close

Вход

Забыли?

вход по аккаунту

?

Fundamental studies of the sampling process in an argon inductively coupled plasma- and a helium microwave-induced plasma-mass spectrometer

код для вставкиСкачать
INFORMATION TO USERS
The most advanced technology has been used to photograph and
reproduce this manuscript from the microfilm master. UMI films the
text directly from the original or copy submitted. Thus, some thesis and
dissertation copies are in typewriter face, while others may be from any
type of computer printer.
The quality of this reproduction is dependent upon the quality of the
copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by
sectioning the original, beginning at the upper left-hand corner and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in
reduced form at the back of the book.
Photographs included in the original manuscript have been reproduced
xerographically in this copy. Higher quality 6" x 9" black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly
to order.
U n iversity M icrof ilms In ternational
A Bell & H ow ell Information C o m p a n y
3 0 0 N o rth Z e e b R o a d . A n n Arbor. Ml 4 8 1 0 6 - 1 3 4 6 U S A
313 761-4700
800 521-0600
Order N u m b er 9120489
F u n d a m e n ta l s tu d ie s o f t h e s a m p lin g p ro c e s s in a n a rg o n
in d u c tiv e ly c o u p le d p la s m a - a n d a h e liu m m ic ro w a v e -in d u c e d
p la s m a - m a s s s p e c tr o m e te r
Chambers, David Michael, Ph.D.
Indiana University, 1991
Copyright © 1990 by Cham bers, D avid M ichael. A ll rights reserved.
UMI
300N. ZeebRd.
Ann Arbor, MI 48106
fundam ental
argon
St u d ie s
In d u c t i v e l y
of the
S a m p l in g P r o c e s s
in a n
c o u p l e d p l a s m a - a n d a h e l iu m
M i c r o w a v e -I n d u c e d P l a s m a - M
a ss
Sp e c t r o m e t e r
BY
D a v id M ic h a e l C h a m b e r s
S u b m it t e d
t o th e f a c u l t y o f th e
Gr a d u a t e Sc h o o l
in p a r t ia l f u l f il l m e n t o f
THE REQUIREMENTS FOR THE DEGREE
D octor
in t h e
of
P h il o s o p h y
D epartm en t
of
C h e m is t r y ,
I n d ia n a u n i v e r s i t y
OCTOBER, 1990
A c c e p t e d by t h e G r a d u a t e F a c u l t y , I n d ia n a U n i v e r s i t y , i n
PARTIAL FULFILLMENT OF THE REQUIREMENTS OF THE DEGREE OF
D o c to r o f P h il o s o p h y .
GARY M.'ffiEFTJE, PH. D
RONALD A . HITES, PH. D .
DOCTORAL COMMITTEE
%
DENNIS G . PETERS, PH. D .
Ja m e s
date of
O r a l E x a m in a t io n :
O c t o b e r 17,1990
p
. Re il l y , P h .
d
.
© 1990
D a v id M ic h a e l C h a m b e r s
A l l R ig h ts R e s e r v e d .
iv
To my parents, John and Lupe
A CKNO WLEDGEMENTS
I would like to express may gratitude to all those individuals who unselfishly provided
intellectual, financial, and moral assistance throughout the duration of this research effort.
From a scientific standpoint I am grateful to those who were influential in guiding the
direction of my research; Dr. Pengyuan Yang, present and past members of my research
group, and especially my adviser, Professor Gary Hieftje, whose enthusiasm and integrity
served as a model. Furthermore, I am thankful for the intellectual resources provided by
Dr. Ken Busch, Dr. James Reilly, and members in both their research groups.
I would like to acknowledge the various groups that formed the technical support
network at Indiana University; the Mechanical Instrument Group, the Electronics
Instrument Group, and the Glass Blowing Group for their expertise. I wish to name the
following specifically;
Mechanical Instrument Group. John Dorsett, group director, for assisting in the
design and construction of the Langmuir probe assembly (Chapter 2), the retardingplate optics (Chapters 3 and 5), the interface sampling plates and skimmer cones
(Chapters 2-5), and the upkeep of the mass-spectrometer hardware needed to
successfully complete this study. Richard Martin for the construction of the Langmuir
probe Gimbal-type assembly and the retarding-plate optics. Larry Sexton for the
vi
Acknowledgements
construction of the sampling plates and skimmer cones and for teaching me how to
machine metal. Gary Fleener and Delbert Allgood for the upkeep of the rotary-vane
pumps. Kenneth Bastin for the construction of a mock interface used in preliminary
experiments.
Glass Blowing Group. Don Fowler, group director, and Don Garvin for making
the argon pneumatic nebulizers, the Langmuir probes (Chapter 2), and the helium
Beenakker cavity plasma torches (Chapter 5).
