Mixed clusters from the coexpansion of C2F6 and N2 in a pulsed, supersonic expansion cluster ion source and beam deflect...


Material Information

Mixed clusters from the coexpansion of C2F6 and N2 in a pulsed, supersonic expansion cluster ion source and beam deflection time-of-flight mass spectrometer a first application
Physical Description:
xiii, 149 leaves : ill. ; 29 cm.
Thompson, Steven Dane, 1962-
Publication Date:


bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1994.
Includes bibliographical references (leaves 144-148).
Statement of Responsibility:
by Steven Dane Thompson.
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 - 001973114
notis - AKE9906
oclc - 31743900
System ID:

Full Text







To Lee Ann, Justin, and Tara

Born in leaks, the original sin of vacuum technology, molecular beams are collimated

wisps of molecules traversing the chambered void that is their theatre like companies of

players framed by some proscenium arch. On stage for only milliseconds between their

entrances and exits, they have captivated an ever-growing audience by the variety and

range of their repertoire.

John B. Fenn


The completion of this work would not have been possible if it were not for the

help and support of many individuals; their assistance is therefore cheerfully


First, I would like to thank Dr. M. L. Muga for his invaluable help and guidance

throughout the course of these investigations and subsequent preparation of this

dissertation. I am grateful to Dr. F. E. Dunnam, Dr. R. J. Hanrahan, Dr. M. T. Vala and

Dr. W. Weltner, Jr. for their constructive suggestions offered as members of the

supervisory committee.

Additionally, I am indebted to David Burnsed for his tireless efforts in the

development of the data acquisition code for the mass spectrometer, and for his assistance

in troubleshooting, repairing, and in several instances completely redesigning faulty

electronic circuits. For the myriad support he has provided to insure the success of this

project, I am truly grateful.

I thank Chester Eastman, Vernon Cook and Dailey Burch for their skillful

fabrication of countless components necessary for the construction of the experimental

apparatus. I am also thankful for the assistance Shawn Hsu and Matthew Litman provided

as participants of the Florida Foundation for Future Scientists Student Science Training

Program. The efforts of Larissa Brown through the Summer Research Apprenticeship

Program are appreciated as well. I am also grateful for the extensive literature searches

conducted by Carol Drum.

I would especially like to thank the U. S. Air Force--specifically, Capt. Steven

Payne of the Air Force Institute of Technology, and Drs. Stephen Rodgers and Patrick

Carrick of the Phillips Laboratory--for the cooperation and support I received during my

assignment at the University of Florida. The equipment loan from the Philips Laboratory

is also gratefully acknowledged.

For their considerable effort in the initial design and construction of the cluster ion

source, I thank Dr. Young Bae of the Stanford Research Institute, and Joe Newnham and

Bob Frazier of the Jet Propulsion Laboratory. I am also grateful to TOFTEC, Inc. for the

loan of the time-of-flight mass spectrometer and associated hardware, without which this

project simply could not have been completed.

I offer my heartfelt thanks to my friends in the Nuclear and Radiation Chemistry

groups--Dr. Avi Gupta, Craig Hoag, David Miko, Harish Narain, and Ma'an Raja--who

have been of great assistance on countless occasions during the course of this work. It

has truly been a pleasure to work with each of them.

I thank my mother and father for their abiding confidence and reassurance .

Finally, I gladly acknowledge the support and encouragement of my wife and

children; their fresh and dynamic approach to life has provided the inspiration and

motivation needed to complete this challenging project.


In the spring of 1990, a new untested hydrogen cluster ion source constructed to

support a rocket propulsion project whose funding had recently been canceled was

delivered to the Phillips Laboratory. Just a few months earlier, three chemists at

Brookhaven National Laboratory (BNL), Robert Beuhler, Gerhart Friedlander and Lewis

Friedman had announced achievement of deuteron-deuteron (D-D) fusion by impacting

200 to 325 keV D20 cluster ions on TiD targets.' This new "cluster-impact fusion"

technique drew immediate attention. Unlike the cold fusion claims made earlier by Pons

and Fleishmann,2 the BNL results had come from a respected research group which was

very familiar with nuclear chemistry techniques. Even though the possibility that light,

high-velocity beam artifacts may have been responsible for the observed fusion events was

pointed out almost from the start,3 the BNL results certainly merited further investigation.

And so the stage was set. A cooperative effort between the Phillips Laboratory

and the University of Florida was initiated with the goal of attempting to reproduce the

BNL experiments. The cluster ion source was shipped to Gainesville in the fall of 1990,

and the installation and equipment modifications were begun while anxiously watching the

progress of cluster-impact fusion in the scientific literature.

While theorists offered explanations for the new fusion method,4,5 Fallavier et al.6

published the results of a failed attempt to reproduce the BNL results using pure

deuterium clusters. They also suggested that beam contamination problems with the BNL

experiment might be very serious.

In an honest attempt to answer their critics, the BNL group published more

positive results with other deuterium-containing target materials along with the results of

extensive tests performed to exclude artifacts as the cause of the observed D-D fusions.7

In July 1991 the BNL group published additional evidence from time-of-flight experiments

which ruled out low molecular weight ions containing oxygen (i.e. D2O0, D30O) as

responsible for the observed fusion events.8 Yet the possibility of stray deuterons as the

possible cause persisted.

By this time we were having our own problems as it became increasingly obvious

that the ion source we received was in need of major redesign and repair. During the

conduct of any sizable undertaking, unanticipated problems and difficulties often arise--

however, this research effort proved more than typical in this regard. Almost from the

very start, ominous obstacles were encountered with the ion source design. Extensive

modifications of the pulse valve, ion optics assembly and numerous other subsystems were

required. Additionally, a completely new electron gun had to be designed and

constructed. In short, the apparatus we received from the Phillips lab was far from being a

"turn key" system.

In addition to the equipment problems with the ion source, the TOFTEC time-of-

flight mass spectrometer we intended to use to characterize the ion source was aging and

required a complete overhaul; this entailed the replacement of badly cracked microchannel

plates along with substantial modification of the detector configuration to improve

detection efficiency.

While our equipment modifications were being effected, the first independent

experimental confirmation of cluster-impact fusion was reported by Young Bae et al. at

SRI International.9 However, the SRI experiment, just like the one at BNL, did not

include post-acceleration beam analysis.

Young Bae joined the BNL research group which now began experiments in which

the post-acceleration beam was analyzed by both magnetic and electrostatic deflection.

Their initial findings, as published in an errata submitted to Physical Review Letters, were

that artifacts were primarily responsible for events previously ascribed to cluster-impact

fusion.10 Ironically, a comment based on the proton peak broadening reported in the

original BNL paper appeared in the same issue of Physical Review Letters in which Lo,

Petrasso and Wenzel strongly suggest light-ion beam contaminants were likely responsible

for the observed fusion events." The BNL announcement was followed by the

publication of yet another failed attempt to find evidence of a collective enhancement of

D-D fusion for carbon-cluster impact on CD2 targets beyond that predicted by a knock-on

model.12 A complete, detailed report of the BNL post-acceleration experiments has just

recently been published in which the D-D fusion rates with small (D2O)nD+ and (H20)nH+

ions (for up to n = 10) were carefully studied.13 They observed no enhancement for the

fusion rates of the small D20 clusters after oxygen knock-on corrections were made.

However, the fusion yields for knock-on processes produced by (H20)nH+ clusters (n = 4

to n = 10) showed approximately a two-fold enhancement over the yields for H30+ ions at

the same velocity.

Thus, cluster-impact fusion may yet prove to be a genuine route to small

enhancements in D-D fusion yields. Careful experiments with large clusters still remain to

be investigated.

As it turned out, the overall source design of our equipment precluded the

formation of large water clusters without major redesign; also we lacked an accelerator

column to reach the required kinetic energies cited as necessary to achieve cluster-impact

fusion. Thus we altered our goals in accordance with what was reasonably within reach,

and remained flexible to pursue promising leads.

Because of time and financial restraints, the solutions presented throughout this

dissertation to the myriad problems which arose are, in some cases, not optimal; however,

in the context of the current goals and available resources, they represented viable

alternatives which permitted the research to progress in a timely and economical manner.

In other words, though much of our research was indeed conducted in a vacuum, our

business decisions were not. Simple and ingenious solutions were relentlessly sought.

In the words of Dr. Muga, "A good scientist must know his equipment!" I feel I

can truly say the process of identifying, isolating, and ultimately solving the countless

equipment deficiencies encountered over the last three years has afforded me a broad and

rich educational experience which simply could not have been acquired in any other way.

Additionally, the imperative to remain flexible and pursue goals which are realistically

within reach (regardless if they were the original goals or not) has served well to introduce

me to this fascinating, yet ever changing realm of human endeavor we call scientific


Steven D. Thompson
Gainesville, Florida
April 1994



ACKNOW LEDGM ENTS ........................................................ iv

P R E F A C E .............. ............. ..................... ................ .. ........... .. .. ............ .. vi

AB STRA CT .................................. ....... ........ ............. ........................... ... xii


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

Background ............... ....................... 1
Cluster Ion Genesis ........................................................... ............... 2
C luster Ion D election ... ..................................................................... 5

2 EXPERIMENTAL APPARATUS ......................................... 8

In tro d u ctio n ............................................................................ .............. 8
Ion Source ..... ... .. ... .... ................................................. 11
P ulse V alve ................................................. ............................... ... 11
E -g u n ...... ...... ................... ................................................ 14
Ion O ptics .......................... 16
Mass Spectrometer ... ....................................... 23
G general Configuration ............................................. .. .... 23
D etector ................................. ............................................ ... .......... 2 3

MASS SPECTROMETRY ............................. 29

Introduction ............ ................................................. ...... 29
A First Order Model .. ....................... 29
A M odified B akker's M odel .................................. ............................. 32
Applications ............................ ............................ ........... 33
Approach ....... .................. ..34
Experiment ........................... ..... ........ .. .............. .. 34
Theory ........ .. ................. .............................. .......... ............ ..... 40
E xp erim ental .......................................................................................... 4 4
Results and Discussion ........ .......................... 47

4 SURVEY OF INITIAL CLUSTER ION RESULTS ...... .............. 57

Introduction ................ ..................... ... ................. ........... 57
Experimental .................. ......................... ............ 58
Results and D discussion ........... ................................. ............... .................. 59


Introduction ...................... .. .............. ............. ....... ....... ............ 73
Experim ental .................................. ....................... ............. 74
Results and Discussion ............................ 77


A MATHEMATICAL DERIVATIONS ....... .. ......... 92

B TRAJION SOURCE CODE ............... ......... 109

C TRAJION OUTPUT DATA ... .......................... 120

REFERENCES ............... ............................ 144

BIOGRAPHICAL SKETCH ................... ...... .............. 149

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



Steven Dane Thompson

April 1994

Chairman: M. Luis Muga
Major Department: Chemistry

The design and characterization of a pulsed, supersonic jet cluster ion source are

detailed. A commercial solenoid valve modified to provide electrical isolation between the

nozzle and solenoid mechanism provided the pulsed gas flow at a nominal duty cycle of

8 Hz, 200 Ps under source stagnation pressures of 1 to 4 atmospheres. Cluster ion

formation was effected through the generation of seed ions via electron impact ionization

of the gas as it exits the nozzle. The cluster ions were accelerated in a weak electric field

(- 1 V/cm) towards a 2 mm i.d. conical skimmer; the extracted core of the expanding gas

flow was then accelerated to 550 eV and focused coaxially through a pair of Einzel lenses.

