Citation
Sputtering and ionization by helium and argon in the microsecond pulsed glow discharge using time-of-flight mass spectrometry

Material Information

Title:
Sputtering and ionization by helium and argon in the microsecond pulsed glow discharge using time-of-flight mass spectrometry
Creator:
Mohill, Matthew, 1974-
Publication Date:
Language:
English
Physical Description:
viii, 208 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Argon ( jstor )
Atoms ( jstor )
Electric potential ( jstor )
Glow discharges ( jstor )
Helium ( jstor )
Ionization ( jstor )
Ions ( jstor )
Mass spectra ( jstor )
Plasmas ( jstor )
Signals ( jstor )
Argon ( lcsh )
Chemistry thesis, Ph. D ( lcsh )
Dissertations, Academic -- Chemistry -- UF ( lcsh )
Glow discharges ( lcsh )
Helium ( lcsh )
Mass spectrometry ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2001.
Bibliography:
Includes bibliographical references (leaves 203-207).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Matthew Mohill.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
026991083 ( ALEPH )
47283626 ( OCLC )

Downloads

This item has the following downloads:


Full Text










SPUTTERING AND IONIZATION BY HELIUM AND ARGON IN THE
MICROSECOND PULSED GLOW DISCHARGE USING TIME-OF-FLIGHT MASS
SPECTROMETRY











By

MATTHEW MOHILL


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE
UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE
REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY


UNIVERSITY OF FLORIDA


2001











TABLE OF CONTENTS

Page

ACKNOWLEDGEMENTS........................................................................v

A B ST R A C T ... ................. ........ ... .. ... ................................. ................vii

CHAPTERS ...........................................................................................

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

Solids Elemental Mass Spectrometry .....................................................1
Solids Analysis Techniques for Mass Spectrometry ..................................3
Glow Discharge and GDMS History ....................................................5
Scope of this Dissertation ..................................................................7

2 THE GLOW DISCHARGE ...............................................................9

Gaseous Discharges ......................... .......... ........................ ...... 9
The Abnormal Glow Discharge ........................................................... 12
Regions of the Glow Discharge .................................................... 14
Glow Discharge Processes ......................... ....... ...........................18
Glow Discharge Sputtering ................................................... 18
Mass of sputtering ion ..................................................23
Angle of incident ion bombardment .................................. 23
Incident ion energy ........................ ........ ..................25
The target material ................................................... .....29
Glow Discharge Excitation/Ionization .... .......... .........................31
Collisions in Gases ............ .......................................... ......... 31
Sputtered Species Excitation .............................................................35
Sputtered Species Ionization ............................................................ 38
Secondary Excitation/Ionization Processes ...........................................42
Types of Glow Discharge Devices .....................................................44

3 EXPERIMENTAL CONSIDERATIONS ...........................................48

Po%\ering the Glow Discharge ..........................................................48
D irect C current ......................................... ................... ....... 48
RadioFrequency ......................................................... ........ ..50
The Microsecond Pulse ..........................................................51







Pulsed Glow Discharge Mass Spectrometry ....................................... 51
DC vs p.s-Pulsed Experimental Considerations ....................................... 52
Sputtering and Ionization in the Microsecond Pulsed GD ...........................53
Analytical Advantages of DC GD ................................................... 54
Analytical Advantages of Microsecond Pulsed GD ...................................59
Atomic Emission Signal Enhancement ..................... ................59
Sputtered Spcies Ion Signal Enhancement ....................... .........59
Temporal Consideratinos in GDMS ............................................61
Glow Discharge Time-of-Flight Mass Spectrometry ...............................68
Introduction ........................................... ........................ .. 68
Time-of-Flight Mass Spectrometry ...........................................73
T theory ................ ........................................ ... ........ 73
Outpur Characteristics ........................................... ....... 74
Performance Limitations and Potential Capabilities ...............77
Inherent Advantages of TOFMS ...................... .............78
Material Transport and Instument Design ...........................................80
Ion Source .................................. ................................ 80
Direct Insertion Probe ........................................................80
Discharge Cell ...................................... ................. 82
Ion Extraction ........................................................... 84
Ion Optics and TOF Sampling ..........................................87
Electronics and Timing .............................................. 88
V acuum System .................................. ............ ........... 90
Ion Detection ............................................................. 90
General Procedure and Data Analysis .................................................91

4 CHARACTERIZATION OF THE MICROSECOND PULSED GD PROCESSES
IN HELIUM .............................................................. ......... 93

Introduction ............................................ ............................ ... .. 94
Background ............ ........................................ .................... 96
Microsecond Pulsed Glow Discharge Processes: Comparing Ar and He .........99
Discharge Initiation ............................. ....................99
Electron Ionization .................................. ... .........100
Voltage effect ... ... .................................. 105
Pressure effect .................................... ............ 107
Sputtering of Atoms ........... ....................... ................07
Incident ion energy .............................................113
Pressure effect .............................................. 115
V oltage effect ....................... ........... .............. 17
Diffusion of Atoms .................................................... 119
Ionization of Atoms ................................................. 126
Sam pling ........................................................... ..... 127
Sum m ary ................................................................. ................ 134







5 THE HELIUM DISCHARGE ANALYTICAL CHARACTERISTICS ........... 138

Introduction .......................................................... .. ..............138
Identifying Interferences .................................... ............................138
Interfering Ion Formation ..................................................... ....... 139
Rem oval of Interferences ................................ ................................ 141
Determination of Calcium in the Presence of Argon .................................147
Relative Sensitivity Factors ............................................................ 151
Relative Ionization Factor ............ .................... ................. ......... ... 160
Summary ...................................... ................ ....... 163

6 CLUSTER FORMATION AND IONIZATION IN A iS-PULSED GD .........164

Introduction ....................................... ................................. ...164
Background .......... ....................................................... 165
Sputtering Theory ................... .......... .... ...............................167
Cluster Form action Theory ................................................ ........ ........ 170
Direct em mission ............................ ............................... 170
Atomic Combination (AC) Single Collision .............................173
Atomic Combination by Double Collision ................................. 174
Theory Application to a Glow Discharge ................... ................... 176
Diffusion, Dissociation, and Ionization of Polyatomics .............................180
Matrix Species Distribution: A Cathode-to-Orifice Distance Study ..............188
The Effect of Pressure on Polyatomic Formation ................................. 195
The Effect of Voltage on Polyatomic Formation ...................................97
Sum m ary ............................... ................... ....................... ... 197

7 CONCLUDING REMARKS AND FUTURE DIRECTIONS ....................200

8 REFERENCES .......................................................... ....... 204

9 BIOGRAPHICAL SKETCH ................................. ........................ 210














ACKNOWLEDGEMENTS

It is with much pleasure and much sorrow that my graduate career is coming to an

end. I am indebted to many people who have not only been mentors but also friends. I

have the utmost respect for my advisor, Dr. W.W. Harrison, who has taught me scientific

integrity and discipline, and has guided me through the process of writing this

dissertation.

The work here is but one of many that have been performed in the Harrison/

Winefordner group laboratory, and would not have been possible without the support and

help from my peers. Special recognition goes out to Dr. Ben Smith. He has been an

active part of the success of my laboratory work and has enriched my time here at

University of Florida. Dr. Wei Hang and Dr. Kristofor Ingeneri introduced me to GDMS

and taught me much about the technique. I must thank Kris for many late night talks. Dr.

Chenglong Yang deserves special attention. Chenglong was a great laboratory mentor

and friend. His hard work and help in the lab were deeply appreciated. I must extend

thanks to the rest of the members of the Harrison and Winefordner group, new and old. I

will miss their friendship and lively banter.

I also wish to thank Dr. Christopher Barshick. He has been nothing short of a

great friend over the last five years, and was the one who convinced me to join the

Harrison group. I thank him.






A final word must be said about my family and close friends. I would not be the

person I am today without the love and support of my parents; through the very lows and

very highs. they have kept me on an even plane. I will forever be indebted to them for

this and I honor them with this dissertation. Thanks, Mom and Dad!!!

This work has been supported by the US Department of Energy, Basic Energy

Sciences, Hewlett-Packard Laboratories, and LECO Corporation.









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

SPUTTERING AND IONIZATION BY HELIUM AND ARGON IN THE
MICROSECOND PULSED GLOW DISCHARGE USING TIME-OF-FLIGHT MASS
SPECTROMETRY

By

Matthew Mohill

MAY 2001


Chair: Willard Harrison
Major Department: Chemistry

The focus of this dissertation is the study of the sputtering and ionization

phenomena in the microsecond pulsed glow discharge using helium and argon as plasma

gases. The glow discharge is an electrical plasma which directly atomizes and ionizes

solid metal samples. In this dissertation, a short, intense voltage pulse is applied to the

sample creating temporal aspects that are both fundamentally and analytically useful. An

in-house built reflectron time-of-flight mass spectrometer is used in these experiments in

which several ion source configurations can be used, including a direct insertion probe

and a Grimm-type device.

The ionization phenomena in helium and argon are studied. The high helium

metastable atom energy results in higher ionization capabilities. The pts-pulse mode must

be used in order to produce higher analyte intensities, that are not achievable in the DC

mode. The temporal advantages and selective ionization of sputtered species versus ga.i







species afforded by argon are lost when helium is used as the plasma gas. The removal

of mass peaks in critical parts of the mass spectrum, without losing sensitivity, is a

benefit of the use of helium as the plasma gas.

The impact of argon and helium ions, and the resulting collision cascade, were

studied relative to mass spectra observed for a copper matrix. Polyatomic clusters of the

copper matrix form and interfere with several elements in the mass spectrum. The

formation, dissociation, and ionization of these species are detailed in this dissertation.

The analytical advantages of using helium as the plasma gas to effect detection of

iron, selenium, arsenic, tellurium, gold, and lead are discussed. For lead and tellurium, a

reduction in the formation of polyatomic clusters results in the removal of these mass

interference. Relative sensitivity factors are calculated for both helium and argon,

showing that the higher metastable atom energies of helium make it a less selective

ionization agent resulting in signal enhancement of hard-to-ionize elements such as As,

Se, and Au. A relative ionization factor is defined which measures the sputtering and

ionization efficiency of helium compared with argon. In general. argon ions are efficient

at sputtering a copper target, but argon metastable atoms are only moderately able to

ionize the atom population produced. Helium is an inefficient sputtering agent. but an

excellent ionizing agent for the species present in a glow discharge source.













CHAPTER 1
INTRODUCTION

Solids Elemental Mass Spectrometry

Elemental analysis of solids employs a wide variety of detection techniques, of

which mass spectrometry is one. A mass spectrometer requires the production of ions for

analysis and subsequently separates them by their mass-to-charge (m/z) ratio. Inorganic

elemental analysis has traditionally been dominated by optical methods in which the

excitation and subsequent radiative relaxation of the analyte produce photons

characteristic of the energy difference between the accessible excited and relaxed states

of the analyte atoms. These photons are measured using spectrometers, diode arrays, and

other similar detection devices that convert photons to electrons. The difficulty for optical

methods is not in the analysis of pure materials, but rather in the analysis of multi-

element materials in which transitions of several elements may have similar energy such

that an overlap of wavelength occurs. This spectral complexity, with many transitions for

a single element, is quite opposite of what is seen in mass spectrometry, where the

elements are typically singly ionized and typically have few isotopes. Mass spectrometry

also suffers from mass interference; however, there are many schemes that combat

isobaric interference in glow discharge mass spectrometry (GDMS). In general, spectral

simplicity in MS is advantageous.

Mass spectrometry has been utilized since the early 1900's \whcn J.J. Thomson

observed "positive rays" from a low-pressure electrical discharge.' What we call positive










ions were studied by Thomson for isotopic abundance measurements, material analy sis.

and the physics of the electrical discharge. The electrical discharge used by Thomson

continued to be developed as an ion source for material and isotopic analysis, and

eventually a need for improved precision and higher sensitivity in mass spectrometers

arose. The ability to detect lower abundance isotopes of various elements was satisfied by

the continuing development of mass spectrometers, in particular the quadrupole.2 The

development of quadrupole mass spectrometry, in the 1950s, would become important

for organic analysis. Inorganic analysis would also benefit from the maturation of the

technique and become a popular mass spectrometric technique in the field.

With improvements in mass spectrometers, the construction of ion sources more

suitable than the Thomson electrical discharge were eventually developed. Aston created

an ion source for low-volatility solids such as refractory metals and metal oxides,3 and

small improvements continued with the Thomson electrical discharge.4 The big

breakthrough in ion sources for solid samples was the vacuum spark.56 The spark is

produced by a high voltage at high frequency between two electrodes (typically wires) to

vaporize and ionize solid material from one of the electrodes. The spark was originally

used as an optical emission device, and was slowly developed into a source of ions for

mass spectrometry. The spark source had the advantage of covering the full elementIal

range with typical detection limits of 0.1 ppm reported by Boumans and Gijbels.78












Solids Analysis Techniques for Mass Spectrometry

Techniques for the direct analysis of solids must fulfill three basic requirements:

1) vaporization/destruction of the sample; 2) atomization of the sample; and 3) ionization.

The techniques that are capable of performing all three in a single environment are ones

in which there is interaction of a plasma and or gas ion with a solid surface, where

particle bombardment at the surface and particle collisions in the gas phase satisfy the

requirements.

Energy is needed in order to effect requirement #1 above. The energy applied in

plasmas occurs in two distinct forms: electrical energy and optical radiation. Glow

discharges, spark sources, microwave and ICP plasmas all use electrical energy, while

pulsed lasers use optical energy produced by photon flux. The energy in particle beam

sources is strictly kinetic. Secondary ion mass spectrometry (SIMS) and fast atom

bombardment (FAB) mass spectrometries both are particle beam sources that take

advantage of surface chemistry to produce ions. A discussion of kinetic energy transfer

via elastic collisions follows in Chapter 2.

As is the case for the development of the spark source for mass spectrometry, the

development of new ion sources is driven by the status of material science, which

requires precision and accuracy in analyzing a sample, particularly for sample impurities.

Surface techniques used in materials science have also become critical areas of study in

analytical chemistry. Requirements for knowing the amount and spatial distribution of

impurities on the surface of a solid material have created the need for better










characterization techniques. In developing these new ion sources, the complexity of the

mass spectrometer system and type of mass spectra observed are all determined by the

analysis sought and the ion source used. These entities are then responsible for the

calibration capabilities and the quality of the mass spectra observed. In the development

of these sources, research has focused on the decoupling principle in which the

atomization and ionization steps are physically separate. For example, in using a laser for

solid sample analysis, the laser is used to vaporize the sample; the sample particles are

then transported into an ICP plasma for subsequent ionization. Ionization in direct laser

analysis does occur, but an ICP is more often used because of the wide range of ion

kinetic energies that are produced during laser analysis. In the case of a glow discharge,

the atomization and ionization steps don't occur at the same time even though they take

place in the same plasma volume in which no physical transport of the atoms is required

for ionization to occur.

The spark source mass spectrometer system was the first big breakthrough in

solids analysis. The source was originally coupled to the Mattauch-I lrzog double

focusing mass spectrometer, capable of energy and angular focusing in a single plane.

Due to the wide energy distribution and the erratic nature of the spark source, this mass

spectrometer system was the only one capable of reducing this energy distribution for

accurate analysis of the ions produced simultaneously.

In the late 1940s and earl\ '50s, secondary ion mass spectrometry (SIMS) \\a.

developed to solve the lack of precision and accuracy of the spark source.9 J.J. Thomson

x as the first to observe the ejection of secondary ions from a surface after primary ion










bombardment,9 and the first instrument of this kind was developed by Herzog and

Viehbock in 1949.10 In the 1970s, development of SIMS as a method for detecting

organic compounds on a surface became the primary focus and continues to be one of the

leading applications. 1-12 The development of SIMS was most likely headed in the

direction of organic analysis and left the inorganic elemental analysis field open for

improvement.

Plasma source development included the work of Fassel13 with the ICP and later

its combination with a quadrupole mass analyzer for the elemental analysis of solutions.

The ICP combined with mass spectrometry has developed into the most powerful

elemental analysis technique. Microwave induced plasmas (MIP) were also developed

for solution-based elemental mass spectrometry,14 yet the MIP didn't develop into a

commercial success. The MIP is still developing new directions including the plasma

torch15 and other configurations that take advantage of its excitation capabilities as

opposed to its atomization and ionization capabilities. Over the same time period, glow

discharge development led to its use in the plasma mass spectrometry field. The

developmental history of glow discharge into an analytical technique follows in the next

section. The use of a glow discharge as an ion source was a critical point in this history

and is important for this dissertation.



Glow Discharge and GDMS History

Glow discharge devices have long been used in atomic spectroscopy and are

routinely used for analysis of solid materials. Hollow cathodes were used by Paschen to










investigate atomic structure by atomic emission and absorption.16 The use of hollow

cathodes was limited to analytical atomic emission as a means of determining low

concentrations of elements across the periodic table, including nonmetals. In the five

decades that followed Paschen's work, the applications of GD devices in chemistry

developed slowly, mainly due a lack of instrumental advancement and a poor

understanding of the analytical capabilities of the glow discharge. The last 30-40 \ ears

has seen the slow growth and maturation of the glow discharges as a technique for solids

analysis, including the development of commercial instrumentation. A need for more

routine analysis resulted in the development of "restricted" glow discharges by Grimm,'7

in which flat samples were used directly as the cathode material. Just as hollow cathodes

grew in popularity when Paschen first introduced them, so did the Grimm-type GD

device. It has become the metal industry standard for bulk analysis and particularly for

depth profiling by atomic emission spectrometry.



Glow Discharge Mass Spectrometry

Analytical glow discharge mass spectrometry has been around since the early

1970s, \%hcen Coburn reintroduced the glow discharge as a solids ion source for the

analytical community.'8 Coburn was able to show the utility of monitoring the ions

produced by a sputtered target in both direct current (DC) and radiofrequency (RF)

discharges. In the mid-to-late 1970s, the development of a quadrupole-based GDMS

s stem was undertaken by the Ilarrison group at the Uni\ersity of Virginia. in whichh bulk

solid and solution residue analysis were the main focus. Radiofrequency (RF) and direct










current (DC) modes are the traditional and more commonly used modes of operating the

glow discharge. RF is primarily used for non-conducting materials, which cannot be

analyzed in the DC mode. The DC mode is more widely used, especially in the metals

industry, where trace analysis is important. Pulsed discharges have been around for many

years, primarily for enhancing signal from hollow cathode lamps. Many improvements

have been undertaken to develop better GD ion sources for mass spectrometry. One of

the most important developments for elemental analysis is the resurgence of time-of-

flight mass spectrometry (TOFMS). With rapid analysis time of the full elemental range,

the TOFMS system has been shown to be quite advantageous in analytical chemistry.19-20

Chapter 3 will describe the operation of the glow discharge with respect to analytical

performance and its coupling to the time-of-flight mass spectrometer.



The Scope of this Dissertation

Developmental and fundamental studies of the pulsed glow discharge have been

of primary interest in the Harrison group over the last twelve years. With the advent of

the microsecond pulsed source a new series of studies have become of interest.21'23 A

thorough comparison of the helium gas characteristics in the microsecond pulsed plasma,

including sputtering and ionization, has yet to be accomplished. The focus of this

dissertation is the characterization of the [ts-pulsed glow discharge using argon and

helium as plasma gases. The following chapters describe the differences observed from

the use of helium and argon as discharge gases for mass spectrometric purposes.

Microsecond pulsed discharge processes are considered for each gas. with special





8



attention given to atomization and ionization. One of the problems facing DC and RF

GDMS is the high background from the discharge gas and related species. One goal of

using different gases and the microsecond pulse mode is to alleviate these interference

problems reducing and removing interference in key parts of the mass spectrum, thereby

facilitating element detection.












CHAPTER 2
THE GLOW DISCHARGE

Gaseous Discharges

When a sufficiently high voltage is established between two electrodes in a

gaseous medium, a breakdown of the gas forms positively charged ions and free

electrons. The positive anode and the negative cathode create a voltage gradient in the

plasma, such that an electrical circuit has been created where the discharge carries the

current. This is what is called a gaseous discharge, and these discharges can be defined

by the current at which they operate. Figure 2-124 shows a schematic diagram of the

characteristics of gaseous discharges based on current and voltage. The three main

discharges shown in the diagram are: 1) the Townsend discharge, 2) the glow discharge,

and 3) the arc discharge.

The Townsend discharge operates in the sub-millitorr pressure range and is

characterized by the production of very few ions and electrons, or low current. This

discharge is low in luminosity due mainly to the low production of excited species which

give a gas discharge its "glow" characteristic. The Townsend discharge is not self-

sustaining, but relies on external X-rays and UV light to produce electrons and ions in the

plasma. The Townsend discharge has been used in analytical mass spectrometry in two

ways: first as a way of ionizing the fill gas instead of high energy electrons emitted from

a hot filament;25 and second as a chemical ionization source.26 The transition from the

Townsend discharge to the glow discharge is distinguished as a region in \which electrical






10









T titT
I I I l
I ,, ( I t
C C C tr
)I I
b .o4I I I ,-
> v) ( I -I I
C &4 I I i
c X, E0 03 3
o., 2. oo u0





9 -7 -5 -3 C-1
01 0-4 W W












10 10 10 10 10 10
0




Current (







Figure 2-1. The current-voltage characteristics of gaseous
discharges showing the plasma coverage on the sample surface.
Vb= breakdown voltage, Vn= normal operating voltage,
=operating voltage of the ar discharge [Ref. 24
10 10 10 10 10 10
Current (A)


Figure 2-1. The current-voltage characteristics of gaseous
discharges showing the plasma coverage on the sample surface.
Vb= breakdown voltage, V = normal operating voltage,
Vd=operating voltage of the arc discharge [Ref. 24]










current increases as the discharge voltage decreases. This increase in current and decrease

in voltage are mostly due to the increase in collision energy exchange at higher gas

pressure.

The glow discharge has two modes, a normal glow discharge and an abnormal

glow discharge. The normal glow discharge is characterized by its luminous glow.

Increases in current do not result in a current flux change because the plasma hasn't as

yet covered the surface of the cathode entirely (see Figure 2-1). No voltage increase,

therefore, is required to sustain the plasma, as long as current increases. Once the entire

surface is covered by the plasma due to increasing current, the current density undergoes

change and a transition to the abnormal discharge occurs. Increases in current result in

increased current density in the abnormal glow discharge region. Increases in voltage

result in increases in current, and help sustain the plasma. The abnormal glow discharge

region, where plasma atomic spectroscopy is performed, is the region of gaseous

discharges used in this body of work. The following sections will be devoted to

describing this region.

As the current is increased, thermal vaporization of the cathode occurs due to the

high current density and increased bombardment by the filler gas. The availability of

high analyte number densities affect the potential fields, and the current-voltage

characteristics become "normal", such that current increases, but the maintenance voltage

decreases. This defines the transition from the glow discharge to the DC arc.

The arc discharge is commonly operated at atmospheric pressure, has large

currents (10-1000 A), and a bright plasma. The high current heats the surface so much










that thermionic electron emission becomes a prominent current carrying mechanism. Gas

temperatures are typically as high as 2 x 104 K and charged particle densities about 1016

cm-3. The arc discharge has been used in atomic emission analysis of solid metals27 and

in vapor deposition of thin films.28 Recent papers have reviewed this old technique for

new applications.29



The Abnormal Glow Discharge

The glow discharge source is a simple device. At its simplest, it consists of an

anode, a cathode, a fill gas at low pressure, and a power supply (Figure 2-2). In our case

the cathode consists of the material to be sampled, the anode is the glow discharge

chamber housing, and the power supply is operated in either a continuous DC, RF, or

pulsed DC mode. When a negative voltage is applied to the cathode, electrons released

from the surface, or spontaneously emitted free electrons, undergo acceleration to\ ard

the relatively positive anode chamber housing. During this acceleration, electrons collide

with fill gas atoms producing positive gas ions. These ions experience the attraction of

the cathode at high negative potential and are accelerated toward it and eventually

bombard the cathode sample. If the gas ion has enough energy (greater than the binding

energy of the atom in the matrix) it will cause the release of several species from the

cathode, including neutral atoms and molecules, ionized atoms and molecules, photons.

and electrons. The electrons will continuously be accelerated toward the anode while the

voltage is on, thereby sustaining the flow of current, creating a "self sustained glow

discharge." The glow discharge can operate at a pressure from 0.01 50 Torr depending










Power Modes


Pulsed DC


DC /


Figure 2-2. A simple glow discharge device showing the various power modes


RF










on the discharge gas, and has average currents in the low milliampere range in DC mode

and less than 1 mA in pulsed mode (depending on duty cycle).



Regions of the glow discharge

A glow discharge consists of 7 distinct regions: the Aston dark space (ADS), the

cathode layer (also called cathode glow, (CG) ), the cathode dark space (CDS), the

negative glow (NG), the Faraday dark space (FDS), the positive column (PC), the anode

dark space (ANDS), and the anode glow (AG). The research in this dissertation requires

operation in the abnormal glow discharge range, as described above. Atomic mass

spectrometry and other analytical spectroscopies probe the collision rich negative glow

for analytical information about the sample. The close proximity of the cathode and the

anode result in the disappearance of the faraday dark space, the positive column, the

anode dark space, and the anode glow particularly in MS when cathode to anode

distances are less than 1 cm. When the cathode and anode are brought closer and closer

together, the cathode dark space remains unchanged in thickness. The positive column

shrinks until it and the Faraday dark space (FDS) are consumed. Figure 2-330 is a

representation of the glow discharge and its distinct regions and their electrical

characteristics. Figure 2-4 shows the abnormal discharge and the regions that are most

important for mass spectrometry and how the negative glow will be probed in this

dissertation.

Nearest the cathode surface is the Aston dark space, which is a largely non-

luminous region, due to the lack of collisions that cause excitation to occur. This is an







ADS = Aston dark space
CG = Cathode glow
CDS = Cathode dark space
NG = Negative glow
PC = Positive Column
ANDS = Anode dark space
AG = Anode glow


PC


HANDS



li


Figure 2-3. The regions of the normal glow discharge and
the corresponding electrical properties [Ref. 30]


C(
ADS
\V



II


SCDS


./^*


AG


[G






Sampler (Anode)

I Ion Skimmer


To Mass Spectrometer


: >


Figure 2-4. The abnormal glow discharge setup for mass
spectrometric detection


n
C11
0
CP
CD










area of negative space charge due to the secondary emission of electrons from surface

collisions of bombarding gas ions, as well as the presence of electrons released upon

neutralization of gas ions impinging on the target. Regardless, the electrons are

accelerated away, and current is primarily carried by positive ions in this region. A short

distance later the space charge becomes positive and remains that way through the

cathode dark space (see Figure 2-3).

The cathode glow or cathode layer separates the Aston dark space from the

cathode dark space. The low luminosity is due to the recombination of incident ions with

slow electrons. The cathode dark space is the next zone of little luminosity. The negative

glow is an electrically neutral zone; therefore, the voltage difference between the cathode

and the anode is reduced in the small region between the cathode and the negative glow,

this being the cathode dark space. This region is often called the cathode fall region, due

to the decrease of the voltage over this small area. The negative glow is the region with

the most analytically useful information. It is a collision-rich, field-free environment in

which excitation and ionization processes occur. It is this region in which glow discharge

ionization studies occur. The relative luminosity of these regions is determined by the

radiative relaxation processes occurring within that region. Hence, a "dark' region has

few excitation processes, while a "glow" region has many. The lack of luminosity in the

dark space is due to the small number of electron-atom collisions. The electrons have

been accelerated by the cathode potential and the efficiency with which discharge gas

ionization occurs decreases. Hereafter, all discussion will pertain only to the use of the

abnormal glow discharge, and mainly its negative glow region.










Glow Discharge Processes

The mechanisms that lead to atomization, excitation, and ionization in a glow

discharge occur within discrete zones of the discharge structure. Ionization of discharge

gas occurs in both the negative glow and in the cathode dark space. Atomization occurs

at the surface of the cathode, a result of the ionization and subsequent bombardment of

the discharge gas. Excitation and ionization both occur in the field-free negative glow as

a result of collisions with electrons, ions, or excited atoms with sufficient energy. Figure

2-5 is a diagram summarizing the processes occurring in the glow discharge and the

zones with which they occur.



Glow Discharge Sputtering

The approach or acceleration of an ion toward a cathode surface can lead to 1 of 5

possible interactions called collectively the collision cascade: 1) the ion may be reflected

or scattered away, 2) the impact of the ion may cause the release of an electron from the

surface (necessary for maintenance of the self-sustaining glow discharge), 3) the ion can

bury or implant itself permanently in the crystal lattice, 4) the ion may cause

rearrangement of the top few layers (called "radiation damage"), and 5) the impact of the

ion will eject a target atom. This last one is known as sputtering and is the most

important analytically. For any ion bombardment event, more than one of these

interactions occur.

The set of interactions above imply that an ion is the only species impinging on

the surface, when in fact either an ion or a neutral species can do so."' It is easiest to use









































Figure 2-5. Summary of GD processes










an ion in the model, because it can be accelerated across the dark space via the attraction

of the negative cathode, whereas explaining how a neutral species is attracted to the

surface is a good bit more difficult. It is believed that the Auger emission of an electron

from the surface neutralizes a majority of the incident ions just prior to striking the

surface. With all that can occur during the collision cascade, it is not surprising that

sputter ejection comprises only 1% of all the energy transferred at the surface of the

cathode.30

Sputtering occurs when an ionized (or neutralized) gas atom bombards the surface

of the cathode, embedding itself in the lattice of the solid, and then transferring enough

energy to create a "shock wave" cascading through the lattice. This shock wave has been

compared to the break in the opening of a game of billiards except in three dimensions"3

(See Figure 2-6). Three species representative of the sample can be sputtered upon

bombardment; a sputtered ion, a sputtered neutral atom, or a polyatomic cluster. In the

case of glow discharges, the negative voltage on the cathode will immediately attract the

sputtered ion for redeposition onto the cathode surface, leaving only a "cloud" of

sputtered neutral species to diffuse, without any field effect, across the cathode dark

space and into the negative glow. The sputtered ions produced upon bombardment of the

gas ions are called "secondary ions", and are the species used for analysis in the surface

technique called secondary ion mass spectrometry, or SIMS. The sputtered atoms are of

interest for spectroscopic purposes, as they are the primary species that leave the surface

during the sputtering process. The release of atoms from one medium (a solid surface)

and the subsequent dissolution into a gaseous medium gives glow discharges and



















------------- T












Cathode
Surface






Figure 2-6. Schematic of sputtering in a glow discharge based
on a billiards break










particularly GDMS a relative freedom from matrix effects, which often plague other

surface and solid analysis techniques. While not completely true, theoretically the

atomization and ionization processes are separate in the glow discharge plasma and

ensure the lack of chemical memory for the sputtered species once they leave the surface

of the cathode.2023

Sputtering yield is the primary measure of sputter efficiency in a glow discharge,

and is usually measured as the number of atoms sputtered per incident ion. Two

equations exist in order to determine this efficiency. The first, developed by Sigmund,32

takes advantage of the energy/mass transfer term (bold in the equation). In Equation 2-1,

the sputter yield (S) is given by:

S = (3/4)7T2(c)[4MiM2/(MI + M2)2](E/Uo), (Eq.2-1)

where a is a function of the relative masses and the angle of incidence of the incoming

ion, MI and M2 are the respective masses (g) of the ion and the sputtered atom, E is the

incident ion energy (eV), and Uo is the surface binding energy that must be overcome for

sputtering to occur (eV).

Experimentally, Boumans33 developed an expression for the sputter yield (S)

represented by Equation 2-2:

S = 9.6 x 104 (W/M i t), (Eq. 2-2)

where W is the weight loss (g), M is the relative atomic mass of the sputter species (g), it

is the ion current (A), and t is the sputtering time (s). The bombarding ion current i+ is

rclactd to the total discharge current i by Equation 2-3:

i+= i/(1 + Yi) (Eq. 2-3)










where yi is the number of secondary electrons released, on the a\ crage, by one ion. The

following subsections will detail the parameters used in these two equations.



1. A. h .s.s of Sputtering Ion

One might assume that as you increase the mass of the sputtering ion, that the

sputter yield would increase because more momentum can be delivered to the surface.

The mass transfer term predicts that sputtering is dependent upon both the sputtering ion

and the target atom, and that sputtering efficiency will reach a maximum when the value

of Mi/M2 is unity. This would make Ar+ an ideal sputter ion for the first row transition

metals and Kr+ for the second row. In Sigmund's theory (Equation 2-1), the alpha (a)

factor offsets the mass transfer term such that the Ar ion is a better overall sputtering

agent than other inert gases such as krypton and xenon even though these two heavier

gases would deliver more kinetic energy and momentum to the surface under similar

conditions. The mass of the sputtering ion is critical to this work. The lighter mass of

helium will be an important factor as a sputtering agent in GDMS.



2. Angle of incident Ion Bomnhardichnt

The collision cascade describes a variety of sputtering ion-surface interactions

that can occur, and it makes sense that sputtering would be dependent upon the angle at

which the gas ion bombarded the surface, thus maximizing momentum and energy

transfer to the top few layers of the surface. Figure 2-734 shows the effect of angle of

incidence in ion beam sputtering on sputter yield. It should be noted that ion beam












>25 E=I OkV




Eo 105/ e ,V





S(dree)





sputter yield at normal incidence[Ref. 34]
0 30 60 90
s a (degrees)


Figure 2-7. The effect of ion angle of incidence (a) on sputter
yield (atoms/ion). Y(a) = sputter yield at angle a, Y(O) is
sputter yield at normal incidence[Ref. 34]










sputtering is much different than glow discharge sputtering and is used here solely as a

means to explain sputtering in the glow discharge system. Figure 2-8 sho m two possible

paths for an incident ion in a glow discharge upon application of the discharge voltage.

One argument for the incident nature of ions is illustrated in the picture of a glow

discharge sputtered surface (Figure 2-935). Microscopic cone-like spires have formed on

the surface of the cathode. The most likely way formation of such structures could occur

is for the impinging ion (or atom) to have a bombardment angle 90 degrees to that of the

surface proper, such that "holes" are dug into the surface. This means that regardless of

the trajectory of an incident ion, in the region between the cathode surface and the

negative glow, the field attraction interacts strongly with the ions, that any effect of its

initial trajectory is eliminated. So, while ion bombardment studies may indicate relative

sputter patterns in a vacuum, they are not necessarily indicative of what occurs in a low

pressure discharge.



3. Incident Ion Energy

The energy of the bombarding ion also plays an important role in sputter yield.

Sigmund's theory predicts that the sputter yield of a given target material is proportional

to the bombarding ion energy. Carter and Colligan have shown that proportionality exists

over only a small ion energy range.36 Figures 2-10a37 shows the effect of low incident

ion energy on sputter yield for a Cu target with noble gas ions. It has been predicted that

glow discharge ions bombard the surface with energies that lie in this linear region of the

curve. The sputter yield line in the figure shows a range of proportionality from about

















03

>


X


Y


Figure 2-8. Possible paths for incident ion (A). X = No voltage
effect on trajectory. Y = Voltage controls trajectory













* d1.,4



1- A


f 4 j


Figure 2-9. A copper target sputtered by a 1 torr argon discharge
showing microscopic cone formation [Ref. 35]


ft. 41E





















20
E 20-
0
Lineor region
8 A

C"

I-0- A

A


'Knee' of curve


0 500 1000

Id on energy (eV)
Threshold
energy
(ET)


Figure 2-10. a) The effect of low incident ion energy on sputter yield
for a copper target and argon ions. Triangle and circles are various data
points[Ref. 36]










100 eV to 750 eV. The range of this proportionality will likely change basIed on the size

of the incident ion. Figure 2-1Ob36 shows a higher energy range for sputter yield \ eruM

incident ion energy. As ion energies get much higher, ion implantation occurs and a

maximum of energy transferred to the top few layers of the lattice is reached, and the

number of sputtered atoms reaches a maximum.



4. The Target Material

The target material is the one parameter in glow discharge sputtering that is the

most variable, as there are many matrices that can be analyzed. Therefore, multi-element

analysis can be affected by the matrix. The Sigmund theory suggests that sputter yield is

dependent on the ion to atom mass ratio. Wehner3 believed that other processes are

involved aside from just the mass relationship. Having produced charts of the sputter

relationships with a 400 eV Ar+ gun, he found a dependence of sputter yield on d-shell

filling. The explanation says that as the d-shells are filled with electrons, the atomic radii

decrease, thereby increasing the atomic density in the matrix. The increased atomic

density prevents the bombarding ion from implanting or penetrating too deeply into the

matrix. The crystallographic orientation of the matrix most likely does not change with

the increase in density as the orientation of the atoms does not change, only their size.