Electronic Instruments Group. Bob Ensman, group director, and Andy Alexander
for assisting in the design of the Langmuir probe, voltage-buffer circuit (Chapter 2)
and for technical maintenance of the system electronics for the mass analyzer, the
instrument detection system, and the source rf and microwave generators.
«
From a financial standpoint I am grateful to the Perkin-Elmer Corporation for the
donation of the ICP-MS instrument, the National Science Foundation for both my support
and my research, and most importantly Indiana University for providing an environment
conducive for research.
From both an emotional and financial standpoint I wish to acknowledge my parents,
John and Lupe Chambers, and the rest of my family, Eric Chambers, Harold Yelle, Mary
Chambers, Jay and Helen Chambers, Larry and Stacia Semeter. Additionally, I would like
to thank some very close friends; Jocelyn Dunphy, Marci Briles, Mary Carroll, Donna
Carter, Shawn Gelsinger, Susan Hieftje, Barbara Ross, and close members of my family
Tom Aldridge, Alice Dexter, Mary Parish, Muriel Tabler, and others who remained with
me throughout my graduate career.
fundam ental
s t u d ie s o f t h e
S a m p l in g
p r o c e s s in a n
ARGON INDUCTIVELY COUPLED PL A SM A - AND A HELIUM
MICROWAVE-INDUCED PLASM A-M ASS SPECTROMETER
D a v id M ic h a e l C h a m b e r s
In d ia n a U n iv e r sit y
B l o o m in g t o n , In d ia n a
Fundamental studies of the ion-transport mechanisms in an argon inductively coupled
plasma- (ICP-MS) and a helium microwave-induced plasma-mass spectrometer (MIPMS) are presented. The studies performed with the argon ICP-MS are focused primarily
on the examination of those forces (i.e., gas kinetic and coulombic) that influence ion
movement in a three-stage differentially pumped mass spectrometer interface, whereas
those with the helium MIP-MS are meant to assess the feasibility of sampling a helium
plasma. Experiments included Langmuir probe measurements, ion kinetic energy
measurements, and the examination of ion-beam density and composition; all as a function
of source and interface conditions. Results from these experiments, when combined, have
permitted the development of a qualitative ion-transport model, which describes those
forces that control ion movement in each stage of the mass spectrometer interface. In
addition, this fundamental investigation has permitted the identification of several
phenomena that contribute to matrix-induced interferences, mass-bias effects, and the lack
of long-term stability in both the argon ICP-MS and helium MIP-MS instruments. Based
on these findings future design improvements are proposed that are expected to reduce the
magnitude or sev<
viii
TABLE OF CONTENTS
C h a p t e r 1.
I n t r o d u c t io n
1.1. Introduction
1
1.1.1. Langmuir Probe Analyses
5
1.1.2. Retarding-Plate Analyses
7
1.1.3. Beam-Composition and Density Analyses
8
1.1.4. Helium Plasma-Source Mass Spectrometry
9
1.2. Literature Cited
11
C h a p t e r 2. L a n g m u ir P r o b e M e a s u r e m e n t s
2.1. Introduction
14
2.2. Experimental
2.2.1. Modification of ICP-MS for Langmuir Probe Measurements
2.2.2. Spatially Adjustable Langmuir Probe Assembly
2.2.3. Langmuir Probe Circuit
2.2.4. First-Stage Vacuum System
2.2.5. Sample Introduction System
16
16
16
23
23
26
2.3. Background
2.3.1. Current-Potential Measurements
2.3.2. Charge Separation at the Sampling Plate
2.3.3. Influence of Shockwave Structure on Electrostatic
Features in the Expansion Core
29
29
32
35
Table of Contents
ix
2.4. Results and Discussion
2.4.1. Stability and Characteristics of Measured Floating Potentials
2.4.2. Space and Floating Potentials
2.4.2.1. Effect of Solvent Load and First-Stage Backing Pressure
2.4.2.2. Effect of Solvent Load and Inner-Gas Flow Rate
2.4.3. Mapping Floating Potentials in the First-Stage
2.4.4. Local Floating Potential Variation with First-Stage
Backing Pressure
41
41
44
44
52
53
59
2.5. Conclusion
60
2.6. Literature Cited
64
C h a p t e r 3.