Collinear modulation of the collimated beam pulse for time-of-flight mass analysis

was accomplished by beam deflection across an interchangeable slit (0.5, 1, 2, or 3 mm

wide) positioned near the entrance of the mass spectrometer. The beam was swept back

and forth as it traversed a varying electric field perpendicular to the beam axis. The field

was generated between two parallel plates, 2.54 cm apart, by applying a positive square-

wave potential to one plate while the other plate was held constant; the beam deflection

rise-time (0 -- +200 V) could be varied continuously between 30 and 2200 ns. A first

order computational model was developed to predict the behavior of the chopped beam.

Both flight times and peak shape information were calculated for a range of experimental

conditions; theory and experiment are shown to be in substantial agreement.

Characterization of the ion source was accomplished by successfully creating

positive cluster ions from He, N2 and CO2 expansions. The beam pulse composition for

the CO2 expansion was also examined by generating a temporal cross-sectional profile

from mass spectra taken by sampling (chopping) different regions of the pulse.

Finally, the formation of mixed clusters from the coexpansion of Freon 116 (C2F6)

and N, is reported. Peaks at 97 and 147 amu are identified as CF3N2' and C2F5N2

respectively. Isotope shifts observed when substituting 15N2 (99 atom %) for 14N2 (natural

abundance) confirm the peak assignments.



Since the first published report of the formation of cluster beams by Becker, Bier

and Henkes14 nearly 40 years ago, cluster research has grown into a multidisciplinary field

attracting attention from chemists, physicists and material scientists alike. Cluster beams,

and molecular beams in general, are particularly appealing because they offer a method to

study molecular systems in a collisionless environment which is free from matrix effects,

they also provide a controlled way to introduce perturbations and observe their effects,

and by extraction from the core of a supersonic expanding gas flow, very low temperature

molecular populations concentrated in one, or only a few, internal states can be achieved.

In fact, at present, clusters represent perhaps the only tractable bridge between isolated

molecules and the condensed phase.

Today, research opportunities abound as attempts to further our understanding of

amorphous solids, surface interactions, catalysis, liquid structure, solvation effects, and

many other finite-system properties through the study of small, isolated cluster systems. A

recent Department of Energy materials science panel report concluded that the range of

basic scientific opportunities in understanding clusters and cluster-assembled materials is

both broad and deep."5 The new chemistry of reaction processes occurring during the

collision between a cluster and a solid surface is a particularly new and exciting research

area which has been variously described as "chemistry with a hammer" or even "chemistry

with a jack-hammer." Both theoretical and experimental results indicate novel reactive

processes can occur as severe shock waves from the impulse delivered as a fast cluster

impacts a solid surface create local pressures up to 100,000 atmospheres and temperatures

as high as 4000 K.16-19 The experimental and theoretical aspects of the application of

relatively large ionized cluster beams in particle-driven inertial confinement fusion have

also been discussed.20 Recent attempts to effect cluster-impact fusion through collisions

of large D20 cluster ions on deuterated targets have received much attention and were

briefly discussed in the Preface. 1,7-10,13

However, even in light of these and other intense research efforts, characterization

of specific clusters themselves in terms of structure, bonding, and reactive properties is

still in the early stages; much more particle-specific information is needed. Charged

clusters, or cluster ions, play a particularly important role in such studies because specific

clusters can be selected through standard mass filtering techniques.

Cluster Ion Genesis

Many methods for producing beams of cluster ions (both subsonic and supersonic)

have been developed and are described in an exhaustive review by Mark and Castleman.21

Electrospray methods have been successfully applied to liquid samples to generate either

positively or negatively charged particles composed of a monomer unit and attached

solvent molecules.22-24 Partial pyrolysis of a liquid sample before the expansion has also

been demonstrated as a thermospray technique for creating cluster ion beams of involatile

materials.25 Corona discharge ionization of gas monomers within the stagnation chamber

prior to expansion has been shown to result in beams with significant cluster ion content.26

Beuhler and Friedman27,28 employed a corona discharge to produce positively charged

cluster species (A.+) with up to 103 monomer units in their studies on argon, nitrogen, and

water. These represent just a few of the methods which have been successfully used to

generate cluster ions.

Electron impact (El) ionization of an expanding supersonic gas very near the

nozzle exit has also been successfully utilized to generate both positive and negative

cluster ions from various source gases,29-31 and is the technique employed in the present

research effort. Like the corona discharge technique, this method produces charged

monomers, or seed ions, which behave as condensation nuclei during the expansion.

Growth of the cluster ions is assumed to proceed by a series of three-body associative

collisions of the type:

A- A, +A +M-- A A, +M where n = 1, 2,... (Eq. 1-1)

in which the excess heat of reaction is carried away as translational or internal energy in

the third body M; this is in marked contrast to the direct ionization of neutral clusters in

which the excess heat typically results in extensive fragmentation in most cases. Because

these association reactions are primarily driven by ion-molecule interactions, which are in

general much stronger than the van der Waals interactions that control neutral cluster

growth, expansion conditions can be chosen in which cluster ions can be grown with the

near exclusion of neutral clusters. The final structure of small cluster cations has been

predicted to be principally determined by competition between two strong interactions--

hole delocalization and ion-neutral polarization.32,33 Shell structures around dimer or

trimer ion cores have been observed in the early growth stages of various non-metallic

cluster ions.34 The "magic numbers" often observed in cluster ion distributions reflect the

stability of the ions, and a growing body of evidence suggests cluster ions may have

structures quite different from their neutral analogs.32,33,35-37

In order to insure that sufficient gas densities in the jet remain for rapid three-body

collisions, the seed ions must be generated very early in the expansion (or prior to the

expansion as is done in the corona discharge method); supersonic expansions typically

retain adequate densities over a free expansion distance of about 10 nozzle diameters.31

Precipitous changes in the local pressure, temperature and density, which can

easily span several orders of magnitude during a supersonic expansion (cf Figure 1-1),

have made theoretical modeling of cluster nucleation a difficult task since most kinetic

parameters are temperature dependent, and cluster densities are a highly non-linear

function of the monomer density.38 Predictions and limitations of classical nucleation

theory for cluster beam sources have been discussed by Stein.39 Molecular dynamics and

Monte-Carlo simulation techniques provide two additional methods to model cluster

nucleation and growth. Practical scaling laws to predict cluster growth under different

experimental conditions have been derived, and are presented in two excellent reviews by


1.0 -.0

0.6 .01

0.4 .001

02 Q Ct-OOt
O2 00001

0 2 4 6 8 10 12 14 16 18 20

Figure 1-1 Free jet centerline properties versus distance (in nozzle diameters) from the
source. A rapid and continuing decrease in temperature (7), density (n), and hard sphere
collision frequency (v) occurs with increasing distance from the source. To, no and V0 are
the source stagnation values. The mean velocity (V) rises quickly and asymptotically
approaches the terminal velocity (V,,). C /C, = 5/3. (Taken from Miller.42)

Many factors have been found to affect the degree of clustering which occurs

during the adiabatic expansion of a gas in a supersonic jet, and are discussed by both

Stein,39 and Hagena,40-41 as well as Ryali and Fenn.43 Though the nozzle geometry has an

important effect on the cluster beam intensity and average cluster size, the cluster

composition in free jets is determined largely by the gas stagnation pressure, the gas

stagnation temperature, and the nozzle diameter. In general, the cluster content and

average cluster size increase with increased stagnation pressures and larger nozzle orifices,

but decrease with higher reservoir temperatures.

Much higher source stagnation pressures can be reached without resorting to

inordinately large vacuum systems by simply pulsing the beam. Pulsed beam intensities are

orders of magnitude higher than continuous beam sources and have a much greater cluster

content.38 Since the first pulsed expansion work by Hagena,44 pulsed nozzles have been

popularized by Gentry and Giese,45 as well as Liverman et al.46 Pulsed beams, with all

their obvious advantages for producing beams of clusters, do suffer from one minor

problem--the beam velocity, temperature and cluster distribution are not uniform

throughout the pulse. Saenger and Fenn47,48 have provided theoretical estimates of the

minimum time required for a fully open pulsed valve to generate a beam comparable to a

continuous jet under the same source conditions.

Cluster Ion Detection

The method of choice for characterization of a pulsed cluster ion source is TOF

mass spectrometry. It offers a simple, yet efficient means to measure ion mass with high

throughput (and corresponding high sensitivity) and low noise over a large mass range.

And because the pulsed valve imposes a relatively low duty cycle on the experiment, TOF

methods are particularly well suited since they also require a low duty cycle in order to

permit the complete detection of a given ion bunch before a new ion bunch is injected into

the mass spectrometer.

The principle of operation of TOF mass spectrometers is actually quite simple. A

bunch of ions are accelerated in an electric field through a potential, V, and are then

injected into a drift tube. After traversing the drift tube, the ions are detected with some

type of fast electron multiplier device (i.e. microchannel plates). The kinetic energy of all

of the ions, eV, is essentially the same and the velocity of each ion is inversely

proportional to the square root of the mass. Thus, the light ions will arrive at the detector

first while the heavier ions will arrive last. The detector response is recorded as a function

of time and the mass for each resulting mass peak is assigned through the relationship:

2 V (Eq. 1-2)

where L is the length of the drift tube and t is the flight time of a particular isomass ion


The temporal pulse width of the ion bunch injected into the flight tube limits to a

large degree the mass resolution obtainable. Typical pulse durations from pulsed valve

sources range from 50 ps to 1 ms and are quite long compared to typical ion transit

times in a TOF mass spectrometer. Thus, for a continuously El ionized pulsed cluster ion

source, the beam must be modulated, or chopped, before it is introduced into the mass

spectrometer. Beam deflection techniques offer an especially satisfying solution to this

problem since they allow for a collinear experimental configuration and yet avoid, for

instance, the many problems associated with operating a high speed rotating disk

mechanical chopper in a high vacuum. Beam deflection TOF mass spectrometry is

discussed at length in Chapter 3.


A superb review of TOF detection methods, cluster ion genesis and molecular

beam methods in general, has recently been published.49 This very well referenced

monograph, edited by Giacinto Scoles, Davide Bassi, Udo Buck and Derek Laine, is

written from an experimentalist's perspective and presents many options to the beam

researcher who is setting out to design a new experiment. The essential principles and

salient features of various techniques are presented in a simple and clear manner, and this

compilation has been an invaluable guide in the conduct of this work.