The maximum amount of energy is then transferred to the top few layers of the matri\.

increasing sputtering. The difference in sputter yield between iron and copper is argued in

this manner. The mass of copper is greater than that of iron, and according to the mass

transfer term, the sputter yield of iron would be larger than copper if argon \\ re used as














Copper


Almei & 3ruce
(1961)


- Xe


. 0


"- N e


_/


10 20 30 40 50 50 70 k.V

Incident Ion Energy (keV)



Figure 2-10. b) The effect of high incident ion energy on
sputter yield for various gases [Ref. 37]


o / -- -- N
i i i










the bombarding gas. The sputter yield of iron is less because copper has more d-shell

electrons that, ultimately, increase the density of the matrix. Wehner's plots for argon and

helium are shown in Figure 2-11 a and b. The different sputter yields amongst elements

may cause a problem when analyzing alloys. Preferential sputtering3 will lead to an

enrichment of the lower sputter yield material at the surface. Due to the inherent

sputtering characteristics of the glow discharge, an eventual steady state will be

established and the gas phase population will be representative of the bulk sample.



Glow Discharge Excitation/Ionization

For analytical glow discharges, the primary concern is the excitation and

ionization of the sputtered species, and one would hope that these were the only

processes that occurred in the plasma. However, ionization of the plasma gas and its

impurities, and interference from sample impurities, can make analysis difficult. It is

important to understand the processes that maximize analyte excitation and ionization

and minimize discharge gas excitation and ionization. First a brief discussion of the

nature of collisions in the glow discharge is given below.



Collisions in Gases

Collisions in a glow discharge are the most essential processes that occur and are

relied upon to ionize sputtered species in the negative glow region of the plasma. Belo\\

are definitions taken from Howatson:24 Elastic Colli.-ion. are simple, mechanical. gas-

kinetic collisions in which the energy is always kinetic; Inelastic colli. ion. are those in












2-6
2.4- Argon
2-2-
2-0-
1-8
1-6
1-4

1-0-
08^ *AI

0 e4 SI
0-2- C
0 10


---




Cu 'Pd

Cr NI
Co
Fe &Ge Ru
Mo
V Zr b
Ti Nb


I I ____ ,


20 30 40
Atomic


50 60 70
number


80 90


\0.24- Heeium
-22- *Be
-20-

0-16
0-14- A
0-12-
0-10-
0-08-
0-06- .C
0-04-
0-02-

0 10


t



T


Au


20 30 40 50 60 70 80 90
Atomic number


Figure 2-11. Wehner sputter yield plots for a) argon
and b) helium at 400 eV incident ion energy [Ref.39]


I










which some of the energy of the collision is transferred into internal energy of the particle

struck, thus producing a reduction of kinetic energy of the system. Elastic collisions can

be compared to the collision of two billiard balls, much like that shown in Figure 2-6 for

sputtering. Figure 2-1230 shows schematically the two collision possibilities of an

electron and an atom. The electrons in orbit around an atom give an approximate uniform

field about the nucleus, and when an electron approaches, it can be repelled or deflected

in a way that is similar to the deflections of two billiard balls or two ideal elastic spheres.

The fundamentals for perfectly elastic collisions comes from the Maxwell-Boltzmann

kinetic theory, in which each atom has a definite radius and cross-sectional area such that

the probability of the two atoms colliding can be calculated.40 However, in such a

collision, changes in internal energy are not considered. For example, if an electron with

energy less than the ionization potential of helium were to strike a helium atom at zero

energy, then the helium atom would have to sustain the shock of the electron collision

(via kinetic energy) because there is no mechanism to transfer that energy (via potential

energy) under the theory of elastic collisions.

Now consider an electron with exactly the amount of energy needed to remove an

electron (IE of the atom). In order for the atom to be converted into a positive ion, the

electron must have enough energy to not only remote e the electron, but also some excess

energy to remove the electron and to compensate for recoil from the atom. Therefore,

electrons with energies that don't exceed that of the ionization potential of the atom they

are striking will result in an elastic collision.














a.








b.







Figure 2-12. a) Elastic collision of an electron and an atom.
b) Electron ionization [Ref. 30]










In the negative glow region of the analytical glow discharge, collisions that excite

and ionize are inelastic. Inelastic collisions fall into two categories: collisions of the first

kind and collisions of the second kind. Collisions in the negative glow plasma with

electrons of various kinetic energies are collisions of the first kind. Collisions of the

second kind occur when atoms and massive particles collide, resulting in potential energy

transfer. Table 2-140 lists the most likely excitation and ionization phenomena in the glow

discharge. A number of factors determine the relative role of these processes including

discharge conditions and source geometry.



Sputtered Species Excitation

Atoms that have diffused across the dark space into the negative glow, face a

series of inelastic collisions by electrons, metastable atoms, and other ions (gas and

sputtered). The glow discharge plasma is an electrically neutral plasma with a high

population of electrons. Electrons in the negative glow are very low in energy (Figure 2-

1341) as the Electron Energy Distribution Functions show under various conditions.

Hence, excitation mechanisms that involve energy transfer to levels at, or less than 3-4

eV (above ground state) will be preferentially populated over those that exist at higher

energies.41 Transitions from non-ground state energy levels to higher levels do occur, but,

without the populations as most of those that are ground state resonance transitions.

These ground state transitions are mostly responsible for the atomic transitions that occur

in the UV-VIS (k = 200-700 nm) regions of the electromagnetic spectrum, and are

responsible for the primary emission from the negative glow.















I. Primary excitation/ionization processes
A. Electron impact
Me + e- (fast) M* + e~ (slow)/M' + 2e-
B. Penning collisions
Me + Ar* -* M* + Ar/M + Ar+ e-
II. Secondary processes
A. Charge transfer
1. Nonsymmetric
Ar+ + M0 -, M' (M+*) + Are
2. Symmetric (resonance)
X+ (fast) + XO(slow) -, Xo(fast) + X+ (slow)
3. Dissociative
Ar+ + MX -* M+ + X+Ar
B. Associative ionization
Ar*, + MO -* ArM+ + e
C. Photon-induced excitation/ionization
Me + hv -, M*/M+ + e-
D. Cumulative ionization
Me + e- -* M* + e- -+ M+ + 2e-


Table 2-1. Glow discharge excitation and ionization phenomena




















AA ** a
AA A
0 o oo a 5 mA
= o o 8 mA
030 o*0 & 12 mA
30 0 &
u. so o o 15 mA
um 20 % *% A 18 mA
*1 00 0 0

20 6.0
10 0 o ca'--


0.0 1.0 2.0 3.0 4.0
Electron energy, eV


60- a---
o 00oo b
s50- 0
0 *e o
S40 o* 10 mA
" vi0eo- o 12 mA
30 o % o 15mA
So 18 mA
U 20 MC oa o a 21 mA

10 0 o

0-
0.0 1.0 2.0 3.0 4.0
Electron energy, eV




Figure 2-13. Electron energy distribution function
(EEDF) for the negative glow region [Ref. 41]










Sputtered atoms are not the only species that are excited in the negative glow.

Excitation of ionized sputtered atoms can also occur. As shown in the list of processes,

the collision of an electron and a gas atom can produce a metastable gas atom that has

enough energy to ionize and or excite sputtered atoms. These ionized sputtered atoms

can undergo similar collision with electrons and metastables that produce excited state

ions. The collision must overcome the energy required to raise an electron above the

ground state ion energy (IE) to the next available energy level.



Sputtered Species Ionization

As previously mentioned, the quantification of an total ion number density via

mass spectrometry is much easier than by atomic emission because relatively few

isotopes exist for each element; whereas multiple transitions represent the excitation of

the ion population. Therefore, fewer measurements are needed in mass spectrometry to

analyze trace elements. The energy of such ions is not entirely important, except when

sample focusing for mass spectrometry is considered. For quantitative analysis in MS, a

maximization of analyte ionization, while maintaining matrix representation, increases

the informing power of the technique. As Table 2-1 illustrates, several collision

imcchaniiisms exist for analyte ionization. The most likely sputtered species ionization

mechanisms are by electron ionization and Penning-type collisions.

For electron ionization, the electrons must have enough energy to ionize the

analyte atom. However, the electron energy distribution function (Figure 2-1341), shows

that very few electrons have enough energy to ionize most metals with I.P.s in the 5-10










eV range. Electron and ion densities have been measured in a RF glow di'chlirge plasma,

based on electrons in the 0-5 eV range.30 The density measured, 10'I cm-3, is 3-4 orders

of magnitude higher that electron densities measured in the 5-10 eV range.30 The lower

electron population in the 5-10 eV range, would produce very low relative analyte ion

populations compared to what have been typically measured in the negative glow of

argon plasmas.30 Vieth and Huneke42 have studied the production of an electron via the

Lovett rate model in which a 7.1-7.6 eV electron is produced that could ionize many

metals. This model, based on electron impact and three-body collisions, proposes a

double Penning collision as shown below in Equation 2-4.

Arm* + Arm* -- Ar+ + Aro + e- (7.1-7.6 eV) (Eq. 2-4)

No experimental evidence has been provided to assume that this mechanism occurs for

analyte ionization as the probability of this process is very low; even though Penning-

type collision (inelastic, with transfer of potential energy) are assumed in many

experiments as the primary analyte ionization mechanism in the negative glow. Table 2-2

lists the low lying metastable energies of various noble gases. The energies are well

above the I.P.s of most of the solid periodic table, and for noble gases from argon to

krypton, the energy is below that of most gas impurity I.P.s. Metastable gas atoms are

formed when inelastic collisions raise the energy to this long lived excited state (11.5 and

11.7 eV for argon).

Quantitative estimations of Penning ionization in a glow discharge have been

performed and have given numbers anywhere from 40-95% depending on discharge

conditions.43 A variety of reasons exist to support Penning ionization as the primary









Gas Spec IE (eV)
02(dimer) 12.0697
H20 12.621
OH 13.017
O 13.618
H 13.598
Ni 14.53
Ar2(dimer) 14.5
N2(dimer) 15.581
Ar 15.759
He2(dimer) 22.22
He 24.587


Ar met 11.5
11.7
He met 19.8
20.6


Table 2-2. Gas species ionization energies and helium and
argon metastable atom energies










mechanism of sputtered atoms. Mass spectra from argon discharges have analyte ion

signals higher than those of the discharge gas species even though the gas species are in

much higher quantities than the sputtered species.44 This reasons is applicable to an argon

DC discharges and not a helium DC discharge because the energies of the helium

metastables are higher. It's been postulated that preferential ionization of the sputtered

species is the reason for this observation. The metastable energy levels of argon are high

enough to ionize a majority of the elements in the periodic table, whereas the metastable

energies of argon lie below the ionization potential of any impurities from air leaks. The

discrimination in energy is one of the main reasons why argon has become the primary

discharge gas in most analytical discharge mass spectrometric instruments. Chapters 3

and 4 will discuss the role of metastable energies on analyte and gas species ionization

further relative to the use of helium as discharge gas.

The best evidence for the Penning ionization mechanism arises from laser

depopulation studies in which lower ion signals were observed for species with ionization

energies less than the metastable level of argon at 11.72 eV, while those with higher

energy were unaffected.45 The same result was observed for a neon discharge except that

the species with ionization energy above 16.71 eV were unaffected.46-49 Experimental

methods of determining the role of Penning ionization, including varying the discharge

conditions,46-49 and quenching the metastable states with methane46-49 have supported the

claim that Penning ionization plays a prominent role in sputtered species ionization. In

Chapter 3, a discussion of the pulsed glow discharge will show even more dominating










ex idence for metastable ionization based on the nature of the events that occur in the

pulsed discharge.



Other secondary excitation/ionization processes

Sy)Immeric charge e.\change is thought to be limited to the region closest to the

cathode such that argon ions that are normally accelerated across the dark space collide

with neutral atoms close to the surface. The incident argon ion continues on its path to the

cathode where it contributes to sputtering. Asymmetric charge exchange also contributes

to ionization in the negative glow. These mechanisms are summarized as equations in

Table 2-1 and shown in Figure 2-14.30 The contribution of each charge exchange process

varies depending on the ion source. This is particularly true for hollow cathode and

Grimm-type glow discharges,50 where asymmetric charge exchange is thought to play a

greater role than symmetric charge exchange. As described before, the ionization

provided by Penning-type collisions is preferred, because of the strong uniformity of the

ionization among elements (RSFs), and the selectivity against background gas species.

Associative inization is the primary mechanism for the formation of polyatomic ions that

contain a di,,charge gas and sputtered species. This process usually takes place within the

negative glow and its contribution in the mass spectrum is in some proportion to the

population of the sputtered species. For a typical diode discharge, the relative role of each

of the ioni/ation mechanisms is uncertain, except that Penning ionization appears to be

the primary ionization mechanism for the sputtered species. As discharge parameters






















Electrode


Ion strikes neutral,
charge exchanges 0, Ion enters dark space
Sfand accelerates

+O +

tt



Forms fast
neutral and slow 0.:*:A: ~
Ion G

Sheath Glow
Region


Figure 2-14. Charge exchange processes [Ref.30]










change. so do the magnitude of these mechanisms; whereas Penning-type collisional

ionization is considered to be the favored process regardless of the discharge conditions.



Types of Glow Discharge Devices

Pin-type/Disc or coaxial (Figure 2-15a): This device is the most commonly used for trace

analysis of metal and environmental samples (usually non-conducting soils) \where a

conducting binder is required. In the case of the pin type, the ease of using the chamber

as the anode makes it the most versatile of these devices. The pins are 1-2 mm in

diameter, with about 5 mm in length exposed to the plasma. The high density of atoms

and ions above the tip of the pin or disc sample have made this device common amongst

commercial glow discharge devices; one example is the VG 9000 Double Focusing Mass

Spectrometer. The disc samples are mounted using a holding device that exposes only

the flat front surface to the plasma, via shielding that surrounds the diameter of the top

surface (typically 4 mm). The work described here uses these small flat discs for metal

analysis.

Planar diode (Figure 2-15b): This is the simplest of the glow discharge devices yet, it is

not used very much. Coburn used an RF power supply with a planar diode for

imcasurement, via mass spectrometry, of ion populations in a sputter deposition

chamber. 8 This is also the basis for the de\ elopment and production of the Grimm-type

glow discharge source.

Grimm Discharge Lamp (Figure 2-15c): The Grimm-type GD device was first introduced

by Grimm in 196817 and has gained in popularity mostly because of the ease of sample










interchange and the ability to analyze thin films.51 The Grimm discharge is often called

an obstructed discharge because the anode is positioned within the dark space distance of

the cathode. This limits any sputtering to the area within the anode shape. It has become

very popular for bulk analysis. Similar sources that resemble the Grimm source have

been constructed for use in atomic emission and mass spectrometries.30

Hollow Cathode (Figure 2-15d): The hollow cathode lamp is probably the best known

and one of the most widely used GD devices. It is commonly used in atomic absorption,

emission, and fluorescence studies. This device is characterized by low voltages and

higher currents compared with other glow discharge devices, due to the constricted nature

of the plasma within the cathode and the static nature of the discharge gas. Called the

"hollow cathode effect," the plasma constriction produces an increase in the radiation

emitted from the negative glow, making this source attractive for optical spectroscopy.52

Hollow Cathode Plume (Figure 2-15e53): The hollow cathode plume was developed by

the Harrison group, and was primarily used for atomic emission studies. The plasma was

formed by the constriction of a hollow cathode discharge to a 1.5 mm orifice in the base

of a normal hollow cathode. Sample cathodes were machined to fit in the base of graphite

cylinders. At a certain pressure, the plasma plume would form through the top end of

the cylindrical hole. Typical argon pressure was about 1 Torr, and currents betw cen 50

and 200 mA at 1000 V were used.







MHV
connector







Silicone
fiberglass
tubing


MACOR
probe
tip





Sample
pin
High-vacuum
feedthru


0-=


CATHODE


Sample


Q,2mm thick
teflon sheath
y to pump 1


DISCHARGE CHAMBER


d.


+
D.--


Figure 2-15. GD devices, a) coaxial cathode b) planar diode c) Grimm
discharge lamp d) hollow cathode lamp


a.


b.


ANODE












CTr
CL
Q-


Figure 2-15 (cont). e) The hollow cathode plume


e.











CHAPTER 3
EXPERIMENTAL CONSIDERATIONS

Powering the Glow Discharge

Direct Current (DC)

DC is the most familiar and most widely used method of powering glow

discharge plasmas because of the availability of inexpensive, stable power supplies that

are very easy to use. The direct current glow discharge plasma is a self-cleaning plasma

in which little sample preparation is required before discharge initiation. Elemental

analysis in the DC mode usually consists of a preburn period in which contaminants from

the sample surface are sputtered off, such as oil from the users hands, oxides from the air,

and any non-uniformity of the sample surface. The DC plasma can be compared to an

electrochemical cell in which the cathode material is eroded or dissolved into a solution.

In the case of an argon glow discharge, the atoms dissolve into what is essentially an

argon gas-phase solution.

The roles of atoms, ions, and metastables have been extensively modeled by

Bogaerls and Gijbels,54-56 and have shown a good correlation with experimental data.

Their results show that mctastable ionization and electron impact ionization are the

primary mechanisms for analyte ionization, of which electron ionization can contribute

up to 40%. In their models, fast argon atoms are also considered to be a sputtering agent,

not just argon ions. The models define the role of each species present in the plasma and

\which ones are most important for the analyst. The most important feature of the glow










discharge is its ability to provide a stable concentration of atoms and ions for analysis,

the advantage being that in most GDMS instruments, the analyzer always samples the

same region of the plasma, leading to higher precision and higher accuracy. For GDMS

chemists, the DC glow discharge is an excellent ion source because of its 100% duty

cycle, which, with the appropriate choice of mass spectrometer, is a great advantage in

elemental analysis.

DC glow discharges have two disadvantages; one is low power compared to

techniques such as inductively coupled plasma, electrothermal vaporization, and laser

induced plasmas; and the other is the need for an electrically conduction material. The

lack of a conductivity requirement for these other techniques allows universal sampling

of any sample or matrix, giving these techniques a great advantage over the DC glow

discharge which for now is limited to conducting metal matrices. In order to sustain a

current in a DC discharge, the sample being analyzed must at least have semi-conducting

properties, and even then accurate analysis is not guaranteed. The conductivity

requirement has not prevented the analysis of non-conducting samples in the DC mode.

The use of a conducting matrix binder has been very successfully used in the analysis of

non-conducting powders."7 The trouble faced in these analyses is heterogeneous

distribution of analyte in the matrix. A secondary cathode has also been emplo) ed to

sample non-conductive solids. A "dummy" hollow ring cathode is positioned above a

non-conductive surface allowing accelerated atoms and ions to strike the non-conductive

target. Sputtering of the non-conductive surface results in the ejection of material into the










plasma formed in the middle and above the surface of the conducting ring cathode,

effecting excitation and ionization of the non-conducting material.



Radiofrequency (RF)

With radiofrequency powered glow discharges, most concerns about sample non-

conductivity are minimized. The RF glow discharge was created by Wehner, who

proposed the use of a rapidly oscillating voltage that would cause a DC bias at the surface

of the cathode thereby creating a plasma much like that in the DC mode. The ability of

the RF glow discharge to analyze both conductive and non-conductive samples makes it

the most versatile of the sampling modes of the glow discharge. The RF discharge is less

affected by surface contamination and reaches a stabilization point faster than the direct

current mode of operation.8

The main drawbacks of RF glow discharges include expense and power

efficiency. RF power supplies are significantly more expensive than DC power supplies

and require a matching network so that efficient coupling of the power to the sample

occurs. Often, reflected power can dissipate the effectiveness of RF po\cer and the

analyst usually tries to minimize or eliminate it. Additionally, if shielding of the source

and sample is not complete, the source can act as a radiotransmitler and disrupt the

capabilities of other detecting devices such as ion counters and microchannel plates on

mass spectrometer devices. Lo\ er sputter rates have been reported for the RF glow

discharges compared to the DC glow discharge, although increases in ionization have










been observed compared with the DC discharge.59,60 There are several comparisons of

the fundamental and applied aspects of these two powering methodes.58-60



The Microsecond Pulsed Mode

Thus far the use of a glow discharge device has been described in a continuous

fashion where voltage and current are maintained throughout the analysis. While there

are advantages to such analysis, pulsing the glow discharge can create additional

advantages for elemental analysis. A pulsed cycle consists of a short on time, eg. 10 Its,

followed by a long off time, before another 10 uts pulse is initiated. In a one second time

period, the frequency of this cycle is 100 -1000 Hz for the work in this dissertation. It is

difficult to maintain a plasma at frequencies less than 20 Hz and often instability occurs

at frequencies higher than 1000 Hz. Sampling the sputtered species in mass spectrometry

occurs anywhere from 80-300 uts after the pulse period. This delay will be referred to

here as the "deflection time", and corresponds to the time between the glow discharge

pulse and time that ions are deflected into the time-of-flight mass spectrometer. There are

several analytical advantages to using the glow discharge in this mode and they are

covered in the next few sections.



Pulsed Glow Discharge Mass Spectrometry

Pulsed GDs have been studied by the Harrison group for about 30 years, starting

mainly with hollow cathode lamps. The goal of the experiments was to obtain higher

emission signals by decreasing the on time and increasing the peak pot~er. At that time,










pulsed periods \\ere milliseconds long. In the last 6-7 years, the pulsed discharge has

moved from the millisecond to the microsecond time regime. A variety of techniques

have been used to probe the microsecond pulsed plasma, including atomic emission (AE),

atomic fluorescence (AF), and mass spectrometry (MS).61'62 The immaturity of the us-

pulsed plasma in analytical chemistry necessitates some explanation of the fundamental

and analytical aspects that distinguish this power mode from the more conventional

continuous modes of operation.



DC vs. las-pulsed Experimental Considerations

The microsecond pulse is characterized by a peak voltage of about 2 kV, and a

peak current of about 200 mA (avg. 0.4 mA at 200 Hz). The DC discharge typically

operates at about 1 kV and 4 mA average current. The high peak current in the uts-

pulsed mode causes an instantaneous heating of the surface, followed by a long relaxation

period. In the DC mode, there is no cooling or off time, and sample over-heating can be

a problem, causing the sample to melt despite the low average current. The average

current in the pulsed discharge mode, approximately 0.4 mA, is less than that in the DC

mode, even though the peak current is 50 times higher. The lower a\ eragc current

reduces overheating in the pulsed mode and extends the operation time. Sample

overheating in the DC mode often results in the need to cut short analysis, which is why

many GDMS instruments have cryogenic cooling s) stems installed. The long-term

stability of the pulsed discharge is advantageous for bulk and trace anal\ si.s. With a

longer operation time, diagnostic studies, like those in this dissertation, are more easily










performed. In addition, the time consuming cooling or replacing of an overheated sample

is not needed in the pulsed mode.

Different instruments are required to operate in the two modes. The power

supplies are different, in that an internally pulsing unit is required to create the short

transient signal desired in the pulsed mode. Leading edge rise times must be fast, on the

order of 30 ns, in order for the pulsing unit to be effective. The maturity of the pulsed

discharge is dependent on decreasing these rise times so that shorter pulses can be used

with much higher power. There are mass spectrometers that are better suited for the DC

mode than for the pulsed mode. An ideal MS system for the DC mode might be a Fourier

Transform Ion Cyclotron Resonance Mass Spectrometer (FT-ICR-MS) where inter-

element and isobaric interference can be eliminated with extremely high resolving

power.63 In the pulsed mode, taking advantage of every pulse for mass analysis is

advantageous. A mass spectrometer that accepts short transient signals and analyzes all

masses simultaneously would be the best fit. The time-of-flight mass spectrometer

(TOFMS) is the right choice. The system offers full elemental coverage and is fast with a

linear dynamic range of about 5 orders of magnitude. A later section will discuss the

fundamentals of the TOFMS system and cover the advantages of using it as a detection

device for the microsecond pulsed mode.



Sputtering and Ionization in the Microsecond Pulsed GD

The sputtering mechanism doesn't change between the continuous and pulsed

modes. Argon ions and fast argon atoms are still the primary sputtering agents.










However, the time regime and the energy with which sputtering occurs do change. While

the DC mode results in continuous injection of atoms into the plasma, the tps-pulsed

mode supplies the plasma with atoms in a transient fashion, allowing for certain temporal

advantages. The sputtering ions and atoms will have a greater energy when bombarding

the surface in the pulsed mode as compared to the DC mode, owing to the higher

instantaneous power for pulsed operation. This leads to higher instantaneous sputter rates

in the pulsed mode as compared to the DC mode. In the ps-pulsed mode, analyte

ionization most likely occurs by Penning ionization. Argon metastable atoms are the

dominant ionizing agent existing after the discharge is extinguished. Electrons rapidly

dissipate after the 10 pIs pulse period, and by 100 p.s (analyte sampling time) electrons

have most likely thermalized. This means that electron impact makes little contribution

to analyte ionization unlike in the DC mode.

In the DC mode, there is always a constant ion flux toward the cathode surface,

and there is always a constant electron flux away from the cathode surface, which leads

to a dark space larger than in the pulsed mode and forces the optimal sampling distance to

be longer in the DC mode than in the pulsed mode. This may contribute to the lower ion

signal observed in the DC versus pIs-pulsed modes for mass spectrometry since the

atomic ion concentration is greater at 3 mm abo\ e the sample surface as opposed to 7 mm

due to the concentration gradient of atoms away from the cathode surface.










Analytical Advantages of the Direct Current GD

DC glow discharge analysis is a proven technique. The constant source of atoms

and ions is reliable and increases the precision and accuracy for trace elemental analysis.

It is a good atom generator despite its relatively low power compared with other

atomization techniques such as the ICP. The DC discharge is inherently self cleaning,

therefore, requiring little sample preparation to the surface. This is one of the advantages

that can be very useful for samples that readily oxidize. With the constantly applied

voltage and continuous bombardment of the sample surface, oxides are less likely to

form, allowing for greater sputtering of the surface. Figures 3-la and 3-lb show a

comparison of a DC and a uts-pulsed glow discharge mass spectra for a tantalum cathode.

The [ps-pulsed spectrum shows a series of oxides beyond the elemental species while the

DC spectrum has only an elemental peak. The elemental peak is greater in the DC mode

than in the pulsed mode showing that oxides are masking the sputtering of the tantalum

elements in the pulsed mode, due to their formation during the relaxation period. If TaO

formed in the area above the surface, then its contribution would be observed in the DC

mass spectrum. However, it is not observed, and it must be assumed that TaO is

sputtered off the surface in the p[s-pulsed mode, likely being formed immediately after

perturbation of the surface by a voltage pulse.

DC glow discharge operation is simple, requiring few fast electronics to gate

signals for collection. For analytical use, it has very good sensitivities for all elements

and allows detection by multiple analytical techniques (Figure 3-261'62). DC glow






0.35 +


0.30


0.25


0.20

0 -
'r 0.15


= 0.10


0.05


0.00
I I I i I
0 100 200

m/z

Figure 3-1. a) DC mass spectrum of tantalum at I kV and 1 torr Ar





















0.05 -




0.00 -


100


200


m/z


Figure 3-1. b) gs-pulsed mass spectrum of tantalum
at 2 kV and 1 torr Ar


0.20-




0.15-


0.10-


>-
(I,
a,
C
CD
c-,
.


Lj--,









Glow Discharge Spectrometries




MASS SPECTROMETRY


+
+

+


hv


ATOMIC
FLUORESCENC


ATOMIC
ABSORPTION


hvo
E



Po P

-I I-,


RIMS = Resonance Ionization Mass
Spectrometry
OGE = Optogalvanic Energy


t
OGE
_t,


Figure 3-2. Glow discharge analytical techniques


ATOMIC
EMISSION


LASER
RIMS OGE


r J










discharges have become very useful for thin layer analysis providing excellent depth

resolution with multi-element capabilities.51



Analytical Advantages of the Microsecond Pulsed Mode

Atomic Emission Signal Enhancement

There is a two-fold advantage in performing AES in a microsecond pulsed GD

plasma: first, there are instantaneously more atoms produced at the surface than there are

in a DC plasma. Secondly, the microsecond pulsed GD produces electrons having more

energy than do DC GD electrons, resulting in greater excitation of atomic states, allmo ing

for better detection limits and increasing informing power. Furthermore, the wide range

of populated states can be used to determine how energetic microsecond pulsed GD

electrons are compared with DC and RF GD discharges. More energetic electrons also

lead to a greater population of metastable argon atoms. Metastable atoms are capable of

ionizing gaseous metal atoms, and have excitation capabilities. These factors result in

emission signals up to 20X greater in the pulsed versus DC modes for certain elements.62

It has also been shown that the signal enhancement experienced by the sputtered atoms is

much greater than that experienced by the fill gas. This leads to better signal-to-noise

and signal-to-background measurements.



Sputtered Species Ion Signal Enhancement

Figure 3-3 shows the comparison of a DC signal at 1 kV and a pis-pulsed signal at

2 kV. The more energetic electrons produced in the pulsed mode lead to an increased


















Cu'


I I I I I a
20 40 60 80 100 120

Mass/Charge





Figure 3-3. Glow discharge mass spectra for copper in
argon; a) ls-pulsed mode, b) DC mode










production of Ar ions available for sputtering. Therefore, a greater instantaneous

population of atoms occur during the pulsed period, or a greater sputtered atom density at

the surface than produced in the DC mode. The more energetic electrons also serve to

create a greater metastable argon atom population, increasing the ionization of sputtered

species. All these factors lead to the 3-4 fold signal enhancement seen in Figure 3-3 for

the copper ion. The signals for Sn at 400 ppm in a NIST brass sample are shown in

Figure 3-4. Figures 3-3 and 3-4 show that about the same signal enhancement is obtained

in the pulsed mode compared the DC mode for both bulk and trace signals. One concern

was that the transient nature of the ion source would cause high noise. However, a

reduction in noise compared to the DC mode is observed because of the long relaxation

of the discharge before sampling occurs.



Temporal Considerations in GDMS

One of the important temporal aspects of the pis-pulsed mode is the ability to

separate the various ionization events that occur during the pulse cycle, beginning with

the ionization of the discharge gas species by electron ionization. In the 10 Ps pulse

period, electron impact ionizes the discharge gas and its impurities (Figure 3-5). Once

the discharge has been turned off, mass spectra show that no significant ionization occurs

for about 100 ats. There is no significance to 100 p.s time period, however, during this

period, electrons sustaining the plasma current completely dissipate. This means that little

or no electron impact ionization of gas species or sputtered species occur. After the 100

ps relaxation period, other ionization events such as charge exchange occur just prior to
















700-






0-


100-


0-


I
116


I
118


I
120


vZ


Figure 3-4. Mass spectra of tin at 460 ppm for a) p[s-pulse b) DC


I
12D)








600 @







- 300







0 J ----

0 50

m/z


Figure 3-5. Electron ionization spectrum
of argon at 1 torr and 10 ps delay time










the maximum signal for the sputtered species (Figure 3-6). Associative ionization is

mostly responsible for the observed polyatomic and molecular species present in the

plasma and is the result of metastable association with sputtered species. Penning

ionization dominates ionization processes and is considered the primary mechanism for

sputtered species ionization at deflection times greater than 100 p.s (Figure 3-7). Figure

3-7 shows an ion counting mass spectrum where copper ions have saturated the detector,

and showing the small presence of gas species as compared to Figures 3-5 and 3-6.

Associative ionization will last as long as there are metastable atoms present in the

discharge volume. The separation of electron and Penning ionization events is due in

part to the temporal attributes of pulsing the glow discharge. When the discharge is not

pulsed there are contributions in the mass spectrum from all of these ionization

mechanisms. Figure 3-8 is a DC spectrum showing the combination of all the above

ionization processes that occur separately in the ups-pulsed mode (Figures 3-5 through 3-

7) and simultaneously in the DC mode.

Sputtered species from the cathode diffuse toward the anode and are c\ cnlu.tLlly

deposited. In this case, sputtered species diffuse into the negative glow region, are

excited and ionized, and then sampled in the same region. In a DC glow discharge, the

negative glow region is always present and ionization of gas and metal species is always

occurring. This means that species containing the discharge gas and gas impurities will

always be in the mass spectrum. In the pLs-pulsed mode, the negative glow only exists for

a 10 p.s period and is then extinguished due to the rapid loss of electron population when

the pulse period ends. During the short on-period, long lived metastable atoms are formed










800-


*i--





- 400-








0


U


M/Z


Figure 3-6. Associative and Penning ionization for
the analysis of copper in argon at 90 ps delay time


+









2000


+
C U
C
1000 +

0





0 50


m/z

Figure 3-7. Penning ionization of copper atoms in argon
for the ps-pulsed GD at a delay time of 140 gs


















50 100


M/Z


Figure 3-8. DC GD mass spectrum of a copper sample in
argon


30000 -



20000 -


10000-



0-


JI-


i 1LJ


i-jj,-X-j










and become the ionizing agent after the pulse period ends. These metastables (in argon)

have enough energy to ionize the metal species but not the discharge gas or impurity

species, including those molecular in nature. This results in spectra dominated by the

discharge gas species at early sampling times and dominated by sputtered species ions at

later deflection times (after 100 lts). By selecting an appropriate deflection time,

reduction, but not removal, of these gas discharge interference can be effected. Figures

3-9 a and b show the schematic diagram of how the separation of gas and sputtered

species is effected in tis-pulsed GD-TOFMS. The first figure shows that gas species are

sampled prior to sputtered species in the glow discharge cell. Then by using a short or

long deflection time, a mass spectrum either rich in gas species or rich in analyte species

is obtained. Figure 3-10a shows the mass spectrum of an aluminum sample. At early

delay times, very little aluminum and lots of discharge gas species and impurity ions are

observed. By delaying the ejection of the ions into the TOFMS, the spectrum shown in

Figure 3-10b can be obtained; a spectrum consisting mainly of the aluminum ion. There

is no exact value to which the deflection time can be set to perfectly maximize the signal.



Glow Discharge Time-of-Flight Mass Spectrometry

Introduction

Time-of-flight mass spectrometry (TOFMS) is a well known technique in

biomolccular chemistry for the analysis of large biomolcculcs. It began in this field

mainly because other types of mass spectrometers suffered from limited mass range. The

time-of-flight has a theoretically unlimited mass range. As long as an ion is produced and






Electron
Ionization


Sputtering




Diffusion and
Ionization


"bu.n
O4 O

u


Cathode
I j1


*4 C


Sampler


Figure 3-9. a) Electron ionization, sputtering, diffusion and ionization
in the ps-pulsed GD


Ar+


Cuo



Cu+








Discharge Gas Sampling I f


0

0


Cu+


Repelling Pulse


Ar+


Sputtered Species Ion
Sampling


Cathode
I v


Repelling Pulse

"--- ---
Cu+
Sampler


Figure 3-9. b) TOF Sampling. Short delay times sample gas ions. Longer
delay times sample analyte ions







0.05


0.04


0.03


0.02


0.01


0.00

0 20 40 60

m/z


Figure 3-10. a) TOFMS spectrum of aluminum at 70 is





0.07


0.06


0.05


0.04


0.03


0.02


0.01


0.00


m/z


Figure 3-10. b)TOF delayed (230 jis) mass spectrum of aluminum


1 I ,~--


m l


I i 1 I 1 ----










can be accelerated into the flight tube, mass analysis can be accomplished. At that point.

detecting the ions is the limiting factor in the success of a particular analysis. GDMS has,

typically been dominated by quadrupole scanning systems and double focusing magnetic

sector instruments. These provide high sensitivity and fairly rapid analysis and peak

hopping capabilities, allowing the selective monitoring of se% eral peaks in a short time

period. In the last decade, TOFMS systems have been explored for elemental analysis,

and this dissertation uses TOFMS for ion analysis and for diagnostic studies, namely to

measure the effect of discharge gas composition on the uts-pulsed GD mass spectra and

general GD phenomena.



Time-of-Flight Mass Spectrometry

Theory

TOFMS works on the principle that if all ions are given the same energy then the

time it takes for them to travel a certain distance is determined by their mass. Thus,

equations can be used, starting with the kinetic energy equation, to calculate the time it

takes for an ion to reach the detector at some distance beyond the ion sampling point.

The kinetic energy equation states that

KE = 1/2mu2 (Eq. 3-1)

where m is mass (g) and u is velocity (m/s). By rearranging this equation, we can

calculate the velocity of an ion in the time-of-flight tube

v = (2zeV/m)/2 (Eq. 3-2)










where z is the charge on the ion and eV is the kinetic energy the ion is given upon being

accelerated into the TOF. Ion flight time is given by

TOF = L/u = (m/2zeV)/2 D (Eq. 3-3)

where D is the distance the ion must travel before being detected. A heavier ion will

travel slower and take longer to reach the detector. Ions having multiple charges will

experience more acceleration and will hence have greater velocity. This equation is the

basis for calibrating the mass spectrum (see section on Data Analysis) and will be used in

the next section to address mass resolution and signal broadening.



