Aqueous solution sampling and the effects of water vapor in glow discharge mass spectrometry


Material Information

Aqueous solution sampling and the effects of water vapor in glow discharge mass spectrometry
Physical Description:
ix, 228 leaves : ill. ; 29 cm.
Ratliff, Philip H
Publication Date:


Subjects / Keywords:
Glow discharges   ( lcsh )
Mass Spectrometry   ( lcsh )
Trace analysis   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1992.
Includes bibliographical references (leaves 221-227)
Statement of Responsibility:
by Philip H. Ratliff.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001800846
oclc - 27669793
notis - AJM4604
sobekcm - AA00004745_00001
System ID:

Full Text







This work is dedicated to my parents, whose love
and support helped me get through
my many years of education.


I wish to express my gratitude to my advisor, Dr. Willard W. Harrison, for

his guidance and encouragement during my graduate career. The qualities that

he has instilled in me as a scientist and a person will not be forgotten. Thanks

to his commitment to the field of glow discharge mass spectrometry and staying

active in research, I have been allowed to attend many national scientific

meetings. This has not only made me a better speaker, but has permitted me to

grow professionally.

I would like to thank the faculty and staff in the University of Florida

Chemistry Department. I would particularly like to thank Evelyn Butler, Susan

Ciccarone, Jeanne Karably, Beverly Lisk, and Wanda Douglas for making the

bureaucracy a little easier to bear. The personnel in the machine shop and

electronics shop are acknowledged for their expertise in constructing many of the

devices required for my research.

I would like to thank my colleagues and friends with whom I have worked

over these years. My colleagues who have gone before me, Christopher

Barshick, Jeffrey Klingler, and Yuan Mei, are especially acknowledged for their

friendship and assistance in the laboratory. The levity they brought to the

research laboratory made every day an adventure. I also would like to thank my

current co-workers, Sue Ohorodnik and Bill Walden, who have had to put up with

me during the writing process when things can get a little hectic. I especially

thank Sue for proofreading this dissertation.

Finally, I would like to express my love and gratitude to my family who have

been very supportive during my graduate career. Although they never pushed me

to pursue this degree, they were with me all the way.

This research has been supported by a grant from the United States

Department of Energy.



ACKNOWLEDGEMENTS .....................................

A BSTRAC T ...............................................



Introduction ..................................... 1
General Description of Gaseous Discharges ......... 2
Fundamental Processes in the Glow Discharge ........... 12
Glow Discharge Sputtering ..................... 12
Glow Discharge Excitation and Ionization ........... 22
Analytical Considerations for GDMS ................... 28
Internal Standards ........................... 29
Relative Sensitivity Factor (RSF) Values ............ 30
Instrum entation .................................. 31
Mass Spectrometer .......................... 31
Neutral Species in the Glow Discharge Plasma ...... 57


Introduction ..................................... 63
Electrothermal Vaporization ......................... 65
Furnace Techniques .......................... 65
Introduction of Micro-Samples into an ICP .......... 69
Solution Samples in the Glow Discharge ................ 70
Introduction ................................ 70
Electrothermal Vaporization/Glow Discharge
Techniques ........................... 71
Ion Source Development ........................... 75
Experiments in a Six-Way Cross ................. 75
Filament on a Probe .......................... 79
Experim ental .................................... 82
Sample Preparation .......................... 82
Sampling Sequence .......................... 82

Data Acquisition ........
Filament Materials ......
Filament-Plasma Interactions ...
Ion Transmission .......
Internal ETV Sample Introduction
Effects of Different Filament
Response to Concentration


Pressure Effects ................
Cathode to Exit Orifice Distance ....

. ... ... ..... 84
. .. ..... .... 84
. . 86
. . 86
.. ... .. .. ... 92
. ... .. ... ... 92
. .. 96
. . .. 98
............ 100

Effects of Sample Positioning on the Filament .....

Memory Effects of the ETV/GD Technique .........
Effects of Contaminant Gas introduction ..........
External ETV Sample introduction ................. ...
Multi-Element Solution Mixtures .....................
Mass Spectral Scanning ......................
Separation of Multi-Element Species with
Filam ent Current ......................
C conclusions ...................................


Introduction ....................................
The Dissociation of Water in Gas Discharges .......
Studies of Water Vapor Effects in Glow Discharges ..
Experim ental ...................................
Results and Discussion ..........................
Effects of Water Vapor on Mass Spectra ..........
Effects of Water on Neutral Species ..............
Effects of Water on Various Plasma Species .......
Effects of Water on Various Cathode
Materials (Matrix Effects) .................
Effects of Water on Plasma Species in
Different Matrices ......................
Removal of Water with a Cryogenic Coil ..........
C conclusions ...................................



.......... 173

Introduction .................................... 173
Experim ental ................................... 174
Results and Discussion ........................... 180
Transient Behavior of Matrix Species ............. 180
Effects of Pulsed Water on
Other Plasma Species .................. 190
Use of Liquid Nitrogen Cooling ................. 198







Effects of Pulsed Water on the
Atomic Population ..................... 202
Power Supply Operation ...................... 207
C conclusions ................................... 211


Final Rem arks ................................. 214
Aqueous Solution Samples ................... 214
Effects of Water Vapor on GDMS ............... 216
Future Directions .................... .. ......... 218
Aqueous Solution Samples .................... 218
Effects of Water Vapor on GDMS ............... 219

REFERENCES ........................................... 221

BIOGRAPHICAL SKETCH ................................... 228

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




AUGUST, 1992

Chairman: Willard W. Harrison
Major Department: Chemistry

Glow discharge mass spectrometry is an accepted technique for the

analysis of trace elements in solid materials. In this dissertation, the sampling of

small volume aqueous solution samples has been explored. This method uses

electrothermal vaporization of a solution residue for atomization, while a glow

discharge provides the excitation and ionization. The main advantage of this

technique over other glow discharge solution analysis schemes is the increase in

sensitivity for a given sample since the analyte is atomized in a short time. The

effects of the electrothermal filament current on the plasma processes were

studied, since this could influence the discharge processes as well as ion

transport to the mass spectrometer. Variables such as pressure, cathode-to-exit

orifice distance, atomization current, and sample placement on the cathode were

evaluated and the best parameters presented. The method was found to have

relative standard deviations between 15-20 %. Multi-element samples may be

analyzed using either mass spectral scanning or separation of the elements by

their vaporization temperature. The biggest disadvantage of this method is the

irreproducibility, which is attributed to the introduction of contaminant gases with

each sample.

The effects of water vapor on the processes of the glow discharge were

also investigated. This is important since water vapor is always present in the ion

source to some extent. Water vapor exhibits detrimental effects on both

atomization and ionization in the plasma. Mass spectra taken with less than 5%

water vapor resulted in ion signals primarily from H20, H30, ArH, and 02. This is

due to ion molecule reactions and water dissociation occurring in the plasma. A

liquid nitrogen coil was constructed to aid in the removal and control of water

vapor in the ion source. Mass spectra obtained while cooling the source

contained ion signals mainly from the cathode material. Different cathodes were

investigated to observe the varying effects of the water vapor. It was found that

when sputtering reactive metals (such as getters) the water problem may be

minimized. Both steady state and pulsed addition of water were examined to

determine the processes occurring in the plasma.



The glow discharge is an old source that has been used over the years

primarily for applications that benefit from conversion of a solid sample into

atoms, ions, and photons that are characteristic of the sample's content. Glow

discharge devices have been constructed in a variety of configurations and have

shown promise in many applications ranging from optical spectroscopy to mass

spectrometry. The earliest reported works examined using the glow discharge for

optical spectroscopy, with one of the first publications appearing in 1916 by

Paschen [1]. Aston [2] used the glow discharge source in the 1940's for its ion

production ability. This source aided Aston's experiments to evaluate and

develop the technique of mass spectrometry, which was then in its infancy.

The majority of the studies reported in this dissertation use mass

spectrometry to sample the ion population produced by a glow discharge source.

Optical methods, such as atomic absorption (AA) and atomic emission (AE), were

also performed in some instances to obtain complementary data on the atomic

population of both ground state and excited species that coexist with ions in the

plasma. The results obtained in these experiments help provide an insight into

the fundamental processes occurring in the glow discharge plasma as well as


give some indication to the types of analytical studies that may be conducted and

the considerations that must be made when using this source.

General Description of Gaseous Discharges

Gaseous discharges are obtained upon establishing a flow of electric

current through a gaseous medium [3]. Gas discharges may be classified

according to the current with which they operate. Figure 1 shows a schematic

representation of the characteristics of gaseous discharges with respect to their

voltage and current [4]. There are generally considered to be at least three types

of discharges: 1) the Townsend discharge, 2) the glow discharge, and 3) the arc

discharge. The Townsend discharge operates at very low currents, which give

rise to its nonluminous appearance, since the population of excited species that

provide a discharge with its bright appearance is at very low concentrations. The

Townsend discharge is not self-sustaining and requires external phenomena such

as X-rays or UV light that will produce electrons in the gas to aid in ion

production. A Townsend discharge for analytical mass spectrometry has been

used [5] to ionize the reagent gas rather than high energy electrons emitted from

a hot filament. The central sections of Figure 1 comprise the areas of glow

discharge operation (both normal and abnormal). Abnormal glow discharges are

used in the studies presented here and will be further explained below. The final

section is the area in which an arc discharge operates. The arc discharge has

a very intense luminosity and operates at very high currents, where thermionic

emission from the cathode provides the electrons for this source. This type of



109 10-7




Figure 1 The relationship between voltage and current in a
gaseous discharge [4]. Vb: breakdown voltage, V,: normal cathode
fall potential, Vd: arc voltage.


discharge has found use for analytical spectrographic emission [6] and for the

vapor deposition of thin films [7].

The basic schematic of a glow discharge system is shown in Figure 2. The

main requirements for glow discharge operation are shown in the figure. The

requirements include a low pressure inert gas atmosphere (most often argon), a

cathode (usually the sample to be analyzed), an anode (usually the glow

discharge source housing), and a power supply to provide the high voltage to the

cathode. The dc/pulsed power supply develops a potential difference between

the cathode and the anode, which are both in contact with the low pressure inert

gas. In the dc mode, the power supply provides a constant voltage to the

cathode, while in the pulsed mode the discharge is switched on and off rapidly

with a given frequency. Upon application of the voltage, free electrons that exist

in this low pressure gaseous environment will accelerate toward the anode, which

has a positive bias due to the potential gradient that exists in the chamber. As

these electrons pass through the gas, they collide with and sometimes ionize the

gaseous species. Positive ions are created and accelerate toward the cathode,

which has a negative bias. As the gas ions strike the cathode a variety of species

is liberated, such as neutral atoms and molecules, ionized atoms and molecules,

electrons, and photons. The emitted electrons will accelerate toward the anode

and complete the electrical circuit. With this flow of electrons, the discharge will

maintain a continuous flow of current and be self-sustaining. The pressure regime

that may be used to operate a glow discharge is from about 0.1 10 Torr.











Discharge formation at these pressures requires a potential of 500 2000 V, which

results in a current in the low milliamp range.