I on K in e t ic E n e r g y M e a s u r e m e n t s
3.1. Introduction
66
3.2. Experimental
3.2.1. Conversion of the ICP-MS into a Retarding-Plate Energy Analyzer
3.2.2. First-Stage Vacuum System
3.2.3. Sample Introduction System
3.2.4. Retarding-Plate Analysis
3.2.5. Plasma-Source and Operating Parameters
69
69
72
72
73
75
3.3. Background
3.3.1. Use of Retarding Plates to Determine Ion Kinetic Energies
3.3.2. Gas Dynamic Acceleration
75
75
82
3.4. Results and Discussion
3.4.1. Retarding-Potential Curves and "Potential Effects"
3.4.2. Retarding-Plate Analyses
3.4.3. Second-Stage Retarding-Plate Analysis
3.4.4. Third-Stage Retarding-Plate Analysis
3.4.5. Induced Ion-Ion Interaction in the Mass-Spectrometer Third Stage
3.4.6. Effect of Load-Coil Configuration on Mean Ion Kinetic Energy
84
84
96
105
108
113
125
3.5. Conclusion
130
3.6. Literature Cited
136
C h a p t e r 4.
M o n it o r in g t h e io n b e a m
4.1. Introduction
138
4.2. Experimental
4.2.1. First-Stage Pumping Configuration
4.2.2. Samples and Sample Introduction System
140
140
143
x
Table of Contents
4.2.3. First-Stage Skimming
4.2.4. Ion-Optic-Lens Configuration
4.2.5. Plasma-Source and System Operating Parameters
143
143
146
4.3. Background
4.3.1. Skimming the Free-Jet Expansion
148
148
4.4. Results and Discussion
4.4.1. Experimental and Calculated Optimal Skimming Conditions
4.4.1.1. Dry Plasma
4.4.1.2. Wet Plasma
4.4.2. Analyte Ion Beam
4.4.3. Interaction of Optics with Ion Beam
4.4.4. Effect of Inner-Gas Flow Rate on Analyte Signal
4.4.5. Oxides and Doubly Charged Species
151
151
152
160
165
175
183
189
4.5. Conclusion
192
4.6
196
Literature Cited
C h a p t e r 5.
H e l iu m p l a s m a S o u r c e - M ass S p e c t r o m e t r y
5.1. Introduction
198
5.2. Experimental
5.2.1. Ionization Source
5.2.2. First-Stage Vacuum System
5.2.3. Sampling-Plate Material
5.2.4. Sample Introduction System
5.2.5. Retarding-Plate Analyses
5.2.6. Potential Effects
5.2.7 First-Stage Pressure Measurements
201
201
201
202
202
204
205
205
5.3. Results and Discussion
5.3.1. Selecting the Sampling-Plate Orifice Diameter
5.3.2. Skimming the Free-Jet Expansion
5.3.3. Ion Kinetic Energy Measurements in the Second Stage
5.3.4. Effect of Air Entrainment on Ion Kinetic Energy
5.3.4.1. Influence of Off-Axis Sampling on Background Spectrum
5.3.4.2. Influence of Off-Axis Sampling on Helium Ion
Kinetic Energies
5.3.4.3. Effect of Air Intentionally Added to the MIP
208
210
214
220
235
235
238
5.4. Conclusion
248
5.5. Literature Cited
250
243
xi
Table of Contents
C hapter 6.
F uture W
ork
6.1. Introduction
6.2.
252
6.1.1. Single-Stage Interface
253
6.1.2. Multistage Interface
260
Conclusion
266
6.3. Literature Cited
270
LIST OF FIGURES
C hapter 2
18
Figure
2.1. Instrumental diagram of the ICP-MS configured for Langmuir
probe measurements
Figure
2.2. Langmuir probe Gimbal mount and first-stage vacuum
arrangement
Figure
2.3. Voltage-buffer-amplifier circuit
Figure
2.4.
Solvent load calibration plot for desolvation system
28
Figure
2.5.
Semilogarithmic plot of electron-accelerating region for a
current-potential curve
31
Figure
2.6. Free-jet expansion diagram
37
Figure
2.7. Floating potential versus time for a dry plasma
43
Figure
2.8. Effect of first-stage pressure on space and floating potentials
at different condenser temperatures deduced from I-V curves
46
Figure
2.9. Effect of first-stage pressure and condenser temperature on
floating potential taken with an electrically isolated probe
48
Figure
2.10. Effect of inner-gas flow rate on space and floating potentials
at different condenser temperatures deduced from I-V curves
50
Figure
2.11. Floating potential versus radial and axial position measured
under different plasma conditions
55
20-21
- 25
List of Figures
Figure 2.12. Floating potential as a function of radial and axial position
at different first-stage pressures
58
Figure 2.13. Floating potential versus first-stage pressure measured at
different lateral positions 10.0 mm downstream from the
sampling-plate orifice.
62
C h apter 3
Geometry of the ICP-MS interface configured for retardingplate analyses
71
Figure -3.2.
Method used to calculate mean ion kinetic energy from a
retarding-potential curve
77
Figure
3.3.