Output Characteristics

The resolving power of the TOF instrument can be considered a function of the

flight tube length. The resolving power of any mass spectrometer is given by

m/Am = t/At (Eq.3-4)

where t is the calculated flight time and At is the time difference of when the first and last

ions of a particular mass reach the detector. By increasing the flight tube length. t

increases. However, the separation of the ions in the mass bundle changes negligibly,

only because At will be determined by their initial energy distribution before acceleration.

Therefore as the flight time (t) increases, so does resolution. The resoli ing power is

affected by the accelerating voltage and the spatial and energy distribution of the ion

bundle. Figures 3-1 la and 3-1 lb64 show the problems of ion splaial and energy

distributions in the TOF s) stcm. Ions that are separated in space can experience a

different accelerating voltage. This affects its flight path and/or its flight time. Ions in

















SSource S-c


- Drift rion I)---


Ei-- =2s -



E=V/s E=O


Figure 3-11. a) Spatial distribution effects in TOF [Ref 64]


--I
1-


- -


@f-






















Source S 14







S=\V/s


D)rit region 1)







E= 0I
^V,


Figure 3-11. b)Energy distribution effects in TOF [Ref. 64]


.0 :


4










which flight path is affected result in noise accumulation at different parts of the nma-.

spectrum, resulting in poorer S/N and detection limits, or they simply hit the flight tube

walls and are neutralized. Ions in which flight time is affected result in poor mass

resolution. Ions closer to the accelerating plate will experience more voltage that those

further away. This results in a small difference in drift velocity, causing the faster ion to

reach the detector sooner than the less affected ion. By reducing this spatial distribution,

an increase in resolving power can be effected. Broadening of the mass peak generally

occurs when there are thermal energy variations in the incoming ion packet, and/or there

are spatial orientation differences caused by either space charge effects or as a result of

ion optic misalignment. These broadening factors can be reduced by judicious tuning of

the ion optics.



Performance Limitations and Potential Capabilities

The spectral production rate is a function of the slowest ion. For example, with a

2 m flight tube and accelerating voltage of 2 kV, an ion of mass 500 has a flight time of

90 jis. This means that the next pulse of ions into the TOF cannot occur until this one has

reached the detector because overlap of the ion packets may occur. Therefore the

repetition rate must not exceed 11 kHz. For the case of the microsecond pulsed glow

discharge, this factor is of no concern. Operating frequencies are generally not higher

than 1 kHz for the [is-pulsed GD source.

Wiley and Mclaren65 developed a TOF system that would increase the resol, ing

power over typical TOF systems of their time. The hardest factor to overcome was the










spatial distribution of the ions in the accelerating region. Figure 3-11 a shows that the

ions closest to the accelerating plate will experience a greater field than those further

away, creating an energy distribution and affecting different flight times once in the drift

region. Wiley and McLaren65 solved this problem by creating and separating the energy

and spatial distributions prior to the packet reaching the accelerating region by turning

the ions around after a certain flight distance so that the ions moved in a V shape. The

reflectron resulted in the distribution of ions being reduced in the turn around (reflectron)

region. This is achieved by applying positive and negative voltages in a manner such that

faster ions will spend more time in the reflectron than slower ions within a single mass

packet. Faster ions will require a longer flight path to slow them down and turn them

around as compared to slower ions. Thus, as Figure 3-12 shows, the ions of each mass

packet will reach the detector within a much shorter time and as a "tighter" ion packet.

and much reduced velocity distribution. The spatial and velocity distributions could be

ultimately minimized by using the reflectron. The TOF instrument used in this

dissertation has a mass resolution of 1600-1700 FWHM (full width half max) in the

reflectron mode.



Inherent Advantages of TOFMS

Cotter et al.66 studied many aspects of TOFMS and listed several advantages of

time-of-flight \ crsus all other mass spectrometric techniques.

1. The TOFMS is much more conservative of the sample in that every ion in an

ion packet is detected






























(b) / Spat






)Deflector



SI -,---
V


c-focus plane


, !4-..-d1
, I


-- I
_I .4d.]
L'o


_______ C-L2 *.....-..~

111111 /


Figure 3-12. Reflectron TOF with spatial and temporal
correction by the reflectron lens [ReT. 64]


----.,2


I










2. The open flight tube with the absence of slits presents a very wide aperture to

the source of ions

3. The TOFMS has no fundamental m/z limit, other than detectability, on the

range of m/z values analyzable. Additionally the mass/charge scale can be

calibrated accurately based on equations 3-1,2, 3, and 6

4. The process of creating an ion packet or bunching can result in low sample

utilization via a low duty cycle and can improve mass resolution

5. Very fast electronics are required for detection of transient detector outputs.



Instrument Design

GDMS requires the physical invasion of the plasma in order to measure the ions

produced. The extraction of ions from a glow discharge plasma would benefit from a

design that maximizes the number of ions sampled. Optical techniques have some

advantage in that the measurement of a photon does not require invasion of the plasma.

However, mass spectrometry is considered to be a much more sensitive technique for the

detection of ions. The following section describes the design of the prototyc GD-TOFMS

instrument used in this dissertation.



Ion Source

Direct Insertion Probe (DIP)

The direct insertion probe is a generic device used in GD devices. The essential

parts are shown in Figure 3-13. An insulated stainless steel rod serves as the interlace







insulated
wire


solder
connection


copper
sample
holder

( ,,T,


stainless steel
body


Figure 3-13. Direct insertion probe


[I


sample


macor
shield


electrical
feedthru










between the powered lead and the sample. A small handle (not shown in Figure 3-13) is

used to couple the power to the back of the probe and is used for easy removal and

position manipulation of the sample in the glow discharge chamber. The sample is

mounted into a small holding chuck that connects the sample and high voltage lead at the

probe tip. The holding chuck is covered with teflon tape so that no surface at high

voltage, other than the sample surface, is exposed. The samples were machined in the

University of Florida Chemistry machine shop so that they could fit into the loading

chucks. The top surface of the sample has a 4 mm diameter. The entire probe tip and

loading chuck are protected by a Macor shield that minimizes surface exposure to only

the top of the sample surface. The shield was not in contact with any part of the sample

or any other high voltage surface to prevent shorting the power supply. These shields

were also machined here at the University of Florida. To prevent arcing when gas creeps

inside of the shield, the loading chuck was wrapped with teflon tape.



Discharge Cell

Two discharge cell designs have been used in this dissertation. The first as a

modified six-way cross with a 450 viewing port that allowed viewing of the sample while

in operation. The source was large and bulky, containing several vacuum interfaces so as

to minimize the amount of contaminant leaks in the system. Figure 3-14 shows a picture

of this source. The coupling of this source to the mass spectrometer had one severe

problem. There was no way during sample exchange to protect the 2"d and 3rd vacuum

stages necessary for ion detection. This is the primary reason why various vacuum







































Figure 3-14. Photograph of direct insertion
probe source #1










interlocks were needed during sample interchange or source cleaning and maintenance.

Hence, the need for vacuum protection led to the use of a slide valve that would ensure

the vacuum integrity while sample interchange or source cleaning was required. This

slide valve was nearly identical to the one constructed for the interface of the quadrupole

GDMS system in this laboratory. The new slide valve interface was designed around the

development of a Grimm-type ion source. This meant that the six-way cross design

could no longer be used. A simple source plate was developed to fit the Grimm-type

designed interface and allowed the same operation of the direct insertion probe. The

interface design is shown in Figure 3-15. Regardless of design, the interface of the probe,

sample, and mass spectrometer is the same for both sources. The only severe limitation

of the modified source is its exposure to air during sample exchange. In our efforts to

maintain vacuum integrity and create source simplicity, we sacrificed cell exposure to air

and its contaminants.



Ion Extraction

Ions are extracted from the glow discharge plasma by moving the sample surface

(and plasma) within 8 mm of a sampling orifice. The sampling orifice is 1 mm in

diameter and is the first of two high vacuum interfaces. The source is operated at

approximately 1 torr; beyond the sampler, the pressure is at 10-4 torr. As ions pass

through the sampling orifice, the rapid drop in pressure causes a jet expansion of the

plasma into a field free region. A skimmer cone placed beyond the sampling cone

samples approximately 1% of the ions produced in the expansion jet and leads to the ion







































Figure 3-15. Direct insertion probe source #2. Designed
for exchangability with a Grimm source interface










focusing lenses. The skimmer cone lies at a distance, 0.7 cm, equal to the mach disc of

the jet expansion of the plasma. At the mach disc distance, ions have reached their

maximum velocity. The skimmer cone also serves as the interface between the jet

expansion and lens and ion optic region that lie at a pressure of 105 torr. Figure 3-16

shows a schematic of the source interface and ion optics region.



Ion Optics and TOF Sampling

The ion optics for the time-of-flight system were constructed at the University of

Florida. The sampling plate interface could either be grounded or floating during ion

extraction. The choice was dependent on the mode of operation. A higher signal was

obtained for the jps-pulsed discharge when the sampler was left floating. Once beyond

the skimmer cone, ions were focused via a set of ion optics including a set of steering

plates. The ions passed through a plate with a small slit. The plate has a small potential,

which minimized the spatial distribution of the ion packet just prior to TOF sampling.

The function of the slit was to reduce the spatial distribution of the ions just before they

reach the deflecting region where ion distribution can severely affect resolution. This

opening measured 10 x 2 mm. The steering plates controlled the position of the ion beam

and adjusted it into an elliptical cross section from its round cross section, allo\\ ing a

greater number of ions to be pulsed into the TOF accelerating grid and creating a

"tighter" ion packet. Sampling in the TOF system occurred orthogonally to the

production of ions and ion focusing. The ions focused in the ion optics filled a repelling

region in which a 2-5 jis long, 70-80 V pulse deflected ions into the TOF accelerating




Samy


Direct
Insertion
Probe

1zi

Samp
1 torr


)ler
Skimmer
Slit
Steering
j Plates


- c- -
Optics


2x104 torr


8x106 torr


1.5


Repeller -


SOV '~i'" G

-2000V 25 mm G


m


Flight Tube


L2 L3
II0


To Detector


Figure 3-16. Schematic of the source interface and ion optics region


1


Faraday
- Detector

2

__Deflectc
Plates





_Microchanr
Plates










grid. A constant 1.8 kV voltage accelerated the ions into the linear portion of the flight

tube. After a 1.5 m flight, ions were detected. For the studies here the ions traveled into

a reflectron lens system in which they were reflected into the reflectron flight tube for an

additional 0.97 m flight. At the end of the reflectron flight tube, ions were detected by a

microchannel plate.



Electronics and Timing

The temporal aspects of the pulsed GD have been discussed. To take advantage

of these aspects, coordination of the GD pulse and TOF sampling must be optimized. In a

DC glow discharge source, the need for fast electronics is minimal because signal

variation and timing are insignificant. Once transient signals are produced, especially in

the microsecond regime, the need for fast electronics in order to capture the signal is

critical. Triggering the time-of flight mass spectrometer to sample the ion population at a

selected time after the pulse of the glow discharge source is necessary (Figure 3-17).

Additionally, in order to study the ionization processes outlined above, the delay betw een

GD pulse and TOF sampling (deflection time) must be adjustable. Therefore fast

electronics were needed. Figure 3-17 shows an example of the timing scheme necessary

to obtain spectra rich in analyte with reduced or no contribution from the discharge gias

species.







A
Applied voltage







B:
Ion profile




C:
Repeller signal


Figure 3-17. ts-pulsed GD timing scheme










Vacuum System

Plasma sources can operate anywhere from the millitorr to atmospheric pressure

range. This means that differential pumping is required in order to maintain the

necessary pressure to transport the ions from their source to the detector. For this system,

source pressure is about 1 torr, second stage at 10-4 torr, and third stage at 10-6 torr. The

vacuum pressure in the 2nd and 3rd stages allow the drift of ions from the source to the

deflecting region, 14 cm, and then from the deflecting region to the reflectron detector

2.5 meters in distance. Vacuum integrity is important for all MS systems. The ability to

maintain a high vacuum in the 2nd and 3rd stages will determine the 1st stage or source

pressure. Our GD-TOFMS system uses 3 turbo pumps. A 1000 L/s turbo pump

(Tubovac 1000 C, Leybold, Export, PA, USA) is connected to the second stage, and a

250 L/s (Varian Vaccum) and a 170 L/s (Varian Vaccum) turbo pump are connected to

the third stage flight tube. All the turbo pumps are backed by 13 1/s mechanical roughing

pumps. The flow rates correspond to the maximum pumping capabilities for nitrogen at

the operating pressure for the GD.



Ion Detection

TOFMS systems commonly use an electron multiplying device such as a

microchannel plate as the detector. Many types of software are used to obtain spectra and

perform elemental analysis. A dual-chevron microchannel plate (MCP) (Galileo,

Electrooptics, Sturbridge. MA, USA) is located at the end of the linear section of the

flight tube behind the reflectron lens and at the end of the reflectron flight tube. The










linear mode was not used in the work presented here. Much better resolution can be

obtained in the reflectron mode due to the longer flight distance of the ions.19

Two detection methods exist that can take full advantage of the multiplex

capabilities of the TOF mass analyzer. The first is a real-time averaging mode in which

the ions from many pulses are averaged and viewed on an oscilloscope, which allows the

user to view real-time spectra and analyze the fundamental processes occurring in the

glow discharge. This is a running average in which every pulse is averaged to the 99

pulses before it. This mode is also used for ion lens focusing and signal optimization.

The second method is an ion counting or accumulation method, commonly used to

measure trace elements in samples. EG&G has developed a Picosecond Time Analyzer

(PTA) (6109, Eg&G Ortec) that can accommodate the transient signals that the

microsecond pulsed GD-TOFMS system produces. The bulk and trace mass spectra

presented in this dissertation were obtained using these two detection schemes.



General Procedure and Data Analysis

Samples were prepared for analysis as described in a previous section. Once the

DIP was in the source, evacuation by a 13 L/s roughing pump ensues. The discharge gas

was then added and the pressure was measured via a wide range capacitance manometer

(WV-100-2 Varian Vacuum). After pressure stabilization, the source was opened to the

2nd and 3rd stage by opening the slide valve. A 3 or 5 kV pulse power supply (IRCO,

MD) was used for discharge powering. The sample was sputtered for 10-15 minutes, to

effect sample surface cleaning and stabilization of the discharge. Spectra from the










oscilloscope were obtained using software developed in our lab from a National

Instruments Data Acquisition program. Trace spectra were taken using the PTA program

software. Generally, 50,000 pulses were accumulated using the PTA counting mode.

Spectral data were downloaded into Microcal Origin*(TM), where files were conm erted

into mass spectra. Mass scaling was completed using Origin according to the equations

shown in the TOF theory section such that.

Mass = at2 + b (Eq. 3-6)

where a and b are constants that are determined when two separate masses are used to

calibrate the mass spectral scale, and t is the flight time of the ion.

Sputter rates were determined by sputtering a copper sample for 30-45 minutes.

Samples were weighed on a milligram analytical balance, with a range of 0.001 mg to 20

mg 0.0001 mg. Sputtering, by pulse, was calculated by dividing the weight loss in mg

by the number of pulses over the given sputtering time. At least three replicates of each

gas were performed.

The gases used in all experiments were research grade 5, 99.999 % pure from

Spectra Gases and from BOC Gases.




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EV3CDAY4V_IH0IQ9 INGEST_TIME 2013-03-05T20:17:05Z PACKAGE AA00013534_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

SPUTTERING AND IONIZATION BY HELIUM AND ARGON IN TH MICROSECOND PULSED GLOW DISCHARGE USING TIME-OF-FLIGHT MA SPECTROMETRY By MATTHEW MOHILL A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2001

PAGE 2

TABLE OF CONTENTS A KNOWLEDGEMENTS ........... .. ...... ... ... .. ... .. ............. . . ...................... AB TRACT .. .. . .... .. ...... ~ ..... .. ... ... .. .............. . .. .. ............ .. ... ......... ii CHAPTERS ............. ........... ........ .. ... ... ........ ........ . ............... ... ........... 1 INTRODUCTION .. .. ... .. ... .. ... ... .. ............... .. .... .......... ........... .... ... 1 olids E lemental Mass Spectrometry ............... ...... .... ..... ... ... ............... 1 olids Analysis Techniques for Mass Spectrometry ...................... .. ........... 3 Glow Dischar ge and GDMS History ... .......... ... ............ .. . .................... 5 Scope of this Dissertation .. ... .... ... .. ... ........ ...................................... 7 2 TH GLOW DISCHARG E . ... .. .. .... ..... .......... ..... .. ................... .. .. ... 9 Ga eo us Di charges .. ... .. .. .. ....... ........... ... ..................... ... .............. 9 The A bnorm a l Glow Discharg e ..... .. .. ... . ............ ... ..... .......... .. ....... .. 1 2 Regions of the Glow Discharge .. .. ... .... ..... .. .. .. .. ..... ... .. .. ....... ......... 14 Glow Di charge Proc esses .. . .. .. .. ...... .. ....... .. .. .. .................. .. ... ....... 18 Glow Discharge Sputtering .. .. .................. ..... . ............... ... ..... 18 Ma s of s putt e ring ion .. ... .. .. ..... .. .. ..... ...................... .. . 23 Angle of incident ion bombardm e nt ....................... ........... 23 In c ident ion e ner gy .... .. .......... ........... .. .. .... .. ............. 25 T h e target mat rial ... . ... .. .... .. .. .. .... ........................... 29 G l ow Di charge Exc itation/Ioni zat ion .... ... .. ......... ...................... ,., 1 olli ion in Ga . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 putt r d p ci x citation ........ ...... .............................................. 35 putt r d p ci Ioni za tion ... .. ... ..... ... .. ... . ....... .... ......... ....... .. .. ... 3 8 c ndar y xc itati n/I ni za ti n Pr c .................... .... ......... . ...... .42 y p of 1 w Di c h arge D v i ce ...................... ............................... 44 3 XP .. RIM .. NT L N ID TI N ............................................. 48 p 1 w Di c h a r g ...... ... .. ...... ...................................... 4 urr nt .. ................................................................... 4 r qu n y ................................................................... 50 Th Mier c nd Pul ...... .. . .... ...................... .... . ............ .. .. 51 II

PAGE 3

Pul ed Glow Di charg Ma p ctr m try ........................................ ... 51 D v -Pul ed xp rim ntal n id rati n ................... ........... ........ 52 puttering and Ionization in the M i cro econd Pu l d D ........................... 53 Analytica l Advantage of D GD ... . ... .. ... .... .. ... ........ ....................... 54 Analytica l Advantages of Micro cond Puls d GD ................................... 59 Atomic Emission Signa l Enhancement .... .. .. ... ... ..... ..................... 59 Sputt red pcies Ion Signa l Enhancem nt .................................... 59 Tempora l Cons id erati no s in GDMS ........ .... ................................ 61 Glow Discharge Time-of Flight Mass Spectrometry ................................. 68 Introduction .............. ........................ .. ........... ... .............. .. 68 Time of Flight Mass Spectrometry ................................. ....... ..... 73 T h eory .. .. .................. .... . .. .. ..... .. .. ...... . ................... 73 Outpur Characteristics .. .. .. .. ..... .. .. .... . ... .. .................... 7 4 Performance Limitations and Potential Capabilities ................. 77 Inherent Advantages of TOFMS ............ ........ ..... .... ... .... 78 Materia l Transport and In stument Design ... .. .. . ........ ..... ................... 80 Ion Source .... .................. ......................................... 80 Dir ect Insertion Probe . .................................... ............ 80 Di sc har ge Cell .... ...... ... . .. .. .. .. . ............................. 82 Ion Extractio n .... .. ... .. .. .. . .. ..... . .. . ... .. . . .. .......... ... 84 Ion Optics and TOF Sampling ........ ................................. 87 E l ectron ics and Timing .. . ... .. ....................... ................ 88 Vacuum System .... .. .. ...... .. .. ...... ............................. 90 Ion D etec tion . ... ..... .... .. .............. ............. ...... . .. .. ... .. 90 General Proc e dur e and Dat a Analysis ........................ . ... .. ............ ....... 91 4 CHARACTERIZATION OF THE MICROSECOND PULSED GD PROCE SES IN HELIUM ................ .. ...... .. .. .............................. ................. 93 Introduction .... ...... .. .. .. ... ....... ..... ... ... ... .... ..... . .... . ........ ... ....... 94 Background . ........ .. .. .. .. ......... .. . ..... ... ... .... . ... ...... .......... .. .. . .. 96 Microsecond Pulsed Glow Di sc h arge Processes: Comparing Ar and He .... .... 99 Di sc h arge Initiation .... .. ...... .... .. ........ ........... ............ 99 E l ectron Ionization ....... ......... .. ................................... 100 Voltage effect .. .. .... .. .. .. . .. .................... .. . .. .. .. 105 Pressure effect . .. ............... ....................... ..... ... 107 Sputtering of Atoms . .. .. ...... .. ... .. ...... ........ .. ............ . 107 Incident ion energy . .. ..... .. ... .. ... . . ... ........... ....... 113 Pressure effect .... ..................... ........ . . .. ..... .... 115 Voltage effect . .. ...... .......... ... ............... ......... 117 Diffusion of Atoms ..... .. ..... .......... ... ........... ............... 119 Ionization of Atoms .. .. .. .. .. .. ......... ... .... ... ............... 126 ampling ................... .. .. .. .... ........... .... .................. 127 umm ary ............. . .. .. ..... ... ..... ............................................... 1 4

PAGE 4

5 H H L I M DI H A RG E LYT I CAL HA RA T RI T I .......... 1 38 In tro du ct i o n ............. . . .... .. .. ..... ... ... . ...................... .. ... ............ 1 3 8 I d e n t i fy in g Int e r fe r e n ces . ... .. ... .... .. .. .. .. ... .. ..... ... .. ..... .. ..... ........... 1 3 8 In t r rin g Ion F orm a tion . . ... ................................. .. ...... ....... .. .... 1 39 R m o a l of In te r fe r e nc e ... ...... ..... ..... . . ... .. ................. ... ............ 1 4 1 D eter m i n a tion of C alcium in th e Pr ese nc e of Arg on . ...... .. ... ... .... ........... 1 4 7 R e l a ti e S e n s iti it y Fa ctor s .. .. ... ..... .. .... .. ... ... . .................. ... . ...... 151 R e l a ti ve Ioni za tion Fa ctor .... ........................ ... ............................... 1 60 umm ar .... .. ...... ........ ........... .. ... ... .. .... ... ..... ....................... 1 6 6 CLUSTE R F ORMATION AN D IO N I ZA TIO N IN A -P L D G D ......... 1 64 Introdu c tion ................... ...................... .. ......... .. ...... . ................ 1 64 B ac k gr ound .. .. .. . . .. . . ... .. ... .... ... ............. ..... .. .. .. .... ............ 1 65 S putt e rin g T h e or y . .. ... .. ... ... ... .. ... .... ... . .... ............................... 1 67 C lu s t er F ormation Th e or y . ..... .. ... .. . ... .. ....... . .. .. .. .. ........... ....... 1 70 Dir e ct e mi ss ion ...... ......................... .. ... . .... .. . .. .. ............ . 1 70 A tomic Combination ( A C) Singl e C olli ion . .. . .. .. ...... ............. 1 73 A tomic Combination b y Double Collision ....... .. .. . . .. .. ............ 1 74 Theo r y A pplication to a Glow Di s char g e . . .. ... ... .. . .. .. . .. .. .. ........... 1 76 Di ffus i o n Di s sociation and Ioni z ation of Pol y atomic s . .. ... .. ............. ... . 1 8 0 Ma tri x S p ec i es Di s tribution : A C athode-to-Orific e Di t a n ce t ud . .. ........... 188 T h e ffec t o f Pr ess ur e on Pol y atomic F ormation ..... ...... ....... .. .. ........... 1 95 T h e Effe ct of Volt age on Pol y atomic F orm a tion .. .. .. ............................. 1 97 urnm a r y .... .. .... .. .. ...... . . .. ... ... .. ... ... .. ... . ................. ..... ......... 1 97 7 C O NCLU DING REMARKS AND FU T U RE DI RE T IO . . ....... .. ....... 200 8 REFEREN S ............... . ... ...... .. . . .. ... .... ................................. 204 9 B I OGRAPH I CAL KETC H ... ........ .. ... .. ... .. ............................... 2 1 0

PAGE 5

ACKNOWL DG M NT It i with much pleasure and much sorrow that my graduat car r 1 coming to an end. I am indebted to many people who have not only been mentors but al fri nd I have the utmost respect for my advisor Dr. W.W. Harrison who ha taught me cientific integrity and discipline, and has guided me through the process of writing this dissertation. The work here is but one of many that have been performed in the Harri on/ Winefordner group laboratory and would not have been possible without th upport and help from my peers. Special recognition goes out to Dr. Ben Smith. He has been an active part of the success of my laboratory work and has enriched my time here at University of Florida. Dr. Wei Hang and Dr. Kristofor Ingeneri introduced me to GDM and taught me much about the technique. I must thank Kris for many late night talk Dr. Chenglong Yang deserves special attention. Chenglong was a great laboratory mentor and friend. His hard work and help in the lab were deeply appreciated. I must extend thanks to the rest of the members of the Harrison and Winefordner group new and old. I will miss their friendship and lively banter. I also wish to thank Dr. Christopher Barshick. He has been nothing hort of a great friend over the last five years and was the one who convinc d m to join th Harrison group. I thank him.

PAGE 6

final ord mu t b said about my family and clo friends. I ould not b th p r on I am toda ithout th lo e and upport of my parent through the ry low and r high the hav kept me on an even p l ane. I wi ll fore r b indebt d to th m r this and I honor them with this dissertation. Thanks Mom and Dad!!! Thi ork has been supported by the US Department of nerg Ba ic nerg i nces H wlett-Packard Laboratories and LECO Corporation.

PAGE 7

Abstract of Diss rtation Pre ented to the raduate chool of the Univ rsity of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philo ophy SPUTTERING AND IONIZATION BY HELIUM AND ARGO IN TH MI ROSECOND PULSED GLOW DISCHARGE USING TIME-OF-FLIGHT MA SPECTROMETRY Chair: Willard Harrison Major Department: Chemistry By Matthew Mohill MAY2001 The focus of this dissertation is the study of the sputtering and ionization phenomena in the microsecond pulsed glow discharge using helium and argon as plasma gases The glow discharge is an electrical plasma which directly atomizes and ionizes solid metal samples. In this dissertation a short intense voltage pulse is applied to the sample creating temporal aspects that are both fundamentally and analytically useful. An in-hou e built reflectron time-of-flight mass spectrometer is used in these experiment in which everal ion source configurations can be used including a direct ins rtion prob and a Grimm-type device. Th ionization phenomena in helium and argon are studied. Th high h lium meta tab! atom energy re ult in higher ionization capabiliti s The -pul m d mu t be u din ord r to produce higher analyte intensiti that ar not achi vabl in th D mode. Th temporal advantag and el ctiv ioni z ation of putt r d p ci I I

PAGE 8

p c1 afforded b argon ar lo t wh n helium is used as th pla ma ga The remo al f ma peaks in critical part of the mass spectrum without lo ing sen iti ity 1s a b nefit of the u of h lium as th plasma gas. Th impact of argon and helium ions and the resulting collision ca ade tudied relative to mass spectra observed for a copper matrix. Polyatomic clu ter of th copper matrix form and interfere with several elements in the ma s sp ctrum. Th formation dissociation and ionization of these species are d tailed in thi di rtation. The analytical advantages of using helium as the plasma gas to ffect det tion of iron selenium arsenic tellurium gold and lead are discussed. For lead and tellurium a reduction in the formation of polyatomic clusters results in the r mo al of th ma interferences. Relative sensitivity factors are calculated for both helium and argon showing that the higher metastable atom energies of helium make it a l ionization agent resulting in signal enhancement of hard to-ionize elem nt uch a A e and Au. A r lativ ionization factor is defined which mea ur the putt ring and ionization fficiency of helium compared with argon. Ing n ral arg n 10n ar ffi nt at puttering a copper targ t but argon metastable atom ar only mod rat l abl t ioniz th atom population produced. Helium i an inefficient sputtering ag nt but an xc 11 nt ionizing ag nt for the pecie pre ent in a glow di charg ourc

PAGE 9

CHAPTER 1 INTRODUCTION Solids Elemental Mass Spectrometry Elemental analysis of solids employs a wide variety of detection techniques of which mass spectrometry is one. A mass spectrometer requires the production of ion for analysis and subsequently separates them by their mass-to-charge (m/z) ratio. Inorganic elemental analysis has traditionally been dominated by optical methods in which the excitation and subsequent radiative relaxation of the analyte produce photons characteristic of the energy difference between the accessible excited and relaxed state of the analyte atoms. These photons are measured using spectrometers diode arrays and other similar detection devices that convert photons to electrons The difficulty for optical methods is not in the analysis of pure materials but rather in the analysis of multi element materials in which transitions of several elements may have similar energy such that an overlap of wavelength occurs This spectral complexity with many tran ition for a single element is quite opposite of what is seen in mass spectrometry where the elements are typically singly ionized and typically have few isotope Ma s spectrometr also suffer from mass interferences ; however there are many scheme that combat isobaric interferences in glow discharge mass spectrometry (GDMS) In general pectral simplicity in MS is advantageous. Mas spectrometry ha been utilized since the early 1900 hen J.J. Thom n ob erved positive rays from a low-pre sure electrical discharg 1 What all p iti

PAGE 10

2 10n re tudi d by Thomson for isotopic abundance measurements material anal i and the ph 1c of th lectrical discharge. The electrical discharge used b Thom on continued to be developed as an ion source for material and i otopic anal si and e entually a need for improved precision and higher sensitivity in mass spectrom t r arose. The ability to detect lower abundance isotopes of variou elements wa ati fied b the continuing development of mass spectrometers in particular the quadrupol 2 The development of quadrupole mass spectrometry, in the 1950s would become important for organic analysis. Inorganic analysis would also benefit from the maturation of the technique and become a popular mass spectrometric technique in the field. With improvements in mass spectrometers the construction of ion source mor suitable than the Thomson electrical discharge were eventually develop d. A ton creat d an ion source for low-volatility solids such as refractory metals and m tal oxide 3 and small improvements continued with the Thomson electrical di charge 4 The big breakthrough in ion sources for solid samples wa the vacuum spark. s 6 The park i produced by a high voltage at high frequency between two l ctrode (t picall wir ) t vap rize and ionize solid material from on of th el ctrod Th park wa originall u ed a an optical mission device and wa lowly dev lop d into a ource f ion for ma p ctr m try. Th park ourc had th advantag of co ring th full l m ntal rang with typical d t cti n limit of0.1 ppm r p rt d by B uman and ijb 1 7 -&

PAGE 11

3 Solid Analysis Techniques for Ma Techniques for the direct analysis of solid must fulfill thr basic r quir m nt : 1) vaporization/destruction of the sample; 2) atomization of the sample and 3 ) i ni z ati n. The techniques that are capable of performing all three in a single environment ar e on in which there is interaction of a plasma and or gas ion with a solid surface wh r particle bombardment at the surface and particle collisions in the gas phase satisfy the requirements. Energy is needed in order to effect requirement # 1 above. The energy applied in plasmas occurs in two distinct forms: electrical energy and optical radiation. Glow discharges spark sources microwave and ICP plasmas all use electrical energy while pulsed lasers use optical energy produced by photon flux. The energy in particle beam sources is strictly kinetic. Secondary ion mass spectrometry (SIMS) and fast atom bombardment (F AB) mass spectrometries both are particle beam sources that tak advantage of surface chemistry to produce ions. A discussion of kinetic energy transfer via elastic collisions follows in Chapter 2. As is the case for the development of the spark source for mass sp ctrom try th development of new ion sources is driven by the status of material cience which reqmr preci ion and accuracy in analyzing a sample particularly for ampl impuriti Surface technique used in material science have also become critical ar a of tud m analytical chemi try Requir ment for knowing the amount and patial di tributi n f impuritie on the urfac of a olid material hav er ated th ne d fi r b tt r

PAGE 12

4 charact rization techniques In de eloping these ne 10n sourc s the com pl it of th ma s spectrometer system and type of mass spectra observed are all determined b th anal i sought and the ion source used These entities are then respon ibl for th calibration capabilities and the quality of the mass spectra obs r ed. In th d elopment of these sources research has focused on the decoupling principle in hich the atomization and ionization steps are physically separate. For example in using a la er for solid sample analysis the laser is used to vaporize the sample the sample particle ar then transported into an ICP plasma for subsequent ionization. Ionization in direct la r analysis does occur but an ICP is more often used because of the wide range of ion kinetic energies that are produced during laser analysis. In the case of a glo di charg the atomization and ionization steps don t occur at the same time e en though th tak place in the same plasma volume in which no physical transport of th atom i r quir d for ionization to occur. The spark source mass spectrometer sy tern wa the fir t big br akthrough in olids analy i The source was originally coupled to the Mattauch-H rzog doubl fi cu in g ma pectrometer capable of nergy and angular focu ing in a ingl plan u t n rgy di tribution and the erratic nature of th park urc thi ma p ctr m t r y t m wa the only on capabl of r ducing thi n r g di tributi n fi r a ur a t anal f th ion pr duced imultan u l In th p ctr m tr IM ) a d lack f pr ci i n and a uracy f th park urc 9 J.J. h m 11 wa th fir t t th j cti 11 of c ndary i n fr m a urfa aft r primar i n

PAGE 13

5 bombardment 9 and the first instrument of thi kind was d e v lop e d b y H r zog a nd Viehbock in 1949. 1 0 In the 1970s development of IMS as a method ford t ctin g organic compounds on a surface became the primary focus and continu to b on f th leading applications 11 1 2 The development of SIMS wa most likely head d in th direction of organic analysis and left the inorganic elemental analysis field open for improvement. Plasma source development included the work ofFassel 1 3 with the ICP and lat e r its combination with a quadrupole mass analyzer for the elemental analysis of solution s. The ICP combined with mass spectrometry has developed into the most powerful elemental analysis technique. Microwave induced plasmas (MIP) were also developed for solution-based elemental mass spectrometry 14 yet the MIP didn t develop into a commercial success. The MIP is still developing new directions including the plasma torch 1 5 and other configurations that take advantage of its excitation capabilities a s opposed to its atomization and ionization capabilities. Over the same time period glow discharge development led to its use in the plasma mass spectrometry field. The developmental history of glow discharge into an analytical technique follow in the ne x t section. The use of a glow discharge as an ion source was a critical point in this hi tor and is important for this dissertation. Glow Discharge and GDM History Glow discharged vices have long been u ed in atomic p ctro op and ar routinely used for analysis of solid material Hollow cathod w r u d b P a h n t

PAGE 14

6 m tigate atomic structure by atomic emission and absorption. 1 6 The u e of hollo cathodes as limited to analytical atomic emission as a means of det rmining lo concentrations of elements across the periodic table including nonmetal In the fi e decade that followed Paschen s work the application of GD d vice in chemi tr de eloped slowly mainly due a lack of instrumental advancement and a poor understanding of the analytical capabilities of the glow discharge. The la t 30-40 year has een the low growth and maturation of the glow discharges as a techniqu for olid analysis including the development of commercial instrumentation. An ed for more routine analysis resulted in the development of' restricted glow discharge b Grimm 1 7 in which flat samples were used directly as the cathode material. Ju ta holl cathod grew in popularity when Paschen first introduced them, so did the Grimm-t pe GD device. It has become the metal industry standard for bulk anal i and particular! r depth profiling by atomic emission spectrometry Glow Di charge Mass Spectrometry Analytical glow di charge mass spectrometry has been around me th arl 1970 wh n oburn reintr due d th glow di charg a a olid i n our r th analytical mmunity. 1 8 oburn wa abl to h w th utilit of m nitoring th i n pr due d by a putt r d targ tin b th dir ct curr nt (D ) and radi fr qu nc (RF di char g In th mid-t -lat 1970 th d v l pm nt fa quadrup l -b d DM und rtak n b th Harri on gr up at th ni r it f Virginia in hi h bulk lid and luti n r idu anal main focu Radi fr qu n RF) and dir ct

PAGE 15

7 current (D ) modes are the traditional and more commonly u ed mod s of op e r a tin g th glow discharge. RF is primarily used for non-conducting materials which cannot be analyzed in the DC mode. The DC mode is more widely used especially in th metal indu try where trace analysis is important. Pulsed discharge have been around for m a n y years primarily for enhancing signal from hollow cathode lamps. Many improvement have been undertaken to develop better GD ion sources for mass spectrometr y. One o f the most important developments for elemental analysis is the resurgence of time-of flight mass spectrometry (TOFMS). With rapid analysis time of the full elemental ran ge the TOFMS system has been shown to be quite advantageous in analytical chemistry 1 9 20 Chapter 3 will describe the operation of the glow discharge with respect to analytical performance and its coupling to the time-of-flight mass spectrometer. The Scope of this Dissertation Developmental and fundamental studies of the pulsed glow discharge have been of primary interest in the Harrison group over the last twelve years With the ad v ent of the microsecond pulsed source a new series of studies have become of interest. 2 1 23 A thorough comparison of the helium gas characteristics in the micro econd pul ed pla ma including sputtering and ionization has yet to be accompli h d. The focu of thi dissertation is the characterization of the -pulsed glow di charge u in g argon and helium as plasma gase The following chapter describ th dif r nc b m the u e of helium and argon a discharge gase for ma p ctrom tric purp Microsecond pul ed di charg proce e ar con id r d for ach ga ith p ci a l

PAGE 16

8 att ntion gi en to atomization and ionization. One of the problem facing D and RF GDM i the high background from the discharge gas and related sp cie One goal of using different gases and the microsecond pulse mode is to alleviate the e interferenc problem reducing and removing interferences in key parts of the mas pectrum ther b facilitating element detection.