The glow discharge is one type of plasma that is used for analytical

studies. A plasma is a partially ionized gas volume that consists of a generally

equal concentration of positively and negatively charged species, as well as a

large number of neutrals. The glow discharge contains a variety of zones that are

evident from Figure 2. These different zones have been characterized with

respect to their light intensity, potential distribution, field strength, space charge,

negative and positive charge regions and the negative current density [8]. The

layers that are present in the glow discharge (from the cathode to the anode) are

the Aston dark space, the cathode layer, the cathode dark space, the Faraday

dark space, the positive column, the anode dark space, and the anode glow. For

all of the regions to be noticeable the anode-cathode distance must be of

sufficient length. In commercial neon lights, the emission that is viewed is from

the positive column, not the negative glow shown here. For the GDMS ion source

the regions from the Faraday dark space out to the anode are missing due to the

close proximity of the anode to the cathode. Also, the Aston dark space that is

near the cathode is not visible because it is surrounded by the cathode glow,

which appears to "coat" the cathode since it is so close to its surface. The

analytically significant processes for GDMS operation occur in the cathode dark

space and the negative glow regions of the discharge, so the discussions here

will only refer to those regions. These regions are named "dark" and "glow"

according to the relative presence or absence of luminosity caused by the

radiative relaxation occurring in these regions [9]. The cathode dark space is

dark because there is very little emission from the electron-atom collisions that

occur here, due to the electrons that have been accelerated by the cathode

potential and can no longer sufficiently ionize the discharge gas [10]. In the

negative glow region, however, excitation and ionization will occur. The primary

color of the negative glow is a result of the light emitted upon the radiative

relaxation of the discharge gas. Argon glow discharges are normally a light blue

color, while the emission from neon is red in color.

As previously mentioned, and shown in Figure 1, there are two operational

regimes for a glow discharge: a normal and abnormal cathode fall. In a normal

glow discharge there is a constant increase in current even as the applied voltage

remains steady. The increasing current is compensated for by a subsequent

increase in the area of the cathode that is surrounded by the negative glow

region. Thus the current density, the amount of current for a given area of the

cathode, remains constant [3]. Once the complete cathode area is covered by

the negative glow, any further increase in current requires an increase in the

current density and thus an increase in the discharge potential. This mode is

called an abnormal discharge and is the mode that is almost always used for

analytical glow discharge devices because it is the most stable form. All studies

described in this dissertation used an abnormal glow discharge ion source.

Types of glow discharge devices

The sample to be analyzed is made to be the cathode in the electronic

circuit, while the anode can be constructed of any material. The ion source

chamber housing serves as the anode in the mass spectral studies described in

this dissertation. This particular type of glow discharge configuration is called a

coaxial cathode glow discharge. Other types of glow discharges include: 1) the

planar diode, 2) the hollow cathode, 3) the Grimm discharge lamp [11], and 4)

the hollow cathode plume. These configurations are shown in Figure 3.

Coaxial cathode. This cathode is illustrated in Figure 3A and the majority

of the GDMS work that has been done uses this type of cathode. This is due to

the ease of using the ion source chamber of the mass spectrometer as the anode.

Samples for this type of glow discharge are in the form of a 1 2 mm diameter

pin with about five millimeters left exposed to the plasma, or a pressed disk that

is shielded to allow only the top surface of the sample to be exposed to the

plasma. Conducting, bulk samples can be machined into the pin form directly,

while powders (conducting and nonconducting) are pressed into the disk form.

The nonconducting samples must be mixed with a conducting powder before

pressing unless an rf power supply is used. When using rf, care must be

employed to assure that sputtering is localized to the target area of the electrode.

Planar diode. This is the simplest configuration (shown in Figure 3B)

that is used analytically and may also be operated with either dc or rf voltage.

Planar diode glow discharges using an rf power supply were used by Coburn et

al. for mass spectrometry studies of the ion population present in sputter

deposition plasma chambers [12].

Hollow cathode. The hollow cathode lamp, shown in Figure 3C, is

probably the most familiar glow discharge that is used in chemical analysis. It





,} :-- .. ::.
*.. i*

*'tk : 1 1: i ~



Figure 3 Various glow discharge configurations.
cathode; B) planar diode; C) hollow cathode lamp; D)
discharge; E) hollow cathode plume.

A) coaxial
Grimm glow








/ \







r" -rJa


Figure 3 -- continued.

often serves as the light source for atomic absorption or atomic fluorescence

studies, as well as emitting the characteristic lines for emission studies of the

lamp material. For its use as a light source, the only requirement is that the

hollow cathode lamp contains the same material as the atom of interest in the

sample to be analyzed. This source operates at lower voltages and higher

currents than other glow discharge devices due to the constraint of the plasma

within the cathode. This is often referred to as the "hollow cathode effect" [13].

This effect produces a large increase in the intensity of the radiation emitted from

the negative glow, making it attractive as an optical emission source. This results

in a large increase in the excitation and ionization that is observed in the hollow

cathode cavity. This effect has been studied and some theories published


Grimm discharge lamp. The Grimm glow discharge configuration was

first introduced in 1968 [11] and is illustrated in Figure 3D. It is often called an

obstructed discharge due to the fact that the cylindrical anode is positioned within

one cathode dark space distance from the cathode. The sputtered region of the

sample is limited to the region of the cathode enclosed by the anode. This

source has found primary use for atomic emission studies. The anodic sampling

of this source produces ions with a narrow energy distribution and is well suited

to analysis in a quadrupole instrument, which behaves best with ions having

narrow energy bands.

Hollow cathode plume. This source is shown in Figure 3E. This source

was developed by Marcus and Harrison [16] primarily for atomic emission studies.

The hollow cathode plume (HCP) results from the restriction of a hollow cathode

discharge to a 1.5 mm orifice in the base of a normal hollow cathode. Samples

must be formed in the shape of a disk, with a central orifice, that may be mounted

in the base of a graphite cylinder and used as the cathode. These disks may be

machined or pressed depending on the nature of the sample. HCP discharges

operate at about 1 Torr pressure and currents in the 50 200 mA range. The

applied potential is typically around 1000 V.

Fundamental Processes in the Glow Discharge

The diagram in Figure 4 shows the structure and some of the processes

occurring in the analytical glow discharge. The various mechanisms, atomization

and excitation/ionization, are shown to occur in very distinct areas of the plasma

and its surroundings. The atomization process, known as sputtering, occurs at

the surface of the sample when argon ions strike the cathode surface. The

excitation and ionization processes are shown in the negative glow region and

occur as the result of collisions in this relatively field free area. There is little or

no potential gradient that exists in the negative glow region since virtually all of

the discharge potential is dropped across the small dark space distance (a few

millimeters, a distance that depends on the operating pressure of the plasma).

Thus, the potential gradient is large within this dark space distance. The following

sections will discuss the fundamental processes in more detail.

Glow Discharge Sputtering

Glow discharge sputtering occurs when the discharge gas ions strike the

surface of the cathode. When an argon ion strikes the surface it will penetrate a


Illustration of the fundamental processes in the glow



Figure 4

few atom distances into the lattice sending "shock waves" out in three dimensions

around the impact point. Actually, it has been reported that the species striking

the surface of the cathode is really a "neutralized ion" [17]. These "neutralized

ions" are formed as the ion gets close to the metal surface and is neutralized by

a field-emitting electron.

If sufficient energy is transferred by the surface collision (in excess of the

target's binding energy), a variety of species will be dislodged from the cathode.

The masses of the collision partners and the individual cross sections (which

depend on the ion velocity and the electronic structure of the partners) will play

a role in the sputtering process [17]. Positive secondary ions are often emitted,

but the strong potential gradient in the cathode dark space will cause these ions

to return to the surface of the cathode and redeposit. These are the species of

interest in the technique of secondary ion mass spectrometry (SIMS), where the

ions formed at the surface of the sample are analyzed. Electrons are also emitted

in GDMS and they are accelerated away from the cathode by the field gradient

and pass into the negative glow region. The emission of electrons from the

cathode surface sustains the discharge and allows it to continue running. Thus,

the glow discharge is a self-sustaining discharge that requires no external

excitation to operate. The third particles that are emitted from the surface are

ground state neutral atoms from the cathode material. These atoms are the

analytically interesting species that are studied with GDMS and make up the main

component of the sputtered particle flux that leaves the cathode surface [18,19].

These species migrate into the negative glow by diffusion, where they may be


subsequently atomized and ionized. It is this distinct production of discrete atoms

from a solid medium that gives GDMS its relative freedom from matrix effects.

Essentially, the solid is being converted into a "gaseous solution" of sample atoms

(the analyte) within the argon discharge gas (the solvent).

The efficiency of sputtering (i.e., the number of cathode atoms removed

from the surface per impinging ion) is determined by a parameter called the

sputter yield (Y) [20]:

S9.6 x 104 W (1)
M i' t

where W is the weight loss of the cathode (in g) due to the sputtering process, M

is the atomic weight, i+ is the ion current (in A), and t is the total sputtering time

(in s). The ion current, i+, is related to the total current of the discharge, i, by the

following equation:

i (2)

where y is the number of secondary electrons released by one ion. Studies of
the experimental factors that influence sputtering have been performed mainly

using focused high energy ion beams similar to those used in SIMS [21]. The

sputter yield is mainly dependent upon the incident ion mass, the angle of

incidence of the incoming ion, the energy of the incoming ion, and the target

material [17,22]. A rigorous theoretical consideration of sputtering and sputter

yields has been published [23]. In the next few sections, some of the factors that

influence sputtering will be discussed.

Incident ion mass

The incident ions that are encountered in glow discharge sputtering are

most frequently those from noble gases. Noble gases are used in GDMS

because their relatively inert nature will minimize any interactions with the cathode

material and they have the ability to provide a good sputter yield since they

behave like hard spheres in the "billiard ball-type" collisions. Bay et al. [24] have

published a theory to explain the dependence of a sample's sputter yield on the

mass of the ions that are bombarding the sample. The maximum sputter yield of

a given target material is proportional to an energy transfer factor, y, which may
be determined as follows:

4M M2
Y. (3)
(MA + M2)2

where M, and M, are the masses of the incident ion and the target atom,

respectively. It can be seen that the maximum energy transfer will occur when M,

= M2, and the highest sputter yield should be obtained. For this reason argon

has been found to be the best choice for sputtering most medium weight atomic

elements [22,25]. Sputter yields have been measured using a variety of noble

gases by Wehner et al. and were shown to have a periodic trend with the best

sputter yields obtained with the noble metals Cu, Ag, and Au [17]. Zinc and

cadmium have still higher yields due to their low heat of sublimation but that

makes these elements undesirable to have in a high vacuum system [17]. Typical

sputter yields obtained using 400 eV Ar for a variety of target materials are


shown in Figure 5 [26,27]. As the mass of the incoming ions is increased, greater

sputter yields are obtained for heavier target elements.