Retarding-potential curves obtained using the quadrupole
pole bias as a retarding field
81
Figure
3.4.
Second- and third-stage retarding-potential curves for lithium
86
Figure
3.5.
The effect of high solution concentrations on third-stage
retarding-potential curves measured for lithium and uranium
88
Figure
3.6.
Third-stage retarding-potential curves for argon while water,
1000 ppm lithium, and 1000 ppm uranium are nebulized into
the plasma
91
Figure
3.7. Retarding-potential curves for argon measured in the second
and third stage at different first-stage pressures
93
Figure
3.8. Second-stage retarding-potential curves for lithium, argon,
manganese, and uranium
98
Figure
3.9. Third-stage retarding-potential curves for lithium, argon,
manganese, and uranium
100
Figure
3.10. Mean ion kinetic energies deduced from retarding-potential
measurements performed in the second and third stages
102
Figure
3.11. Mechanism for the suppression of gas dynamic acceleration
in second stage
104
Figure
3.12. Second-stage mean ion kinetic energy curves taken at
different first-stage pressures and condenser temperatures
107
Figure
3.13. Third-stage mean ion kinetic energy curves obtained at
different condenser temperatures
110
Figure
3.1.
List of Figures
Figure
3.14. Third-stage mean ion kinetic energy curves at different
first-stage pressures
Figure
3.15. Retarding-potential curves measured in the third stage with
a 0.5-mm second-third-stage differential pumping aperture
at different first-stage pressures
Figure
3.16. Retarding-potential curves measured in the third stage with
a 1.0-mm second-third-stage differential pumping aperture
at a moderate first-stage pressure and solvent load
Figure
3.17. Retarding-potential curves measured in the third stage with
a 1.0-mm second-third-stage differential pumping aperture
at a high first-stage pressure and a low solvent load
Figure
3.18. Retarding potential curves measured in the third-stage with
a 1.0-mm second-third-stage differential pumping aperture
at a low first-stage pressure and high solvent load
Figure
3.19. Mean ion kinetic energies measured in the third stage with
a 1.0-mm second-third-stage differential pumping aperture
Figure
3.20. Mean ion kinetic energies measured in the mass spectrometer
third stage for an asymmetrically grounded and a center-tapped
load coil
Figure
3.21. Qualitative ion transport model
C hapter 4
Figure
4.1. Instrumental configuration for monitoring the ion-beam
composition
Figure
4.2. Second-and third-stage ion-optic lenses
Figure
4.3. Diagram of skimmed free-jet expansion in the first stage
Figure
4.4. Argon signal-pressure curves for a dry plasma with and
without an inner-gas flow taken at a 6.0-mm skimming
distance
Figure
4.5. Argon signal-pressure curves for a dry plasma with and
without an inner-gas flow taken at a 10.0-mm skimming
distance
Figure
4.6. Argon signal-pressure curves for a wet plasma taken at
different condenser temperatures and a 10.0-mm skimming
distance
List of Figures
xv
Figure
4.7. Argon signal-pressure curves for a wet plasma taken at
different condenser temperatures and a 6.0-mm skimming
distance
164
Figure
4.8. Manganese signal-pressure curves for a wet plasma taken at
different condenser temperatures and a 10.0-mm skimming
distance
167
Figure
4.9. Light and heavy elemental signal-pressure curves with the
ion-optic lenses removed
169
Figure
4.10. Lithium signal-pressure curves with no optics and at different
solvent loads
172
Figure
4.11. Optimal first-stage pressure for different mass ions at several
condenser temperatures
174
Figure
4.12. Maximum signal for elements of different mass
177
Figure
4.13. Effect of grounded ion-optic lenses on signal-pressure curves
for lithium and uranium
179
Figure
4.14. Effect of third-stage grounded optic lenses on signal-pressure
curves for lithium and uranium
182
Figure
4.15. Effect of inner-gas flow rate on first-stage backing pressure
185
Figure
4.16. Manganese signal versus inner-gas flow rate at several
pressures taken at a 10.0- and 6-mm skimming distance
188
Figure
4.17. Effect of first-stage pressure on oxide signals for cerium and
uranium
191
Figure
4.18. Effect of first-stage pressure on doubly charged ion signals
for barium
194
C ha pter 5
Figure
5.1.
Effect of high solution concentrations on third-stage retardingpotential curves measured for helium
207
Figure
5.2.
Calculated helium and argon gas flow rates through different
size sampling-plate orifices
213
Figure
5.3.
Required first-stage pressure versus skimming distance needed
to optimize skimming for a helium bath gas
216
Figure
5.4.
