraction occurred within the first 10 ms 
with the coarse fraction requiring 25 ms for complete devolatilization. 
The heating rate for these particles was 1.3 x 10^ K/s which is 
similar to heating rates found in this study. This time for devolat- 
ilization was faster than the times observed by Kobayshi et al. (151) 
which may be due to their smaller particle size or due to their use of 
an active oxygen containing medium (burnt fuel oil) for heating of the 
particles . 

McLean et al. (292) made direct observations on the early stages 
of combustion of 65 ym bituminous coal particles. These particles were 


introduced into a high temperature transparent, laminar flow reactor 
fed by a pre-mixed gas flame using mixtures of methane, hydrogen and 
air. These were used to produce uniform reactor carrier gases at 
tenperatures, frcm 1100 K to 1800 K with oxygen mole fractions of 0 to 
10^ K/s. In this study high speed photographs of particle emission, 
high magnification shadowgraphs of burning particles and micrographs 
of partially burnt captured material were obtained. 

The ignition of bituminous coal particl^es were characterized by 
a bright diffuse emission, attributed to burning of ejected volatile 
matter, which ceased after, approximately 5 ms which was replaced by 
incandesence attributed to heterogeneous char oxidation. Shadowgraphs 
of burning coal particles indicated that upon ignition, volatile matter 
ejected from the particle formed a condensed phase surrounding the 
particle. This condensed phase was thought to be a soot-like material 
resulting from pyrolytic cracking of hydrocarbons in the volatile matter. 
Under oxidizing conditions (fuel lean) the condensed volatile matter 
was oxidized during the early stages of char burning, while under 
reducing conditions (fuel rich) the matter persisted throughout the 
flow reactor. 

Fu and Blaustein (293) studied the reactions of bituminous coal in 
a microwave discharge in the presence of hydrogen, water vapor and argon. 
In the microwave-generated discharges in H^, H^O and Ar, coal was 
gasified to give gaseous hydrocarbons and carbon oxides plus tar and 
residual char. H2 was also produced either by dissociation of the 
water vapor or by devolatilization of the coal in the H^O and/or Ar 
discharges. In the H2-Ar discharge a net increase of in the gas 
phase is observed only for high volatile bituminous coal. 


The yield of hydrocarbons was highest in the H^-Ar discharge and 

that of carbon oxides was highest in the H^O-Ar discharge. Both the 

H -Ar and the H 0-Ar discharge gave greater extents of gasification 
2 2 

cind produced more hydrocarbon products than the Ar discharge, indicating 
the occurrence of gas phase reactions of H, OH and active O species 
with the active carbon sites in the coal. In a H^O-Ar discharge the 
eSctent of gasification and hydrocarbon production varied with initial 
H^O concentration, increasing until a level was reached whereupon 
greater amounts of H^O decreased the production of hydrocarbons due to 

These four studies indicate fast devolatilization times at high 
temperatures and heating rates, and reactions that are influenced by 
the molecules contained in a flame, especially radical type reactions. 
We have seen in Chapter III, the accepted mechanisms for combustion of 
coal and char particles. What new reactions could be occurring in the 
methane-air-coal dust to explain the observed devolatilization 
time of 5.5 ms which is about a factor of 2 faster than the values 
obtained by Ubhayakar et al. (152) in a somewhat reactive medium. 

In the studies by Ubhayakar et al. and McLean et al. (292) coal 
particles were heated up in the reaction products frcm a gaseous fuel 
similar to the fuel in the present study. The difference in the above 
two studies ccmpared to the current study is that the gases used in the 
early studies were completely reacted and were quite some distance away from 
the primary reaction zone , therefore allowing a substantial amount of 
radical recombination reactions to occur thereby reducing the concen- 
tration of O, H, OH, etc. In the present study the coal is injected 


the same time as the primary heating gases thereby allowing the coal 
particle to be present when the radicals are first formed in the 
primary reaction zone. 

The reduction in devolatilization time seen in this study can be 
attributed to attack of the coal particle by free radicals formed in 
the primary combustion zone. Experimental evidence (294) has shown 
that at 1100 K to 2000 K oxygen atcms oxidize carbon from 5 to 80 times 
faster than oxygen molecules and, although direct evidence is lacking, 
it is likely that OH radicals have similar reactivity (295) . The 

C + 20H -> CO^ + 


C + 20H CO + H^O 
are both highly exothermic. 

The high reactivity of OH towards carbon containing species is also 
indicated by the work of Fenimore and Jones (296) who studied the 
cOTibustion of soot in flame gases. These authors burned a fuel rich 
ethylene-oxygen-argon mixture on a porous burner to give a gas contain- 
ing soot equal to about 1% of the carbon fed as ethylene. Sixteen 
percent of the fuel ranained in the gaseous products as hydrocarbons, 
mostly as acetylene and the rest as carbon oxides, steam, and hydrogen. 
The products were cooled, mixed with H^ and 0^ to obtain various different 
(j)'s and then burnt on a second burner. 

The rate of combustion of soot, measured by solids sampling at 
different heights above the second burner, was found to be little 
affected by changing the partial pressure of oxygen in the flame gases 


from 0.04 to 0.30 atm. A good correlation was found however with 
the calculated equilibrium concentration of OH radicals and a some- 
what better one with the measured OH concentrations (the measured 

concentrations being 4 to 8 times the calculated) . Even in a fuel 


rich flame with the partial pressure of O^ about 10 atm the soot 
disappeared only 2 to 3 times less rapidly than in the oxygen rich 
mixtures. Frcan this evidence, along with the fact that dry oxygen- 
nitrogen mixtures (no OH) consumed the soot at a notably lower rate 
than that expected on the basis of the oxidation by the flame gases, 
Fenimore and Jones postulated that the soot is mainly oxidized by OH 
radicals- A calculated value shows that a carbon aton is removed from 
the soot particle by about every tenth collision of an OH radical with 
the surface. 

From data obtained in this study and the above mentioned results 
of others it is most likely that the radicals produced in a methane- 
air flame greatly assist the combustion of the coal particle in many 
ways. During devolatilization the flame gas radicals may help stabilize 
the free radical species formed in a coal particle due to thermal cracking 
of the linkages forming the aronatic clusters (143) . Attack of the coal 
molecule by radicals such as OH, O and H may also induce the production 
of more volatiles by increasing the susceptibility of carbon bonds , 
not normally reactive, or by reducing the amount of secondary reactions 
occurring which normally produce carbon deposition or polymerize into 
large soot type particles therefore reducing combustion efficiency. 


The reaction of gas phase free redicals with coal may proceed via 
the following mechanisms 

1) C + OH C(OH) 


C(OH) C(0) + C(H) 
C(0) CO + C 


C(H) -> CH + C 


2) C + H ->■ C(H) 


C(H) -> CH + C 


3) C„ + 0 C(0) 


C(0) CO + C 


The attadk by 0 and H atom should proceed faster than the similar reactions 
with molecular species as experimental data has indicated (293) . Hydrogen 
atom could also attack lone methyl groups present as substituents upon 
coal ring structures or produced in pyrolysis by the following mechanism 

4) C-CH^ + H ^ C^ + CH^ 

The work of Fu and Blaustein (293) shows that in the presence of H 
atoms the majority of the products are hydrocarbons which may be due to 
hydrogen attack in a localized area on more than one carbon. A typical 
reaction mechanism would be 

5) 2C + 2H -> (H)C C(H) 


(H)C+C(H) ^ C^H^ + C^ 

Many H atoms depositing at once at structurally weak points on the coal 

molecule could enhance production of higher hydrocarbons such as C^H^ 

and C H which have been observed in devolatilization studies (146, 157, 
3 8 

293) . Other small hydrocarbon species may be formed via the following: 


6) C + H -s- C(H) 


C(H)+C C(CH) 

C(CH) ^ C H + 2C 
2 F 


7) C + H C(H) 


C(H)+H ^ C(H) (H) 

C (H) (H) ^ CH_ + C 
2 F 

In the case of OH radical attack (293) the reaction products 
contain more carbon oxides thain hydrocarbons as reaction scheme 1 
would indicate. OH could also form C^O via 

8) C + OH -> C(OH) 


C(OH) -s- C(0) + C(H) 
C(0)+C(H) -> C(0)-C(H) 

C(0)-C(H) C O + H + 2C 

2 F 

where C(0)*C(H) is a mutually attracted pair which react together. 
Comparison with Computer Kinetic Code Simulation Data 

A computer code to simulate the kinetic processes occurring when 
coal is combusted in a methane-air flame has been developed, independent 
frcsn this study, by Green and Pamidimukkala (297) in the laboratory 
where this study was accomplished. This computer code has been adapted 
from the type of kinetics used in atmospheric studies, suitably 
modified to simulate the combustion chemistry. The kinetic simulation 
of methane-air ccanbustion uses a reaction set of 52 reactions with 
a total of 17 molecular and atomic species derived from Smoot et al. 
(61) with additional reactions from Jachimowski and Wilson (69). 


Reaction time (s) 
Figure 61. Kinetic model for CH /Air flame. 








I -6 
- 10 

























^ A A A A 

• • • 


A ▲ ▲ ▲ A 





■ ooo oooooo 


I Raw coal 


O Ash 


I I 1 1 1 1 1| \ — I I 1 1 1 HI I I I I 1 1 1 II 





10-^ 10-^ 

Reaction time (s) 


Figure 62. Kinetic rtiodel for CH. /Air/Coal flame. 


The model used to describe the combustion of the coal is that of 
Kobayashi et al. (151). In this model, described in Chapter III, the 
coal is devolatilized by 2 routes, a fast reaction which leads to char 
and the volatiles CO, and CH^, and a slow reaction which yields 
the same products but at a much slower rate. The slow reaction will 
dominate at low temperature with the fast one dominating at high 
temperatures. Thus the net rate of disappearance of raw coal is the 
negative of the sum of the rate of char and volatile formation according 
to the 2 reactions. The net concentration of char depends upon the 
fast devolatilization reactions and the slow oxidation reactions of 
char with 0^, CO^ and H2O. The total reaction scheme to model the 
combustion of a methane-air-coal dust flame contains 57 reactions and 21 
species, with the coal related species being raw coal, total volatiles, 
char, and ash. 

Figure 61 shows the results of a kinetic run of a methane-air 
flame using experimental flow rates, while Figure 62 shows the same 
flame with the addition of 30 mg/min of bituminous coal. From Figure 
62 one can see that the char is completely formed (all possible volatiles 
removed) at '\^7 ms which is in quite good agreement with the experimental 
value of 5.5 ms for devolatilization in a stoichiometric flame. 


This study reports the first use of high resolution spectroscopic 
techniques in the ultraviolet and visible regions of the spectrum, to 
obtain absolute number densities of radical species in a methane-air- 
coal dust stationary flame. The only other previous investigation of 


a methane-air coal dust flame (298) involved propagating methane-coal 
dust inhibitor flames. The majority of the spectral work in this 
earlier study was in the infrared region although spectra of OH (0,0) 
were obtained. However, no quantitative measurements of the band were 
made due to the poor resolution of the rapid scanning spectrometer used 
(approximately 25 8) . Blackbody coal dust temperature profiles were 
calculated from spectral radiance values at 435 and 436 nm. 

The present study has demonstrated that a laboratory size burner 
can be used to perform spectroscopic observations on the simultaneous 
combustion of gas and coal. By observing the spectral emissions of the 
species OH, CH, C^, and CO seme understanding of the coal burning 
processes occuring in the presence of gas can be obtained. This study 
has pointed out the complex relationships between flame composition 
and the enhanced combustion of coal. We have seen in this study how 
the devolatilization time is a function of both flame fuel-air ratio 
and temperature with the flame optimum value found in a stoichiometric 
flame . 

The fast devolatilization and char ignition times found in this 
study indicate the feasibility of replacing oil boilers with a mixture 
of gas and coal. These fast times would cause a gas-coal flame 
to require less of a distance than a coal flame, enabling a higher 
power density to be developed in the boiler. 

Evidence has been given which indicates a major role for radical 
species, OH, 0 and H in the attack and combustion of a coal particle 
and various reaction schemes were postulated to account for experimental 


The advantages gained when burning coal in the presence of a 
methane-air flame can be given as follows: 

1. High temperature of the methane-air flame causes a rapid 
heating effect, thereby producing more volatiles from a 
coal particle than would normally be expected from proximate 

2. A large production of volatiles will result in a smaller 
amount of char which will react faster, therefore decreasing 
the length of flame needed for complete combustion. 

3. The high concentration of radicals in the methane-air flame 
gases help promote faster devolatilization and possibly 
help produce a greater amount of volatiles. 

4. Stoichiometric and lean flame comditions promote the ignition 
of the devolatilized char particle resulting in a much shorter 
char heating time ('^3 ms) when compared to a rich flame 

('^'7 ms) . This is probably due to the higher concentrations 
of 0 atoms found in a lean and stoichiometric flame when 
compared to a rich flame. 

Future Work 

The most intriguing aspects of the present study involve the 
presence of the flame radicals OH, O and H and the effect of these 
species. E\'idence has been put forth in this study for a major role 
for these radicals in the devolatilization and combustion of the coal 
particle. One way to study these effects in greater depth is to 
increase the concentrations of these radicals in the flame. By spiking 
the flame with suitable gases one should be able to increase the 
concentrations of a specific radical and observe its action. 

Hydrogen and oxygen radicals could be increased by adding the 
resepctive molecular species and OH and H atcm could be generated by 
the aspiration of water into the flame. To make sure the effects of 
radicals, and not molecular species, were being observed the respective 
gases could be passed through a flowing microwave discharge and then 
to the flame, insuring a large concentration of radicals in the flame. 


One problem with the present study is how to distinguish between 
the effect of temperature and the effect of radical species as the 
flame stoichiometry was changed. One way to separate the temperature 
effect from the chemical one is to run flames using methane, oxygen, 
nitrogen and argon mixtures. By varying the argon to nitrogen ratio 
the flame temperature could be varied over a sufficient range to 
determine the effect of temperature. As long as the methane-oxygen 
ratio was kept constant the population of radicals would be fairly 
constant. Another possible variation would be to vary all four gases 
in order to obtain different (j) values at a constant temperature. 

Another question which could be addressed is the effect of particle 
size on the devolatilization and char ignition times. It would be 
interesting to see how the amount of volatiles produced (as measured 
by CH and emissions) varied with particle size and at what size 
the reaction went from a chemical controlled reaction to a mass transfer 
controlled reaction. The effect of (J) could alsp be studied with 
respect to particle size and devolatilization times. A certain size 
particle may produce one amount of volatiles in a flame with (t)=1.0 
and may produce more or less in flames with different <J)'s. The coupling 
of particle size to flame stoichiometry should be optimized in order 
to maximize the amount of char burned. 

The emissions observed in some studies might be an indicator of 
how ccmplete the combustion of the coal particles. emission could 
be monitored as a function of coal mass flow and particle size. 
Further work is needed to understand the significance of the emission 
when coal is introduced to the flame. 


An important series of measurements would measure the ground 
state concentrations of the species observed in these studies. 
Absorption data would give information on ground state concentrations, 
the predominate reaction species, and the changes observed with the 
addition of coal should further elucidate the chemical reactions 
occuring in coal combustion. Ground state concentrations could also 
be used, with the temperature measurements, to determine what percentage 
of the excited state is thermally produced as opposed to the amount 
chemically produced. 

Another useful study would involve observations of the complete 
char burning region. This would probably require a large burner to 
insure a tall enough flame to provide uniformly hot surroundings for 
the particles. Emission measurements would probably not be of any 
use in the upper regions due to the noise of the coal dust flame. 
Particle densities could possibly be measured by a scattering technique 
using a He-Ne or He-Cd laser to monitor densities at different heights 
and therefore the burnout rate. Another way to accomplish the same 
experiment would be to contain the flame in an oven and sample the 
flame at several different heights in order to determine burn out rate 
under different flame equivalence ratios or with different additives 
(i.e. H^, O^). 








C*************************** ******************************** 




C******************************** ********* ******************** 
REAL* 4 Mh.llU 200) ,IN1 ( 200 ) 
DIME1<SI0N ICK (14) ,X(200),X1(200) 

ACCEPT 2, J , (ICH(I) .1 = 1 ,J) 

2 FORMAT ( Q. 1 4A1 ) 

CALL ASSIGN ( 2 , ICR . J , ' RDO ' ) 


ACCEPT 2. J, (ICH(I) .1 = 1 , J) 

CALL ASSIGIl ( 3, ICR. J, 'MEW' ) 

TYPE 10 






1 2 IKS/MIK) MOTOR 2/1 RATIO XF=0.25,IF 5/1 RATIO ',/. 
1 XF=0.01. FOR NO SCALING XF=0.1 ') 
TYPE 35 


ACCEPT 40. lA 


20 FORMAT (F10.5) 

N = C 

100 READ(2,997 ,END=110) XP.XI 
997 FORMAT (2F1 0.5) 

102 IF(XP.LT.XO) GOTO 101 


X( I )=XP*XF*10 . 0 
N = I5+1 


101 GOTO 100 
110 IF=1 





GOTO 102 
111 R=X(I)-X(1) 

DELTA=R/ 100.0 

DO 900 1=0,99 
J = l 

650 XL=X(J) 


J = J + 1 

GO TO 650 




WRITE (3, 30) DELTA 
3 0 FORKAT(F10 . 5 ) 


XE=X1 (99 ) 

DO 700 1=0,99 

X2=XE*IA +N1*X1(99*IA +Nl*l) 
XI=IN1 (99**IA+K1*I) 
WR1TE(3,25) X2.XI 

25 F0RliAT( IX, 2F10 . 5 ) 

• 7 00 COllTIllUE 















SAVITZKY - GOLAY METHOD ( AK AL . CHEH . 3 6 , 1 6 27 ( 1 9 64 ) 


REAL *4 HDATA( 1000 ) ,MDATA( 1000) .KP(9) .liSUK 
TYPE 110 

ACCEPT 220 , J , (NAME( I ) , 1 = 1 , J ) 
CALL ASSIGN ( 2 , M AHE , J , ' RDO ' ) 
TYPE 120 

ACCEPT 220,J,(KAME(I).I=1.J) 
K = 100 

READ(2,400) DELTA 

WRITE (3, 400) DELTA 

DO 50 J=1.100 

READ(2,40 0)Xl(J),liDATA(J) 



FORMAT ( Q, 14A1 ) 
M = K-4 

DO 10 1=2,5 
J = I-1 


I = 1 , M 



DO 200 
J = I + 4 

DO 11 K=l,4 

HP(5)=KDATA( J) 

1 J S U M = 1 7 . * N P ( 3 ) + 1 2 . * ( K P ( 2 ) + N P ( 4 ) ) - 3 . * ( K p ( 1 ) + N P ( 5 ) ) 


FORMAT ( IX, 2F1 0.5) 


WRITE ( 3 , 40 1 ) XI ( 1 ) , 0 . 0 , XI ( 2 ) , MDATA ( 1 ) / 2 . 0 
DO 9 00 I=1,M 

V;RITE (3,401) Xl(I+2) , MDATA ( I ) 

WRITE (3, 40 1 ) XI (99) , MDATA ( 9 6 ) *0 . 5 , XI ( 1 00 ) . MDATA ( 9 6 ) *0 . 0 







,. .. X *************** 









REAL*4 IN, INI . , v 

DIMENSION FORK(7 2).NAME(14),A(5).T(3).R(3.5).IN(100),FK(100) 

l.Xl(lOO) .Y(3),X(100) ,IN1(100) 

TYPE 110 

ACCEPT 220.J,(NAME(I),I=1,J) 
CALL ASSIGN ( 2 . N AME . J , ' RDO ' ) 

TYPE 120 

ACCEPT 220,J,(NAME(I),I=1,J) 
READ(2,400) DELTA 

DO 50 J=1.100 
READ(2.40 0)X1(J).IN1(J) 

GOTO 600 

C************ CALCULATE FK(K) 

85 DO 60K=0.49 

FK(K) =0 . 0 
DO 60 J=K,49 
XJ = J 
XK = K 


Al = ( ( (XJ+1 )**2)-(XK**2) )**1 . 5 

Al=Al/( 2*XJ+1 ) 

A2=4*XJ*( (XJ**2)-(XK**2) )**1.5 

A2=A2/(4*XJ**2-1 ) 

A3 = ( (XJ-1 )**2-(XK**2) )**1 .5 

A3=A3/( 2*XJ-1 ) 

ALPHA=4*(A1 -A2+A3 )/3 

GOTO 60 

70 ALPHA=(4*(2*K+1. )**.5)/3. 

60 FK(K)=FK(K)+ALPHA*IN( J) 



C************* FIT FK(K) TO A+BK 2+CK 4 ************** 


DO 5 K=0,49 
DO 20 1=1,3 
DO 20 J=l,3 
Y ( I ) = 0. 0 
20 R(I,J)=0.0 



IF (K.LE.2) GOTO 8 

35 DO 30 I=ICN-2,Kti+2 

XI = I 

Il(1.2)=R(1.2) + XI**2 

R( 1 .3 )=R( 1 .3 )+XI**4 

R( 2,3 )=R( 2,3 )+XI**6 



Y( 2)=Y( 2)+FK(l)*XI**2 

Y(3 )=Y(3 ) + FK(l)*XI**4 





D DO 2 11=1,3 

D DO 2 Jl =1.3 

D 2 TYPE 400,R(I1,J1) 

DO 40 1=1,3 


DO 40 J=l,3 


F1=R( 2, 1 ) 


DO 45 J=l,3 

45 R(3.J)=R(3,J)-R(1,J)*F2 
R ( 3 , 3 ) =R ( 3 . 3 ) -R ( 2 , 3 ) *R ( 3 , 2 ) 
Y(3)=Y(3)/R(3.3 ) 

F=-(B+2*C*K**2)/(3. 141 5 9*DELTA) 
D GOTO 901 



GOTO 500 
KN = 2 
GOTO 35 

DO 700 1=0,99 
X1(I)=X1(I) -RA/2. 
DO 800 1=0,49 
FLAG=1 . 0 
GOTO 85 
DO 925 1=0,49 
DO 950 1=0,49 
IIUI-1) = IK1(50 + I) 
FLAG = 0 . 0 
GOTO 85 




FORMAT( 13 ,F5 .3 ) 

FORMAT ( Q, 14A1 ) 


FORMAT ( IX, 2F1 0.5) 

FORHATC •(1X,2F1 0.5)' ) 

DO 540 1=1,50 

INI(50+I) = IN(I-1 ) 


URITE(3,5 50) 

DO 675 1=1,99 

WRITE (3, 400) Xl(l) ,(lNl(l)+INl(99-l))/2. 




DO 902 1=1,50 

TYPE 400 ,IK(I) 


GOTO 903 




1. A. E. S. Green, Ed., "An Alternative to Oil: Burning Coal 
with Gas" (University Presses of Florida, Gainesville, 1982). 

2. G. J. Minkoff, C F. H. Tipper, "Chemistry of Ccmbastion Reactions" 
(Butterworths , London, 1962). 

3. R. M. Fristrcm, A. A. Westenberg, "Flame Structure" (McGraw-Hill, 
New York, 1965) . 

4. A. G. Gaydon, "The Spectroscopy of Flames" (Chapman and Hall, 
London, 1974). 

5. A. G. Gaydon, H. G. Wolf hard, "Flames: Their Structure, Radiation 
and Temperature " (Chapman and Hall, London, 1979). 

6. M. W. Thring, "The Science of Flames and Furnaces" (Chapman and Hall, 
London, 1962) . 

7. B. Lewis, J. Grumer, Indust. Eng. Cham. 40, 1123 (1948). 

8. B. Lewis, G. von Elbe, "Combustion, Flames and Explosions of Gases'' 
(Academic Press, New York, 1961). 

9. B. Lewis, G. von Elbe, J. Chem. Phys. 11_, 75 (1943). 

10. G. M. Hieftje, Appl. Spectrosc. 25_, 653 (1971). 

11. M. S. Uberoi, A. M. Kuethe , H. R. Menkes, Phys. Fluids 1^, 150 (1958). 

12. F. W. Ruegg, W. W. Dorsey, Combust. Flame 10, 1 (1966). 

13. T. L. Cottrell, J. C. McCoubrey, "Molecular Energy Transfer in Gases" 
(Butterworths, London, 1961). 

14. W. R. Kane, H. P. Broida, J. Chem. Phys. 21, 347 (1953). 

15. R. Mavrodineanu , H. Boileux, "Flame Spectroscopy" (Wiley, New York, 1965). 

16. E. Mallard, H. L. Le Chatelier, Ann. Mines 4_, 274 (1883). 

17. W. Nusselt, Z. Ver. Deut. Ing. 59, 872 (1915). 



18. E. Jouguet, L. Crussard, Ccmpt. Rend. 168, 820 (1919). 

19. P. J. Daniell, Proc. Roy. Soc. A126, 393 (1930). 

20. G. Damkohler, Z. Elektrochm. 46_, 601 (1940). 

21. Y. B. Zeldovich, D. A. Frank-Kamenetsky , J. Phys. Chem. U.S.S.R. 
12, 100 (1938). 

22. Y. B. Zeldovich, D. A. Frank-Kamenetsky, Acta Physiochim. U.S.S.R. 
% 341 (1938). 

23. y. B. Zeldovich, N. Semenoff, J. Exp. Theor. Phys. U.S.S.R. 10, 

1116 (1940) . 

24. Y. B. Zeldovich, N. Semenoff, J. Exp. Theor. Phys. U.S.S.R. 10_, 
1427 (1940). 

25. N. Semenoff, Progr. Phys. Sci. U.S.S.R. 24, 433 (1940). 

26. Y. B. Zeldovich, Z. Tech. Phys. U.S.S.R. 17_, 3 (1947). 

27. A. Th. Belayev, Acta Physicochim. U.S.S.R. 8_, 763 (1938). 

28. E. Bartholcm^ Naturwissenschaf ten, 36_, 171 (1949). 

29. E. Bartholome, Naturwissenschaf ten 36.* 206 (1949). 

30. S. F. Boys, J. Corner, Proc. Roy. Soc. A197 , 90 (1949). 

31. R. C. Murray, A. R. Hall, Trans. Faraday Soc. 47_, 743 (1951). 

32. D. B. Spalding, Fuel 33_, 255 (1954). 

33. C. Tanford, R. N. Pease, J. Chem. Phys. 15_, 431 (1947). 

34. C. Tanford, J. Chem. Phys. 15_, 433 (1947). 

35. C. Tanford, R. N. Pease, J. Chem. Phys. 15^, 861 (1947). 

36. A. Van Tiggelen, Bull. Soc. Chim. Belg. 55^, 202 (1947). 

37. A. Van Tiggelen, Bull. Soc. Chim. Belg. 58_, 259 (1949). 

38. N. Sanenov, "Chemical Kinetics and Chain Reactions" (Clarendon Press, 
Oxford, 1935). 

39. M. V. Smoluchowski, Ann. Physik 21_, 756 (1906). 

40. H. F. coward, F. J. Hartwell, J. Chem. Soc. P401, 2676 (1932). 

41. D. B. Spalding, Phil. Trans. A249 , 1 (1956). 

42. A. G. Gaydon, Proc. Roy. Soc. A196 , 105 (1949). 


43. N. Manson, Compt. Rend 226 , 230 (1948). 

44. J. 0. Hirschf elder , C. F. Curtiss, Adv. Chem. Phys. 3_, 59 (1961). 

45. J. 0. Hirschf elder, C. F. Curtiss, R. B. Bird, "Molecular Theory 
of Gases and Liquids" (John Wiley, New York, 1954). 

46. S. Chapman, T. G. Cowling, "The Mathematical Theory of Non-Uniform 
Gases" (Cambridge University Press, London, 1970). 

47. J. O. Hirschfelder, C. F. Curtiss, J. Chem. Phys. 17_, 1076 (1949). 

48.. J. O. Hirschfelder, C. F. Curtiss, R. B. Bird, E. L. Spotz, 

"The Properties of Gases" (John Wiley and Sons, New York, 1953). 

49. G. K. Adams, G. W. Stocks, Fourth Symposium (International) on 
Combustion, The Combustion Institute, Pittsburgh, 1953, p. 239. 

50. D. B. Spalding, P. L. Stephenson, Proc. Roy. Soc. A324 , 315 (1971). 

51. D. B. Spalding, P. L. Stephenson, R. G. Taylor, Combust. Flame 17_, 
55 (1971). 

52. S. V. Patankar, D. B. Spalding, "Heat and Mass Transfer in Boundary- 
Layers" (International Textbook Co., London, 1970). 

53. G. Dixon-Lewis, Proc. Roy. Soc. A298 , 495 (1967). 

54. G. Dixon-Lewis, Proc. Roy. Soc, A307 , 111 (1968). 

55. C. S. Wang Chang, G. E. Uhlenbeck, J. de Boer, "Studies in Statistical 
Mechanics, Vol. II" J. de Boer, G. E. Uhlenbeck Eds. (John Wiley 

and Sons, New York, 1964). 

56. E. A. Mason, L. Monchick, J. Chem. Phys. 36 , 1622 (1962) . 

57. E. A. Mason, J. T. Vanderslice, J. M. Yos, Phys. Fluids 2_, 688 (1959). 

58. L. Monchick, E. A. Mason, J. Chem. Phys. 35_, 1676 (1961). 

59. L. Monchick, A. N. G. Pereira, E. A. Mason, J. Chem. Phys. 42, 
3241 (1965). 

60. L. Monchick, R. J. Munn, E. A. Mason, J. Chem. Phys. 4^, 3051 (1966). 

61. L. D. Smoot, w. C. Hecker, G. A. Williams, Combust. Flame 26_, 323 
(1976) . 

62. G. Tsatsaronis, Combust. Flame 33_, 217 (1978). 

63. R. M. Fristrom, C. Grunfelder, S. Favin, J. Phys. Chem. 64_, 1386 
(1960) . 


64. G. E. Andrews, D. Bradley, Combust. Flame 19, 275 (1972). 

65. J. Peeters, G. Mahnen, The 14tli Symposium (International) on 
Combustion, The Combustion Institute, Pittsburgh, 1973, p. 133. 

66. C. T. Bowman, Fourteenth Symposium (International) on Canbustion, 
The Combustion Institute, Pittsburgh, 1973, P. 729. 

67. D. B. Olson, W. C. Gardiner, Jr., Combust. Flame 32_, 15 (1978). 

68. C. K. Westbrook, J. Creighton, C. Lund, F. L. Dryer, J. Phys. Chem. 
81 , 2542 (1977). 

69. C. J. Jachimowski, C. H. Wilson, NASA Tech. Paper, 1974 (1980). 

70. A. A. Westenberg, R. M. Fristrom, J. Phys. Chem. 65, 591 (1961). 

71. K. J. Laidler, "The Chemical Kinetics of Excited States" (Clarendon, 
Oxford, 1955). 

72. P. L. Lijnse, "Electronic-Excitation Transfer Collisions in Flames" 
(Diss. Utrecht, 1974). 

73. J. C. Polanyi, Appl. Opt. 1£, 1717 (1971). 

74. J. C. Polanyi, Diss. Faraday Soc. 44, 293 (1967). 

75. J. C. Polanyi, J. J. Sloan, J. Wanner, Chem. Phys. 13_, 1 (1976). 

76. A. G. Gaydon, H. G. Wolfhard, Proc. Roy. Soc. 213 , 366 (1952). 

77. H. P. Broida, A. G. Gaydon, Proc. Roy. Soc. 218 , 60 (1953). 

78. S. L. N. G. Krishnamachari, H. P. Broida, J. Chem. Phys. 34_, 1709 
(1961) . 

79. R. P. Porter, A. H. Clark, W. E. Kaskan , W. E. Browne, Eleventh 
Symposium (International) on Combustion, The Combustion Institute, 
Pittsburgh, 1967, p. 907. 

80. F. E. Belles, M. R. Lauver, J. Chem. Phys. 4£, 415 (1964). 

81. Y. Hidaka, S. Takahashi, H. Kawano, M. Suga, W. C. Gardiner, Jr., 
J. Phys. Chem. 86, 1429 (1982). 

82. C. W. Hand, G. B. Kistiakowsky , J. Chem. Phys. 37_, 1239 (1962). 

83. E. M. Bulewicz, P. J. Padley, R. E. Smith, Proc. Roy. Soc. A3I5 , 
129 (1970). 

84. T. Kinbara, K. Noda, Thirteenth Symposium (International) on 
Ccsnbustion, The Combustion Institute, Pittsburgh, 197X, p. 993. 


85. K. H. Becker, D. Kley, Chem. Phys. Lett. 4_, 62 (1969). 

86. K. A. Quickert, J. Phys. Chem. 76, 825 (1972). 

87. R. E. Ferguson, J. Chem. Phys. 23, 2085 (1955). 

88. A. R. Fairbairn, Proc. Roy. Soc. 267 , 88 (1962). 

89. R. Bleekrode, W. C. Nieuwport, J. Chem. Phys. 43^, 3680 (1965). 

90. J. Peeters, J. F. Lambert, P. Hertoghe, A. Van Tiggelen, Thirteenth 
Symposium (International) on Combustion, The Combustion Institute, 
Pittsburgh, 1971, p. 321. 

91. M. I. Savadatti, H. P. Broida, J. Chem. Phys. 45_, 2390 (1966). 

92. D. M. Mann, Chem. Phys. Lett. 47_, 106 (1977). 

93. P. H. Kydd, W. I. Foss, Eleventh Symposium (International) on 
Combustion, The Combustion Institute, Pittsburgh, 1967, p. 1179. 

94. K. H. Becker, K. D. Bayes , J. Chem. Phys. 48_, 653 (1968). 

95. The American Society for Testing and Materials, "1976 Annual Book 
of ASTM Standards" Part 26, D388-77 , Philadelphia, 1976. 

96. E. A. Hattman, W. E. McKinstry, H. Schultz, Anal. Chem. 5^, 135R (1979). 

97. H. Schultz, A. W. Wells, T. W. Bergstresser , Anal. Chem. 53_, 233R 
(1981) . 

98. A. H. Edwards, Fuel, 37_, 415 (1958). 

99. G. Ingram, "Methods of Organic Elemental Microanalysis" (Chapman 
and Hall, London, 1962). 

100. 0. I. Milner, R. J. Zahner, Anal. Chem. 32., 294 (1960). 

101. P. J. Jackson, Fuel 35., 212 (1956). 

102. A. C. Fielding, W. A. Selvig, U. S. Bureau Mines Bull. 492 , 16 (1951). 

103. W. R. Kirner, Ind. Eng. Chem. 8_, 57 (1936). 

104. M. Z. Schutze, Anal. Chem. 118 , 245 (1939). 

105. J. Unterzaucher , Ber. Dtsch. Chem. Ges. 73B , 391 (1940). 

106. L. C. Bate, Nucleonics 21_, 72 (1963). 

107. 0. U. Anders, D. W. Briden, Anal. Chem. 36_, 287 (1964). 

108. M. C. Stopes, Proc. Roy. Soc. 90B, 470 (1919). 


109. M. C. Stopes, Fuel L4, 4 (1935). 

110. H. R. Brown, A. C. Cook, G. H. Ta^^lor, Fuel 43_, HI (1964). 

111. L. G. Benedict, R. R. Thcxnpson, J. J. Shigo, R. P. Aikman, Fuel 47_, 
125 (1968) . 

112. A. Davis, W. Spackman, P. H. Given, Energy Sources 3_, 55 (1976). 

113. P. H. Given, M. E. Peover, W. F. Wyss, Fuel 4£, 425 (1965). 

114. D. G. Murchison, J. M. Jones, Fuel 42_, 141 (1963). 

115. R. Millais, D. G. Murchison, Fuel 47_, 247 (1969). 

116. M. F. Kessler, Fuel 52_, 191 (1973). 

117. D. W. Van Kreuelen, "Coal" (Elsevier, New York, 1961). 

118. B. N. Nandi, T. D. Brown, G. K. Lee, Fuel 56, 125 (1977). 

119. H. H. Lowry, "Chemistry of Coal Utilication" Suppl. Vol. (John Wiley 
and Sons, New York, 1963). 

120. M. Menser, H. O'Donnell, S. Ergun, Proc. 5th Carbon Conf. Vol. 2, 
Pergamon Press, London, 1962, p. 493. 

121. H. Brusset, J. Devaux, A. Guinier, Compt. Rend. 216 , 152 (1943). 

122. S. Ergun, I. Wender, J. Appl. Chem. 1£, 189 (1960). 

123. L. Cartz, P. B. Hirsch, Philos. Trans. R. Soc. London A252 , 557 (1960). 

124. R. E. Franklin, Acta Crystallog. 3_, 107 (1950). 

125. J. B. Nelson, Fuel 32_, 153 (1954). 

126. P. B. Hirsch, Proc. Roy. Soc. London A226 , 143 (1954). 

127. M. M. Roy, Fuel 36^, 249 (1957). 

128. C. G. Cannon, Nature 171 , 308 (1953). 

129. R. A. Friedal, J. A. Queiser, Anal. Chem. 28_, 22 (1956). 

130. J. K. Brown, P. B. Hirsch, Nature 175, 229 (1955). 

131. H. S. Rao, P. L. Gupta, F. Kaiser, A. Lahiri, Fuel 41_, 417 (1962). 

132. L. Czuchajowski, G. J. Lawson, Fuel 42 , 131 (1963). 

133. S. Fujii, Y. Osawa, H. Sugiiaura, Fuel 49_, 68 (1970). 


134. L. Czuchajowski, Arch. Corn. 6_, 257 (1961). 

135. S. Fujii, H. Tsuboi, Fuel 45_, 369 (1966). 

136. R. M. Elofson, Can. J. Chem. 35_, 926 (1957). 

137. J. K. Brovm, J. Chem. Soc. (London) 744 (1955). 

138. A. G. Sharkey, J. T. McCartney in "Chem. of Coal Utilization" 2nd 
Suppl. Vol., M. A. Elliott, Ed. (John Wiley and Sons, New York, 1981). 

139. I. Wender, L. A. Heredy, M. B. Neuworth, I. G. C. Dryden in "Chem. 
of Coal Utilization" 2nd Suppl. Vol., M. A. Elliott, Ed. (John Wiley 
and Sons, New York, 1981). 

140. L. A. Heredy, A. E. Kostyo, M. B. Neuworth, Fuel 44, 125 (1965). 

141. H. L. Retcofsky, Appl. Spectrosc. 3^, 116 (1977). 

142. P. H. Given, Fuel 39./ (1960). 

143. W. H. Wiser, G. R. Hill, N. J. Kertamus, Ind. Eng. Chem. Process. 
Des. Dev. 6_, 133 (1967). 

144. W. Peters, H. Bertling, Fuel 44_, 317 (1965). 

145. J. F. Jones, M. R. Schmid, R. T. Eddinger, Chem. Eng. Prog. 60, 
69 (1964). 

146. D. B. Anthony, J. B. Howard, H. C. Hottel, H. P. Meissner, Fifteenth 
Symposium (International) on Ccxnbustion, The Ccmbustion Institute, 
Pittsburgh, 1975, p. 1303. 

147. S. Badzoich, P. G. W. Hawksley, Ind. Eng. Chem. Process. Des. Dev. 
9, 521 (1970) . 

148. G. M. Kimber, M. D. Gray, Combust. Flame 11_, 360 (1967). 

149. R. T. Eddinger, L. D. Friedman, E. Rau, Fuel 45_, 245 (1966). 

150. H. N. Stone, J. D. Batchelor, H. F. Johnstone, Ind. Eng. Chem. 46_, 
274 (1954). 

151. H. Kobayashi, J. B. Howard, A. F. Sarofim, Sixteenth Symposium 
(International) on Combustion, The Combustion Institute, Pittsburgh, 
1977, p. 411. 

152. S. K. Ubhayaker, D. B. Stickler, C. W. von Rosenberg, Jr., R. E. 
Gannon, Sixteenth Symposivmi (International) on Combustion, The 
Combustion Institute, Pittsburgh, 1977, p. 427. 

153. M. G. Skyler, V. I. Shustikov, I. V. Virozub, Intern. Chem. Eng. 9_, 
595 (1969). 


154. G. J. Pitt, Fuel 41, 267 (1962). 

155. R. Rennhack, Brennstof f-Chemie 45_, 300 (1964). 

156. V. Vand, Proc. Phys. Soc. (London) A55 , 222 (1943). 

157. E. M. Suuberg, W. A. Peters, J. B. Howard, Ind. Eng. CHiem. , Process. 
Des. Dev. 17^, 37 (1978). 

158. B. K. Maziimdar, N. N. Chatter jee. Fuel 52_, 11 (1973). 

159. P. C. Lewellen, Ph.D. Thesis, MIT, 1975. 

160. D. B. Anthony, J. B. Howard, H. C. Hottel, H. P. Meissner, Fuel 55_, 
121 (1976). 

161. J. B. Howard, R. H. Essenhigh, Ind. Eng. Chem. Process. Des. Div. 6_, 
74 (1967) . 

162. D. Bhaduri, S. Bandyopadhyay , Canbust. Flame 17_, 15 (1971). 

163. L. D. Smoot, M. D. Horton, G. A. Williams, Sixteenth Symposium 
(International) on Combustion, The Conbustion Institute, Pittsburgh, 
1978, p. 375. 

164. J. B. Howard, R. H. Essenhigh, Eleventh Symposium (International) 
on Combustion, The Combustion Institute, Pittsburgh, 1967, p. 399. 

165. W. Nusselt, Z. Ver. Deut. Ing. 68_, 124 (1924). 

166. S. P. Burke, T. E. W. Schuman, Pore. 3rd Int. Conf. Bituminous Coal 2_, 
484 (1931). 

167. R. H. Essenhigh, Sixteenth Symposium (International) on Combustion, 
The Combustion Institute, Pittsburgh, 1977, p. 353. 

168. H. Gan, S. P. Nandi, P. L. Walker, Jr., Fuel 51, 272 (1972). 

169. J. M. Thomas, W. J. Thomas, "Introduction to the Principles of Hetero- 
geneous Catalysis" (Academic Press, New York, 1967). 

170. D. O. Hayward, B. M. W. Trapnell, "Chemisorption" (Butterworths 
Scientific Publications, London, 1964). 

171. F. Bonner, j. Turkewich, J. Am. Chem. Soc. 73 , 561 (1951). 

172. S. Ergun, J. Phys. Chem. 60_, 480 (1956). 

173. M. A. Kanter, Phys. Rev. 107, 655 (1957). 

174. R. Phillips, F. J. Vastola, P. L. Walker, Jr., Proc. 3rd Industrial 
Carbon Graphite Conf. 257 (1971). 


175. F. J. Vastola, P. J. Hart, P. L. Walker, Jr., Carbon (Oxford) 2_, 
65 (1964) . 

176. P. J. Hart, F. J. Vastola, P. L. Walker, Jr., Carbon (Oxford) 5_, 
363 (1967). 

177. P. L. Walker, Jr., F. J. Vastola, P. J. Hart, "Fundamentals of 
Gas-Surface Interactions" (Academic Press, New York, 1967). 

178. L. Bonnetain, X. Duval, M. Letort, Proc. 4th Conf. Carbon (1960) 
p. 107. 

179. F. J. Long, K. W. Sykes, Proc. Roy. Soc. (London) A193 , 377 (1948). 

180. P. L. Walker, Jr., F. Rusinko, L. G. Autin, Adv. Catal. 
133 (1959). 

181. I. Langmuir, J. Am. Chem. Soc. 40, 1361 (1918). 

182. G. S. Rushbrooke, Trans. Faraday Soc. 36_, 1055 (1940). 

183. J. E. Lennard- Jones , A. F. Devonshire, Proc. Roy. Soc. A156 , 6 (1936). 

184. S. Glasstone, K. J. Laidler, H. Eyring, "The Theory of Rate Processes" 
(McGraw-Hill, New York, 1941). 

185. C. Kemball, Adv. Catalysis 2_, 233 (1950). 

186. J. Gadsby, F. J. Long, P. Sleightholm, K. W. Sykes, Proc. Roy. Soc. 
A193 , 357 (1948). 

187. A. E. Reif, J. Phys. Chem. 56_, 785 (1952). 

188. S. Ergun, U. S. Bureau of Mines Bulletin 598 (1961). 

189. M. Mentser, S. Ergun, U. S. Bureau of Mines Bulletin 664 (1973). 

190. J. F. Strange, P. L. Walker, Jr., Carbon 1£, 345 (1976). 

191. S. Ergun, M. Mentser, in the "Chemistry and Physics of Carbon", 

P. L. Walker, Jr. Ed. Vol. 1 p. 203 (Marcel Dekker, Inc., New York, 1965). 

192. J. Gadsby, C. N. Hinshelwood, K. W. Sykes, Proc. Roy. Soc. A187 , 
129 (1946). 

193. J. P. Blakely, L. G. Overholser, Carbon 3_, 269 (1955). 

194. R. F. Strickland-Constable, J. Chim. Phys. 47_, 356 (1950). 

195. J. F. Johnstone, C. Y. Chen, D. S. Scott, Ind. Eng. Chem. 44, 1564 
(1952) . 

196. J. D. Blackwood, F. K. McTaggart, Aust. J. Chem. 12, 533 (1959). 


197. G. O. Ingles, Trans. Faraday Soc. 48_, 706 (1952). 

198. J. D. Blackwood, Aust. J. Chem. 12_, 14 (1959). 

199. J. D. Blackwood, Aust. J. Chem. 15_, 397 (1962). 

200. C. W. Zielke, E. Gorin, Ind. Eng. Chem. 47_, 820 (1955). 

201. F. Moseley, D. Paterson, J. Inst. Fuel 38_, 13 (1965). 

202. J. L. Johnson in Coal Gasification, Advances in Chemistry Series 

No. 131, p. 145, (American Chemical Society, Washington, DC, 1974). 

203. J. D. Blackwood, B. D. Cullis, D. J. McCarthy, Aust. J. Chem. 20, 
1561 (1967). 

204. C. W. Zielke, E. Gorin, Ind. Eng. Chem. £9, 396 (1957). 

205. M. F. R. Mulcahy, I. W. Smith, Rev. Pure Appl. Chem. 19, 81 (1969). 

206. C. G. Von Fredersdorf f , M. A. Elliott in "Chemistry of Coal Utiliza- 
tion", H. H. Lowry, Ed., Supplementary Volume, p. 754 (John Wiley 
and Sons, New York, 1963). 

207. A. B. Ayling, I. W. Smith, Combust. Flame 18_, 173 (1972). 

208. J. ARthur, Trans. Faraday Soc. 47_, 164 (1956). 

209. R. Phillips, F. J. Vastola, P. L. Walker, Jr., Carbon 8_, 205 (1970). 

210. N. R. Laine, F. J. Vastola, P. L. Walker, Jr., J. Phys. Chem. 67_, 
2030 (1963). 

211. M. Rossberg, Z. Elecktrochem. 6£, 952 (1956). 

212. G. Blyholder, H. Eyring, J. Phys. Chem. 63_, 1004 (1959). 

213. F. Rodrigues-Reinoso, P. A. Thrower, P. L. Walker, Jr., Carbon 12_, 
63 (1974). 

214. I. W. Smith, R. J. Tyler, Comb. Sci. Tech. 9_, 87 (1974). 

215. E. A. Gulbransen, K. F. Andrew, Ind. Eng. Chem. 4£, 1034 (1952). 

216. M. Faraday, C. Lydell, Phil Mag. 26, 16 (1845). 

217. W. Galloway, Proc. Roy. Soc. 24, 34 (1876). 

218. W. Galloway, Proc. Roy. Soc. 28_, 410 (1878). 

219. F. Abel, Proc. Roy. Inst. 10, 88 (1882). 



220. Bedson and McConnel, Trans. Inst. Min. Egnrs. 1_, 27 (1894). 

221. Bedson and Widdas, Trans. Inst. Engrs. }2_, 529 (1906). 

222. H. M. Cassel, A. K. Das Gupta, S. Gurreswarny, Third Symposium 
(International) on Combustion, The Ccmbustion Institute, 
Baltimore, 1949, p. 185. 

223. H. M. Cassel, I. Liehman, W. K. Mock, Sixth Symposium (International) 
on Combustion, The Combustion Institute, New York, 1957, p. 602. 

224. R. H. Essenhigh, J. Csaba, Ninth Symposium (International) on Combustion, 
The Combustion Institute, Pittsburgh, 1963, p. 111. 

225. W. F. Marshall, H. B. Palmer, D. H. Seery, J. Inst. Fuel 37_, 342 (1964). 

226. D. Bhaduri, S. Bandyopadhyay , Combust. Flame 17_, 15 (1971). 

227. L. D. Smoot, M. D. Horton, Prog. Energy Combust. Sci. 3_, 235 (1977). 

228. L. D. Smoot, D. T. Pratt, Eds. "Pulverized-Coal Combustion and 
Gasification" (Pleniam Pres, New York 1979). 

229. M. D. Horton, F. P. Goodson, L. D. Smoot, Combust. Flame 28_, 187 (1977). 

230. L. D. Smoot, M. D. Horton, G. A. Williams, Sixteenth Symposium 
(International) on Combustion, The Combustion Institute, Pittsburgh 
1976, p. 375. 

231. C. T. Crowe, L. D. Smoot, Ch. 2 in "Pulverized-Coal Combustion and 
Gasification", L. D. Smoot, D. T. Pratt, Eds. (Plenum Press, New York, 
1979) . 

232. R. Edelman, O. Fortune, G. Weilerstein, Symposium on Emissions from 
Continuous Combustion Systems, (Plenum Press, New York, 1972). 

233. C. Veillon, J. Y. Park, Anal. Chem. 42_, 684 (1970). 

234. J. J. Horvath, J. D. Bradshaw, J. N. Bower, M. S. Epstein, J. D. 
Winefordner, Aneil. Chem. 53 , 6 (1981). 

235. C. Th. J. Alkemade, R. Herrmann, "Fundamentals of Analytical Flame 
Spectroscopy" (John Wiley, New York, 1965). 

236. G. Cadle, R. D. Magill, Ind. Eng. Chem. 43^, 1331 (1951). 

237. W. A. Roach, J. Soc. Chem. Ind. London 65_, 33 (1946). 

238. V. A. Fassel, G. W. Dickinson, Anal. Chem. 40, 247 (1968). 

239. N. Qmenetto, J. D. Winefordner, Prog. Analyt. Atom. Spectrosc. 2_, 
1 (1979). 


240. C. Th. J. Alkemade, W. Snellman, G. D. Boutilier, B. D. Pollard, 
J. D. Winefordner, T. L. Chester, N. Omenetto, Spectrochim. Acta. 
33B , 3831 (1978). 

241. C. L. Wyatt, "Radiometric Calibration: Theory and Methods" (Academic 
Press, New York, 1978). 

242. W. Snelleman, Ph.D. Thesis, Utrecht, 1965. 

243. J. J. Ball, Rev. Sci. Instrum. 44, 1141 (1973). 

244. D. B. Vaidya, J. J. Horvath, A. E. S. Green, Appl. Opt. 21_, 3357 (1982). 

245. G. H. Dieke, H. M. Crosswhite, J. Quant. Spectrosc. Radiat. Transfer, 
2_, 97 (1962). 

246. H. J. Kostowski "Self-Study Manual on Optical Radiation Measurements, 
Chap. 7, The Relative Responsivity and Slit-Scattering Function on a 
Spectroradicmeter" (Nat. Bur. Stand. Tech. Note 910-4 1979). 

247. J. F. James, R. S. Steinberg, "The Design of Optical Spectrcmetors" 
(Chapman and Hall, London, 1969). 

248. L. M. Garrison, D. D. Doda, A. E. S. Green, Appl. Opt. 18_, 850 (1979). 

249. D. B. Yaidya, J. Samuels, A. E. S. Green, J. Opt. Soc. Am. 7J^, 1584 
(1981) . 

250. K. M. Pamidimukkala, Private Communication. 

251. R. W. B. Pearse, A. G. Gaydon, "The Identification of Molecular Spectra" 
John Wiley and Sons, New York, 1976). 

252. G. Herzberg, "Molecular Spectra and Molecular Structure I. Spectra 
of Diatomic Molecules" (Van Nostrand Reinhold Co., New York, 1950). 

253. W. L. Dimpel, J. L. Kinsey, J. Quant. Spectrosc. Radiat. Transfer 21^, 
233 (1979). 

254. A. M. Bass, H. P. Broida, Nat. Bur. Stand. Circ. 541 (1953). 

255. E. E. Whiting, A. Schadee, J. B. Tatum, J. T. Hougen, R. W. Nicholls, 
J. Mol. Spec. 80, 249 (1980). 

256. J. B. Tatum, Astrophys. J. Suppl. 14 , 21 (1967). 

257. H. Honl, F. London, Zr. F. Phys. 33_, 803 (1925). 

258. R. S. Mulliken, Rev. Mod. Phys. 3_f 8^ (1931). 

259. I. Kovacs "Rotational Structure in the Spectra of Diatcmic Molecules" 
(American Elsevier Pub. Co. Inc., New York, 1969). 


260. A. Schadee, B. A. N. 17_, 341 (1964). 

261. J. Hinze, G. C. Lie, B. Liu, Astrophys. J. 196 , 621 (1975). 

262. G. C. Lie, J. Hinze, B. Liu, J. Chem. Phys. 59^, 1872 (1973). 

263. C. E. Moore, H. P. Broida, J. Res. NBS. 63A , 19 (1959). 

264. A. M. Bass, H. P. Broida, NBS Monog. 24 (1961). 

265. D. M. Cooper, J. Q. S. R. T. 26^, 113 (1981). 

266. J. O. Arnold, S. R. Langhoff, J. Q. S. R. T. 19, 461 (1978). 

267. J. G. Phillips, S. P. Davis, "The Swan System of the Molecule" 
(Univ. of Calif. Press, Berkeley, CA 1968). 

268. M. J. Mumma, E. J. Stone, E. C. Zipf, J. Chem. Phys. 54, 2627 (1971). 

269. P. H. Krupenie, "The Band Spectrum of Carbon Monoxide", NSRDS-NBS5, 
(U. S. Govt. Printing Office, Washington, DC 1966). 

270. G. A. Hornbeck, R. C. Herman, Ind. Eng. Chem. 43_, 2739 (1951). 

271. L. B. Headrich, G. W. Fox, Phys. Rev. 35^, 1033 (1930). 

272. K. M. Pamidimukkala, A. E. S. Green, Unpublished work. 

273. R. M. Fristrom, A. A. Westenberg, "Flame Structure", (McGraw-Hill, 
New York, 1965) . 

274. R. H. Tourin, "Spectroscopic Gas Temperature Measurement" (Elsevier, 
Amsterdam, 1966). 

275. A. Savitzky, M. J. E. Golay, Anal. Chem. 3£, 1627 (1964). 

276. W. L. Barr, J. Opt. Soc. Am. 52, 885 (1962). 

277. W. Magnus, F. Oberhettinger , "Special Functions of Mathematical 
Physics" (Chelsea Publishing Co., New York, 1949). 

278. R. N. Bracewell, Aust. J. Phys. 9^, 198 (1956). 

279. M. P. Freeman, S. Katz, J. Opt. Soc. Am. 53_, 1172 (1963). 

280. J. B. Tatum, Pub. Dom. Obs. 13^, 1 (1966). 

281. R. J. Cattolica, Combust. Flame 4£, 43 (1982). 

282. W. M. Heffington, G. E. Parks, K. G. P. Sulzmann, S. S. Penner, 
Sixteenth Symposium (International) on Combustion, The Combustion 
Institute, Pittsburgh, 1977;. p. 729. 


283. J. Warnatz, Eighteenth Symposium (International) on Combustion, 
The Combustion Institute, Pittsburgh, 1981, p. 369. 

284. D. S. Coe, B. S. Haynes, J. I. Steinfeld, Combust. Flame 43_, 
211 (1981). 

285. A. D'Alessio, A. Di Lorenzo, A. F. Sarofim, F. Beretta, S. Masi, 
C. Venitozzi, Fifteenth Symposium (International) on Combustion, 
The Combustion Institute, Pittsburgh, 1975, p. 1427. 

286. G. E. Andrews, D. Bradley, Combust Flame 18_, 133 (1972). 

287. D. M. Cooper, R. W. Nicholls, Spectros. Lett. J, 139 (1976). 

288. J. Carozza, R. Anderson, J. Opt. Soc. Am. 67_, 118 (1977). 

289. D. R. Stull, H. Prophet, JANAF Thermochemical Tables, 2nd ed. 
NSRDS-NBS-37 , U.S. Government Printing Office, Washington, DC 1970. 

290. K. P. Huber, G. Herzberg, "Molecular Spectra and Molecular Structure 
IV. Constants of Diatc3mic Molecules" (Van Nostrand Reinhold Co., 
New York, 1980) . 

291. H. J. Fischbeck, K. H. Fischbeck, "Formulas, Facts and Constants" 
(Springer-Verlag, Berlin, 1982),. 

292. W. J. Mclean, D. R. Hardesty, J. H. Pohl, Eighteenth Symposixim 
(International) on Combustion, The Combustion Institute, Pittsburgh, 
1981, p. 1239. 

293. Y. C. Fu, B. D. Blaustein, Fuel 47_, 463 (1968). 

294. D. E. Rosner, H. D. Allendorf, Carbon 3_, 153 (1965). 

295. J. A. Graham, A. R. G. Brown, A. R. Hall, W. Watt, Conf. on Industrial 
Carbon and Graphite, 1957, Society of Chemical Industry, London 1958. 

296. C. P. Fenimore, G. W. Jones, J. Phys. Chem. 71, 593 (1967). 

297. A. E. S. Green, K. M. Pamidimukkala, "Proceedings of International 
Conference on the Combined Combustion of Coal and Gas", Cleveland, 
Ohio, 1982. 

298. Z. J. Fink, L. E. Dalverny, J. Grumer, U. S. Bureau of Mines, Report 
of Investigations, 1956. 

John Joseph Horvath was born on March 24, 1954, at Bridgeport, 
Connecticut. He received his elementary and junior high school 
education in the public school system of Fairfield, Connecticut. Upon 
graduation from the Fairfield College Preparatory School in May, 197 2, 
he received a scholarship from Windham College in Putney, Vermont, 
where he attended until May, 1973. In September, 1973, he transferred 
to the University of South Carolina in Columbia, South Carolina, and 
obtained his Bachelor of Science in chemistry (ACS certified) in May, 
1976. In September, 1976, entered the University of Florida, receiving 
the Master of Science in analytical chemistry in 1980. Since that time 
he has pursued the Ph.D. in chemistry under the guidance of Drs. A. E. S. 
Green and W. B. Person. Presently he has six publications on studies 
of flames. He has been awarded a National Research Council Fellowship 
at the National Bureau of Standards. 

John J. Horvath is a member of the American Chemical Society, 
Optical Society of America, and the Astronomical Society of the Pacific. 


I certify that I have read this study and that xn my opxnion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 

A. E. S. Green, Chairman 
Graduate Research Professor of Physics 
and Nuclear Engineering Sciences 

I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 

Willis B. Person, Co-chairman 
Professor of Chemistry 

I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 

Martin Vala 

Professor of Chemistry 

I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 

>s.>fetant Professor of Chemistry 

I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 

Richard A. Yos-t^ 

Assistant ProTessor of Chemistry 

This dissertation was submitted to the Graduate Faculty of the 
Department of Chemistry in the College of Liberal Arts and Sciences 
and to the Graduate School, and was accepted as partial fulfillment 
of the requirements for the degree of Doctor of Philosophy. 

April, 1983 

Dean for Graduate Studies and