Incident ion angle of collision

Oechsner [28] has reported a large quantity of data concerning the

influence that the angle of incidence has on the sputtering yields. At angles of

60 700 from the normal, it was found that sputter yields (in the energy range
from 0.5 2 keV) rise to maximum values which are 1.5 2.4 times higher than

that obtained for normal incidence ions (see Figure 6). The sputter yield showed

increases in this range for metals that normally would produce low sputter yields

[17]. In these ion beam studies, it was determined that the ion current density

decreases with the cosine of the angle a between the beam and the surface
normal. Thus, if there is a sputtering rate increase, it is much less than the yield

increase. The dotted line in Figure 6 shows the reciprocal of the cos a
dependence of Y(a)/Y(0). Above this curve is the region where a higher sputter

removal rate is achieved with an ion beam. However, unless the surface is

rotated or sputtered from different angles, it is most often best to sputter with

incoming ions normal to the sample to alleviate problems that occur with surface

roughening [17].

Incident ion energy

Wehner has performed experiments using a high vacuum beam to

determine that the minimum kinetic energy that is required to dislodge an atom

from its matrix is approximately four times the heat of sublimation of the material

[17,29]. For argon, the threshold for sputtering ranges from 13 eV for aluminum


_.-.______. (------0


>O --- -- m

Z 0 *0
- ----- Q -- c z?


o o E

0 o

N 0 w 0 v No 0 w w0

CJ ci c i :- :- -: O O





1 .5 -0 /

00 300 60 900
a (with respect to normal)

Figure 6 Sputter yield increase of Cu with bombardment of various
noble gas ions under oblique incidence. The dashed line is the 1/cosa
curve, separating rate increase from rate decrease. [28]


to 33 eV for tungsten. Once this threshold is surpassed, the sputter yield will

increase rapidly with increasing ion energy, reach a plateau, and then decrease

as the energy continues to rise. This plateau region generally occurs with about

1 keV ion energy, but will vary slightly with the discharge gas and the sample.

The yields for some elements under different conditions are shown in Table I.

These values will show some scatter because the sample may have different

surface roughness, a preferred crystalline orientation within a nominally

polycrystalline sample, or the possible buildup of impurity layers on the target

during sputtering [17].

As previously mentioned, when an ion hits the sample surface it becomes

imbedded in the matrix and its kinetic energy is absorbed by the atoms in the

vicinity. At lower ion energies the penetration is limited to the surface of the

sample where the atoms are disturbed and ejected from the matrix. However, at

higher energies the penetration depth of the ion increases, which results in

damage to the sample matrix at a distance too far from the surface to affect these

atoms. Thus the sputter yield will decrease at these higher energies [17].

In the glow discharge, the energy of the incoming ion is not easily

determined because of the energy losses caused by the collisions that the

incoming ion encounters as it passes through the discharge dark space.

Attempts to determine this energy in the glow discharge have been addressed

[30,31]. Calculations (based on the pressure and cathode fall potential of the

discharge) have been performed to determine the average energy of the incoming

ions hitting the cathode under typical discharge conditions [25]. An ion energy

TABLE I. Sputtering Yields in Atoms/Ion for Some Elements
Under Different Argon Ion Bombardment Energies

Element 500 eV 1 keV 10 keV
C 0.12
Al 1
Si 0.5 0.6
Ti 0.5 2.1
Cr 1.2
Fe 1.1 1.3
Ni 1.45 2.2
Cu 2.4 3.6 6.6
Mo 0.8 1.1 2.2

Ag 3.1 3.8 8.8
Au 2.4 3.6 8.4

Source: Reference [17].


range of 100 eV 575 eV, corresponding to an applied voltage range from 400 -

2000 V dc, was found to be the typical value for the bombarding ions.

Target material

The dependence of sputter yield versus target material was shown in

Figure 5, where different elements under identical ion bombardment conditions

showed periodic trends in their sputter yield. The sputter yield will increase within

any one period of the periodic table. These differing sputter yields may become

a problem when analyzing alloys or other materials that contain many elements

and is called preferential sputtering [32,33]. It was found that preferential

sputtering will cause the surface composition of the sample to change and be

enriched by the species with the lower sputter yield. The initial consequences of

this process is a difficulty in obtaining an atom population in the gas phase that

is proportional to the concentrations of the species in the bulk sample. However,

the processes inherent to the glow discharge will passively correct for this

difference, because the cathode surface enrichment of the lower sputter yield

species will lead to an eventual equilibrium state [34]. Thus, after the initial

sputtering time required to establish this equilibrium has passed, the gas phase

atom population will be proportional to that of the bulk sample.

Glow Discharge Excitation and Ionization

Aside from being an efficient source of atom generation, the glow

discharge will also excite and/or ionize the sputtered sample atoms through a

variety of energy transfer processes. These processes occur primarily in the

negative glow region, because the negative glow region is an environment rich


in collisions that may result in the ionization of the sample atoms. Kinetic and/or

potential energy may be transferred by a collision partner that is sufficient to

excite electrons of the target species to higher energy states [35]. In some cases

the electron may be removed completely. It was determined by Harrison and

Loving that ionization is occurring throughout the volume of the plasma, but only

the ionization that takes place directly adjacent to the exit orifice will be observed

by the mass spectrometer [36]. This is because an ion that is formed in other

regions of the plasma will undergo further collisions on its way through the source

and would never make it to the ion exit orifice.

There are three primary types of collisions that are occurring in the

negative glow region of the discharge. They are collisions involving discharge

electrons, collisions involving excited discharge species, and collisions involving

discharge ions. These collisions will lead to excitation, ionization, and some

recombination of discharge species. Table II lists some of the most important

ionization mechanisms that occur in the negative glow region. The two

mechanisms that are thought to provide a major contribution to the overall

ionization of sputtered species in the plasma are electron ionization (El):

M + e- M' + 2e- (4)

where M is a neutral analyte atom, e- is an electron, and M+ is an analyte ion,

and Penning ionization (PI):

M + Ar* MA + ArO + e-

TABLE II. Possible Ionization Mechanisms in
the Glow Discharge

I) Electron Ionization:

M + e M+ + 2e

II) Penning Ionization:

M + Ar* M+ + Ar + e

III) Associative Ionization (Homonuclear):

Ar* + Ar Ar2+ + e

IV) Associative Ionization (Heteronuclear):

MO + Ar* ArM+ + e

V) Nonsymmetric Charge Transfer:

Ar+ + M -. M + Ar

VI) Symmetric (Resonance) Charge Transfer:

Ar+ast) + Ar (w) Ar ft) + Ar+(s

VII) Dissociative Charge Transfer:

Ar+ + MO M+ + O + Ar

M = neutral analyte atom
M+ = analyte ion
Ar = argon metastable atom
Ar = neutral argon atom
e = electron


where Ar* is an argon metastable atom and Ar is a neutral argon atom. These

modes of ionization will be discussed in the following sections.

Electron ionization

Electron ionization (El) in the glow discharge is accomplished through

electrons that are accelerated across the cathode fall region or by electrons

existing in the discharge that possess energy in excess of the ionization potential

of the target atom. This mechanism is an important source of ionization in the

glow discharge [37], and is thought to be the primary ionization path for the

discharge gaseous species in the plasma [38]. If a collision occurs involving an

electron possessing an energy below the ionization potential, but above the

lowest excitation potential of the target atom, electronic excitation of the species

will occur [39]. If the excited species can radiatively relax, then the transition may

be probed by optical spectroscopy. If radiative relaxation is forbidden, the excited

state may exist in the plasma for milliseconds before an alternative deexcitation

process may occur [40]. These species are referred to as metastable species

and they may lead to further excitation and ionization of other species. If an

electron possesses little or no kinetic energy, it is subject to recombination

collisions with discharge ions. The excess energy may result in the creation of

excited neutrals or metastables [41].

The electrons that are present in the glow discharge differ in both their

origin and their energy. There are three general classes of electrons based on

these differences [37]. First, there are the fast electrons with energies exceeding

25 eV. These have been accelerated away from the cathode across the dark


space. The second class is the electrons that were created by ionizing collisions

in the negative glow region. These are referred to as secondary electrons and

have an average energy about 7 eV. The final class of electrons is called

"ultimate electrons." These are electrons that have been thermalized (with a mean

energy less then 1 eV [42]) by collisions with other plasma species.

Penning ionization

As mentioned in the previous section, long-lived metastable species may

be formed within the glow discharge plasma and be deexcited by collisional

relaxation processes. When this occurs, the transfer of energy from the

metastable species to the collision partner results in excitation or ionization of the

atom. Two such processes are considered to play a major role in glow discharge

ionization. These are Penning ionization, [43] which was shown earlier in

equation (5), and associative ionization [44], which will be discussed in the next


If the internal energy of the metastable atom is greater than the ionization

energy of the collision partner, then an electron-ion pair is created. As mentioned

previously, this process is believed to be the primary ionization mechanism for

sputtered species in the glow discharge. Table III shows some of the energy

levels of the noble gas atoms that are used for glow discharge operation. Argon

is usually the preferred gas for GDMS because of its metastable internal energies

of 11.5 and 11.7 eV. These energy levels are above the first ionization energies

of most elements, below the second ionization energies of many elements, and

are below the ionization energy of atmospheric contaminants.

TABLE III. Low-lying Metastable Levels
of Rare Gas Atoms

Gas Metastable ionization
Energy, eV Energy, eV
He 19.8, 20.7 24.58
Ne 16.6, 16.7 21.56
Ar 11.5, 11.7 15.76
Kr 9.9, 10.5 14.00
Xe 8.3, 9.4 12.13

Source: Reference [8].

Other ionization mechanisms

Associative ionization. Associative ionization, shown in equation (6),

MO + Ar* ArM' + e- (6)

is the mechanism that is responsible for the production of polyatomic species that

contain discharge gas atoms (i.e., the argides in an argon discharge) [38]. The

mechanism is similar to the Penning process except that the collision partners

remain together and form an ion cluster. The electron that is released during the

reaction will carry off any excess energy.

Charge transfer reactions. The charge transfer reactions listed in

Table II are of two types [37]: symmetric and asymmetric. Symmetric reactions

involve collisions of like partners. These reactions may lead to the formation of

fast neutral atoms as an accelerated gas ion collides with a neutral gas atom [37].

It is believed if this happens in the dark space region, sputter enhancement will

occur because of an increase in the number of species bombarding the surface.

Asymmetric reactions are a potentially important mechanism for the ionization of

sputtered species in some glow discharge configurations. Asymmetric reactions

are believed to be the principal ionization mechanism for sputtered species in an

obstructed discharge [45] and are responsible for the production of excited ions

of sputtered material in hollow cathode vapor lasers [46].

Analytical Considerations for GDMS

Glow discharge mass spectrometry has found wide application for the

analysis of a variety of samples over the past decade, from pure metals and

semiconductors to thin films and ceramics. This section will be used to discuss

the considerations that must be made for analytical work with a GDMS system.

These considerations are vital for those applications where an accurate

concentration value is needed to determine impurities in high purity samples [47]

or the abundances of certain critical components, such as in bulk alloys [48] and

semiconductors [49].

Internal Standards

Internal standards in GDMS employ the matrix element as the internal

standard for semi-quantitative analysis. It is assumed that the concentration of

the matrix element of a sample is known. In this method, the sensitivity (the ratio

of the signal intensity to its concentration) observed for the reference element

(e.g., iron in a stainless steel sample) will allow an estimation of the concentration

of the analyte using the following equation:

C X M (7)
where Cn,, is the unknown concentration of the analyte, I,, is the signal intensity

of the analyte, C,, is the known concentration of the reference element, and I, is

the signal intensity of the reference element. These direct comparisons are valid

in GDMS since the variation in elemental sensitivities for this technique is on the

order of three. In some techniques the sensitivity among the elements varies by

orders of magnitude and the direct comparison shown above would not give even

semi-quantitative results.

Relative Sensitivity Factor (RSF) Values

For precise quantitative analysis, external standards must be used for

GDMS. Two approaches are possible: 1) the use of relative sensitivity factor

(RSF) values obtained from standard materials that are similar in content to that

of the unknown, and 2) using sets of standards to construct a calibration curve

from which the unknown concentration is related to the ion signal intensity.

RSF values were first employed in spark source mass spectrometry (SSMS)

and correct for reproducible variations in the observed sensitivities among the

elements. These variations can arise from physical, chemical, and instrumental

differences. To use RSF values, an element of known concentration must be

present in the sample and the analysis conditions used for the unknown and the

reference must be the same. The RSF is defined as a ratio of the sensitivity of an

analyte element to the sensitivity of the reference element [50]. The concentration

of the analyte and the reference must both be known in the standard, while only

the reference, element concentration needs to be known for the unknown. The

RSF value is calculated as follows:

RSF ( Ca (8)
( / C.. )
where the components of the equation are the same as for Equation (7) shown

for semi-quantitative analysis. The RSF value is then employed as a correction

factor for the intensity ratio calculation shown in equation (7). The unknown

concentration of the analyte is then calculated as follows:

C -x x re (9)

RSF values have been obtained for a variety of sample types and were

determined in this laboratory for stainless steel samples by using an NIST

standard reference material 1264A [51]. RSF values in GDMS have also been

reported in the literature for a variety of elements [52]. For RSF values to be most

accurate, separate RSF values should be calculated for every instrument on which

the analysis is to be performed. The calibration curve method has been reported,

using a Grimm glow discharge, with the use of seven stainless steel standards



Mass Spectrometer

The research presented in this dissertation is centered around the use of

a mass spectrometer for fundamental studies of the glow discharge ion source

processes. As mentioned earlier, the glow discharge plasma is an efficient

atomization source containing processes that may excite and ionize many of the

atoms of interest. The mass spectrometer provides an efficient detection system

for the ions produced in this source, as well as providing the isotopic information

for the elements of interest. The mass spectrometer provides spectra that are

simpler and easier to interpret than optical instruments, since each element has

specific mass-to-charge ratios for each of its isotopes. Thus the only interference

problems occur with different elements having an isotope at the same nominal

mass or from polyatomic species that occur at the same nominal mass as an


element of interest. The former problem can be avoided by using an isotope that

does not overlap with another element. Almost all elements in the periodic table

have at least one isotope with a nominal mass that is apart from any other

elemental isotope. The problems of polyatomic interference are more difficult to

solve. Studies have been performed previously using a triple quadrupole mass

spectrometer with a glow discharge source to investigate the collision-induced

dissociation of some polyatomic species that are present in the plasma [54]. A

double quadrupole system has also been reported for sampling an rf powered

glow discharge source, with one quadrupole being used as the dissociation

chamber [55]. Both of these methods were successful in using low energy

collisions with argon target atoms to reduce and/or eliminate species such as M2+,

MAr+, MO+, Ar2+, and residual gas species.

The experiments for this dissertation were performed on two different mass

spectrometers. One of the instruments is based on Extranuclear (now called

Extrel) components and is a more research oriented instrument since its source

components are readily accessible and easily modified. This instrument was

constructed at the University of Virginia and has been described in detail

elsewhere [56,57]. This instrument is shown in Figure 7. The second instrument

is a prototype mass spectrometer built by Finnigan MAT in Bremen, Germany

using Balzers components. This instrument has a more commercial design (i.e.,

the components are not as easily accessible or modified) and has previously

been described [35,51]. A cross section of the Finnigan instrument is shown in

Figure 8. This instrumentation section of the dissertation will be used to compare









4- -


0 0
















Q Cn
co ..1



and contrast these two instrument designs and discuss their unique components

as well as any special considerations that must be made for each. This section

is not meant to be a complete and detailed outline of each individual instrument,

although the primary mass spectrometer components will be discussed, since this

generally applies to either instrument as well as any other quadrupole based

instrument. The Extrel instrument, being more accessible to changes and

modifications, is the instrument where new ideas and new source designs are first

attempted. Once successful, the modified instrumentation or experiments may

be adapted and performed on the Finnigan. The conversion of an experimental

scheme from one instrument to the other gives the researcher an idea of the

difficulties that might be encountered when taking a new idea and trying to use

a commercial instrument to run the experiments. The aspects of instrument

modification and successful adaptation to a more restricted and commercially

designed instrument are always of high concern. This is especially a problem that

may be experienced in an industrial environment where applications usually

require adaptation of an experimental scheme to a commercial instrument.

Ion source

One of the most apparent differences between the two instruments is the

design of the ion source. On the Extrel instrument the ion source is external to

the mass spectrometer chamber and may be isolated by a sliding gate valve that

can be opened and closed between the ion source and the analyzer chamber.

The position of the gate is shown in Figure 7. When the gate is down, a 1.5" hole

is in position between the ion source exit orifice and the skimmer cone of the


instrument. With the gate closed the ion source may be vented to atmosphere

while maintaining a high vacuum in the mass spectrometer chamber. The source

is usually vented with argon gas when changing the sample being used for

experimentation. This alleviates some contamination that may occur if air is

allowed into the source. The ion source, which is constructed from a 2 3/4" six-

way cross (MDC Vacuum, Hayward, CA), is larger in volume than the Finnigan


The Finnigan instrument has its ion source mounted on a stainless steel

tube and inserted into a position within the first pumping chamber of the mass

spectrometer. The Finnigan ion source is constructed from a 1" Cajon ultra-torr

tee and its position in the instrument is shown in Figure 8. This source cannot

be vented to atmosphere without venting the entire mass spectrometer. Thus a

sample introduction assembly was constructed [51] to allow the sample to be

changed without venting the instrument. This assembly is shown in Figure 9.

The principal components of this assembly are labeled in the figure. The ion

source is attached to a stainless steel tube that is welded to a vacuum flange and

attached to the instrument through an adjustable bellows. The positioning of the

discharge chamber in three dimensions is controlled by three turnbuckle screws

located 1200 apart around the bellows. This allows the ion exit orifice of the
discharge chamber to be aligned coaxially with the skimmer cone orifice, as well

as allowing the distance between the two to be adjusted. The ball valve in the

assembly allows the mass spectrometer to maintain its vacuum after removing the

direct insertion probe. While the probe is in place, a 1/2" Cajon connector seals






Figure 9 Glow discharge mass spectrometer sample
introduction assembly for the Finnigan instrument.

around the probe with an O-ring allowing the vacuum to be maintained. A

roughing pump is required to pump air out of the small volume between the

Cajon connector and ball valve when a new sample is inserted. Otherwise, the

short burst of air will cause a high pressure condition in the instrument, causing

the protection circuit to shut off the pumping system to protect the turbomolecular

pumps. During normal operation the roughing pump is isolated from the system.

The design of the direct insertion probes used in these experiments has been

described elsewhere [51]. The probes are identical for both instruments other

than their length. A different probe was constructed for the electrothermal

introduction of solution samples and will be described in Chapter 2.

Einzel lenses and Bessel box

The einzel lenses are constructed from stainless steel tubes and are used

to direct the ions formed in the source into the mass spectrometer. These lenses

are only used for the transport of ions and are only present on the Extrel

instrument in this laboratory. Einzel lenses are required in this system due to the

relatively long distance from the ion exit orifice to the skimmer cone. This

distance cannot be minimized because the Extrel instrument includes the gating

valve to isolate the source from the mass analyzer. These lenses are not shown

in Figure 7 but are contained within the arm of the six-way cross between the ion

exit orifice and the mass spectrometer skimmer cone. The einzel lens system is

shown in the left half of Figure 10. The lenses are powered by a 1000 volt power

supply (Hewlett-Packard Model 6521A) that allows all of the lens voltages to be

independently adjusted from 0 to -1000 V. Einzel lenses are not required in the











0 (D





Finnigan instrument since the bellows assembly allows the ion exit orifice of the

source to be placed very close to the skimmer cone.

After the skimmer cone in both instruments lies a Bessel box energy

analyzer, which allows only a small bandwidth of ion energies to pass into the

quadrupole mass filter. The construction of a Bessel box is shown in the right

half of Figure 10, illustrating the complete lens system for the Extrel instrument.

The Bessel box lenses are powered by an Extrel Ionizer Control Model 020-2.

The lensing system for the Finnigan is shown in Figure 11 that illustrates the lens

layout as well as the position of the quadrupole analyzer. These lenses may each

be biased from 0 to +/- 140 V and this value is determined by the control circuit

illustrated in the figure.

The principle behind using an energy analyzer is that the quadrupole

system will operate most efficiently with ion energies from 0 10 eV and the glow

discharge source produces ion energies from 0 20 eV [57,58]. Also any neutral

species and photons that come from the ion source will collide with the center

beam stop that is positioned in the middle of the second lens (housing). The

Bessel box will allow ions within 1 2 eV of the voltage applied to the housing to

be passed into the quadrupole; this may be tuned not only to give maximum ion

signal to the detector but also to allow "tuning" of the mass spectral peaks to give

a Gaussian distribution around the nominal mass number. Ions with a small

energy will be repelled back out of the Bessel box by the voltage on the housing.

Ions with large energies will collide with the housing or the central beam stop

since they are too energetic to be curved through the device. Ions in the correct


I[ |a|IE
-6 I

F- c7 ? 100:1
0. o .

S& & > 1 0 :
75 =

0 0 0 C .2 9 9 a
I 0

2 4r-

a + E*

I I oI

A* U

E C5 E ce W -- c us c 9
r S O p 0 ) -
a, -C



energy band will be deflected around the central beam stop and then focused

through the exit lenses into the quadrupole mass analyzer.

Quadrupole mass filter

The mass separation system for both instruments is based on a

quadrupole mass filter. This mass separation technique has widespread use and

has been previously described [59]. Mass separation with a quadrupole system

is based on the mass-to-charge ratio (m/z) of the various ions produced from the

sample. The major difference of the quadrupole systems on the two instruments

in this laboratory is the size of the rods that are used. The Extrel system has

stainless steel rods that are 220 mm in length and 19 mm in diameter. This

provides an internal field radius (the distance from the central axis of the rods to

the closest surface of each rod) of 8.22 mm providing a mass range of 1 380

amu. The Finnigan instrument has 200 mm long rods with a diameter of 8 mm,

resulting in a field radius of 3.45 mm that passes specific masses from 1 511

amu. The Finnigan system uses a QMA 150 quadrupole system (Balzers,

Liechtenstein). Thus, both systems give a wide enough mass range to pass all

of the elemental species of interest (approximately 1 250 amu). The basic

operating principles of both quadrupole systems are the same and will be

described below.