Experimental and calculated retarding-potential curves for the
helium MIP-MS
223
List of Figures
xvi
Figure
5.5. Retarding-potential curves for helium, manganese, barium,
and uranium at a first-stage pressure of 0.80 Tore
225
Figure
5.6. Retarding-potential curves for helium, manganese, barium,
and uranium at a first-stage pressure of 1.32 Ton-
227
Figure
5.7. Calculated helium and argon velocities 10.0 mm downstream
from a 0.4-mm sampling-plate orifice
231
Figure
5.8. Retarding-potential curves for helium, manganese, barium,
and uranium at a first-stage pressure of 0.66 Ton-
234
Figure
5.9. Background taken at the analyte sampling position
237
Figure
5.10. Background spectral scans for a dry and wet helium plasma
sampled on axis
240
Figure
5.11. Helium retarding-potential curves for a dry and wet plasma
sampled on axis
242
Figure
5.12. Background spectral scans for a dry and wet helium plasma
sampled on axis with 22 and 60 L/min of air added
245
Figure
5.13. Helium retarding-potential curves for a dry and wet plasma
sampled on axis with 22 and 60 L/min of air added
247
C hapter 6
Fi
- an re
6. 1.
Single-stage ICP—MS instrument
256
Figure
6. 2.
Required pumping speed in a single-stage ICP-MS instrument
259
Figure
6.3. Cost of turbomolecular pump versus pumping speed
262
Figure
6.4. Two-stage ICP-MS instrument
268
LIST OF T A B LE S
C h apter 2
Table
2.1. Operating conditions for the ICP-MS
Table
2.2. Calculated Geometrical Dimensions of a Pure Argon Expansion
C h a pter 3
Table
3.1. Operating conditions for the asymmetrically grounded and centertapped ICP-MS.
C h a pter 4
Tj>Wp»
V
A' . a1 . Orv»rj»tin<r p n n rtitro n « *w.
f o r thf» TPP—M S
C h a pter 5
Table
5.1. Operating conditions for the helium MIP-MS
Table
5.2. Ratio of Background Ion Signals Collected at Different First
Stage Pressures
1
1
INTRODUCTION
1.1. I n t r o d u c t i o n
Since the early 1960s mass spectrometric analysis of atmospheric-pressure gas
discharges and flames has been popular in the physics and physical chemistry fields.
Plasma source-mass spectrometers (PS-MS) used in these fields have been subjected to a
wide range of applications from the production of high-density ion beams for fusion
research [1] to fundamental experiments in plasma physics [2-4]. The potential usefulness
of this technique for analytical chemistry was first realized by Gray [5] who showed that
PS-MS could be used for trace elemental analysis of aqueous solutions. These studies
prompted the development of PS-MS techniques for analytical chemistry and led to the
production of a commercial inductively coupled plasma-mass spectrometer (ICP-MS) nine
years later. The developmental history of this technique has been the subject of several
reviews and will not be covered here [6-9].
Over the years the ICP-MS instruments have experienced a swift pace of development
and have achieved almost immediate and overwhelming acceptance into the analytical
chemistry community. Attributes of this technique that make it an attractive alternative to
other methods of elemental analysis include high sensitivity (ng/L for most elements), few
Chapter 1
2
spectral interferences, elemental detectability that nearly spans the periodic table, and
isotopic analysis capabilities [10]. A comparison of ICP-MS with other atomic
spectroscopy techniques—inductively coupled plasma-optical emission spectroscopy
(ICP-OES) and atomic fluorescence spectroscopy (ICP-AFS), and atomic absorption
spectroscopy (AAS)—has revealed that ICP-MS is able to determine more elements at
lower concentrations than the other techniques [11].
There exist several manufacturers of PS-MS instruments for elemental analysis. The
instruments produced today are equipped with either an ICP or a glow discharge (GD) or
both (i.e., Turner Scientific), which serve to vaporize, atomize, and ionize a sample.
These sources are among the most powerful for elemental analysis and handle a wide range
of sample types (the ICP is best suited for solutions, whereas the GD is better suited for
solid samples). The atmospheric-pressure ionization sources such as the ICP typically are
coupled through a three-stage differentially pumped interface to a quadrupole mass analyzer
[10]. Those instruments that employ the low-pressure GD (1 Torr region) as an ionization
source commonly are coupled through a two-stage differentially pumped interface to either
a quadrupole or a magnetic-sector mass analyzer [12].
Other types of ionization sources have been employed with various levels of success.
The argon microwave-induced plasma (MIP) was among the first ionization sources to be
coupled to a mass analyzer for elemental analysis [13]. This class of plasma-sustaining
devices is of particular interest because it can readily support plasmas in many different
bath gases (e.g., helium [14], nitrogen [15], xenon [16], krypton [16]), which have
unique and useful properties as discharge gases. The primary focus of the work presented
in this treatise will be on the PS-MS techniques that use the atmospheric-pressure plasmas,
specifically the argon ICP and helium MIP as ionization sources.
