Reactivities and structures of several molecular cations relevant to combustion and soot formation


Material Information

Reactivities and structures of several molecular cations relevant to combustion and soot formation
Physical Description:
ix, 236 leaves : ill. ; 28 cm.
Brill, Fred William, 1952-
Publication Date:


Subjects / Keywords:
Soot   ( lcsh )
Ion cyclotron resonance spectrometry   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1983.
Includes bibliographical references (leaves 230-234).
Statement of Responsibility:
by Fred William Brill.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 000378774
notis - ACB8515
oclc - 10223640
System ID:

Full Text







Copyright 1983


Fred William Brill


This work is the final result of the efforts and influ-

ence of many throughout all the years. I must first thank

my parents and family for providing the encouragement and

atmosphere for pursuing the goals of higher education.

Special thanks go to my uncle, Dr. Robert H. Brill, now

Curator of the Corning Museum of Glass, Corning, New York,

for setting the example and directly fostering my interest

in the physical sciences. So, too, I wish I could thank

that group of extra-inspiring teachers and professors whom

I have had over the years in my academic studies--wherever

they may be now.

Any results obtained at an institution such as the Uni-

versity of Florida are the product of a massive team effort.

All too often essential team members do not get the recogni-

tion they deserve. I would like to acknowledge their efforts

here. First, without the personnel in the various shops,

there would be no icr, electronics, or even sample tubes to

work with. My thanks go to Art Grant, Chester Eastman, and

Dailey Burch of the machine shop; Rudy Strohschein of the

glass shop; and Walter Johnson and the others of the elec-

tronics shop. However, I would like to pay special recogni-

tion to Ed Whitehead, the best machinist and one of my best

friends at the University of Florida. In addition to his


excellent work in fabricating parts, Mr. Whitehead took time

to teach me as much as he could about machine tools and shop

techniques. I am a far better experimentalist by a factor

of at least two or three as a result of his teachings. Also,

putting this work into the final form was a major task in

itself. I would like to acknowledge my typist, Pam Victor,

for her fine work.

For the theoretical aspects of this work, I benefitted

from the years of wisdom and experience of many professors.

Professors Weltner and Muschlitz of Chemistry and Professor

Bailey of Physics never failed to lend their ideas (or spare

parts) when questions or difficulties arose. I would also

like to acknowledge the hours that Professor Battiste spent

in discussion with me. Physical chemists like myself often

are inexperienced and out of their element when dealing with

sophisticated physical-organic reaction mechanisms. His

assistance played a great role in attempting to assign the

structures of the various carbonium ions encountered in my

work and explaining their behavior.

In doing this work I have also had the priviledge of

working with some of the leading experts of the world in

theoretical physical chemistry, Dr. Zerner and Dr. Edwards

of his group. It was Dr. Edwards who taught me to use their

MINDO/3 program and ran the spectra prediction programs.

Both Dr. Edwards and Dr. Zerner also assisted me in the in-

terpretation and incorporation of the results into my


Next, I wish to acknowledge the many years of assistance

and guidance of Dr. John Eyler, my research advisor. His

encouragement and ideas, not to mention research funds,

played the most important role in the work and results


Finally, there is my wife and best friend, Amy. She

has been an endless source of strength and encouragement

of the moral kind. These last years have been the best.



ACKNOWLEDGEMENTS ................................... iii

ABSTRACT ........................................... viii


1 INTRODUCTION ................................. 1

2 INSTRUMENTATION ............................. 7

Introduction ................................ 7
The Ion-Cyclotron Resonance Mass
Spectrometer ............................. ... 7
ICR Control and Data Handling Systems ....... 16
Laser/Icr Instrumentation .................... 23

AN ICR MASS SPECTROMETER ..................... 29

Introduction ................................. 29
Theory of Ion-Molecule Reactions ............ 29
Theory of Pseudo-First Order Reactions ...... 35
Experimental Procedure ....................... 41
Pressure Measurements/System Calibrations ... 49


Introduction ................................ 58
General Discussion of Flames ................ 59
Mechanisms of Soot Formation ............... 65
C3H Structures and Discussion .............. 73
C5H3 Structures and Discussion .............. 77


C2H2 + C2H2 System ........................... 87
C3H3 Systems ............................... .. 92
C5HH Systems ................................. 103
CsH6 Results ............................ .. 118
MINDO/3 Results ............................ 118


Flames and Sooting Mechanisms ...............
C5Ht Structure Assignments ..................

INVESTIGATIONS ..............................

Structure Assignments C5H5 Isomers .........
Flames and Soot Formation ...................



Background ..............
Results .................
Discussion ..............


Introduction ................................
MINDO/3 Results .............................
Predicted Spectra ...........................



Introduction ................................
Description .................................
Program Map and Description .................
List .........................................

BIBLIOGRAPHY ......................................

BIOGRAPHICAL SKETCH ...............................











.... .......... .. ............. ....... 206





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



Fred William Brill

April 1983

Chairman: Dr. John Eyler
Major Department: Chemistry

+ '-
Kinetic studies of two species, C H3 and C H5, have been
3 3 5 5'
carried out in a pulsed ion cyclotron resonance (icr) mass

spectrometer. Over 50 rate coefficients of these ions reac-

ting with 12 neutral gases including methane, propane, ace-

tylene, ethylene, benzene, toluene, naphthalenes, indene,

butadiene, and allene are reported. Various proposed

mechanisms of soot formation in flames are examined and

the kinetic results are discussed in relation to Calcote's

ionic model of soot formation. The data obtained tend to

support this model, but raise important questions concerning

the structures and states of the C H3 and C H species found
33 55
in flames. Future lines of investigation are suggested.
Two structures of C H the cyclopropenyl and propargyl,
were produced from electron impact on, or CO charge exchange
were produced from electron impact on, or CO+ charge exchange


with, C3H X, where X is C0 or I. The propargyl cation is

seen to react with a rate near the Langevin limit while the

cyclopropenyl is relatively unreactive with the above gases.
Four different isomers of C5H, H distinguished by their reac-
5 5'
tivity, are believed to be produced from electron impact on

cyclopentadiene, dicyclopentadiene, l-penten-3-yne, and nor-

bornadiene. These isomers and their probable structures are


MINDO/3 calculations were performed on more than 15 of

the possible C H structures. A table of relative heats of
5 5
formation has been constructed, with optimized bond lengths

and geometries for each structure included. LCAO-MO-SCF-CI

calculations have also been performed on four selected struc-
tures of C5H5, predicting their spectra for future use in

distinguishing them experimentally.

Finally, some photodissociation studies on bromoethane

are presented as an example of future lines of investigation
+ +
for the C H and C H species. These studies provide specific
55 33
information on the potential surfaces of the bromoethane cation.

There appear to be two low lying excited states, one of them

repulsive. The other may lead to internal conversion to high

vibrational levels of the ground state. This may be due to

the higher density of these vibrational states since bromo-

ethane possesses a doublet ground state.


Mankind has been both fascinated and puzzled by fire

since before the dawn of history. Although its general

characteristics and behavior (and destructive potential)

have been well known and applied throughout the ages, only

today is the actual chemistry of the reactions which are

occurring within the flame itself being investigated. Only

recently have the powerful tools of the scientific profession

been developed to provide the ability to probe inside a flame,

a most common phenomenon of everyday life throughout the world.

With the industrialization of the world in the last few

centuries, fire, or more accurately combustion, has been har-

nessed with ever-increasing efficiency. This harnessing not

only sparked the development of many scientific fields such

as thermodynamics back in the days of James Watt, but it

started investigations into the engineering aspects and na-

ture of combustion. Improved machines, the development of

central heating, and quantum leaps in military firepower

were the result.

Despite these advances, almost nothing was known about

the inner processes of the flame itself until the tools of

mass spectroscopy and laser spectroscopy became available.

The study of flames was something that engineers did to

develop new burners or that regulatory agencies did to test

the combustibility of some new product. With the advent of

these new tools, a much more sophisticated investigation of

flames became possible. However just the availability of a

tool does not necessarily mean that it will be applied with-

out another motive. Throughout man's experience with fire

and flames, one of the byproducts, soot, has been uniformly

recognized as something ranging from a mere nuisance to a

significant danger.

In the decade of the 1970's, interest developed concern-

ing exactly how soot might actually form in a flame. Cleaner

combustion, internal or external, became an important goal

for many reasons. As a result it was recognized that more

should be learned about all of the reactions involved in

combustion. Consequently, mass spectroscopic and laser-

spectroscopic methods were used to identify and study the

species inside the flame. With these and other probe methods,

theories were proposed to explain the formation of soot.

Many of them, in light of the latest data, have been demon-

strated to be unlikely, but still there is no one definitive

mechanism for explaining soot formation. Some of these mech-

anisms and one which is gaining more and more support for

its acceptance will be examined in detail later.

One of the instruments used in mass spectroscopy is the

ion-cyclotron mass spectrometer (icr). While it has not been

applied to studying flames directly, it is an excellent instru-

ment for investigating ionic species found within a flame.

These ionic species present in flames are suspected of playing

an important role in soot formation. By creating some of

these species in the icr and studying their ion-molecule

reactions, researchers can gain information concerning the

viability of ionic pathways to soot. In addition, much can

be learned about the chemistry of these species.

Chapter 2 describes the pulsed-icr mass spectrometer at

the University of Florida, where these studies were performed.

Details concerning its construction and capabilities are given

there. Also discussed are the modificationswhich allow laser

spectroscopy studies to be performed on the ionic species which

the icr can trap for periods of seconds in favorable cases.

Chapter 3 gives a description of the kinetics involved in

studying ion-molecule reactions. First the various models of

reactions on a molecular scale are presented, and then a back-

ground of the pseudo-first order kinetics of these ion-molecule

reactions is included. Following this, the experimental use

of the icr instrument is outlined and the procedure for obtain-

ing kinetic data is described. One of the essential parameters

for any kinetic analysis is an accurate measurement of pressure

for the reactant neutral involved. This presents some diffi-

culty for the mass spectroscopist. Absolute pressure in the

ranges used is very difficult to measure. The last section of

Chapter 3 gives a complete account of how this problem was

addressed and resolved for this system.

Chapter 4 contains the background material on flames and

what is known about the species and reactions within them.

The question of "What is a flame?" is examined along with

some of the definitions involved. Then the various mechanisms

proposed to explain soot formation are discussed. First there

is a general discussion of the neutral and free radical pathways.

Then, in more detail, one of the presently most promising mech-

anisms, the ionic mechanism proposed by Calcote, is described

(Calcote, 1981).

Two of the ionic species which are suspected to be of

prime importance in this ionic mechanism are the CH + and CH
33 55
ions. It is their reactions and chemistry to which this dis-

sertation is primarily addressed. This is explained in

Chapter 4. Necessary background material for studying these

two species is also laid out. There have been some previous

studies of C3H both theoretical and experimental. However,

only a few semi-empirical calculations have been performed on
C H in the past. Both of these species are very interesting

and'very important chemically in their own right.

The results of this study are presented in Chapter 5.
+ +
For C H and C5H, H over 50 rate coefficients for reactions
3 3 5 5'
with various neutral gases are reported. Also the result of

a large set of MINDO/3 studies on many of the numerous proposed
C5H5 isomeric structures are included. Experimentally, two dif-
5 5
ferent isomers of C3H+ are obtained, and four different C5H +
3 3 5 5
isomers are believed to have been seen. Also, one set of rate

coefficients for C Hg reacting with 13 neutral gases is


Chapter 6 takes the results of the C H3 experiments and

applies them to the ionic model of soot formation. While they


tend to support the model, some very important questions arise

concerning the structure and state of the C H in flames. The
3 3
results of the C H experiments are likewise analyzed with
5 5
respect to their involvement in flames. More questions con-

cerning the structures and states of C H within flames are
5 5
posed. However, in this case, there is more chemistry to be

discussed. Assigning structures to the four experimentally

observed isomers is a formidable task, one that cannot be

done with certainty given the evidence at hand.

Finally, Chapter 7 presents some final conclusions and

questions which have arisen as a result of this research.

Also several future lines of investigations are proposed.

Some of these are elaborated in the Appendices.

Appendix A is a presentation of laser-induced photo-

dissociation studies of bromoethane. These, while not directly

related to the flame question, are typical of the future direc-

tions that certain proposed continuations of the investigations

of the species found in flames should follow. This photodis-

sociation study gives exactly the information on bromoethane
cation that is required for CH +. These studies also stand

on their own strength.

Appendix B is primarily a listing of the MINDO/3 results.

The bond lengths and geometries of many C5H+ and two C 5H

species are listed. Also included are four predicted spectra
for four different isomers of C5H. H They could prove to be a

basis for distinguishing between the experimentally observed

C H cations.
5 5


Finally, Appendix C lists the BASIC language computer

program used in the kinetic analysis of the data for this

work. Both a description and a program map are given.

This program was written to analyze the single or double

kinetic decay curves encountered in an icr work.