PAGE 17

CHAPTER2 THE GLOW DISCHARGE Gaseous Discharges When a sufficiently high voltage is established between two electrodes in a gaseous medium a breakdown of the gas forms positively charged ions and free electrons. The positive anode and the negative cathode create a voltage gradient in the plasma such that an electrical circuit has been created where the discharge carries the current. This is what is called a gaseous discharge and these discharges can be defined by the current at which they operate. Figure 2-1 24 shows a schematic diagram of the characteristics of gaseous discharges based on current and voltage. The three main discharges shown in the diagram are: 1) the Townsend discharge 2) the glow discharge and 3) the arc discharge. The Townsend discharge operates in the sub-millitorr pressure range and is characterized by the production of very few ions and electrons or low current. This discharge is low in luminosity due mainly to the low production of excited species which give a gas discharge its glow" characteristic. The Townsend discharge i not elfustaining but relies on external X-rays and UV light to produce lectron and ion in th pla ma. The Town end discharge has been u ed in analytical ma p ctrom tr int o ways: first as a way of ionizing the fill gas instead of high energ l ctron mitted fr m a hot filam nt ; 25 and econd a a ch mical ionizati n ourc 26 Th tran iti n from th Town nd di charg to the glow di charge i di tingui h d a a r gion in hich l ctri al

PAGE 18

10 (l) {l\ 111 1\1 c:1 ,. O' Ill \1 IJ o:: I t1' 0 .... IJ 11' o~ ~G ~, IJ e-, a f-, a: ;;:: lJ ,< (1> 0-, j Vb t 0 > Vn -9 10 -7 10 ' -5 10 ' -) 10 Current (Al l 10 10 Figure 2-1. The current-voltage characteristics of gaseous discharges showing the plasma coverage on the sample surface. Vb = breakdown voltage V n = normal operating voltage V d= operating voltage of the arc discharge [Ref. 24]

PAGE 19

11 curr nt incr ases as th discharge voltage deer a e This incr a e in curr nt a nd d er a in voltage are mostly due to the increase in collision energy xchange at high r g a s pressure. The glow discharge has two modes a normal glow di charge and an abnormal glow discharge. The normal glow discharge is characterized by its luminous glow Increases in current do not result in a current flux change because the plasma hasn ta yet covered the surface of the cathode entirely (see Figure 2-1) No voltage increase therefore is required to sustain the plasma as long as current increases. Once the entire surface is covered by the plasma due to increasing current the current den ity undergoes change and a transition to the abnormal discharge occurs Increases in current result in increased current density in the abnormal glow discharge region Increases in voltage result in increases in current and help sustain the plasma. The abnormal glow discharge region where plasma atomic spectroscopy is performed is the region of gaseous discharges used in this body of work. The following sections will be devoted to de cribing this region. As the current is increased thermal vaporization of the cathode occurs due to the high current density and increased bombardment by the filler gas. The availability of high analyt number densities affect the potential fi lds and the current-voltage characteristics become "normal" such that current increases but the maintenance oltage decreases. This defines the transition from the glow discharge to th D arc. The arc di charge i commonly operated at atmosph ric pre ur ha lar g curr nts (10-1000 A) and a bright plasma. The high current h at th urfa much

PAGE 20

12 that th rmionic electron emission becomes a prominent current carr ing mechani m. Ga t mperatur s are typically as high as 2 x 10 4 Kand charged particl den itie about 10 1 6 cm3 The arc discharge has been used in atomic emission anal is of solid m tal 27 and in vapor deposition of thin films. 28 Recent papers have reviewed thi old technique for new applications. 29 The Abnormal Glow Discharge The glow discharge source is a simple device. At its simple t it consi ts of an anode a cathode a fill gas at low pressure and a power supply (F igure 2-2). In our a the cathode consists of the material to be sampled the anode i the glo di charg chamber housing and the power supply is operated in either a continuou D RF or pulsed DC mode. When a negative voltage is applied to the cathode el ctron rel a d from the surface or spontaneously emitted free electrons undergo accel rati n t ard the relatively positive anode chamber housing During thi ace leration l ctr n ollid with fill ga atoms producing positive gas ions. The e ion exp ri nee th attracti n f th cathod at hi g h negativ potential and ar accel rat d to ard it an d ntuall b mbard th cathode ampl If the ga i n ha no ugh n r gy (gr at r than th bindin 0 n rgy f th at m in the matrix) it will au fr m th athod including n utral atom and m and l ctr n h l ctr n will c ntinu lt ag i di h r g h g l u t aining th fl di ch rg can p rat t pr and m 1 cu l ph t n ward th an d hil th er atin l f u tain d g l ur fr m 0.01 0 T rr d p ndin

PAGE 21

Power Modes RF Pulsed DC DC :t, :::s 0 m 13 Figure 2-2. A simple glow discharge device showing the various power modes

PAGE 22

14 on the di charge gas and ha average currents in the low milliampere range in D mode and less than 1 mA in pulsed mode (depending on duty cycle) Regions of the glow discharge A glow discharge consists of 7 distinct regions: the Aston dark space (AD ) the cathode layer (also called cathode glow (CG)) the cathode dark space (CD ) the negative glow (NG) the Faraday dark space (FDS) the positive column (P ) the anod dark space (ANDS) and the anode glow (AG). The research in thi dissertation requir operation in the abnormal glow discharge range as described above. Atomic mas spectrometry and other analytical spectroscopies probe the collision rich negative glo for analytical information about the sample. The close proximity of the cathod and the anode result in the disappearance of the faraday dark space the positive column the anod dark space and the anode glow particularly in M wh n cathode to anod distanc are less than 1 cm When the cathode and anode ar brought clo r and tog ther the cathod dark pace remains unchanged in thickn column hrink s until it and th Faraday dark spac (FD ) are con urn d. Figm 2-3 i a r e pr ntation f th glow discharge and it di tinct region and th ir 1 trical charact ri tic igur 2-4 how th abnormal di charg and th r 10n that ar m t imp rtant fi r ma p ctr m try and how th negativ glow ill b prob d in thi di rtati n. ar t th cath d urfa th ton dark pac hich i a lar g 1 n nlumin u r g i n du t th la k f c lli i n that cau citati n t ccur. hi i a n

PAGE 23

CG CDS ~l NG Luminou Intensity s ,. j~ Potential Field Space Charge Density Current Density Ve / de .............. D< --n+ ~nADS = Aston dark space CG = Cathode glow CDS = Cathode dark space NG= Negative glow PC = Positive Column ANDS = Anode dark space AG= Anode glow PC I r .. V Ex --+ J -+: J j+ jFigure 2-3 The regions of the normal ,glow discharge and the corresponding electrical properties LKef. 30] 15 .... 1 """'

PAGE 24

16 Sampler (Anode) Ion Skimmer To Mass Spectrometer Figure 2-4. The abnormal glow discharge setup for mass spectrometric detection

PAGE 25

1 7 area of negative space charge due to the secondary emission of electron from ur fa c collisions of bombarding gas ions a well as the presence of electron r e l e a s ed up n neutralization of gas ions impinging on the target. Regardles s the electron s ar e accelerated away and current is primarily carried by positive ions in thi s region A hort distance later the space charge becomes positive and remains that way through the cathode dark space (see Figure 2-3). The cathode glow or cathode layer separates the Aston dark space from the cathode dark space. The low luminosity is due to the recombination of incident ions with slow electrons. The cathode dark space is the next zone of little luminosity. The negative glow is an electrically neutral zone ; therefore the voltage difference between the cathode and the anode is reduced in the small region between the cathode and the negative glow this being the cathode dark space. This region is often called the cathode fall region du e to the decrease of the voltage over this small area. The negative glow is the region with the most analytically useful information. It is a collision-rich field-free environment in which excitation and ionization processes occur. It is this region in which glow di charge ionization studies occur The relative luminosity of these regions is determined by th radiative relaxation processes occurring within that region. Hence a dark re g ion ha few excitation processes while a glow region has many. The lack of luminosity in th e dark space is due to the small number of electron-atom collisions The electrons hav been accelerated by the cathode potential and the efficiency with which di char g g a ionization occurs decrea es. Hereafter all discu sion will pertain only to th u o f th abnormal glow discharge and mainly its negative glow region.

PAGE 26

18 Glow Discharge Processe The mechanisms that lead to atomization excitation and ionization in a glow discharge occur within discrete zones of the discharge structure. Ionization of discharge gas occurs in both the negative glow and in the cathode dark space. Atomization occur at the surface of the cathode a result of the ionization and subsequent bombardm nt of the discharge gas. Excitation and ionization both occur in the field free negati e glow a a result of collisions with electrons ions or excited atoms with sufficient energ igur 2 5 is a diagram summarizing the processes occurring in the glow di charge and the zones with which they occur. Glow Discharge Sputtering The approach or acceleration of an ion toward a cathode urfac can l ad to 1 of 5 possible interactions called collective l y the collision ca cade: l) th ion may be r fleet d or scattered away 2) the impact of the ion may cause the rel ea of an el ctron fr m th urface (n ce sary for maintenance of the selfustaining glow di charg ) 3 th i n can bury or implant itself permanently in the crystal lattice 4) th 10n may cau r arrangement of th top w layer ( call d radiati n damag ) and 5) th impa t f th i n will j ct a target atom. Thi la ton i known a sputt ring and is th m t imp rtant analytically For any ion b mbardm nt v nt mor than on f th int racti n ccur. t f int r ti n ab v imply that an i n 1 th nl p . impingin g n th urfa wh n in fact ith ran 1 n ran utral p c1 can d I It i

PAGE 27

19 + Figure 2-5. Summary of GD processes

PAGE 28

20 an ion in the mod I because it can be accelerated across the dark spac ia the attraction of th negative cathode whereas explaining how a neutral speci s is attract d to th surface is a good bit more difficult. It is believed that the Auger emi sion of an lectron from the surface neutralizes a majority of the incident ions just prior to striking the surface. With all that can occur during the collision cascade it is not surpri ing that sputter ejection comprises only 1 % of all the energy transferred at the surface of the cathode 30 Sputtering occurs when an ionized ( or neutralized) gas atom bombard th urfa of the cathode embedding itself in the lattice of the solid and then transferring enough energy to create a shock wave cascading through the lattice. This shock wave ha b n compared to the break in the opening of a game of billiards except in three dimen ion 30 (See Figure 2-6). Three species representative of the sample can be putt red upon bombardment ; a sputtered ion, a sputtered neutral atom or a polyatomic clust r. In th ca e of glow di charges the negative voltage on the cathode will imm diatel y attract th putt red ion for red position onto th cathode surface leavin g only a cl ud f putt e r d n e utral sp ci s to diffu without any field effi t aero th cath d dark p ac a nd into th n g ativ g low Th putt r d ion produc d up n bombardm nt o f th ga 1 n a r call d condary i n and ar th p c1 u d r an a l i in th urfa t chniq u call d c ndar y i n ma p ctr m tr r IM int r t D r p tr cop1c purp durin g th putt rin g proc ar th pnmar p putt r d at m ar f that 1 a th ur fa f atom fr m n m dium ( a lid ur fa c ) a nd t h ub qu nt di luti n int a g a u m dium g i g l di har g and

PAGE 29

Cathode Surface Figure 2-6. Schematic of sputtering in a glow discharge based on a billiards break 2 1

PAGE 30

22 particularly GDM a relati e freedom from matrix effects which often plague other urface and solid anal ys is techniques While not completely true theor ticall y th atomization and ionization processes are separate in the glow di charge pla ma and ensure the lack of chemical memory for the sputtered species once the y lea e th urfac e of the cathode. 20 23 puttering yield is the primar y measure of sputter efficienc in a g lo di harg and is usually measured as the number of atoms sputtered per incident ion. Two equations exist in order to determine this efficiency. The first developed b y Si g mund 32 takes advantage of the energy / mass transfer term (bold in the equation). In E quation 2 -1 the sputter yield (S) is given by: S = (3/ 4)n 2(a) (4M1M2/(M1 + M2)2] (E/U o ) (E q .2 -1 ) where a is a function of the relative masses and the angle of incidence of th incomin g ion M 1 and M 2 are the respective masses (g) of the ion and th putt r d atom i th incident ion energy ( e V) and U 0 is the surface binding energy that mu t b o ercom fo r puttering to occur ( V) xp rimentally Boumans 33 d v lop e d an xpression for th putter i ld ( ) r pr nt d by quation 2 2: = 9 .6 X 10 4 ( W / M t* t) q 2--) wh r W i th weight lo (g) M i th r lativ at mic m a f th putt r (g) t i th i n c urr nt (A) an d t i th putt ring tim ( ). Th b mbardin g i n urr nt t i r l at d t th t t I di c h arg c urr nt i by quati n 2-3: = i i I Yi) q 2)

PAGE 31

23 h r Yi 1 th number of condary el ctron r lea d on th a ra g b y n 1 0 n h following ub ections will d tail the parameters used in th two equ a ti n 1 Ma of ,puttering Jon One might assume that as you increase the mass of the sputtering ion that the putt r i ld would increase because more momentum can be delivered to the surface Th mass transfer term predicts that sputtering is dependent upon both the sputtering ion and the target atom and that sputtering efficiency will reach a maximum when the value of M i/ M 2 is unity. This would make Ar + an ideal sputter ion for the first row transition metals and Kr + for the second row. In Sigmund s theory (Equation 2-1) the alpha (a) factor offsets the mass transfer term such that the Ar ion is a better overall sputtering agent than other inert gases such as krypton and xenon even though these two heavier gases would deliver more kinetic energy and momentum to the surface under similar conditions The mass of the sputtering ion is critical to this work. The lighter mas of helium will be an important factor as a sputtering agent in GDMS. 2 Angle of Incident Jon Bombardment The collision cascade describes a variety of sputtering ionurface interaction that can occur and it makes en e that sputtering would be dependent up n the angl at which the gas ion bombard d the surface thus maximizing momentum and en rg trans r to the top few layer of the surfac Figure 2-7 3 4 how th f ct of angl f incid nc in ion b am putt ring on sputt r yield. It hould b not d that i n b am

PAGE 32

-... 2,5 t-------+-----......f--L---~ Eo= 1.05 l
PAGE 33

25 puttering i much diffi rent than glow di charg puttering and i u d h r mean to explain sputt ring in the glow discharg tern Figur e 2-8 ho t p ibl paths for an incident ion in a glow discharge upon application of the di charg e olta g One argument for th incident nature of ions is illu trated in the picture of a glow discharge sputtered surface (Figure 2-9 3 5 ) Microscopic cone-like spire have form e d on the surface of the cathode. The most likely way formation of uch tructures could occur is for the impinging ion ( or atom) to have a bombardment angle 90 degree to that o f th urface proper such that holes are dug into the surface. This means that regardles of the trajectory of an incident ion in the region between the cathode surface and the negative glow the field attraction interacts strongly with the ions that any effect of its initial trajectory is eliminated So while ion bombardment studies may indicate relative putter patterns in a vacuum they are not necessarily indicative of what occurs in a low pressure di charge. 3 Incident Ion Energy Th energy of the bombarding ion also plays an important role in putter y ield. igmund theory predicts that the sputter yield of a given target material i propo11i nal to the bombarding ion nergy. Carter and Colligan have hown that proportionalit o v r only a mall ion n rgy range. 36 Figures 2-1 0a 37 how th ef ct of l incid nt 1011 nergy 011 putt r yield for a Cu target with 11obl ga ion It ha b 11 pr di t d that glow di charg 10n bombard th surface with en rgi that li in thi li11 ar r g i n f th curve. Th putt r yi ld line in the figure how a rang of pr porti 11 lit fr m b ut

PAGE 34

~ 0 . . 26 Figure 2-8. Possible paths for incident ion (A). X = No voltage effect on trajectory. Y = Voltage controls trajectory

PAGE 35

2 7 Figure 2-9. A copper target sputtered by a 1 torr argon discharge showing microscopic cone formation [Ref. 35]

PAGE 36

3 0 g e 2-0 0 0 "O "ii :;. 0' C -~ "5 a. 1 V) 0 Threshold energy (ET) Linear region A 500 1000 Ion energy (eV) 28 Figure 2-10. a) The effect of low incident ion energy on sputter yield for a copper target and argon ions. Triangle and circles are various data points[Ref. 36]

PAGE 37

29 100 Vt 750 e V. Th range of thi proportionality wi11 lik l y chang ba d n th 1 z of th incid nt i n. Figur 2-1 0b 36 shows a high r n rgy range for sputt r yi ld r u incident ion nergy A ion nergies get much high r ion implantation occur and a maximum of energy tran ferred to the top few layers of the lattic e i reach d and th number of sputtered atom r aches a maximum. 4. The Target Material The target material is the one parameter in glow discharge sputtering that i th most variable, as there are many matrices that can be analyzed. Therefore multi-elem nt analysis can be affected by the matrix. The Sigmund theory suggests that sputter y ield i dependent on the ion to atom mass ratio. Wehner 38 believed that other processe ar involved aside from just the mass relationship. Having produced charts of the putt r relationship with a 400 e V Ar + gun he found a dependence of sputter y i ld on dh 11 filling. The explanation says that as the d-shells are filled with electrons th atomi radii decrease thereby increasing the atomic density in the matrix. The incr a d atomi density prevents the bombarding ion from implanting or penetrating too d epl into th matrix The cry tallographic orientation of the matrix mo t like! do n t chang ith th incr a in d n ity as th orientation of th at m do not chang onl th ir i z The ma x imum amount of en rgy i then tran fi rr d to the top f th matri , incr a ing putt ring Th difference in putt r yi Id b tw n iron and opp r i r 0 u d in thi manner. Th ma of copp r i gr at r than that f iron and ace rdin g t th m transfer t rm th putter yi ld of iron would b lar g r th n c pp r if arg n r u d

PAGE 38

s 2 15 ... 0 ......... O? E 0 ..., 10 5 0 C per ~.1 e & Bruce 1 6 0 / / /) ... 1<1 0 I/ 00 1 "' Ar / :-, -'v .,, 0 / .,,-C V / _. :J I ..... '] -, ,/ .., 1e ~ N 1 2 30 50 '"'O 70 Incident Ion Energy (ke V) Figure 2-10. b) The effect of hi_gh incident ion energy on sputter yield for various gases [Ref. 3 7] 30 k V

PAGE 39

3 1 the bombarding ga h putter yield of iron i l b cau copp r ha m r dh 11 l ctron that ultimat ly increa e the density of th matrix. Wehner pl t r arg n and h lium ar h wn in F i g ur 2-11 a and b The differ e nt sputt r y ield among t l e m nt ma cau a probl m when analyzing alloy Preferential sputtering 39 will l ea d to an nrichment of the low r sputter yield material at the surface. Due to the inherent putterin g charact ri tic of the glow discharge an eventual steady state will b e tabli h d and the gas phase population will be repre entative of the bulk sample. Glow Discharge Excitation/Ionization For analytical glow discharges the primary concern is the excitation and ionization of the sputtered species, and one would hope that these were the only processes that occurred in the plasma However ionization of the plasma gas and it impurities and interferences from sample impurities can make analysis difficult. It i important to understand the processes that maximize analyte excitation and ioni zat ion and minimize discharge gas excitation and ionization First a brief discussion of th e natme of colli ion in the glow discharge is given below. ollisions in Gases lli i n in a g low di charge are th m t th at c m and ar r li ed upon to i ni z putt r e d peci in th n ga ti g l w r g ion f th pla ma are definitions taken from Howatson: 24 Ela tic olli ion ar impl m chani kinetic collisions in which the energy i alway kin ticJn e la tic colli ion ar th 111

PAGE 40

2 6,------------+---=-~------~ 2 -4 Argon 2 1 2 1 +J ro 1 :9 4 0 1 2 i-.. lO o -sl.AL 0 6 ~ Be 0-4r s, 0 2~ C Cu Cr Ni Co Fe .Ge V Ti Zr Pd \0 4 ~ He l ium 8 0 ~ 22r- Be Au cd 1 20~ ] 0 I 0-14 Pl I 0 12 I 5. 0 10 u! r:ri o-os Th1 0 06 .c 0 04 I 0 02f'---:-::--~:-~:--L~-1--....1.__..J__1__J I 0 10 20 30 40 50 60 70 80 90 0 10 Atomic number Cu Cr Aq 20 N i Co Fe V Pd Rh Zr Ru Mo Nb 30 40 so 60 Atomic number Figure 2-11. Wehner sputter yield plots for a) argon and b) helium at 400 eV incident ion energy [Ref.39] 70 80 w N

PAGE 41

33 which ome of the n rgy of the colli ion i tran ferr d into int rnal nerg y f th particl truck thu producing a r duction of kinetic energy of th y tern la tic colli i n c a n b compared to the colli ion of two billiard balls much like that shown in Figur 2-6 for puttering. Figure 2-12 30 hows schematically the two collision pos ibilities of an el ctron and an atom. The electrons in orbit around an atom give an approximate uniform field about the nucleus and when an electron approaches, it can be repelled or deflected in a way that is similar to the deflections of two billiard balls or two ideal elastic spheres. The fundamentals for perfectly elastic collisions comes from the Maxwell-Boltzmann kinetic theory in which each atom has a definite radius and cross-sectional area uch that the probability of the two atoms colliding can be calculated. 40 However in such a collision changes in internal energy are not considered. For example if an electron with energy less than the ionization potential of helium were to strike a helium atom at zero energy then the helium atom would have to sustain the hock of the electron colli ion (via kinetic energy) because there is no mechanism to transfer that energy (via potential energy) under the theory of elastic collisions Now consider an electron with exactly the amount of energy needed tor mo an ctron (I of the atom). In order for the atom to be conv rt d into a po iti 1 n th l ctron mu t ha nough n rgy to not only r move the electron but al o m energy to remov the lectron and to compensate for recoil from the atom Th r for lectron with energi that don t exceed that of the ionization potential of th at m th are striking will result in an ela tic collision.

PAGE 42

34 b. Figure 2-12. a) Elastic collision of an electron and an atom. b) Electron ionization [Ref. 30)

PAGE 43

35 In th n gativ glow r gion of th analytical glow di charg c lli i n that xc it and ioni ze are inelastic. Inela tic collision fall into two categori : colli ion of th fir t kind and col Ii ion of the econd kind. Colli ions in the negativ e g lo w pla ma with lectron of various kinetic energies are collisions of the first kind Colli ion of the econd kind occur when atoms and massive particles collide resulting in potential energy transfer. Table 2-1 40 lists the most likely excitation and ionization phenomena in the glow discharge. A nwnber of factors determine the relative role of these processes including discharge conditions and source geometry. Sputtered Species Excitation Atoms that have diffused across the dark space into the negative glow face a series of inelastic collisions by electrons, metastable atoms and other ions (gas and sputtered). The glow discharge plasma is an electrically neutral plasma with a high population of electrons. Electrons in the negative glow are very low in energy (F igure 2 13 4 1) as the Electron Energy Distribution Functions show under various condition Hence excitation mechanisms that involve energy transfer to levels at or less than 3-4 e V ( above ground state) will be preferentially populated over tho e that exi t at higher energies. 41 Transitions from non-ground state energy lev ls to higher I v l do occur but without the populations a most of those that are ground tater onance tran iti n These ground state transitions are mo tly r pon ibl for the atomic tran ition th t c ur in the UV-VIS (A = 200700 nm) regions of the lectromagn tic p ctrw11 and ar responsible for the primary emis ion from th n gativ glow.

PAGE 44

I. Primary excitation/ionization processes A. Electron impact M + e(fast) M + e(slow)/M+ + 2e B. Penning collisions Il. Secondary processes A. Charge transfer 1. Nonsymmetric M + Ar!-+ M* + Ar 0 /~ + Ar 0 +e. Ar++ M-+ M+ (M+.) + Ar 0 2. Symmetric ( r~onance) x+ (fast) + X(slow) --+ x_-O(fast) + x+ (slow) 3. Dissociative B. Associative ionization Ar!+ M 0 -+ ArM+ + C. Photon-induced excitation/ioniz.ation D. Cumulative ioni7.ation 36 Table 2-1. Glow discharge excitation and ionization phenomena

PAGE 45

;:; c u: 0 w w u. C w w 60 -------:---------------, ...... a 50 40 'o<:Poo 0 0 Ao-. 0 oe o o ~.. o ... 0 0.. ... ...,,.,_ 0 a...--..Cb o 0 0 Do 30 20 10 D Q:J!e o ..L--..------~Mltaliiieu.....--....-~ 0 0 1 0 2.0 3 0 4.0 Electron energy, eV 60 -------,---------------, 00 Oo 50 40 30 20 10 0 00 00 0 00 o .. o .. : 0 a::,Q:J'1 i:! D .. 0 ~ D D .... 0 D O 0 D ri_ 0 b --c 0 C1tJ !ao-~""'-...,....,_~ 0 0 0 0 0 0 0 0 1.0 2 0 3 0 4 0 Electron energy, eV SmA 8mA l0mA 12mA 15mA 18 mA 10mA 12mA 15mA 1BmA 21 mA Figure 2-13. Electron energy distribution function (EEDF) for the negative glow region [Ref. 41] 37

PAGE 46

38 puttered atoms are not the only pecies that are excited in the negati e glo citation of ionized sputtered atoms can also occur. As sho n in the li t of proc e th collision of an electron and a gas atom can produce am ta table gas atom that ha nough energy to ionize and or excite sputtered atoms. These ionized putt red atom can undergo similar collision with electrons and metastables that produce excited stat ions The collision must overcome the energy required to rai e an electron abo the ground state ion energy (IE) to the next available energy level. Sputtered Species Ionization As previously mentioned the quantification of an total ion number den ity ia ma s spectrometry is much easier than by atomic emission becaus r lati I fe isotopes exist for each element; whereas multiple transitions r present th xcitation f the ion population. Therefore fewer measurements are needed in ma pectrom tr to anal y ze trace elements. The energy of such ions i not entir ly important c pt wh n ampl focu ing for ma p ctrometry is con idered. For quantitati e analy i in M a ma x imi z ation of analyt ionization while maintaining matri r pr ntation incr a th informing power of the technique A Tabl 2-1 illu trat everal c lli ion m h a ni m x i t r anal y t ioni z ation Th mo t lik ly putt r d p i ni za ti n m h a ni m ar b y I tr n i ni z ation and P nning-t F r ctr n I ni z ati n th l ctr n mu t ha nough n r g t rn z th a n a l y t at m H l ctr n 11 rg di tribution functi 11 Fi g ur 2 -13 4 1 ) h th a t tr n h 11 u g h n r g t i 11i z m tm tal ith I.P in th 5-10

PAGE 47

3 9 V range. El ctron and ion den ities have be e n mea ured in a RF g lo w di char g pla m a based on electrons in the 0-5 eV range. 3 0 The density measured 10 11 cm 3 i s 3-4 ord e r of magnitude higher that electron densities measured in the 5-10 e V range 30 Th e lo we r electron population in the 5-10 e V range would produce very low relative analyt e ion populations compared to what have been typically measured in the negative glow o f argon plasmas. 3 0 Vieth and Huneke 42 have studied the production of an electron via the Lovett rate model in which a 7.1-7.6 eV electron is produced that could ionize man y metals. This model based on electron impact and three-body collisions proposes a double Penning collision as shown below in Equation 2-4. Ar + + Ar 0 + e(7.1 7.6 eV) (Eq. 2-4) No experimental evidence has been provided to assume that this mechanism occurs for analyte ionization as the probability of this process is very low; even though Penning type collision (inelastic with transfer of potential energy) are assumed in many experiments as the primary analyte ionization mechanism in the negative glow Table 2-2 lists the low lying metastable energies of various noble gases. The energies are well above the I.P .s of most of the solid periodic table and for noble gases from argon to krypton the energy is below that of most gas impurity I.P.s. Metastable ga atom ar e formed when inelastic collisions raise the energy to this long lived excit d tate (11 5 and 11.7 eV for argon). Quantitative estimations of Penning ionization in a glow di charg ha b n performed and have given number anywhere from 40-95 % d p ndin g on di ch a r g conditions. 4 3 A variety of reasons exi t to support Penning ioni z ation a th primar

PAGE 48

Gas Spec IE (eV) O2(dimer) 12.0697 H2O 12.621 OH 13.017 0 13.618 H 13.598 Ni 14.53 Ar2(dimer) 14.5 N2(dimer) 15.581 Ar 15. 759 He2(dimer) 22.22 He 24.587 Ar met 11.5 11. 7 He met 19.8 20.6 Table 2-2. Gas species ionization energies and helium and argon metastable atom energies 40

PAGE 49

4 1 m chani m of putt r d atom Mas pectra from argon di charg ha analyt 1 0n ignal higher than tho e of th di charge ga spec1 even though the ga p c1 ar m much higher quantitie than the sputtered species 44 This rea on i s applicable to an argon D di charge and not a helium DC discharge because the energies of th e helium m ta table are higher. It s been postulated that preferential ionization of the sputtered pecie is the reason for this observation. The metastable energy levels of argon are hi gh enough to ionize a majority of the elements in the periodic table whereas the metastabl e energie of argon lie below the ionization potential of any impurities from air leaks. The discrimination in energy is one of the main reasons why argon has become the primar y discharge gas in most analytical discharge mass spectrometric instruments. Chapters 3 and 4 will discuss the role of metastable energies on analyte and gas species ionization further relative to the use of helium as discharge gas. The best evidence for the Penning ionization mechanism arises from laser depopulation studies in which lower ion signals were observed for species with ioni zation energ1 les than the metastable level of argon at 11. 72 e V while those with higher nerg were unaffected. 4 5 The same result was observed for a neon discharge except that the p ec ie with ionization energy above 16.71 eV were unaffected. 46 49 Experimental m thod of determining the role of Penning ionization includin g varying th di charg conditions 46 4 9 and quenching the metastable states with m than 46 49 have upport d th claim that Penning ionization plays a prominent role in putter d p c.1 ioni za tion. In hapter 3 a discussion of the pulsed glow di charg will how nm r d minatin 0

PAGE 50

id nc for m ta tabl ioni zat ion based on the nature of th pul d di charg Other condary excitation/ionization proces es 42 nt that occur in th ymmetric charge exchange is thought to be limit d to the re g ion closest to th cathode such that argon ions that are normally accelerated across th dark spac collid with neutral atoms close to the surface. The incident argon ion continue on it path to the cathode where it contributes to sputtering. Asymmetric charge exchan e al o contribut to ionization in the negative glow. These mechanism are ummari z d a eq u ati n in Table 2 -1 and shown in Figure 2-14. 30 The contribution of each charge exc han ge proc aries depending on the ion source. This i particularly true for hollow cathod and Grimm-type glow discharges so where asymmetric charge exchang i thought to pla a greater role than symmet ric charge exchange. As d scrib d b efor th ioni zati n provid d by Penning-type collisions i preferred, becau e of the tron 0 uniformit f th ionization am ng lements (R F ) and the l ect ivi ty again t backgr w1d ga p A ocialive ionization i the primary m chani m for th formation f pol at that c ntain a di charg ga an d putt r d p ci hi pro u u a ll tak plac i thin th n gati gl wand it contributi n in the ma p ctrum i 111 m pr p rti n t th p pulati n f th putt r d p ci F r a typical diod di harg th r l ati f th i ni ati nm chani m i unc rt in pt that P nning i ni ati n app ar t b th e prim ry i ni ati n m chan i m fi r th putt r d p di h rg param t r

PAGE 51

Electrode !'!.6"!..~tf..=1 on strikes neutral, ::::~::::::::.:: charge exchanges .. Ce =!.~~~ ~= ~.~.!it, 0 + I j ..... . ................... ---~--0 0 .. F f t ................. orms as trat and ton =-f!. Sheath ... Glow Region 43 Ion enters dark space and accelerates Figure 2-14. Charge exchange processes [Ref 30]

PAGE 52

44 chang o do the magnitude of these mechanism ; whereas Penning-t pe collisional ionization is considered to be the favored proce s regardles of the discharge condition Types of Glow Discharge Device Pin-type / Disc or coaxial (Figure 2 1 Sa): This device is the most commonly u ed for trac analysi of metal and environmenta l sample (usually non conducting oil ) her a conducting binder is required. In the ca e of the pin type the eas of u ing the chamber as the anode makes it the most versatile of these devices. The pin are 1-2 mm in diameter with about 5 mm in l ength exposed to the plasma. The high den ity of atom and ions above the tip of the pin or disc sample have made this device common among t commercial glow discharge devices one example i the VG 9000 Double Focu ing Ma pectrometer. The disc samples are mounted using a holding de ice that po on! th flat front surface to the plasma, via shielding that surrow1d the diam ter of th top urfac (typically 4 mm). The work describ d here u th e mall flat di c for metal analy i Planar diod (Figure 2-1 Sb): This is th simpl t of the glo di harg d 1c not u d ry much. oburn u d an RF power upply with a planar diod for m a ur m nt via ma p ctrom try of ion p pulati n in a putter d po ition it i chamb r. 1 8 Thi i al th ba i fi r th dev lopment and pr du ti n f th rimm-t p g l w di charg ourc rimm Di charg Lamp (Figur 2-1 Sc): Th rimm-typ D d 1c a fir t intr due d b y rimm in 1 8 1 7 and ha gain din p pul rit m ti b cau fth a f ampl

PAGE 53

45 int rchange and th ability to analyze thin film 5 1 The Grimm di charge i oft n call d an obstructed di charge because the anode i positioned within the dark space di s tance of the cathode. This limit any puttering to the ar a within the anode s hap e It ha b come v r y popular for bulk analysis. Similar sources that resemble the Grimm so urc e ha ve been constructed for use in atomic emission and mass spectrometries. 30 Hollow Cathode (Figure 2-l 5d): The hollow cathode lamp is probabl y the best known and one of the most widely used GD devices It is commonl y used in atomic absorption emission and fluorescence studies. This device is characterized by low voltages and higher currents compared with other glow discharge devices due to the constricted natur e of the plasma within the cathode and the static nature of the discharge gas. Called the hollow cathode effect," the plasma constriction produces an increase in the radiation emitted from the negative glow making this source attractive for optical spectroscopy. 2 Hollow Cathode Plume (Figure 2 -l 5e 5 3 ): The hollow cathode plume was developed b y the Harrison group, and was primarily used for atomic emission studies. The plasma wa formed by the constriction of a hollow cathode discharge to a 1.5 mm orifice in th e ba e of a normal hollow cathode. Sample cathodes were machined to fit in the ba e of grap hit cylinders. At a certain pressure the plasma plume would form through the top nd of the cylindrical hole Typical argon pressure was about 1 Torr and current b tw n 50 and 200 mA at 1000 V were used.

PAGE 54

a. b. C. d. MHV connector C A THOD E Silicone fiberglass tubing \ MACOR probe tip I Sample pin High-vacuum f eedl h ru DISCHARGE!CHAMB E R 46 A ANO DE to p u mp 2 02mmthic'k teflon sheath I D I SCHARG E CHAM BE R CA T HODE, j I l I \ I --~'-t to pump Figure 2-15. GD devic es a) co a xial cathode b) planar diode c) Grimm di s cha rge lamp d) hollow cathode lamp

PAGE 55

e. w ::E =.l _j o_
PAGE 56

CHAPTER 3 XPERIMENTAL CONSIDERATION Powering the Glow Discharge Dir ct Current (DC) DC i the most familiar and most widely used method of powering glow discharge plasmas because of the availability of inexpensive stabl pow r supp lie that are very easy to use The direct current glow discharge plasma is a self-cl aning pla ma in which little sample preparation is required before discharge initiation Elemental analy is in the DC mode usually consists of a prebum period in which contaminant from the sample surface are sputtered off such as oil from the users hands oxides from th air and any non-uniformity of the sample surface. The DC plasma can be compared to an 1 ctroch mical cell in which the cathode material is eroded or di olv d into a oluti n. In th ca e of an argon glow discharge the atoms di olv into what i ntiall an ar g on ga -pha lution 10n and meta tables have b n xt n i el m d 1 d b B g a rt and ijb l 54 56 and hav hown a good correlation with p rim ntal data Th ir r ult h w that m ta tab! ni z ation and l ctron impact i ni z ati n ar th primary m chani m fi r analyt i ni z ation of whi h 1 tr n i ni z ati n an ntribut up l 40 % In th ir m d I fa t arg n atom ar al con id r d t b a putt ring ag nt hi h n a r m t imp rtant fi r th anal t. h m t imp rtant atur f th g l 48

PAGE 57

4 di charg i it ability to pro ide a table cone ntration of atom and ion fi r an a l y i th ad anta 0 e being that in mo t GDM in trum nt the analy z r alwa y ampl th am region of the pla ma leading to higher preci ion and high r accurac For DM chemi t the DC glow di charge i an excellent ion source because of it 100 % dut y cycle which with the appropriate choice of mass spectromet r is a great advantage in elem ntal analysis DC glow discharges have two disadvantages; one is low power compared to t chniques such as inductively coupled plasma electrothermal vaporization and laser induced plasmas ; and the other is the need for an electrically conduction material. The lack of a conductivity requirement for these other techniques allows universal samplin g of any sample or matrix giving these techniques a great advantage over the DC glo discharge which for now is limited to conducting metal matrices In order to sustain a cw-rent in a DC discharge the sample being analyzed must at least have semi-conducting properties and even then accurate analysis is not guaranteed. The conductivity r quirement has not prevented the analysis of non-conducting sample in the DC mode The u e of a conducting matrix binder has been very successfully used in the analy i of non-conducting powders. 57 The trouble faced in these analyses i hetero 0 en ou di tribution of analyt in the matrix. A secondary cathode has al o b d t sampl non-conductive olid A dummy hollow ring cathod i p iti n d ab a non-conductive mface allowing accelerated atom and ion to trik th non-conducti targ t. putt ring of th non-conductiv urfac r ult in th j ction of mat rial into th

PAGE 58

pla ma form d in th middle and above the surface of th conducting ring cathod ffecting excitation and ionization of the non-conducting material. Radiofreguency (RF) 50 With radiofrequency powered glow discharges most concerns about sampl non conductivity are minimized. The RF glow discharge was created by Wehner who proposed the use of a rapidly oscillating voltage that would caus a DC bia at th urfac of the cathode thereby creating a plasma much like that in the D mode. The ability of the RF glow discharge to analyze both conductive and non-conducti e samples make it the most versatile of the sampling modes of the glow discharge. The RF di charg i le affected by surface contamination and reaches a stabilization point faster than the dir ct current mode of operation. 58 The main drawbacks of RF glow discharges include expense and po er fficiency. RF power suppli are significantly more expen i than D pow r uppli and require a matching network so that efficient coupling of the pow r to th ampl ccur Oft n refl cted power can di ipat the ef cti ne rand th analy tu ually tri to minimize or liminate it. Additionall if hi ldin f th urc and ampl i not c mpl t th ourc can act a a radiotran mitt rand di rupt the capabiliti f th r d t cting d v1c uch a ion count r and mi r hann 1 plat 11 ma p ctr m t r d v1c w r putt r rat hav b n r p rt d fi r th RF gl di harg c mpar d t th gl w di charg alth ugh incr a in 1 nizati n ha

PAGE 59

5 1 b nob r d compar d with the DC di charge. 59 60 Ther ar se v ral comp a ri on f th fundam ntal and applied aspects of these two powering m e thod e 58 60 The Micro econd Pulsed Mode Thus far the use of a glow discharge device has been described in a continuou s fashion where voltage and current are maintained throughout the analysis. While th e r e are advantages to such analysis pulsing the glow discharge can create additional advantages for elementa l analysis A pulsed cycle consists of a short on time eg 10 s, followed by a long off time before another 10 s pulse is initiated. In a one second tim e period the frequency of this cycle is 100 -1000 Hz for the work in this dissertation. It is difficult to maintain a plasma at frequencies less than 20 Hz and often instability occurs at frequencies higher than 1000 Hz Sampling the sputtered species in mass spectrometry occurs anywhere from 80-300 s after the pulse period. This delay will be referred to here as the deflection time ", and corresponds to the time between the glow discharge pulse and time that ions are deflected into the time-of-flight mass spectrometer. There ar e several analytical advantages to using the glow discharge in this mode and the y are covered in the next few sections. Pulsed Glow Discharge Ma s Spectrom try Pul ed GDs have been studied by the Harrison group for about 30 ar tartin b mainly with hollow cathode lamps. The goal of the experim nt a t btain hibh r emi sion ignals by decreasing the on time and incr a in g th p ak p r. t th a t tim

PAGE 60

52 pul d period r millisecond long In the la t 6 7 ear th pul ed di char 0 ha moved from the millisecond to the microsecond time regime. variety of technique ha e been u d to probe the microsecond pulsed plasma including atomic emi ion ( ) atomic fluore cence (AF) and mass pectrometry (MS) 6 1 62 Th immatmit of the pulsed plasma in ana l ytical chemistry necessitates some explanation of th fundamental and analytical aspects that distinguish this power mode from the more con entional continuous modes of operation. DC vs. us-pulsed Experimental Consideration The microsecond pulse is characterized by a peak vo lt age of about 2 kV and a peak current of about 200 mA (avg. 0.4 mA at 200 Hz). The D di charg t picall p rates at about 1 kV and 4 mA average current. The high peak current in th pul ed mode causes an instantaneous heating of the urfac follow d b a l ng r la xa ti n p riod In the D mode there i no cooling or off time and ampl o r-h a ting can b a problem cau ing the samp l to melt despite th low average curr nt. Th a rag curr nt in th pul d discharg mod appro imat ly 0.4 m m d v n th u g h th p ak curr nt i 50 tim high r. Th than that in th D er a ra g urr nt r due v rh atin g in th pul d mod and xt nd th p rati n tim ampl rh atin g in th D mod n ed to cut h rt nal hich i h m a n y M in trum nt ha me lin g t m in tall d Th t bilit f th pul d di char ad anta g u fi r bulk and tra nal i With a I n g r p rati n time dia g n tic tudi lik th in thi di rt ti

PAGE 61

53 p rform d. In addition th tim con urning cooling or r e placing fan ov rh e at d a mpl not ne ded in the pul d mod Different instrument are required to operate in the two modes The pow r supplies are different in that an internally pulsing unit is required to create the short tran ient signal desired in the pulsed mode. Leading edge rise times must be fast on the order of 30 ns in order for the pulsing unit to be effective. The maturity of the pul e d discharge is dependent on decreasing these rise times so that shorter pulses can be used with much higher power. There are mass spectrometers that are better suited for the DC mode than for the pulsed mode An ideal MS system for the DC mode might be a Fourier Transform Ion Cyclotron Resonance Mass Spectrometer (FT-ICR-MS) where inter element and isobaric interferences can be eliminated with extremely high resolving power. 63 In the pulsed mode taking advantage of every pulse for mass analysis is advantageous. A mass spectrometer that accepts short transient signals and anal y zes all masses simultaneously would be the best fit. The time-of-flight mass spectrometer (TOFMS) is the right choice. The system offers full elemental coverage and is fa t with a linear dynamic range of about 5 orders of magnitude. A later section will di cu the fundamentals of the TOFMS system and cover the advantages of using it as a detection d vice for the microsecond pulsed mode Sputtering and Ionization in the Microsecond Pul ed GD The sputtering mechanism doesn t change between the continuou and pul d mod Argon ions and fa t argon atoms are till the primary putt ring ab nt

PAGE 62

54 Ho e er th tim r gime and the energy with which puttering occurs do chang Whil th DC mode result in continuous injection of atoms into the plasma the s-pul ed mode suppli s the plasma with atoms in a transient fashion allo ing for certain t mporal ad antage The sputtering ions and atoms will have a greater energy when bombardin g th urface in the pulsed mode as compared to the DC mode owing to th higher in st antaneous power for pulsed operation This lead s to higher instantaneous putt r rat in the pulsed mode as compared to the DC mode In the s-pul ed mode analyt ionization most likely occurs by Penning ionization. Argon m tastable atoms are the dominant ionizing agent existing after the discharge is extinguished. E l ectrons rapidl di ss ipate after the 10 s pulse period and by 100 s (analyte ampling time ) e l ctron ha ve mo t likely thermalized. This means that e l ectron impact make little contribution to analyte ionization unlik e in the DC mode In the DC mode there is always a constant ion flux to ard the cathod urfac and there i a lwa y a constant electron flux away from the cath d e ur fac hich l ad t a dark p ace lar ge r than in the pul ed mod and force th ptimal ampling di tan t b lon g r in the D mode than in th pulsed mode Thi ma c ntribut t th r 1 n din th D v r u -pul ed mod r ma p ctr m 111 th at mic i n ntration i gr at r a t 3 mm a b th a mpl pp d t 7 mm du t th c n ntration gra di nt of atom away from th cath d e urfac

PAGE 63

55 Analytical Advantages of the Direct urrent GD DC glow discharge analysis is a proven technique. The constant ource of atom and ions i reliable and increases the precision and accuracy for trace el mental anal y i It is a good atom generator despite its relatively low power compared with other atomization techniques such as the ICP. The DC discharge is inherently elf cleaning therefore requiring little sample preparation to the surface. This is one of the advantage that can be very useful for samples that readily oxidize. With the constantly applied voltage and continuous bombardment of the sample surface oxides are less likely to form allowing for greater sputtering of the surface. Figures 3-1 a and 3-1 b show a comparison of a DC and a s-pulsed g lo w discharge mass spectra for a tantalum cathode. The s-pulsed spectrum shows a series of oxides beyond the elemental species while the DC spectrum has only an elemental peak. The elemental peak is greater in the DC mode than in the pulsed mode showing that oxides are masking the sputtering of the tantalum elements in the pulsed mode due to their formation during the relaxation period. If TaO formed in the area above the surface then its contribution would be observed in the DC mass spectrum. However it is not observed and it must be assumed that TaO i sputtered off the surface in the s-pulsed mode likely being formed immediately after perturbation of the surface by a voltage pulse DC glow discharge operation is simple requiring few fast electronic to gat signals for collection. For analytical use it has very good sensitivitie for all elem nt and allows detection by multiple analytical techniques (Figure 3-2 6 1 6 2 ). DC glow

PAGE 64

0 35 +ro 0 30 + 0 25 < ,,..-.__ > 0.20 0 rfl 0.15 0 Q.) 0 0.10 0 05 0 00 I I 0 100 200 m/z Figure 3-1. a) DC mass spectrum of tantalum at 1 kV and 1 torr Ar

PAGE 65

> .._,, ...... en c::
PAGE 66

Glow Discharge Spectrometries ATOMIC FLUORESCENCE ATOMIC ABSORPTION hv RIMS = Resonance Ionization Mass Spectrometry OGE = Optogalvanic Energy MASS SPECTROMETRY + + + I CATHODE Power Supply t OGE t ATOMIC EMISS I ON LASER RIMS/OGE Figure 3-2. Glow discharge analytical techniques 58

PAGE 67

5 di charg have become v ry u ful for thin layer analy i providing x c 11 nt d pth r solution with multi-element capabilitie 5 1 Analytical Advantages of the Microsecond Pulsed Mode Atomic Emission Signal Enhancement There is a two-fold advantage in performing AES in a microsecond pulsed GD plasma: first there are instantaneously more atoms produced at the surface than there are in a DC plasma. Secondly the microsecond pulsed GD produces electrons having more energy than do DC GD electrons resulting in greater excitation of atomic states allowing for better detection limits and increasing informing power. Furthermore the wide range of populated states can be used to determine how energetic microsecond pulsed GD electrons are compared with DC and RF GD discharges More energetic electron also lead to a greater population of metastable argon atoms. Metastable atoms are capable of ionizing gaseous metal atoms and have excitation capabilities. These factor re ult in emission signals up to 20X greater in the pulsed versus DC modes for certain element 62 It has also been shown that the signal enhancement experienced by the sputtered atoms is much greater than that experienced by the fill gas. This leads to better signal-to-noise and signal-to-background measurements. Figure 3-3 show th compari on of a D ignal at 1 kV and a -pul d ignal at 2 kV. The mor energetic lectron produced in the pul ed mode lead to an incr a d

PAGE 68

60 1 4 1 2 1 0 0 8 0 6 0 4 0 2 0 0 t----.-----.-----.-----.-----,------............ --,----.-----.-----.-----.----r 120 a 20 40 60 80 100 Mass/Charge Figure 3-3. Glow discharge mass spectra for copper in argon; a) s-pulsed mode, b) DC mode b

PAGE 69

6 1 production f Ar ion availabl for puttering. Ther fi r a gr at r in tantan u population of atom occur during the pulsed period or a gr at r puttered at m d n it at th urface than produced in the DC mode. The more energ tic electron al o rv to create a greater metastable argon atom population increasing the ionization of sputtered pecies. All these factors lead to the 3-4 fold signal enhancement seen in Figure 3-3 for the copper ion. The signals for Sn at 400 ppm in a NIST brass sample are shown in Figure 3-4. Figures 3-3 and 3-4 show that about the same signa l enhancement is obtained in the pulsed mode compared the DC mode for both bulk and trace signals. One concern was that the transient nature of the ion source would cause high noise. However a reduction in noise compared to the DC mode is observed because of the long relaxation of the discharge before sampling occurs. Temporal Considerations in GDMS One of the important temporal aspects of the s-pulsed mode is the ability to eparate the various ionization events that occur during the pulse cycle beginning with the ionization of the discharge gas species by electron ionization. In the 10 pul e p riod electron impact ionizes the discharge gas and its impurities (Figure 3-5). Once the di charge ha been turned off mass spectra show that no ignificant ionization occur for about 100 There is no significanc to 100 tim period ho r during thi p riod el ctron u taining th pla ma current completely di ipat Thi m an that littl or no electron impact ionization of ga pecies or putt r d p c1es occur. ft r th 100 s relaxation p riod other ionization nt uch a charg chang cur ju t pri r t

PAGE 70

62 a 700 0 -_fl_ __J 116 118 12) b an 100 -~~f~.J\1J 0 ,,...._,,,,..116 12) Figure 3-4. Mass spectra of tin at 460 ppm for a) s-pulse b) DC

PAGE 71

en C 600 2 300 C 0 z 0 50 m/z l Figure 3-5. Electron ionization spectrum of argon at 1 torr and 10 s delay time

PAGE 72

64 th ma imum ignal for th puttered species (Figure 3-6). A ociati e ionization i mo tl r ponsible for the obser ed polyatomic and molecular pec1e pr nt in th pla ma and i the re ult of metastable association with puttered peci Penning ionization dominates ionization processes and is considered the primary mechani m for sputtered species ionization at deflection times greater than 100 s (Figure 3 7). 1gur 3-7 shows an ion counting mass spectrum where copper ion ha e saturated th detector and showing the small pr esence of gas species as compared to Figures 3-5 and 3-6. A ociative ionization wi ll l a t as long as there are metastable atoms pre nt in th discharge vo lum e. The separation of e l ectron and Penning ionization part to the temporal attributes of pulsing the g lo w discharge. When the di charg i not pulsed there are contributions in the mass spectrum from all of the ionization mechanisms. Figure 3 -8 is a DC spectrum showing the combination fall th ab ionization processes that occur separately in the s-pulsed mod (Figur 3 5 tlu ugh 37) and simultaneou ly in the DC mode. putt red p cies from th cathode diffu e toward th an d and ar 11tuall d po it d. In thi ca e sputtered pecies diffu int th n gati glo r g1 11 ar x cit d and i ni z d and th n amp! d in th am r gion. In a D gl di char g th n g ati g l r g 1 n 1 alway pr nt and i nization f ga and m tal ccurnn g hi m an that p c1 ntaining th di harge ga and g a impuriti ill I way b in th ma p ctrum. In th -pul d m d th 11 g ati nl fi r 10 p ri d nd i th n tin ui h d du t th rapid I f I ctr n p pulati n h n th pul p ri d nd Durin th h rt n-p ri d 1 n Ii d m ta t bl at m ar fi rm d

PAGE 73

(/) C 800 2 400 C 0 0 + 0 50 M/Z + ;::j u + 0 ;::j u Figure 3-6 As s ociative and Penning ionization for the analysus of copper in argon at 90 s delay time

PAGE 74

2000. (/) C Q,) +-' C 1000 0 I 0 + ;:::3 u + 0 + 0 ;:::3 u I I I I i~ II. I I 50 m/z Figure 3-7. Penning ionization of copper atoms in argon for the s-pulsed GD at a delay time of 140 s A I

PAGE 75

30000 20000 en ...... C ::J 0 () 10000 0 + 0 50 + ::::3 u M/Z + i u 100 Figure 3-8. DC GD mass spectrum of a copper sample in argon

PAGE 76

68 and b come the ionizing agent after the pulse period ends. These metastables (in argon) ha enough energy to ionize the metal species but not the discharge gas or impurity pecies including those molecular in nature This results in spectra dominated by the di charge gas species at early sampling times and dominated by sputtered species ion at later deflection times (after 100 s). By selecting an appropriate deflection time reduction but not removal of these gas discharge interferences can be effected Figure 3-9 a and b show the schematic diagram of how the separation of gas and sputtered species is effected in s-pulsed GD-TOFMS. The first figure shows that ga pec1es are sampled prior to sputtered species in the glow discharge cell. Then by using a short or long deflection time a mass spectrum either rich in gas species or rich in analyte species is obtained. Figure 3-1 0a shows the mass spectrum of an aluminum sample. At earl delay time very little aluminum and lots of discharge gas species and impurity ion ar observed By delaying the ejection of the ions into the TOFM th pectrum hown in igure 3-1 Ob can be obtained ; a spectrum consisting mainly of th aluminwn ion. Th r is no x act value to which th deflection time can be et to perfectl maximi z th i g nal. Glow Di charge Time-of-Flight Ma Introduction 1m f-fli g ht ma p ctrom try (T FM ) i a 11 kn wn t chniqu m bi m I cul a r h mi try for th analy i o f larg bi m 1 ul It b g an in thi fi ld m a inl b cau th r t y p f ma uf r d fr m limit d m a ran g Th tim f -fli g ht ha a th r tically w1limit d ma rang l n g a an i n i pr du d and

PAGE 77

Electron Ionization Sputtering Diffusion and Ionization i Cathode i I] b Sampler Figure 3-9. a) Electron ionization, sputtering, diffusion and ionization in the s-pulsed GD Ar+ 0 Cu 0

PAGE 78

Discharge Gas Sampling ._I __..D Sputtered Species Ion Sampling Cathode t b D /.. = . . . Rep :~ i ;; Pulse .. -D . c:::::::lli -c::::a Cu+ Ar + D / ____ -i.. Repelling Pulse / . .. .. .. -D -c::=c::=: cu + Sampler Figure 3-9. b) TOF Sampling. Short delay times sample gas ions. Longer delay times sample analyte ions --.J 0

PAGE 79

0.05 ArH 0.04 HO+ 2 0 03 > >i u, C 0.02 Q) ..., C: 0.01 \ 0.00 0 20 40 60 m/z Figure 3-10. a) TOFMS spectrum of aluminun at 70 s

PAGE 80

> ... U) C: G) ... C: 0.07 + ....... -< 0.06 0.05 0.04 0.03 0.02 0.01 0.00 0 20 40 m/z Figure 3-10. b)TOF delayed (230 s) mass spectrum of aluminum 60 --.J N

PAGE 81

73 an b ace I rat d into th flight tub ma anal y i can b ace mpli h d t that p int d t cting th i n i th limiting fact r in th of a pa1 i i ular a n a l y i DM ha typically b n dominated b quadrupol canmng y tern and doubl cu in g magn tic ctor in trum nt Th pro id high en itivity and fairly rapid analy i and p ak h pping capabilitie allo ing th monitoring of v ral peak in a hort tim p riod. In th la t decad TOFM y terns have been explored for elemental analy i and thi di rtation u e TOFMS for ion analysis and for diagnostic studies namel y to m a ur the effect of di charge gas composition on the s-pulsed GD mass spectra and g n ral GD ph nomena. Time-of-Flight Mass Spectrometry Theory TOFMS works on the principle that if all ions are given the same energy then the time it takes for them to travel a certain distance is determined by their mas Thu equation can be used starting with the kinetic energy equation to calculate the tim it take for an ion to reach the detector at some distance beyond the ion sampling point. Th kinetic energy equation states that KE = l/2mu 2 (Eq. 3-1) wh r m i mass (g) and u i velocity (ml ). By r arrangin g thi quation w can calculate the velocity of an ion in the time-of-flight tube v = (2zeV / m ) 1 12 ( q 3 2)

PAGE 82

74 h r z i th charg on the ion and e V i the kinetic energ th n 1 g1 en up n b mg accel rat d into th TOF. Ion flight time i given by TOF = LID = (m/2zeV) 1 12 D (Eq. 3-3) here D i the distance the ion must travel before being
PAGE 83

75 So u rce Dri ft r egi o n D .1 2 ()--. E = \ /s 1 ::: 0 1 \ Figure 3-11. a) Spatial distribution effects in TOF [Ref 64]

PAGE 84

76 ) Source S Drift region D EB EB c, EB-. EBe\ + 01 E = \ / j_ E=O j_ V Figure 3-11. b)Energy distribution effects in TOF [Ref. 64]

PAGE 85

77 hi h flight path i affect d r ult in noi e accumulation at dif r nt part o f th m a p ctrum r ulting in poorer SIN and detection limit or they imply hit the fli g ht tub wall and are neutralized. Ions in which flight time is affected result in poor mas re olution. Ions closer to the accelerating plate will experience more voltage that tho further away. This results in a small difference in drift velocity, causing the faster ion to reach the detector sooner than the les affected ion. By reducing this spatial distribution an increase in resolving power can be effected. Broadening of the mass peak generall y occur when there are thermal energy variations in the incoming ion packet and / or there ar patial orientation differences caused by either space charge effects or as a re ult of ion optic misalignment. These broadening factors can be reduced by judiciou tuning of the ion optics Performance Limitations and Potential Capabilities The spectral production rate is a function of the slowe t ion. For example with a 2 m flight tube and accelerating voltage of2 kV an ion of mass 500 has a flight tim of 90 This means that the next pulse of ions into the TOF cannot occur until this on ha reached the detector because overlap of the ion packet may occur. Therefor th rep tition rate must not exceed 11 kHz. For the case of the microsecond pul d glo discharge thi factor i of no concern. Operating frequencies are generally not high r than 1 kHz for the -puls d GD source Wiley and Mclaren 65 develop d a TOF y tern that would in r a power over typical TOF sy t ms of th ir tim The hard t fa t r t th

PAGE 86

78 patial di tribution of the ions in the accelerating region. Figure 3-11 a hows that th ion closest to the accelerating plate will experience a greater field than tho e further away creating an energy distribution and affecting different flight times once in the drift region. Wiley and McLaren 6 5 solved this problem by creating and separating the energ and spatial distributions prior to the packet reaching the accelerating region by turning the ions around after a certain flight distance so that the ions moved in a V shape. The reflectron resulted in the distribution of ions being reduced in the turn around (reflectron) region. This is achieved by applying positive and negative voltages in a manner such that faster ions will spend more time in the reflectron than slower ions within a ingle ma packet. Faster ions will require a longer flight path to slow them down and turn them around as compared to slower ions. Thus, as Figure 3-12 shows the ions of each ma packet will reach the detector within a much shorter time and a a tighter ion packet and much reduced velocity distribution. The spatial and velocity distributions could be ultimately minimized by using the reflectron. The TOF instrument u ed in thi di sertation has a mass resolution of 1600-1700 FWHM (full width half max) in th reflectron mode. ofTOFM tt r t al. 66 studi d many a pect of TOFM and li t d ral ad antag tim -of-fli g ht r u all oth r ma p ctr m tric t chniqu f 1. Th T F M i much m r con rvati of th ampl in that ry 1 n 111 an i n pack t i d t ct d

PAGE 87

(b) = / pace focu plane J J I I I I --~d ~.____ __ L 1 ----i 1-+-d 1 [ I 1 ~2 I 4t::r e ~S1 I 0 1 l : ---, : ----L 2 ~d2 ~ i i i I I I i Figure 3-12. Reflectron TOF with s12atial and temporal correction by the reflectron lens [Re t 64] 7 9

PAGE 88

80 2 The op n flight tube ith the absenc of lits pr s nt a er wid ap rture to the ource of ion 3. The TOFM has no fundamental m/z limit other than detectabilit on the range of m/z values analyzable. Additionally the ma s / charge scale can be calibrated accurately based on equations 3-1 2 3 and 6 4. The process of creating an ion packet or bunching can re ult in lo ampl utilization via a low duty cycle and can improve ma re o luti on 5. Very fast electronics are required for detection of tran ient det ctor output Instrument Design GDMS requires the physical in vas ion of the plasma in order to meaure th 10n produced. The extraction of ions from a g low discharge pla ma wou ld b nefit from a d ign that maximizes the number of ions samp l ed. Optical t chniq u es hav ome advantage in that the measur m nt of a photon doe not r quir in a ion of th pla ma. However ma p ctrometry is considered to b a much mor n iti t hniqu for th d t cti n of ion Th e following ction de crib th de ign f th protot D-T in trum nt u d in thi di rtati n. h dir ct in rti n pr b i a g n ri d 1c u d in d lC ntial p a rt r h w n in i g ur -1 n in ulat d tainl th int rfa

PAGE 89

copper insulated solder sample sample macor wire connection holder shield [) D I shrink wrap ceramic electrical (2 layers) Insula tor feedthru stainless steel body Figure 3-13. Direct insertion probe 00 .....

PAGE 90

82 between the po ered lead and the sample. A small handle (not shown in Figure 3-13) i used to couple the power to the back of the probe and is used for easy removal and position manipulation of the sample in the glow discharge chamber. The ample is mounted into a small holding chuck that connects the sample and high voltage lead at the probe tip. The holding chuck is covered with teflon tape so that no surface at high voltage other than the sample surface, is exposed. The samples were machined in the University of Florida Chemistry machine shop so that they could fit into the loading chucks. The top surface of the sample has a 4 mm diameter. The entire probe tip and loading chuck are protected by a Macor shield that minimizes surface exposure to onl the top of the sample surface. The shield was not in contact with any part of the sample or any other high voltage surface to prevent shorting the power supply. The e shi ld were also machined here at the University of Florida. To prevent arcing when ga er ep inside of the shield the loading chuck was wrapped with teflon tape. Discharge Cell Two di charge cell designs have been used in this diss rtation. The fir t a a modified s ix-way cros with a 45 vi wing port that allowed vi wing of the ampl hil in p ration The ourc wa larg and bulky. containing ral vacuw11 int rfac a t minimi z e th amount of contaminant leaks in the igur 3-14 ho a pictur ofthi urc Th coupling of thi ource to th ma pectrorn t r had on a n way during ampl chang to pr t ct th 2 nd and 3 rd acuum t ag n c a r y fi r i n d t cti n. Thi i th primar r a on h an u a uwn

PAGE 91

Figure 3-14. Photograph of direct insertion probe source #1 83

PAGE 92

84 int rlock er needed during sample interchange or source cleaning and maint nanc Hence the need for vacuum protection led to the use of a slide valve that would ensur the acuum integrity while sample interchange or source cleaning was required Thi lide al e was nearl y identical to the one constructed for the interface of the quadrupole GDMS system in this laboratory. The new slide valve interface was designed around the development of a Grimm type ion source. This meant that the six-way cross design could no longer be used. A simple source plate was developed to fit the Grimm-type designed interface and allowed the same operation of the direct insertion probe. The interface design is shown in Figure 3 -1 5. Regardless of design the interface of the probe sample and mass spectrometer is the same for both sources. The only se ere limitation of the modified source is its exposure to air during sample exchange. In our efforts to maintain vacuum integrity and create source simplicity we sacrificed cell expo ure to air and its contaminants. Ion xtraction Ion are extracted from the glow di charge pla ma by moving the ample urfac (and pla ma) wit hin 8 mm of a samp lin g orifice. The an1pling orific i 1 mm in diam t r and i th fir t of two high vacu um int rfac The ourc i appr x imat ly 1 t rr b yond th amp! r th pr ur i at 10 4 torr. ion pa thr u g h th amp lin g rific th rapid dr p 111 pr ur cau a J t pan i n f th pla ma int a fi ld fr r g1 n. A kimm r cone plac d b ond th ampling c n amp J appr x imat ly 1 % f th i n pr due din th xpan 1 nJ t and l ad to th 1 n

PAGE 93

Figure 3-15. Direct insertion probe source #2. Designed for exchangability with a Grimm source interface 85

PAGE 94

86 focu ing 1 n es. The skimmer cone lies at a distance 0.7 cm equal to the mach di c of the jet expan ion of the pla ma. At the mach disc di tance ions haver ach d their maximum velocity. The skimmer cone also serves as the interface between the jet expansion and lens and ion optic region that lie at a pressure of 1 otorr. Figure 3-16 hows a schematic of the source interface and ion optics region. Ion Optics and TOF Sampling The ion optics for the time-of-flight system were constructed at the Univer it of Florida. The sampling plate interface could either be grounded or floating during ion extraction. The choice was dependent on the mode of operation. A higher signal wa obtained for the s-pulsed discharge when the sampler was left floating. Once be yon d the skimmer cone, ions were focused via a set of ion optics including a et of teering plate The ions passed through a plate with a small lit. The plate ha a small pot ntial which minimized the spatial distribution of the ion packet just prior to TOF ampling T h e function of the slit was to reduce the spatial distribution of th ion ju t b for th r ach the deflecting region where ion di tribution can everely affi ctr op ning m asured 10 x 2 mm. The teering plat controlled th po iti n f th i n b an1 and adju t d it into an lliptical cro s ction from it round cro ction all mg a gr ater numb r f ion to b pul ed into the TOF ace l rating gr id and er ating a ti g ht r i n pack t. amp lin g in th y t m occurr d rth g n a ll t th pr ducti n f i n and ion fi cu ing. Th ion fi cu d in th i n ptic filled a r p llin g r g 1 n in whi ha 2 -5 l ng 70-80 V pul d fl ct d ion int th T ace 1 rating

PAGE 95

Sampler Sk i mmer Di r ect Insertion Probe I 1 torr 2x10 4 torr Slit Steering Plates Repeller Jl Faraday F Detector OV 1 ~m G1 -2000V 1 ___ 25 mm I G2 8x10 6 torr 1.5 m Deflectc -Flight Tube Plates icrochanr L!!~~~:r;Plates To Detector Figure 3-16. Schematic of the source interface and ion optics region 00 ...J

PAGE 96

88 grid. A constant 1.8 kV voltage accelerated the ions into the linear portion of th flight tube After a 1.5 m flight ions were detected. For the studies here the ion tra eled into a reflectron lens system in which they were reflected into the reflectron flight tube for an additional 0.97 m flight. At the end of the reflectron flight tube ions were detected b a microchannel plate. Electronics and Timing The temporal aspects of the pulsed GD have been discu sed. To take ad antage of these aspects coordination of the GD pulse and TOF sampling must be optimized. In a DC glow discharge source, the need for fast electronics is minimal because ignal variation and timing are insignificant. Once transient signals are produced e peciall m the microsecond regime the need for fast electronics in order to capture the ignal is critical. Triggering the time-of flight mass spectrometer to sample the ion population at a selected time after the pulse of the glow discharge source is necessary (Figure 3-17). Additionally in order to study the ionization processes outlined abo the dela betw n GD pul and TOF ampling (deflection time) must b adju tabl Th r for fat 1 ctronic were needed. Figure 3-17 how an example of the timing ch m n ar to obtain p ctra rich in analyt with r due d or no contribution from th di harg ga p Cl

PAGE 97

A:. Applied voltage B : Ion profile C : Repeller signal ~!Ous : 100us I t i I i ; Sms -------------~ l ii ,, :i :l ii --------~---.--, Figure 3-17. s-pulsed GD timing scheme l it :: ii 'i I i n jj ii I I 00 \0

PAGE 98

90 Plasma sources can operate anywhere from the millit01T to atmospheric pre sur range. This means that differential pumping is required in order to maintain the necessary pressure to transport the ions from their source to the detector. For this s tern source pressure is about 1 torr seco nd stage at 104 torr and third stage at 106 torr. The vacuum pressure in the 2 nd and 3 rd stages allow the drift of ions from the source to the deflecting region 14 cm and then from the deflecting region to the reflectron detector 2.5 meters in distance. Vacuum integrity is important for a ll MS systems The ability to maintain a high vacuum in the 2 nd and 3 rd stages will determine the 1 t stage or source pressure. Our GD TOFMS system u ses 3 turbo pumps. A 1000 L / s turbo pump (Tubovac 1000 C Ley bold Export PA USA) is connected to the econd stage and a 250 L / s (Varian Vaccum) and a 170 L / s (Var ian Vaccum) turbo pump are connect d to the third stage flight tube. A ll the turbo pumps are backed by 13 1/s mechanical roughing pumps The flow rates correspond to the maximum pwnping capabi li tie for nitr g n at the op rating pressure for the GD. I TO M y t m commonly u an 1 ctron multiplying d 1c uch a a microchann 1 plate a th d t ctor. Many types of oftwar ar u d to obtain p ctra and ntal analy i dual-ch ron microchann 1 plat (M P) ( alil -1 c tr ptic turbrid g M U ) i locat d at th nd of th lin ar e ti n f th fli g ht tub b hind th r fl ctron 1 n and at th nd f th r fl ctron flight tub Th

PAGE 99

linear m d was not u ed in the work presented here. Much better re olution can b obtained in the r flectron mode du to the longer flight di tance of the i n 1 9 9 1 Two detection methods exist that can take full advantage of the multiple x capabilities of the TOF mass analyzer. The first is a real-time averaging mode in which the ions from many pulses are averaged and viewed on an oscilloscope which allow the user to view real-time spectra and analyze the fundamental processes occurring in the glow discharge. This is a running average in which every pulse is averaged to the 99 pulses before it. This mode is also used for ion lens focusing and signal optimization. The second method is an ion counting or accumulation method commonly u ed to measure trace elements in samples. EG&G has developed a Picosecond Time Analyzer (PTA) (6109 Eg&G Ortec) that can accommodate the transient signals that the microsecond pulsed GD-TOFMS system produces. The bulk and trace ma s spectra presented in this dissertation were obtained using these two detection schemes General Procedure and Data Analysis amples were prepared for analysis as described in a previous section. Once the DIP was in the source evacuation by a 13 L / s roughing pump ensues. The discharge ga wa then added and the pressure was measured via a wide range capacitance manom t r (WV-100-2 Varian Vacuum). After pressure stabilization the ource wa op ned to th 2 nd and 3 rd stage by opening the slide valve. A 3 or 5 kV pul e power upply (IR 0 MD) was used for di charge powering. The ample wa putt r d for 10-15 minut t effect ample surfac cleaning and stabilization of the di char 0 p tra fr m th

PAGE 100

92 o cillo cop re obtained using software developed in our lab from a ational Instruments Data Acquisition program Trace spectra were taken using the PT A program oft are Gen rally 50 000 pulses were accumulated using the PTA counting mode pectral data were downloaded into Microcal Origin* (TM) where file er con ert d into mass spectra. Mass scaling was completed using Origin according to the equation shown in the TOF theory section such that. Mass = at 2 + b ( q. 3-6 where a and b are constants that are determined when two separate masse are u d to calibrate the mass spectral scale and tis the flight time of the ion. Sputter rates were determined by sputtering a copper sample for 30-45 minute amples were weighed on a milligram analytical balance with a range of 0 001 mg to 20 mg. 0.0001 mg. Sputtering by pulse was calculated by dividing the eight I m m 0 by the number of pulses over the given sputtering time. At lea t three replicat of each gas were performed. The ga es used in all experiments were res arch grad 5 99. 999 % pur fr m pectra a and from BO Ga e

PAGE 101

CHAPT R4 CHARA TERJZATION OF THE MI RO OND PUL E D GD PRO H LIUM Introduction The glow discharge is a miniature reaction cell in which the study of the physical and chemical phenomena of sputtering and ionization occur. By using inert gases such as argon and helium chemical reactivity of the cell is mostly eliminated leaving physical phenomena to dominate Sputtering is the only mechanism in a glow discharge source that liberates sample from the matrix surface. Once liberated the vaporized sample is ionized by collisions and subsequently analyzed via mass spectrometry The most abundant sputtered species are atoms. This atom population made up of matrix and trace species will be the primary comparison point for the study of argon and helium as discharge gases. The use of helium as a discharge gas presents some interesting comparisons with argon the gas normally used in glow discharge devices The argon glow discharge ion ource in both pulsed and DC modes is very well characterized and will be the basis by which the helium discharge is compared. The elemental characteristics of helium are much different than argon including mass atomic size ionization energy and meta table atom energy. By using helium as a plasma gas the sputtering and ionization dif r nc occurring between argon and helium are studied in order take ad an tag of th proc that occur in helium. Table 4-1 shows the phy ical and chemical properti fh lium and 93

PAGE 102

Table 4-1. Physical and chemical properties of helium and argon Helium Argon Mass (g/mole) 4 40 Atomic radius (nm) 32 76 Ionization Energy (eV) 24.587 15.759 Metastable Atom Energy (eV) 19.7,20.3 11.5,11.7 94

PAGE 103

95 argon. Th ma of argon i t n time that of h lium and th r fore th putt rin g i expect d to be great r for argon than helium under similar condition A r view of th a pect of putt ring from Chap. 2 shows that the mo t important term in the putt r y i ld equation i the energy transfer term Energy Transfer Function = 4mimt / (mi + mt)2 (Eq 4-1 ) where mi is the mass of the incident ion and mt is the mass of the target atom. Therefore the heavier ion being closer to the mass of an element of interest like copper will sputter more than the lighter ion if each has equivalent energy when bombarding the surface. Barshick 67 has shown that by changing the matrix to carbon an increase in ion signal is observed when helium is used versus argon. This is an example of how the energy transfer function works. The use of copper here is based on the reasonable sputtering yield and the small number of isotopes for mass spectrometry. Wehner has shown that the sputtering yield of argon versus helium will be about 8 times greater for copper under vacuum conditions when the ions have the same energy (see Wehner plots in Chap 2) The glow discharge is certainly not in a vacuum and the ions bombarding the surface ill have a much larger range in energy than tho e from an ion gun. It s po ibl that th average energy of the ions will be close to that of the ion gun y t th y will ha a larb range. Therefore the high range of incident ion energies in th glow di charb ma affect puttering differently than for the vacuum ion gun that ha a n rg range. As mention d pr viously th ionizati n n rg of th h Ii um at m i mu h greater than that of argon Therefore mor n rg i n d d t pr du

PAGE 104

96 lectrons to a i tin the selfu taining nature of the GD This r ults in a ignificantl high r n rgy pla ma with a much higher current. Thi higher energy pla ma can b achie ed b increasing oltage or pressure. By ignoring The ionization b thermal effects an a umption can be made that helium would hav better ionization capabilitie than argon due to the high metastable energy levels. The metastable level of argon ar only on the order of 2 e V above the higher ionization energies of some metal in the periodic table including Sn As Pb Sb and Se wherea helium has meta tabl nerg levels nearly twice the value of the first ionization level of the e elements. Ther for higher ion populations of elements in helium compared to argon plasmas should occur. Background Helium has become widely used in plasma spectroscopy as an additi e ga in order to enhance emission lines in ICPs 68 glow discharge pla ma 69 and MIP 70 The u e of helium n on nitrogen and krypton in RF GDM ha b n tudi d by igli and Cam o 71 who reported sputter rat s for Ar and He of 28 and 23 / min r p cti el for a copp r ample The author a urned that helium at m w r n t good n u g h putt rin 0 ag nt and that mo t of th puttering was du to putt r d u atom ioni z d cl t th urfac and ace l rated back toward th urfac a in id nt i n F r th -pul di harg it additi nal putt rin g b matri i 11 aft r ha in 0 b urfac durin g i n bombardm nt b ga i 11 Th pul h uld b h rt n u g h t liminat thi ffi ct and a ll w diffu i 11 of at m thr u g h th d rk p putt r rat h a e b n r p rt d b Bar hi k 7 in a g l di har g i ni ati n

PAGE 105

97 ourc that show an approximat 6 fold dif r nc m putt r rat at 1 t rr and lkV r argon v r u helium. The helium glow discharge has been limited in u e to Grimm-typ ource wh r ery high ourc pres ure ar used (20-50 Torr He). 72 The higher pr ur 1 n d d to produce enough sputtered atoms for analysis by atomic emission. The vacuum requirements for mass spectrometry have difficu l ty supporting this high pressure and thi i lik Iy why helium hasn t been widely used in GDMS. However the use of helium as an ionization source for GCMS is we ll known. 73 For spectroscopic purposes it ha been used for the determination of chlorine in ha l ogenated hydrocarbon vapors 73 as a deve l opmenta l scheme for GC ana l ysis. The determination of nonmetals in vapors and in solids i a new and growing challenge in atomic spectrometry. These sources have been called gas sampling g l ow discharges They take advantage of the Grimm type or hollow cathode configurations and are able to sample liquid or vapor directly into the plasma. In uch cases the excited metastable helium atoms are then able to excite and ionize these halogens which are not efficiently excited and ionized by argon meta tab] The Grimm source offers high current densities and a fairly confined plasma. The e attribut s made it particularly useful as an excitation source. In a separate study of the helium plasma Tsuji Wagat uma and Van Gri k n used a 5 kV discharge at l ow helium pressure to create fa t 1 ctron which pa d through th negative glow of their hollow anod ource. 74 Th fa t l tron indu d X-ray that w re u ed for electron induced X-ray mi ion analy i of F -M binar allo

PAGE 106

98 The helium discharge has proven to be an interesting source because of it s hi g h citation capabilities. The focus of this dis ertation is to in ve ti ga te the s putterin g and ionization properties of helium. As mentioned above helium show a reduction in putt r rate compare to argon. There is evidence to support the data showing a 6-8 fold difference in sputter rate between argon and helium. Both vacuum and glow discharg e data illustrate this even though Giglio and Caruso report similar sputter rates between helium and argon for the RF discharge. The energy transfer term would support their claim in that copper ions would be the most efficient at transfe1Ting energy to the lattic e thereb y maximizing sputtering This might explain the different results obtained b y Giglio and Caruso, compared to what Barshick repmts for the D C g low di charge. R ega rdless when helium is used as a sputtering agent the sputter ield t er m pr e dict a reduction in the atom/ion ratio for helium compared to argon. Typical sputtering studies are performed most routinel y b y mea urin g th i g ht lo s of a sa mple after sputtering for a measured time Most sputter i ld tudi h a u d an ion gun as opposed to a glow discharge source. Be id mea urm g i g ht lo puttering can b e m eas ur e d u ing thin layer m tallic s ample and r cording th tim it tak for th g lo w di charg to putter away th d pth of the thin la r. ha b n typically u d a a d etect ion m thod h r although ma p tr m tr ha b n u d t ana l yze m thin l ay r 7 5 T h r ar variou way t m nitor putt ring in a g l di charg ur mo t n tably mea ur m nt f th a t m p pul at i n b a t mi i 11 2 1 r fi r th i ni za ti n f th at mp pulati n 1 9 for ma p tr m tric purp B th hav b n u d in thi l ab rat r y fi r tra c a n a l and fundamental tudi f th

PAGE 107

99 di charg param t r For th t chniqu monitoring chang 111 i gna l utput pro id ary mean to a ume that puttering ha b n af:fi ct d b y a chan g m ither pre ure or voltage. Probing a glow discharge pla ma with a la er 7 6 and mea uring th an1ount of ionization p ctro copically has also been explored. These conventional m thods ar reliable and wid ly u ed. The next section details the comparisons of argon and helium as plasma gase m the s-pulsed glow discharge ion source. The conventional processes are studied using TOFM The section defines the processes that are similar and different when using either argon or helium. The processes occuring in the s-pulsed glow discharge are compared from discharge initiation through sputtered atom ionization and sampling. The final section summarizes via three dimensional plots the entirety of the processes discussed. It is shown that under certain conditions helium can be used as a plasma ga for bulk and trace analysis in the pulsed mode Pulsed Glow Discharge Processes: Comparing He and Ar. Discharge Initiation Breakdown voltages were determined by sustaining the discharge for 5 minut and slowly lowering the voltage until the plasma was extinguished and appr ciabl lo of current wa ob erved. Initial pressures for this study wer 1-2 torr for h lium and 1 torr for argon ignificantly less then the 20-50 torr used for h lium in th at mic mi ion exp riments abov In the DC mod both arg n and h lium had imilar breakdown voltage averaging about 300 V. Ho v r nl a mall pp r i n i gna l

PAGE 108

100 and ignificant gas species signal were observed in the mass sp ctrum for the helium di charge an indication of very little sputtering. In the pul ed mode di charg breakdown occurred at 800 V for argon and 2 kV for helium at th ame pre ure The higher oltage observed for argon is most likely due to the transient nature of the pulsed discharge. The much higher voltage observed for helium is probably a combination of the higher ionization energy of helium and the inabilit of the helium discharge to sustain a high enough electron and ion population during the 10 s pulse period. The collisional cross section of atoms at 1 2 torr helium is much smaller than the collisional cross section of atoms in argon at the same pressure. At 1 torr the mean fr e path for an electron in argon is 0.584 mm, and 3 02 mm for helium. At 2 torr the mean free path for helium is 1.5 mm, and 0.44 mm at 7 torr. Therefor the mean free path for an electron is much larger in helium at low pressure and the possibility of n utralization at the walls of the chamber is more significant at short sampling di tance In th e tudi the optimal pre ure for helium was selected to be 7 torr ( e Diffu ion) and 1. 5 torr for argon. Th se pressures equalize the mean free paths of th el ctron t th final pr ures the breakdown voltag is 1 kV in the pulsed mod for helium. Th mall r m an fr path cau e greater ionization of th h liwn atom and u tain th pla ma at l wer v ltag l ctr n impact i nizati n i th fir t i nizati n proc during th pul d c l and th primar y m th d f pr ducing di charg ga i n fi r b mbardm nt f th cath d

PAGE 109

IOI urfac When a high n gative voltage is applied to th cathod a br akdown f the di charg ga occur and l ectron a re accelerated away from the cathod ur face and collid with discharge gas atoms an d gas impurities such as water and nitrog en. Time of-flight mass spectra tak n at ear l y delay times consist of the ions representative of th discharge gas composition Figures 4-1 band 4-2b show the electron ionization spectra for argon and helium respectively. T h e significant difference in int ensity for the helium ion is most lik e l y due to the high er number of helium atoms and the higher applied voltage In both spectra, H 2 0 + are present ; however in helium N + and OH + also appear w ith appreciable intensit y This could be a result of helium metastable atoms immediately ionizing these species. The production of discharge gas ions is critical in sustaining the glow di charge plasma for the ten microsecond period. Both electrons and ions maintain the circuit in th e cell. As gas ions strike the surface, e l ectrons are emitted as we ll as atoms. In fact some e l ements have a greater ability to induce electron emission from the target surface than others The secondary e l ectron emiss ion coefficient for ion bombardment is given b y Yi ( e l ectrons / ion). This measures the number of electrons emitted per incident ion. Figure 4 3 30 shows Yi va lu es for the noble gases on clean tungsten and molybdenum metal a a function of incident ion energy. In the energy range for a pulsed g l ow di charge H + produce about 2.4 times a many e l ectron as Ar + Electron emi sion i e ntial in ionizing the gas ions during the hort pulse period. U ltimat ely maximi z ing th am unt f di charge gas ion for bombardment of the surface i beneficial which d p nd on increasing the current. Curr nt in the s-pul ed mod i ffi ct d b g lo di har

PAGE 110

0.18 0 16 a 0 14 0 12 0 10 0.08 0.00 0 04 ~0.02 ~0 00 -~ 0 0.16 b 0.14 0 12 0.10 0.08 0.00 0.04 0.02 0.00 0 20 40 60 20 40 60 wz Figure 4-1. Electron ionization mass spectrum of an argon discharge at a) 1.2 kV and b) 1. 7 kV 102 80 80

PAGE 111

07 a 0.6 05 0.4 0.3 02 01 ~00 f 01 b 5 10 15 25 0.6 05 04 0.3 02 0.1 00 0 5 10 15 a) MZ Figure 4-2. Electron ionization mass spectra in a helium discharge at a) 2.5 kV and b) 3.5 kV 103

PAGE 112

. in e l ectrons I 0.32 0.28 0.24 0.20 per i on 0.16 0.12 0.08 0 04 --_____ ..-------Ne -He --Tungsten --Molybdenum ---__,,, __ ....-~ -___ __ ,_,__ ..... ... --_,_ --Kr Xe -------.._~----...... -O't--..----...----r----0 200 400 600 800 1000 Ion k i etic n rgy { eV 104 Figure 4-3. The number of electrons emitted per incident ion for the noble gases [Ref. 30]

PAGE 113

10 5 param t r Th a rag curr nt i affi cted by voltag pr ure and frequ nc y. Pul width can al o affi ct the av rage current ; however it is kept con tant through ut th p riment The peak cunent i only affected by pressure and voltage. The peak cun nt 1 of primary concern because it is the main driving force of the s-pulsed di charge B y adjusting the pressure and voltage in the s-pulsed GD more ions for bombardment are produced by electron impact. Optimal conditions for the helium discharge are obtained with an applied voltage at 3 5 kV 405 mA peak cunent compared to the 1.7 kV applied voltage for argon and 210 mA peak current. The peak cunent in the helium discharge i two times higher than in argon and approximately 100 times higher compared to the DC mode creating a higher energy plasma and allowing the use of helium as an analytical plasma gas. A Voltage Effect Figures 4-1 and 4-2 show the effect of voltage on the electron impact ion signal in the s-pulsed discharge for argon and helium. The delay time s lected for the e pectra was 10-20 s after discharge initiation. Aside from the major masses for argon at 40 and 20 and for helium at 4 the most noticeab l e effect in the mass spectra i the contribution of th gaseous dimer (m/z 8 in helium and m/z 80 in argon). The dimer i much larg r in helium than it i in argon. Both ions have a net bond order of ; how v r th increa d signal for th He 2 + i a re ult of a more stable mol cular orbital filling at th 1 I 1 than in the 2 and 2p levels for the Ar 2 + in argona com par d to th r r p ti at mi i n Ov rall an incr ase in voltage for both di charge r ult in a larg r contributi n f i n

PAGE 114

106 from el ctron ionization Thi can al o be obs r ed as an incr a e in peak curr nt a a function of th increase in voltage as shown in Figure 4-4. The higher voltage for the h lium discharge increases the current. Also note here that the pre sure is higher in the h lium discharge but the atom density and ionization cross section are the ame a that in argon (see Diffusion Section in this Chapter). B. Pressure Effect Figures 4-5 and 4-6 are mass spectra showing the effect of increasing the pres ur on the electron ionization process for argon and helium. By incr a ing th number of discharge gas atoms an increase the number of gas ions formed an increa e in p ak current at a fixed voltage Figure 47, and an increase in ion signal for both ga observed. The increase in pressure causing a shorter mean free path for the electron lowers the voltage at which the discharge initiates. puttering of Atoms N t puttering in the exp riments i det rmined by m a uring th ight lo a ampl putt r d v r a giv n time. Th total w ight lo i di id db th nw11b r f pul c rr p nding to th total putter tim u ually 45 minut r 540 000 pul Th putt ring tudi ar ummariz d in Tabl 4-2. ptimal nditi n of 7 l rr .5 kV fi r h lium and 1.35 t rr 1.7 kV fi r argon r u d. Th r ult h that ar g n pull r 0 29 ng / pul and h lium putt r 0.04 ng / pul h 7 :6 ld difD r n f

PAGE 115

420 400 380 360 340 320 f300 -280 c Q) 260 I;;. 'a 240 220 ~200 180 160 140 120 1 0 ,I I / 1.5 2 0 2.5 Voltage 3.0 3 5 Figure 4-4. Pulse peak current for argon and X helium 107

PAGE 116

108 0.03 a O a2 0.01 0.00 t---~--r--------,---r-----.-----r-0 20 00 00 0.3 b 0.2 0.1 0.0 0 20 m/z Figure 4-5. Electron impact mass spectra in an argon discharge at a) 0.8 Torr and b) 1.5 Torr

PAGE 117

w C (]) ... C 0 1a 0 6 0 5 0.4 0 3 0.2 0 1 0.0 0 10 20 30 ., 0 1-= b 0.60 5 0.4 0 3 0.2 0 1 0.0 0 10 20 30 mlz Figure 4-6. Electron ionization mass spectra o f a helium discharge at a) 3 Torr 3.5 kV and b) 7 Torr 3.5 kV 10 9 40 40

PAGE 118

110 240 a 220 /. 200 /. 180 ~
PAGE 119

111 Table 4-2. Sputter rates for argon and helium under optimal conditions SAMPLE GAS Press/Cur Voltage # of Pulses Wt Loss (mg) ng/Pulse A\erage (ng/pu l RSD 1 Ar 1.35/0.17mA 1 7KV 540000 0 17 0 32 -2 Ar 1.35/ 0 17mA 1.7KV 540000 0 155 0 29 3 Ar 1.35/ 0.17mA 1.7KV 540000 0.145 0 27 0 29 4 30% 4 Ar 1.35/ 0.17mA 1.7KV 540000 0 156 0 29 1 He 7T /0 28 mA 3.5kV 540000 0.025 0.05 2 He 7T /0 28 mA 3.5kV i540000 0 022 0.04 3 He 7T /0.28mA 3.5kV 540000 0.02 0 04 0 038 14 50% 4 He 7T /0 28 mA 3 SkV 540000 0.015 0.03

PAGE 120

m puttering between the two illustrates the effect of the lighter mass of helium on sputtering. 112 In Chap 2 the factors that affected sputtering were considered. Yet the application of these factors to glow discharge mass spectra was not obvious. The pul ed glow discharge relies on the formation of discharge gas ions via electron impact and the subsequent bombardment by these ions in order to sputter atoms from the cathode surface. Just like the formation of discharge gas ions the sputtering of atoms from the surface is affected by changes in pressure and voltage. Pulse frequency in this case should affect net sputtering of atoms over a given time. Frequency will not affect sputtering within each pulse because the discharge is off for a relatively long period e en at a high frequency like l kHz. The effects observed on sputtering when changing the pressure and voltage parameters are a direct result of the electron impact phenomena and gas ionization. The amount of Ar + or He + in the discharge is affected by el ctron ionization. When changing the conditions that affect electron ionization sputt r ield and net s puttering are affected. With the glow discharge ma s pectrom t r here incident ion energies can not be determined yet they ar ti mated to be 1/3 of th applied voltage. 77 Having optimized voltage and pre ure for h lium and argon parat Jy it can be a urned that the averag incid nt ion n rg1 in argon ar about hal f th in helium. R gardl by varying the 1z of th bombarding ion chang 10 th am unt of putt ring ar ob rv d.

PAGE 121

11 3 Incident Ion nergy The putter yi ld (atom / ion) is affected by the energy of th incid nt ion a wa di cu d in hap 2 During the 10 s pulsed period all the voltage i dropped fr m th cathod surface to the edge of the negative glow where a majority of the incident ion a r formed. As ions feel the influence of the voltage they are accelerated toward th urfac reaching a maximum kinetic energy up to the applied voltage Obtaining a maximum kinetic energy ( equal to the applied voltage) requires the ion to traverse the dark space / negative glow boundary without experiencing any collisions This become more likely at lower pressures as the system becomes more like that of an ion gun that fires ions at the surface. Certainly ion bombardment in a glow discharge does not have the conditions used in a vacuum but on a microscopic scale near the surface it is similar The long mean free path at lower pressures allows ions to strike the surface without encountering many collisions. In the glow discharge case the size of the dark space increases when ions are formed farther away from the surface allowing them to be affected by the electric field for a longer period of time increasing the energ y with which they bombard the surface. While the larger energy may cause a larger sputter yield ( atoms / ion) it doesn t mean there is an increase in the overall or net putt r rate of th glow di charg y tern. As pre sure decreases the discharge will not likely su tain it I f during the hort pul e period. A current decrea es the di charg i dri en furth r out of the abnormal di charger gion and into the normal glow di char g r gion m anin g that the plasma cov rage on the urfac of th cathod i r due d. hapt r 6 mi g ht gi an timation of the incident ion energy ba d on th ratio of u 2 t u h r n u,PuP g l

PAGE 122

di charge n ironment dra tically affects thi ratio An accurate anal n arly impo sible with the system used for thi study. B. Pres ure Effect 114 of thi nerg 1 There are several factors to consider when assessing the effect of pre sure on puttering these are : the sputter yield net sputter rate the number of plasma collision and the current. The sputter yield (atoms/ion) is determined by the size and n rgy of th incident ion. Disregarding size increasing pressure will decrea e the nerg y of th incident ion because it will experience more collisions on its path to the cathode urface that reduce its energy. Wehner et al. have shown that reduced ion energ decreased s putter yield 38 However when pressure was increased signal for the copp r ion increased which was not predicted for the effect of pressure on putter yi ld. Thi because the copper ion signal is not determined by a single event uch a th impinging of an argon ion rather it is determined by the net sputtering of a number of impinging ion Therefore as the pres ure increases e en though th th numb r of bombarding events increases and there i a great r n t putt rin g and an mer a d c pp r ion ignal. he effect of pr ur on the copp r ion ignal i h wn in 1 g ur 4-8 fi r h lium and ar g on. Hi h r pr ur al 111 r a th numb r of atom in th 0 1 di har g a m r lli i n p 1 atomic p c1 into th ir r p at m h a bility t br a k d wn th p l y at m1c m an hi g h r ion fracti n f th putt r d p i

PAGE 123

> Z' w C: Q) C: 11 5 0 6 a 0 5 0.4 0 3 0 2 0.1 0 0 0 7 3 4 5 6 7 b 0.6 Helium Pressure (Torr) / 0.5 0.4 0.3 0 2 0.1 0 0 0.8 0 9 1.0 1.1 1 2 1 3 Argon Pressure (Torr) Figure 4-8. Effect of pressure on copper ion signal in a) helium and b) argon 1 4

PAGE 124

116 High r ionization is obser ed a an increase in current a the pre ur 1 mcrea d at con tant oltag C. Voltage Effect The voltage controls the average energy of the plasma When the oltag i increased electron energy acceleration away from the cathode urfac and incr a m energy of the bombarding ion and correspondingly an increase in current are m a ur d. An increase in electron energy has one significant effect: an increa e in the numb r of argon ion formed that are accelerated toward the cathode surface cau ing a high r n t sputtering. The increased kinetic energy of the argon ions also affects a higher n t sputtering because more energy can be transferred to the surface increasing the putter yield. Figures 4-9 a and b shows the effect of voltage on the copper ion ignal. Th defl ction time at each voltag is set to maximize the ion ignal. In hap 2 it a ho n there i a maximum to the energy transfer for sputtering by ga ions ; ho e r n rg1 in the -pulsed GD ion source ar not really clo e to th kn of that n rg cur ig 2-3 and 2-4). D pite the imilarity in Figur 4-9 a and b kn that h lium v n at high voltage and current putt r w rat m und r ptimal onditi n numb r f b mbarding nt that occur b cau in th ltag Th f ct of lt ag i li g htly dif r nt than th ffi ct f pr ur in that it i difficult t d t rmin whic h i m r r p n ibl r th incr a m c i g nal th putt r ld r th n t putterin rat In thi ca th t m mplim nt r

PAGE 125

0.7 b 0 6 a 0 5 /I 0 6 /' 0.4 I 0 5 ,,-... > ,,--.. > 'cii 0 3 -~ 0.4 C Q) Q) +" ..... C C 0 2 0 3 0 .1 ~ 0 2 I 0 0 2 0 2 2 2 4 2 6 2.8 3 0 1 1 1 2 1 3 1.4 1 5 V olt a g e ( kV ) Voltage (kV) Figure 4 9. Eff e ct of voltage on copper ion signal in a) hel i um a n d b ) a r go n

PAGE 126

118 The collision cross section and mass of the discharge gas influence the diffu ion of copper atoms through the dark space and into the negative glow. Figure 4-10 78 h the mobility of ions through different gases. Ions can be used to represent atoms in thi figure. The smaller cross section of helium will cause a much faster diffusion of specie from the surface to the sampling orifice if the same pressure in this case 1 torr is u ed for both gases. The much larger argon atom cross section causes a much slo er diffu ion of sputtered atoms. When examining the shape of the ion mobility curves a slight dependence on the size of the ion diffusing is seen but a much greater dependence on the medium in which the ion is diffusing is also apparent. The curves show that the diffu ing mass has a greater effect in helium than in argon. Therefore there i some dep ndenc on sputtered atom size when the cross section of the two gases is equal i.e. wh n th pressure of helium is increased the diffusion of atoms is reduced. The colli ion of an argon atom with a copper ion will effect a slower diffusion than the colli ion of a helium atom with copper ion. The helium atom is more easily push d out of the way r ulting in a fast r diffusion of the sputtered atoms. This effect i rather mall in the lo pr ur di charg u ed here but has an effect at similar pres ures for argon and helium. In ord r t quali z th diffu ion rat of atom w mu t con ider th cro ction and / or th ar a of th arg n and h lium atomic nuclei. By calculating th total ar a f th and dividing it by the total ph re area of helium a calculation of th approximat pr ur that will create imilar atom d n ity and diffu ion charact ri ti for b th ctional ar a i d t rm d by 4nr 2, and th f argon

PAGE 127

lB 7.0 --00 c 1 2 N 8 C) '-" 8 :E 0 4 I I I \ :-----..__ I I H E..1..ll)M I i I 1 \ I I j ....._ 1---._ i j NEON I, ARGON !>0 LOO 150 200 MASS OF ION Figure 4-10. Mobility as a function of ion mass in He Ne and Ar. [Ref 78] 119

PAGE 128

1 2 0 corr sponding radii for each gas are 0.33 A for helium and 0.76 A for argon: the ratio of th collision ar a of argon to helium is 5.05. This means that fiv times the pressur i required for helium in order to create similar diffusion characteristics for both pla ma The optimal pressure for helium then becomes 7 Torr if 1.35 Torr Ar is used for tandard analysis. The effect of pressure on diffusion can be measured by noting the deflection time of the maximum peak for the copper ion signal. As pressure increases the peak signal occurs at later times. The number of collisions increases as pressure increases causing a slower diffusion of sputtered atoms through the plasma Figure 4-11 shows the deflection time versus pressure for the maximum of the copper ion signal in argon Figure 4-12 shows that the copper ion signal is first detected about 80-90 s after the di charge pul period in the argon discharge while in helium it is about 30 -40 s a shown in Figure 413. This feature can be attributed to the small dependence on the mas difference between argon and helium when considering the diffusion of atoms. The pectra indicate that the ma x imum copper ion ignal is obtained by setting the deflection time to about 140 ~l for both h lium and argon di charge This i an indication of fairly equal cro the t pr s ures of 1.35 and 7 Torr for argon and h lium r pecti ly. ction at In pr vi u chapt r it wa n t d that th primary ioni z ation m chani m for putt r d p ci ar P nnin g and ctron impact ionization. P nnin g i ni z ation i n id r d th m r d minant f th tw proc n thou g h l ctr n impa t ha b n

PAGE 129

160 .,-.. en 140 (1) E I120 C 0 .:; 0 (1) .;:: 100 Q) 0 80 60 / 0 6 0 8 1 0 1 2 Argon Pressure ( torr) / / / / / 1.4 Figure 4-11. Deflection time for the maximum copper ion signal versus argon gas pressure 121

PAGE 130

0 8 0 6 02 0 0 122 I I I' I I -~ '/....f I I 100 S I I I I I I I I I I I 0 40 00 00 100 Figure 4-12. 3-D microsecond pulsed mass spectra of copper in argon at 1.7 kV 1.35 Torr and 3 mm cathode-to-orifice(C-O)

PAGE 131

1.0 0 8 0 6 Ch C: 0 4 -'= 0 2 0 0 0 !t I I I 10 20 30 I 40 I 50 M/2 I I 60 70 1 23 ,l 200 100 () d' I I I 80 90 100 Figure 4-13. 3-D microsecond pulsed mass spectra of copper in a helium discharge at 3.5 kV, 7 Torr, and 3 mm C-O

PAGE 132

124 stimated to cause 40 % of the ionization in the continuous DC mode In the -pul d mode TOF deflection of ions into the flight tube occurs 100 to 300 s after the di charg has been extinguished. By this time there are few electrons with enough energy to ioni ze the metal atoms. Hence the only ionizing agents left after the pulsed period are metastable gas atoms of helium and argon Measuring the metastable population of helium or argon atoms formed during the pulsed period is impossible with the instrumental setup used here but has been performed on both helium and argon discharges using the laser depopulation method 14 There are however some generalizations that can be made about metastable atom formation with respect to pressure and voltage. In general for both helium and argon the parameter that affect electron ionization of argon and helium also affect the population of metastable atoms in a similar manner. As pressure is increased the number of bombarding ions increases releasing more electrons from the surface during th colli ion cascade. A higher current and greater number of electrons tran late to a g r at r numb r of metastable atoms being formed. Electron impact is the primary m e thod b y which meta tabl gas atoms are formed and as voltage is increa ed current and h nc ioni zat ion do a l In add ition a ion strike the surfac mor oltage mer a a nd a n incr a in current is observed. The re ult i a two-fold b n fit fro m incr a ing pre ur and voltage; an incr a e in putt ring and an incr a m 10mzmg ag nt ccur in both ca and r ult in th mer a in copp r ion i gna l ob r d.

PAGE 133

1 25 ampling Thu far spectra have only been shown that illu trate the fundam ntal a pect o f the -puls d discharge separately for helium and argon. While analyzing each proc s individually i necessary the ability to take advantage of them is determined by how ampling occurs. The time-of-flight mass spectrometer allows definition of the exact ionization events occurring in a particular space above the surface of the sample at a certain time. By collecting mass spectra at sequential delay times we can determine the optimal sampling position and conditions for trace analysis. To fully understand and take advantage of the effects of pressure and voltage the deflection time of the time-of-flight mass spectrometer is used to obtain a complete profile of the pulsed cycle and its processes. The mass spectra are compiled into 3 dimensional graphs of signal intensity versus deflection time for all masses in the elemental range. The mechanism by which sputtered species are ionized in helium and argon i similar, except for the energy of their metastable atoms. The higher energy of metastable helium atoms ionizes the sputtered population more efficiently than the argon metastabl atom The efficiency of ionization depends upon the collision cross ection of th ionizing metastable atom and the metal atom In other word how efficient i th geom tric colli ion of the metastable colli ion of argon ver us that f h lium int rm of the tran fer of energy? It has been suggested 5 0 that the closer th n rg of th arc n metastable is to that of an excitation level th more efficient th tran r f n rg ill be. However when considering ionization th only valu c n id r di th an1 unt f

PAGE 134

126 n rg it tak to remo e an electron. Therefore much like el ctron impact ioni za tion there ill b a d pendence of electron energy on the ionization cro s ection of th meta tabl atom. Yet the metastable atoms of helium and argon are so lo m energ m comparison to the e l ectron energies that an increase in meta table energy i going to mcrea the ionization efficienc Figures 4-12 and 4-13 show these 3 dimensional plots of ma s pectra at sequentia l deflection times from 10 s to 300 s for both argon and helium. The critical feature of the mass spectra is the intensity of the copper ion peak. The inten s ities are ery close for each gas. However the sputter rate for helium i 7-8 tim than for argon indicating that the higher energy metastable helium atoms must ioni ze an sputtered population 7-8 times more efficiently than argon. The figure hown ar a result of maximizing the pressure and vo ltage parameters in both helium and argon. Figur 4-12 and 4-13 represent the optimal condition for argon and h lium thi b e ing 3 mm cathode-to -orific e distance (C -O) 3 .5 kV and 7 Torr for h lium and 3 mm -0 1 7 kV and 1.35 Torr for argon. It is evident that the cone ntration f ga p c1 H 3 O + OH + N + i much high e r in th h lium di char g imilar c n c ntration of th l ak impuritie xist in th t di charg !ta g u d w ith h lium will certainly ioni ze mor of that population. urnin g that th nth hi g h r h hi g h m ta tabl at m n rgi impuri ti I ni z rn ga f h lium will al o incr a th ioni zat i n f th a unlik rg nm ta tab! hi h mak it m r diffi ult t tak a d va nt ag f th t mp ra l r p n n rma ll y a hi d ith a r g n In a dditi n th app aranc f c pp r i n at a rl d fl cti n tim in h lium pr that h liu m

PAGE 135

1 27 m ta table ar contributing t th high ga p c1 observed at the e e arly d e fl cti n tim The distanc between the cathode urface and the sampling orifice is being call e d the C-O di tance. Correctly choosing this distance is critical in obtaining the ma x imum ignal for the sputtered p cies. When operating in the DC mode the C-O distance is 6 mm and an explanation for this large distance has a l ready been discussed. In the s pulsed mode this distance is 3-3.5 mm. If the C O distance was set to 6 mm in the pulsed mode the spectra shown in Figures 4 14 and 4-15 would be the result for argon and helium respectively The signal in both cases is much smaller as compared to the signal observed in Figures 4-12 and 4 13 where the C O distance is 3 mm. Copper atoms experience more co ll isions when the C O is increased causing a much more diffu e packet of sputtered atoms. Therefore their concentration at the sampling orifice is maller at 6 mm as compared to 3 mm. Figure 4-16 and 4 17 show the effect of pressure at 3 5 kV 3 mm and 1.7 kV 3 mm for helium and argon respectively The ionization cross sections for the lower pre sures of argon and helium are similar. The figures show that the copper ion signal in helium is slightly larger than in argon This is most likely due to the higher voltage and current observed the helium p l asma. The most significant factor for th h lium discharge when lowering pressure is the inability to separate gas pecie from th matri copper ions and is the reason why higher pressure i requir d The mix of el ctron ionization and Penning ioni z ation proce se i partly a r ult of fa t r diffu ion f th

PAGE 136

1 0 0 8 0 6 en C: <1) 0.4 C: 0 2 0 0 128 200 0 tf 100 S u d' 0 20 60 80 100 120 Mass/Charge Figure 4-14. 3-D microsecond pulsed mass spectra of copper in argon at 1.7 kV 1.35 Torr and 6 mm C-O

PAGE 137

"' C: 1 0 0 8 0 6 C 0.4 0 2 0 0 0 20 40 60 80 100 Mass/Charge Figure 4-15. 3 .. 0 microsecond pulsed mass spectra of copper in helium at 3.5 kV, 7 torr, and 6mm C-0 129 300

PAGE 138

1 0 0 8 0 6 u; c:: Q) C: 0 4 0 2 0 130 i' 0 ---'-L----'-~_____._. ___ ---.-,......__ __________ 100 If S 'fj c5' 20 40 60 Mass/Charge 80 100 120 Figure 4-16. 3-D microsecond pulsed mass spectra of copper in argon at 1.7 kV, 0.8 Torr, and 3mm C-0

PAGE 139

1 0 0 8 0 6 in C (l) c 0 4 0 2 0 0 I 0 I .. ,I d ll 1 I I I Ii i I I I I I I I I 20 I I JI II II I .I I I I I I I I I I I I I I I 40 I I I J I I ,I I I I I I 60 Mass/Charge I I I 1 l l 100 d' 1 I I 80 100 120 Figure 4-17. 3-D microsecond pulsed mass spectra of copper in helium at 3.5 kV, 3 Torr, and 3 mm C-O 13 1

PAGE 140

matri atom Therefore the separation in time of electron impact and meta table 1 nization i much smaller at low pressure ve r us high pres ure 132 A imilar situation occur when voltage is adjusted as shown in Figure 4-18 and 4-19. When voltage is decreased the signal for the copper ion decreases a result of lower putter y ield and lower net sputtering. It also seems that a lower oltag fa or an en ironment of more gas species interferences such as H 2 0 OH and 2 H mo tl molecular in nature. The lower energy plasma can not breakdown the e molecular species and therefore has just enough energy to ionize them. The spectra how that it i more difficult to cause a separation of gas and sputtered species as the vo ltag e i d ec rea se d Summary The fundamental s-pulsed GD proces es ha ve b ee n d fin d for both h lium and argon as plasma gases. The size and mass of th h e lium atom affect b th puttering and ionization in that a higher en rgy pla ma i achi ed with h lium but a much lo w r sp utt ring effici ncy. A 7-8 fold dif r nee in n t putt rin g i r p rt d d pit having coppe r ion i g n a l of th e am inten ity It i th hort int n pul th at mak h lium a v iabl di charg ga for analytical mea ur m nt T h r uld b n n in u in g a ga tor m v int r r nc i f n iti v it i ac rific d. T h n t hapt r r ana l ytica l a d vantag f th h lium di charg in g n ral h lium i n ar in ffi i nt putt rin g ag nt mpar d i th ar n i n and

PAGE 141

1 0 0 8 0 6 &:' en C (1) C o 4 0 2 0 20 40 60 Mass/Charge 80 133 100 120 Figure 4-18. 3-D microsecond pulsed mass spectra of copper in argon at 1.2 kV, 1.35 Torr, 3mm C-O

PAGE 142

1 0 0 8 0 6 'in 0 4 .f: 0.2 0.0 I I t 0 ----I II I I I ii I I II I I I I I I 20 I ii II I I I I I I l I I l I I II I I I I I I 111 I I II I I 111 I 1111 I II I I 40 t 60 Mass/Charge iii I f 100 I I I J I t 80 100 Figure 4-19. 3-D microsecond pulsed mass spectra of copper in helium at 2.5 kV, 7 Torr, and 3 mm C-O 134 200

PAGE 143

arg n meta tabl atom ar only mod rat ionizing ag nt compar d with th hi g h fficiency ionization capabiliti of th h lium metatabl atom. 1 35

PAGE 144

CHAPTER 5 THE HELIUM DISCHARGE ANALYTICAL CHARACTER! TIC Introduction The processes of the pulsed cycle have been discussed in the last chapter and it now important to show how we can take advantage of using helium as a pla ma ga There are three aspects of the helium plasma that are considered here: remo al of interferences Relative Sensitivity Factors (RSFs) and Relative Ionization Factor or RIF. Removing interferences without a loss in sensitivity of trace elements in the copper matrix is essential in proving that helium can be used as a analytical pla ma gas. Th last chapter howed similar signals for the copper ion in both argon and helium di charg The RIF is a measure of the ionization efficiency of the helium pla ma r u th arg n pla ma. RSFs have been used abundantly in GDMS as correction factor for el m ntal analysis. 42 Considered here are the important differences between th valu obtained in h I ium er u argon. Identifying Inter rences Int r r nc from argon ari from th combination f di har g ga p i and an l ak ga p c1 uch that Ar + ArN + ArH 2 +, Ar H + and r + ar rm d. th r int rfi r nc ar a r ult f th putt red p ci mbinati n u h a u W and uAr + Th c mbination f putt r d atom a p l atomic p ci al o ffi t th ana l i in th g l w di charg Th 136

PAGE 145

+ + U 3 u OH + and u 3 0 + The influ nc of th 1 37 dim r and trim r i in the det ction of l ad tellurium and bi muth. Unlike om of th oth r int rfi r n Ii t d abov these interferences are not removable It seems however that we can tak advantag of the di charge gas by using helium to reduce the effect of the e p l yatomic in the mass spectrum ( ee Chap. 6). Interfering Ion Formation From spectra previously shown, it s been illustrated that there are ioni zat ion processes and association reactions that occur well after the pulse period that produce argon based and other interfering ions The goal is to reduce these interferences on the mass spectrum and we propose two ways of accomplishing this. The first takes advantage of the temporal effects afforded by the short transient signal produced The temporal advantage can reduce but not remove contributions of argon and argon-ga interferences An alternative method involving the use of helium a the di charge gas can be used for removal of such interferences. In comparing how these interferences are formed an under tanding of which ioni za tion processes are dominate in helium and in argon is neces ary to take advantage of interference removal. It s also necessar y to under tand the difference in the proce in order to explain the spectra observed in both discharges. In the DC mod th u of h e lium i a problem becau e it has an xtrem ly low sputtering rat Th con tant pres nc of th plasma in th DC modem an that ioni za tion proc int e rfi ring ions mu t take into account electron ioni za tion. From th pr 1 u h pt r

PAGE 146

138 lectron ionization is affected by pressure and voltage and therefore the contribution of interferences in the mass spectrum can be attributed to these effects in a GD plasma that i constantly producing electrons. In the pulsed mode electron ionization is only a factor for 10 s. In this time frame some contribution to the formation of argon ions and related interferences can be attributed to electron ionization. However during the sampling period (100-200 s deflection times) electron ionization is negligible and formation of argon and its related interfering ions is small. In the s-pulsed mode argon ions that associate to form interfering ions are themselves formed from charge exchange and oth r secondary ionization pathways. Argon metastables are the dominant ionizing agent for sputtered atoms but other ionization mechanisms are dominant in leading to the formation of argon ions in the pulsed mode In the DC mode argon ions are formed primarily by electron ionization while the other ionization mechanisms are n gligible in comparison. In a helium discharge electron impact ionization dominate in the D mode for helium ion formation and impurity gas species ionization. In the pulsed mode meta table ionization as opposed to charge exchange in argon i the dominate ionization proc for interferences becau e the helium metastables have nough n rg to i ni z mo t int rfi rences and gas sp cie Therefore the primary m chani m for putt r d p c1 ioni z ation is al o the primary ionization mechani m for impurity ga p c1 Takin g advantag of th t mporal a p ct in th pul d di charg i mu h m r difficult in th h lium di charg than th argon di charg Th lack f int rfi r nc in critical ma ran g ma y b advantag ou in th h lium di charg

PAGE 147

1 39 Th r lative ionization proc se that occur during ach pul ed ycl cau th formation and ionization of isobaric interferences in argon. It wa hoped that an incr a in pressure would serve to reduce these interferences by increasing the number of collisions causing dissociation into their atomic constituents. However at any pressure mass spectra contain interferences at critical masses particularly those related to argon such as argon oxide, argon nitride and argon water combinations. Figure 5-1 shows a 3D graph at high sensitivity for a copper sample showing the elemental mass range against TOF deflection time for argon. A small signal at m/z 56 remains in the mass spectrum regardless of delay time. This is a pure copper sample such that iron couldn't be responsible for the peak at 56 and therefore it must be the result of an ArO ion. Therefore accurate trace element analysis can be affected by the presence of these interferences particularly for iron in argon. Much of the literature discussing the alternative use of gases other than argon cites the interference at mass 56 a b ing the primary focus. 71 The easiest way to relieve this problem is by removing the interferences compl tely and can be accomplished by changing the discharge ga to helium. H we er a change in di charge ga often results in a loss of ensitivity A change to helium in th se exp riment do not end with thi re ult. Figur 5-2 ho a 3-D graph at high sensitivity for the elemental mas spectrum ver us defl ction time in a helium di char g of the ame copper sample. The di app arance of the mass at 56 d m n trat th advantag of u ing helium a the di charge ga in th r mo val f unwant d i bari

PAGE 148

0 15 0 10 u; C Q) 0 05 0 00 20 ----------" ----. :===.: ====-= ::=:===::i: -------~:= ,,it' = 1 1' ~ : = ,1 = 1 = 1 : 1 : r 40 60 Mass/Charge 80 100 Figure 5-1. 3-D s-pulsed mass spectra for copper in argon 140 300 250 200 150 If 100 qj Q 50

PAGE 149

0 10 ?;0 05 ci) C Q) .... C 0 00 o I He 50 20 40 60 80 100 Mass/Charge Figure 5-2. 3-D s-pulsed mass spectra of copper in helium 1 4 1 200 w -::t 150 t 100 .JP cJ

PAGE 150

142 int rfi r nc imilar formation of interferences such as HeO and HeH 2 O ma tak place but i limited to masses under 30. With the majority of helium interferences limited to masses under 30 b tt r i otopic ratios can be obtained for the elements interfered with by argon r lat d polyatomics These elements include iron selenium nickel chromium anadium cobalt and of course calcium Table 5-1 lists the major isobaric interference for th pecific elements in argon. The elements of interest here are iron and calcium. B id these specific elements other elements are also affected by impurit gas specie uch a N 2 H N 2 and CO that appear in the mass spectrum for argon. ilicon is the major element that suffers in this mass range. These interferences are relati ely small in argon but become a much greater problem in a helium discharge where the strong ionizati n f ga pecies is effected by higher energy metastable atoms It seem that man m r metal are affected by argon isobaric interferences than in a helium di charg 1 advantag for the use of argon results from the difference in ionization en rgi f ga pecie in the 1-32 mass range and the argon metastable atom energ Th arg n m ta table nergy is higher than th IE of the e pecies liniting ioni z ation f th p ci Thi doesn t mean that ga p c1 don t app ar in the ma p ctrum ltimat ly in a trul y cl an nvironm nt without air and at r ap r I ak th r m al f int rfi r nc in ar g on wouldn t be n c ar mall l ak in ur ful in in g th p iti attribut f u ing th -pul d di har g lium in rd th pr bl m f p ibl int rfi r nc h mall 1 ak al h th

PAGE 151

143 Table 5-1. Isobaric interferences in argon GDMS Metal Isotope Interference 4oca+ 40Af+ s2cr+ Arc+ 54fe+ ArN+ 56fe+ Aro+ 57pe+ ArQH+ 58Ni+ ArH 2 Q+ 69Co+ ArH o+ 3 sose+ Ar+ 2

PAGE 152

144 di ad antages of helium in that helium metastable atoms ha e high enough energ to ionize most gas impurities. Table 5-2 shows the result of isotope ratio analysis of a NIST Brass 1116 sampl containing 460 ppm of iron. In the first column under each gas is the actual ratio 56 Fe / 5 4 Fe for eight samples. The standard isotope ratio value for 56 Fe / 5 4 Fe is 15.8. The second column shows the relative deviation in % between the measure and calculated values. The isotope ratios were determined by measuring the peak intensity for the respective isotopes in the mass spectrum For helium a lower RSD value of 12.79 is obtained versus that of argon at 19 7 illustrating that the isotope ratios for iron in helium are more accurate than those in argon. This leads us to believe that ArO and ArN type interferences have been removed in the helium discharge. Though the study of elements such as Li Na and Be wa not undertak ninth case of the helium discharge the ability of the helium metastable atoms to ioniz ju t about any discharge gas combination would make analysis difficult for the e l m nt Interfer nces may include He 2 H HeH 3 O and HeO. In Figure 5-1 and 5-2 th ga ignal response from impurities such as water and air are much small r in th argon di char g than in the helium discharge. The ionization pot ntial of th e ga eou p c1 ar ab v th en rgy of the meta table argon atom but b low th m ta tabl n rg of th h lium atom Th ga p c1 r pon eat arly delay tim from Figur 5-1 and 2 i due m t lik ly t l ctron impact ionization which i ti ioni z ati n Aft r 50 to 70 th r pon of the impurity ga pec1e ha di mini h d in th ar g n di char g Y t in th h lium di charg th ga p c1 r p n ntinu 11

PAGE 153

145 Table 5-2. Isotope analysis of iron in argon and helium Helium Argon Fe56/54 Rel. 0 /o Diff Fe56/54 Rel.%Diff 14.38 8.99 11.19 29.18 14.09 10.82 11.2 29.11 13.05 17.41 13.09 17.15 11.72 25.82 14.82 6.2 14.89 5.76 10.51 33.48 17.99 13.86 11.12 29.62 14.34 9.24 14.47 8.42 14.15 10.44 15.17 3.99 Ave. 14.32 RSD Ave. 12.69 RSD 12.79% 19.64%

PAGE 154

146 be ond the d fl ction time at which the copper ion appears and actuall increa es a the peak copper ion signal increases. Th i s corresponds to an increase in ionization from th helium metastable atoms. A slight return of the gas impurities in the argon di charge corresponding to the appearance of the copper ion is observed. The author thinks that some charge exchange and / or other secondary ionization processes are responsible for these ions yet there is no definite evidence. Regardless of the ionization selectivity that helium has for the gas impurities we are still ab l e to achieve temporal separation of the gas and sputtered species ionization events only on the bulk level. Examination of the trace signal for the low-ma r gion of a helium discharge reveals ions at almost every m/z even when copper ion signal i at it maximum as shown in Figure 5-3a. The helium metastable atom has high energ and can ionize nearly everything in the gas phase as illustrated by the saturation (peak ith very broad bases) of many e l ements in the mass spectrum. The H + and He / ar not saturated like these other impurities because the IE of He and He 2 are higher than th helium metastable atom energy. Figure 5-3b shows the l ow mas rang for argon at th optimum delay time for the copper ion signal. The large ignal at m/z 23 i mo t like! a + The pre enc of odium in the di charge i cau d by handling of th ampl and th dir ct in ertion probe. The inability of argon meta table to ioni z e the ga p ci manifi t d in much 1 w r ignal for ju t about ry ma

PAGE 155

147 6000 (1) + a :r: + + N 4000 (1) 0 :r: + z 0 + (1) N + :r: z u 2000 I l ""'u I I I . I I I I 0 5 10 15 20 25 8000 b + t,j z 6000 4000 2000 0 0 5 10 15 20 25 M/Z Figure 5-3. Low mass region for a) a helium discharge and b) an argon discharge at the optimum delay time for the copper ion signal I 30 30

PAGE 156

148 Determination of Calcium in the Presence of Argon Time-gated detection or temporal selectivity in a pulsed glow discharge allow the reduction but not the complete removal of discharge gas species. The remaining argon ions are mainly the result of secondary ionization processes such as charge exchange that continue as long as ions are present within the plasma volume. Figure 5-4 shows the argon isotope at m/z 36 at a deflection time of 150 s for a copper sample. Copper is one of the noble" elements in which sputtering is greater then expected and in which gas species reduction is the greatest at the peak of the copper ion signal. Howe r the spectrum clearly shows that argon ions are detectable The ion signal of 40 Ar + in thi case has saturated the detector as has the signal at m/z 39 believed to be K + Additionally Figure 5 4 shows that detection of calcium in a D discharg would b even more complicated by the presence of 40 Ar + Typically in ma s sp ctrom tr a different isotope could be selected for quantification; however for calcium there i a 45 times loss in sensitivity in using 44 Ca + In addition mass 44 is typically interfer d ith by CO 2 The availability of the 4 Ca + isotope is critical for trace analy i Figures 5-5 and 5-6 show the -pulsed GD mass spectra for argon and h lium with a pur calcium cath de The pectra how that it is impossible in the argon di charg to di tingui h b tw n th 40 Ar + i otope and the 40 a + isotope. A re olution of n arly 190 000 i r quir d to parat the e two ma In the helium di charg the ignal at m /z 40 mu t b du t calcium b cau n rtw1at l th r i t p n t abundant en u g h t b m a ur d on th o cillo c p and ar dra ticall af ct d b th hi g h matri x at m/z 40 and th r int r ring ion uch a 2 in th c w1ting mod

PAGE 157

100 80 60 Y) C ::J 0 (.) 40 20 36Ar 35 36 37 M/Z 40Ar 38 Figure 5-4. Argon isotopes at 150 s deflection time 39 40

PAGE 158

0 05 0 04 0 03 en C: Q) C 0 02 0 01 I I I 0 I I I I I I l A I 20 h h A l A ll Jl 1 II h 90 40 60 M/Z Figure 5-5. 3-D microsecond pulsed mass spectra of calcium in argon 150 290

PAGE 159

I l I I I I l I I l I I I I 1 I I l I l l 110 I I l I I, I l I I 0 20 40 60 M/Z Figure 5-6. 3-D microsecond pulsed mass spectra of calcium in helium 1 5 1 230

PAGE 160

o lo m en iti it is ob er d by using the helium gas. Thi for the purpo es of detecting calcium at trace levels Relative ensitivity Factors (RSF) 152 certainl ad antageou In quantitati e glow discharge mass spectrometry sputt ring and ionization ar two critical factors that determine the accuracy with which trace elements are analyzed and quantified in a given matrix. As the ions are introduced into a mas spectrometer a measure of the relative concentration of elements in a matrix hould be obtain d 1a signal response of the mass spectrometer. However this requires that the respon of the mass spectrometer be uniform for all masses. An equal response is not typicall observed and therefore, a relative sensitivity factor is introduced. Th signal obtained b the mass spectrometer is a reflection of the ion concentration at the sampling point. Th refore both the physical act of sputtering and the chemical act of ionization ill affect the ion signal observed and will create biases for certain elem nt ba don th ch mical characteri tics of these elements. A i n are mea ur d by the ma sp ctromet r the int n it of th ignal i m a ur d again t a matrix trace elem nt of which the mo t mm n i F th ion ignal ar u d to gauge the relativ r pon e of the gl di char 0 ma p ctrom tr he R F are u d to quantify trac 1 m nt c n ntrati n fi r a g i n amp! fi r a given da R ar d fin d b th q uati n b 1 R Fx = ( xi R) / (I / IR) f

PAGE 161

1 53 Wh re X and R indicate the analyt elem nt and th refi r nc m nt r p cti e l I and ar th ignal inten ity and concentration by weight in th ampl Th ffi ct f pr ur voltage ample shape and ample matrix R Fs hav been inv sti g at e d for th ir ffect 42 A one of the critical factors in considering the difference of the helium and ar g on plasmas ionization becomes critical in this work in determining RSFs. By changing th discharge gas composition the sputtering and ionization characteristics of the plasma have changed and a change in the RSF is expected. RSF values are usually based upon iron as the trace element in the matrix. In these experiments a NIST 500 copper sample is used which contains only 41 ppm iron in the matrix and whose signal may be interfered with in the mass spectrum for argon. Th sample contains several elements of interest in the mid and upper mas range The single isotope of silver 107 has been selected to be the relative point of reference for calculating RSF because it is not interfered with and has a significant concentration. The R F alu for silver is chosen as 1. Values higher than 1 are an indication that a decrea e in sensitivity has occurred. In other words not enough signal is obtained for that elem nt relativ to the concentration and response of silver. Tho e elements with R F I r than 1 are the result of obtaining too large a signal and in som ca e thi can b an indication of interfering pecies causing a higher signal. Signal can al o b affi ct d b other factor uch a ionization energy of the lement. For the xperiment performed here optimal condition r u d t m a ur 1 n ignal in each gas. These ar th same optimal condition a tabli h d in th pr 1 u

PAGE 162

154 chapt r in which pla ma processes in the pulsed discharge are di cus ed A look at Fi g ur 5-7 of the R Fs for both gas shows quite different r suit for the alue bet een argon and helium Most R Fs are considered to be between 1 and 5 for argon ho e er a quite larger range is obtained for argon with surprisingly low values for iron and a er high alue for arsenic. The low RSF for iron is most likely due to the pre ence of an interference contributing to the signal at m / z 56 making trace quantification difficult and inaccurate. As for the high value for arsenic a high ionization energy and a lo sputtering yield force the small signal observed in argon. Arsenic i kno n as a difficult element in solids mass spectrometry and is why the RSF is so large in argon. In the helium discharge a low RSF value is also obtained for iron. An explanation is difficult to come by but it may be an effect related t the ma tran port and detection of the iron. Elements such as iron and nickel are v r clo e in ma to the matrix ion of copper. In ion counting mode the matrix element aturates th det ctor cau mg 10n to broaden and spill over to adjacent bin in th d t ctor corr pondin g to other mas thereby increa ing the signal content of the e lem nt Th am pr bl m al applie to the argon di charge. The other elem nt con id r d h r ar ignificantl diffi r nt in ma and are not affected by aturation. Th ran g f R F r h lium in Figur 5-8 how th t th ran 0 i ab ut .2 5 t 1.5 ith a maj rity n ar 1. R F ar u ed t corr ct for gain and l and i ni z ati R fi r th h lium di char g t th ar g n h r m t a l u ary if all th alu t lfira an r fr m 1-5 ith ir n and ar 111 in th putt ring 1. Th mpar d pti n hi h

PAGE 163

u.. U) a:: Fe As RSFs for Argon and Helium Se Sn Element Sb Au Pb Figure 5-7. Relative sensitivity factors for argon and helium 155 Bi

PAGE 164

156 2000 1.800 1 600 1.400 1 200 14. 1.000 r 0 800 0 600 0 400 0 200 0 000 Fe As Se Sn Sb Pb Bi Element Figure 5-8. RSF of several elements in a helium glow discharge

PAGE 165

1 57 r fl ct the lective ionization capabiliti of the argon discharg v nth u g h ar g n m tastables have enough energy to ionize all the element shown they are onl y 2 V high r than ome of the elements with higher ionization energi s Ther for h e lium with its higher energy metastable atoms is less selective in it ionization m anin g that total ionization is increased in helium. This means that there i an increase in the abilit y of the helium discharge to ionize these more difficult' elements whereas argon has a more difficult time ionizing them. Figure 5-9 shows the relationship between ionization energy and RSF. Generally as the ionization energy increases so does the RSF value for atoms in the argon discharge The same is not true for the helium discharge where a fairly flat response is observed. It would be interesting if there were elements with ionization energies close to that of the helium metastable atom to see if the same increa e in RSF value is observed. Relative Ionization Factor (RIF) The Relative Sensitivity Factor is a measure of an individual ga ability to putter and ionize elements in a matrix. The relative ionization factor on the oth r hand is a measure of how efficiently each plasma ionizes a given sputtered atom population and more specifically a single element. The RIFs calculated h re will c rtainly upport the results obtained for the RSF values in terms of increased ioni z ation for c rtain element The RIF i calculated as follow : RIF = (I1-1 e/ I A r) / ( R A/ R1-1 c )

PAGE 166

158 20 X 15 u.. 10 CJ) a:: X X 5 X X .. 0 7 0 7 5 8 0 8 5 9.0 9 5 10.0 Ionization Energy Figure 5-9. RSFs for and xAr versus ionization energy

PAGE 167

1 59 wh re I r pr nts the ignal intensity of a given elem nt and R i the putt rin g rat f th argon and helium pla mas at optimal condition Ther ha to be om e additi nal ioni z ation of certain elements (those with higher IE) in order to explain the lo wer R obtained for helium versus argon. Figure 5-10 shows without question that th r l ati e ioni za tion factor is above 7-8 for As Se and Au, all of which had high R F va lu e in argon and all of which have high ionization energies. A plot of the IE vs RIF i s shown in Figur 5-11 an increase in RIF is observed as ionization energy gets lar ger indicatin g that the efficiency with which the helium discharge ionizes the sputtered population i s much greater. The RIF value must be about 6-8 in order for the ionization efficiency of the helium to match the Ar / He ratio of the sputter rate As an example if iron has similar sig nal in both argon and helium then its RIF values would be about 7 becau e the putt r rate of argon is about 7 times that of helium If the RIF is below 7 then one of two things are possible ; the signal in helium is less than it should be or the signal for the e lement in argon is higher than it should be. Considering the discussion of int erference it is more likely that excess signal is obtained for the element in argon decr eas in g th RIF. Summary Thi ection showed that isotope ratios for iron could b impro ve d b u m g helium as a discharge gas as a method to remove interference from argon i obaric int e rferenc es The use of h lium for the detection of calcium i s hown to b an ffi cti

PAGE 168

160 Relative Ionization Factor Bement Figure 5-10. Relative ionization factors for helium

PAGE 169

60 50 L.. 0 0 ca 40 LL C 0 :;:; ca N C 0 Q) > :;:; ca Q) a:: 30 20 10 0 7.0 7.5 8 0 8.5 9.0 9.5 10.0 Ionization Energy (eV) Figure 5-11. Relative ionization factor in helium versus ionization energy 161

PAGE 170

16 2 alt rnati e to time delayed sampling techniques including the temporal aspects afforded b using the microsecond pulsed glow discharge. The ability of helium gas more pecifically the helium metastable atom to be a less selecti e ionizing agent reduces th need for RSFs in quantitative GDMS. The helium discharge has shown to have greater ionization efficiency as compared to the argon discharge specifically for difficult elements such arsenic and selenium.

PAGE 171

CHAPTER6 LU TER FORMATION AND IONIZATION IN A -PULSED GD Introduction The primary focus of this dissertation is the study of the fundamental aspect of the pulsed helium glow discharge and how applying them can provide analytical utility ersus the s-pulsed argon glow discharge. Sputtering and ionization are the two critical aspects in determining the analytical nature of a glow discharge ion source. This chapter discusses sputtering based on ion signal observed and the aspects of the s-pulsed GD plasma that influence not only the intensity of the signal observed but also what specie are ionized To discern differences between the sputtering of argon and helium more than just the atomic ion population must be considered. The surface interaction of a target and an incident ion either helium or argon, are considered here along with the influence of the glow discharge parameters. The processes that occur after contact of the ion and target are critical to the analytical utility of the s-pulsed helium GD. The glow discharge is a somewhat unique device in that the vaporization and ioni zatio n of the sample take place within the same volume without the chemical memory effects experienced in a technique like SIMS. In SIMS the ionization tep occurs simultaneously with the vaporization step where as in a glow discharg ioni zatio n of the putt red pecies requir diffusion of atom through a virt ual dark pace into a region rich in meta table ga atom Henc th r 1 a parati n in th tim 163

PAGE 172

164 o f th aporization and ionization events that virtually eliminates the chemical memor o f the sputt red sp cies to its matrix. Microsecond pulsed glow discharge mass spectra reveal that the ionized population is not wholly monoatomic. In nearly all spectra taken for a copper sample in an argon di charge dimers and trimers are observed. These are species that have typically shown up in glow discharge mass spectra ; however special attention wa g1 en here because these polyatomic species interfere with three elements : tellurium lead and bismuth. The major isotopes of tellurium are 126 128 and 130 which are directl y interfered with by the copper dimers at the same nominal masses. The lead i otope at 206 207 208 and the bismuth isotope at 209 are not directly interfered with by the copper trimer but rather by its combination with 0 OH and H 2 O. The greatest contribution of these is the Cu 3 0 which has masses at 205 207 and 209. These combinations are generally formed by oxides already attached to the urface or in a g a phase association reaction. Normally another lead isotope either 206 or 208 could be u se d for quantification but because there are small contribution from other trimer combination s, analysis is difficult. Background or th e ak e of 1 m ntal analy i and quantification a tudy of th formation of th matri x pol y atomic wa und rtak n. tud y in g th formati n of dim r i n t n nor i th u fa g low di char g e for po t ioni z ati n of th dim r and trim r b y 1 ctr n d bl . 79-83 d. imp ac t a n m t a ta 1 m za t1 n. tu 1 of i n b mbardm nt in low pr ur

PAGE 173

1 65 di charg utiliz d pre ure in th 1-10 mtorr range cau ing lar g bombardin g n r g 1 b cau of large dark space di tance Thi means th el ctric fi Id gra di nt can affi ct the en rgy of the ion to a greater xtent as compared to th hi g h r pr ur di charg used in the e xperiments. The formation of polyatomic speci s in a hi g her pr ur di charge has not been considered mainly because of the ubsequent interaction of the dimer with th glow discharge plasma. The formation mechanism of pol yatom ic copper pecies is discussed in a pulsed glow discharge as well as their rol in the GD plasma The formation diffusion and ionization of these polyatomic specie has not been studied in the s-pulsed glow discharge that is typically used for analytical mea urements. Dimers have been used as a means of quantification in a de GD b y Barshick and Eyler. 84 In much lower pressure (1-10 mtorr) studies post ionization b y metastables and electrons is small in comparison to the higher pressure plasmas becaus of the smaller population of gas atoms. A low population of gas atoms is advantageous for ion dimer measurement because there is very little dissociation through meta sta bl e collisions of the dimer species The effect of increasing voltage on bombarding ion energy and its effect on dimer sputtering has been discussed using a Langmuir prob to m eas ur e th actual ion energy and species flux from the surface. 85 Th ffect that pul e di charge parameters have on diffusion di sociation and ioni za tion of th man analytical glow discharg pla ma has y t to be tudi d. It ha b n r cogni z d that th r are sputtering differenc e betwe e n argon and h lium and by hara t ri z in g th fi rm ti n of thes e polyatomic in ight into the b neficial u f h liwn a th di h arg b discerned. Mas p ctra taken in both h lium and arg n pla ma ill d t rmin th

PAGE 174

166 relati e role that various formation mechanisms have on the population of polyatomic pec1es. Sputtering Theory Sputtering is caused by the impingement of a discharge gas ion on the surface of a metal target in our case copper. In Chap. 2 there is a brief discussion of what occurs during the collision cascade that results from ion bombardment and the factors that affect sputtering comparing the process to the break in a game of billiards As tated before the energy with which this ion bombards the surface is going to affect the processes that occur during this collision cascade. Various factors including pressure and vo lta ge wi ll affect this incident ion energy in the glow discharge. In the case of glow discharge analysis, the effect that these parameters have on the result of a single impacting e ent i not measured but rather the net effect of countless bombarding events. Therefore determining the formation mechanisms in the GD will be difficult d pit indication ba ed on the data as to which processes play a major role in th formation of pol atomic pec1es. Wh nan atom or ion of mass 40 hits a metal target with 400 V nerg a c rtain numb r f atom will b r l a d from th s urfac becaus n u gh n rgy a pro id d t v rcom th binding n rgy of tho e atom with the urfac Th bindin 0 nerg i timat d to b 4 tim th h at of ublimation of th targ t I m nt. F r th rang of th binding en rgy i b twe n 11-33 V. In hap 2 W hn r plot illu trat d th f fi ct f h lium and arg n n putt r yi ld h wing that b hangin 0 th

PAGE 175

1 67 ma of the bombarding ion t 4 the amount of atom putter d i d er a d w h nth energy of th incident ion remain the sarn Figur 6-1 how experimental putt r yie ld s of argon and h lium on copper versu incident ion energy a produc d by W hn e r et al. 39 The average incident ion energy in a glow discharge i appro x imat e d a 1/ 3 o f th applied vo lt age 77 In the studies here the s-pulsed argon discharge would have incid e nt ion energies at about 500-600 e V and helium would have energies between 1-1.2 ke V. There is a point where only a certain amount of energy is transferred to the surface despite increases in incident energy. This changes with the size and mass of the incident ion yet the incident energies for the helium and argon discharges are well below this threshold. The sputter yield from Figure 6-1 can be extrapolated for helium to determine a more comparable value than the sputter yield obtained at 400 e V. At 1100 e V the sputter yield for helium is about 0.39 Based on incident ion energy the sputter rate of argon is predicted to be 5-6 fold greater than helium rather than the predicted 10 fold difference if only va lu es at 500 eV for both gases were used Weight loss measurements for each pulse indicate a 7 fold difference of argon over helium (see Chap. 4). These Wehner plots predict the number of atoms ejected in a single bombardin g vent and from Fig 6-1 the value is about 2.2 for argon at 500 eV. These plot do not indicate whether bound atoms are released as single entities nor what happens immediately after atom ejection. It is these concerns that have led to variou tudie o f formation of polyatomic pec1 s using ion bombardment. There have been ome tudies and comparisons of th putter yi ld and dim e r population in ion bombardment experiment along with variou r ar hint th

PAGE 176

,,-.._ c 0 ..... Cl) s 0 ,:,j ..._,, "'O Q) 5= "Q) ::J en 168 2.5 Ar 2.0 1.5 ./ 1.0 / 0 5 I .He A A 0.0 1- 0 100 200 300 400 500 600 Incident Ion Energy ( e V) Figure 6-1. Experimental sputter yields of argon and helium on a copper target [Ref. 39]

PAGE 177

mechani m of dimer formation in SIM and other po t ionizati n t chniqu ertainly the removal or reduction of these specie would be beneficial for glo 7 -8 3 1 69 discharge analy is. These dimers and trim rs are matrix related and th re re can t b e removed (as in a mass change) by simply changing the composition of the discharg e g a a we might do to remove gas related (ArO ArN ... etc .. ) interferences. By determinin g the formation mechanism in the glow discharge and how various paramet r ef:fi ct th e formation of polyatomics a reduction of their contribution in the mass spectrum could b e possible. This reduction wou l d increase the sensitivity for elements such as lead and tellurium Cluster Formation Theory D E . 8 7 -8 9 1rect m1ss10n In the direct emission event ,' the bond energy of the two atoms in a dimer pair must be sufficiently strong for the dimer pair to leave the surface as an entit y Th energ delivered to one atom will overcome its lattice energy and release it in an outward direction. The ejection of the atom will either result in the pulling of the other atom m th ca of trong bond energy or leave it behind in the case of weak bond nerg Th e refore targ e t materials with high interatomic bond str ngths are candidat for dir e ct mis 10n. v n then a urface molecule survives sputtering only for mall relati en rgie b twe nit atomic constituents. This condition i onl y fulfill d for lar 0 differ nee in th atomic ma and wh n the mom ntum of th incid nt i n 1 trans rr d to th h avier atom. Thi th or m illu trat d in Fi 0 ur 62 h a tr nu u

PAGE 178

a b Figure 6-2. a) Direct sputtering of bound copper atoms in argon. b) Atomic combination of non-bound atoms above the surface after ion bombardment 170

PAGE 179

1 7 1 r quir ment and is g n e rally wh y th direct emission ev nt i b coming l a ce pt e d a a mechanism in the formation of dimers and trimers at least in ion bombardm nt tudi Basically one atom is set in motion to an outward direction b y th e binar y collision cascade and under the above energy considerations the co-emis ion of the 2 nd particle can occur resulting in the sputtering of a dimer. Oechsner 86 used a et of theories to prove the unlikely hood of the direct emission of dimers from a ingle el e m e nt matrix. Set atom A in motion with energy W 1 ab and with atom B at rest. The energy of atom A relative atom B is (Eq 6-1 ) Co-emission of atom A is possible if W M is smaller than the binding energy of the cluster AB (Eq 6-2 ) Emission of AB must follow that W 1 ab is larger than the energy (U 0 ) with which the pair AB is bound to the surface wlab > U o ( Eq. 6-3 ) Because Wbi nd and U 0 are the same co-emission is only po sible when M >> M 8 therefore in an elemental matrix atomic combination must b the re ulting mechanism. Molecules formed from sputtering of target with low bond en rgi and r similar mass con tituents are proposed to be formed by atom j ct d from a in g l colli ion event in which th two atoms receive enough en rgy to overcom th ir bindin b nergy.

PAGE 180

172 Atomic ombination (A )-Recombination-Single Collision 87 92 The AC model a first proposed by Oechsner and Gerhard 8 7 and independentl elaborated on by Konnen et al. 9 1 and by Gerhard. 92 As illu trated in Figure 6-2B thi mechanism requires that when the bombarding ion strikes the surface only atom are released As such the atoms in the sputtered cloud resulting from ion bombardment a sociate after they have left the surface. This means that the requirement of atoms b ing neighbors is not applicable and the possibility of association with atoms from a different surface positions very close by is possible but very unlikely. The requirements for combination are fulfilled if atoms are properly correlated with respect to their individual momenta such that the surrounding molecular system is lower in energy (more favorable) in comparison with the single atoms. The sputtering ca cade i er hort (101 2 s) and therefore the combining atoms are always situated in time. The probability that atom 1 and atom 2 combine will depend on their kinetic properties assumed during the collision cascade the distribution of each ithin th colli s ion cascade volume and their corresponding interaction potential. 1 Th totalit f the se effect s can b d scrib din stationary terms and by the raw a rage numb r of atom in tead of individual ejecting events This means that the number f dimer form d i s a proportionality in which th combination depend nth t tal numb r of putt r d a t m of atom 1 and th total number of putter d at m f at m 2 Th total pr babilit y fi r th fi rmati n in th A mod l i ( q 6-5

PAGE 181

1 7 3 wh r a 1 2 and a 2 1 ar th combination probabilitie of atom 1 with atom 2 and 1 c r a and k i th a rag number of atoms ejected in a single colli ion ca cade Atomic Combination by Double Collision The third model is loosely called the double collision mechani m and was fir t postulated by Konnen et al. 9 1 and by Gerhard. 92 Instead of a single bombarding event releasing atoms or dimer pairs from the surface the double collision model proposes that the dimer pair be made up of atoms resulting from two independent collision cascade This requires the almost simultaneous bombardment of ions in a near surface collision in which the energy transfer causes the same results observed in the first two theories. It potentially means that the collision cascades of two separate incident ions affect the sam atom space in the lattice. The action of two atoms combining via two different colli ion cascades has been called clustering ," and the probability of such action the clust ring probability. The experimental evidence supporting such a mechanism is valid for sub keV ion bombardment but falls apart above that energy. The evidence concludes that yields of sputtered neutrals and ionic clusters show a power law dependence on the number of atoms in the cluster. The slope of the power law fit correlates with the total sputter yi Id of the target material/incident ion combination ( e Figure 6-3). A variety of molecular dynamic (MD) and statistical mod ling imulation ha been performed on the sputtering process in order to di cem th dim r fi rmati n mechanism. Re ults indicat that multiple colli ion model d n t a g r 11 ith M studies because of the non-comparabl kin tic n rgy di tribution pr du d up n i n

PAGE 182

QJ n Figure 6-3 A least squares fit to a power law for the relative yields of polyatomics versus cluster nuclearity [Ref.93] 174

PAGE 183

1 75 bombardm nt. It n arly impossible to model or predict the full influ nc f tw parat bombarding events on the resulting collision cascade in the ame lattice pace and in th same time frame. Statistical or recombination models predict a linear dependence of Y 2 / Y I on the sputter yield which is most commonly observed in experiments testing th AC model. As a whole experiments have shown that nearly 30 to 40 % of the sputtered cloud from bombardment can be made up of dimers when bombarding ion energy is from 1 5 to 3.5 keV. At lower energies, from 1 keV and below 2-10% of the sputtered cloud is estimated to be made up of these dimers and trimers taking into account all formation mechani ms. Theory Application to a Glow Discharge As mentioned a majority of sputtering studies and polyatomic formation studie were produced via ion guns. There have not been any glow discharge mass spectrometric studies on the formation of polyatomic sputtered species in helium and argon. The large majority of puttering studies use ion guns onto a target carrying no voltage including that of Wehner who uses 400 eV ion guns for both helium and argon. While w might con ider ion bombardment at the surface in a glow discharge to b very clo ely r lat d to thi ch me it still isn t the same because of factors such as th voltage appli d to th urface and the gradient it creates beyond the surface. It th opinion of thi auth r that the puttering of polyatomic peci is most lik ly af ct d by th iz of th in id nt i n

PAGE 184

176 and thi d pendence can therefore be used to reduce the effect of these species in the mas pectrum The glow discharge environment allows the elimination of one of the possible dimer formation mechanisms simplifying the model of bombardment in the glow discharge source. The sputtering mechanisms proposed above were de eloped ba ed on tudies performed in a vacuum using an ion gun. The low pressure environment in the glow discharge has few advantages in studying dimer formation Disadvantage include a wide range of bombarding ion energies and many dissociating collision However the processes that occur in the GD allow us to eliminate the double collision mod 1 a a possible dimer formation mechanism. The sample used in these studies is a flat disc sample 4 mm in diam t r ith a total area of 1.26 105 m 2 The atomic radius of a copper atom is 1 .4 101 0 m. uppo em a single pul e 2.73 x 10 12 atoms are sputtered (ba ed on sputtering m a ur din thi dissertation). If a square planar arrangement of copper atoms exi t on th urfac (resulting in 68% coverage of the total surface) of the ol id di c th nth total urfac f ur amp! w uld contain approximately 1.39 x 10 1 4 atom Thi m an that nl 1. 6% f th urfac atoms would b puttered in a single pul If th bombardm nt of th urfac i uniform in tim and in pac th n th ar a or di tanc b tw n b mbardin g nt arg n ha a putter y i Id gr at r than 1 ig 6-1) uld b qual t 7 pp rat mi di tan Thi i c 1 ul t db di ding th t tal urfac ar a f th ath d by th numb r atom putt r din a ingl pul a wning that putt r n at m r add d n t ma al c n id r th b mbardm nt r 10 in hi h

PAGE 185

1 77 th numb r of copp r atom di tance would urely incr a e. hi m an th a t th d ubl colli ion model i liminated imply becau e there i too much pace b e tw e n c lli i o n to affect the ame copper atom during the collision ca cade. Thi leaves onl y th in g l bombarding e ent to form polyatomic species .. As discussed above the ability to transfer energy to the surface and break lattic e bond is critical in sputtering atoms and will determine the amount of sputtered atom s released. There is similar probability of bound atoms and non-bound atoms to end up in a dimer pair in both the direct sputtering and the ACM theories. Consider Figure 6-4 twelve possible next neighbor combinations are listed based on a square planar arrangement of atoms at the surface and these atoms can either be bound in the matrix and sputter released as an entity or be non-bound atoms and released and recombined just above the surface if only dimer formation is considered. Of course there i a much greater probability that only atoms are ejected upon single ion bombardment for the energies of the glow discharge. In a helium discharge increased pressure and therefore increased current mean more bombarding events compared to that in argon. The helium discharge u tain a current that is approximately twice that of the argon discharge under anal tical conditions meaning approximately twice the number of ion bombard the urfac in th 10 s pul e period. The increased current in helium results in an 11 at m di tanc between bombarding events based on the same calculations a th e r ar 0 on Th putter yield for helium on copper shows that d spit the numb r f b mbardin 0 nt

PAGE 186

178 Figure 6-4. Nearest neighbors combinations to bombarded atom #1: 1) 2-3, 2) 2-4, 3) 3-4, 4) 4-5, 5) 4-6, 6) 5-6, 7) 6-7, 8) 6-8, 9) 78, 10) 8-9, 11) 8-2, 12) 9-2

PAGE 187

1 79 and de pit th ion energy ( extrapolated to 1 ke V) that on averag l than n at m 1 produced per ion bombarding event (Fig 6-1) Several factors influence sputtering one of which is the ability to transfer e n e r gy to the lattice as determined by the energy transfer function. Helium ions will tran fi r energy less efficiently than argon ions and the effect that this has on puttering i illustrated in Figure 6-SA and 6-SB. In A the amount of energy transferred by the helium ion is not great enough to cause the release of bound copper atoms simply because the lattice energy of the second atom is stronger than the bond energy between two copper atoms which is just strong enough to prevent bond breaking and ejection of the first atom. It's been stated however that the direct emission mechanism likely doesn t occur for the pure copper matrix. The end result in Fig A is a rearrangement of the top layer of copper atoms. In B a single atom is released but there is not enough energy to break the lattice binding energy of another atom in order for atomic combination to occur above the surface. Figure 6-6 shows the non-bound atom release scenario for the argon ion bombardment. Figures 6-2 through 6-6 could explain the difference i putter rate between argon and helium. Considering Figures 6-2 and 6-4 through 6-6 th r i a possibility that a single bombardment by an argon ion could 1 ad to 1 of ix po ibl at m r lease ev nt while for helium only one possible atom release v nt xi t Th figur are only approximations of what really occurs but th y refl ct th proc esses th a t differ betw en argon and h lium for atom and dim r putt ring m chani m and m a also explain th 7 fold differ nee in sputter rat ob rv d b t nth tw a

PAGE 188

180 0 a 0 ~ -. .. b -----Figure 6-5. a) Direct sputtering of bound copper atoms in helium b) single atom ejection by helium bombardment

PAGE 189

181 0 0 Figure 6-6. Sputtering of non bound atoms by an argon ion

PAGE 190

182 Diffusion, Dissociation, and Ionization of Polyatomics The glow discharge gas is not only responsible for the amount of pol atomic copper species formed but also for their subsequent destruction in the glow di charge plasma. Once formed polyatomic copper species diffuse across the GD dark space into the negative glow. In this region collisions result in either dissociation or ionization Changes in voltage and pressure will show why the collision rich negative glow i so important in the removal of these species from the mass spectrum. This is of particular interest for the detection of lead and tellurium. The copper ion signal can be used as an indicator of how net sputtering is affected by changes in glow discharge parameters. However it is ultimately weight loss tudie that prove these effects on net sputtering Copper atomic ion signals are similar for both argon and helium pulsed GD plasmas (Chap 4) yet argon sputter 7 times more atom than helium. This means that the ionization efficiency of the helium plasma is 7 time greater than argon. Therefore, species formed in the plasma that have I.E s clo e to that of the copper atom (7. 87 e V) will be subject to the same ionization efficiency For comparison of sputtering efficiency between argon and helium (without con idering putter rate) the copper dimer and trimer populations can be u d for sputtering compar1 on b caus th ir formation i a dir ctr sult of th int raction of ach ga 1 n and the cath d urfac A impl pr of how that th dim rand trim r ion ignal mu t b th am for b th ar g n and h lium if th bulk ion ignal ar th am und r ptimal condition if w ar to a um that th dir ct mi ion v nt ha no b aring on th p pulation of dim r

PAGE 191

18 3 and trim rs. The recombination model would hav to pr diet that dimer signa l in h lium and argon are the same if the putter yield for helium wa a bo v one. Becau e th putt er yield at 1-1.2 keV is well below one the recombination model i ineffectiv for predicting the similarity in dimer signal for each gas. B y lookin g at the spectra under optimal conditions in Figure 67 we see that the dimer signal is approximately 10 time higher in argon than in helium. The small mass difference between the ma s p ectra i most likely due to small variations in the flight tube voltage. The energy of the ions is not responsible for this shift only because the orthogonal pulse of ions into the time-of-flight should render all kinetic energy equal. Figure 6-7 shows the combination of a Cu 2 and an OH species formed as explained before This combination is used because the analysis of the dimer is affected by the presence of tellurium. This would imply that more dimers are produced in the argon discharge per incident ion event than in the helium under optimal conditions. In other words helium doesn t have enough momentum ( a function of mass ) or energy to sputter these larger polyatomic species. Under the same conditions no signal is d etected for the trimer in the helium discharge while the visible isotope pattern for the copp r trimer at m/z 189 191 193 is apparent for the argon discharge in Figur 6-8 The calculated distribution is shown in Figure 6-8B. This calculation i ba d on a binomial expansion relative to the number of atoms in the cluster and the natural isotop e abundance. In other words there is only one way to produce 63 Cu 3 wh r a th r ar three way to obtain 63 Cu 2 65 Cu ; and because the 65 u i otop i mall r in a bund anc the

PAGE 192

400 300 en 200 ..... C: :l 0 (.) 100 0 63 Cu 2 0H t. , I I I I I I I I I I I I :, , o I , I\ I I I I I I I I '' I I ' I I I I I I I ,_ I ' ,, ~ \, , ............ ... 144 ; " " , '' I I '' I I Io ' ' ' ' I I I mlz \ . ... ... ,_ ... -, _ Ar --He 65 Cu 2 0H I _,,_,,, ,, ,, I I I ' I I I 147 Figure 67. Dimer ion signal for helium and argon pulsed GD 184

PAGE 193

u, C ::l 0 (.) 63Ct1i6scu + Ar I He I \ 63Cu + ,, 40 ,, 3 ,, ,, I 1 }, I I I I ', I I I \ I I I I I I 63Cu6scl + I I I I I I I I I I I II 20 I I I II I I I I II I I I I I I I I I I I I I I I I ,. \ I I I I 65Cu + I I I I I I 3 I I I ,, I 0 188 190 192 194 M/Z Figure 6-8. a) Copper trimer ion signal in argon and helium s-pulsed GD 185

PAGE 194

186 100. 90 80. 70 . 60. 50. 40~ 30. 201 o. 0 1 I I I I I 188 190 192 194 196 Figure 6-8. b) Calculated isotope distribution for Cu 3 +

PAGE 195

1 87 di tribution i not three times larger than the 63 u 3 di tributi n Th se calculation r e p e rformed u ing Isopro (Cornell University). Th data don t prove the dominance of one mechani s m over another b cau e the application of the AC model to helium is ineffective and predict no dim er formation. However both mechanisms may contribute to the intensity with which copper polyatomics how up in the mass spectrum for argon. The negative glow region be gins about 0.5 mm be y ond the surface and extends beyond 1 cm for both argon and h e lium Measuring ions from the matrix beyond about 7 or 8 mm is difficult because the distribution of ions has become so diffuse. The region from 3 -7 mm above the cathode surface is generally considered the analytical zone for mass spectrometry. However we can take advantage of the sampling distance to study some of the fundamental aspects of sputtering. B y choosing distances shorter than 3 mm we obtain information about dimers and trimers and possible sputtering mechanisms. In this section th e sam plin g distance is varied against the parameters of voltage and pressure for both argon and helium. It will be immediatel y evident that the distribution of copper dimer and trimer ions is different than copper atomic ions with respect to the distanc e above the s ur fa e. While not an expected result it provides insight into the advantage of bein g able to ar the ampling distance to tudy fundamental and anal yt ical a p ect of both th h lium and argon plasma.

PAGE 196

188 Figure 6-9 shows the ion signal from the 1 3 Cu 2 ion peak and the 58 i ion peak ersus cathode to sampling orifice distance. The copper dimer ha three i otope p ak at 126 128 and 130. The distribution of polyatomic species is different than that ofth atomic species. What is surprising is that all these species are formed in the same 10 pulse and diffuse the same distance to the sampling point. The question i : why are the distributions different? The data indicate that the polyatomic species are most lik 1 dissociated into copper atoms as they move through the collision-rich di charge environment. It must be noted that at the time that sampling occurs for th e polyatomics the discharge has been off for more than 70 s. till excit d meta table and fast argon atoms exist to break up polyatomics and / or to ioni ze them. If the ga atoms and metastables were not responsible for the breakdown of the polyatomic th n the distribution of the atoms and polyatomics would be the same. It eem that th distribution of sputtered species indicates that th polyatomic b ing brok n do n int atom are contribut in g to the copper ion signal via meta table ioni zation con r ati timat of the population of these polyatomic i betw n 2 7 % f th t tal putt r d i n p pulation Thi m ans that in th -puls d di charge th r mi 0 ht b a 1 of i ni za ti n ffici ncy f trac lement in th ample becau m ta tabl at m ar partl r p n ibl fi r th br akd wn f th p l yatom ic Th r fi r b r du ing r limin tin the putt rin g or fi rmation f th p l yatomic an incr a n iti it f up t 2 -7 % mi g ht b p ibl ch mg c nditi n and pla ma ga b c m ignificant r tra

PAGE 197

co I.() I Q) ~ z 189 2200 63Cu 800 2000 2 58Ni+ 1800 600 1600 () 0 "O 400 1400 .., 0 ~f 1200 CD 200 .., 1000 800 D -----. 0 600 1 0 1 5 2 0 2.5 3 0 3 5 4 0 Cathode-to-Orifice Distance (mm) Figure 6-9. The distribution of atomic and polyatomic species m the s-pulsed GD

PAGE 198

190 anal i e pecially if up to 7% of ionizing agent (argon) i con urned in the di ociation and/or ionization of unwanted species such as these copper polyatomics. Figure 6-10 shows mass spectra of lead and bismuth ith Cu 3 O + a th pnmar interfering species. The figure shows the utility of the being able to adju t the ampling distance in order to obtain atomic and polyatomic information in a ingle ource. More often than not commercial instrumentation and other glo di charg ion sources require a fixed cathode to sampling orifice distance. It also illustrates the abilit of the glow discharge plasma to atomize / dissociate most molecular pecie through multiple collisions. At 1.6 mm the spectrum clearly shows the strong isotope pattern of a Cu 3 species with masses at 205 207, 209, and 211. Masses for lead at 206 and 208 ar visible at 1.6 mm but don t have the correct isotope ratio most like! b cau e of interferences from Cu 3 OH +. As the C-O distance is increased to 3.5 mm th contributi n of Cu 3 + i diminished at least 10 fold leaving only atomic lead and bi muth p c1 As before sma ll shifts are observed in these mass spectra. Th mall hift ob er d at mas e 206 and 208 can be contributed to small day to day chang in flight tub olta 0 r defl ction voltage as the e mass both repre ent 1 ad i otop larg r hift i b erv d at ma 207 The hift her can be contribut d to th di ffi r nc m ma b tw nth u 3 and th 207 Pb + Additionally th re i n t unit r luti p ak at 20 and th u 3 O + p ak at 207 indicatin g th diffi r nc in p c1 b th r p cti ath d -t rifi di tan h r du ti n of th u ignal a cau d b in r a ing th di tanc H w r it i n t ju t th chang in di tanc that f ct th chang in ign I rath r it i

PAGE 199

"' .... C :J 0 u \ ., ,, I I 200 I I I I I I I I I I I I I I I I I I I I 100 I I I I I I I I I I I I I I 204 206 ,, I I I I I I I I I I I I I I I I I I I I I I I I I mlz I I I I I I I I I I --Pb Bi at 3 5 mm Cu 3 O at 1.6 mm ,, I I I I I I I I I I / I I I I I I I I 208 210 Figure 6-10. Lead and bismuth at 3.5 mm and Cu 3 O + signal at 1.6 mm 191

PAGE 200

192 the additional plasma dissociation that reduces the signal and allows the detection of lead and bismuth. The same cannot be said for the analysis of tellurium and its primary interference of the copper dimer. Figure 6-11 shows that the same 10 fold decrease in polyatomic signal can be caused by changing the C-O distance from 1.6mm to 3 .5 mm. However there still remains a high enough signal for the Cu 2 species to affect tellurium analysis. Figure 6-12 shows a mass spectrum of Te at 3.5 mm for both argon and helium. The isotope features for tellurium are still interfered with slightly in helium; however the isotope distribution of the copper dimer is no longer observed, indicating a reduction in the sputtering of these polyatomic dimers in helium This is one of the advantages of using helium as a plasma gas. The reduction of polyatomic sputtered species allow th complete analysis of elements from mass 45 and above with little cone rn for isobaric ga species (defined here as helium+ air leak gas impurities) interferences and with minimal matrix species interferences ( defined here as Cun+ discharge gas and / or air leak ga impurities). Th ffect of pressure on polyatomic formation As pr s ure incr ase the contribution of polyatomic in th ma p ctrum d er a s as n in Figur 6l 3A. Quite contrary to that i the f ct pr ur ha on th atomic ion p pulation which incr a e with pr ure and con qu ntl curr nt (Figur 13B) Th incr a m copp r ion signal i du to an incr a d numb r f b mbarding v nt at th ath d ar af ct d wh n pr ur 1 incr a d: 1) th av rag n rgy f th bombarding p

PAGE 201

,, I I : I I I I I I 1000 ti) .... C :::J 0 (..) I I I I I I I I I I I I I I I I, I I I I I I I '., I I I I I \ I I I I I ., / 0 126 I I I I I ,. ,, , I I I I I I I I I I I I I I I I 128 m/z I \ I I \ ., --3.5 mm 1.6mm I I I ~ , I I I I I I I I I I I I I I I I / I / I I I I I \ I / __ .,. 130 193 Figure 6-11. Mass spectra of Te + and Cl + in argon at 1.6 mm and 3.5 mm

PAGE 202

194 150 l 1----I ,, II ,, II II ,, 11 , 1 I I I I I I I 100 I Cl) C ::::, 0 I (.) II 50 -! I I I I I I ,, I I / I I I / I I I \ I I I \ 0 126 127 128 129 130 m/z Figure 6-12. Mass spectra of Te + in helium and argon at 3.5 mm

PAGE 203

a 1400 1 / 0 126 b 1800 Cl) c 8 900 0 54 127 1?R m/z m/z {\ 129 0 8 Torr Ar 1 Torr Ar 1 25 Torr Ar 130 1.25 Torr Ar 1 Torr Ar 0 8 Torr Ar 57 Figure 6-13. The effect of pressure on a) copper dimer and b) . rron ions 195

PAGE 204

196 i reduc d b cause it experiences more collisions on its acceleration path to the cathode urfac due to a greater atom density and 2) sputtered pecies encounter a gr at r number of collisions between the sample surface and the sampling orifice A reduction in bombarding energy will result in a decrease in the sputter yield as illustrated in Figur 61. This also means that there is less energy to transfer to the surface and a small r probability of sputtering dimers as single entities and even less probability of putt ring enough atoms for atomic combination. As the polyatomic specie diffuse thought the plasma the effect of pressure is two-fold: 1) a higher pressure will decrease the energ of the collisions in the plasma and 2) there will be more collision The dissociation nerg of copper dimers is 2 eV. Therefore, any collision that has energy b tween 2 eV and 7 89 eV will result in atomization/dissociation. Collisions with energ abo e 7 89 V ill result in ionization. Hence the overall effect of pressure is a reduction in the contribution of polyatomic pecies and an increase in atomic specie Th effect of voltage As oltage i increa d the average bombarding ion n rg incr a hich m an an mer a in sputter yi ld is f ct d a hown in Figur 6-1 It al o m an that n rgy that can b tran ferr d to th urfac at th cath d er ating a gr at r p ibility of putt ring or forming polyatomic p c1 at l a tin arg n. th p I at m1 p c1 diffu int th pla ma the ill fa high r a lta g pp ur ltag ill m r g r at r i ni z ati n f th p l at m1 p b au f high r

PAGE 205

1 97 av rag n rgy of colli ion Figw-e 6-14 shows an increa in copper dim r i g n a l ith an incr a e in voltage. Originally it was assumed that a higher energy pla ma w ould upport a greater dis ociation of polyatomic species however the spectra indicat different! y. Summary This chapter defined the possible formation mechanisms of polyatomic matrix species in a pulsed glow discharge plasma It seems that the AC model is the primary mechanism for their formation in this ion source arrangement. However direct sputtering may still play a role in dimer formation in argon. The transfer of energy of the incident ion is the most critical factor in determining what and how many species are ejected from the copper target. The inability of the helium ion to transfer energy efficiently resulted in low signal of matrix polyatomics and showed the potential analysis of tellurium lead and bismuth. The diffusion dissociation and ionization of polyatomics in the argon discharge were discussed. The distribution of matrix polyatomic peci 1 different than atomic species most likely a result of the breakdown of the polyatomic dimers and trimers once exposed to plasma collisions The glow discharge i very g ood at atomizing mo t ga eous polyatomics and results in mass spectra relati el fr fr m matri x specie a we observed in the analysis of lead

PAGE 206

a 150 f 1 I I A 100 j l 8 50 125 126 127 b 160 8 80 0 116 117 ,~ I I I I I 128 m/z 118 m/z 198 2.5kV 2.75kV 3.0kV 129 130 1 I 11 119 120 Figure 6-14. The effect of voltage on a) copper dimer and b) tin at --2.5 kV, -2.75 kV, and -3.0 kV

PAGE 207

CHAPT R 7 ONCLUDING REMARKS AND FUTURE DIRECTION This dissertation has focused on the differences in the ionization and putt rin g processes that occur in argon and helium s-pulsed glow discharges. In order to produc ome of the same processes in argon a helium discharge must be run at about 5 tim es the pressure of that in argon. The higher pressure results in equivalent mean free paths for both gases and creates a similar diffusion environment for sputtered species. A higher voltage is neces ary to sustain the helium plasma This results in a higher energy plasma having higher current. There are several factors that are beneficial when using helium compared to argon. First isobaric interferences observed with argon can be removed without a los m sensitivity. The ionization processes in argon and helium which cause interferences are different. In argon metastable ionization is the primary mechanism of sputtered specie ionization after the pulse period. Secondary and associative ionization mechanism dominate in the formation of interferences because the metastable atom energy is not high enough to ionize impurity gas species. In helium metastable ioni z ation i responsible for both sputtered species and interference ionization becau e it meta tabl e atom do have nough nergy to ioni z most peci in the glow di charg On of th primar y r a on why argon is u ed a a pla ma gas i b cau o f th i ni z ation di crimination again t ga impurity p cie 199

PAGE 208

200 The s-pulsed argon glow discharge is capable of discriminating against most discharge gas species. However it s been shown here that argon ions still exist after the pulsed period and may contribute to formation of isobaric interferences. By using helium as the discharge gas better isotope ratios for iron were obtained. In addition it was shown that the determination of calcium with an argon discharge is impossible whereas the use of helium can eliminate this interference without any loss in sensitivity. Relative sensitivity factors and relative ionization factors were determined for both gases. The RSFs showed that helium is a less selective ionizing agent compared with argon most likely because of its higher metastable ion energy. The helium discharge was able to ionize elements with higher ionization energies more efficiently than argon, and provide increased sensitivity for these elements resulting in a reduction in RSFs Aside from a few elements most RSFs were close to 1 Sputtering differences in argon were studied on two le ve ls Sputter rat determined and showed that argon sputters about seven times more efficiently than h lium i.e. the sputter rate per pulse was seven times great r in argon. The am e l ementa l sensitivi ty was achieved for both discharge ga e hawin g that h lium ioni z a given puttered population seven times more efficiently than argon a nd l e a di cu ed abov lu t r formation wa di cu ed with re p ct to argon ion and ah lium ion b m barding th urfac ffici ntly t th urfac b cau of it mall ize and h nee mall individual mom ntum. H lium ion d n t hav th mom ntum that argon ion hav wh n triking th urfac Th

PAGE 209

20 1 application of th clu t r formation to a glow di charg how d that th d ubl lli i n mod l a an unlikely candidat for dimer formation. Th re wa no e x p rim nt a l pr f that th dir ct emission mechanism contributes to dimer formati n ; how v r Oech n r 86 pro d that ingle elem nt targets could not not have single entity ejection This m e an that the atomic combination of nearest neighbor sputtered atoms i the mo t likel mechanism for polyatomic formation in a glow discharge. Incident ion energy was determined to be the most critical factor in polyatomic formation in a glow discharge. This energy is affected by both pressure and voltage of the glow discharge. Higher pressures resulted in a reduction in polyatomic signal while higher voltage resulted in an increse in signal for the same species. By taking advantage of the distribution of sputtered species in the s-pulsed plasma the detection of lead was affected. The detection of tellurium was much more difficult in argon. Plasma dissociation did not reduce the dimer signal to reveal Te isotopes at 126 128 and 130. However tellurium could be analyzed by using helium to discriminate against the formation of polyatomic species. Future direction for the this work include: 1) Thin-layer analysis with helium. The Grimm-type source require much higher pressures than the simple diode DIP ion source. Therefore. atomic mi ion would work well but some severe vacuum limitations exist for ma sp ctrom tr 2) additional puttering studies on polyatomic formation. Th u f n on uld b interesting becau e of ome of the imilariti betwe n it and h lium in t m1 f energetic Hydrogen (H 2 ) would mak an int r ting compari on to that f h liun1

PAGE 210

2 0 2 3 mi x ed helium and argon studies. This is an acti e area of res arch for glow di charges. The moderate sputtering rate of argon combined with th ionization efficiency of helium can create some very high sensitivities for ome el ment om e of this work has already been performed. It seems that the addition of 10 30 and 50 % helium to an already sustained argon glow discharge doe n t affect the net sputtering. The signals for most elements are enhanced with increased ignal for e As and Au These are the same enhancements observed in th helium di charge meaning that ionization efficiency is increased with the addition of the helium to the argon plasma. Future experiments might include a look at th effect of helium pres ure on bombarding energy and the detection of polyatomic

PAGE 211

REFERENCE 1. Thomson J.J. Rays of Positive Electricity and Their Application to hemical Analysis; Longmans Green and Co. : London 1913. 2. Paul W. and Stinwedel H. Z. Naturforsch 1953 8a 448 3 Aston F. M. Proc. Roy. Soc. A. 1937 163 391. 4. Thomson J.J. Phil Mag. 1921 42 857. 5. Dempster A.J. Phys. Rev. 1918 11 316. 6. Dempster A.J. MDDC 370 U.S. Department of Commerce Washington DC 1946. 7. Boumans P.W.J.M. in Ana l ytical Emission Spectroscopy: Part II ; Grove E.L. Ed. ; Marcel Dekker Inc.: New York 1972; Chapter 6. 8. Ramendik G and Verlinden J.; Gijbels, R. in Inorganic Mass Spectrometry; Adams F.; Gijbels R; Van Grieken, R. Ed.; John Wiley and Sons: New York 1988 Chapter 2. 9. Thomson J.J. Phil. Mag. 1910 20 752. 10. Herzog R. and Viehbock F.P., Phys. Rev. 1949 76 855L. 11. Magee C.W. Harrington W L. and Honig R.E. Rev Sci Instrum. 1978 49 477. 12. Wittmaack K Maul J. and Schulz F. Int. J Mass Spectrom. Ion Ph ys 1973 11 35. 13. Houk R.S. Fassel V.A. Flesch G.D., Svec H.J. Anal hem. 1980 52 2283. 14 Douglas D.J. French J.B. Anal. Chem. 1981 53 37 15. Pack B.W and Hieftje G.M. Spectrochim Acta 52B 1997 2163. 16. Paschen F. Ann. Phys.1961 50 901. 17. Grimm W. Spectrochim. Acta Part B 1968 23 413. 203

PAGE 212

204 18. oburn J.W. and Kay E. Appl. Phy Lett. 1971 19 350. 19. Hang W. and Harrison W.W. JAAS. 1996 11 835. 20. Hang W. Baker C. Smith B W Winefordner J.D. and Harri on W. JAA 1997 12 143. 21. Hang W Walden W.O. Harrison W.W. Anal.Chem. 1996 69 1148. 22. Pollman D. Ingeneri K. and Harrison W.W. JAAS. 1996 11 849. 23. Hang W. Yang P. Wang X., Yang C Su Y. and Huang B Rap. ommun. Ma Spectrom. 1994 8 590. 24. Howatson A.M. An Introduction to Gas Discharges Pergamon Pre s: Elmsford Y. 1976 25. Morrison J. A. and Ede l son D. J Appl Phys. 1962 33 1714. 26 Iida Y ., Daishima S. and Kanda F. Anal. Sci 1993 3 391. 27. Mahan C.A. JAAS. 1997 12 247. 28. Florian K. Habler J. and Surova E. JAAS 1999 14 559 29. Becker J. and Dietze H.J. Int. J Ma ,pectrom. 2000 197 I 1. 30. Chapman B. Glow Discharge Processes. John Wi l ey and on : N w York 1980 figure adapted from Nasser (1971). 31. Bo gaert and Gijbels R. J Appl. Phy. 1996 79 1279. 32. i gmu nd P Phys R ev 1969 184 383 33. uman P.W.J.M.Anal. h e m 1972 44 1219. 34. h n r H. A. Ph y ic 1973 261 37. 5. ar hi k .M. H K R Z k . and Kin g F .. App l. pee. l 9 53 5. 6. hapt r 7 H in mann

PAGE 213

205 37. Im n 0. and Bruce G Nu cl. In trum Meth. 1961 11 279 38. W hn r G.K. Ph y R ev. 1956 102 690. 39. Betz G and Wehner G.K. in : Sputtering by Particle Bombardm e nt II. R. B hri ch Ed. Springer-Verlag: Berlin 1983. 40 Marcu R. K. Glow Discharge Spectroscopies Plenum Pres s New York 1993. 41. Dang D and Marcus R.K . Spectrochim Ac ta 45 B 1990 1053. 42. Vieth W. and Huneke J.C. Spectrochim Ac ta 46 B 1991 137. 43 Bogaerts A and Gijbels R. A nal. Chem. 1996 68 2676. 44. Hess K.R and Harrison W W. Anal. C h e m. 1988 60 691. 45 Palmer A.J and McGowan. Appl Ph ys. 1972 43 4084. 46 Simth R L. Serxner D. and Hess K.R. Anal. Chem. 1989 61 1103. 47 Levy M.K. Serxner D Angstadt R. L. Smith and Hess R.K. Spectrochim Acta 46B 1991 253 48. Ferreira N. P. Strauss J.A. and Human H G.C ., Spectrochim. Acta. 37B 198 2 273. 49. Piper L.G ., Vela z co J.E. and Stetzer D W. J. Chem Ph y 197 3 59 3323. 50. Wagatsuma K. and Hirokawa K. Spectrochim Acta 46 B. 1991 269. 51. Bengston A and Lundho lm M. J AAS 1988 3 879. 52. Weston G F Cold Cathode Glow Discharge Tubes ILIFF E Book s Ltd. Lon d on 1968 pp. 107-111. 53. Marcu R. K. and Harrison W W. Spectrochim Acta 40B 1985 933. 54. Bogaerts A. and Gijbels R. Freseniu J A nal Chem 1996 355 85 3. 55. Boga ert A., Guenard R ., mith B. Winefordn e r J.D Harri on W. and ijb I R ,p ec tr ochim. Acta 5 2 B 1997 219. 56. Bogaert A Gijb I R. and Goehe r W. Anal. h e m 1 9 6 68 22

PAGE 214

206 57. ch 11 W and Van Griek n R. JAAS 1997 12 49. 58. D G ndt Van Grieken R Hang W and Harrison W.W. J AA 1995 10 689. 59. Perez C. Pereiro R. Bordel N. anz -Medel A. JAA 1999 14 1413. 60. Pan X. Hu B. Ye Y. and Marcus R.K J AAS 1998 13 1159. 61. Walden W.O ., Hang W. Smith B.W. Winefordner J.D. and Harr ion W.W Freseniu J Anal Chem., 1996 355 442 62. Yan X. Hang W. Smith B W Winefordner J.D. and Harri on W W J AA 1998 13 1033 63. Barshick C.M. Goodner K. Watson C. Ey ler J. Int J Ma pectrom. 1998 178 73. 64 Cotter R.J. Time-of-flight Mass Spectrometry: Instrumentation and Application m Biological Research ; American C h emica l Society Washington DC 1997 65. Wiley W.C. and McLaren I. H R ev. Sci Instrum. 1955 26 1150 66. Cotter R.J A nal. Chem. 1996 71 445A. 67. Bar s hick C.M Ph.D. Dissertation University of F lorid a 1991. 68. Nam Masamba R. and Montaser A. Anal. h e m 1993 65 2 784 69. Wagatsuma K. and Hirokawa K. Anal. hem 1988 60 70 2. 70. B nakk r .I.M. p ectrochim Ac ta 32B 1977 173 71. igli J. and aru o J.A. App l. '])ec. 1995 49 900 72 Ichikawa Y. and Teii J Phy D 1980 13 2031. 73 u z w kijr. J.P and Hi ftj .M. A nal. h e m 2000 7 2 3 81 2. p lnik Z. W agat uma K. Van Gri k n R. nd Vi R D Anal. h m pg.? . Yang .. and Harri n W W JAA In Pr 7 P t l B M. mith B W. and Win fi rdn r J. 'fJ tr him. A ta 40B 1985 119

PAGE 215

207 77. Davi W.D. and Vand rslice T.A. Phy Rev. 1963 131 219. 78. Chanin L.M and Biondi M.A. Phy Rev. 1961 122 843 79. Woodyard J.R. and Cooper C.B. J Appl. Phy 1964 38 1107. 80. Coburn J.W. and Kay E. Appl. Phys Lett. 1971 19 350. 81. Coburn J.W. Taglauer, and Kay E. J Appl. Phys. 1973 45 1779. 82. Coburn J.W. Taglauer and Kay E. Proc. 6 th Int. Vac. Congr. lap J Appl. Ph ys. Suppl.2 Pt. 1 1974 501. 83 Coburn J.W and Kay E. Appl. Phys. Lett. 1971 18 435 84. Goodner K.L. Eyler J.R. Barshick C.M. and Smith D.H. Int. J Mass Spectrom. Ion Proc 1995 147 65. 85 Hanison D.E. Avouris P. and Walkup R. Nucl. Instr And Meth. B 1987 18 349. 86. Oechsner H in : Physics of Ionized Gases 1984 Eds M Popovic and P. Krstic (World Scientific Singapore 1985) p. 571. 87. Oechsner H and Gerhard W. Surf Sci 1974 44 480. 88 Urbassek H.M. Nucl. Instr. Meth. B 1987 18 321. 89. Benninghoven A. and Muller A. Surf Sci 1973 39 416. 90. Hofer W.O. in Sputtering by Particle Bombardment III. Eds. R. Behri ch and K. Wittmaack (Springer Berlin 1991) p.15. 91. Konnen G P ., Tip A and de Vries A.E. Radiat. Eff. 1974 21 269. 92. G rhard W. Z. Ph y B. 1975 22 31. 93. Lill Th Calaway W.F ., Ma Z. and Pellin M.J. urf ci. 1995 322 361

PAGE 216

BIOGRAPHICAL SKETCH Matthew Mohill was born on March 22 1974 in San Diego A though he gr up in Portland OR. He is an avid outdoorsman who loves golf snow skiing hiking and whitewater sports such as rafting and kayaking. Matt attended Trinity College in Hartford CT as an undergraduate receiving his Bachelor of Arts d gree in 1996. In hi senior year he spent a semester at Oak Ridge National Lab working under Dr Christopher Barshick and considers it one of his best semesters in college. For the la t 4 ears he has been exploring the fundamentals of the s-pul ed h lium glo di charg and has plans to use his graduate knowledge and experience in upcoming endea or For th time being Matt will make his fifth cross-country journey back to Oregon in hi 1976 Toyota Landcruiser. 208

PAGE 217

I certify that I have read this study and that in my opinion it conform s t o accepta bl standards of scholarly presentation and is fully adequate in sc op e a n d quality as a dissertation of the degree of Doctor of Philosophy Harrison Will ar d W. P h.D Prof essor, hemi tr y I certify that I have read this study and that in my opinion it con fo rm s to acceptab l e standards of scholarly presentation and i s full y adequate, in scope and quality a s a dissertation of the degree of Doctor of Philosophy Winefordner J a es D u d l ey Ph.D Graduate Research Profe s sor Chemistry I certify that I have read this study and that in my opinion it confo1ms to a c ce pt able standards of scholarly presentation and is fully adequate, in scope and quality as a disse11ation of the degree of Doctor of Philosophy Eyler John Ro rt, Ph D. Profe s so r h emis try I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality a s a dissertation of the degr e e of Doctor of Philo ophy ~d A l an, Ph: Professor, Che mi try

PAGE 219

I certify that I have read thi study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality as a dissertation of the degree of Doctor of Philosophy ~
PAGE 221

a20 rs r 430 0 7/ 30 / 0 1 34760 SUr

PAGE 222

I UNIVERSITY OF FLORIDA II I II IIIII I Ill I l l ll l ll l l lll I I IIII II I I II II I I II I I I I II III I II I I III I I 3 1262 08554 4517