Chapter 1
3
Despite the long list of positive attributes of ICP-MS and GD-MS, analysis of realworld samples often is limited by deficiencies such as matrix-induced interferences, massbias effects, the lack of long-term stability, and interferences caused by the formation of
argide, oxide, and doubly charged species [9,10,17]. Although there have been many
empirical studies aimed at minimizing these effects [18,19,20], little has been done to
identify the physical processes that are responsible for them. These shortcomings either
might arise in the discharge or might be the result of sampling the plasma into the lowpressure environment of a mass spectrometer vacuum chamber. Because these deficiencies
are a characteristic more associated with ICP-MS than with ICP-OES, they have been
perceived to be an artifact of the ion-transport mechanisms that exist in the mass
spectrometer interface. Unfortunately, these precepts have not been substantiated.
In the physics and physical chemistry fields researchers have documented the
complications encountered when charged species from a plasma or a partially ionized gas
interact with the walls of a sampling probe or plate. Specifically, these interactions can
influence population and mobility of plasma constituents. To understand and control such
interactions, information learned from other related areas—gas dynamics [21], conductivemetal-plasma interaction studies [22], ion-optic-lens theory [23], and vacuum technology
[24]—has been integrated into this field. The combination of these different areas has led
the way for the development of techniques and theories that describe the sampling of
plasma sources to create an intense ion beam representative of the discharge.
A main consideration in the development of an atmospheric pressure PS-MS
instrument is the production of an intense ion beam at a reduced pressure (nominally 10'^
Torr) that is representative of the plasma. One of the best sampling methods available that
forms an ion beam with these requirements is the skimmed fiee-jet expansion technique
[21]. The technology that has been used to develop this kind of sampling process for
Chapter 1
4
current PS-MS instruments is firmly based upon the principles of gas dynamic theory [25—
27]. Unfortunately, most of the existing theory describes the sampling of pure neutral
gases and does not consider ion-ion interactions that may occur in the expansion. For this
theory to describe ion movement and the sampling of plasmas, it must be assumed that ion
trajectories can be approximated by the movement of neutral species [28]. In the
continuum-flow region of the first-stage expansion (as the plasma is initially sampled into
the mass spectrometer), collisional frequency is sufficient that this assumption is valid [29].
However, as ions move to the transition- and molecular-flow regions of the expansion and
into the lower pressure stages of the interface, ion-neutral collisional frequency decreases
and permits electrostatic forces to influence ion movement These electrostatic fields can
emanate from either ion-optic lenses or from the ion beam itself in the form of ion space
charge [30].
The current theoiy that is used to describe ion movement in the mass spectrometer
interface only takes into account electrostatic fields from ion-optic lenses and not ion
interaction with other charged species [31]. The effect of these ion-optic-induced
electrostatic fields on ion movement can be described with ion-optic-lens theory [23]. In
the analytical chemistry field both gas dynamic and ion-optic-lens theory have been
combined to model ion movement through the mass spectrometer interface [31]. This
model entails the use of gas dynamic theory to describe ion movement through the mass
spectrometer first stage. As an ion enters into the second or third stage of the interface
where ion-optic lenses are present, electrostatic lens theory is then used to describe the
focusing of ions through to the mass analyzer. Unfortunately, this model fails to describe
accurately ion movement through the mass spectrometer interface [31]. What is even more
unfortunate is that this model commonly is used to guide instrumentation design and to help
interpret experimental results that characterize the instrument. As a result, researchers have
Chapter 1
5
been forced to take an empirical or semi-empirical approach to achieve innovations in
instrument design and operation.
For this reason, a comprehensive fundamental study of the ion-transport process is
needed to advance this field further and make way for innovative improvements. Studies
presented in Chapters 2 through S are aimed at understanding the fundamental properties of
the sampling process and identifying those mechanisms that are responsible for the
shortcomings of this technique so they may be controlled The approach used to achieve
this goal involved monitoring the composition and physical properties of the bath gas and
analyte ions in different regions of the mass spectrometer interface. This particular
approach has led to the development of an ion-transport model that encompasses both the
argon ICP and the helium microwave-induced plasma (MEP) mass spectrometers. To
deduce this model, information from plasma physics, gas-dynamic theory, conductivemetal-plasma interaction theory, electrostatic-lens theory, and vacuum technology were
used to guide and interpret results from experimental measurements. Ion-transport
processes were deduced from three types of measurements performed in the argon ICPMS; Langmuir probe (Chapter 2), ion kinetic energy (Chapter 3), and ion-beam
composition (Chapter 4) measurements. In Chapter 5 ion kinetic energy measurements and
ion-beam composition are studied to determine the feasibility of helium PS-MS.
1.1.1. Langmuir Probe Analyses
In plasma-diagnostic endeavors, physicists were concerned not only with gas-dynamic
interferences that occurred when sampling a discharge, but also with interferences from
electrostatic forces. They learned that as the plasma diffuses through a small differentialpumping-plate aperture, charge separation occurs because electrons have a higher
collisional frequency than ions and strike the plate walls surrounding the orifice at a greater
Chapter 1
6
rate [32]. The separation of charged species produces an partially ionized gas that has a net
positive space potential that will affect ion movement [20,33,34]. A similar phenomenon
should be expected to occur in analytical PS-MS instruments. Surprisingly, in the
analytical chemistry community charge separation has been stated to be negligible [35].
This misconception has prevented an accurate account of the ion-transport processes that
occur as the plasma is sampled into the mass spectrometer first stage.
A Langmuir probe examination of the electrostatic characteristics in the mass
spectrometer interface was used to verify the existence of and provide information about the
degree of charge separation. From such studies, localized space potentials were
determined This space potential was found to exert an influence on ion movement in every
stage of the mass spectrometer interface. In these experiments, it was shown that ion space
potential, more specifically ion space charge, in the interface could be altered by adjusting
the first-stage backing pressure, the inner-gas flow rate, or the solvent load. This
parameter dependence has revealed methods to manipulate ion-beam density
advantageously for the highest ion throughput and the suppression of matrix-induced
interferences and mass-bias effects (see Chapters 2 and 3).
Finally, spatial mapping of the electrostatic characteristics throughout the ffee-jet
expansion in the first-stage was performed to determine the location of different expansion
zones—the isentropic expansion core and the shockwave structure (see Chapter 2).
Because electrostatic characteristics were found to vary among these different regions, it
was then possible to compare the dimensions of the ffee-jet expansion for an expanding
partially ionized gas with that for an expanding neutral gas [36]. A close correlation
between the dimensions of these two expansions was found. This similarity indicates that
the expansion of the partially ionized gas maintains the geometry of a pure-gas expansion;
Chapter 1
7
that is, neutral flow in the partially ionized-gas expansion can be described by gas dynamic
theory.
1.1.2. Retarding-Plate Analyses
The most straightforward method to measure ion kinetic energies in the mass
spectrometer interface is retanding-plate analysis. This method was borrowed from similar
experiments performed by physicists interested in deducing energy distributions of both
ions and electrons in discharges [37-39]. Because the physical characteristics of argonICP and helium-MIP discharges are already well documented, those forces that influence
ion movement (i.e., gas-kinetic and electrostatic) can be determined by monitoring ion
kinetic energies. To perform these experiments in a conventional PS-MS such as the one
used in this work, the instrument was converted to a retarding-plate energy analyzer [40].
In this new instrument, disk-shaped retarding plates, which were placed perpendicularly to
the ion beam at selected locations within the interface, took the place of ion-optic lenses.
Ion kinetic energy measurements were performed by applying a voltage across a particular
retarding plate, which had a centered orifice. Those ions with kinetic energies higher than
the applied potential passed through the sampling aperture to the detector, whereas ions
with lower energy were reflected and not detected.
In the ICP-MS field, ion kinetic energies typically have been measured by using the
quadrupole bias potential as a retarding field [41-45]. However, retarding-plate analysis
was chosen in this work because it was found to be a more reliable method in which to
measure ion kinetic energies. First of all, the retarding plate emits an even electrostatic
field, whereas the quadrupole bias potential produces a hyperbolic field that can distort ion
kinetic energy measurements. Also, for the low mass-range quadrupole (0 to 511 amu)
used in these studies, ramping the quadrupole bias potential can distort ion focusing rf and
Chapter 1
8
dc fields and can potentially make trajectories of heavy ions less stable than lighter species.
Finally, retarding plates can be positioned in different regions of the mass spectrometer so
ion energies can be spatially followed.
Retarding-plate analysis was used to monitor kinetic energies of bath-gas and analyte
ions in the second and third stages of the argon ICP-MS and in the second stage of the
helium MIP-MS. In these studies, forces that influence ion movement (i.e., gas-kinetic or
electrostatic) and their relative strengths were deduced from retarding-potential curves.
Coulombic interactions among ions have been found to be so strong in the mass
spectrometer second stage that they can overpower gas-kinetic forces. In contrast, ion
movement in the third stage was found to be influenced less by these coulombic
interactions. However, electrostatic interactions could be intentionally exaggerated in the
third stage by increasing ion flux. Results from these experiments were useful not only for
the development of a new qualitative ion-transport model for a three-stage ICP-MS
instrument, but also have uncovered ion-transport mechanisms that foster matrix-induced
interferences and mass-bias effects.
1.1.3. Beam-Composition and Density Analyses
The sampling method currently used in commercial PS-MS instruments is designed on
the basis of the skimmed free-jet expansion model developed as a jet spectroscopy
technique [21,25-27]. The pertinent relationships developed from this theory have been
used to determine design parameters such as sampling-plate and skimmer-cone orifice
diameters, skimming position, and first-stage backing pressure in current PS-MS
instruments [28]. However, the sources used in these two techniques differ. In jet
spectroscopy, the source is essentially a pure neutral gas, whereas in PS-MS the source is
a plasma to which a considerable amount of sample material and solvent have been added.
Chapter 1
9
The emphasis of the two techniques differs as well. In jet spectroscopy, the objective is to
cool the bath gas and seed species through collisions with other expanding species, and
thereby reduce their velocity distribution; in contrast, in PS-MS, one desires a large ion
flux directed toward the quadrupole.
Optimal skimming in jet spectroscopy is achieved under those conditions where gas
density and collisional frequency are the lowest and on-axis ion flux is the greatest [26]. In
the present study the mathematical relationships taken from gas dynamic theory used to
describe the best skimming conditions were compared with experimentally determined
conditions. In argon ICP-MS the optimal skimming conditions were found to be quite
different from what gas dynamic theory would predict. Instead it was found that
coulombic forces among ions in the lower pressure zones of the interface could completely
alter ion trajectories predetermined by gas-kinetic processes in the higher pressure zones.
1.1.4. Helium Plasma Source-M ass Spectrometry
Although the most widely used plasma source in PS-MS is the argon ICP, there has
been growing interest toward the development of a helium PS-MS instrument [14,46-48].
This interest exists because there are several potential advantages of using helium as a
plasma gas. The primary advantage is that helium has a higher first-ionization potential
(24.6 eV) than argon (15.8 eV). As a result, a helium plasma should efficiently ionize
species with high first ionization potentials (F, Cl, Br, I, Ru, P, Hg, S, As, Se, Be, Au,
Pt, Cd, Zn, and Os) [48]. In addition, a helium bath gas is expected to exhibit a lesscomplex mass spectral background and should simplify the detection of elements (K, Ca,
V, Fe, As, and Se) whose mass-spectral peaks overlap with those of argon and argoncontaining polyatomic ions (Ar+, ArC+, ArN+, ArO+, ArOH+, ArCl+, Ar+2) found in the
argon ICP [48].
Chapter 1
10
Unfortunately, a helium PS-MS instrument that rivals the conventional argon ICP-MS
system has not yet been developed. On the contrary, progress in its development has been
slow. One of the most recent of these systems has been characterized as being
unpredictable and behaving differently from the argon ICP-MS [48]. This unexplained
behavior likely arises from differenfces between helium and argon gas kinetics and plasma
characteristics. The few studies that have been performed to characterize the sampling
process in PS-MS instruments for analytical chemistry pertain only to the argon ICP-MS
[28,44,49-52]. Despite the numerous works concerning helium PS-MS [46-48], no
progress has been made toward understanding the fundamental aspects of sampling a
helium plasma.
In Chapter 5 the sampling process of a helium PS-MS instrument is examined to
determine if the concept of helium PS-MS is fundamentally sound. The helium ionization
source used in these studies was a Beenakker-type MIP [53]. This type of MIP is an
attractive plasma source because it readily supports a helium plasma, generates little heat,
has a simple design, and is inexpensive to construct and operate [54]. The fundamental
aspect of this particular helium PS-MS instrument have been investigated and are compared
to those of the argon PS-MS system. The results from these experiments not only provide
fundamental information about the helium-plasma sampling process, but also guide future
development of this technique. It has been determined that if helium PS-MS is to be a
viable alternative to argon ICP-MS, two obstacles need to be overcome: The first is to
eliminate air entrainment and second is to reduce mass-bias effects.
11
Chapter 1
1.2. L it e r a t u r e C i t e d
[1]
Y. Okamoto and H. Tamagawa, Rev. Sci. Instr. 43, 1193 (1972).
[2]
D. L. Hamm, Sol. State Technol. 4, 109 (1979).
[3]
H. W. Drawin, Plasma Diagnostics, Ed. W. Lochte-Hogreven, North-Holland,
Amsterdam, (1968) pp. 111.
[4]
A. N. Hayhurst and N. R. Telford, Proc. Roy. Soc. Lond. A. 332, 483 (1971).
[5]
A. L. Gray, Proc. Soc. Anal. Chem. 11, 182 (1974).
[6]
A. L. Gray, European Spectrosc. News 43,13 (1982).
[7]
A. L. