Sie sind auf Seite 1von 164

UNIVERSITY OF CALIFORNIA

Los Angeles

A FIBFIS Sampling Study of

Formation of Nitric-Oxide in a Premixed

Fuel-Rich Methane-Air Flame

A thesis submitted in partial satisfaction

of the requirement for the degree

Raster of Science in Engineering

by

Ivan A. Gargurevich

1980
The thesis of Ivan A. Gargurevich is approved.

U, D. Van Vorst

Ouen I, Smith

Eldon L. Knuth, Committee Chair

University of California, Los Angeles

1980
To Horn and Dad,

Lucia,

Cynthia,

Irene,

Sergio,

and Carmen,
TABLE OF CONTENTS

PAGE

LIST«OF FIGURES . . . . Vi

LIST OF TABLES ... viii


ix
LIST OF SYMBOLS . .
xii
ACKNOWLEDGMENTS

ABSTRACT OF THE THESIS xiii


CHAPTER 1. INTRODUCTION . ,.",'," i
CHAPTER 2. PREVIOUS STUDIES OF NO FORMATION IN

HYDROCARBON FLAMES . . .. 10
2.1 The Formation of NO in Fuel-Lean Flames . j_g
2.2 Prompt NO Formation ........... 15

2.3 Formation of NO in Fuel-Rich Flames ... 22


2.4 Mechanisms of Methane Oxidation ..... 30
2.4.1 Methane Oxidation in Fuel-Lean

Flames .............. 30
2.4.2 Methane Oxidation in Fuel-Rich
Flames .............. 34
CHAPTER 3. EXPERIMENTAL APPARATUS 44
3.1 An Introduction to Sampling Techniques
in Combustion Studies .......... 44

3.2 Possible Composition Distortions in MBMS


Sampling of Flames ........... 53
3.2.1 Free-3et Expansion ........ 54

1
PAGE

3.2.2 Chemical Relaxation in Supersonic

Expansion ............. 59

3.2.3 Species Condensation ....... 63


3.2.4 Pressure Diffusion ........ 65

3.2.5 Relaxation Phenomena ....... 69

3.2.6 Skimmer Interference . . . . . 0 . 71

3.2.7 Mach-Number-Focusing „ . 72
3.2.8 Background Scattering ....... 76

3.2.9 Flame Perturbations by Sampling

Cones 77
3.3 The U.C.L.A. MBMS Sampling System .... 83

CHAPTER 4. EXPERIMENTAL PROCEDURE AND DATA RE-

DUCTION ......... 87
4.1 Composition Measurements ......... 87

4.2 Temperature Measurements ......... 91

4.3 Calibration Procedures .......... 93

CHAPTER 5. EXPERIMENTAL RESULTS ... 96

5.1 Temperature Measurements in the Flame . . 97


5.2 Stable Species Profiles . . 110
5.3 Formation of Nitric-Oxide in the Flame . . 112

5.4 C2H2, C2H6 Profiles . .. . 124


5.5 CH. and H2CO Species in the Flame .... 130

CHAPTER 6. CONCLUSIONS .... 136

REFERENCES . . . . . . 140
LIST OF FIGURES

FIGURE PAGE

1. node of coal formation . 3

2. The dependence of CH20, C2H4, C2H2, and H2

peak concentrations on flame equivalence

ratio ................... 40
3. The dependence of peak C2H,, C2H2, and H2

on peak JCH-, } for a range of flames $ =


0.56 to 1.25 \ 41

4. The U.C.L.A. MB[*]S sampling system ..... 50

5. Temperature ratio as function of dimension-

less flou time for free jet gas flows from

orifice 52

6. Free-jet shock structure .......... 55

7. Schematic diagram of model considered in

Mach-number-focusing 74

8. Intensity vs. electron energy for Argon . . 89


9. Temperature profile measured using TDF tech-

nique ..... ......... 93


10. Temperature profile measured using TOF tech-

niques 100

11. Normalized Oxygen profile (tip dia. = 2.0


104
mm, d^ = 0.24 mm)

12. Normalized Oxygen profile (tip dia. = 0.5 mm,

d^. = 0.30 mm) ........ . . ..... 105


FIGURE PAGE
.
13. C02 normalized profile (tip dia. = 2.0 mm,

d# = 0.24 ram) ................ IDS


14. Normalized C02 profile (tip dia. = 0.5 mm,
d^. = 0.30 mm) . . . . . .107
15. 02, CI-L, and C02 mole-fraction profiles for
CH^-Air flame ................ ill
16. NO concentration profile .......... 114
17. NI-U concentration profile .......... 115
18. HCN concentration profile .......... lie
19. Normalized m/e = 30 profile 117
20. Normalized m/e = 17 profile ......... 119

21. Normalized m/e = 43 profile ......... 123


22. C2H2 concentration profile ......... 126
23. Normalized m/e = 27 profile . .... . . . . 127

24. Normalized m/e = 30 profile 129


25. C-atom concentration profile ........ 131
26. CM concentration profile .......... 132
27. CH2 concentration profile 133
28. Normalized m/e = 15 profile 134
LIST OF TABLES

TABLE PAGE

1. Methane oxidation in fuel-lean mixtures ... 35


2. Methane oxidation in fuel-rich mixtures ... 36

\M 1
-

LIST OF SYMBOLS

a speed of sound
C. calibration constant appearing in Eq. (56)
£•
d effective source-orifice diameter
d.£ cone orifice diameter
d, skimmer entrance diameter
D binary diffusion coefficient
E C3 electron energy in ev
#
E activation energy
f mole fraction of heavier species
h total enthalpy
I. ion signal intensity of species i
ID electron emission in mamp
k. foruard reaction rate coefficient for the ith
reaction
k__ • reverse reaction rate coefficient for the ith
reaction
K equilibrium constant
Kn Knudsen's number
'"Id skimmer entrance - detector distance
m mean molecular ueight
m. molecular ueight of species i
m/e mass to charge ratio
n number density
n, number density at detector
n, number density at skimraer entrance
p pressure
j mass density
Re Reynolds number
S speed ratio
i
Sj_ speed ratio perpendicular to jet axis
Sc Schmidt number
T temperature
Tj^ static temperature perpendicular to jet axis
T,, static temperature parallel to jet axis
t time
u most probable velocity of molecules
v diffusion velocity
X nach disk location

Y r-lach disk diameter


YDD diameter of the barrel shock

thermal diffusivity
u~~t enrichment factor as defined in Eq. (49)
n calibration factor appearing in Eq. (56)
0 (Fuel/Air) actual/(Fuel/Air) stoich., equiva-
lence ratio
0 heat capacity ratio, C /c
'Vu in,p relaxation time at constant h,p
CT relaxation time at constant T.p
I »P
L AB mean time between collisions of A and B mole-
cules
\ mean free path
cr hard sphere collision cross section

Subscripts

eq denotes equilibrium values


0 denotes stagnation conditions
1 denotes conditions at skimmer entrance
frozen process
ACKNOULEDGPIENTS

I uould like to express ray gratitude to Dr.


Eldon Knuth for his continued guidance and encourage-
ment during the time this uork was performed.
Also, I uant to thank all members of the Mole-
cular Beam Lab. for their cooperation, and especially

to Dr. 0. Smith, Fir. Uayne E. Rodgers, and Ph.D. stu-

dent S. Yoon for sharing their knowledge and useful

information.
This study uas supported by NSF under Grant

No. ENG-7709190.

XII
ABSTRACT OF THE THESIS

A MBWS Sampling Study of

Formation of Nitric-Oxide in a Premixed

Fuel-Rich Methane-Air Flame

by

Ivan A. Gargurevich

Plaster of Science in Engineering


University of California, Los Angeles, 1980

Professor Eldon L. Knuth, Chair

The formation of NO in a fuel-rich, premixed

methane-air flame at one atmosphere uas studied using

molecular-beam-mass-spectrometric (MBMS) sampling tech-

niques.

A flat flame of equivalence ratio 1.37 uas

sampled uith particular attention given to NO, HCN, HNCO,


and MB., species. The onset of formation of HCN uas

found to occur prior to the formation of NO, approxi-

mately at the same time as the formation of NH,, and very


near the primary reaction zone of the flame. Decay of

HCN leads to NH- formation, and, to a lesser extent, to


iJ

NO formation. HNCO is observed at very lou concentrations

in the post-flame gases, and acts as an intermediate in

HCN decay. The peak mole fractions of HCN and NH^


about 67 and 44 ppm., respectively,and Nitric-
Oxide reaches a mole fraction of 56 ppm. after all of the

HCN, and NH^ have decayed. The Zeldovich mechanism,

taking into account radical overshoot, accounts, at

most, for 10 ppm. of the NO formed in the flame; the

rest is the result of HCN decay.

The same flame uas sampled for C, CH } CH,-,,


and CI-U« The peak mole fractions of C, CH, and CH-

were found to be about 206, 481, and 908 ppm., respec-

tively, uith an uncertainty of a factor of approximately


2. Because of their greater concentrations, CH and CH~

are the most likely to lead to HCN formation. The con-

tribution to the CH., profile from CH, could not be to-

tally eliminated, even at electron energies as lou as

13.3ev. No effort uas made to estimate its mole frac-

tion in the flame. The CH-, profile peaks uithin the pri-

mary reaction zone.

The flame uas also sampled for H-CO. Its

formation is observed to occur before the formation

of CHr, and upstream from the reaction zone of the flame.


J

The rate of formation CH-, does not reach significant


levels until most of the formaldehyde has decayed.
A relatively large concentration of acetylene

(2100 ppm.) uas found; its formation and decay occurred

mostly uithin the primary reaction zone. Ethane, in

much smaller concentrations, uas also observed in the

same region. The tuo-carbon species provide additional

xiv
paths for the decay of CH- and must be included in the
kinetic modeling of fuel-rich hydrocarbon flames,
Visible attachment of the flame to the quartz

cone uas observed for the flame of equivalence ratio

1.37. Temperature measurements, made using time-of-

flight (TOF) techniques, shoued that the effect of


attachment is to decrease the measured temperature gra-

dient within the reaction zone. A cone of smaller

tip diameter minimizes this effect. Relative inten-


sity profiles shoued also smaller measured gradients

uithin the primary reaction zone for a cone of relative-

ly large tip diameter. Attachment of the flame to the


cone persisted approximately until the end of the pri-

mary reaction zone.

xv
CHAPTER 1

INTRODUCTION

In the last decade, in order to satisfy an


ever-increasing demand for energy, and as a result of
continuing decline in the domestic production of crude
oil, the United States has been forced to import crude
oil and refined products in increasing amounts. It is
83
forecast that, in 1980, thirty percent of our total

energy uill come from the Middle East Nations. Fur-


thermore, since 1972, the fraction of the energy used
for producing the U.S. GiMP uhich has been based on
imported oil has doubled, increasing from a level of
about 13£o of the U.S. GNP in 1972 to about 26% of
the U.S. GNP in 1978. Since the end of 1978 the OPEC
cartel has raised the price of its exported crude by
61j£, and it is estimated that by 1985 it uill reach
nA
340 per barrel .
In 1985 the U.S. could be import-
83
ing as much as 15 million barrels per day and, with
an estimated price of $40 per barrel, the cost for
1985 would be $220 billion. By comparison, the value

of oil imports into the U.S.A. in 1972 uas $4.5 billion,


The finite limits of U.S.A. oil and natural
84
gas reserves are fairly well established ; both could
be exhausted in ten years if they uere worked to their

fullest capacity. However, the estimated resources

of coal amount to approximately 90,000 x 10 Btu.

(or about 430 billion tons) which at the present con-

sumption rates would last for about 666 years.

In the near future, the U.S.A. will have to

find or exploit alternative sources of energy if it


is to play an important role in the world community.

In Duly 1979, President Carter announced his Energy

Plan. The main objectives of the Plan were to dimin-

ish oil imports, to accelerate the conservation of oil

fuel, and to spend $88 billion over ten years to pro-

mote synthetic-fuel production from coal and shale

rock. According to Carter's Energy Plan for about the

next ten years, oil imports would be kept at the 1977

level of 8.2 million barrels per day; 2,5 million bbl./

day would be gained by synthetic fuel production and 2

million bbl./day through conservation and conversion

measures. Carter's Plan would require coal production

to double by 1985 (in 1979 coal production was 625


million tons per year) °r, in the seven-year period

1979-1986, 107 million tons of new capacity would have

to be added every year.


ORIGINAL PLANT MATERIAL CCA' ?/AC c RA.'
(VISUAL MiCHOSCCPY)

FESINGUS MATERIAL
CUT1CL£S,S?OH£S
RESETS

WOOC, CGnX
j

CiRSCNIZATION

V,iCF.CN!TE

RANK F EAT— L i G N 1 TH U S —• 3 i ~J M ! N G U S —"A r-JTH= AC ~


KIGH M=D.
C5A
%C 60 ra 80 S3
%0 25 15 3
Ci!-CR:riC I2CCO 1200 o I4OCO ISOOa S55GC
VALUE

Fig, 1. Mods of coal formation . ( Source : Ref. 85 .)


It is clear then that combustion of coal and

products derived from coal and shale rock will be play-

ing an important role in the U.S. energy picture for years

to come. (Nuclear energy supplies about 3% of the

total energy produced in the U.S. today. Its devel-

opment, however, has encountered much opposition and it

is unlikely that construction of new nuclear plants

uill be started in the 1980's). Coal originates pri-

marily from plants. Fig. 1 shows the steps in coal


formation. Through a series of evolutionary changes,

the primary products of the original decomposed plant

matter becomes transformed sequentially into peat,


lignite, subbituminous coal, bituminous coal, and fi-

nally to anthracite. Uith these transformations the

carbon content increases and the oxygen content decreases.


n j-
Studies ' indicate that there is some correlation be-

tween coal rank and its aromaticity, 40-50^5 aromatic

carbon content for subbituminous coal and 90% aromatic


carbon content for anthracite. These studies showed

no correlation between H/C mole ratio and percent aro-


matic carbon. Oxygen occurs predominantly as pherolic

or etheric groups in coal. Sulfur found in coal has

a similar functional chemistry to Oxygen. Sulfur con-

tent in coal varies from 2 to 3 percent. Nitrogen

occurs predominantly as pyridine or pyrrolic type rings,

and coal contains typically from 1 to 2 percent nitrogen


by ueight. Coal structure is consistent with highly
I-5
substituted aromatics uhich are not highly condensed.
I
Coal undergoes primary decomposition in the tempera-
ture range 400 - 450 C.

The realization of limited and non-reneuable

energy resources and the increasing role of combustion

in the years ahead uill require extensive research


in the areas of energy conservation and combustion

problems. As a result of President Carter's Energy

Plan, there uill be a shift to direct coal burning

by the Utilities Companies and Industry. The health,

safety and environmental threats posed by mining and

burning of coal are severe. Increased coal burning

uill result in larger emissions of sulfur dioxide,

carbon dioxide, and nitric oxide into the atmosphere.

The accumulation of carbon dioxide might result in


the so called "Greenhouse" effect predicted by many

scientists. The large amounts of ash that uill result

from coal burning uill have to be disposed of in an


environmentally safe manner. The production of syn-
thetic fuel uill also cause environmental problems.

Furthermore, the production of synthetic fuel might


require huge amounts of uater for hydrogen production.

The strain on existing uater supplies might be consider-

able and one can foresee legal battles for uater rights

betueen the coal-producing states.


In 1974, transportation accounted for 54/5
n £!
of the total liquid fuel consumption ' in the U.S.
It is most likely that transportation uill depend pri-
marily on liquid fuels in the near future. Most of
the liquid fuel being consumed today in America is
lou in aromatics content and has a high H/C content
(the average H/C ratio being 1.9). Houever, unless
extensive hydrocracking and hydrogenation are conducted,
the properties of synthetic fuels are different from
those of the fuels being used today in transportation.
Synthetic fuels are high in aromatics content and have
a H/C ratio of about 1.2-1.5. Transformation of syn-
thetic fuels from coal into more "gasoline-like" fuel
Rfi
results in an energy loss of about 2Qfa of the heat
of combustion of the syncrude being processed. The
elimination of restrictions on H/C ratio and boiling
range in fuel manufacture uould result in a consider-
able reduction in the cost of the fuel. As a result
of its properties, synthetic fuel combustion uould re-
sult in the emission of large amounts of soot by the
internal combustion engine . Thus, extensive research
is needed to eliminate the need for high levels of hy-
drogenation and boiling range conversion in the manu-
facture of fuels from coal. Furthermore, more research
is needed in the chemistry of soot formation and burn-
out.
It is evident that, in the future, ue uill

have to burn prime fuels more efficiently with less pollu-

tion and ue shall also have to burn much material that

is not nou considered suitable fuel at all. As Uein-


oY
berg points out, progression in combustion systems

is in the direction of more controllability. The pro-

gression has been from diffusion flames (the reaction

zone is aluays near stoichiometric, corresponding to


the maximum possible flame temperature and hence the

largest NO emission) to premixed flames, uhich allows

one to vary flame temperature, burning velocity, ig-

nition temperature, and stability criteria by controlling

the fuel/air ratio to the uell-stirred reactor. Uein-

berg proposes that the next step should be breaking auay

from this dependence on initial fuel/air ratio and con-

trol the reaction rates in the flame independently.

To take this desirable step on combustion systems, ue

must control the reaction rate independently by making


the temperature in the reaction zone a separate para-

meter, using variable heat recirculation, injection


of free radicals, radiation, etc.
This study is concerned with the kinetics of

formation of nitric oxide in an atmospheric, fuel-rich,

premixed, methane-air flat flame, and it is part of a


series of studies that are being conducted in U.C.L.A.

with regards to nitric oxide formation from fuel-ni-


trogen in flames. To give the reader an idea of the

magnitude of this NO formation, in the study of methane-

air flat flames by R. Gay at U.C.L.A., a concentration

of 140 ppm. uas reported in the post-flame gases of a

mixture of equivalence ratio 1.1. In this same flame,

the primary reaction zone uas characterized by the

rapid formation and decay of formaldehyde uith a max-

imum concentration of 540 ppm., and NO formation uas

restricted to the post-flame region. The primary reac-

tion zone uas located uithin the first 5mm. from the

burner surface. In the study of NO formation in a fuel-


rich flame, the coupling betueen the NO-formation reac-

tions and the combustion reactions must be considered.

This requires a knouledge of the combustion mechanism,


and for methane combustion, although extensively studied,

there are questions regarding the rate-determining

step, the f^CO-decay mechanism, the fate of CH~ species,

and the role of t^ hydrocarbons. Nitric oxide above

that predicted by the Zeldovich mechanism is said to


form via the formation and decay of HCN. The forma-

tion mechanism for this species has been studied but

very little is knoun about its decay mode. Experiments

indicate that NHj_ species and OCN-species might play


an important role in the decay of HCN.

It is the aim of this study to ansuer or

clarify some of the questions regarding the kinetics


of NO formation in fuel-rich, premixed methane-air flames.

It is hoped that this uill be of some value in the study

of NO formation in flames containing fuel-nitrogen.

The study of flames uill be conducted using

a molecular-beam-mass-spectrometer (MBNS) sampling


system. This system has been shoun to meet all the

requirements for sampling of high temperature systems

uith excellent balance. The use of MBFIS sampling sys-


tems could undoubtedly play an important role in com-

bustion research in the near future. MBPIS sampling

techniques allou one to follou stable and radical spe-

cies uith minimum distortion. This is valuable help

in the study of (a) oxidation mechanisms and (b) rate

coefficients of elementary reactions. As mentioned

above, MBNS sampling techniques are being used at U.C.L.A.

to study the fate of nitrogen-bearing species and,

more recently, sulfur-containing species in flames.

These studies uill no doubt help determine more effi-

cient and more clean uays to burn fuels such as coal

or liquid fuels derived from it.


CHAPTER 2

PREVIOUS STUDIES OF NO FORMATION

IN HYDROCARBON FLAMES

2.1 The Formation of NO in Fuel-Lean Flames

It is generally accepted that the formation


of nitric oxide in the post-flame gases of lean or

near-stoichiometric hydrocarbon-air flames is modeled

by reactions originally proposed by Zeldovich .

0 + N2 # NO •»• N (1)

N + 02 £ NO + 0 (2)
From experimental data, Zeldovich and co-uorkers con-

cluded that the rate of MO formation in fuel-lean or

near-stoichiometric hydrocarbon flames uas much slouer

than the combustion rate, and that the NO-formation

rate could be calculated assuming equilibration of


2
the combustion reactions. Lavoie, Heyuood and Keck

have indicated that the reaction

N -f OH £ NO + H (3)
also can contribute to NO production, especially in

near-stoichiometric and fuel-rich mixtures, and it is

nou generally included in kinetic studies of NO forma-

tion.
3
Uestenberg studied the kinetics of NO forma-

10
tion in lean, premixed hydrocarbon flames. Using re-

actions (1), (2), and (3), he derived equation (4)


for the rate of NO formation

= 2k1(D)(N2) 1"(ND)2/K(02)(N2) (4)


dt
k2(02)+k3(OH)

where K is the equilibrium constant for the overall


reaction Nj + Oj v1 2ND. In the derivation of equation
(4), it is assumed that (i) the reactions occur under
constant-volume conditions, (ii) d(N)/dt is close to
zero; and (iii) Reaction (5) is equilibrated in the
post-flame gases.

0 + OH ^ H + 02 (5)
The fact that condition (iii) is quickly attained in

the secondary zone of flames has been amply demon-


n Tn
strated ' . Calculation of the NO formation rate
by Equation (4) requires values of the local tempera-

ture and concentrations of 02, N 2 , 0, and OH. Houever,


as first suggested by Zeldovich, Uestenberg took final
equilibrium values of the temperature, and of the N 2 ,
0~, OH, and 0 concentrations.
Using the equilibrium approximation of IM2,
0?, OH, and 0 concentrations, Bowman obtained the

following NO formation rate (using data from Ref. 4)

11
16
d(NO) 6x10 T ~"2expf^69,090/T
r "l x ( 0 0J)
dt eq k eq/ 2 'eq eq

moles (6)
cm sec

Elevated temperatures and high 02 concentrations in


the post-combustion gases result in relatively high

NO-formation rates. Experimental studies of NO for-


mation in the post-combustion zone of laboratory flames
support this simplified mechanism.
Neuhall and Shaded studied the formation
of nitric oxide in the immediate vicinity of a flame
front propagating through a high-pressure combustion
vessel in order to investigate the formation of nitric

oxide under conditions similar to those occurring in


internal combustion engines. Spectroscopic techniques

uere used to determine NO concentrations. Experimental


data uere obtained for hydrogen-air mixtures at pres-
sures of 240, 325, and 340 psia, and 0.7, 0.9 and 1.0
equivalence ratios respectively. Temperatures ranged

from 2000°K to 2500°K. They found that under these


conditions NO is formed primarily in the post-flame

gases, and that the rate of NO formation can be pre-


dicted by the Zeldovich mechanism, uith oxygen- and
nitrogen- atom concentrations set equal to their chem-

ical-equilibrium values.
*>
Livesey et al s t u d i e d the f o r m a t i o n of NO

12
in atmospheric-pressure, premixed, oxy-propane flames.
They used many small diameter stainless steel tubes as
flameholders. An outer shield of argon uas used to

H§r -
prevent outside nitrogen from reaching the combusted
gases. Gas samples uere taken through a uater-cooled
probe and analyzed for N0Xv by the Saltzman method.
Temperatures in the burned gas region uere determined
spectroscopically from the temperatures of rotationally
excited OH. Temperatures uere in the vicinity of 2800°K.
They shoued that reactions (1) and (2) are sufficient
to explain the rate of formation of NO in the post-
flame gases.
The effects of combustion -modifications on
the formation of nitric oxide in CH^/air flames uere
studied by England . A 250,000 Btu/hr ring-type burner
uas used. Nitric oxide in the flue gas uas measured
uith a long-path infrared analyzer at concentrations
from 2 to 500 ppm. Gas samples uere uithdraun by a
uater-cooled quartz probe. He found that nitric oxide
emissions could be reduced by 9Q?0 from peak levels by
burning very fuel-rich flames, and that by reducing
the air preheat from 650°F to 50°F, peak nitric-oxide
levels dropped 67%. The louer NO levels at louer air
temperatures uere due to louer flame temperatures.
England also obtained data for NO concentrations uhen
up to 32/o of the combustion gases uere recirculated.
Nitric oxide in a flame burning uith a combustion air

temperature of 650°F uas reduced by B2% from the peak

levels by 32^ recirculation. Nitric-oxide production

uas lowered uith product-gas recirculation because

of a reduction in the methane-air flame temperature.

These results could have been predicted from Equation

(6); louer flame temperature, Teq


Dn , results in a smaller

rate of formation of nitric oxide. Reduction in the

oxygen level has a similar effect on the rate of NO

formation.

14
2.2 "Prompt" NO Formation

That the formation of NO in the primary re-

action zone might involve a mechanism other than the


g
Zeldovich mechanism uas first indicated by Fenimore

in the Thirteenth Symposium (international) on Combus-


tion. He measured the formation of NO in premixed

atmospheric flames of ethylene, methane, and propane

uith N2 " ^2 mixtures. Post-flame gases were uithdraun


by a quartz microprobe and collected in an evacuated

flask. NO and N02 concentrations uere determined by


chemical analysis. Temperatures uere measured by the

Sodium-line reversal method 20 and uere constant over

the region investigated. Fenimore found that the kin-

etics of NO formation in the post-flame region is de-

scribed by the Zeldovich mechanism, Houever, extra-

polation of the measured NO formation curves back to


zero time gave positive intercepts, indicating, as Feni-

more suggested, a transient formation of NO in the

primary reaction zone, the growth rate of uhich uas


too fast to be explained by the Zeldovich mechanism.

He labeled this intercept NO as "prompt" NO. Since

positive intercepts uere not found in hydrogen or CO

flames, Fenimore suggested reactions such as

CH + N2 ^ HCN + N (7)

C2 + N 2 ^ 2CN (8)

15
to explain the "prompt" NO.
Controversy aroused regarding "prompt" NO,
and numerous investigators have made experimental stud-
ies to prove or disprove "prompt" NO formation.
Bowman 12
" investigated the formation of NO
in shock-induced combustion of H2/02/N2
Concentration time-histories of NO, OH, and H?0 uere
measured during reaction using spectroscopic techniques,,
NO formation rates uere found to exceed the rates pre-
dicted by the post-flame zone mechanism. It uas shoun
that the observed rates uere consistent uith reactions
(1) through (3), if 0 and OH concentrations uere cor-
rectly determined. In the region of rapid NO formation,
observed radical concentrations uere in excess of cal-
culated equilibrium values. Hence, the rapid NO forma-
tion rates uere found to be the consequence of non-equil-
ibrium combustion chemistry.
Further evidence for non-equilibrium chemistry
in the reaction zone of flame uas gathered by Bouman
and Seery 21 . They studied shock-induced combustion
of CH/L - 0? - N 2 mixtures dilutad in argon. Concen-
tration histories of NO, OH, and C02 uere measured
using spectroscopic techniques. Initial temperature
uas in the range 2600 - 3200^, and the initial pres-
sure uas 3.5+ 0.5 atm. They found NO-formation rates
to be consistent uith reactions (1), (2), and (3).

16
Furthermore, discrepancies between observed NO forma-

tion rates and those using an "equilibrium" model for

NO formation uere explainable in terms of "radical

overshoots" in the reaction zone. They did not find

evidence to suggest that other reactions uere of im-

portance in the NO-formation mechanism.

Sarafim and Pohl studied NO formation in


one atmosphere premixed methane-air flat flames of

equivalence ratios from 0.89 to 1.15, Gases uere burned


on a water-cooled porous-disk, flat-flame burner.

Gas samples uere withdrawn through a water-cooled stain-

less-steel probe. NO concentrations were determined


by a modified Saltzman method; other species were mea-

sured using a gas chromatograph. They found that the

peak rates of NO formation in the flame zone were one

to two orders of magnitude greater than the values

in the post-flame zone, and to be consistent with the

modified Zeldovich mechanism (Reactions (1) - (3)),

when free-radical concentrations (of 0, OH, and H)

were correctly estimated.


The "Radical Overshoot Theory" requires a
detailed consideration of the kinetics of the fuel

combustion, or the use of other suitable approximate

methods to compute the non-equilibrium concentrations

of 0, OH, and H near the primary reaction zone in order

to determine NO formation rates. Detailed combustion

17
mechanisms are known for only a feu fuels — H?, CO,

and CFL (methane oxidation will be discussed in a la-

ter section), and a simplified approach involving the


"partial equilibrium" of Reactions (9) - (11) is used
in most cases.

H + 02 £ OH + 0 (9)
0 + H2 ^ OH + H (10)
H2 + OH £ H20 + H (11)
In the "partial equilibrium" approach, the radical

reactions (9) - (11) are considered to be equilibrated


in the flame. This is not to say that the radical

concentrations are equal to their equilibrium values,


rather the ratio of the products to reactants concen-
trations for each of the reactions (9) - (11) is equal
to the reaction equilibrium constant. The concentra-

tions of 0 and OH then can be related to the concen-


trations of stable species uhich are readily measured.
11
Sarofim and Pohl obtained good agreement betueen
observed and calculated "partial equilibrium" NO forma-
tion rates in lean and slightly rich hydrocarbon flames,
Pesters and Flahnen 27 conducted sampling of lean and
.
stoichiometric CH,/02 flames and observed "partial
equilibrium" of Reactions (9) - (11).
Recently, Gay et al 13
' studied one-atmosphere

premixed methane-air flat flames using a molecular-


beam-mass-spectrometer (MBPIS) sampling system. A ua-

18
ter-cooled, copper porous-plug burner urns used. The

gas temperature immediately dounstream from the primary

reaction zone uas measured using the molecular-beam-


time-of-flight technique. For equivalence ratios of

0.8, 1.0, 1.1, tuo distinct regions uere found: a

primary reaction zone uhere formation and consumption

of H~CO takes place, and a secondary reaction zone

uhere NO is formed. Their studies shou that contrary

to the "prompt NO" involved by other investigators,

the formation of NO (if any) in the primary reaction

zone is negligible for flames of equivalence ratio

<I.1. They concluded that the modified Zeldovich mechan-

ism explains NO formation rates in the secondary re-


action zone if radical concentrations and temperature

during combustion are properly evaluated.

Iverach, Basdin, and Kirov studied the

validity of the Zeldovich mechanism in fuel-lean and

fuel-rich hydrocarbon flames in order to end the con-


troversy surrounding "prompt" NO. Propane, ethylene,

acetylene, 6Qfa Ho/40/o ChL, and mixtures of hydrogen


and carbon monoxide uere burned uith air on a uater-

cooled, Neker-type burner. Samples uere taken through

a uater-cooled aluminum tube. NO, CO, C^ concentra-

tions uere determined uith infrared analyzers, 0^ uith


a paramagnetic analyzer, and hydrocarbons uith a flame-

ionization detector. Temperatures uere determined by

19
the sodium-line reversal method and theoretical pro-

cedures. Their experiments showed the following:

(1) for nonhydrocarbon flames, the extended Zeldovich

mechanism is applicable for all fuel-air equivalence

rates, (2) for hydrocarbon flames, the mechanism is

applicable, for both in-flame and post-flame regions,

for #£l,15, and for post-flame region for 0<1.5, pro-

vided that the B^-atom concentration is determined cor-

rectly, and (3) for all regions of fuel-rich hydrocar-

bon flames ($>1.5), and for regions in the vicinity

of the primary reaction zone of moderately rich flames

($>1.15), the mechanism is not tenable because it re-

quires improbably high &-atoms concentrations. They

concluded that reactions such as (7), (8), and possi-

bly (12)

C + N£9 £ CN + N (12)
might be responsible for the large rate of NO forma-

tion in fuel-rich flames and the primary reaction zone

of moderately rich flames (j8>1.15).


28
Ay and Sichel carried out a numerical analy-

sis of the combustion of methane in nitrogen-oxygen

mixtures in order to clarify the relationship between


prompt NO and overshoots in radical concentrations.

Diffusion and heat conduction, which are important

in laminar flames, were neglected in their analysis;

nevertheless, their results show the link between rad-

20
ical overshoot, flame temperature, and the NO forma-

tion rate. Their analysis agrees with the experiments

of Iverach et al , namely, "prompt" NO in near-stoichi-

ometric and lean flames, which is related to the high

rates of NO formation in the recombination and post-

flame zones, can be explained by the Zeldovich mechan-

ism provided the appropriate free radical concentrations

are used; for rich mixtures the NO concentrations com-

puted using the detailed kinetics were far belou experi-

mentally observed values, and the Zeldovich mechanism

fails to explain NO production in rich mixtures.


2.3 Formation of MO in Fuel-Rich Flames

As indicated first by Fenimore and later


supported by the experiments of Iverach et al , "prompt"

NO formed in fuel-rich hydrocarbon flames cannot be


explained by super-equilibrium concentrations of the
radicals 0, OH, and H since the concentrations required

to explain the observed NO-formation rates uould be

significantly larger than "partial equilibrium" values.


The forementioned investigators proposed that reactions

such as (7), (8), and (12) played an important role


in the NO formation process

CH + N HCN + N ( 7)
+ N 2CN ( 8)
C + N 2 £ CN 4- N (12)
uith subsequent reactions of N-atoms-via Reaction (3)

to form nitric oxide. The production of NO in these


flames then requires a detailed combustion mechanism
for the fuel in consideration. (This is also true

of some leaner flames, see Section 2.2). Methane com-


bustion mechanisms are considered in the next section,
and as the reader uill find out, there still are un-

certainties regarding the rate determining step and


the rate coefficients of reactions involved in the
mechanisms. The rapid NO formation rate near the re-

action zone of rich flames requires that reactions

22
such as (7) - (12) be very fast. The experimental

data of Bachmaier et al 29 support this proposal.

Hayhurst and McLean studied a fuel-rich


flame of unburnt composition H2/02/N2: 3/1/4.75 and
a flame of the same composition except that 1% of the
hydrogen uas replaced by acetylene. The flame uas

burnt at atmospheric pressure on a uater-cooled burner


made from stainless-steel hypodermic tubes. The burnt
gas temperature reached 2000°K. Gas samples uere taken

through a uater-cooled probe (made of quartz) into a


chemiluminescent NO analyzer. They found that the
amount of NO in the acetylene seeded flame uas over

that formed in the Hz


0/00/N0 flame via the modified
z /.
Zeldovich mechanism. Based on thermochemical data
alone, they proposed reactions

CH + N2 £ HCN + N ( 7)
CH2 + N 2 ^ HCN + NH (13)
folloued by

N + OH ^ NO + H (3)
NH + OH £ N + H20 (14)
to explain the "prompt" NO formation. Furthermore,
they suggested that HCN is coupled to CN by equilibra-

tion of
HCN + H £ CN + H2 (15)
uith NO resulting from one of the tuo exothermic reac-

tions

23
CN + 02 £ CO -f NO (16)

CN + OH ^ CO + NH (17)
Reaction (17) is followed by (14) and then by reaction

(3).
22
Haynes et al detected the presence of cyano
species in fuel-rich hydrocarbon flames. Uarious pre-

mixed hydrocarbon-air and nonhydrocarbon-air flames

(including a CH, - H2 air flame with equivalence ratio

of 1.42) uith and without the addition of small amounts


of pyridine were burnt on a water-cooled Fleker-type

burner. Samples were withdrawn from the post-flame


gases via water-cooled probes (silica or Alumina), ana-

lyzed continuously for CO and C02 (NDIR analyzers),


-.- -

02 (paramagnetic), and NO (chemiluminescent). The con-

centrations of cyano species and NH. species in the


post-flame gases were determined chemically, and H-

atoms concentrations by the Li/LiOH absorption tech-


f~y *~f O /i

nique ' . Temperatures uere determined uith Na-line

reversal measurements. They found NO, HCN, and NH.

concentrations in excess of equilibrium values. (The


measured values of NH. concentration are viewed with

caution, however, since they were highly susceptible

to sampling conditions). The observed NO formation

rates were larger than predicted by the Zeldovich mechan-

ism with 0-atom concentrations inferred from measured

H-atoms concentrations using a partial-equilibrium ap-

24
proximation for the system H - OH - 0 - H2 - H20. The
formation of cyano-species uas found to be related
to the decay of hydrocarbons and in very rich flames
occurs well into the post-flame gases. Furthermore,
it uas concluded that HCN decays via CN and the rate-
determining reaction
CN + C02 £ OCN + CO (18)
Morley also studied fuel-rich flames, in-
cluding CH, - 02 ~ MO flames with equivalence ratios
from 1.20 to 1.95. A method involving direct sampling
of natural flame ions into a mass spectrometer uas
used to calculate the concentrations of HCN, NH 3 , and
NO. He found that the HCN concentration is proportional
to the N2 concentration independent of the fuel type,
that the HCN concentration increases uith increasing
equivalence ratio, and that HCN-decay leads to both
INUand NO, the relative proportions depending on the
equivalence ratio. Morley suggests the reaction se-
quence,
CH + N2 <* HCN + N (7)
HCN + H ** CN + H2 (15)
HCN + OH £ CN -i- H20 (19)
uhere (15) and (19) are balanced, to explain the forma-
tion of HCN and CN in the flame. The subsequent de-
cay of HCN is consistent uith the reaction
CN + OH £ NCO + H (20)

25
being the rate-limiting step, and it leaves the nitro-

gen from HCI\1 as N H - . Nitric oxide and nitrogen are

produced by reactions involving the NH. species and


Reaction (3)

N + OH ,-• NO + H ( 3)
17
("ligauchi et al modeled the combustion of

CH^ - Qj " No f-'-ames using a 40-reaction mechanism.


They studied flat flames burning on a cooled stainless-
steel porous plate at 0.1 atm. pressure. They used a
quartz probe to sample the flame. NO uas detected
X
using chemiluminescence; H, OH, and 0 using ESR; CN-

species using chemical means, and stable species using

gas chromatography. For equivalence ratios from 0.88

to 1.42, they concluded that the Zeldovich mechanism

is inadequate to predict "prompt" NO formation, and

that HCN is formed prior to NO. They favored Reaction

(13) over Reaction (?) on thermochemical grounds and

pointed out the violation of the spin-correlation rule

by Reaction (7). Other species like C, C2 did not con-

tribute significantly to "prompt" NO formation because


of their low concentrations in the flame. They sug-

gested that

H2CO + M CO + H2 H- M (21)

is the rate determining step in methane oxidation.


95
Pecters, Blauuens, and Smets measured

the concentrations of NO, 0, and fuel fragment radi-

26
icals such as CH, CH2, C2, by means of molecular-beam
sampling and mass spectrometric detection in a number
of hydrocarbon/02/N2 flames. The fuel uas ethylene,
ethane, or methane. They found that even in hot, fuel-
lean flames, the Zeldovich mechanism cannot account
for the rapid NO formation in the flame front. They
derived an expression for the NO generation rate in the
flame front of all flames investigated in terms of the
CH2 or CH radical concentrations. They concluded that
both of the processes
CH + N 2 * HCN + N (7)
CH2 + N 2 £ HCM + NH (13)
in each case followed by rapid oxidation of the prod-
ucts, may be the principal source of NO in the flame
front, and that other hydrocarbon fragments, C, C2,
C 2 H do net contribute to "prompt" NO formation because
of their lou concentrations in the flame front.
The actual mechanism of decay of hydrogen
cyanide in hydrocarbon-air mixtures is uncertain.
The studies by Haynes and Morley previously mentioned
are in contradiction in regards to uhich reaction is
the rate-determining step in the decay. In order to
elucidate the HCN-decay mechanism uhich concludes uith
2 fi
the formation of NO, Haynes recently conducted a
study of premixed hydrocarbon flames burning on a ua-
ter-cooled neker-type burner seeded uith fuel-nitro-

27

'
gen in the form of pyridine, ammonia, or nitric oxide

in amounts up to 2000 ppm. of nitrogen species in the

post-flame gases, A water-cooled silica probe uas

used to sample the gases. NO uas determined continu-


ously by a nondispersive infrared analyser (l\IDIR).
Cyano species and total amine species uere determined
using chemical means. H-atom concentration in the
hot gases uas determined by the Li/LiOH absorption method,
Post-flame temperatures were determined by the Sodium-
line reversal method. Based on the experimental re-
sults, Haynes proposed two parallel rate-controlling
steps: one which is first order in OH, possibly reac-
tion (22)
HCN + OH *• HOCN + H (22)
the other which is second order in OH:

HCN + OH * CN + H20 (23)


CN + OH *• OCN + H (24)
The relative contributions of (22) and (24) vary with
temperature, uith Reaction (24) dominating at high
temperatures. The OCN- species decay very rapidly
to form NH.-species
1
+H
HOCN (cyanic)^HIMCO (isocyanic)^ NH2 -f CO (25)
OCN + H % NH + CO (26)

As Haynes points out, measurements of OCN and NH.


species are required to confirm the proposed mech-

anism.

23
("lorley considers the fate of the NH. spe-
cies. He indicates that the nitrogen hydrides can
be interconuerted by reactions of the type,
NHXV + H ± NHX , + H^-9, x=l,2,3
V —X (27)
which may or may not be balanced, and the oxidation
and the formation of N^ taking place by
N + OH £ NO + H ( 3)

'2 + NO ^ N2 + H20
NHo (28)
NH + NO * N 0 + OH (29)
N + NO ^ N 2 + 0 (3U)
Furthermore, at relatively lou temperatures, the above
reactions might lead to a build-up of amine species.
Experimental measurements are needed.

29
2.4 Mechanisms of Methane Oxidation

In order to predict the formation of nitric oxide

in fuel-rich flames and some lean flames of hydrocarbons

(see Sections 2.2, 2.3), the concentration histories of the

free radicals are necessary. The kinetics of methane oxygen

systems have been studied extensively over the years.

Nevertheless, many uncertainties remain, both with respect

to the oxidation mechanism and the rates of the elementary

reactions which form part of the mechanism. Furthermore,

in the richer methane flames, C^-hydrocarbon decay and

formation has to be accounted for, and the mechanism grous

in complexity.

2.4.1 Methane Oxidation in Fuel-Lean Flames

Studies include shock-tube experiments by Cooke


3D 31 32
and Williams , Bouman , Heffington et al , Brabbs and

Brokau , molecular-beam-mass-spectrometer sampling of


27
flames by Peeters and Mat-men , numerical analysis and flou
34
reactor experiments of Uestbrook et al , the kinetic sur-
35 9697
vey of Engleman , and recent uork by other authors .

In Table 1 (p. 35), the 23 reaction mechanism for


31
the shock-initiated CH,-oxidation proposed by Bouman " is

reproduced. This mechanism has been used extensively by

other investigators. Bowman's analysis shows that, in the

temperature range 1900 - 2400°K, this mechanism correctly

3D
predicts the time of occurring and magnitude of the radical
(0, OH, H) concentration overshoots, and also predicts a
departure from a "partial equilibrium" state during the
initial stages of the radical concentration overshoots.
Boni and Penner ^ ficonducted a sensitivity ana-
lysis of the mechanism proposed by Bowman at 2QOO°K. Their
analysis showed that Reactions (Rl), (R7), (R16) and (R25)
are the most important reactions for all of the species
CH4 + n £ CH3 + H + N (Rl)
H + 02 £ 0 + OH (R7)
CH3 + 02 £ H 2 CO + OH (R16)
H2CO + PI £ HCO + H + M (R25)
In addition Reaction (R15) is important for the major free
radical species, i.e., OH, 0, and CH20. The 0-atom profile
is dominated by Reactions (Rl), (R7), (R16), (R25), (R15)
(as cited by Bowman) plus Reactions (R24) and (R3)
CH3 + 0 £ H2CO + H (R15)
HCO + fl £ CO + H + N (R24)
CH4 + H £ CH3 + H2 (R3)
98
The mechanism of Reaction (R16) has recently been studied ,
and the oxidation of methyl radicals was shown to occur
through the steps
CH^\J + 0 * CH<-90 + H
CH3 + 02 ^ CH30 + 0
CH3 + 0 £ CH20 + H2

31
The methoxy radicals produced react by means of
CH3D + M £ CH20 + H + M
CH30 + 02 £ CH20 + H02
The reaction of CM, uith 02 and the thermal decomposition
of CI-UO are the most important pathways of methyl radical
oxidation,
Peeters and Mai-men 27 measured the concentration
of all species, stable as uell as unstable, throughout the
reaction zone of lou-pressure lean methane-oxygen flames.
The results of their analysis shou that (i) CH, reacts 15%
uith OH, 15% uith 0, and 10% uith H; (ii) CH,
O
is destroyed
by 0 atoms; (iii) CH90 disappears 4Q?o by thermal decompo-
*
sition, 40% by reaction uith OH, and 20% by 0 attack; (iv)
CHO reacts mostly uith 02; (u) H02 is destroyed 45% by OH,
40% by 0, and 15% by H, Their studies shou that the rate
limiting step is the decomposition of formaldehyde
CH20 + N = CO + H2 + 1*1 (31)
They argued that Reaction (31) explains quantitatively the
function of Hz
9 in the flame, and the activation energy of

Reaction (31) is in good agreement uith the overall "acti-


vation energy" of lean-CH^/02 flames (about 40 kcal/mole).
On this basis, they rejected reactions such as
CH20 + 0 £ CO + H + HO (32)
CHO + N ^ C O + H + N (R24)

Furthermore, they pointed out that Reaction (R25) of Bouman


may uell be the rate-determining step in shock studies,

and Reaction (31) cannot contribute significantly to the


removal of CH20 in shock tubes since the major part of the
CH20 is consumed by H atoms in a chain mechanism initiated
by Reaction (R25). Their results also shou that C02 is
formed mostly via Reaction (R9)
CO + OH £ C02 + H (R9)
The important reactions in Peeters mechanism are
CH4 H- OH £ CH3 + H20

CH4 + 0 £ CH3 + OH
CH3 + 0 £ CH20 + H
CH20 + N $ CO + H2 + M

CH20 + OH(0) £ CHO + H20(OH)


CHO + 02 £ CO + H02
H02 + OH(0) * 02 + H20(OH)
H02 + H ^ 20H
H + 02 ^ OH + 0
0 + H2 ^ OH + H
OH + H2 £ H20 + H
OH + OH £ H20 + 0
CO + OH £ C02 + H
•7 C
(For extensive kinetic data see England ). Further
support for Reaction (31) as the rate-determining step
in flames can be found in the investigations of Migauchi
et al . They studied premixed CH4/02/N2 flames at 0.1

33
atm. and equivalence ratios between 0.88 and 1.42. They

found the rate-controlling reaction of methane oxidation

to be Reaction (31) uith a rate constant k31=2.1x!015 exp

(-17,620/T) mol"1cm3sec"1.

2.4.2 Methane Oxidation in Fuel-Rich Flames

In the combustion of fuel-rich hydrocarbon

flames,the -concentrations of hydrocarbon fragments are high


enough that their recombination reactions become important

and hydrocarbons higher than the original fuel are found as


intermediates.

In CH^-oxidation, the chemistry of C2-hydrocar-

bons becomes important in the fuel-rich mixtures. Gardiner

and Olson 37 have conducted a series of shock tube experi-

ments and numerical analysis of fuel-rich CH,/02/Ar mix-

tures. The reactions of CH 2 » C^H,, and CoH- were found to

play an important role in their shock tube experiments.

The result of their experiments can be modeled by a sixty-

three-reaction mechanism which is reproduced in Table 2

(18000<T<2700°K). They omitted species CH and C2 on the


basis that their concentrations and reaction rates are too

small to affect the overall course of reaction. The spe-

cies ChUO and d-U02 were omitted on similar bases. Their


sensitivity analysis showed primary sensitivity to the py-

rolysis reactions (1) and (30), and to the oxidation reac-

tions (7) and (48). Additional sensitivity is present for


TABLE 1,

Methane oxidation in fuel lean mixtures

Reaction Rate constant, kf*

CH,-i-H+M 2X1017 exp(- 44 500/77)


2. 6X10" exp(- 6290/2T)
3. CH4+H-»CH,+H, 2.24X10*7'' ex p ( - 4 400/7")
4. CH,+O— CH,+OH 2.1X10" exp( -4 560/r)
5. H,+OH—H+H,O 2.9X10" exp( -5530/T)
6. O+Hj—H+OH 3.2X10'< exp( -7540/r)
7. H+Oj—O+OH 2.2X10" exp( -8 450/r)
8. O H + O H —O+H.O 5.5X10" exp( -3520/r)
9. CO+OH— CO,+H 4.0X10U exp( -4030/T)
flO. CO+O4-M— CO.+M 5.9X10"exp( -2060/D
11. H+OH+Ar— H,O+kr
12. H+OH+H,O-»HzO-}-H,O
f!3. Oi+Ar-^O+O+Ar 7.9X1011 exp(-52770/r)
tH. H,+Ar—H+H+Ar 2.2X10" exp(-4S300/r)
15. CH.+O-^HjCO+H 1X10"
16. CH.+O)— H,CO+OH 2X10"
f!7. CH.+OH— HjCX^+H, 4X10"
18. H,CO+O— HCO+OH 5X1011 exp(-2300/r)
19. H,CO+OH— HCO-1-Hrf) 5.4X10" exp(-3 170/r)
20. H-CO+H— HCO+H, 1.35X10" exp(-l 890/7")
21. HCO+O— CO+OH 1X10"
22. HCO+OH— CO+H^) 1X10"
23. HCO+H—CO+H, 2X10"
24. HCO+M— CO+H+M 5XlOue.xP(-9570/r)
25. H,CO+M-» HCO+H +M 4X10 U exp(-lS500/r)
26. H+HO,-*OH+OH 2.5X10"exp(-950/r)
27. H+O.+M—HO.+M 1.5X10" exp(500/r)
|2S. C,H,— CH4+CH, 8X10" exp(-4 500/r)
f29. C-H.+O— C^H^+OH 4X10"exp(-32SO/r)
t30. HCO+O,— CO+HO, 4.2X10"exp(-7200/7')

Units: cm, cal, °K, mole, sec. t Re.iction not included in final mecnanism.

35
TABLE 2.

Methane oxidation in fuel rich mixtures

Rate Constant0

Reaction IogIOX n 0

(1) CH 4 + M = CH3 + H + M 17.67 0 46,900


(2) CH 4 + H = CH3 H- H 2 14.86 0 7,600
(3) CH4 + O = CH3 + OH 7.07 2.1 3,840
(4) CH4 + OH = CH3 + H 2 O b 3.54 3.1 1,010
(5) CH4 + CH 2 = CH3 + CH3C 13.00 0 0
(6) CH3 + M = CH 2 + H + M c ' d 16.67 0 46,900
(7) CH3 + O 2 = CH 2 O + OH 11.84 0 4,530
(8) CH3 + H = CH 2 + H 2 c ' e 14.86 0 7,600
(9) CH3 + O = CH 2 O + H 13.83 0 0
(10) CH 3 + OH = CH 2 O + H 2 12.87 0 0
(11) CH 3 + C H 2 = C 2 H 4 + H 13.30 0 0
(12) CH 2 + C H 2 = C 2 H 2 + H 2 13.50 0 0
(13) CH 2 + O 2 = CH 2 O + 0 14.00 0 1,860
(14) CH 2 0 + M = CHO + H + M 16.70 0 36,235
(15) CH 2 0 + H = CHO + H 2 13.30 0 1,660
(16) CH 2 O + 0 = CHO + OH 13.70 0 2,300
(17) CH 2 O + OH = CHO + H 2 O 15.70 0 6,540
(18) C H O + M = H +CO + M 14.47 0 7,400
(19) CHO + O 2 = HO 2 -i-CO 12.53 0 0
(20) CHO + H = CO + H 2 13.30 0 0
(21) CHO + O = CO + OH 13.48 0 0
(22) CHO + OH = C O + H 2 O 13.48 0 0
(23) C 2 H 6 + M = CH 3 -t-CH 3 + M / ' 111.29 -25.3 80.020
(24) C2H6 + H = C2H5 + H2 14.12 0 4,715
(25) C 2 H S + O = C 2 H 5 + OH 13.26 0 3,070
(26) CoH 6 + OH = C 2 H 5 + H20g 13.80 0 1,810
(27) C 2 H 6 * CH3 = C 2 H 5 -H CH4h 14.74 0 10,820
(28) C 2 H 5 + M = C 2 H 4 + H + M1 14.67 0 13,390
(29) C 2 H 5 + O 2 = C 2 H 4 + HO 2 12.18 0 2,445
(30) C 2 H 5 + H = CH3 + CH 3 13.57 0 0
(31) C 2 H 5 + H = C2H4 + H2 12.27 0 0
(32) C2H4 + M = C2H2 + H2 + M 17.32 0 39,810
(33) C2H4 + M = C2H3 + H + M 17.41 0 48.600
(34) C 2 H 4 + H = C2H3 + H 2 14.84 0 7,300
(35) C2H4 + 0 = C H 3 i - C H O 13.35 0 1.360
(36) C 2 H 4 + OH = CH3 + CH2O 13.00 0 0
(37) C 2 H 3 + M = C2H2 * H + M 14.90 0 15,850
(38) C->H 3 + H = C 2 H 2 + H2C 13.00 0 0
(39) C2H2 + M = C 2 H - H H + M 16.62 0 53,850
(40) C 2 H 2 + H = CoH + H2;' 0.89 3.2 250
(41) C 2 H 2 + O = CH 2 + CO 13-72 0 1,860

36

-
Rate Constant0

Reaction Iog10>l n 6

(42) C 2 H 2 + OH = CH3 + CO 12.08 0 250


(43) C2H2 + C 2 H 2 = C4H3 + ff 13.00 0 22,650
(44) C 2 H 2 + C2H = C 4 H 2 + H 13.60 0 0
(45) C 4 H 3 + M = C 4 H 2 - | -H + M/ 15.93 0 30,200
(46) C 4 H 2 + M = C 4 H + H + U> 17.54 0 40,260
(47) H 2 + O2 = OH + OH 13.23 0 24,210
(48) H + O 2 = OH + O 17.08 -0.91 8,370
(49) O + H 2 = OH + H - 14.34 0 6,895
(50) OH + H 2 = H 2 O + H 13.72 0 3.270
(51) OH + OH = H 2 O + O 13.74 0 3,520
(52) H + O 2 + M = HO 2 + M 15.40 0 0
/C'i'V T
\,j j ) ti r>_ O
kj U
n -i-T \i — TJ f~}
jvi — rio^-* 4- \t
•*" 23.88 -2.6 0
(54) H2 + M = H + H + M 12.35 0.5 46,600
(55) O2 + M = O + O + M .11.27 0.5 48,165
(56) CO + Oo = CO2 + O 11.08 0 17,615
(57) CO + OH = CO2 + H k 12.60 0 4,025
(58) CO + O + M = CO2 + M 13.45 0 -2,285
(59) H 2 + HO 2 = H 2 O + OH 11.86 0 9,410
(60) H + H0 2 = OH + OH 14.40 0 960
(61) H + H0 2 = H 2 + O 2 13.40 0 350
(62) OH + H0 2 = H 2 O + O 2 13.70 0 500
(63) O + H0 2 = OH + O 2 13.70 0 500

0
Rate constants in the form A X 7" exp (-6/7"), in cm, mol, s, and K units. The reverse reaction rate constants were
computed using J A N A F thermochemica! data where possible and locally computed thermochemical data otherwise.
b
The T3 factor may extrapolate the rate constant loo high at higher temperatures, but is used here because this
expression goes through the high-temperature data.
c
Estimate. No sensitivity to the rate of this reaction was observed.
^Estimated to be equal to frj/10.
e
Estimated to be equal to Jr 2 .
^ This reaction is in the pressure-dependent falloff region of a unimolecular decomposition. This rate expression is
from Olson et al. [9] for experiments over the range 1330 < T < 2500 K and 0.10 <P < 0.50 a tm. Care must be taken
that this rate-constant expression be used only for the pressure and temperature conditions for which it is valid.
* This is a low-temperature value that agrees with the few high-temperature data.
h
Strongly curved Arrhenius behavior is observed for this reaction.
1
This rate-constant expression was obtained by RRK calculations of the falloff from the experimental high-pressure
limit expjession of Glanzer and Troe [ 4 7 J .
^ Preliminary results.
k
See Baulch and Drysdale [621 for a review of the literature on this reaction.

37
the CH2 reactions (ll) and (13), and reactions (14) and (2).

Literature values for rate constants uere used for Reac-

tions (1), (2), (13), (14), (30) and (48). The authors

cautioned that the H2CO-decay mechanism is uncertain and

insufficient experimental data is available. The numeri-

cal values for the rate constants of Reactions (7) and (11)
uere found by numerical analysis. They believe their re-

sults to give k? to uithin +5Q% and k, , uithin a factor of


2.
38
Harvey and Maccoll studied the formation of C?

hydrocarbons uithin methane-oxygen flames of equivalence

ratios 0.56 to 1.25 at 20 torr. Their experiments shoued

that for their experimental conditions the reaction

CH3 + CH3 + N £ C2Hg + 1*1


is uell into the fall-off region resulting in a lou re-

action rate (the maximum mole fraction of ethane in the

stoichiometric flame uas 5 x 10 ). Their results enabled

them to propose the following ethylene forming reactions

CH3 + CH2 £ C2H4 + H


The ChU and CI-U concentrations in these flames are in

equilibrium via
-x pit
H v* L»MO n/p

Figure 2 shows the dependence of CH20, C2H4, C2H2 and


H2 peak concentrations on flame equivalence ratio.

33
Figure 3 shows the dependence of C 2 H < , C2H2, and hU

on peak fCH for a range of flame equivalence ratios.

They also pointed out that at higher pressures the rate

of the combination of methyl radicals (CH, -f- CH,, + PI


O O

£ C 2 Hg + M) to ethane uill become competitive uith the

other sources of C2 hydrocarbons, and acetylene is formed

in the flame via

+ CHo x~ CoHo 4- HO

39
0-4

0-3
o
o

0-2

O
5

0-1

0-5 0-6 O-7 O-8 09 1-0 11 1-2 1-3


EQUIVALENCE RATIO

Fig. 2. The dependence of CH2O ,C2 H 2 , and H2


peak concentrations on flame equivalence ratio .

40
(PEAK CH3 MOLE FRACTION) 2 i 10s

Fig. 3. The dependence of peak C_H 4 C H and HS on peak


2 2 2 2 '
( CHS ) for a range of flames £ Z 0.56 to 1.25 .

41
Shock-tube studies of the early stages of

pyrolysis and oxidation of methane were conducted by


39
Tabayashi and Bauer . Temperatures ranged from 1950

to 2770°k, total densities (2.37 - 8.91) x ID"6 mol


— *Z

cm , and CH^/CU ratios 3.3 and 6.7. From their pyroly-


sis experiments in CH^/Ar mixtures, they concluded that
methane decomposes mainly via

CH + Pi £ CH, + H + n

C
2H5 + ^ * C2H4 *H * N
C2H3 + Fl £ C2H2 + H + M
methylene is produced mostly by the reaction

CH2 -r H2 £ CH3 + H

and consumed via reactions with CI-L and species gener-


ated during ChL-pyrolysis. Their oxidation studies

show that oxygen reacts only with the fragments of meth-

ane pyrolysis, and under the methane rich condition

oxidation starts predominantly via the reaction

CH3 + 02 £ H2CD + OH

rather than

H + 02 £ OH + 0
and the resulting OH radicals react rapidly with meth-

ane uhich is present in large excess,

CH4 + OH £ CH3 + H20


thereby generating more methyl radicals, uhich reinforce

the pyrolysis chain. Their calculated species profiles


best fit the observed profiles if the reaction rate

constant for the methyl radical-oxygen reaction is given

a value of 2.7 x ID12 x exp (-12,000/RT) cm3moi""1s~1.


Regarding the fate of formaldehyde, they found that

thermal decomposition of formaldehyde uas in agreement

with their experimental results and computer simulations

CH20 + M = CO + H2 + PI
Houever, an alternative mechanism consisting of the

following reactions

CH 2 0 + M £ CHO + H + 1*1

CHO + M ^ C O + H + IV1

CHO + 0 2 £ CO + H 2 0

CHO -f H £ CO + H2

CH20 + H £ H2 + HCO
could not be excluded.

The role of CH2 in CH^-oxidation uas found

to be ambiguous. This study also shous that oxidation

of CO to CO- occurs mainly via the reaction

CO + OH £ C02 -f H
CHAPTER 3

EXPERIMENTAL APPARATUS

3.1 An Introduction to Sampling Techniques in Combus-


tion Studies
40
As Knuth has suggested, specifications for
the ideal instrumentation for flames studies might in-
clude the following broad requirements: (a) monitor
all chemical species, including intermediate and free-
radical species, (b) function in temperature, pressure
and composition ranges typical of flames, (c) resolve
spatial measurements small in comparison uith the flame
thickness, and (d) monitor uithout disturbing the (non-
equilibrium) thermodynamic state of the system.
Shock tubes have been shoun to be suitable
for studies of high temperature processes, and excellent
revieus and books are available that describe their ap-
plicability in chemistry and physics. The shock-tube
method of studying high-temperature processes possesses
at least four major advantages over other methods41
~:

(1) The equilibrium temperature in the after-shock


region can be calculated accurately and re-
liably from the measured shock speed and
conservation of mass, energy, and momentum
considerations,
(2) The "slug" of gas behind the primary shock
is very uniform in pressure and temperature.
Uall effects are negligible because of the
short time scale.
(3) Practically any gas mixture can be heated
to any selected temperature by using the ap-
propriate shock strength.
(4) The heat release by the shock impact is ini-
tially in the form of kinetic energy, i.e.,
translational energy of the molecules.
The major disadvantage is the short observa-
tion time, which for the after-shock region, is usually
betueen 1QD and 400^sec, Thus, optical spectroscopy
is by far the most useful method of following the pro-
duction and/or consumption of individual species behind
shock waves. Typically, three or four species are mon-
itored. If only one optical path is available, then
a oiven set of conditions must be repeated for each of
the species monitored.
Ono of the principal problems with spectro-
scopic studies is that of determining absolute concen-
trations, since absorption coefficients are known only
for atoms, most diatomic molecules, and a few simple
poly-atomic species. In the absence of this knowledge
it is necessary to calibrate the system. This is dif-

45
ficult because radical concentrations of knoun values

are hard to obtain, the production of the radical it-

self might be an experimental challenge, and it is neces-

sary to calibrate over the temperature range to be stud-

ied since the temperature dependence of absorption coef-

ficients of complex molecules is unknown. One also

has to find absorption lines in the available spectral

regions uhich are not obscured by other species. Hence


the use of shock tubes in conjunction with optical spec-

troscopy satisfies requirement (d) most closely, require-

ment (a) least closely.


In an alternative technique, gas samples are

withdrawn from the flame using a sampling probs. Probe

sampling is a straightforward process; the sample is


withdrawn, quenched, and analyzed. The chief advantage

of this technique is that a wide variety of analytical

instruments can be used to follow the fates of species

within the flame. The main disadvantage is that compo-

sition changes due to additional chemical reactions

during transport from the flame to the analytical in-

strument^) are inevitable. Probes also have perturb-

ing effects on the flame that have to be considered:

aerodynamic, thermal, and catalytic effects. Some of

these have been eliminated by the use of very fine mi-


9n
croprobes made of quartz. The relative effects of
several different probe designs on flames have been studied
by England , Malta , and Allen . The use of sampling

probes meets requirement (a) relatively closely and re-


quirement (d) relatively poorly.

Sampling of ions from flames is another tech-


nique used by investigators to study the processes oc-

curring in the combustion mixture. In a typical appa-

ratus, the gas is sampled through a small hole at the

tip of a cone and expanded into a vacuum chamber at

10 to 10 torr. The system is designed to obtain

critical flow through the nozzle. The ions are formed

into a beam by an electrostatic lens and focused through


a small hole, typically 2mm. in diameter, into a second

chamber at 10 torr. which holds a quadrupole mass spec-

trometer and electron multiplier. Both positive and

negative ions can be detected by reversing the polari-

ties of various voltages. Ideally the expansion of the

gas into the vacuum should be fast enough to quench any

chemical reactions so that the expanded gas has approx-

imately the same composition as that in the flame.

This is usually the case for neutral molecules, but


because ion-molecules reactions are typically 1-2 orders

of magnitude faster than those betueen neutral molecules

considerable distortion of ion signal can occur .


This is particularly important for negative ions because

of the possibility of fast associative reactions. Fur-

thermore, the boundary layer that exists on the sampling

47
cone uill have an additional effect on the flame gas

composition before it expands through the nozzle .

This and the effects previously described can be min-

imized by proper design of the sampling system, giving

this technique an advantage over some other flame-sampl-

ing techniques, probe sampling for example. The chief

disadvantage of ion sampling is that it can be used

only for species uhose concentrations are known func-

tions of measurable ion concentrations. Hence this

technique satisfies requirement (d) most closely and

requirement (a) least closely.

The molecular-beam-mass-spectrometer (MBFIS)

sampling system satisfies the abovementioned requirements

uith better balance. The U.C.L.A. MBfIS sampling system

has been employed in the study of reactive mixtures since

1969, and numerous studies have been conducted in this

laboratory to eliminate or minimize all those effects

uhich might either distort the gas composition during

sampling or make the interpretation of measurements

a confusing task. A discussion of these perturbing


effects is left for the next section, and only a descrip-

tion of the system uill be made here.

The FIBMS sampling system consists typically

(see Figure 4) of a sampling cone (uhich might be en-

closed in a chamber for lou-pressure studies), source

chamber, skimmer, collimating orifice, detector chamber,


and quadrupole mass spectrometer detector. Inclusion
of an effusive source and a uheel-chopper in the collima-
tion chamber allows calibration of gas mixtures and
phase-sensitive detection. The source gas (at pressures
of up to several atmospheres and temperatures of several
thousand degrees Kelvin) expands through the sampling
orifice into the source chamber (uhere background pres-
sure is typically 10—3 to 10-1 torr.) and the core of
the resulting jet is skimmed and transferred into the
collimation chamber (uith background pressure from 10"~
to 10 ' torr.). The molecular beam is chopped in the
collimation chamber for phase-sensitive detection, and
the collimating orifice admits these molecules flying
along the system centerline into the detector chamber.
The mass-spectrometer detector generates a signal pro-
portional to the species density in the beam at the de-
tector.
As uas mentioned in the discussion of ion
sampling from flames, the intention here is to lower
suddenly the temperature and pressure of the reaction
mixture, so that further chemical changes are prevented
and the composition of the expanded gas approximates,
that of the flame.

49
FABRITEK WAVE
COMPUTER GENERATOR

TOP SIGNAL
EAI 2 5 0 QUAD. BEAM EFFUSIVE MOVABLE
MASS SPECTROMETER CHOPPER SOURCE SKIMMER

\ WATER-COOLED
HEAT SHIELD
ION
SIGNAL

DETECTOR EXCITOR

V A C U U M PUMP V A C U U M PUMP V A C U U M PUMP

LOCK-IN LSI-1 1
COMPUTER BURNER POSITION SIGNAL
AMP.

Fig. 4. The UCLA MBMS sampling system.


Thus, applicability of MBMS sampling techniques

to the study of chemical reactions in flames results

largely from the rapidity with uhich the gas tempera-

ture and pressure decrease during the expansion through

the orifice in the sampling cone. A plot of the tem-

perature ratio T/To versus dimensionless time, taken

from reference 40, is shoun in Figure 5. T is the gas


temperature, a is sound speed, t is flou time (zero

at the throat), x is axial distance (zero at the throat),


d is throat diameter, and subscript o refers to source

(stagnation) conditions. Furthermore, a point that

should be stressed is that, uhile in the sampling of ions

one is actually measuring average properties of the

gas that flous through the sampling orifice, in MBMS

sampling only that small fraction of the sampled gas

that flous in the vicinity of the orifice centerline

reaches the quadrupole mass spectrometer detector.

This results, as uill be explained in more detail in the

next section, in the detection of molecular-beams that

are much more representative of the flame gas.

51
Fig. 5 Temperature ratio as function of dimensSonless flow time
for free-jet gas flows from orifice.

52
3.2 Possible Composition Distortions in_ flBNS. Sampling
of Flames.

In an idealized model4 of molecular-beam

sampling, the sampled gas undergoes an isentropic free-

jet expansion through the source orifice and the speci-

fic-heat-ratio, ^Tf remains a constant throughout this

expansion. Flou remains a continuum upstream from the


skimmer entrance and becomes free-molecular downstream

from this surface. There is assumed to be no background

scattering of the free jet and the molecular beam.


The beam composition at the detector is the same as

the source-gas composition.

In the design of a flBHS sampling system, var-


ious departures from the above model must be considered.

As the combustion mixture flous from the source to the

detector, the following are encountered: sampling-probe-


induced distortions of the gas composition, chemical

reactions during the expansion of the gas into the vac-

uum, species condensation, pressure diffusion, freez-


ing of vibrational, rotational or translational degrees

of freedom (in this order), skimmer interference, Flach-

number focusing, and scattering by background gases


(in any one of the lou-pressure chambers). Except for

probe-induced distortions, all of the above considera-

tions are treated in the paper by Knuth , and the dis-

53
cussion is based mostly on this reference and uork by

other authors with the U.C.L.A. MBMS sampling system.

3.2.1 Free Get Expansion.

Before proceeding into the discussion of com-

position distortions in MBMS sampling, the reader might

benefit from a brief introduction to the problem of

unconfined expansion from a sonic orifice into a lou-pres-


sure chamber.

As Sherman indicated, the problem of trans-

onic flou through an orifice or axisymmetric nozzle re-


mains unsolved. Houever, the supersonic region of the

flou is very little influenced by the transonic region,

and the method of characteristics can be applied in

the inertia-dominated region of high-Flach-number isen-

tropic flou.

The free-jet shock uave structure is shoun

in Figure 6. As pointed out above, the method of char-

acteristics is applied to the region inside the barrel

shock. Ouen and Thornhill were the first to carry


out the computations for gases uith T = 7/5 assuming

slightly supersonic flou at the orifice (M = 1 leads

to a singularity). Sherman and Anderson report


calculations for other values ofj. The main feature

of these calculations is that about one nozzle diameter


from the orifice the streamlines appear to radiate from

a "source" at x . Furthermore, the density decreases

along each streamline in proportion to the inverse square

distance from this "source." The isentropic core of the

flou appears to pass from the nozzle exit through a bar-

rel, the sides of uhich are the shock uaves generated

by the coalescence of compression uaves originated at


the jet boundary as the reflections of the initial ex-

pansion fan. The Mach disk is formed by the intersec-

tion of the lateral shock uaves.

55
SONIC MACH
ORIFICE DISK

BARREL
SHOCK

Fig. 6. Free-jet shock structure.

56
The variation of the mach number, M, along

the jet axis with distance x has been carried out by

Sherman and Ashkenas ' for'"/" = 5/3, 7/5, and by An-


derson49 for If = 1.30, 1.20, 1.10, and 1.05. They used
the best fitting formula
f TlP —1 ~'°
i«M»™»» I ^
d* xl

where A, and x are constants which can be obtained


o
from the above references. For other values ofT, one
may use the following equation suggested by Knuth ,

(34)

Note that Equation (33) has a singularity at x = x ,


and therefore it is only applicable if x >1.
d* '
It is important to point out effects that are

not considered in the ab'ove analysis of free-jet expan-


sion. (1) Boundary-layer growth on the converging nozzle.
The effect of this boundary layer will be to change
the effective orifice size (see Ref. 51 for a compre-
hensive analysis) and distort the flow pattern near the
orifice. In equations (33) and (34) an effective ori-
fice diameter then must be used.
(2) Freezing of vibrational, rotational, and transla-

tional degrees of freedom (always in this order). Freez-

ing of degrees of freedom will cause T to change in the

57
expansion and not be constant as assumed In Thornhill's

analysis. Freezing of translational degrees of free-

dom causes the flou to transit from continuum to free-


molecular. Equations (33) and (34) for the centerline
Flach number do not hold in the free-molecular flou re-
gion. It should be noted, houever, that the density
keeps decreasing as the inverse square of the distance
as a consequence of the source-point nature of the flou.
(3) To the author's knowledge the determination of
jet boundaries for gas mixtures uith diffusive separa-
tions (e.g., ba rodif fusion) has not been solved theo-
retically.
Sherman 46 determined the location of the Mach

disk empirically,

p = stagnation pressure
p = ambient pressure
This is independent of v and holds for 15 ^ P0/0 ^17,000.

52
Bier and Schmidt determined, from photographic studies
of jets, empirical relations for the nach-disk diameter
y
7 and the maximum diameter of the barrel shock y Q
m a
3
for 10 ± p / <. 10 .

v ^ 0.598
£n = 0.252 fp I (36)
d*

53
0.591
- - 0 316
d
*

Sometimes, values for the Mach number near


the sonic orifice (upstream of the effective "source"

at x ) and upstream of the orifice are needed. As in-

dicated above, Equation (33) is restricted to x/d*^l.


Values for M in these tuo regions have been determined
experimentally by Sherman and Ashkenas for o = 7/5,

and Anderson ' f or 1) = 5/3. Sharma arrived at ex-

pressions for T/T in terms of x/d* in these tuo regions


from theoretical considerations. His calculated Plach

numbers are larger than experimentally determined ones.

3.2.2 Chemical Relaxation in Supersonic Expansion

Successful nBHS sampling requires that all

chemical reactions occurring in the sampled gas mixture

are quenched, i.e., that they "freeze," as early as

possible in the free jet expansion through the sampling

orifice. If the chemical reaction is frozen upstream

of the sonic orifice, then the sampled gas is most repre-


sentative of the undisturbed gas composition. However,

if the reaction proceeds at large rates it uill be able

to follow changes in temperature and pressure up to

the nozzle throat, and in some cases several throat

diameters downstream.

59
Hayhurst studied the occurrence of chemical

reactions in the supersonic expansion of a gas into


a vacuum. His investigation focused on the hydration
of HjO4"

H30++ H20 + M # HgQ* + PI (38)


in an atmospheric premixed H2/02 flame uith nitrogen
added as a diluent. Reaction (38) is fast enough to
be equilibrated in the flame. For this purpose it is
necessary to calculate the variation of temperature,
pressure, and density of the gas in the free jet expan-
sion. Hayhurst arrived at a complete description of
the isentropic flou field by the method of character-

istics. (The free-jet problem has been treated by other


authors ' ). Note that in the high temperature and
pressure regions of the expansion, reaction (38) can
be assumed to be equilibrated at all times. Hayhurst

concluded, "The calculations described above shou that


there is continuum flou in the expansion of the gas through
-.7
the sampling nozzle for about 5 x 10 sec, after uhich

the flou is molecular. Consequently, if the relaxa-


tion time of any chemical reaction is much greater
_7
than 5 x 10 sec at atmospheric pressure, the reaction
X

uill not proceed during the expansion. If such condi-


tion holds for all possible chemical reactions, com-

position uill not be affected by sampling. This repre-


sents the case of ideal sampling. Houever, a reaction

60
_
uith a relaxation time less than 5 x 10 sec uill be

at equilibrium in the flame, where the time available


_3
is about 10 s, and uill be able to follow any changes
of temperature, etc., up to the nozzle throat. The
relaxation time of a reaction similar to (38) uill

increase once inside the nozzle, because of the fall


in gas pressure and temperature, so that a point even-
tually be reached at uhich equilibrium can no longer
be maintained and the composition uill become "frozen."
Consequently, the composition observed uill not neces-
sarily be that of the original sanple, but uill corres-
pond to conditions at some point in the sampling nozzle."
Ions from flames have also been sampled by f'lorley 45 .
His results are in agreement uith those of Hayhurst.
In vieu of the importance of chemical relax-
ation in systems at high temperature and pressure,
Knuth studied the chemical relaxation process in
flou systems. He identified the proper relaxation
time as being^ . , the relaxation time at constant

enthalpy and pressure, rather thariTy p , relaxation

time at constant temperature and pressure. (Relaxa-


tion time is defined as the time required for the de-
\
viation from equilibrium to reduce to 1/e its initial
value if the existing thermodynamic constraints uere

maintained.) Knuth combined Bray's sudden-freezing

model uith Phinney's 57 relaxation-time-freezing point

61
criterion, i.e.,

§tYh>p -C (39)

uhere D/Dt is the hydrodynamic derivative and C is


a constant, to arrive at a chemical relaxation equa-
tion, with C given a value of 0.5. Equation (54) of
40
Knuth provides a criterion for the early freezing

of a reaction. This equation is characterized by tuo


parameters; namely, the kinetic parameter defined as
the ratio of the chemical activation energy to the
random thermal energy, and the scaling parameter defined
as the ratio of the floui time to the collision time.
58
Young used a last-activated-collision model
to determine a criterion for the freezing of a chemi-
cal reaction. In this model, a given chemical reac-
tion freezes approximately at the location uhere the
sample molecule encountered its last activated collision,
Young studied chemical freezing of bimolecular reac-
tions involving an activation energy. The number of

activated collisions of a molecule uas computed along


the free-jet centerline. The results of this calcula-
tion indicate that chemical freezing in the free jet
is characterized by the kinetic and scaling parameters
as defined by Knuth. Furthermore, the analysis reveals
that complete chemical freezing in the free-jet is
obtained if the ratio of the logarithmic scaling factor

62
divided by the kinetic parameter is less than 0.5.

0 ~ / E* (40)
•V" = scaling parameter = d*/a YAR
o
E* = kinetic parameter = E*/kT
— . Q

Too — mean time between collisions of A and B molecules


AD

E* = activation energy

d* =, diam. of sonic orifice


( ) denotes stagnation conditions

These results can be used as a design criterion to

develop a sampling system uhich avoids composition

distortion by chemical reaction in the expansion.

3.2.3 Species Condensation

Species condensation in supersonic molecular


59
beams uas first observed by Bier et al in 1956.
Condensation uould make the interpretation of measure-

ments very complicated, but fortunately, as Knuth con-

cludes in his revieu, the probability of condensations


in FIBFIS sampling of high temperature systems is very

small .

The classical theory of homogenous conden-

sation postulates that a critical size nucleus is spon-

taneously formed uhich has the properties of the bulk

phase. Such theory cannot be applied to the rapid ex-

63
pansion occurring in a free jet for it leads to crit-

ical sizes that are of the same order of magnitude as

the molecular diameters . Instead a kinetic theory

approach ' has been used to predict condensation


in the free jet, and the rate limiting step is the form-

ation of dimers by termolecular collisions, not only

for monoatomic gases, but also for polyatomic gases.

Once the dimer is formed, tuo body reactions can produce

trimers and larger clusters which lead to the formation


cf a condensate.

Knuth shows that the ratio of the flou time

d*/a to the mean time,'f" , between consecutive three

body collisions is the relevant dimensionless scaling

parameter in dimer formation. The smaller the flou time,


relative to the mean time between termolecular collis-

ions, the smaller is the dimer production. This hy-

pothesis is supported by experiments conducted by several


{• I /- o
authors ' . They found that the excess dimer mole

fraction (relative to the equilibrium mole fraction

in the source) is approximately a linear function of


p d*. Note that the scalina parameter d*/y a is re-
~o o o
r\

lated to P by equation (56) of Reference 40. Further-

more, these experiments show that for d*/^Oa0 ^50


dimer formation is insignificant (see Table I of Ref.

40 for example), and that this rate decreases as TQ


.63, 64, 65
increases

64

-
3.2.4 Pressure Diffusion

For theoretical treatment of separation ef-


fects in free jets and shocks, the reader is referred
Fi R fi 7
to the continuum analysis by Sherman ' , Mikomi and
Takashima , and the kinetic theory approach to a shock-
uave structure in binary mixtures by Oberai . Our aim
here is to present the reader with the main results
of the above experiments and theory.
In his treatment of the free-jet of a binary
gas mixture, Sherman combines the continuum flou
conservation equations (Navier-Stokes), uith the bi-
nary diffusion equation (41) based on the Chapman-Enskog
theory of diffusion,

fv A f D A B Tf(l-f) fmA"mB V Inp


1 m
-7 (41)

Uhere the subscript A refers to the heavier species;


f is the mole fraction of the heavier species, m is
the mean molecular mass given by
m = fm ft + (l-f)m Q

D Q R and(A-are the binary diffusion coefficient and


H cj

thermal diffusivity, respectively; V. is the diffusion


velocity of the heavier species. Examination of Equa-
tion (41) leads to the following conclusions: (1)
In the free-jet expansion and in a shock transition,

65
the temperature and pressure gradients are in the same

direction. Thermal diffusion opposes barodiffusion

under these circumstances. Houever, an order-of-mag-

nitude analysis of Equation (41), as applied to the free-

jet or the shock uave, shou that barodiffusion is the

dominant separation mechanism. (2) In the free-jet


expansion, the temperature and pressure gradients exist

in the radial direction outuard from the centerline.

The higher pressures and temperatures occur on the


axis and decrease radially outuard as the Mach number

increases. Thus, barodiffusion in the free-jet uill

cause an enrichment of the heavy molecules along the

jet centerline (in barodiffusion the heavy molecules

diffuse to higher-pressure regions).

Sherman obtained an expression for f, the

mole fraction of the heavy species, along the center-

line of the jet, by expanding all variables in terms

of inverse pouers of the Reynolds number based on stag-

nation conditions and orifice diameter, d*, and solv-

ing for the first-order perturbation term in the mole

fraction, f, by integrating the equations (the zeroth


order approximation leads to isentropic flous, and

homogenous mixtures). This expression can be found

in Ref. 66. The theoretical results of Sherman have


70
been validated by the experimental results of Rothe ,

66
71 72
Sebacker , and Campargue . These results show that
the heavier species are enriched along the jet center-
line as predicted from the diffusion equation and that
most of the separation occurs over a distance of three
throat diameters from the sonic orifice. Furthermore,
negligible separation in the free jet was found for

Knuth
40 considers Sherman's results and shous

that for
C = o(i) (42)
Sc m -1
mass separation due to barodiffusion near the source
2
orifice is negligible for Re £ 10 . This criterion
is conservative for x/d* > 3 due to ordinary diffusion
(Sherman neglected ordinary diffusion in his theoreti-
cal treatment. Ordinary diffusion, houever, may have
to be considered for x/d* ^> 3). In Equation (42),
C is a constant, and subscript o refers to stagnation
conditions.
Extensive experimental studies of the shock
structure in binary gas mixtures (see Ref. 40 for an
extensive literature review) shou the validity of the
theoretical treatment of Sherman and Oberai The
following features should be noted: (1) Ouing to the
fact that barodiffusion accelerates the heavier mole-
cules through the shock, and the light species in the

67
opposite direction, the shock and the jet boundary are

enriched with the light gas molecules. (2) The over-

all shock thickness, in terms of upstream properties,

is larger than for a pure gas under similar conditions.


40
Knuth argues, "Effects of pressure diffusion in the

barrel shock are important in sampling studies if the

relative enrichment of light species extends to the

jet centerline." He continues, "The important role

of the pressure gradient through the barrel shock sug-

gests that a suitable criterion for no reversed separa-

tion effects at the jet centerline might be provided

by the criterion for a diffuse shock structure. As shoun


70
by Rothe , the shock thic
thickness, o , is related to the

flou parameters as follows

. VH.0 J3*
S ^

A diffuse shock is realized ifJ /X_5 is sufficiently

large. Furthermore, for

P
I/Re o /^ P«
i
s
reversed separation will be insignificant along the

jet centerline. (Separation effects due to the shock

structure have been called "reversed separation" and

"background invasion.")

68
3,2.5 Relaxation Phenomena

Considerations of relaxation effects in the


free-jet expansion are most relevant in FIBMS sampling
of combustion systems. This uill be more apparent af-
ter the discussion of Mach number focusing and the time-
of-flight temperature measurements.
Consider the free-jet flou of a gas with vi-
brational, rotational, and translational degrees of free-
dom active at source conditions. Recall from the Kinetic
Theory of gases that to achieve relaxation, i.e.,equil-
ibrium, of all degrees of freedom the number of colli-
sions required is the largest for vibration and the least
for translation. One can characterize the relaxation
process by a characteristic time *• (the time for the de-
viation from equilibrium to be reduced to 1/e its ini-
tial value if existing thermodynamic constraints are
maintained). Furthermore, one can define a relaxation
time for each of the degrees of freedom. Kinetic Theory
also shous that as the temperature and pressure of the
system decrease the relaxation time increases, since
collisions necessary for energy transfer between the
degrees of freedom become less frequent.
Thus, if a significant change in temperature
or pressure is realized in a very short time, as it is
in the case of a HEMS free-jet expansion (see Fig. 5),

69
and as uas shown above, the relaxation of specified degrees

of freedom will lag considerably from the change in the


thermodynamic constraints. Uhen this occurs, the degree
of freedom is said to be "frozen," and the proper criterion
for freezing is that I for the degree of freedom be much
larger than the flow time through a given state. Since vi-
bration requires the largest number of collisions for re-
laxation, it freezes first, and it is folloued by freezing
of rotational and translational degrees of freedom.
As uas discussed in Section 3.2.2, Knuth devel-
oped a freezing point criterion, namely

uhere c is a constant of order unity. As Knuth points out,


this criterion is valid for all relaxation processes oc-
curring in flou systems. For each relaxation process, the
terms "partial-equilibrium flou," and "partial-frozen flou"
are used to identify flou upstream and dounstreara, respec-
tively, from the freezing surface of the relaxation pro-
cess. Prediction of the location of each of the freez-
ing surfaces requires criterion (39), and the tempera-
ture dependence of the collision number for the degree
of freedom of interest. Furthermore, the variation
of the centerline Mach number uith distance along
the jet axis must be knoun. Expressions to predict
the Wach number can be found in Section 3.2.1. The

70
prediction of freezing-point locations is beyond the

scope of this discussion, and the reader is referred

to the uorks by Knuth et a!50» 73


» 74
and Sharma54.

3.2.6 Skimmer Interference

Most of the literature pertinent to the free-


jet interference by the skimmer has been reviewed by
4D
Knuth . Criteria uhich facilitate the design of sys-

tems uith minimum skimmer interference can be found

in this reference. As Knuth points out, consequences

of this interference are (1) decrease of beam inten-

sity, (2) widening of the velocity distribution, and (3)

distortion of the beam composition, listed in order

of decreasing sensitivity to skimmer interference.

The aim here is to skim the isentropic core

of the free-jet uith minimum interference by the fol-

lowing considerations: (1) The length of the skimmer

must be such as to minimize stand-off shock (.due to


free-jet-interactions with the chamber wall) interfer-

ence.
(2) The shock structure must be attached to the skimmer

tip in order to effectively sample the isentropic core

of the jet. This puts a restriction on the external

cone angle one can use depending on the Mach number.

(3) The beam can be attenuated effectively by colli-

sions with molecules reflected off the inner surface

71
of the skimmer. This effect is minimized if the in-

ternal angle is sufficiently large.

(4) The effects of orifice size uere studied by Fisher


75
and Knuth . For a conical skimmer, the skimmer-in-
duced decrease in measured Plach number is less than
I
3^ if KnQ^ (= X ol/d,), uhere the mean-free-path A -,
is based on molecular density at the skimmer and on

collision cross section at source temperature, is great-


er than about 0.3-0.4. The first three considerations
lead to Equations (60) - (62) of Knuth.

3.2.7 Mach-Number-Focusing

Composition distortion due to Mech-number-

focusing is an unavoidable consequence of the differ-


ent velocity distributions of the heavier and lighter
species in the free-jet. The lighter species have a

greater tendency to fly auay from the centerline than

the heavier species. The end result is an enrichment


of the beam uith the heavier molecules.
Studies of P l a c h - n u m b e r - f o c u s i n g have b e e n

made by Stern and Waterman for the case of parallel
continuum flou at the skimmer, Sharma for mixtures
73
of C02, Knuth et al for mixtures of monoatomic gases,
and Yoon and Knuth for mixtures of diatomic gases.
In examining the effects of Plach-number-fo-

cusing the follouing flou model is considered as de-

72
picted in Fig. (?):

(1) For a given species, transition from continuum

to free-molecule flou may or may not occur upstream

from the skimmer entrance.

(2) If transition occurs, then it takes place at a

surface uhich is spherical uith origin at the source


orifice.

(3) The molecular velocity distribution at the skimmer


entrance is, in general, a steady tuo temperature Plax-

uellian distribution superimposed on a spherically

symmetric radial hydrodynamic velocity.

Before free-molecule flou is realized in the

free-jet expansion, the translational temperatures

Ti and Ttl are equal. But, as the expansion proceeds,

the flou deviates from isentropic flou due to a large

decrease in the collision rate. T,t becomes frozen

earlier than Tj_ . Further dounstream, the variation

of T_L uith distance suitches from nearly isentropic

to a collisionless type dependence uhere Tj_ decreases

(due to geometrical cooling) uith a slope of -2. In


Rach-number-focusing studies, the relevant temperature

is Tj_
Knuth has shoun that the ratio of molecular

number density at the detector, n ,, to the molecular

number density at the skimmer entrance, n,, for a given

species can be expressed by


SOURCE

{The angles amax and £max are defined respectively by the line of sight from
the detector to the skimmer lip and by a radial line from the source to the
skimmer lip.)

Fig. 7. Schematic diagram of model considered in

Mach-number focusing .

74
(45)
n
Where the geometrical parameters are defined in Fig.
(7), and S^,, the speed ratio for the given species
at the skimmer entrance, is given by

=u (46)
2kT
m
i.e., the ratio of the hydrodynamic velocity to the
most probable random speed perpendicular to U, at the
skimmer entrance.
Writing Eg. (45) for a minor species A and

the major species B and taking the ratio of these tuo


expressions, one arrives at the enrichment factor 0(.,

) (47)

The enrichment factor depends on both system geometry

and the speed ratios. The effect of the speed ratios


on ®C , is knoun as Mach-number-focusing. For the spe-
2 ,.2
cial case that Sj, <^max -^ ^or ^ anc' B » ^-^' C47) 'De~
comes —, 2
(48)

i.e., independent of system geometry.


Three cases are generally considered: (1)

75
Continuum-flou up to the skimmer. (2) Transition

to free-molecule flou upstream from the skimmer. (3)

Transition of free molecule flou for only one of tuo

species in a binary mixture. The analysis through which

^ ± is obtained for each of the above three cases can

be very complicated. The reader uill find comprehen-

sive treatment of Mach-number-focusing effects in free-

jets in the references given at the beginning of this

section.

3.2.8 Background Scattering

Scattering of beam molecules by background

molecules may occur in all of the FIBMS sampling system

chambers. In practice, only scattering in the source

chamber might result in significant beam attenuation.

If the beam has intensity I in the absence of back-


77
ground scattering, Anderson and Fenn showed that

the effect of scattering can be expressed by

I = Io exp (-o- nx) (49)

Uhere I is the observed intensity, n is the number

density of scattering molecules,xis the distance over

uhich scattering occurs, and cr is the effective scat-

tering cross section.

Note that in a gas mixture, scattering might

result in distortion of the relative densities of the

76
several beam species. Knuth has developed criteria

to minimize this effect and the beam attenuation in the

source and collimating chambers.

3.2.9 Flame Perturbations By Sampling Cones

Feu studies have been made regarding the ef-


fect of sampling cones on flames. Furthermore, as

uill be shown, some of the theoretical studies that

have been made are not totally relevant to FIBMS sampling


systems.
78
Biorde et al investigated the effects of

probes on CFL/O^/Ar flames at 0.04 atm. They compared

the relative effects of a quartz-microprobe uith the

effects of quartz sampling cones of various angles and

orifice sizes. Comparisons uere made uith respect to

(1) visual perturbation of the flame, (2) stable species

concentration profiles, (3) temperature profiles, and

(4) sensitivity to radicals. Species profiles uere

obtained by MEMS sampling techniques. The measured


temperatures in the absence of a probe uere compared

uith temperatures measured tuo probe-orifice diameters


upstream of the sampling cone by means of a thermocou-

ple. Biordi et al.argue that this is the relevant

temperature. Their experiment led them to the follou-

ing observations:

•77
(1) As the cone angle increases, the reaction zone

appears to shift dounstreara and becomes narrower, and

the concentration gradients steepen. As they concluded,

"These effects may be rationalized in a qualitative

fashion as due to the cooling ahead of the probe and

'attachment' of the flame to the probe." The cooling

would only result in the translation of the concentra-

tion profile (the probe is actually sampling flame


gases corresponding to a point upstream of its actual

physical location). Attachment causes non-uniform

motion of the flame, and the relative changes in the con-

centration profiles are a result of it. As Biordi

indicates, attachment is disastrous if one requires

the information obtainable from the slopes of the spe-

cies profiles,

(2) The perturbed and unperturbed temperature profiles

differed by about 200°K in the low-temperature region

and by about 100°K in the high-temperature region.

Houever, for the cone for which visible attachment

occurred, the effect on the temperature uas so great


that it would be impossible to relate the temperature

with the composition profiles to the accuracy required

to calculate rates in the primary reaction zone.

(3) It uas possible to detect radical species, e.g.,

0 , OH, H, with all the cones investigated. The mi-

croprobe resulted in a 9Q% loss via wall and internal

78
collisions before a beam uas produced.

They found that a quartz sampling cone uith


40 total angle and thin leading edge minimized all

of the abouementioned distortions for their experimental

conditions, and that radical concentrations uere more

in agreement uith those obtained by other means. Higher

pressures would allou the use of larger cone angles


and thicker leading edges.
79 8D
Hayhurst et al ' studied distortions caused

by a sampling cone (1 to 2 mm. in length) projecting

from a flat plate into 1 atm. H2 - 0- - N« flames.


Ions formed in the flame uere formed into a beam by the

use of electrodes and focused through a small hole

into a detection system. The flow field model consisted

of a hemispherical point sink adjacent to a flat plate.

Their flou-field studies shou that in order to mini-

mize aerodynamic disturbances of the flame and boundary-

layer effects the use of long sampling cones of rela-

tively small external angles uith large orifice sizes

is required. These external effects include heat trans-


fer, catalytic reactions, and composition distortions

due to chemical reaction in the boundary layer. Hou-

ever, as pointed out, efforts to reduce boundary layer

effects are in conflict uith efforts to minimize com-

position distortions during the gas expansion through


the sonic orifice uhich require large internal cone

79
angles and small orifices . This is because the resi-

dence time of the gases before molecular flou is at-

tained is prolonged by making the cone more pointed

or by enlarging the orifice. (See section 3.2.2 on


chemical relaxation).
81
Smith " analyzed the problem of a long sampl-

ing cone in lou-pressure ChL/02 flames. Smith's model


consisted of a sink located at the tip of an infinitely

long cone with arbitrary cone angle. The main result


of his analysis is that the stagnation streamline moves

towards the cone tip as either the cone angle or the

sink strength is decreased. (The stagnation stream-

line separates that portion of the field that enters

the cone orifice from that uhich does not enter it.)

Smith's results shou that a compromise must be made be-

tween disturbances of the sample before and after it

passes through the sampling orifice. This is in agree-

ment uith Hayhurst's results discussed above, namely,


that during the expansion chemical freezing occurs

earlier if large cone angles and small orifices are

used. However, consideration of distortions on the


flame before expansion require the opposite.

The studies by Hayhurst et al,and Smith can

be used to predict the extent of cooling of the sampled

gas and of composition distortion due to the presence

80
of the cone when one is measuring the average properties

of the gas that flows through the sampling orifice.

This is the case for the sampling of ions that are


focused into an ion beam. Thus, for example, Smith
shous that if the fraction of sampled gas that has a
smaller residence time in the thermal boundary layer
than a given relaxation time for some chemical reac-
tion is large, then one can expect little composition

distortion due to cooling in the boundary layer before


sampling. (The thickness of the thermal boundary layer
is approximately equal to that of the momentum boundary
79
layer since Pr/^1), The average temperature drop
in the sampled gas due to boundary-layer cooling can
also be estimated from their results.
In MBRS sampling of flames, the free-jet
is skimmed along its centerline by the skimmer. Thus,
only a very small fraction of the gas that flous through

the sampling orifice ever reaches the detector. This


82
fraction can be uritten as follous
skimmer-orifice flou
= f(T)
sampling-orifice flou

Where d, = diameter of skimmer orifice, x, = distance

from sampling orifice to skimmer orifice, and f(7~)

is a function of specific-heat ratio. The ratio o^/x.


™T «~2
has values that range, from 10 " to 10 , and f(7")

81
has typically values of 1 or less. Thus the fraction

(50) is indeed small, and one may question the validity

of using the analysis by Hayhurst or Smith to predict

distortions in MBF1S sampling systems when measuring

neutral-species concentrations. That the use of aver-

age values is not justified for DBMS sampling of neutral


species is shoun by considering the influence of the

gas flowing along the sampling cone surface on the

gas flowing near the orifice centerline. The extent

of the forementioned influence can be estimated from

the ratio of tp, a characteristic time for flow through

the region in the vicinity of the sampling orifice

d*, to t_., the characteristic time for diffusion from


82
the uall to the centerline. Knuth ' has shoun that

the value of tp/tp is of the order 10—2 . This supports

the above argument. (The lack of information on probe-

induced distortion in MBHS sampling of 1 atm. premixed

flames has motivated a study uhich uas being carried

out at the U.C.L.A. molecular beam lab. at the time this

paper uas being uritten.)

82
3.3 The U.C.L.A. 1*181*13 Sampling System

The MBMS sampling system uhich uas used to

sample the premixed atmospheric methane-air flat flames

described here is depicted in Fig. 4. The background


—3 -1
pressure is of the order of 10 to 10 torr. in the
source chamber, 10~ to 10~' torr. in the collimating
—R —7
chamber, and 10 ' to 10 in the detector chamber. The

flame gas undergoes free-jet expansion through the

sonic nozzle, and only molecules flying along the cen-

ter-line of the system reach the detector.

The design of the U.C.L.A. MBMS sampling sys-

tem used here uas achieved by the considerations dis-

cussed in Section 3.2. The quartz sampling cone uas

selected on the basis of experiments, concurrent with

this study, carried out at the U.C.L.A. Molecular-Beam


Laboratory. It included a cone of total angle of 110°,

height of 50 mm., flat tip of 1.5 mm. of diameter,


orifice diameter of 0.27 mm., and a channel length

of 0.3 mm. More uill be said regarding sampline cones

in Chapter 5. The source chamber is pumped by a 16-


inch-diameter Stokes diffusion pump. The skimmer-to-

nozzle distance can be varied so as to maximize the

signal intensity at the detector. The brass skimmer

(orifice diameter 0.041 in.) transfers the core of

the free jet into the collimating chamber uhich is


pumped by means of a 10-inch diameter CUC diffusion pump.

The skimmer uas designed to minimize its interference

uith the beam. The skimmer internal half angle is 16°;


its external half-angle is 20°. The collimating ori-

fice (6 mm, diameter) can be closed by a valve. The

detector chamber is pumped by a 6-inch-diameter NRC-UHS

oil-diffusion pump. The baffle between the detector

chamber and its oil diffusion pump is refrigerated us-


ing Freon-12 to reduce background interference. The

mass spectrometer ionizer is placed uithin a liquid-

nitrogen cooled copper shroud uhich is itself attached

by means of a "cold finger" to a four-liter deuar (en-

closed by a vacuum uall) containing the liquid-nitro-

gen. The shroud reduces the background pressure in

the immediate vicinity of the mass spectrometer ionizer

by approximately a factor of five. The ionizer is

kept clean by constant heating uith a small coiled-tan-

talum-uire heater. The above features have alloued

detection of species in very lou concentrations. Con-

tamination of the detector-chamber walls can be reduced


by baking the walls externally.
The flat-flame burner has been designed and

its operation characterized by S, Saremi. Briefly,


it consists of a 2,30 inch disk (test section) of porous

brass, a burner housing, and a uater-cooling copper


line. Uniform mixing of fuel and air is achieved by

passing the mixture through a 15-foot teflon tubing

connected to the burner.

An EAI Quad. 250 mass spectrometer is used

for mass separation of the ions produced at the ionizer.

Its output is amplified by an electron multiplier oper-


ating ab -3kv. The signal is further amplified by an

100 mega-ohms, 15pf. pre-amplifier and fed into a lock-

in amplifier. The beam chopper in front of the colli-


mating orifice permits signal modulation. The fore-

mentioned lock-in amplifier is synchronized with the

beam chopper by means of a photodiode triggering system.

The lock-in amplifier allous time-averaging the modu-

lated signal from the pre-amplifier for periods up to

125 sec. The signal is finally fed into a channel

of an analog-to-digital converter connected to a LSI-11

computer uhich stores the data.

An effusive source consisting of a ceramic


tube uith an 0.040-in.-diameter orifice near its bot-

tom can be placed on or off the system centerline.


The effusive beam is used to calibrate the system and/or

to study fragmentation patterns of molecules as a func-

tion of temperature. The ceramic tube is heated by


means of a tantalum uire. A chromel-alumel thermocouple

is used to measure the temperature of the gas in the source

and a thermocouple gage to measure the pressure.

85
The source-gas temperature is measured by

the time-of-flight technique (TDF). For this purpose,

a TOP experimental set-up is placed along the system

centerline in the collimating chamber. This set-up

consists of an aluminum rod 25.6 cm. long uith a Bendix

magnetic strip electron multiplier detector at one end

and an electron-compact excitor at the other end.


The aim is to produce metastable species (mostly M^)

in the molecular beam by electron bombardment. The

ions in the beam are diverted by means of a small mag-

net attached to the aluminum rod before the detector.


Thus, the signal generated at the detector is produced

by the metastables alone. The electrons from the fila-

ment are collected by an anode connected to a uave gen-

erator uhich generates square uaves of uidth 1-5/Xsec.

at 1 msec, time intervals. The signal from the detector

is amplified by a pre-amplifier (100 mega-ohms,

15pf.) and then fed into a Fabritek signal-averaging


computer (model 1602) uhich includes a signal digitizer

and a sueep control unit. The output from the uave


generator is used to trigger the averaging procedure

of the Fabritck computer. The computer stores in mem-

ory the signal intensity as a function of time at


preset time intervals (4^Wsec.) and averages it. The

output can be displayed on an oscilloscope or recorded

on graph paper by a plotter.

86
CHAPTER 4

EXPERIMENTAL PROCEDURE AND DATA REDUCTION

4,1 Composition Measurements

Partial automation and recent addition of


a LSI-11 computer have simplified the collection of

experimental data and data-reduction. A servo-mechanism is


used to move the burner surface closer to or further

from the tip of the sampling cone. The burner can

be moved faster or slower as required by the experi-

ment. A voltage signal indicating the burner position

is sent to one channel of an analog-to-digital converter

of the computer. A total of 100 data points can be

taken in the axial direction of the burner in a span


of 0.6 cm. and stored in the LSI-11 computer. As men-

tioned in Section 3.3, the output from the electron

multiplier is amplified by a pre-amplifier and then


fed into a lock-in-amplifier for phase-sensitive analy-

sis. The net ion signal from the lock-in amplifier


is finally fed into a second channel of the analog-

to-digital converter of the LSI-11 computer and stored

in the computer memory. Computer programs were devel-

oped for this purpose.


The present detection system can resolve

87
species uhich differ by 1 amu. Species which differ

by less than 1 amu. e.g., CN (m/e = 26.0177) and C2H2

(m/e - 26.0378), cannot be resolved; their mass-spec-

trometric signals uill be detected at the same m/e

ratio, i.e., at m/e = 26 for the species in the above

example.

The signal intensity at a given mass-to-charge

ratio might be due to contributions of ions from stable

species, radical species, and fragments from a parent

molecule (fragmentation being a temperature dependent

process). Contributions from isotopes of rnajor species,


1 -Z TO
e.g., C H A , CO , also must be considered. The separa-

tion of these contributions to the signal intensity


at a given m/e ratio is facilitated in some instances

by the formation and decay of the different contribut-

ing species in different spatial regions of the flame,

and in some instances by differences in the electron

energy for the ionization processes (ionization, frag-

mentation) for different species. An extensive com-


pilation of ionization potentials (IP) and appearance

potentials (AP) of ions can be found in Ref. 88.


The electron energy scale uas calibrated

using the spectroscopically determined ionization po-

tential of Ar, 15.76 ev. For this purpose, several

ionization efficiency curves of Argon uere obtained.


One of these curves is reproduced in Fig. 8. The tail
4

6.0
•—»
09

5.0
*^
C

(5
i_
!5 4.0 —

0 •

M 3.0 -

^0
^N

^-» •

2.0 -

1.0 •

)) | - 1 • « 1 1 1 J

15 16 17 18 19 20
electron energy (ev) (uncorrected)

Fig. 8 . Intensity vs electron energy for Argon .

89
of the curve is due to the finite uidth of the elec-

tron energy distribution (about 1.5 - 2.0 ev). No ef-

fort uas made to reduce the uidth of this distribution.

A correction factor for the energy scale uas obtained

by comparing the electron energy at which the Argon

signal first appears uith the literature value of 15.76

ev.

The experiments uer e carried out at an elec-

tron energy uhich minimizes, in some cases, fragmenta-

tion of the molecule and contribution of other species.

This uas accomplished solely on the basis of appear-

ance and ionization potentials. Because of the uidth

of the electron energy distribution, the signal of

species that differ in IP or AP by less than 2 ev could

not be totally resolved (unless the species occurred

in different regions uithin the flame).

90
4.2 Temperature Measurements

The temperature of the flame gases uas de-


termined experimentally by TOF techniques. The exper-
imental set-up uas described in Section 3,3. In this
section, the analytical procedure used to determine
the temperature uill be described. The time-of-flight
for the metastable species uas determined from the
Fabritek computer output. To facilitate data reduc-
tion, the time of rise of the electronics uas deter-
mined and the time scale uas calibrated by means of
room temperature measurements.
The energy equation for adiabatic gas expan-
sion may be uritten

ho + U^
2 = hjL + u^2 ^^

uhere the enthalpy includes rotational, vibrational,

and translational contributions. For nitrogen, the

characteristic vibrational temperature is 3340 K so


that relatively little energy is stored in the vibra-

tional degrees of freedom at flame temperatures. Fur-


thermore, freezing of the vibrational modes is expected
to occur very early in the expansion. Thus, the en-

ergy equation reduces to

+ h
> < > =(

91
so that, uith freezing of rotational and translations!

degrees of freedom,

( c rot,o + c
tr,o + R)T
o = c
rot,oo Trot,eo
+ (C,tr,oo +R)!,.tr,<»3 -t-NU12 (53)
2
where the subscript <^> refers to frozen conditions,

Substituting for the values of the specific heats,


2
+ 3R + R\J = R T + 5 RT. + MU (54)
x
f) ' O rot, oo ft tr.0o 1 '
2

or the flame temperature TO


2
Ton = •=•
2 Trot,©o
r + is
5 T.tr,cx3 + FID, x(55)
1 '
7R

The most probable velocity of the molecules, U-,, is


determined from the TOF experiments. The terminal

rotational and translational temperatures, T


rot, oo
and T.
trjcxo respectively, uere estimated using the "sud-
den freeze" model described in Chapter 3. This proce-

dure is described in more detail in Refs. 74 and 82.

(Note that if no freezing of rotational and transla-

tional degrees of freedom occurs, then the energy equa-


2
tion reduces to T = P1U-, /7R).
o •*•
TOF temperatures measurements have the advan-

tage over other temperature measurements, e.g., ther-

mocouple measurements, that one is measuring the tem-

perature of gas at the same location and under the same

conditions as during the sampling experiment.

92
^•3 C.alibrat 1 on Procedures

The mass spectral signal intensity observed


for a particular ion, i, may be written

NXj^ (56)
Where n is a constant that relates the number den-
IP
sity of the parent neutral species in the ion source

to the number density in the flame, C. a constant that


includes gain factors and geometrical factors pertin-
ent to the ion source, 0*. the ionization cross section,
T. the transmission efficiency of ion i through the
mass filter, 'TP- number of secondary electrons ejected
at the conversion dynodes of the multiplier, N the

number density at the gas source (the flame in this


case), and X. the mole fraction of the species i.
In a methane-air flame, the mole fraction

of Argon remains approximately constant throughout


the different flame regions since it is an unreactive
species. Thus, the Argon signal uas used to normal-

ize all other species intensity signals, i.e.,

(57)
'Ar

uith X A knoun to good approximation, and C. ^ C Ar =

constant.
Calibration of the MBF1S sampling system en-

93
tailed experimental determination of the first tuo ratios
on the right-hand side of Egn. (57). For simplicity
i / Ar
the ratio n /n will be called (n )i, and the ra-
cS VJS i S

tio qT i Y i /0^ r T Ar >^ r uill be called ( n E F F ) i . Then,

(58)
Ar
The factor (ng)i includes effects such as pressure

diffusion in the free jet expansion, Mach-number-fo-

cusing, and skimmer interference. The factor (n^ppji


u
includes effects due only to the detection system.

Calibration of the ["IBMS sampling system for

a given species i can be done in two different manners.


In the first, a source gss of known composition sim-

ilar to the composition of the flame is sampled with

the F1BF1S system. This allows determination of the

product (n )i(r]r-pp)i since all other variables in Eq.

(58) are known. In the other method, an effusive source

is used to determine the detection system factor (hr-rr)i>

and then (n )i can be obtained from the product (h )i

(nrrrji or from a plot of (n )i vs. molecular weight.


rr
(The reader is referred to the discussion in Chapter 3,
where (n-)i is shown to depend largely on the molecular

weight of the species).


Stable species in the flame were calibrated

by the first method above, i.e., mixtures of known

94
composition uere put through the MBMS system and the

product (1)1 ( n r - ) i was determined. The effusive

source allowed the determination of (nrrr)i for the

species i; thus (n )i, and a plot of (n s )i vs. mole-


\P v/
cular weight could be obtained. The mole fraction
of stable species i in the flame was computed from
1 1
r^I^^
Ar i M_miI-Lm"JJI A,n _

where X n , and I./I. are also knoun.


H IT iL M 37

For radicals, the mole fraction uas determined as fol-


lows: First, the beam factor (n s )i uas determined from a
I
plot of (n )i vsi m.. Then, (nrpp\i was estimated from

the literature values for the total ionization cross-

sections of Argon and the species i, and from the sen-

sitivity of the detector T. T"j/T"nr Tfcr » which uas es-


timated from the experiments conducted uith the stable

species as a function of molecular ueight. (TJote that


s a
lso a function of the molecular structure.
Ar
This introduces some uncertainty in this calibration

procedure). The radical mole fraction is then given

by, f
• *i
yi-^x
X
Ar (60)

The uncertainty in this procedure is approximately a

factor of 2.

95
CHAPTER 5

EXPERIMENTAL RESULTS

The measurements were made using a room-tem-

perature, one-atmosphere, premixed flat methane-air

flame of equivalence ratio 1.37. The flou rates uere

I
kept at 5.12 SCFH for methane and 35.6 SCFH for air.

Air was used as the shroud gas in order to minimize


the effect of buoyancy in the post-flame gases. The

flou rate of this shroud gas was set to minimize the

deformation of the edge of the flat flame.

The MBMS sampling system uas calibrated di-


rectly for the stable species CH^, 02, CQ-2, NH 3 , HCN

and NO. For other species, eg, radical species, the

calibration procedure described in Section 4.3 uas

used.

As mentioned before, the species profiles

uere taken uith the burner moving auay from the sampl-
ing cone. The electron energies uere selected to min-

imize fragmentation of the parent molecules, and to

resolve the various contributions at a given m/e ratio.

96
5.1 Temperature Measurements in the Flame

A temperature profile for the flame with


conditions as described above uas measured using the
TOF techniques. Fig. 9 shows this profile when a probe
of total angle 80°, orifice diameter 0.24 mm., channel
length 1.1 mm., and tip diameter of 2.0 mm. is used.
The post-reaction zone temperature corresponds approx-
imately to the adiabatic flame temperature.

97
2000
: :•

1500

2
o
a
o
H

1000

(b 11.37
3-0.24MM
*~
TIPDIA.I2.0MM
500
j_ _L
1 2 3 4 5
Distance (mm)

Fig. 9 . Temperature profile measured using


TOF technique .

98
During experiments, visible attachment of

the flame to the quartz sampling cone uas observed.

The attachment persisted up to a distance of about

0.35 cm. from the burner surface (for the above cone)

at which point the flame could be seen to detach itself


from the cone tip. Furthermore, when the cone uas in

its closest position to the burner surface (0.013 cm.),


the flame surrounding the cone uithin a radius depend-

ent on cone geometry showed discoloration.

Temperature profiles for cones of different


geometrical parameters (total angle, tip diameter, ori-

fice diameter, and channel length) can give useful

information regarding the effects of the forementioned

parameters on the flame. Fig. 10 depicts the tempera-

ture profile obtained uith a quartz cone of the same

total angle as the corresponding to Fig. 9 but uith a

tip diameter of 0.5 mm., an orifice diameter of 0.30

mm., and a channel length of 1.4 mm.

The following observations can be made from

Figs. 9 and 10. The uidth of the primary reaction zone


(defined as the region between the steepest temperature

gradient and the post-flame temperature) for the mea-


surements made with the cone of the larger tip diameter

is approximately 0.5 mm. larger than that observed

from the measurements made with the smaller tip dia-


meter. Furthermore, the slope of the temperature pro-

99
2000
9
e

1500

I
*-
CO

I 1000

e
$-1.37

o
TIPDIAZ0.5MM
0
500

1 2 3 4 5
Distance (mm)

Fig. 10. Temperature profile measured using


TOF technique .

100
file for measurements made uith the cone of smaller

tip diameter is greater. Note, houever, that the post-

reaction-zone temperature is approximately the same for

both cones.

101
Observations of the cones during the temper-

ature measurements shoued that for the cone uith the

smaller tip diameter the surface area of the cone that

uould glou red uas limited to the very tip of the cone

in contrast to the cone uith the larger tip diameter


for uhich a larger surface area uas observed to glou

red. Furthermore, uith the cone in the closest posi-

tion to the burner, the forementioned discoloration

of the flame uas least for the cone uith smaller tip

diameter. (The diameter of this region uas approxi-

mately 0.5-1.0 cm. for the larger tip.) Also, the

detachment of the flame from the cone surface occurred

approximately 0.5 mm. earlier for the smaller tip dia-

meter than for the larger tip diameter as the burner

moved auay from the cone tip.

Further insight into probe-induced distortions

of the flame is provided by the intensity profiles of


C02 and 02 measured uith the tuo different quartz cones.

These are shoun in Figs. 11-14. The effect of the larger

tip diameter is to reduce the measured rate of decay


of 02 and rate of formation of C02 uithin the reaction

zone of the flame. (The intensity profiles of C02 and

02 as uell as all other species intensity profiles,

are normalized by the Ar signal to eliminate the ef-

fects of temperature variation uithin the flame. For

this flame, the Argon mole fraction is approximately

102
8134 ppm.)
In conclusion, for a flame uith conditions
as described here, a cone uith a smaller tip diameter
uill reduce the extent of flame attachment to the quartz
cone. Attachment of the flame to the sampling cone
results in a decrease in the measured slopes of the
concentration profiles and temperature gradient uith—
in the primary reaction zone.

103
j

3D — O.oOmamp

o
Tio
^ 20
CM
co
tl
o

.0

2 3 4 5 6
Distance (mm)
Fig. 11. Normalized Oxygen profile ( tip dia-2.0 mm
dx=0.24 mm).

104
3.0

| zO.30 mamp
II
O
X

•5 2.0
OJ
CO
II
O

.0

i 2 3 4
Distance (mm)

Fig. 12. Normafized Oxygen profile (tip diazO.5 mm,


d ~ 0.30 mm ) .

105
I.!
(ja.37
»e —0.30mamp

\8 I.O

II
0

0.5

2 3 4
Distance (mm)

Fig. 13. CO2 Normalized profile ( tip diaz 2.0mm,


djp 0.24mm ) .

106
.5
Ee 120.50 ev
—0.30mamp

- 1.0
tl
o

II
O

! 0.!

2 3 4
Distance (mm)

Fig. 14. Normalized CC^profile (tip diaz0.5mm


)

n?
The effect of channel length and cone angle
have been studied by Yoon and Knuth 82
' in a study sub-

sequent to the preliminary experiments conducted in

this paper. As a result of their uork, the cone of to-


tal angle 110 , cone tip diameter 1,5 mm,, orifice dia-

meter 0,27 mm,, and channel length p.3 mm, was selected

as that which minimizes probe-induced distortions of


the three cones described here. (The shorter channel

length uas observed by Yoon and Knuth to minimize chem-

ical relaxation effects),

Yoon and Knuth found also that the properties


of the sampled gas for a stoichiometric flame are those

of the gas about five orifice diameters upstream from


the probe tip. Thus, temperature and concentration

profiles shown here and in the chapters to follow should

be shifted upstream by a distance about five tiroes the

orifice diameter of the probe.

Note that the use of a FIBMS sampling system

to obtain concentration profiles and flame temperatures

makes the data so obtained more self-consistent than


concentration profiles and temperatures obtained by

two different experimental techniques, e.g., WBFIS sam-

pling system for sampling of the flame, and a thermo-


couple to measure temperatures. Thus, temperature

and concentration profiles obtained by FIBFIS sampling

108
techniques makes data evaluation an easier task,

109
5.2 Stable Species Profiles

Fig. 15 depicts the mole fraction profiles


a
for the stable species CI-L, CU* nd CCU* Methane and
Oxygen are totally consumed before the end of the reac-

tion zone. The post flame mole fraction of carbon di-

oxide is near its equilibrium value of 5% (see Ref.

19).

110
TTT

Mole fraction

o p p
o
*

§
oo
£
ro o
o 3 1 i
Z (2
1
en
om o
a «y H Sr t / O
3 O ? V I*m M
* ^ o
r\i- of-*"
< 1« .-• *. • • M
ro
3
a •
*





p
O <
* .•
b
O •
o>
0
erOJh *

3 £ •



* •


a>
^t
-t
o
1 ««

2 o •©-
M

o
3
1 , 1 1
: 0.40mamp
; 1.37

•o^ m. *
en-
i

p "*•.. o
(0
o
0
®
i

I . I , ! i

o p p O O
0 O O b b
-^ Cn

Mole fraction
5.3 Formation of Nitric-Oxide in the Flame

The concentration profiles for NO, NH3, HCN


are shoun in Fig. 16, 17, and 18, respectively. The
NO (l.P. 9.25 ev)profile uas taken at an electron energy
of 13.56 ev; at this lou energy lev/el, contributions
of N15N15 and C12018 are effectively eliminated. In
the flame region uhere NO is observed to form, one
could expect a significant contribution from C2Hg.
Fig. 19 shous the relative intensity profile for m/e=30
at 13.56 ev. The first peak is attributed to HoCO,
and the second peak to C2Hg. The C2Hg in the flame
is all consumed at a distance of 0.45 cm. from the burner
surface, thus making no contribution to the NO peak.
(The first peak approximately 0.10 cm. is due to an
ethane impurity, 0.12% in the methane and in this ex-
perimental uork). Further support for the H2CO assign-
ment to the first peak can be found in Fig. 6 of Ref.
13.
The HCN (l.P. 13.73 ev) profile uas obtained
at an electron energy of 14.76 ev. In the region 0.35-
0.45 cm. from the zero burner position, the relative
contributions of HCN and C2H3 (uhich might include
contributions from other C2-hydrocarbons) could not
be resolved. Houever, by extrapolation of the portion
of the HCN profile to uhere there is no hydrocarbon

112
contribution, the onset of HCN formation can be seen
to occur at approximately 0.35 cm. from the burner
surface.

113
Concentration (ppm)

en O en
O O O
3 1 , . . . , - - . , . ,
*P

P Ps —
II I 1 I'
O -I -*•
z 4, 0 b
0 O Oi "^
o 3 J
0
2 o * * • 3
Z « en " • * -Q
2.
3
«
3
/
t •
!• •

£1 °
3' ,-s
*
* •

3 3 en ^ *

* *
»

3 ~
*

*
*

* —_ •

* •

CO ^m • *

*
Concentration (ppm)

cn O
0 0
Z! 1 1 1
*> J
M
<*-•
3"
/^-"
*i 44*

CO
O
o
o 0 ,, u
.•
3 S ^

c_n
=r
5 ^$

. •*

r4- O
5" fl»

•So f^

.
|HK
^
gr
^-JT1"0"
o> II ii M
_» —A
P CO CO

-Nl
?
3 °>
(D
w <
•o
Concentration (ppm)

cn O

•n O O

CO

I
0
/
OJr-

o
o
3 D
O
cn 0) 5T
r* ST
s 3
O
O
5*
3
T>
O^

o O
11 1 1
CO
-xl o O)
3 (D
0)

•o
C
2H6 (j) Z1-37
0.5
E^H 13.56ev

C5 !e z 0.
O 0.4 H 2 CO

II 0.3
O

s
O
CO
02 NO
11
O

* 0.

2 3 4
Distance (mm)

Fig. 19. Normalized m/er30 profile.

117
The concentration profile for NH-,
O
(I.P. 10.19ev)

uas obtained at an electron energy of 13.16 ev. In the

region uhere NHU forms, contributions from H90 frag-


mentation and OH radical species are possible, Hou-
ever, at 13.16 ev, fragmentation of H20 to OH (A.P.
18.3 ev) would be negligible even at flame temperatures.
89
The uork of Milne and Greene supports this assumption.
The hydroxyl radical (I.P. 13.17) might possibly con-

tribute to the peak at this electron energy. The rela-


tive intensity profile for m/e - 17 for the entire
region 0.0 - 0.6 cm. is shown in Fig. 20. Since OH

contribution is limited to the region 0.20-0.40 cm.


from the burner surface, the peak at about 0.55 cm.
is solely due to NH^j.
The concentration profiles for NO, NHg, and HCN have
not been corrected for fragmentation of the parent
molecules. Fragmentation of [10 is not likely to occur

until an electron energy of 20 ev is reached. There-


fore, at 13.56 ev, NO fragmentation uould be negligible
at flame temperatures. The A.P. of CN from fragmenta-

tion of HCN (I.P. 13.73 ev) is quoted to be 19.36 ev90.


Thus, HCN fragmentation at 14,76 ev is likely to be neg-
ligible even at flame temperatures. The absolute par-
tial ionization cross sections for Ammonia by electron
impact, from the threshold energy up to 180 ev, have
91
been determined by Mark et al . At 19 ev, this partial

118
CO 0.4 OH
o <0 Z1.37
T™"
*
x *

O
E e - 13 - 16e v
II
0.3 je Z0.40mamp
o

p 02 NH3
Tic •

4
v^ *
E
0.1 — •

F • ' i. •!• • i i i
I 2 3. 4 6
Distance(mm)

Fig. 20. Normalized m/ez17 profile.

119
ionization cross section for fragmentation of NH X to
O

I\IH2 is about 21% of the total ionization cross section


of NH. At and belou 19 ev, only ionization of MH,
occurs. Thus, IMHU - fragmentation in the flame is
not likely to be important if the electron energy is
kept lou, and at 13.16 ev is believed to be negligible,

120
The relative intensity signal for m/e=43

is shoun in Fig. 21, The electron energy uas 20.5 ev.

The signal in the region 0-0.20 cm. can be attributed

to propane fragmentation, present as an impurity in the

methane gas. The peak at approximately 0.25 cm. is

most likely C^H--,. The last peak at 0.55 cm. from the

burner surface is attributed to HMCO (I,P. 11.60 ev).

The assignment is based upon the fact that HCM is also


seen to form and decay at this distance from the burner

surface, and HCN decay is likely to lead to the forma-

tion of HNCO. Measurements made at 11.7 ev further

support the assignment of the second peak to C-7H


O i7
.

Houever, the peak attributed to HNCO uas not aluays

reproduced. Direct calibration of HNCO uas not carried


out, but its peak mole fraction has been estimated from

Fig. 21 to be approximately 5 ppm.

The results of the experiments conducted

in this study support the mechanism by uhich the onset

of HCN formation preceeds the formation of "prompt

NO," and HCN decay results in the formation of the "prompt


NO" near the reaction zone. Furthermore, PJH-, is observed

as the intermediate product of HCN decay. The rapid

build-up of Ammonia in the post-reaction zone, houever,

seems to indicate that the formation of NO and N 2 from

NH- species does not take place quickly compared uith

the rate of HCN disappearance at a flame temperature

121
of 2000°k and ^ = 1.37. HNCO seems to play no import-

ant role in the decay of HCN since its concentration

in the flame is so lou.

122
(J) m.37
Ee- 20,5ev
CO .0 I e H 0.30 mamp
o
T-
X

|
C
3H7

TiO 0.5
HNCO

_• L
2 3 4
Distance (mm)

Fig. 21, Normalized m/e=43 profile.

123
5.4 C2H2, C2Hg Profiles

The concentration profile for acetylene in

the flame is presented in Fig. 22. The profile uas

obtained at an electron energy of 17.7ev. Since the

system uas not calibrated directly for C2H2, the cross


section data of Tate and Smith 95 uere used to obtain
the mole fraction. This profile uas measured uith the

quartz cone of 80 total angle and 2.0 mm. tip dia-

meter. The peak concentration of C2H2 in the flame

is approximately 2100 ppm. This concentration level


38
is in agreement uith the results of Harvey and Maccoll

There is not sufficient data in the literature regard-

ing the fragmentation of C2H2 and C 2 H- to calculate

the importance of each at flame temperatures. Houever,

as mentioned above, the peak concentration of C-Ho

obtained here is in agreement uith the measurements

of Harvey and flaccoll.


Fig. 23 shous the relative intensity profile

for m/e = 27 at 20.50ev. The HCI\I contribution in

the post-reaction zone can be seen. This profile uas


also obtained uith the cone of 80° total angle, and

2.0 mm. diameter (cf. Section 5.1). Uith the burner

positioned at the point uhere the maximum signal inten-

sity uas observed, an intensity vs. electron energy

124

-
profile uas measured. The signal uas observed to van-

ish at approximately 14.5 ev. Thus, C2H3 (l.P. 9.45ev)

concentrations in the flame are likely to be negligible

relative to those for C2H., and the signal presented

in Fig. 23 is mostly due to the fragmentation of C2H.

species.

125
Concentration (ppm)

— __ rx>
Cn O Cn O
O O O O
o .0 o o
•n 1 1 1 1
iQ*

to
10

O —
10
^om"e"
X
10 II 41 II
CTi
o 0 -* ^
*. %»
• w -Nj
. W
rj
H
3 0 |\J ~ *-. o -^
g o
O
fl>
(ft
l-f ....

» «
«J -^

ao sS~GJ
3- ' . .
• .

* •

3 3 ' .

so a-
~ -^^
' • .

^» •
*

O* [S. —. » *

...-•'



.*

</l

.'
"""•
C
• 2HS

0.03
(j)Z 1.37
E e Z 20.50ev

[e ~0,30mamp

8 0.02
E

HCN
OJ
n
o

O.Ol-

2 3 4
Distance(mm)

Fig. 23. Normalized m/ez27 profile .

127
Figures 19 and 24 shou the relative inten-
sity profiles for m/e = 30 at 13.56 and 12.10 ev. re-
spectively. As was mentioned in Section 5.2, the second
peak is assigned to C9£•H,-
o.
Fragmentation and cross-
section data for C2Hg at lou electron energies are
not available in the literature. Thus, an upper-bound
on the concentration of C0£.H,-b was estimated from the
cross section data of Ref.92 and the fragmentation data
of Berry and Osberghaus 93 . The maximum concentration
of tfle was estimated to be 200 ppm. The uncertainty
uas estimated to be a factor of 2.

128
r 1.37
r I2.10ev
IT O.SOrnamp

1.0
03
O
T-
X
H2CO
O
4

0.5
C
2H6 NO

»• 4 ••

I .

2 3 4
Distance(mm)

Fig. 24. Normalized rn/ezSO profile .

129
5.5 C»i and H2CO Species in the Flame

The system uas not calibrated directly for

the species C, CH, CH 2 « The cross sections uere es-

timated by additive methods from the data of Rapp and


94
Kieffer . The uncertainty in these mole fractions
is estimated to be a factor of 2. The electron energy

uas kept low (about 2-3 ev. above the I.P. of the spe-

cies) to avoid fragmentation of the molecule by elec-


tron impact. The concentration profiles for C, CH,

CH2 are shoun in Figs. 25 - 27. Fragmentation of ChL

to C, CH, and ChU uas observed to occur prior to the


formation of each of these species. Thus, the peaks

identified as C, CH, and CH2 have no methane contribu-

tion. As mentioned earlier, the cross section for

ionization of each of the CH- species uas estimated

by additive methods. It is believed that the cross

sections estimated here are likely to be larger than


the actual cross sections; hence, this estimate uill

lead to lower mole fractions of the species in the

flame.

130
Concentration (ppm)

Cn O cn O
O '-O O O
•n
•P
10
01

IO
O


O
•g 11 H II
O -i
o
^I•*•S
I %
TO

cn
Concentration (ppm)

__ rv> GJ -PS en
0 0 O O 0
O O, O O 0
•n
MAI
i 1i ' ii • ii i ti i ii

0) i- jn -e-
"""" ' ii • i
o
3C O -i -*
O V J^ •
K) O O b NI
3
S
3 0
ru ~~ ft) <D
3 <
•o
§. W
I" «
_• •
TJ —»
21
o 3
3 • ,
~3 3 , •
<§* "^

9
* •

« 0 .*

Cn
Concentration (ppm)

o
O o o O O O o o o o
o o o o o o o o o o
•n
G5'
i r ~r

ro
•-4 -o-
o
X
it H
00
P w o
OJ O ><J1
O o
W] O)
3 3 rt>
O o <
3
U
a
Q)
ts

O
3
TJ

cn
3.0

(t) = 1-37
E Z 13.30ev
rt
o le — O.SOmamp
2.0
o

5
~
to
II
o
CH

\
.0

2 3
Distance (mm)

Fig. 28. Normalized rn/en15 profile.

134
The relative intensity profiles for H?CO
and CH|7 are shoun in Figs, 24 and 28 respectively.
These profiles uere taken uith the 80° total angle
cone of 2.0 mm. tip diameter. No effort uas made to
estimate the absolute concentrations of these species.
From Fig. 28, it can be observed that the CH, contribu-
tion cannot be totally eliminated even at lou electron
energies. (This profile uas obtained at 13.3 ev.)
From Fig. 24 the H2CO formation is seen to occur prior
to ChU formation. This profile uas obtained at an elec-
tron energy of 12.10 ev. The disturbance by the probe
in this region of the flame for the cone used in this
measurement is likely to be large (see. Ref.82), so
that the uncertainty in the mole fraction that one uould
estimate also uould be large.

135
CHAPTER 6

CONCLUSIONS

The experiments show conclusively that,near

the primary reaction zone of fuel-rich flames, NO forma-

tion occurs by a mechanism other than the Zeldovich

mechanism. The amount of NO formed via the Zeldovich

mechanism, taking into account radical overshoot near


2R
the reaction zone, uas predicted by Ay and Sichel in

their computer analysis. From their Fig. 4, the NO

formed by the foreraentioned mechanism is approximately

10 ppm. The experimental NO concentration in the flame

after all HCN and NH3 in the flame have decayed uas

determined in the study conducted here, to be approx-

imately 56 ppm. Furthermore, the rate of NO forma-

tion uas observed to be largest near the reaction zone

of the flame, prior to the decay of HCN, leveling off


after all the HCN and NH 3 have been consumed. Ammonia

uas observed in the post-reaction zone of the 0 = 1.37,


CH./Air flame as a consequence of HCN decay. The

rapid build-up of NH3 (peak mole fraction of 44 ppm.)


suggests that the decay of NH. species, interconverted

by reactions of the type,

NH X + H £ NH X-1 + H2 (27)

136
x = 1, 2, 3
via reactions
N + OH £ NO + H (3)
NH2 + NO £ N2 + H20 (28)
NH + NO £ N2 + OH (29)
N + NO £ N2 + 0 (30)
does not occur rapidly at flame temperatures.
The onset of formation of HCN occurs prior
to that of NO and approximately at the same time as
NH- formation. Houever, the decay of HCN is too slou
for it to be consumed completely before the onset
of NO formation. The flame uas sampled for HNCO.
Its peak mole fraction uas estimated to be approxi-
o fi
mately 5 ppm. This indicates, as suggested by Haynes ,
that HNCO decays by extremely rapid reactions, and
that the initial weakening of the C-N bond in HCN
to form the OCN - species is the rate limiting step
in HCN oxidation.
Concentration profiles uere obtained for
the species C, CH, CH~. The maximum mole fractions
of C, CH, and CH2 are approximately 206, 481, and
908 ppm., respectively. The most likely reactions
leading to the formation of HCN are
C + N2 £ CN + N (12)
CH + N2 £ HCN + N ( 7)
CH3 + N 2 £ HCN + NH (13)

137
The rate constants of Reactions (7) and (13)
oc
were calculated by Peeters et al. from experimental
data at 1820°K. Their values are given by
k? = 8 x ID11 exp (-11,000 /RT)
k13 = 2.8 x 1012 exp (-22,500 /RT)
in units of mole -1 cm3 sec-1 , uith a probable error
of a factor of 2. It is likely that the formation of
HCN can be explained by Reactions (7) and (13), uith
Reaction (13) dominating because of the larger concen-
tration of ChU in the flame. Reaction (12) uould also
contribute, but to a lesser extent.
C^-hydrocarbons uere observed to form and
decay mostly uithin the primary reaction zone of the
mo
flame. The peak £^2 ^e fraction uas estimated to
be 2100 ppm., uncertain by a factor of 2. The C2-hy-
drocarbons provide additional paths for the consump-
tion of the methyl radical uhich contribute to keep-
ing the ratio of CO molecules to hUO molecules formed
from CH* oxidation equal to 0.5. Thus, this study
further indicates that C2 chemistry has to be consi-
dered in the kinetic modeling of methane oxidation
in fuel-rich flames.
The experimental data obtained from the flame
gives insight into the mechanism of CH,-oxidation in
fuel-rich flames. The formation of the methyl radical

138
does not reach significant rates until most of the for-

maldehyde in the flame has been consumed, (The hUCO-

forming reaction(s) keep the CHU concentration level

very low). Furthermore, the HUCO profile attains a

maximum before the primary reaction zone of the flame

is reached. The lou 0-atora concentrations in the re-

gion uhere FUCO rapidly forms suggests that the reac-

tion of CH3 uith oxygen is the main path for H~CO forma-

tion.
Visible attachment of the flame to quartz

cones of total angles 80°, and 110° uas observed in

the course of these experiments, for a flame of equiva-

lence ratio 0 = 1.37. The attachment to the cone per-

sisted throughout the primary reaction zone of the flame

and resulted in a decrease of both temperature and

concentration gradients uithin the primary reaction

zone. Quartz sampling cones of small tip diameters

reduce the attachment effects.

139
REFERENCES

1. Zeldovich, Ya., B., Sadovnikov, P. Ya., and D.A.


Frank-Kamonetskii, "Oxidation of Nitrogen in Com-
bustion," Academy of Sciences of USSR, Institute
of Chemical Physics, Floscou-Leningrad (trans.
by PI. Shelef), 1947.
2. Lavoie, G.A., Heyuood, D.B., and Keck. 3.C., Com-
bustion Science and Technology 1:313 (1970).
3. Uestenberg, A.A., Combustion Science and Technol-
ogy 4:59 (1971).
4. Baulch, D.L., Drysdale, D.D,, Home, D.G., and
Lloyd A.C., "Evaluated Kinetic Data for High
Temperature Reactions," Vol. 2, CRC Press, 1973.
5. Bouman, C.T., "Kinetics of Pollutant Formation
and Destruction in Combustion," Prog. Energy Com-
bust. Sci., Vol. 1, Pergamon Press, Great Britain,
1973, pp. 33-45.
6. Neuhall, H.K., and Shaded, S.H., "Kinetics of
Nitric-Oxide Formation in High-Pressure Flames,"
Thirteenth Symposium (international) on Combus-
tiont The Combustion Institute, Pittsburgh, Pa.,
1971, pp. 381-389.
7. Livesey, 3.B., Roberts, A»L., and Williams, A.,
Combustion Science and Technology. 4:9 (1971).
8. Fenimore, C.P., "Formation of Nitric Oxide in
Premixed Hydrocarbon Flames," Thirteenth Sympo-
sium ^International) on Combustion. The Combus-
tion Institute, Pittsburgh, Pa., 1971, pp. 373-
380.
9. Buleuicsz, E.FI«, 3ames, C.G., and Sugden, T.N.,
Proc. Roy. Soc. (London). A235:89 (1956).
10. Kaskon, U.E., Combustion and Flame. 2:286 (1958).
11. Sarafim, A.F., Pohl, 3.H., "Kinetics of Nitric
Oxide Formation in Premixed Laminar Flames," F^our-
teenth Symposium (international) on Combustion.
The Combustion Institute, Pittsburgh, Pa., 1973,
pp. 739-754.

140
12. Bowman, C.T., Combustion Science and Technology,
3:37 (1971).

13. Gay, R.L., Young, U.S., and Knuth, E.L., Combus-


tion and Flame. 24:391 (1975).

14. Iverack, D., Basden, K.S., and Kirov, N.Y., "Forma-


tion of Nitric Oxide in Fuel-Lean and Fuel-Rich
Flames," Fourteenth Symposium (international)
on Combustion. The Combustion Institute, Pitts-
burth, Pa., 1973, pp. 767-775.

15. Hayhurst, A.N., and McLean, H.G., Nature. 251:303


(1974).

16. Plorley, C., Combustion and Flame. 27:189 (1976).


17. Miyauchi, T., Mori, Y., and Imamura, A., "A Study
of Nitric Oxide Formation in Fuel-Rich Hydrocarbon
Flames: Role of Cyanide Species, H, OH, and 0,"
Sixteenth Symposium (international) on Combustion,
The Combustion Institute, Pittsburgh, Pa., 1976,
pp. 1073-1081.

18. Saltzman, D.E., "Colorimetric Microdetermination


of Nitrogen Dioxide in the Atmosphere, Analytical
Chemistry, 26:1949 (1954).

19. England, C», "Quantitative Evaluation of Reduc-


tion Techniques for Oxides of Nitrogen," Jet Pro-
pulsion Lab., Calif. Inst. of Technology, Pasa-
dena, Calif., 1972, Report 1200-37.
20. Fristrom, P.M., and Uestenberg , A.A., Flame Struc-
ture. MacGrau-Hill, New York, 1965.
21. Bouman, C.T., and Seery, D.J., "Investigations
of NO Formation Kinetics in Combustion Processes:
The Methane-Oxygen-Nitrogen Reaction," Emissions
from Continuous Combustion Systems, U. Cornelium
and U.G. Agneu eds., Plenum Press, Neu York, 1971,
pp. 123-139.
22. Haynes, B.S., Iverach, D., Kirov, N.Y., "The Be-
havior of Nitrogen Species in Fuel-Rich Hydrocar-
bon Flames," Fifteenth Symposium (International)
on Combustion, The Combustion Institute, Pitts-
burth, Pa., 1975, pp.1103-1112.
23. Halstead, C.J., Combustion and Flame, 11:362 (1970).

141
24. Zeegers, P.3., and Alkemade, C.T., Combustion
and Flame, 15:193 (1970).

25. Peeters, 3., Blauuens, 3., and Smets, B., "Mech-


anism of "Prompt" NO Formation in Hydrocarbon
Flames," Sixteenth Symposium (international) on
Combustion, The Combustion Institute, Pittsburgh,
Pa., 1976, pp". 1055-1062.
26. Haynes, B.S., Combustion and Flame. 28:113 (1977).
27. Peeters, 3., and Mahnen, G., "Reaction Mechanisms
and Rote Constants of Elementary Steps in Methane-
Oxygen Flames," Fourteenth Symposium (internation-
al) on Combustion, Combustion Institute, Pittsburgh,
Pa., 1972, p. 133.
28. Ay, 3.H., and Sichel, M», Combustion and Flame,
26:1 (1976).
29. Bachmaier, F., Eberius, K.H., and 3ust, Th., Com-
bustion Science and Technology, 7:77 (1973).
30. Cooke, D.F., and Williams, A., "Shock Tube Stud-
ies of the Ignition and Combustion of Ethane and
Slightly Rich Methane Mixtures uith Oxygen," Thir-
teenth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, Pa., 1971,
p. 757.
31. Bowman, C.T., "Non-Equilibrium Radical Concentra-
tions in Shock-Initiated Methane Oxidation," Fif-
teenth Symposium (International) on Combustion,_
The Combustion Institute, Pittsburgh, Pa., 1974,
p. 869.
32. Heffington, U.M., Pork, G.E., Sulzmann, K.G.P.,
and Penner, S,S., "Studies of Methane-Oxidation
Kinetics," Sixteenth Symposium (international)
on Combustion, The Combustion Institute, Pitts-
burth, Pa., 1976, p. 133,
33. Brabbs, T.A., and Brokau, R.S., "Shock Tube Mea-
surements of Specific Reaction Rates in the Branched
Coin CH4-CO-02 System," Fifteenth Symposium (in-
ternational on Combustion, The Combustion Insti-
tute, Pittsburgh, Pa., 1974, 893.
34.
t. Uestbrook, C., Creighton, 3., Lund, C
Uestbrook, C., and Dryer,
n. ,-.1 n i . o c n n f 1 nT~>\
F.L., 3. Phvs. Chem.. 81:2600 (1977).

142
35. Engleman, U.S., "Survey and Evaluation of Kinetic
Data on Reactions in Methane/Air Combustion,"
EPA-600/2-76-003.
36. Bonni, A.A., and Penner, R.C., Combustion Science
and Technology. 15:99 (1977).

37. Gardiner, U.C.. and Olson, D.B., Combustion and


Flame. 32:151 (1978).

38. Harvey, R., and Flaccoll, A., "The Formation of


C2 Hydrocarbons Uithin Methane-Oxygen Flames,"
Seventeenth Symposium (international) on Combus-
t ion« The Combustion Institute, Pittsburgh, Pa.,
1978, p. 857.

39. Tabayashi, K.. and Bauer, S.H., Combustion and


Flame. 34:63 (1979).

40. Knuth, E.L,, "Direct Sampling Studies of Combus-


tion Processes," Engine Emissions; Pollutant
Formation and Measurements, ed. by G.S. Springer
and D.3. Patterson, Plenum, Neu York, 1973, pp.
319-363.

41. Gaydon, A.G., "The Use of Shock Tubes for Study-


ing Fundamental Combustion Processes," Eleventh
Symposium (international) on Combustion. The Com-
bustion Institute, Pittsburgh, Pa., 1967, p. 1.
42. England, C., "Quantitative Evaluation of Reduc-
tion Techniques for Oxides of Nitrogen," Jet Pro-
pulsion Lab., Pasadena, Calif., Report 1200-37,
April 1972.

43. Malta, P.C., Kramlich, 3., Benedict R., Singh,


S., Pratt, D.T., and Robertus, R.3., "Combustion
and Pollutant Kinetic Modeling for Methane, Meth-
anol, Fuel-Sulfur, and Fuel-Nitrogen," USS/CI Pa-
per 75-19, Palo Alto, Western States Section/The
Combustion Institute Fall Meeting, 1975.
44. Allen, 3.D., Combustion and Flame. 24:133 (1975).

45. Morley, C., Uacuum. 24:581 (1974).


46. Sherman, F.S., and Ashkenas, H., "The Structure
and Utilization of Supersonic Free Gets in Lou
Density Uind Tunnels," Proceedings of the Fourth
International Symposium on Rarefied Gas Dynamics,
ed. by 3.H. Leeuu, Vol. II, Academic Press, Neu

143
York, 1966, p. 84.

47. Ouen, P.L., and Thornhill, C.K., "The Flow In


an Axially Symmetric Supersonic 3et from a Nearly-
Sonic Orifice into a Vacuum," Aeronaut. Res.
Council, Great Britain, RM 2616, 1963.

48. Anderson, 3.B., AIAA 3.. 10:112 (1972).

49. Anderson, 3.B,, "Molecular Beams from Nozzle Sources?


Molecular Beams and Lou Density Gas Dynamics,
ed. by P.O. Uegner, Marcel Dekker Inc., New York,
1974, Chap. I.

50. Knuth, E.L., "Rotational and Translational Relax-


ation Effects in Lou-Density Hypersonic Free
Gets," Engineering Dept., University of Calif-
ornia, Los Angeles, Report 64-53, 1964.

51. Tang, S.P., and Fenn, 3.B., AIAA 3.. 16:41 (1978).

52. Bier, K., and Schmidt, B., Z. anqeu. Math. Phys..


13:493 (1961).

53. Anderson, 3.B., AICHE3, 13:1188 (1967).

54. Sharma, P.K., "Energy Transfer in Free-3et Flows


with Applications to Sampling Studies of Combus-
tion Systems," Ph.D. Thesis, School of Engineer-
ing and Applied Science, U.C.L.A., 1976, p. 27.

55. Hayhurst, A.N., and Telford, N.R., Proc. Roy


Soc., A 322:483 (1971).
56. Bray, K.N.C., "Chemical and Vibrational Nonequil-
ibrium in Nozzle Flows," Nonequilibrium Flows, ed.
by P.P. Uegener, Part II, Marcel Dekker, Inc.,
New York, 1970, p. 59.
57. Phinney, R., "Mathematical Nature of the Freezing
Point in an Expanding Flow," Martin Marietta Co.,
RM-122, Baltimore, 1964.

58. Young, U.S., AIAA 3.. 13:1478 (1975).


59. Becker, E.U., Bier, K., and Henkes U., 2. Physik,
146:333 (1956).
60. Cassanova, R.A., and Stephenson, U.B., "Expan-
sion of a 3et into a Uacuum," Eleventh Symposium
(Interna ti o nal) o n Co mbustion. The Combustion
Institute, Pittsburgh, Pa., 1967, p. 577.

61. Good, R.E., and Golomb, D», "Dimerization in


Free 3ets, First Step Toward Condensation," Pro-
ceedings of the Eighth International Symposium
on jterafied Gas Dynamics, ed. by K. Karamcheti,
Academic Press, New York, 1974, p. 87,

62. Milne, T.A., Vandergrift, A.E., and Greene, T.


T., 3. Chem. Phys.. 52:1552 (1970).
63. Greene, F.T., and Milne, T.A., 3. Chem. Phys.,
39:3150 (1963).

64. Hagena, 0.. and Bier, K., 3» Chem. Phys., 56:


1793 (1972).

65. Uilliams, U.D., and Lewis, 3.U.L., "Experimental


Studies of The Reservoir Temperature Scaling of
Condensation in a Conical Nozzle Floufield,"
Progress in Astronautics and Aeronautics, ed.
by 3.L. Potter, vol. 51: Part II, AIAA Publica-
tion, 1976, p. 1137.
66. Sherman, F.S., Phys. Fluids. 8:773 (1965).

67. Sherman, F.S., 3. Fluid, Mech., 8:465 (1960).

68. Mikami, H., and Takashima, Y., Int. 3. Heat Mass


Transfer. 2:1597 (1948).

69. Oberai, M.M., Phys. Fluids, 8:826 (1965).

70. Rothe, D.E., Phys. Fluids. 9:1643 (1966).

71. Sebacher, D.3., AIAA 3.. 6:51 (1968).

72. Campargue, R~., 3. Chem. Phys., 52:1795 (1970).

73. Sharma, P.K., Knuth, E.L., and Young, U.S., 3.


Chem. Phys.. 64:4345 (1976).

74. Yoon, S., Knuth, E.L., "Species Enrichment Due


to Flach-Number-Focusing II. Diatomic Species,"
Eleventh International Symposium on Rarefied Gas
Dynamics, ed. by R. Campargue, Vol. 1, Commis-
sariet a L'Energie Atomique, Paris, 1979, p. 639,

75. Fisher, S.S., and Knuth, E.L., AIAA 3.. 7:1174


(1969).

145
76. Waterman, P.C., Stern, S.A,, 3. Chetn. Phys..
31:405 (1959).

77. Anderson, 3»B», and Fenn, 3.B., "Background and


Sampling Effects in Free 3et Studies by Molecular
Beam Measurements," Fourth International Sympo-
sium on Rarefied Gas Dynamics, ed. by 3.H. Lecuu,
vol. II, Academic Press, Neu York, 1966, p. 311.
78. Biordi, 3.C., Lazzara, C.P., Papp, 3.F., Combus-
tion and Flame. 23:73-82 (1974).
79. Hayhurst, A.M., and Kittelson, D.B., Telford,
N.R., Combustion and Flame. 28:123 (1977).

80. Hayhurst, A.M., and Kittelson, D.B., Combustion


and Flame. 28:137 (1977).
81. Smith, O.I., "Probe Induced Distortions in the
Sampling of One-Dimensional Flames," USS/CI Fall
Meeting, USS/CI 79-52, Berkeley, 1979, p. 15.

82. Yoon, S., and Knuth, E.L., "An Experimental Study


of Probe-Induced Distortions Jh Flolecular-Beam-
Flass-Spectrometer Sampling from Flames," Presented
at the Twelfth International Symposium on Rare-
fied Gas Dynamics (to be published).
83. Gombs, G.C., "The Future of Coal as a Source of
Synthetic Fuel, " Coal Conversion Technology.
ed. by A.H. Pelofsky, American Chemical Society
Symposium Series 110, The American Chemical So-
ciety, 1979, p. 135.

84. Goldstein, U., Energy Policy, 7:275 (1979).


85. Uhitehurst, D.D», "A Primer on the Chemistry
of Coal," Organic Chemistry of Coal, ed. by 3.U.
Larsen, The American Chemical Society Symposium
Series 71, The American Chemical Society, 1978,
p. 1.
86. Longuell, 3.P., "Synthetic Fuels and Combustion,"
Sixteenth Symposium (international)on Combustion,
The Combustion Institute, Pittsburgh, Pa., 1976,
p. 1.
87. Ueinberg, F.3., "The First Half-Million Years
of Combustion Research and Today's Burning Prob-
lems," The Fifteenth Symposium (International)

146
on Combustion, The Combustion Institute, Pitts-
burgh, Pa., 1974, p. 1.
88. Franklin, 3.L., Dillard, 3.G., Rosenstock, H.M.,
Herron, 3.T., Draxl, K., and Field, F.H., "loni-
zation of Gaseous Positive Ions," NSRDS-NBS 26,
U.S. Dept. of Commerce, National Bureau of Stand-
ards, 1969.

89. Nilne. T.A., Greene, F.T., 3. Chem. Phys.. 44:


2444 (1966).

90. Franklin, 3.L., and Field, F.H., Electron Impact


Phenomena, Academic Press, Neu York, 1957.

91. Mark, T.D., Egger, F., and Cheret, M., 3. Chem.


Phys., 67:3795 (1977).

92. Beran, 3.A., and Kevan, L., 3. Phys. Chem., 73:


3866 (1969).

93. Ehrhardt, Van H.. and Osberghaus, 0., Z« Natur-


forschg., 13a:16(1958).

94. Papp, D., and Englander-Golden, P., 3. Chem.


Phys., 43:1464 (1965).

95. Smith, P.T., Tate, 3.T., Phys. Rev., 39:270 (1932).


96. Uestbrook, C.K.. Combustion Science and Technol-
ogy, 20:5 (1979).~
97. Uarnatz, 3., Progress in Astronautics and Aero-
nautics, vol. xx (1980).

98. Uestbrook, C.K., Dryer, F.L., Presented at the


Eighteenth International Symposium on Combustion,
Waterloo, Canada, 1980 (to be published).

Das könnte Ihnen auch gefallen