A quadrupole system consists of a set of four stainless steel electrodes,

which ideally have a hyperbolic cross section, that are accurately positioned in a

radial array. A block diagram of a quadrupole system is shown in Figure 12.

Since the construction and mounting of hyperbolic rods is more difficult and



Schematic diagram of a quadrupole mass analyzer.

Figure 12


expensive, many quadrupole systems employ electrodes that have a circular

cross section. Denison [60] has previously shown that a good approximation to

an ideal hyperbolic field may be obtained if the radius of the electrodes (r) is

related to the quadrupole field radius (ro) by the expression,

r 1.148 ro (10)

The filtering mechanism of the mass analyzer is obtained by the

simultaneous application of dc and ac signals. This will produce an ion trajectory

through the quadrupole that follows a very complicated path with only one

specific mass-to-charge ratio being passed at any one time. The operation of the

quadrupole system will be explained here in a qualitative manner by considering

the physical effects in the X-Z and Y-Z planes separately. The Z axis is defined

along the length of the rods, while the X and Y axes pass through the center of

the opposite rod pairs (see Figure 12). A positive dc potential is applied to the

two rods lying in the X-Z plane, while the rods in the Y-Z plane experience a

negative dc potential [61]. The rf potentials applied to the pairs of rods are

opposite in polarity to one another and are 1800 out of phase.
Here the previously mentioned planes will be considered separately as to

the mass-dependent effects of each pair on the trajectories of some ions. First,

the trajectories in the X-Z plane will be discussed. If an ion is heavy or the

frequency of the rf potential is very high the ion will feel primarily the effect of the

average potential applied to the electrode. In this case heavy ions are mostly

influenced by the positive dc potential and will be focused onto the center axis

of the rods. The times when the dc potential is negative will have a negligible

effect on heavy ions. However if the ion is light the varying rf potential may have

a greater affect on its trajectory. If it is sufficiently light it may experience an

acceleration during the negative voltage cycle that will cause it to collide with an

electrode. By this means ions are filtered according to their mass-to-charge ratio.

Thus ions with a mass below a critical m/z value will be filtered out of the beam

and ions above the critical m/z will be transmitted through the device to the

detector. By increasing the rf voltage, heavier ions can be displaced by the rf

field. Thus a high-pass mass filter is formed in the X-Z plane (see Figure 13A).

Second, the effects in the Y-Z plane will be considered. The potential

applied to the electrodes in the Y axis is always equal in magnitude and opposite

in sign to the potential applied to the electrodes in the X axis. Thus, as

mentioned previously, the rf potentials in the X-plane electrodes are 1800 out of
phase with the potentials in the Y-plane electrodes. Also the dc potential applied

to the X-Z plane is positive, while that of the Y-Z plane is negative. As before,

heavy ions will be primarily influenced by the average dc potential. However, in

the Y-Z plane this potential is negative. This means that the heavy ions will be

defocussed away from the central axis of the system and eliminated from the

beam. Conversely, the ions that are lighter will respond to the focussing action

that occurs when the positive portion of the alternating field becomes larger than

the static negative potential. If the frequency and the magnitude of the rf field are

well chosen, the rf potential may be considered to correct the trajectories of light

ions and prevent them from striking the rods along the y axis. Increasing the rf

voltage increases the mass-to-charge ratio that is allowed to pass through the




Figure 13 Mass filtering in the quadrupole. A) X-Z plane, high-
pass filter; B) Y-Z plane, low-pass filter; C) combination of both
planes, band-pass filter [59].

quadrupole. This brief explanation has been discussed in more detail and some

models proposed by Dawson [62]. Thus in the Y-Z plane, the electrodes and

applied potential combine to result in a low-pass mass filter (see Figure 13B).

If an ion is to travel through the quadrupole to the detector on the other

side, the ion must be stable in both the X-Z and Y-Z planes. Thus an ion must

be sufficiently light so as not to be eliminated by the low-pass filter of the Y-Z

plane but not so light that it is eliminated by the high-pass filter in the X-Z plane.

Thus a band-pass filter is created where a narrow mass range of ions are

transported through the quadrupole filter to the detector (see Figure 13C). The

mass corresponding to the center of the mutual stability region is governed by the

magnitude of the applied ac and dc potentials.

The equations of motion in a quadrupole system can be quite complicated.

Therefore, a brief discussion will be presented here and more information may be

obtained from the references cited within this section. The motion of the ions in

the x and y directions depends on the variation of the potential, 0f, with time as

expressed in the following equation [63],

( U- Vcos( ot) (11)

where U is the dc voltage, V is one half of the peak-to-peak rf voltage and (

equals 2jrf with f being the rf frequency. It was determined that for a sinusoidally

operated hyperbolic mass filter, the potential distribution (p) at time (t) is then

expressed as [59],

< [U + Vcos ( ot)] (x2 (12)

where x and y are the distances along the given coordinate axes, r, is the

quadrupole field radius. Equation (12) is then differentiated relative to the x and

y axes to calculate the intensity of the electric field (E) along the X-Z and Y-Z

planes [59]. The results are as follows:

8x r.


Ey [U + Vcos (ot ) (14)
by r
These equations can then be mathematically converted to the Mathieu type

equations by first considering the equations of the force that an ion feels in each

plane and substituting Newton's law (F = ma) [62]. The final results of these

calculations are shown in Equations (15) and (16),

d + [U + Vcos( t)] ex 0 (15)
dt2 mr


_[U + Vcos ( ot)] 0 (16)
dt2 mri


The solutions to these equations are usually unbounded, which corresponds to

the rejection of ions from the mass filter. Stable solutions may be obtained only

for certain values of m, r,, f, U, and V. These solutions will correspond to

conditions that will give a successful transmission of an ion through the

quadrupole. The following substitutions (known as the Mathieu parameters) are

usually made to allow these equations to be written in a single expression [63].

The Mathieu parameters are:

a -4 (17)


q (18)

These substitutions represent the dc voltage counterpart (a) and the rf voltage

amplitude (q) [59,63]. A typical stability diagram is shown in Figure 14, which is

constructed by plotting dc voltages that yield stable solutions vs. rf voltages that

yield stable solutions. The area below the triangular region represents the values

of dc and rf voltages resulting in a stable ion trajectory for this mass.

To generate a mass spectrum, the alternating voltage, V, is scanned, with

the direct voltage, U, following it with a given ratio along a scan line that is

anchored at the origin in the diagram. Two different scan lines are shown in

Figure 14. The top one will result in more triangular peaks, but will have a lower

intensity since the area of stability traversed is small. The lower scan line will

increase the signal intensity of the ion but will result in broad trapezoidal peaks.

0 0

zq g o
uj- z a
a. Z c

,5 ,o0 2
D D co N

S0 0)

v i d c
T ca
I 0V

1----I----I ---T ||
I I0

o i&


0 .MeA?

6 0 6


Each given mass will have a slightly different stability diagram. Figure 15

(modified from reference [63]) shows the stability diagrams for four different

masses superimposed on the same diagram with actual voltages on the axes.

This diagram may be used to demonstrate the principles of resolution and

intensity and how they interrelate. The mass scan line given is one that may

typically be used in GDMS where unit mass resolution is required. Note that as

the mass scan line passes out of the stable region for mass 28 it just begins to

enter the stable region for mass 29. This will give unit resolution. If the slope of

the line is decreased, then the mass scan line would intercept more of each ion's

stable area (resulting in greater intensity), but for two adjacent ions the mass scan

line would pass through portions of their stability region at the same time. Thus,

the operator must realize that when using a quadrupole system there are trade

offs between good resolution and good signal intensity.

Ion detection

Both instruments have dual detection systems. Each has a Faraday cup

to detect the matrix species and other large ion signals. For the detection of

minor and trace species, the Extrel system utilizes a channeltron continuous

dynode electron multiplier (Galileo Electro-Optics Corporation) and the Finnigan

uses a discrete dynode electron multiplier (Balzers SEV 217, Liechtenstein). Both

of these detectors are set off-axis to the quadrupole axis. These detectors may

be operated in an analog mode that detects the small currents hitting the

detectors or an ion counting mode that will detect discrete events when single

ions strike the detector.


o\ m
\ c(
1 o


CM o

L E.

S\ *E
\\ I E <1

10 0 10 0

I-- -Oo 00
\ E



Faraday cup. The Faraday cup is the simplest detector employed in

mass spectrometry. It consists of a metal plate (electrode) that is used simply to

collect the ions that strike it. The ions are neutralized on the surface of the

electrode by a transfer of electrons, and the current that results is equal to the

incoming ion current. The Faraday electrode is enclosed, except for a small

entrance aperture that allows the incoming ions to pass through. This shielding

will prevent secondary electrons from escaping before striking the electrode,

which could lead to erroneous readings. The detected currents on the Faraday

electrode are usually small and a high gain signal amplifier is needed. The

Faraday detector requires a well insulated system in order to reduce any leakage

currents. The minimum current that is detectable using this detector is

approximately 10"14 amps [63].

Channel (continuous dynode) electron multiplier. Channel electron

multipliers (CEMs) are widely used as detector devices in many quadrupole mass

spectrometer systems [64]. The characteristics of CEMs that have lead to their

widespread use include their excellent signal-to-noise (with dark counts of less

than 0.5 counts per second), a stable dynode surface that can be exposed to air

without degradation of performance, low power requirements, compact size, and

a narrow gain distribution of output pulses. CEMs are constructed from a

formulation of glass that is heavily doped with lead. The fabrication produces a

glass that exhibits secondary emissive properties and resistive characteristics.

The Channeltron detector typically has a resistance of about 109 ohms. Electrical

contacts are deposited on both ends of the channel to allow electrical contact of


the CEM to an external high voltage. The external voltage will replenish the

charge on the channel wall, as well as accelerate low energy secondary electrons

to a level where they will create cascading secondary electrons upon collision

with the wall. Gain on the order of 108 is possible. This detector should be

operated in a vacuum of at least 106 Torr, since higher pressures will lead to

increased background signal and the life of the detector will be shortened.

Ion detection in the CEM is as follows. If a particle has sufficient energy

to cause emission of at least one secondary electron upon striking the interior

surface of the detector then it may be detected. The secondary electron will be

accelerated by the electrostatic field inside the channel until it hits the surface

again, which should result in the release of one or more secondary electrons.

This cascading process may occur 10 20 times in the channel depending on the

design of the multiplier. The gain of a CEM does not depend on the length of the

channel or the diameter of the channel alone. The gain is a function of the ratio

of the channel length to its inner diameter [65] as well as a function of the

potential difference from end to end (i.e., the applied high voltage). Using this

knowledge, channeltron multipliers have been constructed with very small sizes

and in arrays. The detection efficiency of a CEM was found to be a function of

the incoming ion energy, its velocity, and the degree of ionization [66].

Discrete dynode electron multiplier. The secondary electron multiplier

(SEM) operates in a fashion somewhat similar to the CEM. In the SEM, ions

coming from the mass spectrometer are accelerated by an electric field and strike

an electrode (or dynode), which is made of a material with a high coefficient of


secondary electron emission. These electrons will in turn produce more

secondary electrons that are accelerated by the field to a second dynode where

they hit and produce other secondaries. This process repeats until a large

number of electrons from the last dynode strike a Faraday plate, which generates

a large signal. The current is amplified as mentioned previously in the discussion

of the Faraday cup. The first dynode of the detector is usually operated at -1000

to -3000 volts that results in an overall gain of 104 108. Thus a current of 10'18

A would be multiplied to about 10-12 A, which is easily measured by the amplifier.

Analog filtering with a bandwidth of a few seconds is usually needed to obtain a

good signal-to-noise ratio [63]. The SEM, as well as the CEM, gives improved

detection limits and an improved response time over the Faraday cup, which

allows the instrument to be scanned faster.

One of the disadvantages with electron multipliers (both CEMs and SEMs)

is that the gain will change as the detector ages. This can be corrected by

measuring gain curves often and adjusting the operating voltage of the detector.

The gain is usually calculated by comparing the voltage detected on an electron

multiplier with that on the Faraday cup using the same ion beam. The SEM and

CEM can both operate in pulse counting modes, where each individual incident

ion produces one pulse. Each single ion that strikes the detector will produce

about 10' electrons, which is a readily detectable pulse. The detection limit of the

pulse counting mode is determined by the background count rate, which is

usually about 1 count per second. Pulse counting circuits are more expensive

than are analog measurement circuits; as a result, the ion currents should be

below about 1012 A for this method to be worthwhile. Also, if the current is too

high, pulse pile-up will occur in the detection system and the counting becomes


Vacuum system

The vacuum systems of the two instruments differ, but the requirements

and care that must be taken to maintain a vacuum system are virtually the same.

The Extrel system uses two diffusion pumps (NRC, Model VHS-4), while the

Finnigan uses two turbomolecular pumps (the first chamber is evacuated with a

330 L/s pump and the analyzer chamber is pumped by a 170 L/s pump). One of

the difficulties that is encountered when merging a glow discharge system is the

pressure difference required to operate the source and the mass analyzer. This

problem is alleviated by the use of differential pumping through small apertures

between three chambers in the instrument. The ion source operates at a pressure

of about 1 Torr while the first chamber maintains a pressure of about 104 Torr.

These chambers are separated by a 0.5 mm ion exit orifice. The analyzer

chamber is maintained at about 106 Torr during operation and is separated from

the first chamber by the small hole in the skimmer cone. The pressure is

measured in the ion source of both instruments with thermocouple gauges

(Teledyne-Hastings, Model DV-4D).

Data collection

Both instruments were operated primarily using a Hewlett-Packard 9816

computer system to control the mass scanning as well as collect the data. A

block diagram of the detection and data acquisition system for the Channeltron

detector of the Extrel instrument is shown in Figure 16. The system for the

Finnigan is virtually identical in its operation.

For detection in the analog mode the signal is passed through a

preamplifier and an electrometer. The signal is then directed to an oscilloscope

(Tektronix). The Extrel instrument's oscilloscope (Tektronix Model 5223 with

digitizing time base Model 5B25N) allows real-time mass spectra to be observed

since the quadrupole controller operates in conjunction with the oscilloscope to

allow an external voltage ramp (from the time base output) to control the mass

scan. The analog signal may also be routed through a voltage-to-frequency

converter to measure the ion signals on a multichannel analyzer (MCA) (Tracor-

Northern TN-7200). Finally, the system is connected to the Hewlett-Packard

computer via an IEEE bus system to allow the signal to be converted to a digital

signal that may be stored and manipulated with the computer software.

In counting mode, the signal is fed into an amplifier system that will

eliminate any signal pulses that are below a given threshold discriminatorr

feature). The amplification stage provides a high speed source of pulses (30 ns)

for signal counting devices. The signal can then be directed to the MCA and

collected as counts per channel dwell time. The HP computer may then be used

to read the MCA data for subsequent storage, data manipulation, and printing of

hard copies with the laser printer.

Neutral Species in the Glow Discharge Plasma

The mass spectrometer system will only allow the study of species that are

ionized in the glow discharge ion source. However, it is often advantageous to













correlate the ion population that is measured with the mass spectrometer to the

neutral and excited atomic population that exists in the ion source. This is

accomplished by interfacing the techniques of mass spectrometry with atomic

absorption (AA) to measure the ground state atomic species, or atomic emission

(AE) to measure the excited neutral species. These experiments were performed

only on the Extrel system since its six-way cross ion source will allow

simultaneous measurements to be made.

Mass spectrometer/atomic absorption system

A block diagram of the MS/AA system is shown in Figure 17. A

commercial neon or argon filled hollow cathode lamp (HCL) was focused down

to a diameter of about 3 mm that was directly adjacent to the ion exit orifice. After

passing through the negative glow, the beam was focused into a 0.5 M Ebert

monochromator (Jarrell-Ash, Waltham, Massachusetts). The intensity of the light

beam was detected by a photomultiplier tube (Hamamatsu R956, Bridgewater,

NJ) coupled with a photometer (Model 110, Pacific Instruments, Concord, CA).

The HCL was powered at a constant current of 6 10 mA and a peak voltage of

300 V by a DC power supply (Hewlett-Packard Model 6525A). The output of the

HCL was modulated at 50 100 Hz by a mechanical chopper (Scitec

Instruments). A reference signal from the chopper was directed to a lock-in

amplifier (Model 5301, EG&G/Princeton Applied Research, Princeton, NJ), as was

the signal from the photometer. The lock-in will discriminate against any

background emission signal that is produced in the negative glow. The output

signal from the lock-in amplifier was then directed into a voltage-to-frequency

IH l,




converter to allow the signal to be collected on an MCA. The MCA data can then

be read by the HP computer and manipulated or stored. The signal may also be

simultaneously monitored on an oscilloscope.

Mass spectrometer/atomic emission system

A block diagram of the MS/AE system is shown in Figure 18. The emission

experimental setup is basically the same as used for absorption, only much

simplified. With AE, there is no need for an HCL, chopper, or lock-in amplifier.

The emission signal is collected from the negative glow and focused into the

monochromator with one pair of lenses. The detected signal is then fed directly

into the oscilloscope and voltage-to-frequency converter, and subsequently into

the MCA. The collected data is then stored and manipulated by the HP

computer. This system is much simplified since there is no background

subtraction requirement for emission measurements.

2 E

LL )
LL. Q(



As discussed in the first chapter, glow discharge mass spectrometry has

found the majority of its applications in the analysis of solid samples [67,68].

However, this chapter will focus on the introduction of solution samples into a

GDMS system. Solution analysis with a glow discharge is an advantageous

avenue to explore, since many samples already exist in this form. A scheme has

been developed that couples a glow discharge, using its ionization abilities, with

an electrothermal vaporization device, which is very efficient for sample

atomization. This method will reduce any need to manipulate the solution before

glow discharge analysis (i.e., mixing with a solid material), although the solvent

must be removed prior to placing the sample into the mass spectrometer, since

the glow discharge is normally operated at a low pressure (1 Torr). This method

uses electrothermal vaporization of the solution residue to introduce the analytes

into the glow discharge plasma.

The analysis of solution samples is an obviously important facet of

chemistry. Many classic analytical methods were developed from the needs for

solution sample analysis. One of the fastest growing modern techniques for the

analysis of trace elements in solution is the inductively coupled plasma (ICP) [69].


The detection of analytes in an ICP may be performed using atomic absorption

(AA), atomic emission (AE), and/or mass spectrometry (MS). ICP-MS uses

differential pumping and skimmer cones with small orifices (similar to that

described in Chapter 1 for GDMS) to allow the mass spectrometer to operate in

the region of 10e to 10" Torr, while the ion source operates at 760 Torr (1

atmosphere). The ICP technique is amenable to the introduction of a large flow

of solution, since the samples are introduced through pneumatic nebulizers that

remove most of the solvents. Solvent particulates that may enter the source can

produce detrimental effects in the plasma, and the properties of the aerosol

relative to the analytical performance of the ICP are of great concern. Differences

will occur in ICP emission signals depending on whether the incoming aerosol is

wet or dry [70]. It was found that the small amount of residual water remaining

after the desolvator did not change the plasma conditions. When the dry aerosol

was mixed with H20/HNO, droplets, there was a reduction in the ICP emission

signal by 43.1 %. Compared to the GDMS source operation, the ICP is able to

handle solution samples more easily, due to its higher source operating pressure,

larger pumping capacity, and ability to handle large fluxes of material. The argon

flow alone into an ICP is in L/min, whereas the glow discharge only can handle

a few mL/min to maintain its operating pressure.

This chapter will begin by briefly reviewing some of the modern analytical

techniques that are used for the analysis of trace metals in small volume solution

samples. The emphasis of this discussion will center on methods using

electrothermal vaporization of the analyte, since the glow discharge solution


analysis system was developed using an electrothermal filament for sample

atomization. Finally, the experiments performed during the development of this

method will be discussed.

Electrothermal Vaporization

Electrothermal vaporization (ETV) is an important technique that is used to

provide controlled and fast atomization of a sample deposited onto an

electrothermal device. This device can take many different forms, such as a

filament of graphite furnace. ETV occurs when a sample, is heated to a

temperature above the vaporization point of the elements to be analyzed. As a

result, neutral and ionic elemental species are released from the electrothermal

device. The atomic population produced may be detected and analyzed by a

variety of methods, including AE and AA. The atoms may then be subsequently

excited (either from the ETV energy itself, or by other external methods) to

produce ions that can be detected by ionic emission, ionic absorption, or, as in

the present study, quadrupole mass spectrometry. The following sections will

give some background and examples of some techniques using ETV.

Furnace Techniques

Many solution analysis schemes use a heating device (such as a flame or

furnace) to produce excitation of the species to be analyzed. This section will

describe some techniques that use a graphite tube furnace in conjunction with a

glow discharge hollow cathode plasma to study the atoms and ions from a

solution sample. The methods discussed here rely on the glow discharge

processes to provide the excitation or ionization, so they are termed "nonthermal


excitation" methods. The ETV/GDMS method that was developed for the solution

analysis to be described later was based on some of the principles of these


Furnace atomic nonthermal excitation spectroscopy

The solution sampling method reported in this dissertation is a variation on

the furnace atomic nonthermal excitation spectroscopy (FANES) technique [71].

In FANES, a thin layer residue is analyzed after solvent evaporation of the solution

sample. The evaporation of the solvent and removal of the analyte from the

graphite furnace surface is accomplished by varying degrees of heat generated

in the furnace by a high current flow. The analyte is then detected by atomic

emission from the graphite tube glow discharge. A block diagram of the FANES

source is shown in Figure 19. The analysis procedure is as follows: 1) pipetting

the sample at atmospheric pressure, 2) drying the sample at atmospheric

pressure, 3) ashing the sample at atmospheric pressure, 4) pumping out the

source, 5) filling the source with argon, 6) ignition of the hollow cathode

discharge, 7) setting the power supply for atomization, 8) detection of the

emission pulse, and 9) cooling the source. Each measurement cycle lasts about

one minute and is under microprocessor control. The discharge used for FANES

is the hollow cathode type that was discussed in Chapter 1. The FANES

technique was first reported by Falk et al., who described a new emission source

with independent atomization and excitation processes [71,72]. The source was

found to have limits of detection comparable to flameless atomic absorption, while

orders of magnitude better than ICP emission. Detection limits of the FANES


S o z oo-F
X z Or
( F- D m




< C a-



-J o
o Om

z <0 c
w Z C -C


z 'z
m~( -1 O .9
0~C > Om "
O RE o

technique, using a 50 /L sample, were in the range of 0.0007 to 16 /g/L
depending upon the element under study. Also, the FANES method was shown

to exhibit a high dynamic range and multi-element capability. This technique has

been compared to other atomic spectroscopic methods in a paper by Falk et al.

[73]. Similar to FANES, the furnace ionic nonthermal excitation spectroscopy

(FINES) technique has been used to measure the emission from ions that are

produced in the source described above. Since this method is also considered

nonthermal, the ions are produced from energy provided by the glow discharge

plasma in the source.

Molecular nonthermal excitation spectroscopy

The technique of molecular nonthermal excitation spectroscopy (MONES)

was developed for the studies of molecules in the FANES source. This method

has been used for the determination of negative ions that combine with metallic

species. For instance, the halides F and CIl have been determined by the

nonthermal excitation of MgF and MgCI molecules [74]. For this determination,

the glow discharge was formed within a graphite tube operated at 2000 oC. The
solution containing the halide is mixed with a solution containing the Mg2+ ions

as an additive, with a concentration higher than the halide to promote reaction to

form the MgX molecule. Upon formation of the MgX in the gas phase, molecular

absorption measurements were made and detection limits of 0.5 ng of fluorine

and 0.24 ng of chloride were obtained. This method shows how reactions may

be carried out and studied in an electrothermal device. MONES experiments

were developed after preliminary FANES studies indicated the formation of free


atoms and small molecules in these discharges [74,75,76]. Recently a microwave

induced plasma has been coupled with the MONES method [77]. In these

experiments, nitrogen and oxygen were determined in gaseous and aqueous

samples through the formation and emission of NH and OH radicals.

Introduction of Micro-Samples into an ICP

As pointed out in the introduction, ICP methods have provided a means for

analyzing solution samples that are introduced into the torch in vast quantities

with a large supporting gas flow. The ability of the ICP to desolvate a solution

sample and submit ions to the mass spectrometer makes it a very prominent

technique for bulk liquid analysis. However, if the amount of solution to be

analyzed is in short supply and only a small quantity is available, a conventional

ICP cannot be used.

Recently, interest has grown in measuring microliter samples with the ICP

by interfacing it with an electrothermal source. This was performed by attaching

an ETV device to the ICP torch using 1 2 feet of plastic tubing. The ICP argon

gas is passed through the graphite furnace as the sample is vaporized, producing

an atomic population that is representative of the sample. These atoms are then

swept, by the gas flow, into the ICP torch for subsequent ionization. The ICP may

be operated in single ion monitoring mode to obtain the maximum sensitivity for

a given element or in mass scanning mode to provide multi-element analysis.

The advantages of coupling ETV and ICP-MS include detecting elemental sample

amounts in the low femtogram (fg) region, performing trace analysis on very small

volume samples (ug or pL), and alleviating the matrix and solvent interference


that may be encountered with a typical ICP-MS analysis. Sensitivity values from

20 80 counts/fg have been reported, resulting in detection limits in the low fg

range [78]. The measurement precision is in the range of 5 20 %. Calibration

curves have been found to be linear over 2 3 orders of magnitude, although

quantitation usually requires using the standard additions method for analysis.

Memory effects and matrix influences are considered serious constraints for the

ETV-ICP/MS technique and these problems are under investigation.

Solution Samples in the Glow Discharge


As previously mentioned, the most common samples analyzed in GDMS

are solid materials in the form of pins machined from bulk solids or disks formed

in a die using powdered samples. The research presented in this chapter is on

the development of a method to introduce small volume aqueous solution

samples into the GDMS ion source to broaden its range of application. Since the

glow discharge source requires operation in a low pressure environment, it is not

amenable to the introduction of solvents in a sample. If a solution sample is

introduced into the discharge plasma without desolvation, it may extinguish the

negative glow. Therefore, strict control is needed to limit the amount of solvent

that is brought into the source. Some methods have been employed over the

years in an attempt to provide GDMS solution analysis and reduce solvent effects

in the vacuum system. The methods employed generally require placing the

sample onto a metal substrate and evaporating the solvent, leaving a thin film of

solution residue (containing the analyte atoms of interest) on the cathode surface.


Methods that were performed earlier in this laboratory involved the use of a

copper rf hollow cathode source for solution analysis [79], as well as applying a

solution to a cupped cathode [80]. In these experiments, the applied sample was

evaporated, and then atomized and ionized by the discharge processes. This

allowed for analysis of the solution residue by sputter atomization into the glow


Electrothermal Vaporization/Glow Discharge Techniques

When combining electrothermal vaporization and glow discharge ionization,

the strengths of both systems may be used. The electrothermal source provides

a high current that will sufficiently atomize the solution residue in a short time,

while the glow discharge is a very efficient source for excitation and ionization

because of the energetic plasma processes. The glow discharge can be used by

itself to sputter atomize the sample, but this process is limited by the lower

currents allowed in the discharge. Thus, sputter atomization is a slower process

that releases the sample over a longer time, reducing sensitivity. For optimum

sensitivity, the cathode should be carefully chosen to be nonporous. If the

cathode is porous, the solution residue will reside at least partially inside the

cathode and will take even longer to analyze. The analyte signal in this case will

be constant and long lived, but the detection limits will be reduced. An alternative

method has been investigated where the solution was mixed with a conducting

powder, dried, and then pressed into a discharge cathode for subsequent

analysis [81]. The method is thought to be a step forward in the construction of

specific standard cathodes to be used for quantitative glow discharge analysis


with relative sensitivity factors, since solid standards containing the elements of

interest are sometimes hard to find.

Electrothermal vaporization/glow discharge atomic emission

Electrothermal vaporization into a glow discharge (GD) emission source

has been demonstrated, using a tungsten wire for atomization of solution samples

[82]. In these experiments, silver and boron solutions were studied using atomic

emission. The discharge was operated at a current of 60 mA and at 1.2 Torr

argon pressure. The maximum filament current of 16 A was used for sample

vaporization. The effect of various solution additives on the emission signal was

also studied and the presence of up to 1000 ppm of aluminum, calcium,

magnesium or sodium had little or no effect when added to a silver solution.

Electrothermal vaporization/alow discharge mass spectrometry

In ETV/GD analysis, the heating of the filament serves as the primary

atomization step. Although some atomization will still occur by glow discharge

sputtering, the amount of sample loss by sputtering will depend on the

temperature of the filament and the time required for the solution residue to be

removed by thermal methods. The glow discharge provides the only apparent

mechanisms for ionization for the analyte under investigation, but the lack of ionic

detection in the absence of a glow discharge is not conclusive evidence that ions

are not being formed directly from the energy provided by the filament. The ions

formed, if any, may be of the wrong energy to pass successfully through the ion

lenses of the mass spectrometer and make it to the detector. One disadvantage

of the ETV method is the need to devise detection schemes that will accurately


handle transient signals. The signal peak that is produced from a small volume

solution sample will typically last only a few seconds, thus fast detection systems

are needed.

The schematic diagram shown in Figure 20 shows a simplified overview of

the plasma and its processes for ETV/GD analysis. The analyte to be studied (Zn

in this case) is shown to reside on the surface of the cathode (the tungsten

filament). The plasma sputtering process will liberate both tungsten and zinc

atoms from the surface as the argon ions bombard the filament. This results in

some loss of analyte before initiation of the filament current, but the discharge

"on" time will be minimized in the sampling scheme, as described later. As with

a normal glow discharge, Penning and electron ionization will occur with both the

solution residue and filament material atoms.

The results presented in this chapter are intended to show the usefulness

of this method for introduction of solutions into the glow discharge. Most of the

experimental data are concerned with the fundamental processes that occur in the

glow discharge and demonstrate how different characteristics of the discharge

environment (such as pressure, voltage, and current) may affect the ion signal

and change the analytical usefulness of the method. Also, the presence of a

filament carrying current in the vicinity of the glow discharge may show some

positive or negative effects on the ion transmission and the signal detected by the

mass spectrometer.

This ETV/GD method is of interest as well since it provides a separation of

the atomization and ionization steps in the analysis. Experiments that separate







0 W



Figure 20 Simplified diagram of the glow discharge processes
occurring in the ETV/GD system.


- "-'),VO

* o e-

atomization and ionization are always of interest since this provides another

dimension with which to evaluate the fundamental processes of the glow

discharge. Other experiments have been performed to show the utility of

separating atomization and ionization steps in the glow discharge analysis. It has

been shown that laser ablation can provide a means of atomizing the sample into

a discharge plasma, which subsequently provides the ionization mechanism [83].

These studies show that the atomization and ionization processes are both

controlled by the laser and discharge parameters. In parallel, the primary

atomization may be thought to occur upon heating the filament, while the primary

ionization is provided by the glow discharge.

Ion Source Development

Experiments in a Six-Way Cross

The initial studies were performed using the Extrel ion source (six-way

cross) that was described in Chapter 1. A diagram of this source, with the ETV

device in place is shown in Figure 21. The source allows easy access to all of

the components inside and application of the solution sample to the filament. A

dual high voltage feedthrough, mounted in a 2.75" rotating flange (MDC High

Vacuum Products Corp.; Hayward, CA), was used to hold the filament and its

supporting assembly. This allowed for ease in filament alignment and

replacement. The filament loops for these experiments were 4 mm in diameter

and 2.75 mm long and were constructed by coiling a rhenium ribbon 3 turns

around a 3.2 mm metal rod. The filament was held in place by two stainless steel

rods and a small copper support wire in such a manner that the center of the coil










was aligned coaxially with the exit orifice. This was done because the majority of

the solution residue will lie on the inside of the filament loop and ions traveling out

from the center of the coil will most likely be directed through the ends of the coil

and toward the exit orifice. Set screws secured the support rods to the dual

feedthrough as well as mount the filament in a fixed position. Glass shields were

constructed to protect the stainless steel rods from the discharge environment,

allowing only the filament loop and about 10 mm of a copper support wire to be

unprotected. The external cathode was mounted on the end of a probe and

placed in the mass spectrometer through a bellows assembly that allowed for

placement of the cathode in a variety of positions around the filament coil. The

discharge was operated in both constant current and constant voltage mode

using a Heinzinger HNCs 2500-150 ump. power supply, allowing operation up to

a maximum of 2500 V or 150 mA. The studies discussed here were performed

at about 1 mA current and voltages ranging from 400 800 V, with pressures in

the range of 0.45 2.5 Torr. The current loop was powered by a MAT 261

filament power supply (Finnigan MAT; Bremen, Germany), which allows up to 10

A of current to be supplied to the filament for sample atomization. The maximum

current varied from 4 6.6 amps, depending on the filament material and the

analyte to be studied. Table IV shows some materials that were initially studied

in order to find an appropriate filament to provide a "white hot" glow capable of

vaporizing the analyte of interest. The filament power supply was insulated since

it floats at a high potential when the filament was used as the discharge cathode.

A stepper motor assembly was constructed to control accurately the output

TABLE IV. Properties of Some Possible Filament Materials

Material Diameter (mm) Current (A) Observations
Al 0.25 4 glows & melts
Brass 1.2 6.6 -
Chromel 0.35 2.6 red glow
Cu 0.9 6.6 -- -
Ni 1 6.5 ---
Re (ribbon) 0.1 x 0.75 4.2 white glow
Ta 0.015 4.5 white glow
W 0.010 5.5 white glow


current of the MAT 261. The current was supplied to the filament by RG-59

coaxial cable and high vacuum feedthroughs.

Filament on a Probe

A problem with this type of source is that there may be slight differences

in the construction of individual filament coils. Also with the rigid support system

there is no way to optimize the distance from the cathode to the exit orifice, which

is an important parameter for detecting the maximum ion signal with GDMS.

Figure 22 shows how the ion signal of tungsten, sputtered from the coiled

electrothermal filament, varies with the cathode-to-exit orifice distance. It can be

seen that the maximum ion signal occurs when the cathode is about 7 mm away

from the ion exit orifice, a typical value for a 1 Torr glow discharge. With these

problems at hand, it was realized that a direct insertion probe (DIP) should be

used in an effort to solve the problem of positioning. A 0.5" outer diameter DIP,

shown in Figure 23, with a 10-turn tungsten filament was constructed to allow the

introduction of a solution residue sample into the ion source without venting and

with the ability to position accurately the filament coil with respect to the ion exit

orifice. The DIP also will allow the use of an auxiliary cathode, if so desired, since

the probe contains four vacuum feedthroughs. Two of the feedthroughs carry the

electrothermal current and discharge voltage for the filament, while one may carry

voltage for an auxiliary discharge cathode. Initial experiments using the DIP show

comparable results to those obtained in the six-way cross, although the ion signal

can be maximized with the DIP due to its flexibility.

cd (9 CJ C\i V

0 0 0
,- 0 0

(90 [x slunoo) Ieu6!s Uol Mgs9L

cm E



b /


o E
T (U
) -


- /)


u- o

< UJ





- L


I o












4 .


it .


Sample Preparation

Solutions were prepared by mixing an appropriate amount of a metal

nitrate with distilled water to obtain the desired cation concentration. The

chemicals that were used are as follows: nickel nitrate {Ni(N03)2 6H20}

(Mallinckrodt Inc.; Paris, Kentucky; Analytical Reagent Grade), zinc nitrate

{Zn(N03)2 6H20} (Fisher Scientific; Fair Lawn, NJ; Certified A.C.S.), cupric nitrate

{Cu(NO3),2 21/2H20} (Fisher Scientific; Fair Lawn, NJ; Certified A.C.S.), ferric

nitrate {Fe(N03)3 9H20} (Mallinckrodt Inc.; Paris, Kentucky; Analytical Reagent

Grade), and silver nitrate {AgNO3} (Mallinckrodt Inc.; Paris, Kentucky; Analytical

Reagent Grade). Enough concentrated nitric acid (Fisher Scientific; Fair Lawn,

NJ; Reagent A.C.S. grade) was added to the solutions to provide an acidic

concentration of 1 2 % HNO3. Fundamental studies of the source properties

were carried out with analyte concentrations of approximately 1000 ppm.

Sampling Sequence

The sampling sequence that was devised for the electrothermal

introduction of solution samples went through a variety of stages in its

development. There were a few considerations to make in order to ensure the

most accurate analysis of the samples. First, the solvent on the filament had to

be completely evaporated to minimize any detrimental effects that may be

obtained by excess water being introduced into the mass spectrometer. Second,

the current used for solvent evaporation had to be less than that required to

remove the analyte from the filament. Third, the filament material should be

stable for the evaporation step. Some of the filament materials could not

withstand the lower evaporation currents when carried out in an air atmosphere.

This was particularly true for the rhenium ribbon, which would completely oxidize

and break unless the evaporation was performed at a low pressure (= 20 Torr)
or in an inert gas atmosphere. Finally, the discharge operation time before the

filament current is initiated should be long enough to ensure a stable discharge,

but as short as possible to minimize any loss of sample analyte due to sputtering.

The final version of the sampling sequence that was used for sample

introduction on the DIP is as follows: After bringing the source up to atmospheric

pressure, the solution samples (ranging from 10 40 1L) were directly pipetted
onto the filament loop by a microliter pipet [Volac Model # R 880/A, 5 50 UL
adjustable volume; Great Britain]. The solutions were evaporated at about 1.5 A

filament current for 2 min. The current was increased to 2 A, and the evaporation

step continued another 2 min (after this time the coil surface appeared to be dry).

The probe was placed into the instrument and the roughing pump applied for 1

min. The sample was allowed to sit under high vacuum conditions for 2 min after

placing it into the ion source. The argon was then introduce at the working

pressure, usually 1 Torr, and the pressure was allowed to equilibrate for 1 min.

The detector was then turned on for 30 s. The MCA was turned on for 5 s to

measure any background signal that may be present. The discharge was turned

on and allowed to run for 5 s. The filament current supply was turned on to the

appropriate atomization current for the sample. After data collection, the filament

was allowed to continue heating for 2 min to drive off any remaining species and


reduce memory effects. The filament assembly was allowed to cool for about 5

min before another cycle was repeated with the next sample.

Data Acquisition

When studying solution samples with the electrothermal source, the data

obtained consisted of a transient signal lasting a few seconds. A schematic

representation of the data acquisition scheme is shown in Figure 24. The

quadrupole was set to pass the mass of interest and the signal from the detector

was monitored using the MCA. The MCA was first turned on to allow detection

of any background signal. The discharge was turned on, at t = 0 s in the figure,

for 5 s to allow stabilization of the discharge voltage. This time was chosen as

the shortest time to allow a pre-sputtered filament coil to attain a stable discharge.

This time was minimized to allow most atomization to take place due to the

electrothermal vaporization. The filament was then turned on, at t = 5 s in the

figure, and quickly rises to the desired current. It may be noted that the spikes

that are produced at this time are due to electrical noise in the system upon

initiation of the filament current. After about a 1 s delay, the ion signal from the

solution residue is detected as a peak.

Filament Materials

The results that were obtained with different filament materials are

discussed in this section. The materials under active investigation were rhenium,

tantalum, and tungsten. These were the only three filaments that resulted in the

"white hot" glow required for vaporization of the samples (see Table IV for the

filaments attempted and some observations). The initial study used a rhenium

4 8 12 16
Time (s)





Figure 24 Diagram of the relationship between the detected ion
signal with respect to the operation of the glow discharge and
electrothermal filament.












ribbon as the atomization source. Studies were performed two ways: 1) with the

filament acting as the discharge cathode for ionization, and 2) with ionization

being provided by the discharge supported by an auxiliary cathode (as shown in

Figure 21). Both methods produced comparable ion signals for silver solutions,

but the latter technique was subsequently adopted because of its simpler design.

If the rhenium ribbon is subjected to multiple runs, the glow discharge sputtering

process will "eat" through the ribbon and cause a discontinuity in the current loop.

Tantalum wires were also tried as an ETV filament but they too degraded very

quickly. The majority of the experiments was performed using a tungsten wire.

This wire was the most stable of the three in terms of producing sample ion

signals and the long term ability to withstand the sputtering processes.

Filament-Plasma Interactions

The glow discharge processes and operation may be disturbed by the

operation of the electrothermal device within the ion source region. This section

will describe some considerations that must be made when coupling an

electrothermal device, carrying a high current, with the glow discharge device.

The experiments described in this section were performed to observe the changes

in a typical glow discharge signal when applying a current to the electrothermal


Ion Transmission

The first concern when using the static coil experimental setup was the

effect that the filament current would have on the transmission of ions from the

auxiliary pin cathode. In these experiments, the discharge cathode was separate


from the filament and the cathode pin could be positioned around the filament.

Different positions were investigated for the placement of the discharge pin

cathode. These included placing the pin behind the filament, such that atoms

from the pin must traverse through or around the filament to reach the exit orifice,

placing the pin inside the filament and co-axial to the ion exit orifice, causing the

negative glow region to be confined in the cylindrical space defined by the coil,

and placing the pin beside the filament coil. The ion signal transmission was

found to be greatest with the pin beside the filament coil, so this configuration

was used. This orientation was also allowed with the direct insertion probe.

Effects of different filament currents

Changes may occur in the glow discharge processes when the filament is

carrying a relatively high current (three orders of magnitude more than the

discharge itself). A glow discharge pin cathode was operated in the constant

voltage mode while being positioned beside of the filament. As the filament

current was increased from 0 to about 5 A, the discharge current was monitored.

The results of this experiment are shown in Figure 25. The glow discharge

current is observed to decrease slightly over the range of 0 4 A of filament

current. After this point the glow discharge current shows a dramatic increase.

This may be the result of electron emission from the filament, which may add to

the flow of electrons toward the anode and thus increase the measured current

of the gaseous conductor.

The next experiments were performed to observe how the filament current

would affect the ion signals when the electrothermal filament itself was used as




g 0

L.. C



To r ll

O CO C ) C1C C)) :
S 6 (6 4 i d E 7L

) ua a

(vwJ) luejjflQ 86jaL13S!O AMO|


the discharge cathode, since this much simpler experimental design was used

most often. The effect of filament current on the Re+ signal is shown in Figure 26.

The reported signal is the peak height of the 87Re+ isotope after scanning a

mass range containing the rhenium isotopes 25 times. The initiation of the

filament current results in a fairly constant decline of the ion signal, when

operating with a 1 mA constant current discharge. The discharge voltage

changes from 598 V with no filament current to 740 V with 3 A filament current.

Overall, there is about a 67% decrease in the ion signal from 0 3 A, probably

due to changes in ion transport induced by the current present near the plasma,

possibly as a result of an induced magnetic or electrical field in the area.

These changes were also observed when running the discharge at a

constant voltage of 500 V. Figure 27 shows the ion signals for some gaseous

species (H20 at m/z = 18, COH and N2H at m/z = 29, and ArH at m/z = 41) and

one sputtered species (Re at m/z = 187) with increasing filament current. All of

the ions in this experiment respond in a similar manner and seem to follow the

same trend as the discharge current (as seen in Figure 25). Thus, it appears that

the filament current (and the subsequent changes in the discharge current) has

a major effect on the magnitude of the ion signals obtained from the discharge.

This observed increase at the higher current led to subsequent operation of the

discharge in the constant voltage mode for the analysis of solutions, since high

currents will most likely be used and this will result in a higher ion signal.

However, analyte signals from solutions may respond differently depending on

their atomization temperature.



ci o
L() CC

C c E


O = E


-o 4


O) c
7 t L L 6i

C) m E C

(O I. X sluno3) Iu6B!S uo| eOLda




V .0 OQ
II o
cn 3 '|
c ca

0 / > ,1
C Ca

6 >

LO 0


q i -,i- )

LO M 6-

()o o o o oo -
() co c UOi

(A) IbU61S U0|