This chapter provides a general description of the equipment as currently

configured. The cluster ion source (Figure 2-1) was originally designed and constructed

by JPL and was on loan from the U. S. Air Force, while the TOF mass spectrometer

(Figure 2-2) was supplied by TOFTEC, Inc. Since the cluster ion source underwent rather

extensive redesign, and the TOF mass spectrometer was also significantly modified, no

attempt has been made to indicate where modifications to their original designs were

effected. Additionally, because every complex experimental apparatus is continually

subject to modification and improvement to better meet the changing needs of the

researcher, the design detail presented here reflects the latest experimental configuration


Careful attention was given throughout the redesign and modification of the

apparatus to select materials which possessed characteristics best suited for their intended

function within the vacuum system (i.e. low outgassing rate, high melting point, non-

magnetic properties).50,51 In addition, SIMION v4.02 was used extensively as a design

tool to model potential energy surfaces and compute ion trajectories.52 In general,

SIMION predictions were found to be quite reliable and contributed greatly to the success

of the final instrument design.

A detailed treatment of the beam deflection technique and associated hardware, as

well as the data acquisition system is deferred to Chapter 3.







8 -




0-- -
-1 -
^ ---F

Ion Source

A top view of the ion source is shown in Figure 2-1. A 40 cm diameter chamber

defines the vacuum envelope around the jet expansion region. A large Varian 3500 L/s

turbomolecular pump, backed by a Varian 200 L/s turbo and a Varian 300 L/s mechanical

pump, provides substantial pumping speed for the main vacuum chamber. The pressure is

monitored from atmosphere down to operating pressures with a Granville-Phillips 307

gauge controller through the sequential use of a Convectron gauge and a nude ionization

gauge mounted in the top of the chamber (not shown). The ion optics assembly is

contained within a 6 inch diameter beam pipe; a Varian 400L/s StarCell Vaclon pump

provides additional pumping capacity in this region. Pressure measurements in the ion

optics region are supplied by the ion pump controller. Two 6 inch pneumatic swing valves

are positioned over the throat of the ion pump and at the distal end of the optics region

near the start of the TOF mass spectrometer. To reduce the risk of equipment damage

should high vacuum conditions be lost, the gauge and valve controllers are programmed to

scram the system by automatically shutting down the e-gun and closing both swing valves

if the pressure inside the vacuum chamber rises above an adjustable threshold--typically set

at 5 x 10 torr.

Pulse Valve

A cross-sectional view of the pulse valve assembly is given in Figure 2-3. The

assembly is mounted on a precision 3-way manipulator (cf. Figure 2-1). Adjustments

orthogonal to the beam axis are made through micrometer drives and allow for setting the

nozzle attitude. Though the nozzle is fixed with regard to movement along the beam axis,

axial translation of the valve solenoid and attached valve rod can be accomplished by

rotating a fine thread (24 tpi) screw; this capability is crucial for operating the pulse valve


since it permits optimization of the valve travel (nominally 0.008 0.020 inch) while the

valve is in operation and the vacuum chamber is evacuated. This proved to be particularly

important due to valve drift from thermal heating effects caused by the close proximity of

the e-gun to the nozzle. Alignment of the jet nozzle with the skimmer and ion optics

elements is effected through adjustment of the nozzle support plate.

The valve itself is constructed from a modified General Valve Series 9 pulse valve

equipped with a lightweight (1.6 g) nylon valve rod which seats in a brass nozzle. An Iota

One Pulse Driver controls the Series 9 solenoid and, for an unmodified valve, permits a

nominal fast response (time to fully open) of 50 60 /s by overdriving the solenoid with

an initial 300 V input for 165 /s; a holding voltage of about 20 V is then applied in

accordance with the pulse duration selected on the Iota One Pulse Driver. Armature

movement occurs only during the last 1/3 of the high voltage pulse. Thus there is about a

100 .ts delay between the time when the voltage is first applied to the solenoid and when it

actually begins to open. The relatively long time required to build and collapse the

magnetic field within the solenoid also accounts for a minimum open time of 250 300 1s.

A compression spring within the solenoid body restores the valve to the fully closed

position in about 40 50 ps.53 The valve was normally fired at a frequency of less than

10 Hz. The gas volume delivered per 200 ts pulse (measured for air at I atm. and 250C)

is approximately 5 yAL.

The valve rod was fabricated from 5/32 inch diameter nylon stock and was tapered

on one end to match the nozzle valve seat geometry; the valve and seat were lapped

together prior to assembly to ensure a good seal. A groove 1/16 inch wide and 1/32 inch

deep was cut down the entire length of the rod to permit unrestricted gas flow to the


The nozzle body was constructed from brass, but contains a 4.65 cm long stainless

steel capillary tube 0.030 inch in diameter which serves to define the nozzle throat. The

nozzle is easily replaceable, and a second nozzle was in fact constructed with a throat

diameter of 0.018 inch but was not used.

The gas stagnation temperature was monitored via a miniature Type J

thermocouple mounted outside against the valve assembly near the region where the valve

seats. The stagnation pressure was measured with a Bourdon tube mechanical

pressure/vacuum gauge attached to a gas manifold. Cooling of the valve assembly, though

possible through the use of an external dewar and conducting straps within the vacuum

chamber, was never attempted.

The ions were produced at a high potential and then accelerated to ground. The

nozzle and nozzle support plate were therefore electrically insulated from the vacuum

chamber through the use of ceramic stand-offs. A Kepco APH1000M DC power supply

and Lambda C-280M DC supply maintain the nozzle at + 550 V. The skimmer potential

is supplied in parallel with the nozzle voltage through a potentiometer and was typically

reduced 5 V. The results of SIMION modeling of the electric field generated between the

nozzle (and support plate) and the skimmer and ion optics face are shown in Figure 2-4.


The electron beam is generated continuously from a 0.005 inch diameter, 97%

tungsten, 3% rhenium, coiled filament wire. The wire (total length of -~ 5 cm) was wound

seven turns about a 0.040 inch diameter rod to create the coil. A current of 2 A was

passed through the filament which was held at +430V. The filament legs are inserted

into stainless steel posts mounted within a MACOR cylinder fitted within a 3/8 inch o.d.

stainless steel casing closed at one end by a thin stainless steel plate with a 1 mm slit. The

casing and slit plate are at the same potential as the nozzle (+550V) and therefore the

electrons emitted from the filament (positioned approximately 1 mm away from the slit)

are accelerated to 120 eV as they exit the slit. The electrons then drift across the nozzle



I :I__


.. .. ...

,. ".''" ... .


Figure 2-4 SIMION model of jet expansion region. (a) Ion trajectories from a 1200 CO2-
jet spray ; A0.5 V contour lines show smooth E-field between nozzle (NOZ) and skimmer
(SK); (b) 3-D view of potential surface; Nozzle potential = +550 V, skimmer = +545 V.

exit and are detected by a collector with essentially identical external geometry as the e-

gun to minimize electric field perturbations near the nozzle. Electrical leads were also

shielded to minimize field disturbances. Electron currents of 200 ZA were measured

under typical source operating conditions. Both the gun and collector can be moved

radially and axially relative to the nozzle, but can only be adjusted with the system up to


Ion Optics

The central portion of the expanding jet flow is extracted through a standard Beam

Dynamics conical nickel skimmer with a 2 mm entrance orifice located 4.7 cm

downstream from the nozzle exit. The optics used to accelerate and focus of the beam are

shown to scale in the SIMION simulations for two different operating configurations

presented in Figures 2-5 and 2-6. The ion optics assembly end cap and face plate are held

at the same potential as the skimmer (+ 545 V). The two Einzel lens voltages are supplied

independently through two Ortec 556 HV power supplies. The potential applied to the

four pairs of parallel plates can be varied individually by adjusting an array of eight

precision potentiometers connected to two Kikusui PAB 250-0.25A DC power supplies.

The plates are used for making small adjustments to the beam trajectory (steering). The

last pair of plates is also used to modulate the pulsed cluster ion beam for introduction into

the TOF mass spectrometer.

The electrical leads are supplied to each element internally (within the 4 inch o.d.

stainless steel sleeve), but are insulated and shielded with Teflon spaghetti and 1/8 inch

copper tubing. As previously mentioned, a 6 inch swing valve separates the ion optics

from the mass spectrometer.







+ 0





0 0-
0 -

CA to)


IZ-F! t ilt




III ~.-.--.-. --

HI' Ft





I 'P.-



- tili





. &)
1! ; 9
^'A C)
^ \ an
;..^\ i
:: ** ^ \C)
: ... \-


( *pt~~~


I' I.PfrittLJlflf i tT htT." j

/ t -










+ o


0. 0



(' U

HEFf iill' ii Hl-tiln


i t


:1~ f~





iii ''-4' 4-H ~ Fuel"

{J j



.. .

.. .. ., .

" : ".- '" ;

.. .


Mass Spectrometer

A top view of the mass spectrometer was shown in Figure 2-2. Pumping is

provided by a Leybold-Heraeus TOPS 360 turbomolecular pumping system which consists

primarily of a 360 L/s turbo pump backed by a dual stage rotary vane pump.

Thermocouple, cold-cathode and Bayard-Alpert type ionization gauges near the detector

end of the mass spectrometer supply pressure information. The slit is positioned 11 5/16

inches from the edge of the last set of parallel plates--the deflection plates. Titanium slits

of 0.5, 1, 2, and 3 mm can be interchanged, but require the vacuum system to be

interrupted to provide access. The vacuum envelope is defined by standard 4 inch o.d.

stainless steel beam pipe.

General Configuration

Figure 2-7 shows the TOF components in greater detail. Ions enter a grounded,

47 inch long, 3/4 inch diameter stainless steel flight tube after traversing the velocity

compaction region. The velocity compaction portion of the mass spectrometer was

disabled for this work and was not used. As the ions exit the field free region inside the

flight tube, they receive a final 2 kV acceleration before striking a stainless steel impact



Secondary electrons generated by collision of the high velocity ions with the metal

impact plate are collected and amplified by a stacked pair of Galileo chevron microchannel

plates (MCP's) with an active diameter of 25 mm (see Figure 2-8). An electron trap

consisting of a 90% transparent trap grid and stainless steel deflection screen enhance the

d i t-'-

-- -- _


-y -_1 --^- -


~ L. -~

II *


-a >

X o



0 0.



3 &


U0 3

(A 0


o o

.1 O
E| 0
-- -






collection efficiency. The trap grid and deflection canopy are kept at -1.9 kV by an

Ortec 556 HV power supply. The impact plate, first MCP, and second MCP are each

operated at -1.9 kV, -1.7 kV, and -0.1 kV, respectively, through a resistor string

supplied by a separate Ortec 556 power supply.

SIMION negative ion potential energy plots are presented in Figure 2-9, and

simulations for a 550 eV CO2+ beam impact and subsequent detection of secondary ions

ranging from thermal energies to 10 eV are shown Figure 2-10.

The two-stage MCP configuration, operated with an 1.6 kV applied voltage

differential, produces approximately a 106 output gain.54 The detector signal is supplied

through a partially shielded lead to a standard BNC connector. Additional amplification of

the signal is provided by an Ortec 535 quad fast amplifier. The inverted signal is then

supplied to a LeCroy 9400, 125 MHz digital storage oscilloscope (DSO). Triggering and

other timing aspects associated with data collection are discussed in the next chapter.

11111 <


ED A I I 1anI 11 MCP's


Figure 2-9 SIMION plots of (a) detector 3-D negative ion potential energy surface, and
(b) A25 V contour lines in the electron trap region; Deflection Screen (DS) = Trap Grid
(TG) = Impact Plate (IP)= -1900 V; Top MCP = -1700 V; Bottom MCP = -100 V.


(a) 550 eV CO2+ beam


(b) 0.025 eV electrons


(c) 1 eV electrons

(d) 10 eV electrons

Figure 2-10 SIMION simulation of (a) 550 eV CO2+ beam impact with deflection plate
(1/64 inch deflection noted); (b) through (d) depict the collection efficiency of the trap for
stated e- energies with an initial 1800 angular dispersion. Ticks are 1 ns timing marks.





A First Order Model

In 1973, J. M. B. Bakker" published the first paper addressing the theory of beam

deflection as a technique to modulate a continuous ion beam for introduction into a TOF

mass spectrometer. His elegant treatment of this rather complex topic continues to

provide valuable insight into the behavior of ions as they move through a rapidly changing

electric field perpendicular to the beam axis. Bakker described the modulation process in

terms of a 'kink' introduced into a homogeneous beam of ions initially deflected

downwards as it passes through an electrical field which suddenly experiences an inversion

in the electric field which then deflects the beam upwards.

The ion beam profile for this model is shown in Figure 3-1. The kink lies between

e-e and a-a and travels from left to right at the same velocity as the rest of the ion beam

while simultaneously expanding in both the upward and downward directions. The e ions

have the largest upward velocity component while the a ions have the largest downward

velocity. Ions at the center of the kink possess little or no velocity component in the up or

down direction. Of the ions affected by the step-function controlled electric field, only the

ions within the boundary do-co-bo-co will actually pass through the slit S. Finally, it should

also be noted that the velocity of the ions from left to right is unaffected by the

D,. -. ....._ .. ... -- .

cc) t--3 t \\
SK "',V

d) 1, '
(e) t

Figure 3-1 A monoenergetic ion beam of width B and consisting of ions with mass m and
energy eU, traverses an electrical field of strength E controlled by a step function and
produced between deflection plates of length I and separation D. The ions then travel
through a field free space L until they reach a slit S. The position of eight ions at time To
(when the electrical field changes instantaneously) is indicated by ao, ao, bo, co, Co, do, eo
and e0o. The electrical field experienced by these ions as they travel between the deflection
plates is a function of their positions at time To, and the field profile for the ao ions is
shown by Figure 3-1(a). Figure 3-1(b) shows the field encountered by the bo ion, and etc.
The ion positions are shown for three consecutive instants of time, t,, t2 and t,, after the
field change, and are designated by corresponding subscripts. Note: the figure can also be
interpreted as showing the relative positions of three monoenergetic ion beams with
different masses at one instant of time, i.e. b3-c3-d3-c3 represents a packet of very light
ions, b2-c2-d2-c2 represents a packet of medium weight ions, and 1b-c,-d,-c1 represents a
packet of very heavy ions. (Taken from Bakker.55)


It^ J-" -"

perpendicular electric field, and therefore the projected distances e a / and d b I'

are constants.

Bakker not only provided a qualitative model of the beam modulation process, but

he also derived expressions for the ion burst duration (pulse base width), resolution, and

relative ion intensity for a theoretical beam-modulated TOF mass spectrometer (cf Table

3-1). His treatment, however, was restricted to three distinct cases; the first (Table

3-1(a)) treats the idealized case of an instantaneous reversal in the electric field polarity,

the second case deals with the more likely situation in which the deflection plate voltage,

and thus the electric field, changes rapidly with respect to the transit time of the ions

through the plates (Table 3-1(b)), and the final case (Table 3-l(c)) addresses the situation

where the transit time is shorter than the time required to invert the electric field. Many of

the equations are not exact solutions, but are approximations based on certain stated


Table 3-1 Summary of Bakker Results

Waveform V(t) Ion burst duration At (general) At (tsath) Resolution R = tz/2At
Ion intensity Icc Atl/i
tl-- (B+S)mD (B+S)mD VoL2
,--- eVo(2ts+4t(ta+4t22)1/2 2eVot2 2DU(B+S)
(B+S) DU
eV (VolL
- -- -1= C2
tz-/- (B+S)mD (B+S)mD R=Ci
eVo(2t,2+4tit2+4ta2-iJt4)1/s 2eVots
14 V 2(B+S)mDT (B+S)mDT ti I m /\i
VO t eVotx(ti+2t2) eVot ts R = 2- T C e =

T> t, R =f(ml/2)
,., -Vo i I=f(m-i/2)

Bakker concluded that a beam-modulated TOF instrument operating in the regime

described by the first two cases would have a resolution independent of ion mass,

independent of the location where the ions were initially formed, and proportional to the

square of the length on the drift space. In 1974, Baker56 reported the construction of a

beam-modulated TOF mass spectrometer, and provided an experimental confirmation of

his theory.

A Modified Bakker's Model

Two attempts to refine Bakker's theory have been published recently.57'58 Yefchak

et al. explored the added effects of energy spread and angular divergence of the beam via a

computer simulation. Approximate peak widths were calculated for the Bakker effect

alone, the Bakker effect with energy dispersion, the Bakker effect with angular

divergence, and all effects combined. The program modeled the motion of various ions by

means of Runge-Kutta integration through the deflection plate region and onward to the

aperture. Flight times for those ions found to pass through the aperture were added in the

form of a histogram to generate a peak shape. According to Yefchak et al., the simulated

peak variances were not well characterized, but were believed to fall within a few percent

of the true values. The Bakker equations shown in Table 3-1(b) provided the

mathematical basis for all of these calculations.

Ce Ma et al. also developed a computer simulation, based on simplified equations

much like those of Bakker, to model the case when the voltage rise time on the deflector

plate is short in comparison with the transit time of the ions between the plates (cf. Table

3-1(b)). The simulation was used to establish the m/z independence of the ion packet

width, and to investigate the effects of varying the deflection plate voltage, the deflection

voltage rise time, and the initial beam acceleration voltage on the ion packet width when

the ion beam is swept past the slit; the slit was positioned near the entrance to the TOF

mass spectrometer rather than near the detector as with Bakker's work. Ce Ma et al. also

investigated the effect of a Gaussian energy distribution on the ion peak width using a

Monte Carlo method. Since no shutter function was included in any of these simulations,

the peak width data pertained to base peak width. No pulse shape information was



Beam deflection was first used in connection with TOF mass spectrometry as a

simple gate. Futrell et al.59,60 were among the first to use a deflection plate to reduce

baseline noise caused by stray ions from a continuous high pressure ion source which enter

the drift region between pulsed extractions and are accelerated down the flight tube. Glish

and Goeringer61 installed a set of deflector plates in a tandem quadrupole/TOF mass

spectrometer used to study collision activated dissociations; these deflectors were used to

gate the ions exiting the collision cell into the drift region for TOF analysis.

Since beam deflection provides an excellent method for sampling a continuous ion

source, or a source which cannot be pulsed sufficiently fast for TOF measurements, it is

beginning to find more and more applications--especially among analytical chemists.

Beam deflection techniques have been successfully applied in tandem magnetic sector/TOF

mass spectrometry by Enke's group at Michigan State University.62'63 Pinkston et al.64

recognized the advantages of beam deflection for ion pulse formation, and incorporated

the technique into an electric sector/TOF mass spectrometer. Very recent applications

include the design of an atmospheric pressure ionization/TOF mass spectrometer using a

beam deflection method,58 and the combination of gas chromatography with beam

deflection TOF mass spectrometry.57,65,66

No doubt, the utilization of beam deflection techniques will continue to grow in

the future as the need to "chop" continuous, or slowly pulsed ion beams increases.

However, the development of a more comprehensive theoretical model would certainly

facilitate the prudent application of this sampling method. Bakker's model continues to

serve well as long as a given instrument is designed and operated within one of the

regimes where the equations apply. But what about when they do not apply? For

instance, the case where the rise time of the voltage on the deflection plates is about the

same as the transit time of the ions has not previously been treated. Additionally, since the

ion transit time is mass dependent, it is conceivable that over a large mass range some of

the ions might fall within one regime while others do not. In short, a more rigorous

mathematical treatment is needed.

The novel application of beam deflection methods to chop an El ionized, pulsed

cluster ion source for TOF analysis is detailed below, along with the development of a

more rigorous computational model.



The basic components of the cluster ion source and mass spectrometer were

described at length in Chapter 2. This discussion, therefore, will be limited to the beam

deflection hardware and support equipment.

The last pair of steering plates (cf. Figure 2-5) in the ion optics assembly was also

utilized for beam deflection; the high voltage pulse needed to sweep the beam back and

forth across the slit was superimposed upon the experimentally determined steering

voltage required for proper alignment of the beam. A schematic diagram of the basic

experimental configuration along with typical operating parameters is shown in Figure 3-2.

I A 6vrY y SP


0 --- ---- SRS DG 535 GENERAL



C-280M 125 MHz 4DX2-66V


Figure 3-2 Beam deflection schematic. The Fast HV Pulser switches +200 V on and off
the lower plate (LP) while a constant + 100 V is applied to the upper plate (UP). Each
sweep is initiated by a TTL pulse from the pulse driver. The slit plate (SP) is located
38.7 cm from the deflection plates, and the detector (not shown) is positioned 151.4 cm to
the right of the slit.

The beam is bent by the variable electric field generated between the two parallel

plates as the square wave is applied to the lower plate; the parallel plates are separated by

a distance of one inch and are both one inch in length. The beam is then chopped by an

interchangeable slit located 28.7 cm away from the deflection plates, and the resulting

beamlet then enters a field free drift region. While traversing the drift tube, the chopped

beamlet separates into individual isomass packets which are finally detected by a stacked

pair of microchannel plates located at the end of the flight tube (cf. Chapter 2). The

timing is critical in this chopping process since the ion beam is not continuous, but is in

fact mechanically pulsed. However, since the pulse duration is long (nominally a few

hundred microseconds) compared to the time slice chopped out of the pulse (a few

hundred nanoseconds or less) it is only necessary to synchronize the voltage swing so as

to coincide with the pulse arrival time over the parallel plates. Ideally the slice should be

taken from the middle of the beam pulse, thus avoiding the edges which were produced

during non-equilibrium flow conditions in the nozzle.

By adjusting the delay from the SRS pulse generator, it is possible to control

where the slice will be taken within the beam pulse. This truly unique feature makes it

possible to study the spatial and temporal composition of beam pulses by taking mass

spectral cross-sections at arbitrary positions along the entire length of the ion beam pulse.

Just such a study was performed with CO2 and is presented in Chapter 4.

The upper plate is held at a constant +100 V while the lower plate voltage is

driven by a square wave of +200 V amplitude. The rise time (10 90%) of the leading

edge of the square wave can be varied continuously from 30 to 2200 ns through the use of

a novel fast switching circuit.67 The schematic diagram is provided in Figure 3-3.

The electric field potential between the plates for the two limiting conditions, i.e.

when the lower plate voltage is 0 V, and when it is + 200 V, was modeled using SIMION

and the results are shown graphically in Figure 3-4. It should be noted that significant

fringing fields are predicted to exist. Ion beam trajectories were also computed and are

presented in Figure 3-5. Because of the asymmetries involved with the electric field lines,

as well as the difference in the rise and decay times of the high voltage square wave, only

beam sweeps originating from the leading or rising edge of the wave (sweeps from bottom

to top) were used to acquire mass spectra. Thus, when the high voltage is switched on, a

TTL pulse from the Fast HV Pulser is used to trigger a LeCroy 125 MHz DSO which

captures the TOF spectrum and holds it until queried by a Gateway 2000 4DX2-66V

computer via a National Instruments GPIB interface. These single-shot spectra are then

added together to build the final mass spectrum. The entire beam deflection and data

acquisition process is slaved to the pulse driver TTL output. The digital accumulation and

final display of the summed spectra are controlled through software developed by David J.



6.8 pF. 15V

DG403 VL


10k. 2W




Figure 3-3 Fast HV Pulser switching circuit. A regulated power supply (not shown)
provides the constant current needed to quickly charge the considerable input capacitance
of the power MOSFET Q1. A dual analog switch, with its two switches paralleled to
reduce switching times, couples the supply to the MOSFET's gate on command. A TTL
signal applied to the analog switch's IN; and IN2 pins connects the 12 V supply applied
to the switch's DZ and D2 inputs to its S, and S2 outputs. The analog switch's outputs
connect the MOSFET's gate, switching 200 V in nanoseconds ( the pulse rise time, 1, =
32 ns and the fall time, t = 1400 ns). Note: varying R2 affects t, and t,. (Taken from





"0 0





C> 0

c )


0 )


a )




+.00. .SP .

.--^ -.i^^ U "'

.^ LP

Figure 3-5 SIMION beam trajectories and 3-D view of the potential energy surface. (a)y
displacement of 550 eV, 2 mm diameter ion beam when the Upper Plate (UP) voltage =
+100 V and the Lower Plate (LP) voltage -= 0 V; (b) displacement of the beam when UP =
+100 V and LP = +200 V. SP = Slit Plate.


A computer code was developed to model the behavior of ions traversing the beam

deflection TOF mass spectrometer. Unlike previous efforts, the goal from the very outset

was to create a robust algorithm which could accurately describe ion trajectories, and

predict the flight time and pulse shape of any given isomass packet for virtually any

conceivable experimental configuration. To accomplish this, equations of motion were

derived from first principles which address the variable electric field conditions

encountered between the parallel plates as the applied voltage changes. Additionally, the

velocity distribution within an isomass packet was modeled with a one-dimensional

Boltzmann distribution, and a shutter function was constructed to describe the relative ion

flux transmitted by a slit as a cylindrical beam is swept across it. The detailed

mathematical treatment of these topics is provided in Appendix A.

Once the proper equations were in hand, the next step was to find general

solutions within the constraints imposed by realistic beam deflection TOF mass

spectrometer designs. A numerical approach was taken, thus a computer code was

written to model ion beam behavior under a wide set of user defined conditions. The

primary goals while encoding the program were to keep it completely general (i.e. not

restricted to Bakker's three cases) while at the same time providing rigorously correct

solutions, and to allow for maximum flexibility through numerous user defined parameters.

TRAJION substantially meets these goals. Programmed in PASCAL, TRAJION

makes good use of the top-down programming technique; the highly structured PASCAL

environment proved ideal for solving the complex beam deflection problem through the

stepwise implementation of several procedures which address independent subproblems.

TRAJION consists of six procedures and one relatively short main program segment. A

flow chart detailing the general procedure call logic is shown in Figure 3-6. The complete

source code is provided in Appendix B.









Figure 3-6 Simplified TRAJION flow chart. Procedures are indicated by rectangles.

Input parameters are supplied in part through nine user defined constants: 1) XB,

the length of the parallel plates, 2) XC, the distance separating the parallel plates and slit,

3) XD, the distance separating the slit and detector, 4) D, the distance between the

parallel plates, 5) E, the slit width, 6) TAU, the plate voltage decay time (Note:

TRAJION solves the case where the voltage applied to the lower plate falls from its initial

maximum value to a final lower value while the upper plate is held at ground potential, i.e.

the ion beam is swept from top to bottom. However, due to axial symmetry either sweep

direction can be effectively modeled merely by inputting the tr or the tr time value for the

constant TAU), 7) VO0, the initial maximum plate voltage, 8) VTAU, the final minimum

plate voltage, and 9) N, the number of Boltzmann velocity bins and also 1/10 the number

of time slices taken of the beamlet admitted through the slit.

An additional six parameters are supplied through answering prompts: 1) M, the

mass-to-charge ratio, 2) KE, the beam kinetic energy, 3) TFAIP, the beam temperature,

4) R, the beam radius, 5) DX, the number of trajectory loci to be calculated between the

parallel plates, and finally 6) DT, the maximum acceptable flight time error for the two

limiting trajectories which define an isomass packet. Thus, a total of fifteen adjustable

parameters allow for tremendous flexibility.

Once these input values are assigned, the procedure BOLTZMANN is called. It

generates an N X 2 array, VXDIST, which contains the discreet velocity distribution for

the ion beam (the tails are truncated at 5%); the velocity distribution data are also written

to the file "BOLTZMAN.TXT."

The main program segment calculates successive ion trajectories through the beam

deflection apparatus in a step-wise fashion. Each trajectory begins at time zero, at a

predefined position, XO. The ion position is tracked by the main program segment

through a series of appropriate procedure calls. PREACCEL handles the ion motion prior

to entering the parallel plate region, CONSTACCEL addresses ion motion between the

plates before the electric field changes, or after the field change has occurred (i.e. when

the electric field is constant), VARIACCEL addresses ion motion while the electric field is

changing, and POSTACCEL tracks the ion position once the ion emerges from the parallel

plates and terminates when the ion strikes the slit plate or (if the ion passes through the

slit) the detector. Which procedures are called, and when, is determined by the main

program segment logic. By design, the first calculated trajectory will impact the slit plate,

the XO value is then increased by an amount proportional to DT, and the process is

repeated until the first solution trajectory is found which passes through the slit.

Since the beam is swept downward, the fastest ions at the top, rearward portion of

the beam which pass through the slit last establish the maximum initial distance from the

parallel plates, LASTX, while the slowest ions at the top rearward portion of the beam

establish the maximum flight time to the detector, LASTT. The fastest ions at the bottom,

forward portion of the beam which pass through the slit first, establish the minimum flight

time to the detector, FIRSTT. After comparing the initial x-coordinate of the slowest ions

at the bottom, forward portion of the beam first to pass through the slit with the starting

position of the fastest ions in the same region of the beam, the minimum initial distance

from the parallel plates, FIRSTX, is also assigned.

Thus, limiting ion trajectories are used to define the initial spatial extent, and the

final temporal width of the solution ion packet. The two flight times, FIRSTT and LASTT,

bracket the arrival times of all other solution trajectories. This base width time interval is

divided into N equal time segments which define the channels in the array MCADIST

where subsequent trajectory flight times and intensities will be accumulated to create the

final pulse shape histogram. Further, the initial x starting positions, LASTX and FIRSTX,

bracket the starting positions (XO's) for all the remaining trajectory calculations. The line

segment defined by LASTX FIRSTX is divided into O1N equal segments of width

DEL TAX whose midpoints will now serve, in turn, as the final set of XO's. Additionally,

complete trajectory data for the FIRSTX and LASTX trajectories are written to the file


A nested pair of FOR loops now takes control of the remaining trajectory

calculations; the first loop steps through the final set of initial starting positions, XO's, and

the nested loop assigns the x-velocity, VX, the nested FOR loop steps through the entire

Boltzmann velocity distribution, VXDIST, for each XO. Thus a total of 10N2 trajectories

are computed. The main program segment handles these computations through a series of

appropriate procedure calls as before, only now an additional procedure is called for each

solution trajectory to calculate the cross-sectional area of the beam which passes through

the slit. The area calculated by the procedure CHOPPER is then multiplied by DEL TAX

and the appropriate Boltzmann density to yield a relative intensity scalar. The flight time

to the detector and the intensity scalar are then used to build a pulse shape histogram. The

final accumulated data contained in MCADIST are written to the file "PULSE.TXT." A

diagram illustrating the basic histograming process is shown in Figure 3-7.


Before any meaningful cluster ion research could be accomplished, the

performance characteristics of the new cluster ion source and beam deflection TOF mass

spectrometer had to be determined. A dual approach was utilized to combine both

experiment and theory in order to gain the maximum amount of information about the new

system in the least amount of time. Of particular concern was the effect of the collinear

beam deflection technique on the time, intensity, and resolution of the final mass peaks.

Computer simulations were run for a 550 eV monoenergetic ion beam over the

mass range of 1 to 106 amu. Two series of calculations were performed for a 3 mm slit

using different deflection voltage rise times: tr = 2200 ns and t, = 30 ns. All the code

parameters were matched as close as possible to the actual experimental setup. The beam

axis (or centerline) trajectories between the parallel plates for each run were plotted and

S0 -

r .- E

o o d



UQ >l Cl




-w 1- -


.d 0 '-0N

04-. 4-..


.44 1.


.C3 H



0j C.4 )

are shown in Figures C-1 through C-14 of Appendix C. The complete trajectory data are

displayed along with the trajectory plots in accompanying tables.

Preliminary experiments were also run to test the effect of various experimental

configurations on the overall equipment performance. Once the optimum operating

parameters were established (i.e. pulse valve settings, Einzel lens and steering plate

voltages, etc.) a series of experiments was conducted to elucidate the effects of slit size

and deflection voltage rise time on the signal intensity and resolution. Fifteen different

configurations were tested using five slit geometries (none, 3 mm, 2 mm, 1 mm and

0.5 mm) and three different deflection voltage rise times (2200 ns, 500 ns and 30 ns).

Carbon dioxide at a stagnation temperature of about 100F and 2 to 15 psig pressure was

used throughout. The pulse valve was selected to fire at between 2 to 8 Hz with 200 /As

duration. The 0.030 inch i.d. nozzle was positioned 4.7 cm away from the skimmer. The

electron gun and collector were positioned 5.8 mm away radially and produced a

continuous stream of 120 eV electrons through the 1 mm slit opening in the gun face

which was axially positioned flush with the nozzle tip. The nozzle voltage was maintained

at +550 V while the skimmer was held 5 V below it at +545 V. Both Einzel lenses were

kept at +340 V. Steering plate voltages were adjusted at the beginning of each day's run

to maximize the signal intensity of the undeflected CO2 pulsed beam. The final spectra

were accumulated from 500 to 2,000 shots, depending on the signal-to-noise (S/N) ratio

produced by the given configuration being tested. Typical source pressures were 2 x 10-6

torr base, to 2 x 10-5 torr with the valve firing. Pressures in the ion optics and beam

deflection region were generally 5 x 10-7 torrn while the pressure in the detector region

was lx 10-6 torr.

A final set of TRAJION simulations were also performed. Trajectory calculations

for a CO.2 beam with a mean energy of 550 eV were computed for each of the five slit

configurations using all three deflection voltage rise times. The internal beam translational

temperature (Boltzmann energy spread) was set at 10 K and the beam diameter was

defined as 1 mm. The resulting fifteen 101-channel, pulse shape histograms are presented

as Figures C-15 through C-17 in Appendix C.

Results and Discussion

The reduced data for the monoenergetic beam centerline trajectory study are

presented in Tables 3-2 and 3-3 as the length of the segment chopped out, or sampled,

from the beam by the beam deflection technique. Since the beam is assumed to be

homogeneous, the sampled beamlet length for a given ion mass is directly proportional to

the final pulse intensity. As shown in Figures 3-8 and 3-9, the initial TRAJION

calculations predict a mass dependent intensity function for ions up to about 103 amu

when the beam is swept slowly (tr = 2200 ns) across the 3 mm slit; however, an enhanced

sensitivity for lower mass ions is not indicated when the beam is swept quickly (1, =

30 ns). Thus, if a slow deflection voltage rise time and large slit are utilized, intensity

corrections would have to be made before quantitative comparisons based on peak

intensities could be made.

Fortunately, the peak positions (flight times) remain proportional to the square

root of the ion mass as in standard TOF methods. Graphical proof of this statement is

given for both the slow and fast beam sweep cases in Figures 3-10 and 3-11. The sampled

beamlet midpoint positions for the seven representative ion masses (see Tables 3-4 and

3-5) were plotted against time to ascertain whether a common starting position, and thus a

constant flight path length, existed. For both cases, all the beamlet midpoints were

precisely aligned at the same time at a point approximately 12.43 mm into the parallel

plates--just short of halfway. Thus, even for a collinear beam deflection TOF system in

which the slit and detector have not been offset to compensate for the off-axis beam

displacement generated by the electric field between the parallel plates, each isomass ion

Table 3-2 Temporal and Spatial Beamlet Lengths
(tr = 2200 ns, KE = 550 eV, 3 mm slit)

mle w (ns) w (mm)

1 120.999 39.419127
10 120.999 12.465423
100 120.999 3.941913
1000 135.515 1.396077
10000 428.417 1.395784
100000 1355.103 1.395756
1000000 4281.448 1.395752




Beamlet Length

(tr=2200ns, KE=550eV, Slit=3mm)

-U -

U- U U -U

Figure 3-8 Plot of beamlet length as a function of ion m/e.

0.1 1 10 100 1000 1E4 1E5 1E6 1E7
m/e (amu)

Table 3-3 Temporal and Spatial Beamlet Lengths
(tr = 30 ns, KE = 550 eV, 3 mm slit)
mWe w (ns) w (mm)

1 4.283 1.395318
10 13.547 1.395623
100 42.842 1.39571
1000 135.482 1.395748
10000 428.433 1.395752
100000 1354.825 1.395753
1000000 4284.33 1.395751

Beamlet Length

(tr=30ns, KE=550eV, Slit=3mm)


'E 1.425
"c 1.4
c- 3
. 1.375

1 10 100 1000 1E4
m/e (amu)

1E5 1E6 1E7

Figure 3-9 Plot of beamlet length as a function of ion m/e.

1.35 -

0 M 0 0 M a



*3 E E A


** o

0C 0 0 -- 0( 0. 0
Io a- 0 a o

) LU O Go C |

D O 0) o
S 0R0 0 0 0 0
'A Co 0 07

_II 0
1/ -r o n o. i O5-n 1 (



LO -

0 I \ \
o 4 \

o -N 0 e

S a O a n a ( aI

a 0 n O w w I,-
.z (co C) (0 (0 N L
O (0 (0 0 --
F- oo V V) U) C o
a r- O 0 0 0
C- 6 0)

-, 0_ C C- OC
F-0 e

.0 00 0 0 n ( I
-o = ,- o fl- c o

T- T
= ll o n c C < '

packet traverses a constant path length ensuring standard TOF methods still apply for

peak identification.

The results of the experiments conducted to characterize the equipment

performance under various slit and deflection voltage rise time configurations are

summarized graphically in Figure 3-12. The CO2+ peak resolution, as measured by the

full-width at half the maximum intensity (FWHM), shows a marked improvement for all

five slit configurations when the deflection voltage rise time is decreased from 2200 ns to

500 ns, but changes very little when the rise time is further decreased to 30 ns. This same

general trend was predicted by the TRAJION simulations for a 1 mm diameter CO2+ beam

with a Boltzmann translational energy spread of 10 K (see Figure 3-13).

The effect of slit size on peak resolution is not as clear. The relative positions of

the five resolution curves in Figure 3-12 seem almost random; in fact, their relative

positions probably have very little meaning at all. Because of the complexity of the ion

source and mass spectrometer systems, it is very difficult to reproduce exactly the same

operating conditions from one day to the next since the equipment had to be completely

shut down at the end of each day to preclude possible damage from loss of utilities in the

laboratory. Perhaps the largest source of error resulted from the visual alignment of the

nozzle, ion optics, and slit, which had to be accomplished after each slit configuration

change. As the TRAJION calculations bear out, even slight misalignments can cause

portions of the chopped beamlet to miss the detector, or impact the sides of the flight tube

wall. In fact, the TRAJION results indicated that significant wall collisions may occur for

slits larger than 2 mm. In addition, it was later discovered, after these series of

experiments were completed, that Teflon insulating discs located at the beginning of the

TOF mass spectrometer were acquiring a static charge from ion impact and were thus

making it extremely difficult at times to focus and steer the beam onto the detector at the

end of the mass spectrometer. The problem has since been corrected.




C- a)
00 0


0 0 0 0 0 00 0 E
c(su) I\\ HM- \E









0 10 C
C0 -- 0-

(su) lAIHM-A






0 0 0
0 CO 0
C OC )





~---4- I ~

The TRAJION results were based on an estimated 1 mm beam diameter and 10 K

energy spread in the beam translational motion. The beam spot could not be measured

experimentally with available equipment, but the SIMION ion optics model predicted a

converging beam whose focal point was positioned just in front of the parallel plates when

both Einzel lenses are held at +340 V (cf. Fig. 2-5). The SIMION model then suggests

an initial beam diameter of nearly zero which then diverges to just under 1 mm as the

beam exits the parallel plate region. Likewise, the beam kinetic energy spread was not

directly measured, but a plausible value was chosen. Thus, the values selected for input

into the computer code were likely not exact matches to the true experimental values, and

this, along with the aforementioned variability introduced into the experimental effort,

accounts for the modest differences between Figure 3-12 and Figure 3-13. Peak intensity

attenuation when going to smaller slits and faster beam deflection rise times was born out

in both the experimental effort and the TRAJION simulations.

Thus, both experiment and theory agree that by varying the deflection voltage rise

time the sensitivity/resolution trade-off can be changed. This allows for the resolution and

sensitivity to be tailored to meet the needs of each particular experiment.

The physical picture of beam deflection when the beam is not monoenergetic is

actually quite complex to visualize. The differing energies means that even ions of like

mass will experience differing displacements in the electric field between the parallel

plates; thus, one can think of the ion beam as a series of coaxial beams of differing

energies (and velocities) which are swept back and forth across the slit by varying

amounts, thus resulting in an overall elongation of the beam spot at the two sweep

position extremes. This effect is illustrated in Figure 3-14. Thus, just as expected, if the

energy spread is too great, TOF mass spectrometry cannot be successfully employed.

Fortunately, one of the hallmarks of using supersonic expansions is that they generate

exceedingly cold ion beams with closely matched velocities near the center of the

expansion where the beam is sampled by the skimmer.

xXmm ~ma~

\c Vz
*T .1 "im

Figure 3-14 Beam spot diagram for an isomass beam with varying translational energies
and therefore velocities varying between Vx,,,,, and Vx,,,. The maximum beam
displacement (at the slit plane) for the two highest static field conditions is shown.

The TRAJION code is a good design tool which certainly aids the user

tremendously in developing a better understanding, if not intuition, of the behavior of this

rather complex system. By no means has a completely comprehensive understanding of

the beam deflection technique in TOF mass spectrometry been reached, nevertheless a

sufficient working knowledge of the effects of changing various parameters was obtained

to allow the research to progress to the next stage--the formation and characterization of

cluster ions.



Having gained a basic understanding of the general behavior of the experimental

apparatus itself, the next step was to complete the characterization of the new pulsed

source and beam deflection TOF mass spectrometer by actually forming cluster ions and

comparing the results to available scientific literature.

The cluster ions were generated using three separate gases: He, N2 and CO2. The

cluster ions were produced through El ionization of the gas as it exits the nozzle. Since

this process is rather inefficient, many neutral species remain to aggregate around these

seed ions resulting in cluster ion growth. Thus, these rapid three-body ion-molecule

association reactions which occur very early in the supersonic expansion should be quite

sensitive to the initial gas density or pressure.

Saenger and Fenn 48 predicted a minimum pulse duration (valve open time) of

about 10 ls for a pulsed supersonic free jet to reach a fully developed flow comparably

cold to that obtained from a continuous source. Because of the novel in-line switching

technique used to sample the beam pulse "on the fly" for TOF mass analysis, a unique

opportunity existed to actually examine the composition of a pulsed beam by generating a

temporal cross-sectional profile of the beam pulse.

Thus, the effect of the source stagnation pressure on the observed clustering was

investigated, and the effect of where the sample is taken (or chopped) from each beam

pulse was also studied.


Each experiment was conducted under essentially the same equipment

configuration. The gas stagnation conditions were typically 105 F at variable pressures of

2 to 45 psig. The pulse valve was set for 200 ps pulses and fired at 3 or 8 Hz. The nozzle

(0.030 inch i.d.) was held at +550 V, the skimmer (2 mm i.d.) voltage was +545V and

each Einzel lens was maintained at +340 V. The electron gun and collector were

positioned 5.8 mm away radially and produced a continuous stream of 120 eV electrons

through the 1 mm gun slit opening which was axially positioned flush with the nozzle tip.

A 3 mm slit was used in obtaining the He cluster ion spectrum and the beam pulse profile

spectra; the rest of the data were collected with no slit installed. A beam deflection

voltage rise time of 30 ns was used throughout. Expansion chamber pressures varied

widely with the stagnation conditions but were generally < 5 x 10-5 torr when the valve

was firing. Post skimmer pressures in the ion optics and beam deflection regions were

routinely in the mid 10-7 torr range while the pressure near the detector was 1 x 10-6 torr.

Each gas was supplied from pressure regulated K-bottles and was introduced into

the pulse valve inlet via 1/4 inch tubing. The He and CO2 experiments were run without a

filter to remove particulate material and water vapor; the N2 work was done later, with the

benefit of a new gas manifold system and in-line filter (a Sporlan "Catch-all" filter drier

containing 5 in3 of desiccant).

Pressure effects on CO2 clustering were investigated at four different pressures:

2 psig, 15 psig, 30 psig and 45 psig. Each spectrum was accumulated from 500 single-

shot spectra.

Pulse profile data were obtained using CO2 as well. The delay from the SRS

Digital Delay/Pulse Generator into the Fast High Voltage Pulser (cf. Figure 3-2) was

varied from 500 As to 1,000 jIs while collecting data from a stable CO2 pulsed beam.

Since the SRS unit is triggered by the TTL signal from the Iota One Pulse Driver,

changing the delay of the SRS output TTL signal into the Fast High Voltage Pulser

provides a simple, yet very precise method of selecting where the TOF beamlet sample

will be chopped out of the gross beam pulse. These data were collected in a series of

accumulated spectra of 500 shots each.

Great care was taken not to introduce systematic errors while conducting the CO2

stagnation pressure and pulse profile studies. Each of these series was completed in a

single day to eliminate changes in the equipment parameters which accompany each fresh

start-up. Additionally, adjustments to the equipment while running were kept to a

minimum to ensure that the observed effects could be attributed to the variable under

study (i.e. stagnation pressure, deflection voltage delay).

Results and Discussion

The He spectrum is shown in Figure 4-1. Three He containing ion peaks are

clearly evident: He+, HeH+ and He2+. This spectrum (an accumulation of 500 single-shot

spectra) was obtained with a He stagnation pressure of 15 psig and did contain a small air

contaminant as evidenced by the characteristic nitrogen, oxygen and water peaks.

The N2 spectrum presented in Figure 4-2 shows clear evidence of nitrogen

clustering. N+, N2+, N(N2)+, (N2)2+, (N2)3, (N2)4, (N2)5+ and (N2)6+ ion peaks each

appear well above background. This spectrum was obtained with an N2 stagnation

pressure of 15 psig as well, and represents the accumulation of 1,000 single-shot spectra.

Small H20+ and 02+ peaks are also seen.

Figure 4-3 displays the results of expanding CO2 with a stagnation pressure of

45 psig. A rich array of cluster ions resulted. The primary series observed was (CO2)n+

where n = 1 to 7; however, mixed clusters resulting from further additions of H20 and 02

were also detected. Thus, (CO2),(H2O)m+ where m = 1 or 2, and (CO2),O2+ were formed,

and are due to impurities in the gas handling system.





U 2
U U ci)









4. '-N
en 0
Cl 777

7: 0
- 2
6 Ca


+ Cl
C .4.






The results of these three cluster ion experiments demonstrated that the ion source

and beam deflection TOF mass spectrometer were indeed fully operational. The initial

design and construction of the ion source and ion optics was patterned after the high

intensity ion source developed by Young Bae et al. at SRI International.31 Though some

significant equipment differences exist, our results compare quite favorably to those

reported by Young Bae's group for (He),-, (N2)n+ and (CO2)+n cluster ion genesis.

However, slightly smaller clusters were observed in our apparatus (n _< 7) as compared to

theirs (n < 10)

Johnson et al.29 reported the synthesis of CO2 cluster ions by successive nucleation

onto a monomer ion during a free jet expansion as a function of the distance between the

electron beam (200 eV energy, 100 /AA, 1 mm diameter) and the nozzle (1.0 mm diameter,

1.5 atm. CO2 stagnation pressure at room temperature). As per Figure 4-4, the results

from a distance of 2.8 cm show only a single peak from the CO2+ monomer--thus, in the

absence of subsequent collisions, the primary ion produced by electron impact is the

monomer, with very little contribution to higher clusters from the ionization of neutral

clusters formed in the jet. However, they found the degree of clustering increased

dramatically as the electron beam was moved toward the nozzle. Higher clusters were

successively formed until, at 0.2 cm distance, a distribution peaking at n = 7 and

continuing through about 17 was obtained. At that distance (0.2 cm), the pulsed beam

density was estimated to be about 1017/cm3, sufficiently high that three-body clustering

represents a significant loss mechanism for the nascent CO2+ species formed by electron


Though the effect of varying the distance between the nozzle and electron gun in

our system was not investigated, it seems quite probable that a similar proximity effect

would exist for a well collimated e-beam. Because of the simple e-gun design utilized in

this work (electrons are accelerated through a 1 mm wide slit without the benefit of any

focusing lenses or steering plates, cf. Figure 2-3) the electron stream is probably not well

(c) 4
0.4 cm


(d) 0.2 cm


0 20 40
Time of flight (j3s)

Figure 4-4 TOF mass spectra as the distance between the electron beam and the nozzle is
varied. The clusters are the result of successive attachments of CO2 to CO24, not from the
ionization of neutral CO2 clusters. The sharp features between the main peaks correspond
to (CO2)n+ with added 02 or H2 and are due to impurities in the gas handling system;
collision-induced dissociation of the higher clusters during the extraction from the high-
density jet give rise to the broad peaks. (Taken from Johnson et al.29)

collimated, but is more likely characterized as a spray; thus, any attempt to make fine

adjustments to where the e-beam intersects the expanding jet were deemed unpromising,

especially in light of the good clustering results already obtained.

The effect of the source stagnation pressure on the beam intensity and degree of

clustering are shown in Figure 4-5 for CO2 gas pressures of 2, 15, 30 and 45 psig. The

primary (CO2),+ cluster ion series shows increased clustering with increased pressure as

expected. At 2 psig n < 4, whereas at a CO2 stagnation pressure of 45 psig the maximum

cluster size, n, increased to 7. Thus, an effective means of increasing the associative

collisions which result in cluster growth is to simply increase the backing pressure of the

source gas.

The CO2 temporal beam pulse profile data are presented in Figures 4-6 and 4-7.

Beam time slices taken in increasing 100 p/s increments, beginning at 500 /s delay from

the pulse valve driver TTL signal and continuing up to a full 1 ms delay, are shown in

Figure 4-6. As clearly seen from the spectra, the most dramatic change in beam intensity

takes place during the first 100 /s of the beam pulse leading edge. Figure 4-7 contains the

results of a much closer examination of the first 100 ps of the beam pulse leading edge.

The beam slices taken in 10 /s time increments reveal a rather sharp increase in intensity

which reaches a maximum at about 550 ps delay and then begins to slowly decrease. This

behavior agrees quite well with the undeflected beam pulse shapes observed on the DSO

while adjusting the valve travel, steering plate voltages, etc. when the beam deflection

voltage was turned off (see Figure 4-8).

Still the question remains as to whether any portion of the pulsed beam is

comparably cold to that produced by a continuous beam under the same source

conditions--or in other words, how much time is required for the pulsed supersonic jet to

reach fully developed flow? A simple model was developed by Saenger and Fenn48 for

estimating the minimum valve open time for a pulsed nozzle to produce a beam like that

j A ~i. ~ A A I.

it .... ..... .... A.... ... .........

i 1 IL ~a


a .i A _

100 0

TOF (e/s)

TOF (1s)

Figure 4-5 TOF mass spectra as a function of the CO2 source stagnation pressure.
Cluster results from (a) 2 psig, (b) 15 psig, (c) 30 psig, and (d) 45 psig backing pressures
are shown. Each spectrum was accumulated from 500 single-shot spectra and is presented
at the same scale.

500/ts (d)

600tts (e)

700/As (0



0 100 0 100
TOF (/As) TOF (1s)

Figure 4-6 CO2 pulse profile. TOF mass spectra taken at increasing intervals of 100 /s
into the beam pulse. Each spectrum represents the accumulation of 500 single-shot
spectra and is presented at the same scale.


0 100 0 100
TOF (1s) TOF (j/s)

Figure 4-7 Profile of the CO2 pulse leading edge. TOF mass spectra taken at increasing
intervals of 10 /xs into the leading edge of the beam pulse. Each spectrum represents the
accumulation of 500 single-shot spectra and is presented at the same scale. (continued on
next page)

0 TOF (TOF (s)
TOF (/s)

Figure 4-7 (continued)

a I
a a

a I


a __________

a _____________

a I
a I

produced from a continuous source. They conclude by citing a general rule of thumb for

estimating the minimum valve open time:

At = 4d / ao (Eq. 4-1)

where d is the nozzle diameter. The value ao is determined by the stagnation temperature,

To, and is given by:

ao = (ykT0 / m)' (Eq. 4-2)

where 7 is the ratio of the constant pressure and constant volume heat capacities, k is the

Boltzmann constant, and m is the molecular weight of the source gas.

However, since their model assumed the valve was an ideal shutter which opens

and closes instantaneously, it simply serves only to define the minimum time that a valve

must be fully open in order to insure at least some of the molecules reach a terminal state

corresponding to the steady flow prescription for equivalent stagnation conditions. As

Saenger and Fenn pointed out:

...it would seem a reasonably safe rule of thumb to require that the valve
open time be at least 4d/ao. However, even this "conservative" estimate
must be regarded only as a lower limit. No real valve has an opening
mechanism that approximates the ideal shutter assumed in these
calculations. Moreover, all the designs to date have dead volumes that act
as capacitances and/or flow restrictions that may cause the effective values
of Po [the stagnation pressure] and d to be less than their nominal values, at
least for part of the pulse cycle. These factors will greatly complicate the
calculation of true flow time requirements. Even rough estimate of their
effects can only be made with reference to a specific valve design.48

It would appear from the pulse profile data obtained for CO2, that the maximum

flow from the pulsed nozzle occurs about 50 /As after the valve has begun to open. This

result agrees well with the valve response characteristics reported by General Valve53

which suggest -50 ls is required for the valve to fully open. However, the pulse

intensity is never truly "flat topped" but actually begins to decrease slowly beyond this

initial sharp rise. It should be noted, moreover, that the nozzle used in these experiments

is actually a stainless-steel capillary 46.5 mm long and thus a free jet expansion does not

really exist, but rather a capillary expansion. As noted by Saenger and Fenn, flow

restrictions and capacitance effects from such a long nozzle result in exceedingly complex

flow conditions.

The distribution of the heavy-mass beam components appears to be slightly

skewed toward the trailing end of the beam pulse--at least by the time it arrives over the

deflection plates. This may in fact be due to the separation of the isomass groups within

the beam pulse which occurs during the flight to the beam deflection region. If the beam

pulse from our apparatus is indeed peaked (as all the experimental evidence indicates) then

it is not at all unreasonable to imagine that the separation of these peaked isomass

distributions would account for the slightly increased density of higher mass beam

components found when chopping further into the pulse.

Because the beam pulse is affected not only by the pulse valve and nozzle

hardware, but also by the source gas flow properties, the delay between the TTL pulse and

arrival of the maximum beam pulse intensity is not a constant, but in fact changes slightly

depending on the source gas employed. The optimum beam deflection delay can be

quickly found by simply adjusting the SRS Digital Delay unit for the maximum peak

intensities on the DSO. This procedure was, in fact, adopted from the very start and was

generally used to determine the appropriate timing delay prior to the start of any data




Even while calibrations and final equipment corrections were still being made, we

observed that when Freon 116 (C2F6) was used as the source gas, a curious peak at mass

97 frequently appeared; the peak was sometimes quite pronounced while at other times it

was barely visible--or not present at all. It was also noted that trace amounts of H20, N2

and/or 02 were often visible in the mass spectra due to air impurities introduced into the

gas handling system.

Three plausible explanations of the peak origin were considered The peak could

be due to some impurity in the Freon 116 bottle (i.e. a hydrocarbon impurity) in which

case it should appear whenever Freon 116 is present, and its intensity should be

proportional to the Freon 116 concentration. A second possibility was that the peak was

due to C,2F3O, which would ostensibly require an ion-molecule reaction involving C2F5+

and 02, or perhaps 02+ and C2F6, and should therefore only occur in the presence of

oxygen. The formation of the mixed cluster CF3N2+ provides yet a third alternative, and

would of course require the presence of nitrogen. A meticulous literature search indicated

that the existence of CF3N2+ had not been previously reported.

Early data collected without the benefit of gas manifold and filter system suggested

a correlation existed between the appearance of the characteristic air peaks and the

occurrence of the 97 amu peak. Even after installing a manifold and filter, numerous

attempts to completely eliminate any traces of H20, N2 or 02 were unsuccessful, but a

reproducible trend did emerge. The 97 peak could be observed while supplying C2F6 to a

gas mixing vessel initially filled with N2. As the C2F6 concentration increased, the

characteristic El ion peaks would increase in intensity as well, while the N2 peaks

diminished. Of particular interest was the fact that the 97 peak, though initially not

present at all with the pure N2, would grow in and reach a maximum, then die away almost

completely, even as the C2F6 concentration continued to increase.

Additionally, the 97 amu peak maximum appeared to be correlated to the

appearance of high-mass ion peaks (m > 138 amu). These peaks appeared only during a

certain N2 : C2F6 mixture concentration range, and indicate the formation of cluster ions

since their mass exceeded that of any possible parent ion. In fact, a peak at mass 147 was

often observed in coincidence with the 97 amu peak and could conceivably be attributed

to C2F5N2, since CF3+ and C2F5+ are the two most abundant ions produced during the El

ionization of C2F6.

Though the experimental evidence seemed to point toward the existence of CF3N2+

(and C2F5N2), it remained to additional, carefully planned experiments to show

definitively which of the three proposed explanations (or combination of explanations) was

indeed correct. If CF3N2+ and C2F5N2, are indeed being formed during the coexpansion of

C2F6 and N2, then a +2 amu mass shift should occur when 15N2 is used instead of 14N2;

the 97 and 147 amu peaks should move to 99 and 149 amu respectively. The occurrence

of such a shift would suffice to establish the existence of the proposed mixed clusters.


A stainless steel mixing vessel (-40 mL capacity) equipped with a gas manifold

and Bourdon pressure gauge was used to control the source gas composition. The

manifold assembly was constructed from four Nupro SS-4H packless, bellows valves and

1/4 inch o.d. stainless steel tubing; the gas handling system was therefore completely

bakeable. Both the C2F6 and the N2 were supplied from pressure regulated bottles

through 1/4 inch o.d. copper tubing to the gas manifold system. The manifold valve

design allowed the mixing chamber to be evacuated via a mechanical pump without

breaking any of the gas lines. Additionally, the pressure of each supplied gas could be

measured using the single mechanical pressure gauge. The mixed gas then passed through

a filter/drier and entered the stagnation region around the pulse valve.

A final attempt was made to eliminate any trace amounts of H20 or 02 from the

gas handling system through the use of standard vacuum techniques. As in many of the

previous C2F6/N2 coexpansion experiments, the mixing vessel, while not in use, was

evacuated through the mechanical pump and baked with heat tape at 150F; however,

for this particular run, after the mixing vessel had cooled down the morning of the

experiment, the filter/drier was packed in dry ice to create a cryogenic pump to remove

water and other low volatiles from the source gas stream. The Freon 116 used in these

experiments was procured from Matheson and the natural abundance N2 was obtained

from BiTech.

The experiment was begun with only N2 at 15 psig supplied to the mixing vessel--

the C2F6 manifold valve was closed. After the equipment parameters had been adjusted

and a good N2 spectrum had been obtained, the N2 manifold valve was closed and the

C2F6 valve was opened, allowing only 15 psig C2F6 to flow into the mixing vessel as the

pulse valve continued firing. Thus, the C2F6 concentration, though initially zero, gradually

increased over time as the N2 concentration decreased. This dynamic concentration effect

was monitored spectroscopically by acquiring TOF mass spectra of the pulsed beam every

15 minutes; each spectrum was accumulated from 2,000 single-shot spectra acquired at a

collection rate of 4 Hz.

The system was configured with no slit, and a deflection voltage rise time of 30 ns

was used throughout. The pulse valve was set at 200 ps duration and 8 Hz. The source

gas stagnation temperature was 1080F. The 0.030 inch i.d. nozzle was held at +550 V,

while the 2 mm i.d. skimmer was maintained at +545 V. Both Einzel lenses were held at

+340 V. As always, the electron gun supplied 120 eV electrons with a measured current

of 200 iA at the collection plate. Typical pressures during the run were < 5 x 10-5 torr

in the ion source chamber, 5 x 10-7 torr in the ion optics/beam deflection region, and

1 x 10-6 torr near the detector.

Because of the sensitivity/resolution trade off mentioned in Chapter 3, a second

series of spectra were collected using a 0.5 mm slit and deflection voltage rise time of

2200 ns in order to gain better resolution. (A faster deflection voltage rise time was not

used due to unacceptable reduction in S/N). The experimental parameters were essentially

the same except the first Einzel lens was held at +340 V and the second Einzel lens was

kept at +200 V; this Einzel lens configuration was found to yield better peak resolution

when employing a 0.5 mm slit and deflection voltage rise time of 2200 ns. Also, the

filter/drier was not cooled with dry ice for this series of spectra, but was simply left at

room temperature.

A third experiment was conducted very much like the one described above except
15N2 (99 atom %) was used. One liter of 15N2 was obtained from Cryogenic Rare Gas at a

cost of $460. In order to observe a 2 amu isotopic shift, high resolution was deemed

more desirable than high sensitivity; thus, the 0.5 mm slit configuration was employed.

System parameters were optimized using 14N2 before introducing 15N2 into the system.

The filter/drier was not cooled.

As before, the mixing vessel was initially filled with '5N2, but a pressure of only

9 psig was possible due to the rather large volume of the lecture bottle in which the 1 L of

1sN. was supplied. After the initial spectrum of pure 5sN2 was obtained, the pressure had

fallen to 7 psig and this, then, was the stagnation pressure maintained throughout the rest

of the experiment through the addition of C2F6 at 7 psig (the 5sN2 manifold valve was now

closed). As in the previous two experiments, 2,000 single-shot spectra were accumulated

every 15 minutes and used to track the ion beam composition as the relative

concentrations of C2F6 and I5N2 changed with time. The nozzle and skimmer potentials

were at +550 V and +545 V respectively, the first Einzel lens was held at +340 V, and

the second Einzel lens was kept at +200 V. Therefore, aside from the source gas

pressure changes, essentially the same system conditions existed as were noted for the

14N2 experiment using a 0.5 mm slit.

Results and Discussion

The results of the C2F6/14N2 coexpansion experiment conducted without a slit are

displayed in Figure 5-1. Figure 5-1(a) was taken prior to opening the C2F6 manifold

valve, and therefore establishes the mass spectrum arising from the expansion of

essentially pure nitrogen without the presence of C2F6. Figure 5-1(b) was collected

immediately after the C2F6 valve was opened. The series shows the change in the ion

beam composition as the source gas ratio of C2F6: 14N2 was increased. The last spectrum,

Figure 5-1(q), was obtained after evacuating the mixing vessel, and then filling it with

C2F6 at 15 psig; it therefore establishes the mass spectrum arising from the expansion of

essentially pure C2F6.

In an attempt to better depict the overall trend observed during the 3 1/2 hours in

which the experiment was run, every other spectrum (9 of the 17 total) were carefully

analyzed and converted into the stacked, 3-dimensional graph shown in Figure 5-2. Each

spectrum was painstakingly analyzed to establish each peak position as precisely as

possible. Normalized peak intensities were also estimated by triangularizing each peak

and computing the area contained within. Flight times were also converted into ion

masses. The peak width measured for the N2+ peak (28 amu) was -270 ns FWHM for a

resolution (m/Am) of 51.

Figures 5-1 and 5-2 both clearly show enhanced cluster composition within the

beam occurring only during a certain time period; of course, since the source gas

(a) (d)



(b) (e)

(c) (0

0 200 0 200
TOF (jps) TOF (tts)

Figure 5-1 TOF mass spectra of N2/C2F6 coexpansion. (a) Pure N2; (b) Initial spectrum
taken when addition of C2F6 was started and N2 valved off; Spectra (c) (p) were taken at
15 minute intervals as [C2F6] increased; (q) pure C2F6. (continued on next two pages)



0 200 0
TOF (p/s)

Figure 5-1 (continued)

TOF (p/s)

TOF (p/s)

Figure 5-1 (continued)

200 0

TOF (1s)




o G




oD C0 0 C C C0
o) C ) ) C
LO 'T co (N IC


*X 0




C)( a a) C _

V() 'I4CM CNI-*

,kiISNqiN I




I 0

CD C) CD Cj C) C
I i "'- '' -

-.. .

L0 ^- CO C r<-I


composition is gradually changing with time, this indicates that the formation of the heavy-

mass clusters is favored only when a specific C2F6:14N2 composition is present in the

source gas. The 97 peak intensity is also greatest during this period of enhanced

clustering; a closer view of this segment of the mass spectrum is shown in Figures 5-2(b)

and 5-2(c).

A similar series was obtained using the 0.5 mm slit. Though the sensitivity was

reduced significantly from that obtained without a slit, the resolution measured for the N,2

peak increased to 199 (m/Am). Thus, by combining the information from both runs, a

more precise understanding of the system behavior can be had. Since the 0.5 mm slit run

provided better resolution, the spectrum containing the highest intensity 97 peak was

chosen for comparison with the C2F6/15N2 results in which an isotopic shift is predicted.

The spectrum selected is presented in Figure 5-3 along with a calibration curve in Figure

5-4 of the square root of the ion mass vs. the TOF.

The spectra collected during the C2F6/15N2 experiment were characterized by

reduced peak intensities due to the lower source pressure which we were forced to use.

However, the suspected mixed clustering was still observed with a resolution comparable

to that of the C2F6/14N2 work using the 0.5 mm slit. Again, the best spectrum was chosen,

in which the peaks of interest were most pronounced, to compare with the best C2F6/14N2

spectrum. The selected spectrum along with a mass calibration curve are shown in

Figures 5-5 and 5-6 respectively.

Thus, the outcome of the 15N isotope study remained in the careful analysis of two

spectra. A mass shift of +2 amu was indeed observed for the 97 and 147 peaks as shown

in Figure 5-7. As depicted in the graphical comparison of the two spectra, the large CF3

and C2F5 peaks are aligned, as well as the peaks identified as CF3(H2O)+, CF3(C2zF6) and

C2F5(C2F6). It should be noted that a rather large (N,2), peak appears in the C2F6/14N2
spectrum and not in the C2F6/15N2 spectrum. This is not totally unexpected since the '5N2

stagnation pressure was only about half that for the 14N2 run; the stagnation conditions of






z 2


^ -_____ -- -- 00


m U

I \(/)
1 ..
-- 1- ,

w 2 0 0 0 0co 0
O -r

" CO !-- CO CC O ,
(sn) O .1
0 V I--






~ E