As an extremely versatile instrument, the icr mass spec-

trometer can be set up and used in many different configurations.

To begin, the basic theory of its design and operation will be

discussed. Subsequently there follow the details of the in-

strument at the University of Florida. Digital electronics

and a microprocessor-based data acquisition system greatly

enhance this system's performance. Also, the icr has been

modified to conduct laser studies on the ions trapped within

its analyzer cell. Both of these aspects will be covered in


The Ion-Cyclotron Resonance Mass


The principles of cyclotron motion, the motion of charged

particles in a crossed electric and magnetic field, have been

known for many years. Using them in experiments capable of

high mass resolution was first accomplished by Sommer, Thomas

and Hipple (Sommer, Thomas, and Hipple, 1951), their intent

being to study the e/m ratio of the proton. Commercial ver-

sions of the icr mass spectrometer as it is known today be-

came available from Varian Associates, (Lehman and Bursey,

1976) in the late 1960's. Since the physics of cyclotron

motion and, more specifically, such motion in the icr mass

spectrometer have been well documented before, only a gen-

eral overview will be presented here. The reader is referred

to the following works for further details--McIver, 1970;

Sharp, Eyler,and Li, 1972; Morgenthaler, 1979.

Figure 2-1 shows a schematic diagram of the icr trapped

ion cell. The magnetic field is parallel to the z-axis.

Electric potentials, and sometimes other signals, are applied

to each of the six sides, thereby establishing an electric

field inside the cell. Ions are created by electron impact

on neutral gases in the cell. In all cases only singly

charged species will be considered. Once created, these

ions will undergo cyclotron motion in the properly tuned

magnetic and electric fields.

For an ion with mass m, charge e, acceleration a, and

velocity v perpendicular to B, Newton's second law can be

expressed as

S= ma = e(v x B) 2.1

If one substitutes a = v /r, where r is the radius of the

ion's path, this leaves

ma = my /r = e(v x B) 2.2

If one defines wc as the angular frequency in radians/s, then

mv/r = mwc = eB 2.3

0 N

\ O\

\ H



--Iq- { ._-

\o \ H
co I \ -


0 c \ C

O -p
0 0 0
S\\ H

00) 0


wc = eB/m 2.4

Usually the frequency vc is used, where vc = wc/2w7

vc = eB/2rm 2.5

This result is the basic cyclotron equation. A more rig-

orous treatment of the cyclotron motion involves using Equa-

tion 2.6.

F = q(E + v x B) 2.6

Here q is the charge on the ion and the other quantities are

as described before. Solving Equation 2.6 and taking into

account the effects of the trapping potentials on the motion

of the ions gives Equation 2.7.

w2 = (qB/m)2 4qVt/md2 2.7

This expression for w now has a second term, a correction
factor due to the potential on the trapping plates. The

quantity Vt is the potential on the trapping plates and d

is their separation.

Two other motions are seen in addition to the basic

cyclotron orbit. One is an oscillation on the z-axis about

the origin of the ion's formation which is related to the

correction term described above. The second is a slow drift

along the equipotential lines within the cell which causes

the cross section of the ion cloud to oscillate between a

circle and an ellipse in the x,y plane. Morgenthaler de-

scribes these in more detail (Morgenthaler, 1979).


Equation 2.5 indicates that for a given frequency, the m/e

ratio detected varies linearly with B. Ion detection will be

discussed shortly and the control and data handling system

covered later in this chapter. In conventional icr operation

it is customary to hold the frequency vc constant and sweep

the magnetic field. Typically this field ranges from 1 to 15

gauss (0.1 to 1.5 tesla). Other parameters include, for posi-

tive ions, side plate trapping potentials of about 1.5 V,

upper and lower plate potentials of 1.0 V, and end plate

potentials ranging from 1.0 to 3.0 V. During the course of

experiments it is often necessary to fine tune these voltages

in order to optimize ion trapping. For studying negative

ions, all voltages are reversed in sign. Using the above

conditions and at a frequency of 153.57 kHz the m/e vs B

separation should be exactly 10 amu/1000 gauss. In reality

a constant factor, approximately 100 to 200 gauss, must be

subtracted from the observed magnetic field's value in order

to determine the m/e ratic in resonance. This is due to the

effect of the trapping plate potential on the cyclotron fre-

quency, the second term of Equation 2.7 discussed earlier.

At these values of vc and B the radius of the cyclotron

motion is on the order of 0.1 mm.

A pulsed marginal oscillator is used for detection of

ions trapped in the cell. This device, too, has been de-

scribed in detail previously (Mclver, 1973). Essentially

ions trapped in the icr cell and undergoing cyclotron motion

can absorb rf radiation if its frequency is that of their

cyclotron motion. When the detector is gated on, rf is pro-

vided in a burst to the top plate of the cell and is absorbed

by the ions if it is of the proper frequency. It can be shown

that the amount of power absorbed is directly proportional to

the number of ions present. Unfortunately, power absorption

is inversely proportional to the mass, a matter which must be

routinely corrected if two ions are to be compared in intensity.

The marginal oscillator detector provides great sensitivity in

the measurement of the total power lost by absorption from the

ions. The power absorption signal is passed through a gated

integrator whose output is captured by a sample and hold

circuit. From there the data can be digitized and sent to

a recorder or a computerized data handling system.


The pulsed icr mass spectrometer utilized in these studies

was built at the University of Florida. A schematic diagram is

shown in Figure 2-2. Non-magnetic stainless steel was used in

the fabrication of the vacuum system. Valves, gauges, and

pumps were purchased commercially as were many of the fittings.

All construction was performed in the Chemistry Department

machine shops.

A Varian 3400 series 9 in. low impedance electromagnet

is shown on the right of Figure 2-2. It has a V-FR2500 7 kW

power supply and is regulated by a "Fieldial" Mark 1 controller.

Although the power supply allows for operation up to 1.6 tesla,

in actual use it has only been possible to go up to 1.4 tesla.

This is due to the fact that a recirculating coolant system

(1) -P
0) CQ



C *i-H

using distilled water cooled by tap water cannot handle the

heat produced past this value. Ideally a chilled water re-

circulation system should be used.

The icr cell, shown in the schematic of Figure 2-2 and

drawn in detail in Figure 2-1, is located inside the vacuum

chamber. Only two inches separate the magnet pole faces.

This vacuum chamber, or can, is removable at the flange.

One would perform such a removal in order to service the

cell, adjust laser mirrors, or most likely, replace the

filament. Constituting the main portion of the high vacuum

side of the mass spectrometer, the vacuum chamber is pumped

by either one of two pumps. The primary pump is a CVC 2 in.

oil diffusion pump with a liquid nitrogen baffle on top.

During a run the main gate valve would be open, the nitrogen

baffle filled, and the valve to the Vaclon pump shut. An

ultimate background pressure of 1 x 108 torr is routinely

attained. While the mass spectrometer system is not in use,

the main gate valve is closed and the valve to the Vaclon

titanium sublimation pump opened. The Vaclon pump, with its

8 1/s pumping speed, maintains a low background of approxi-
mately 3 x 10 torr. Heating tape is wrapped around the

entire vacuum chamber and all other parts of the mass

spectrometer. The system can be baked out for extended

periodsof time at temperatures exceeding 100 degrees Celsius.

A Bayard-Alpert ionization gauge measures the pressure

in the main vacuum region. This gauge is susceptible to two

serious errors in addition to an overall systematic error.

First, the ionization gauge is not equally sensitive to

different gases. The reading for the same pressure of dif-

ferent gases may differ by a factor of 4. Second, the pres-

sence of fringing magnetic fields can also affect the gauge,

causing reading errors of 20 per cent at 1.0 tesla. Finally,

there are errors due to effects of the vacuum system geometry.

The gauge is at the end of an extended sidearm, downstream

from the trapping cell. Therefore, the pressure at that point

cannot possible be the same as at the cell. All of these prob-

lems can be and have been compensated or adjusted for. This

will be discussed later, at the end of Chapter 3.

The foreline of the mass spectrometer is used for sample

handling. It is pumped by another CVC 2 inch oil diffusion

pump topped by a liquid nitrogen trap. A cold cathode dis-

charge gauge monitors the pressure in the foreline. These

components are also shown in Figure 2-2. Both of the oil

diffusion pumps, the one on the high vacuum side and the one

on the foreline of the system, are protected by a controller

which will turn off power to their heaters should the pressure

in the high vacuum side rise too high. This controller moni-

tors the pressure with a thermocouple gauge, not shown in

Figure 2-2, located between the oil diffusion pump of the

high vacuum side and the rotary oil pump which backs it.

Sample is introduced from pyrex sample tubes or bulbs,

made in the Chemistry Department glass shop. First, the gate

valve is always kept open. The outer valves or the foreline,

shown in Figure 2-2 as OV1 and OV2, are closed when attaching

or removing sample tubes. Then the four valves, OV1, OV2,

IV1, and IV2, are all opened and the entire foreline pumped

down. Following this, one closes the inner valves, labeled

IV1 and IV2, and allows the sample vapors to enter. Two

Varian leak valves, LV1 and LV2, then provide the interface

between the foreline and main vacuum region. Sample vapor

is leaked at a constant rate through a tube terminating

underneath the icr cell. If only one sample is to be entered,

only one set of valves, e.g., IV1, OV1, and LV1, is used.

As a final note, it should be mentioned that the entire

main vacuum system was rebuilt after a catastrophic failure

of an inner seam of the nitrogen trap then installed. A

nitrogen baffle with different dimensions was installed.

Thus this description supercedes previous descriptions of

this particular area of the instrument prior to 1982.

ICR Control and Data Handling Systems

Digital Electronics

The icr electronics have recently been updated and differ

significantly from previous descriptions (Morgenthaler, 1979).

Originally the experimental sequencing and parameters were

controlled from a console where one entered the values on

binary-coded decimal thumbwheel switches. In the new version,

all control is accomplished through a microprocessor system

designed and built by Dr. Thomas Buckley. The reader is re-

ferred to his dissertation for a highly detailed description

of the hardware and software (Buckley, 1982). This section,

however, contains a general description of the electronics,

computer system, and the data acquisition procedure.

The data acquisition system begins at the output of the

marginal oscillator. In both the updated and original versions

a switched integrator integrates the signal to give the total

power absorption of the trapped ions for a given period of

time. A sample/hold molecule then reads this value which is

actually an analog voltage level. This voltage can then be

routed to a simple chart recorder for display or digitized

for data processing. Earlier digital treatment was done

through a single KIM-1 microcomputer and KIMSI S-100 bus.

A dual APPLE II/KIM-1 microprocessing system has evolved

from this which greatly enhances the control and data hand-

ling capability.

In general the APPLE II (48K, DOS 3.3 system) microcom-

puter is the master controller, while the KIM-1, in a slave

configuration carries out the pulse generation, averaging,

and system initialization. The APPLE II handles all user

input and output, system initialization, data storage, and

limited data analysis; the user types in the experimental

parameters and instructs the APPLE to begin the experimental

run. Talking to each other via the S-100 bus where the APPLE

and KIM-1 have an 8K block of common expansion memory, the data

is passed back and forth. Both microcomputers use the same

6502 microprocessor and thus can accept the same machine lan-

guage codes. Also on the S-100 bus area Biomation Waveform

Recorder, a Rockland Frequency Synthesizer, and an experi-

mental interface. Essential to reproducible work is total

synchronization of all components. The frequency synthesizer's

internal 8 MHz clock provides the basis for this.

The previously mentioned analog voltage signal, origi-

nating in the marginal oscillator and held in the sample and

hold circuit, is fed into an internal 8 bit A/D converter of

the Biomation unit. Then the digitized value is stored in

all 2048 channels of the recorder. It can then be interro-

gated by the KIM for its data during the appropriate time

in the experimental sequence. When this occurs the KIM will

take the value in the sixth channel and place it in the common

8K S-100 memory. The choice of the sixth channel is arbitrary;

however, the first channel was not used in case any spurious

noise spike might be located there. If averaging is to be

done, the KIM does that first. From the S-100 block the APPLE

then takes the data, stores it, and can display it in appro-

priate form on the high resolution graphics monitor.

Data Acquisition Methods and Modes

Following the description of the icr mass spectrometer

it is appropriate to outline the typical experimental run.

Ions are created in the icr cell by electron impact. Although

this is inherently inefficient (only 0.1 to 0.01 per cent of

the neutral molecules are ionized) at the pressures used

sufficient numbers of ions are made. Approximately two to

three amps of current from a Kepco regulated supply heat

the rhenium filament, shown in Figure 2-1, to incandescence.

Also, a fixed negative potential, Vf, is placed on this

filament. Between the filament and the entrance to the

cell is the grid. Normally a voltage of (Vf -5) volts

is applied to the grid, effectively preventing electrons

boiled off the filament from penetrating to the cell. When

it is time to create ions, the grid potential change to

(Vf + 5) volts, allowing the beam to enter the icr trapping


Assuming that the other conditions such as the magnetic

field, electric potentials, and the rf frequency are properly

set, these ions will be in resonance, trapped in the cell,

and ready for detection. A pulse from the control system at

a preselected time gates on the marginal oscillator applying

a burst of rf energy into the cell. Then, as described before,

the ions of the proper e/m ratio absorb power. This absorp-

tion is measured and converted to a voltage which is processed

as described in the previous section. Finally, a last pulse

is applied, the quench pulse. This pulse ejects all remaining

ions from the cell and thus prepares it for the next duty cycle.

All of these actions take place on a time scale of 1 ms to 16 s.

One enters these times for the specific pulses into the APPLE,

which, with the KIM, controls the sequence described above.

A diagram showing this scheme is given in Figure 2-3.

The experiment outlined can be performed in one of four


1. Continuous scan. The pulse sequence repeats with

all parameters held constant. Up to 280 points can

be stored on the APPLE II display screen after

which it automatically clears and 280 more points

are added.



.0 C)

0) ) 0
C)~ C


> z

r- 00

0 C)


S 3)

U) M

- 3
* -C o a
CP Cd C)'

4c) Cd
o C CH
) *DH C)

SC0 0

CMr d t
(12 Cd-Q
C) -PT t




C )






a >C
-----1c 1-1

-H -P


2. Signal averaged. Each data point on the screen is

the average of n scans, with n chosen by the operator.

The run terminates when the screen is filled with

280 points.

3. Detect delay sweep. The delay of each successive

pulse is increased by a set interval up to a set

limit. Each data point can also be the average of

n preset passes.

4. Detect delay sweep/baseline. This is the same as

the mode 3 except that the data alternates from

that with the grid pulse on (ions formed), and that

with the grid pulse off (no ions formed). This mode

in effect gives a signal with its baseline.

These four modes provide the basis of all experimentation.

For a conventional mass spectrometer "mass sweep" over a por-

tion of the mass spectrum, one would use mode 1 with the magnet

Fieldial controller gradually increasing or decreasing the

field. When the ions come into resonance the successively

displayed data points form a peak on the monitor screen.

After is completed, the data, if worth anything, can

be stored on a data disk for later viewing and analysis. One

of the pulsed icr's greatest advantages is its ability to

give time resolved mass spectra. These are of great value

for kinetic work. What one obtains is a straightforward

intensity vs. time plot! Modes 3 and 4 give this type of

plot. In use, the decay with baseline is much preferred

since it allows one to be sure of obtaining the absolute

intensity with respect to the baseline at each point. The

APPLE II can also perform a least squares fit of all the

baseline points and subtract the average from all data points

of the ion signal. It was in this mode that almost all of

the kinetic data in this dissertation were obtained. The

laser studies required a slightly modified version of

mode 1.

Data Analysis

Data analysis is now accomplished by a microcomputer.

Software has been written for the APPLE II 48K system to

perform this task. In the ICR-MO experimental control soft-

ware, written by Dr. Buckley, there is a simple procedure

for obtaining rate constants; however, this routine is not

the best and cannot handle double kinetic decays. With the

data digitized and stored on a disk, they can be retrieved

easily and analyzed at a later time. A number of programs

to perform double exponential decay curve fits as well as

to calculate laser induced photodissociation cross-sections

have been developed. The kinetic analysis program is dis-

cussed in Appendix C. The chemistry and physics involved

in these calculations are covered in the next chapter.

Microprocessor data analysis is considerably more accurate

and many times faster than the old methods of analyzing

curves by hand and calculator. Also, it can be performed

in a few minutes between experimental runs, thereby giving

rapid results and directions for immediate follow-up.

Laser/Icr Instrumentation


The icr mass spectrometer has been modified to perform

laser studies on the species trapped within its analyzer cell.

This configuration, described previously (Eyler, 1974), uses

an intracavity technique to maximize the photon flux through

the sample ions. A Candela ED-100U flashlamp pumped dye laser

provides the coherent radiation source. In Figure 2-4 this

device is denoted by FL. More about the laser including

operational details will be discussed in the next section.

The light path is shown to pass between a mirror and a

diffraction grating; this constitutes the laser cavity. It

also contains several microscope slides and a window on the

vacuum can. Mirror M is 99 per cent visible reflecting and

is antireflectivity coated. It is mounted on the back end

plate of the icr cell. In order to maintain electrical in-

tegrity, a fine gold mesh is stretched across the face of the

mirror. No serious degradation of the laser beam results.

The other end of the cell must be open to pass the beam

inside. A square hole, about 3/4 in. on a side, is cut into

the end plate. This hole does not adversely affect the ion

trapping capability of the cell. The diffraction grating,

used for tuning the wavelength of the dye laser throughout

the visible region from 450 nm to 700 nm, is labelled GR.

A low geared motor driven sine arm, denoted SD, rotates

this grating. The region S, inside the icr cell, is where

the ions are trapped; they are within the laser cavity, thus

giving an intracavity mode of operation.



c 0
iM L
T_ a
llo l-





The laser beam passes through several microscope slides.

Slide SL1 diverts approximately 10 percent of the beam into

either an ISAmonochromator, labelled MC, for measuring the

laser wavelength, or a Gentec model 200 fast response Joule-

meter, labelled PM, whose output is sent to an oscilloscope

for manual recording. A second set of microscope slides,

labelled SL2-n, is employed for attenuating the beam without

changing the pumping voltage, laser alignment, or reflectors.

This is useful for studying the effects of varying the photon

flux on the observed ion photodissociation process. These

slides are separated by 1/4 in. space of plexiglass. Sets

of two, three, four and five slides were fastened together

with epoxy cement.


The Candela dye laser has an output of 250 mJ per flash

and the pulse duration lasts 500 ns. The spark gap assembly,

newly installed, now has a replaceable electrode and a second

fitting to accommodate a nitrogen flow system. A 25 kV power

supply energizes the xenon flashlamp. Pulses from the icr

control console or the microcomputer, via a trigger circuit,

can fire the laser at any time during the icr experimental

sequence. A manual trigger can also be used.

A whole range of laser dyes were used to span the visible

spectrum from 450 nm to 700 nm. They consisted of Coumarin

460, Coumarin 480, LD-490, Coumarin 504, Coumarin 540, Rhoda-

mine 560, Rhodamine 590, Kiton Red 620, Rhodamine 640, and

Cresyl Violet 670. All dyes were purchased from the Exciton

Chemical Company.

The use and optimization of the laser became somewhat

of a science in itself over the years. Some of the technique

or "tricks" learned, usually the hard way, are presented here.

Originally the entire driver and flashlamp were enclosed in

a metal case in order to eliminate the large electromagnetic

pulse which affected some of the old electronics. This cage

turned out to be unnecessary with the present system, thus

enormously simplifying the alignment procedure. A good con-

nection consisting of braided wire running from the laser

drive base directly to an earth ground is sufficient to avoid

problems arising from the intense rf radiation produced from

the laser. The condition of the spark gap and nitrogen chamber

proved to be the most important factor in obtaining reliability

and reproducibility of laser pulses. A new module with replace-

able electrodes and a nitrogen purge outlet was installed which

eliminated earlier problems from these sources. The only pro-

cedure necessary before each day's operation is to flush dry

nitrogen through the chamber for five minutes. This clears

the moisture and any residual gas which might interfere with

the spark.

The dyes present some problems too. They are normally

made to solution specifications given by Exciton. This

usually means that they are dissolved in absolute ethanol.

However, some lased successfully in methanol, a first for

this research group. It was observed that dissolving the

dyes in methanol shifted their output wavelengths approxi-

mately 10 nm to the red. The dyes are continuously circu-

lated by a pump through the flashlamp. This circulation

through a warm pump and the heat generated by firing the

laser heats up the dye considerably. The temperature effect

on the performance, or non-performance, of laser dyes is

controversial. Some at Exciton deny this effect's existence

completely. Nevertheless, degradation of dye performance

upon heating has been observed dramatically and consistently

over the years in this laboratory. Thus a glass cooling

coil was installed in the dye system to be immersed in ice

during use. This procedure significantly enhances the laser

performance in terms of power and reproducibility.

Past Establishment of Techniques

Studies of laser induced photo reactions in the pulsed

icr have been done since the mid 1970's. Thus the techniques

have been refined and the matter has become more routine.

Morgenthaler has derived the expressions for the laser-

induced photodissociation of molecular ions (Morgenthaler,

1979; Morgenthaler and Eyler, 1979). One starts with the

probability, P(X) that an ion will photodissociate at

wavelength X:

P(X) = 1 exp(-Ktp(A)o(X)dX) 2.8

where K is an overlap factor between the laser and the ions,

t is the pulse duration, p(X) is the flux of photons of wave-

length A and a(X) is the absolute cross-section for photo-

dissociation at A. The integration is done over the bandwidth

of the laser output. From this expression Equation 2.9 can

be derived

K'a(X) = (Zn(f/f-R))/(AE(X)) 2.9

where f is the fraction of the sample ion population actually

irradiated, R is the fraction of ions dissociated, E(X) is

the energy in a pulse of wavelength \, and K' is a system

geometry constant accounting for the fraction of the beam

which overlaps the ion cloud. In the laser-icr experiments

one measures four quantities, the wavelength, the laser power,

a relative number of ions without the laser on, and a rela-

tive number of ions after the laser has been fired. Only R,

X, and E(X) from the above equation come directly from the


There are considerable problems in defining exactly

what goes on when the laser beam strikes the cloud of

trapped ions. Morgenthaler discusses them extensively in

his thesis. In his experiments it was found that the frac-

tion of ions irradiated, f, is unity. This turns out to be

true in this work also. However, where previous studies

found no significant variation in the fraction of the beam

hitting the ions, part of the K' term, now it is seen that

the beam's nature changes with power. Calibration runs,

all at the same laser wavelength, gave relative cross-

sections that varied with power, clearly not an acceptable

state of affairs. The evidence seems to indicate that the

laser beam profile does not change homogeneously as its power

increases. This matter is discussed later in Appendix A

and account taken in the analysis of results.



In this chapter both the theoretical and practical as-

pects of studying ion-molecule reactions in the icr mass

spectrometer system will be covered. First discussed are

the kinetics of these reactions on a molecular level. The

various models developed are briefly described in order that

a feeling for the various processes taking place inside the

icr cell may be obtained. Next follows a discussion of the

macroscopic kinetics of these pseudo-first order reactions

including some problems and constraints. The experimental

procedure for obtaining bimolecular ion-molecule rate con-

stants is outlined. This discussion is a continuation of

that relating to certain instrumental details found in the

previous chapter. Finally, the serious problem of accurate

pressure measurements, mentioned earlier, is covered in depth

along with the procedure consequently developed for handling

the matter throughout the studies.

Theory of Ion-Molecule Reactions

The intention in this section is to acquaint the reader

briefly with the various treatments of ion-molecule reactions

on a molecular level. While no experiments in this work deal

with any of these theoretical models, nevertheless, the con-

cepts involved are worthy of mention in the interest of being

complete in the discussion of the overall kinetics. The

reactions studied in the icr, ion-molecule reactions, usually

involve a reaction between a singly charged ionic species and

a neutral molecule. Paul Langevin, a French physicist, first

dealt with the interaction of an ion and neutral shortly after

the turn of the century (Langevin, 1903). His model is still

successful in describing certain reactions.

Langevin's model assumes that the neutral has no perma-

nent dipole moment. Instead the ion creates an induced dipole

in the neutral molecule. This gives rise to a charge-induced

dipole potential between the two particles.

V = -(ae2)/2r ) 3.1

where V is the potential, e is the charge on the ion, r is the

center to center distance between the ion and the molecule,

and a is the polarizability of the neutral. Consequently, a

force of attraction, F, exists which can be written

F = -(2ae2)/r5 3.2

Figure 3-1 shows four possible cases describing an en-

counter between an ion and a neutral. The quantity b, the

impact parameter, is defined as the perpendicular distance

from the center of the ion to the projected path of the

neutral if there were no force between them. In that case

there would be no reaction unless a direct collision occurred.

Case a illustrates this. However, as indicated in Equation 3.2,


a b






Figure 3-1. The Four Cases of an Ion-Molecule Encounter.

in the Langevin model there is attractive force between the

ion and the neutral. This being so, there are three types

of encounter possibilities depending on the impact parameter

and the relative translational velocity of the ion and

neutral. First, if the incoming neutral possesses high

kinetic energy, its path is deflected but there is neither

collision nor reaction. Case b illustrates this. Should

the energy be low enough and the impact parameter small

enough, the neutral is sufficiently attracted toward the

ion that it spirals in, and collides. This is case c.

Finally, for certain conditions, i.e., when there exists

barely enough attraction to keep the neutral from flying

away, it will orbit the ion. Case d shows this situation.

A cross-section for these collisions can be derived

from the Langevin results.

a(v) = 2e(a/u) 3.3

Here a(v) is the cross-section as a function of velocity.

The quantity, v, is the velocity and u is the reduced mass

of the system. The cross-section in turn can be related to

the measurable macroscopic rate coefficient as follows

k = I f(v)o(v)vdv 3.4

Here f(v) is the velocity distribution function and a(v)

the Langevin cross-section. Two extreme constraints limit

application of this equation. Both f(v) and the relative

internal energy of the reactants must be constant through-

out the reaction.

If one assumes that each orbiting collision is reactive,

the final rate coefficient is given by

k = 27e(a/u)2 3.5

This is known as the Langevin, Gioumousis and Stevenson

equation. The latter two obtained it by using a Maxwell-

Boltzmann distribution for the velocity function in Equation

3.4. The rate coefficients obtained from this expression

are generally good--considering the two restrictions men-

tioned above. Also, the model has another restriction;

applications must be limited to cases where the neutral

has a small dipole moment or none at all.

Many molecules do possess dipole moments, so refinements

of Equation 3.5 are necessary. One early refinement, known

as the "locked dipole theory," assumed that as the neutral

approaches the ion and the two spiral, the neutral dipole

vector "locks" onto the ion, thus making it always point

inward to the ion (Moran and Hamill, 1963; Gupta, Jones,

Harrison, and Myher, 1967). Unfortunately, the resulting

rate coefficients turn out to be uniformly high and

unacceptable. The next improvement, made by Bowers and Su,

is known as the "Average Dipole Orientation Theory" or ADO

Theory (Su and Bowers, 1973a, 1973b). Here one assumes

that there exists an overall average orientation of the

dipole with respect to the ion. Results from this treat-

ment are reasonably good. Lehman and Bursey compare the

rate constants obtained by these methods with those deter-

mined experimentally (Lehman and Bursey, 1976, p. 161).

One can see that for these examples the ADO theory ranks best.

Statistical mechanics provides another approach to ion-

molecule reactions. When applied to ions created by electron

impact this is referred to as quasiequilibrium theory, QET.

The "quasi" prefix is added due to a few tenuous assumptions

which have to be made. The reader is referred to the general

literature for details about QET (Rosenstock, Wallenstein,

Wahrhaftig, and Eyring, 1952; Kiser, 1965). Suffice it to

comment that QET is used along with other theories such as

the Langevin, Gioumousis and Stevenson, and has been highly

successful in predicting product distributions. Lehman and

Bursey again show this in their text (Lehman and Bursey,

1976, p. 164).

The next section explores the macroscopic kinetics of

ion-molecule reactions. However, before proceeding, some

additional points should be made. One of the difficulties

in applying QET is the necessity for knowing details about

the reactant's electronic structure, vibrational modes, and

specific energy state. Icr technique usually cannot provide

such information. In some cases, though, ions can be created

by charge exchange, thereby giving some knowledge of the re-

actant's initial energy state. Nevertheless, one must assume

that the rate coefficients determined are averaged over those

obtained from all initial states to all final states--whatever

they may be.

The rate coefficients dealt with in this field are con-

ventionally given with the units of cm /mol s. Values typi-
cally measured in an icr mass spectrometer range from 1 x 1012
cally measured in an icr mass spectrometer range from 1 x 10

to 5 x 109. This latter number approaches the collision

rate. Such a reaction would be considered very fast. At

rates near the lower bound, one frequently encounters ion

loss mechanisms other than reaction; determining these rate

coefficients is consequently often quite difficult.

Theory of Pseudo-First Order Reactions


The study of ion-molecule reactions deals primarily

with bimolecular collisions and the rates are second order.

All of the reactions discussed in this work are assumed to

be so. In this section the simple rate expressions for

these systems will be presented along with the method of

analyzing the data.

Assume that A is the ion and C is the reactant neutral

gas. The simple second order reaction and its rate expres-

sion are

A + C --- k 3.6

-dA/dt = k [A][C] 3.7

In the icr the ion number density created is at least three

orders of magnitude less than the neutral number density.

Therefore it can safely be assumed that [C] >> [A]. In this

case [C] is essentially constant during the course of the

reaction and a new coefficient k1 then replaces the terms


-dA/dt = kl[A]


which is referred to as a "pseudo-first order" rate expression.

After separating the variables and integrating, one has

[A] = [A(O)]exp(-k:t) 3.9

where A(0) is the initial number of ions present. Thus, in

analyzing a reaction such as one described by Equation 3.6,

a simple ion signal vs time plot can be fit exponentially

and k' determined. The true rate coefficient, kl, can then

be obtained from this k' by dividing k1 by [C].

Rather than neutral concentrations, a number density is

often used in mass spectroscopy. Using PV = nRT, this number

density can be directly related to the pressure; (n/V) =

P/RT where the (n/V) term is the number density of ions.

The conversion factor relating pressure to number density

is 2.925 x 1016 particles/cc torr at 330 Kelvin, a tempera-

ture which has been measured in an icr cell by a thermocouple.

Thus, to obtain k1 from ki, the [C] is first converted to a

number density prior to division. This number density of a

species will be denoted by pX where X is the species in

question. Figure 3-2 shows a typical intensity (ion number

density of A) vs. time plot which results from examining

a reaction such as one described by Equation 3.6.

Often one finds parallel reactions occurring. This com-

plicates matters slightly. In addition to Equation 3.6 we

may also have

A + D ---- Q k 3.10






Figure 3-2. Typical Intensity vs. Time Decay Curve for the
"Pseudo-First Order" Reactions Studied in the
ICR Mass Spectrometer.

Theoretically there could be a third reactant or more. The

rate expression becomes

-dA/dt = klpC + k2pD + .. 3.11

and the integrated form is

A(t) = A(O)exp[-(klpC + k2PD)t] 3.12

Should this be the situation, one would have a single expo-

nential decay, but the rate coefficient is given by the sum

of two terms, one for each reaction. Figure 3-2 also would

be the typical intensity vs. time curve seen in this instance.

The slope of a log plot of this data would have two terms

contributing to it, klpC and k2pD, from Equation 3.12.

If one rate coefficient is known, the other can be

obtained. In an experimental situation the first rate co-

efficient, usually for the ion reacting with its parent

neutral, is determined by analysis of the ion's signal decay

curve--with no other reactant neutral added. Following this

"control" experiment, the other neutral gas is added and the

resultant decay curve can be analyzed to determine its rate

coefficient since the first one is now already known. Appen-

dix C discusses the computer program which currently does

these calculations.

It happens in mass spectroscopy that one often forms

two or more isomers of a particular ion from the same parent

neutral. Since the icr distinguishes ions solely on the

basis of m/e, the signal intensity will be the sum of the

intensities of all isomers with this same ratio. This is

the case in many of the kinetic experiments described later.

Therefore a treatment for handling this situation in the cal-

culation of rate coefficients has been developed.

Let B be defined as a second isomer of A, for example

linear C H and cyclic CH Both could react with any given
xy xy
neutral species present.

A + C -- P k 3.6

A + D --- Q k2 3.10

B + C ----- P' k3 3.13

B + D ----- Q' k4 3.14

For these reactions it is assumed that A is the faster reac-

ting of the two. Like Equation 3.12, an expression for the

amount of ion B at any time, t, can be written

B(t) = B(O)exp[-(k3pC + k4pD)t] 3.15

Since the icr signal intensity, I(t), is the sum of the sig-

nals of a given m/e, one has

I(t) = A(t) + B(t) 3.16

I(t) = A(O)exp[-(k1pC + k2pD)t] +

B(O)exp[-(k 3pC+ k pD)t] 3.17

Equation 3.17 contains four rate coefficients; their deter-

mination from one set of data would indeed be a formidable

task. The computer program described in Appendix C only has

the capability of calculating two rate coefficients from any

one set of data. This program does, however, incorporate all

four constants. One can enter two and determine two. It also

gives the values of A(O) and B(O).

When there are two isomers present in one data set, they

must be separated. The rate constant for the slower reaction,

by convention involving isomer B, is calculated from the tail

of the double decay curve. So too is B(O), the intercept for

isomer B. Next, using Equation 3.17, the amount of isomer B

at each point in the initial part of the curve can be sub-

tracted from the total ion signal. This leaves only that

part of the curve representing isomer A. From here the

faster rate constant can be calculated.

Determination of Rate Coefficients

While the problem of calculating one or two unknown

rate constants for just one isomer is relatively easy, cal-

culations for multiple ions rapidly grow complicated. With

the case of only one isomer, a routine for handling either

one or two k's is straightforward and adequately covered

above. When there are two ions, more questions or possi-

bilities arise. Should one wish to deal with four unknown

rate constants, a minimum of two separate experiments must

be done. Assume that both A and B ions react with C and D

neutral molecules. One approach is as follows. If alter-

nate sources for producing pure A or B were available, one

could study exclusively reactions of the ion A or B with C

or D to obtain one or more ofthe unknown rate constants.

Regrettably this is not always the case. To handle the

situation, one performs a control run involving both isomers'

reaction with only one neutral, usually the parent C. Then

one set of coefficients is determined, in this case k1 and k3.

Following this, one runs the experiment with both C (the parent),

and D, an additional neutral reactant added. The first set of

constants is entered into the computer and the second set is

determined from them and the data.

Given the above difficulties it might be asked, why react

two species of neutrals at the same time? Usually this is not

done out of choice. When working in mass spectroscopy, espe-

cially icr, the neutral gas from which the ions are formed is

always present--usually in quantities more than sufficient to

react significantly with the ions. So, all too often one is

forced to deal with two neutral reactants in the same experiment.

Most of the C H and C H isomers involved in the work to be
33 5 5
presented exhibit this unfortunate tendency.

Experimental Procedure

Preparation of Instrument

This section describes the experimental procedure. When

one wishes to perform kinetic studies with the icr mass spec-

trometer there is a set routine to be followed. Generally

this routine applies for all of the work presented here.

The only exception is the laser-icr procedure which will

be covered later.

The preparation sequence starts with baking out the

vacuum system. Heating tape raises the temperature of the

spectrometer to over 100 degrees Celsius. This heating is

usually done overnight and at least once a week. Opening

the vacuum can for even the shortest interval, however,

necessitates a bakeout before any more work can proceed.

A day's run starts with charging the liquid nitrogen

traps and closing the Vaclon valve (refer to Figure 2-2).

Usually the Vaclon has been pumping overnight. Within a

few minutes the main gate valve can be opened and pumping

with the oil diffusion pump commenced. The ion gauge is

then degassed for 10 minutes. Finally, within 15 to 30

minutes the high vacuum side of the mass spectrometer
should be at 1 x 10~ torr. If it cannot achieve this

minimum pressure after one hour, either there is a leak, a

virtual leak, or the chamber is dirty and requires a bakeout.
The inlet foreline system usually pumps down to 1 x 10- torr

at the same time.

While the pumping down progresses one activates and pre-

pares the electronic systems. First the main power is turned

on and the computer system booted. The software is input to

the APPLE and the KIM is initialized. Next, one sets the mar-

ginal oscillator frequency. The icr traps ions best at higher

magnetic fields; thus one wants to set the frequency at a value

which allows a field of 10 to 12 kG to be used (see Equation

2.7). Of course, if a sweep is to be performed then it is

best to use 153.57 kHz. At this frequency the icr mass cali-

bration is 10 amu/1000 gauss. There are two advantages here;

first, a wider mass range is available over the usable mag-

netic field; second, numbers convenient to work with are

obtained. Also at this time one checks the cell plate

voltages. Normally for positive ions the side trapping

plates are maintained at +1.25 to +1.50 V. The end plates

can vary between -1.00and-4.00 V. Experience has shown

that end plate voltages do not have much effect on the per-

formance within the limits mentioned. The top and bottom

plate voltages are about -1.00 V. It turns out that the icr

will work within the above ranges, but fine tuning is required

later for each plate voltage. If negative ions are to be

studied, then all of the voltages on the six sides of the icr

cell are inverted in sign.

At this point the magnet and filament are turned on.

Care in this procedure is necessary to avoid undue stress

on the filament. The Fieldial controller is set for a low

value of the magnetic field and the magnet power activated.

Then one should adjust the magnetic. field up to the operating

region. The Kepco filament power supply is then turned on.

One smoothly and slowly increases the filament current, stop-

ping two or three times to allow thermal and electrical sta-

bilization of the system. At a value between 2.0 and 2.5 A,

one should obtain the correct electron beam current reading

from the collector plate. For best results, and often any

results, this beam current should be between 0.02 and 0.10

mV on the scope which, due to the current to voltage cir-

cuitry used, corresponds to an electron beam current of the

same values in uA. Aside from performing adjustments slowly

the main thing is not to turn the magnet on after the filament.

Doing so suddenly stresses the thin rhenium wire which can

cause it to bend or break. Even so, operating at high fields

eventually bends the filament due to the I x 5 force on the

wire. The resulting misalignment can disrupt or eliminate

the electron beam, necessitating its replacement. One suc-

cessful corrective action is installation of the filament

slightly off center, left of the grid hole so viewed from

the side, so that its natural bending moves it back over the

hole. At this point the system is ready for operation.

Preparation of Samples

Samples were prepared, stored, and used in specially

made pyrex sample tubes. These tubes consist of a female

ground glass joint attached to one side of either a teflon

or ground glass stopcock with a large bulb or 6 in. test

tube attached to the other side. The large bulb measures

500 or 1000 ml and is used for gas samples. The test tube

version is used for liquids or solids with sufficiently high

vapor pressure. There are also some designed to hold solid

compounds too. All of these are made in the Chemistry Depart-

ment glass shop.

Most reagents and gases used in this work were obtained

commercially. All samples consisted of reagent or spectral

grade compounds unless otherwise noted. Liquids were simply

pipetted into the cleaned sample tubes. An auxiliary vacuum

line handled the transfer of gases into sample bulbs and was

used to make specific gas mixtures. Dissolved gases in the

liquid and solid samples, usually air, caused serious problems

if allowed to remain. To remedy this, the standard freeze-

pump-thaw procedure was used.

Contaminant gas in gas samples posed a more difficult

problem. If there was a significant difference in boiling

points between the two, they were separated on a vacuum

line. A case in point was air and 1-3 butadiene. Liquid

nitrogen froze the sample but not most components of the

air; thus pumping removed the air. Liquid nitrogen does

condense oxygen, but the pumping immediately removed it

since it is almost at its boiling point. Unfortunately

when both boiling points are close, as with air and methane,

such a procedure-can result in losing most of the sample.

The best cure here is prevention; careful technique in making

the sample is required.

Taking of Data

Taking data starts with establishing an ion signal.

The sample is introduced into the foreline by the manner

described in Chapter 2. It can take a few minutes to sta-

bilize the pressures since the process of sample introduction

is a dynamic one. Typical operating pressures range from

2 x 107 to 2 x 10-5 torr.

At this point the experimental sequence parameters are

entered into the APPLE computer. Figure 2-3 shows the duty

cycle. Usually the grid pulse, allowing the electron beam

to create ions, lasts for 5 ms. Following it is the detect

pulse. One sets it for any desired time after the grid

pulse, depending on the system to be studied. The length

of the detect pulse is typically 3 to 5 ms. Finally the

quench pulse, 5 ms long, is set as close to the end of the

cycle as possible. This might be at 95 ms of a 100 ms duty

cycle. After initiating the pulse sequence, one views the

data being recorded on the APPLE monitor or the actual ana-

log signal itself on an oscilloscope. While doing this, the

magnetic field can be scanned or fine tuned to the signal

peak. At this point it is good practice to fine tune the

plate voltages again.

The creation of too many ions in the cell can lead to

space charge effects. Essentially the charge from the large

number of ions distorts the electric potential lines within

the cell. This causes severe losses of ions and excessive

noise in the signal. Under these conditions the data is

neither accurate nor reproducible. In order to avoid this

problem one first reduces the electron beam current. Some-

times reducing the length of the grid pulse helps too. Also,

different ions behave differently here. For example, the

C3H + species studied extensively was very difficult to keep

under optimum conditions; the C H5 species caused few problems.
5 5
It is speculated that both the large numbers of C H ions cre-
3 3
ated from electron impact on propargyl halides, and their

relative non-reactivity, which caused them to remain trapped

in the cell for long lengths of time, contributed to the dif-

ficulties of tuning the C H signal.
3 3
When performing general survey work, icr mode 1, described

in the section on data acquisition, is used. The magnetic

field is swept over a specified interval while the APPLE

records the signal intensity. In this mode only one delay

time is used for the entire run. This mode primarily gives

information on the location of the spectral peaks and their

relative intensities.

Mode 4, also described earlier, is the most useful for

kinetic studies. First, the peak which was identified in

mode 1 is tuned and conditions for its detection are

optimized. Then the detect delay sweep parameters are set

and the number of passes per point entered. For very slowly

reacting species, the sweep range would extend from 0 to 100

ms or even 1000 ms. In order to have the best analysis of

the reactions, delay times were run as long as possible.

That way if two isomers existed, the faster one would have

a greater likelihood of being depleted in the tail of the

decay curve. Also, longer delay runs insured that a "single"

isomer really was a single isomer; i.e., its signal decayed

to zero.

The computer system's averaging capability provided a

powerful enhancement of the data quality. Yet there was no

firm rule to follow in selecting the number of passes for any

given run. In the end the character of the data determined

it. Although it seemed logical that a high number of passes,

25, would give the best curve, this was rarely the case.

There seemed to be two opposing forces at work. As the num-

ber of passes increased, up to 10 or 15, the data improved,

noise effects averaging out as expected. However, there

usually was a point where the data started degrading again.

Although it is suspected that the electronics may contribute

to this, probably the reason is that experimental parameters

change slightly over time. In order to take just 50 points

averaged 25 times with a duty cycle of 500 ms multiplied

times two because of the baseline points, the time of that

run would have to be over 20 minutes. That can be a long

time for the magnetic field, oscillator, or the filament.

current not to drift.

After determining the detect delay sweep parameters and

the number of passes by trial and error, the actual data runs

were performed. The results were closely monitored; if bad

points due to noise or signal drift appeared, the run was

aborted. Sometimes 15 or more attempts were necessary to

secure acceptable data. In some cases, despite numerous

repetitions, only marginal curves could be obtained. Over-

all, however, the quality of the data was excellent and


Finally, the technique of icr double resonance needs

to be mentioned. By placing an rf burst of the proper fre-

qency on the upper cell plate, specific ions were ejected

from the cell. If they were suspected of being precursors

of another ion, this suspicion was confirmed by the following

procedure. The product ion was tuned in resonance and its

intensity monitored. Then, after the grid pulse but before

the detect pulse, a third pulse called the w2 pulse was

activated. This w2 pulse gated on the power rf burst which

ejected the supposed reactant ion. If the monitored ion dis-

appeared while the w2 pulse was enabled, one had strong evi-

dence that it was indeed the product of the ejected ion's

reaction. The double resonance technique was used in this

work. However, its value was often diminished by large num-

bers of ions and reactions present in the icr cell at one

time. Also the close proximity of two more suspected pre-

cursors led to uncertainty. Sometimes it was difficult to

determine exactly which ion, or ions, were being ejected

under these circumstances. Consequently few double resonance

experiments gave usable information.

Pressure Measurements/System Calibrations


One of the parameters required to obtain rate constants

is the pressure of the reactant neutral. Unfortunately, abso-

lute pressure for the ranges dealt with here is a very diffi-

cult quantity to measure. Inside a vacuum chamber the dynamics

and geometry make this problem many times more complicated.

In addition there are errors associated with the instruments

used to determine pressure. This latter problem will be ad-

dressed first.

Referring to Figure 2-2, one will notice that the pres-

sure in the high vacuum region is monitored by an ionization

gauge tube. This tube is operated by the Varian 862 Digital

Ionization Gauge Control. Electrons emitted by a heated

filament are accelerated to a grid. On the way they may

strike and ionize gases in the region. A collector with

negative bias then attracts these positive ions resulting

in a current which can be measured. The sensitivity of the

gauge depends on a number of factors; most involve the vol-

tages and geometry of the tube. However, not all gases have

the same ionization cross-section and therefore will not pro-

duce the same ion current at the same pressure. The differ-

ence can be almost an order of magnitude (Varian, n.d., p. 19).

For the work presented herein, the ionization gauge was cali-

brated against another pressure measuring device, a Barocel

capacitance manometer. A whole series of calibrations was

performed for each gas employed throughout the pressure

ranges used. This will be covered in depth later along with

the geometry and dynamics problem.

The presence of a magnetic field can also affect the

readings of an ion gauge by deflecting the electrons and

ions within. This effect was observed in the system here.

A magnetic field of 10 kG caused the ion gauge to read 1.15

times its reading with the magnet off. This factor remained

consistent throughout the experiments. For each different

magnetic field intensity, this factor was determined prior

to a run.

Another look at Figure 2-2 shows that the icr vacuum

system is essentially a flow system. The gas enters the

chamber from a rod with holes drilled in it under the cell.

It then passes through an opening in the flange, by the ion

gauge sidearm, by the VacIon sidearm, through 90 degree

elbow, and finally through the gate valve and into the

pumps. Given the hydrodynamics of the system, it is un-

likely that the pressure measured in the gauge sidearm is

the same as the pressure in the cell.

The following equation describes the corrections needed.

Pt = PiCC Cs 3.20

Here Pt represents the true pressure inside the cell and Pi

is the pressure measured by the ion gauge. The correction

due to ionization cross-section is Ci; the magnetic field

correction is Cm; and the overall system geometry correction

factor is C .

Calibration Experiments

The Ci factor was determined by Barocel capacitance

manometer calibrations. This device was attached to a side-

arm located directly above the cell. After preparing the

Barocel and pumping down the icr, the Barocel was zeroed.

Then a small amount of gas was leaked in and allowed to come

to equilibrium by waiting a few minutes. The readings on

both the ionization gauge and Barocel were recorded. Con-

siderable drift existed in the Barocel which required it to

be re-zeroed after every reading on its most sensitive

scale. This and the equilibrium period necessary prevented

one from simply increasing the sample gas continuously and

periodically recording readings. Pressures calibrated

spanned from 1 x 10-5 to 1 x 10 torr. The values of P.

and the pressure from the Barocel were plotted against each

other and a least squares fit performed. The slope was the

Ci correction factor. In most cases the least squares corre-

lations indicated good linear fits. Table 3-1 lists the cor-

rection factors for all substances studied in this work.

The Cm factor was checked daily and each time the mag-

netic field was changed more than 1 kG. Pressure readings

on the ionization gauge with the magnet off were compared to

readings with the field strength set at different values.

Below 7 kG no significant effect appeared after the field

had stabilized. Cm was 1.05 at 7 kG; 1.15 at 10 kG; and

1.18 at 12 kG.

The method of determining the C factor involved obtain-

ing a rate coefficient for a well known reaction, i.e., mean-

ing one for which the rate coefficient has been measured to

a high degree of precision.

Methane Reactions

The rate coefficients for the following two reactions

nave been determined. These reactions are very well suited

for study in mass spectrometry due to their high rate coeffi-

cients and the fact that they are the only reactions occurring

for low energy electron impact on methane.

+ +
CH + CH+ -- CH + CH k5 3.21
+ + 3
CH + CH -- C2H+ H2 kg 3.22
3 25 2 6

Well over a dozen different researchers have obtained values

for k ; about half as many have done so for k6. For complete

Table 3-1. Ci Correlation Factors of Various Substances
for Use in Equation 3.20.

Substance C. Correction Factor

C2H2 acetylene 2.4
C2H4 ethylene 2.0
C3H8 propane 1.5
CH4 methane 2.9
C4H6 1-3 butadiene 1.4
C3H allene 2.6

C6H6 benzene 1.0
C7H toluene 1.0
C10H8 napthalene 4.0
C11H10 1-methylnapthalene 4.0
C11H10 2-methaylnapthalene 4.0
C H indene 3.9

C5H6 cyclopentadiene 1.8
C10H12 dicyclopentadiene 1.3
C H norbornadiene 1.9

CO carbon monoxide 4.0
C3H3CH propargyl chloride 2.0
C2H5Br bromoethane 2.0

All C. correction factors are ratios; PBarocel)/Po Gauge)
1 (Barocel) (Ion Gauge)

details concerning these numbers the reader is referred to

the following references--Clow and Futrell, 1970; Huntress

and Pinizzotto, 1973; Huntress, Laudenslager, and Pinizzotto,


The values for both k5 and kg show excellent agreement.

Those of k5 range from 1.10 x 109 cm3/mol s to 1.20 x 109.

Those of kg range from 9.6 x 10-10 to 1.00 x 10-10. This

being so, the numbers of Huntress and Pinizzotto will be

used since they are the latest and were determined in an icr.

They are

k5 = (1.14 0.03) x 10 c /mol s

k = (0.96 0.04) x 109 c/mol s

Both reactions were run in the icr on two separate dates.

The methane and methyl ions were made by 70 eV electron

impact. Pressures, as given by the ion gauge, ran from
-7 -6
5.0 x 10 to 1.0 x 10-6 torr, yielding two of the cleanest

sets of data obtained. The analysis proved to be relatively

easy. Table 3-2 shows the results. In this table the calcu-

lated rate coefficients k5 and k6, as well as their averages,

are given for each pressure (ion gauge). Equation 3.20 was

used, employing Cm of 1.05 appropriate for the magnetic field

used, and a Ci of 2.9 for methane. A value of 1.0 was used

for C

Both of the data averages demonstrate excellent agreement

for the two different days. For the first set the ratios of

k5(exp)/k5(lit) and k6(exp)/k6(lit) both are 1.1. For the

Table 3-2. Methane Calibration Results.

CH + CH4 Reaction


CH + CH Reaction


Set 1

10.0 x 10-7 torr
10.0 x 10 torr








10.0 x 107 torr




Set 2

9.8 x 107 torr
9.8 x 10 torr





k5 (1st set)
(2n set)
kg (1 set)
nd set
(2 set)

All rate coefficients in

cm3/mol s x 109

C. and C factors included in coefficients listed.
I m

Set 2 was taken two weeks after Set 1.














1.3 x
1.3 x
1.0 x
9.9 x


second set these ratios are 1.1 and 1.0 respectively. Refer-

ring back to Equation 3.11, it can be seen that these ratios

are also the factors by which the pressures must be multiplied

in order to make the experimental values match the literature

values; that is, they are Cs. The average of these four ratios

is 1.1. This then represents the C which must be used in all

succeeding experiments to correct for the system geometry.

Such a low value also implies that these different geometric

factors cancel out to a large extent in this system.


The absolute pressure is the least accurate of all the

quantities involved in using icr to determine an ion-molecule

rate coefficient. Three separate corrections need to be made

on the value directly read from the ion gauge. Other factors,

besides pressure, do not have as much uncertainty. Time, during

.the experiment, is controlled by the computer and is accurate

to 1 ppm, an insignificant error. Error introduced by the in-

herent nature of the electronics, such as in the A/D converter,

is much smaller than that of the signal noise. However, the

effect of signal noise is markedly reduced by the signal aver-

aging and least squares method of curve fittings. This error

should not be more than a few per cent. Obviously the C. and

C corrections are larger, but not by much. The quality of
the experimental k5 and k6 values show that under good condi-

tions, rate coefficients calculated from icr data under good

conditions can contain much smaller overall errors than one


Some of the studies within this work were done under

considerably more difficult conditions than the methane

calibrations. Often the nature of the species forced one

to work on the verge of space-charge effects in order to

see the desired reactions. Also, experimental parameters

drifted, as they will sometimes do when complicated equip-

ment is involved. Despite all attempts to minimize these

problems, in some cases the data suffered. In icr work,

especially with new reactions, the overall general error

of rate coefficients determined is usually considered to

be in the 20 to 30 per cent range.



Until recently scientific examination of flames and other

combustion processes has been limited primarily to engineering

aspects and some of the more gross characteristics of the

phenomena taking place. Thus, these early studies covered

temperature measurement, heat transfer, fuels, stoichiometry,

combustion products, burner design, explosion properties, etc.

Application of new analysis techniques such as mass spectrom-

etry and laser spectroscopy now allows study of the numerous

species and reactions within the flame itself. Such studies

also have given rise to a number of schemes attempting to ex-

plain the formation of soot. Originally neutrals and free

radicals were thought to be responsible for this. Lately,

evidence has mounted indicating not only that the postulated

reaction paths involving these species are flawed, but also

that soot formation follows an ionic route from small primary

ions up to large charged particles which are aromatic in

nature. In this chapter a brief discussion of flames is pre-

sented and the above mechanisms are examined. Then details

are given concerning the structures and chemistry of two of

the initial, important species in the ionic pathway to soot,

C H+ and C H.
3 3 5 5*

General Discussion of Flames

Definition and Nature of Flames

The combustion process occurs in many forms, ranging from

a smouldering ember to a detonating stick of dynamite. Any

form must have as its basis a self-supporting exothermic chem-

ica-1 reaction. Usually such a reaction involves oxygen, but

not always. In the simplest systems there is just fuel and


The ratio of fuel to oxidant turns out to be very impor-

tant in describing a flame. This is generally expressed in

terms of the equivalence ratio

(Fuel/Oxidant)actual 4.1

If the equivalence ratio is less than 1, the flame is consid-

ered to be lean; if the ratio is greater than 1, the flame

is rich. The process of increasing this ratio, or mix, both

cools the flame and eventually leads to the formation of soot.

The reaction of the two components in a combustion sit-

uation can occur under a number of conditions. It can take

place in a combustion chamber, as in the internal combustion

engine, or in free space. If the reaction is or becomes auto-

catalytic, an explosion results. The fuel can be either solid,

liquid, or gas. In order to have a flame, one important re-

quirement is that the combustion reaction has to be able to

self-propagate through space. However, it must also be

stabilized. An example of an unstable flame is when the gas

flow to a Bunsen burner is adjusted too high; the flame leaves

the tip of the burner, flickers, and may blow itself out.

Studying flames necessitates that one have a stable, steady

flame. To achieve this, one needs a good burner which prop-

erly mixes the fuel and oxidant, provides the correct flow,

and stabilizes the flame. In all of the studies of flames

to be cited below, a stable, premixed flame on a burner was


Species in Flames

What is in a flame? Recent research has given both

qualitative and quantitative results describing the species

present in a flame. Of highest concentration are the neutrals

and free radicals. Then come ions and soot particles. When

discussing any of the proceeding components, it is necessary

to specify the equivalence ration since drastic changes in

species concentration occur between lean and sooting flames.

Also, one has to define the regions of the flame in order to

be accurate. J. M. Goodings et al. (Goodings, Bohme, and

Chung-Wai Ng, 1979) label the regions as shown in Figure 4-1.

Region A is upstream of the reaction zone tip and region D

is referred to as the downstream portion. For a lean flame,

there is only one other region, B, the reaction zone, extend-

ing approximately 0.3 mm upstream of the burner tip to approx-

imately 0.05 mm downstream. In the case of a rich or sooting

flame, a fourth region C is added immediately downstream of B.

There, in C, species and reactions not observed in the lean

flame are present. Finally, in the descriptions to follow,

neutral, ionic, and free radical species found in either



(C) (rich and
sooting flames






Figure 4-1. Reaction Regions in a Flame.

acetylene and oxygen or methane and oxygen flames are


In a lean flame, 4 = 0.2, the fuel and oxidant are

found throughout the flame, along with the products, CO2

and H20, found downstream. In region B, one can find HCHO,

HCOOH, CH OH, CH OCH, CO, H2, and ketenes. Numerous free

radical species are also observed. These include CH3, H02,

CH302 in the reaction zone and 0, OH, and H downstream.

Present too are a number of positive ions. The most preva-

lent is the CHO ion. This primary ion, formed upstream in

region A, is believed to be made by chemi-ionization.

CH + 0 ---- CHO' + e- 4.2

Other positive ions can then be formed from CHO+ through a

number of reactions.

CHO+ + X --+ HX+ + CO proton transfer 4.3

X + HCO charge transfer 4.4

--- rearrangement procedure 4.5

CHO+ + M + X --- CHO+:X + M clustering 4.6

CHO :X + Y -- CHO+: Y + X switching 4.7

Through these reactions, of which proton transfer is the

most likely, larger positive ions are formed. Some of them

found in the reaction zone (Goodings et al., 1979; Hayhurst

and Kittelson, 1978), are CH 0 CHO 0 CH 0, C H3, and
2 3 5 2 7 3'
CH 0 These larger ions then are oxidized in the reaction

zone by radicals such as O, OH, and H so that downstream

only H30 and once again CHO+ are seen. These ions then can

be neutralized by recombination with free electrons or nega-

tive ions. As an aside it must be noted that negative ions

also exist in flames. However, they are not believed to play

an important role in the formation of soot and very little

will be said about them in the discussions that follow.

In the C region of rich and sooting flames, a species

with mass number 39 appears in overwhelming magnitude. This

species has been identified as C H+, which is suspected to
3 3'
show two structures, linear and cyclic. Two reactions have
been proposed for the formation of C H The first starts
3 3*
with CHO which then reacts with CH2CO to form C2H30+.

Then follow

C2H 0 + --- C2HO + H2O 4.8

+ +
C2HO + C2H ---- C H + CO 4.9
2 22 33

Large amounts of acetylene are generally seen in rich and

sooting flames, even when a fuel other than acetylene is

being burned. Reaction 4.9 probably does not account for

all of the massive amounts of C H seen in rich or sooting
3 3
flames; thus Calcote adds (Calcote, 1981)

CH + C2H C H + e 4.10
2 2 3 3

to account for the excess C H3. Calcote notes that there is

little proof of Reaction 4.10 so far and that it deserves

further study. Also noted is that Bowserand Weinberg (Bowser

and Weinberg, 1976) have suggested that C H3 formed in pyroly-
sis arises from
sis arises from

C2 + CH --- C H + e 4.11
3 3 3

More evidence is needed to determine whether Equation 4.10

or 4.11 account for the extra C H present.

Also in the C region, more and larger hydrocarbon neutral

and radical species are seen than in the fuel lean case due to

the deficiency of oxygen. These include C2H2, various poly-

acetylenes, CO, and the fuel. Correspondingly, there can be

larger ionic species built up from the neutral and radical

building blocks now present. Some of them are C H, C H5,
5 3' 5 5'
+ +
CH and C H. Ion-molecule reactions involving C2H are
7 7' 6 5 2 2
believed to be responsible (Calcote, 1981).

C H ++ CH -- C5H+ H2 4.12
33 225 3 2

--- C H 4.13
+ +
C H + CH --- C H+ 4.14
5 5 2 2 7 7


Other reactions of different ions with the polyacetylenes

account for the even number carbon atom species such as
+ + +
C H5, C H6, C6H etc. Further build-up to ions with nine,
4 5' 4 6 6 3'
11, 13, and 19 carbon atoms, probably aromatics, is seen

(Calcote, 1981). Finally, these species persist even down-

stream where the usual recombination reactions occur.

Since there is not enough oxidant to complete the oxidation

of all the hydrocarbons, a large number of products result.

In the sooting flame, ( = 2.5 for C2H2, there is even

less oxidant; therefore more carbon products are produced.

Eventually these products change from molecules (or ions) to

macromolecules (or ions) and nucleate into discrete particles

which then continue to grow into what is commonly referred to

as soot. It is to the early steps of this mechanism of soot

formation that this research is addressed. The rates of the

ion molecule reactions of C H and C H isomers with various
33 55
neutral gases are studied. Next follows a general overview

of the different mechanisms proposed to explain the formation

of soot.

Mechanisms of Soot Formation

Neutral and Free Radical Mechanisms

A large number of neutral and free radical mechanisms

have been proposed which attempt to describe the path from

small molecular species to soot. Most of these have been

named and have been well discussed before. The following

references should be consulted for more details--Palmer and

Cullis, 1965; Lahaye and Prado, 1978; Gaydon and Wolfhard,

1979. Of the paths proposed, only a few seem to be free of

thermodynamic or kinetic flaws. In fact they are becoming

less accepted not only in their own right, but also when com-

pared to ionic mechanisms which are gaining prominence.

Finally, as Calcote points out regarding neutral mechanisms

(Calcote, 1981, p. 227), "Most discussions culminate with

complex paths from the fuel via acetylene(s) to soot. This

. is an indication of frustration; one single path which

might, of course, vary from system to system would be more

intellectually satisfying." These neutral and free radical

paths will be briefly reviewed. For more information, a re-

view by Calcote should be consulted (Calcote, 1981).

Essential to any discussion of soot mechanisms is the

nature of soot itself. Observations on soot particles in

flames (Calcote, 1981) indicate that it is a collection of

chainlike spherical units approximately 10 to 50 nm in

diameter. These units exhibit a graphitic structure and

have a carbon-to-hydrogen ratio varying from 8:1 to 12:1;

thus calling soot simply "carbon" is inaccurate. The

graphite-like structure implies that in the steps of soot

formation, ringed structures, aromatics, must form. These

smaller aromatics must then grow in size through successive

addition reactions to give large polycyclic aromatic hydro-

carbons, PCAH, which grow further to form eventually graphite-

like structured species.

The growth of a soot particle requires three steps.

First, nucleation; there must be a transition from a molecu-

lar system to a particle system. Then, the particle must

grow into the 10 to 50 nm sphere. And finally, these par-

ticles must aggregate together into long chains (Calcote,

1981). There is somewhat of an analogy to the buildup of

polymeric substances here.

One of the most attractive paths for soot formation is

the one through polyacetylenes. Here C2H2 reacts to form

higher species


62 82
C 2H2 --- C4H --- 4.15

CH C6H3 C8H3

The reaction continues to form even large molecules. This

mechanism, however, does not lead to PCAH's and cannot be

used to explain soot formation. Rearrangements would be too

slow and the thermochemistry of the later addition reactions

required becomes unfavorable as the size of the species be-

comes larger (Tanzawa and Gardiner, 1979).

A variation of this (Homann and Wagner, 1967) has radi-

cals such as C2H attacking the polyacetylenic species and

forming side chains leading to branched polymerizations and

presumably ring closures. Again, among other problems, there

is still the slow growth of the polyacetylenes; also the spe-

cifics of the ring closure have not been addressed. It too

has been suggested that butadiene-type species may close to

form rings (for references, see Calcote, 1981).

Some models have C2 and C3 radicals combining with each

other or with C2H and C2H2. This then leads to five carbon

atom species which further react. The main objection to

these paths is that while it is true that both C2 and C3

radicals are seen in sooting flames--they are also seen in

lean, non-sooting flames. They give very strong and distinct

spectral lines (for references, see Calcote, 1981).

A final category starts with ringed species and has the

acetylene,polyacetylenes, and free radicals adding on to

form more rings. Although some data exist, there is not

enough yet. It is believed that there may be either thermo-

dynamic or kinetic bottlenecks present. At one point in the

chain it may be thermodynamically unfavorable to proceed to

larger rings or the reaction may be too slow. Also, decom-

position paths could become more favorable.

Overall not one really good neutral or free radical

model exists. There seems to be some promise in a few, but

much more experimentation needs to be done. On the other

hand, ionic mechanisms have been growing in favor lately.

A better model seems to be emerging from recent study.

Ionic Mechanisms

Overall the ionic mechanism is similar to those involv-

ing neutrals and free radicals. To begin, the three essen-

tial steps, nucleation, growth, and agglomeration remain the

same. However, the difference is that the initial building

blocks, the primary ions and such, along with the PCAH and

soot particles, are charged species. Figure 4-2, reprinted

from Calcote (Calcote, 1981, p. 216), is a schematic diagram

showing this evolution of soot from primary ions. The primary
+ +
ions, C H and CHx species, combine with small neutrals such
33 5x
as C2H2 to form larger ions, eventually leading to polycyclic

ions. There is also a chance that they may directly combine

with larger neutrals present as a result of other reactions

in the flame. The PCAH then nucleate and grow in size and

weight forming spherical particles.






Growth from
the nucleating agent.

primary molecular species to soot aggregation, assuming ions as

Reprinted by permission of the publisher from the article
by H. Calcote, Combustion and Flame 42: 215-242, p. 216,
copyright 1981 by the Combustion Institute.

Figure 4-2.

Growth of Soot Particles from Primary Molecular








10 7
106 Z

105 3


*Tany of the objections to the neutral and free radical

mechanisms do not manifest themselves when the reactants are

ions. For example, the rearrangements and other reactions

necessary for ring closure proceed readily. They are both

kinetically and thermodynamically favored (Calcote, 1981).

Recent research has dispelled one of the primary objec-

tions to the ionic path, the small relative concentration of

ions in the flame compared to neutral and radical species.

Figure 4-3, reprinted from Calcote (Calcote, 1981, p. 225),

shows the concentration profiles of key neutral and ionic

species as a function of distance above the burner. Several

significant conclusions can be drawn from this data. First,

although one notices that the concentrations of ions are

orders of magnitude lower than those of neutrals, the soot

concentration is lower than that of the large positive ions.

There are probably enough of these large positive ions to

react with the substantially larger reservoir of neutrals

available to form the levels of soot and soot particles

observed. Other inferences can be made from this figure.

It can be seen that the buildup and decay of relevant species

are consistent with the proposed scheme. The precursors

rapidly peak, as do the ions of 300 to 1000 amu. Meanwhile

the large positive ion curve rises slower, but it peaks just

after the initially very rapid drop of the 300 to 1000 amu

curve. Finally the rise in the soot curves begins later,

after the large positive ions and 300 to 1000 amu ions ap-

proach their maxima. The soot curves then peak downstream

following the decay of the large positive ions.

,1016 C2H2
I ------__ --


1014 CgH2 -


IONS. 300-1000

101 -

0 1 2 3 4 5 6 7
Concentration profiles. 02, CO, C2H,, H80,
C14H8, and "precursors" from Homann and Wagner
[16, 33); C4H,, C6H2, CgH, from Bonne et al. [28|;
large positive ions and charged soot particles from Prado
and Howard (5, = 3.0, u = 38 cm s-1ions, 300-1000
anu from Olson and Calcote [17]; determination of
curve shape by mass spectrometry, concentration esti-
mated by comparison of results with Prado and Howard
[51, Delfau et a. [131,and ersborg et al. [ 11] (see dis-
cusslon following Ref. [17]); soot number density below
3.5 cm from Wersborg et al. [], O = 3.0, u = 50 cm s-.1
Bonne et al. 28O soot number densities continue to rise
to 6 x 1011 cm-3 at 2.8 cm. Above 3.5 cm the data of
above two goups, which were very close, have been

Reprinted by permission of the publisher from the article
by H. Calcote, Combustion and Flame 42: 215-242, p. 225,
copyright 1981 by The Combustion Institute.

Figure 4-3. Densities of Various Species found within



Questions have been raised concerning the first steps

of this proposed ionic mechanism. These first steps may not

occur fast enough to account for the rapid buildup of ions

in the 300 to 1000 amu range. It is argued that the curves

of Figure 4-3 could have other explanations.

Specifically in question are the initial reactions of

C H with CH2 to form the C H species and the further se-
33 i2 5 x
quential reactions of C H ions to give CH then CH ,
5 x 7 x' 9 x'
etc., on up to larger ringed species. Equations 4.12, 4.13,

and 4.14 give these, reactions. Are their rate coefficients

fast enough? If they are not fast enough, can the C H+ and
3 3
perhaps the C Hx react directly with polyacetylenes and the
large numbers of other species such as toluene, indene, and

napthalenes found in flames to form the larger ions? Can

both paths be invoked simultaneously?

The icr mass spectrometer, discussed in Chapter 2, is

well suited to study the reactions involved here and answer
+ +
some of the above questions. Both C H and C H5 ions can be
33 55
made by either electron impact or charge exchange reactions

and then allowed to react with any of the above mentioned

neutrals. The rate constants can be determined, albeit with

considerable difficulty in many cases. Then some answers or

additional specific and directed questions may follow. In

the next section, theoretical details concerning the C H
3 3
and C H5 ions studied are presented. Much interesting
physical-organic chemistry regarding them exists.
physical-organic chemistry regarding them exists.

C H+ Structures and Discussion
3 3
Isomeric Structures

It is the intention in this section to discuss the struc-

ture and nature of the critical C H3 species. Until recently
3 s
there has been little theoretical or experimental information
regarding C H Now, more attention has been given to its

chemistry because of its suspected role in soot formation.

There exist eight reasonable structures of CH As
3 3*
any chemist would know, when dealing with the properties of
+ +
C H and later those of C5H5, the specific structure is
3 3' 5 5
critical to reactivity and other properties. Figure 4-4 on

the following page illustrates and labels each of the eight

isomeric structures discussed by Radom (Radom, Hariharan,

Pople, and Schleyer, 1976). Radom and his co-workers have

done ab-initio calculations on the geometries and energies

of each structure.

Of all eight structures, only (I) and (II) are really

stable; the others, (III) through (VIII), are not likely to

be found experimentally. It is also possible that if they

existed they would rapidly convert into forms (I) or (II)--

probably without activation energy. This leaves only the

cyclopropenyl (I) and propargyl (II) cations for discussion.

These isomers are approximately 1.5 eV apart in relative

energy, the cyclopropenyl cation the most stable. Isomers

(III) through (VII) each successively lie 1.5 eV higher.

Radom predicted the 1.5 eV difference from his STO-3G,

4-31G, and 6-31G bases ab-initio calculations. His cal-

culated heat of formation of 253 kcal/mole for the

/ --



:C + C


H\ +
C ---C ==C-H



C C+
, C C





,C C:

Figure 4-4. The Eight Isomers of CH .
3 3*




H/ /t ,,, ,,,,H


cyclopropenyl cation agrees well with the recent experimental

value of 256 2 kcal/mole determined by Lossing (Lossing,

1972). Thus the linear-C H ion, the propargyl cation, has a
3 3
heat of formation of 287 kcal/mole.

Aromaticity and Antiaromaticity

By now the alert reader will have noted that the cyclic

C3H ion represents the first example of an aromatically

stabilized species. Furthermore, one might also recognize

that the C H ions, if cyclic, are examples of antiaromatic
5 5
species. At this point it is prudent to digress in order

to review the topics of aromaticity and its opposite,


The simple case of benzene illustrates aromatic

stabilization. To be aromatic, a molecule or ion must

have molecular orbitals containing 4n + 2 delocalized T

electrons in orbitalslocated above and below the planar

ring of the species. Such species exhibit striking stabil-

ity and considerably less chemical reactivity than their

nonaromatic analogues. Typically, non-aromatics similar

to benzene such as cyclohexane, cyclohexadiene, and cyclo-

hexatriene undergo rapid oxidation, additions, and

hydrogenations. Benzene undergoes only slow hydrogenation

at high temperatures and pressures. Mainly it experiences

substitution on the ring. It is estimated that benzene is

approximately 36 kcal/mole lower in energy than its analogue,

cyclohexatriene. This is defined as its resonance stabiliza-

tion energy.

The cyclic C3H species has two T electrons above and

below its ring. Even though a three-membered ring is highly

strained, cyclopropane is a stable compound and the cyclo-

propenyl cation is also known to be a stable structure

(March, 1968). Aromatic resonance stabilization is re-

sponsible for this stability. Consequently it then should

be expected that the cyclopropenyl cation will exhibit reac-

tivity behavior similar to other aromatic species. Although

an absolute argument cannot be made that it will react slower

than a linear C H cation because aromaticity only compares a
3 3
species with cyclic analogues, it is generally believed that

cyclic C H should be the slower reacting cation of the two.
3 3
Opposite to aromaticity is the concept of antiaromaticity.

In this case cyclic species which have 4n r electrons exhibit

destabilizing cyclic conjugation. Two types of antiaromaticity

have been advanced, relative and absolute. Relative antiaro-

maticity is when the cyclic system is less stable than a con-

jugated acyclic analogue. An absolute antiaromatic system is

one that is even less stable than its non-conjugated analogues

(Bauld, Welsher, Cessac, and Holloway, 1978). It should be

mentioned that the entire concept of antiaromaticity is still

somewhat theoretical although Bauld points out that evidence

seems to confirm relative antiaromaticity in some small species,

including cyclic C H ions. However, there is no evidence
5 5
establishing absolute antiaromaticity yet (Bauld, Welsher,

Cessac, and Holloway, 1978).

ICR Studies of C H
3 3
The first icr evidence demonstrating the existence of
two C H structures was obtained by Ausloos and Lias (Ausloos
3 3
and Lias, 1981). They created both structures by charge

transfer reactions

+ +
M+ CH X C H + M + X 4.16
33 33

The charge exchange gas M varied; they used C02, Xe, Kr, and

others. The parent neutral was either propargyl chloride

or bromide.

Propargyl chloride gave mostly the non-reactive, cyclic,

ion. Propargyl bromide gave both linear and cyclic ions;

the ratio depending on the charge exchange gas for both the

chloride and bromide parents. The proportion of linear ion

reached a maximum when the ionization potential of the

charge exchange gas was 2 to 3 eV above that of the prop-

argyl bromide.

C H Structure and Discussion
-5 5
Isomeric Structures

Two of the more important intermediates in the ionic
+ +
sooting mechanisms are C H5 and C H In the flame, both
5 5 5 3*
species have been identified by mass spectroscopy (Michaud

et al., 1981). It remains to be determined which of them

is more abundant in flames, or which of them is more impor-

tant to the soot formation mechanism, if either. The C H
5 5
cation more readily lends itself to investigation since it

can be created easily in the mass spectrometer. The CsH+
5 3

ion can be made, but its intensity is far smaller than that

required for any productive reactivity studies with it.

There are many possible structures for C5H certainly

exceeding the eight of C3H Again, when discussing the

properties of C H it is necessary to specify the isomer, or
5 5
suspected isomer. Some of the structures considered in this

work are drawn in Figure 4-5. For future reference they are

each labeled according to a scheme. An L prefix, such as

(L-l), denotes a linear isomer. The T prefix denotes a three-

member ring; the F denotes a four-member-- ring; and a V

denotes a five-member.e ring. A B denotes a bridged structure.

A p suffix indicates that the structure is planar; a T suffix

means that the cation is the triplet, otherwise it is the
+ rI
singlet. In the C5H cation, two electrons are in the el T

molecular orbital. Consequently the low lying electronic

states are: 3A2, 1A1, and a doubly degenerate 1E2 (Borden

and Davidson, 1979). These species are by no means all of

the possible structures. Many can have variations in the

placement of the hydrogens, the + charge, and the placement

of bonds or r electrons.

Despite a number of theoretical and experimental studies
done on the C H5 system, little could be said regarding the

specific relative energies of the isomers. In experimental

work, heats of formation have been obtained for the species

created. Unfortunately, no one has ever actually, positively

identified just which C H isomer they had been studying!
Sometimes, however, it was assumed; this "obviously" must be
Sometimes, however, it was assumed; this "obviously" must be




linear +

0 a------ --





bent up 250




Figure 4-5. Some Structures of the C H System.
5 5




Figure 4-5. continued


(T-l) (T-2)



(F-1) (F-2)

(B-1) (B-2)



the linear, that the five-membered ring. Such assignment

turns out to be premature considering the lack of evidence

and myriad possibilities which exist. Theoretical studies

primarily compared one structure with another. Some used

ab initio calculations; others used semi-empirical methods.

However, no single method has been used to give a comprehensive

ordering for all or most of the structuresof prime importance.

This leads to the proverbial comparison between "apples, or-

anges, and bananas" when attempting to incorporate all results

of these studies into one single energy scale for the isomers.

Past Experimental and Theoretical Studies

A number of experimental studies have been performed on

the C H system. Initially they involved determining the ap-
5 5
pearance potential via mass spectrometric methods (Dorman,

1965). Occolowitz and White (Occolwitz and White, 1968)
created the C H cation by electron impact from several
5 5
sources and calculated the heat of formation. Specifically

they obtained: 271 kcal/mole from cyclopentadiene, 227

kcal/mole from aniline. Then, using an appearance potential

from Pottie and Lossing (Pottie and Lossing, 1963) for the

cyclopentadienyl radical, they also calculated a heat of

formation for the planar C H5 cation of 240 kcal/mole.
5 5
Assuming that 3-penten-l-yne had to produce the linear ion

(L-2 or L-3) and pointing to the similarity of the heats of

formation of the cations produced by all of the parents,

it was concluded that they all were linear structures.

However, they noted that there could be an error of 20 kcal/

mole due to uncertainties in the experimental method.

Harrison et al. studied the mass spectra of methyl-

substituted cyclopentadienes (Harrison, Haynes, McLean, and

Meyer, 1965). They noticed that two distinct groups of heats

of formation seemed to emerge. One group ranges from 280 to

290 kcal/mole and the other ran from 300 to 310 kcal/mole.

Suspecting that something was wrong and that these numbers

were too high (they had estimated the planar cation to have

a AHf = 270 kcal/mole), they wisely abstained from drawing

any conclusions.

Franklin and Carroll obtained more heats of formation

in a similar manner (Franklin and Carroll, 1969). Here again

two groups emerged. Cyclopentadiene gave 271 kcal/mole;

l-penten-3-yne gave 288 kcal/mole; and toluene gave 290

kcal/mole. However another set of parents consisting of

butadienylacetylene, methylcylopentadiene, cycloheptatriene,

2,4-hexadiyne, and 1,4-cyclohexadiene gave numbers in a range

of 299 to 309 kcal/mole. They likewise declined to make any

structural assignments. Tajima and Tsuchiya also observed

a similar set of data from electron impact on a series of

halogen-substituted toluenes (Tajima and Tsuchiya, 1973).

A more recent value of 255 kcal/mole for the heat of

formation for the (V-l) planar cation has been obtained by

Lossing and Traeger (Lossing and Traeger, 1975). They re-

mark that this number is to be preferred over the previous

numbers obtained by using dissociative ionization thresholds.

By ionizing the cyclopentadienyl radical with an electron

monocromator-mass spectrometer, these values for the ioniza-

tion threshold are much more accurate.

Finally, McCreary and Freiser calculated a value of 239

5 kcal/mole from icr photodissociation experiments (McCreary

and Freiser, 1978). The C H+ cation loses C2H to form C5H5
7 7 2 2 5 5
upon uv irradiation. This unexpected photodissociation was

noticed while studying benzyl and tropylium cations formed

from toluene and toluene derivatives.

This then is the situation concerning experimental heats

of formation for the C H5 system. With the exception of the

two, later and better, numbers of 239 and 255 kcal/mole, there

are two divisions, 270 to 290 kcal/mole and 300 to 309 kcal/mole.

As pointed out by Lossing (Lossing and Traeger, 1975), the ear-

lier numbers are all probably too high due to the dissociative

ionization methods used which can leave the products with some

internal excitation. In general, no conclusive statements can

be made based on the earlier evidence. However, the values of

239 and 255 kcal/mole have less uncertainty. This indicates

that the five-membered ring may have a heat of formation close

to 255 kcal/mole. It is not known what structure the 239 kcal/

mole figure applies to.

A number of interesting theoretical studies on some of

the C H5 species have been done. Stohrer and Hoffman (Stohrer
and Hoffman, 1972) performed extended Huckel calculations on

a family of the C H cations. They first suggested the exis-

tence of the pyramid, (F-2) structure. Presumably, the singlet

(V-lp) distorts to either (V-3p) or (V-4p), both shown by

Borden and Davidson to be equivalent, changing through a

pseudo-rotation (Borden and Davidson, 1979). It then bends

to structure (V-5) whose dihedral angle goes past 90 degrees.

From there the pyramid is formed.

All of the above mentioned ions, and others, were pur-

sued further. Kollmar, Smith, and Schleyer performed CNDO

studies (Kollmar, Smith, and Schleyer, 1973). Dewar and

Haddon did MINDO/3 studies (Dewar and Haddon, 1973), and

Hehre and Schleyer did ab-initio studies (Hehre and Schleyer,


The main conclusions were that there should be two minima

on the hypersurface of the sequence proposed by Stohrer and

Hoffman. However, there was disagreement on which species

was lowest in energy. Kollmar, Smith, and Schleyer concluded

that either of (V-lP), (V-3P), or (V-4P) represented the first

minimum and that the pyramid (F-2) was the most stable and

represented the second minimum. Dewar and Haddon predicted

that the bent form (V-5) at an angle of 23.6 degrees (255.3

kcal/mole) and the pyramid (269.7 kcal/mole) were the most

stable; this time the pyramid being of higher energy. Both

also predicted an energy barrier between (V-5) and (F-2);

Dewar's approximately 43 kcal/mole.

Further studies were done on the D5h planar singlet.

The planar singlet (268.5 kcal/mole) (V-lp) undergoes Jahn-

Teller distortion without activation energy. Its energy

levels are considerably lower in one of the three C2v forms,

(V-3p), (V-4p), or the bent (V-5). Dewar (Dewar and Haddon,

1973) placed (V-3p) at 257.2 kcal/mole and (V-4p) at 256.9

kcal/mole. Kohler and Lischkafurther refined the numbers

and determined that the difference between form (V-4p) and

(V-5) was only 0.7 kcal/mole (Kohler and Lischka, 1979).

MINDO/3 Background

MINDO/3, developed by Dewar (Bingham, Dewar, and Lo,

1975) is a semi-empirical SCF MO method designed to opti-

mize geometries and bond lengths, and then give heats of

formation. The current MINDO/3 program at the University

of Florida accommodates all atoms from H to Xe. The results

obtained from MINDO/3 have been validated by Dewar and others.

Studies indicate that in general, heats of formation can be

calculated within 5 kcal (Dewar, 1975; Bingham, Dewar, and

Lo, 1975) and that geometries are accurate to within a few

degrees, bond lengths to within 0.002 Angstroms. MINDO/3,

in addition, gives dipole moments, polarizabilities, and

charge densities, but these, while at least as good as those

obtained from other methods, are not optimized.

The calculation of geometries proceeds via the varia-

tional method of Murtagh-Sargent. Derivatives of the energy

with respect to coordinates are taken. Optimization procedures

are then implemented on the basis of these gradients; adjust-

ments in the atomic coordinates result. Heats of atomization

are derived from comparing the energies of these separated con-

stituent atoms to the energies ofthe system. These results are

converted to heats of formation by use of a table of actual

heats of formation.

ICR Studies of C5H System
5 5
The only icr work on C H has been done indirectly as
5 5
in the previously mentioned photodissociation study of C H
7 7
by McCreary and Freiser. Ausloos (Ausloss, 1982) also stud-

ied the C H cations formed in the charge-transfer-induced

fragmentation of ethylbenzene, toluene, and nonbornadiene.

Two different structures of the C 5H produced were seen.

One, formed from the ethylbenzene, toluene, and the non-

bornadiene, reacted rapidly (10-9 cm3/mol s) with its re-

spective parent. A second, produced by the nonbornadiene,

reacted very slowly. The fraction of this non-reactive

isomer increased from 19 to 50% when Ar+ replaced Kr+ as

the charge exchange gas. It was with this sparse icr back-

ground that these present studies on the reactions of (CH)+

were undertaken.


CH + C2H System
-2-2 2-2

An initial investigation of ion-molecule reactions, sim-

ilar to those found in flames, involved a sequential series

of condensation reactions, starting with acetylene reacting

with its molecular cation. These studies were carried out

both on the icr at the National Bureau of Standards, Wash-

ington, DC, and the icr, previously described, at the Uni-

versity of Florida.

The reactions

+ +
C2H + C --- CH2 + H2
C 2 222 4 2 2

CH+ +H 5.1b
4 3

had been studied earlier and the rate coefficient for the

disappearance of C2H found to be 1.41 x 10- cm /mol s,

reasonably fast (Huntress, 1977). Further reactions of
+ +
the C4H and C H2 ions at a similarly fast rate might sug-

gest that such a sequence could indeed be important to

sooting flames.

These experiments were performed on the icr at the Uni-

versity of Florida prior to establishment of the microcomputer

control and data system. Descriptions of the manual techniques

utilized to study ion-molecule reactions have been published

earlier (Lias, Eyler, Ausloos, 1976). At this time, the UF

icr also had not been calibrated for pressure correction.

Accordingly, the procedure followed was first to obtain the

rate coefficients for the reactions without corrections and

then to measure the already known coefficients for Reactions

5.1a and 5.1b. An overall system correction factor was then

calculated from comparing the observed rate coefficients for

5.1a and 5.1b with the values of Huntress.

Purified acetylene gas from Matheson was used as obtained

after three freeze-pump-thaw cycles, described earlier. Mass

spectral analysis revealed no impurities present in the samples.

In general the ions were produced by 14 eV electron impact for

3 ms. The pressures of parent neutral gas ranged from approx-
-5 -
imately 8 x 10-5 torr to 2 x 10-5 torr. A magnetic field

strength of 8.0 to 13.0 kG was used.


Profiles of the C2H2, CH +, and CH2H ion signal inten-
2 2' 4 3 4 2
sities are shown in Figure 5-1. Rapid buildup and decay of

the C2H2 cation is noted. Then, a simultaneous increase of
+ 2
both C4H and C4H2 follows. Finally, these latter two sig-

nals are seen to decay. Double resonance experiments, ICDR,
+ +
confirmed that both products, C H2 and C4H came only from
the C2H2 ion. Data points, intensities vs. time, were taken

directly from these curves and used to obtain the rate coeffi-

cients for Reactions 5.1a, 5.1b, 5.2, and 5.3 using the method

described above. Averaging nine sets of data gave rate

+ r)

0 0

c u < --
o CU



IO a

S 0 0


+- r


c ,




o ^


coefficients for the disappearances of the ions: k for C H
4 3
-10 3 +
was 2.1 0.3 x 10 cm /mol s; k for C4H was 3.3 0.8 x
4 2
10-10 cm3/mol s.

This decay of the two product ions' signal intensity is

due to their reaction with acetylene.

C H2 + C2H2 ----- C H 5.2
4 2 2 2 6 4

+ +
CH3 + CH --- C6H5 5.3

These reaction products were also confirmed by ICDR; the
+ + + +
C6H+ and CH5 each originated only from C H2 and C4H3

respectively. No evidence of any other products such as
+ +
C6H2 or C6H was seen.

When the pressure of the acetylene was raised to about

2 x 105 torr, the next set of reactions was observed.

C6H + C2H2 CgH 5.4

C H + C -- C H+ 5.5a
6 5 2H2 8 7

--- CH + H. 5.5b

These products also were seen to react further, leading to
+ +
the formation of C0 H+ and C H+ ions. Unfortunately the
10 6 10 8
intensities of the C8 H and C10H species were too low for

performing a kinetic analysis. Such an analysis may be

feasible with the new computerized data system.


Aside from the measurement of two more ion-molecule

reaction rate coefficients and studies of the products, the

results suggested that sequential reactions could indeed be

involved in the chemistry of flames. At the time of these

studies the ionic pathway to soot formation had considerably

less support than it has now. An unanswered question then

was, Should a series of ion-molecule condensation reactions

be seriously considered? All of the reactions measured here

turned out to be quite rapid, and the reactions suspected to

be involved in soot formation are not too different from them.

They, equations 4.9, 4.12, 4.13 and 4.14, are quite similar

and therefore further study along these lines was warranted.

Interestingly, one possible direct contribution to soot

formation is suggested. The species C H3 has been identified
2 3
in flames (Hayhurst and Kittelson, 1978; Goodings, Bohme, and

Chun-Wai Ng, 1979). The C H reacts with acetylene to form
2 3
C 4H

C2H + CH ---- CHH + H2 5.6

This reaction has a rate constant of 2.5 x 10-10 cm3/mol s

(Kim, Anicich, and Huntress, 1977; Goodings, Bohme, and Chun-

Wai Ng, 1979). Conceivably Reaction 5.6 could provide another

channel to feed into the larger ionic species found in sooting


Finally, examination of the C HH curves revealed that
6 5
there may be two isomers produced. The double decay, char-

acteristic of such, was seen in its intensity vs. time plot.

This is the subject of further study (Eyler and Campana,

manuscript in preparation).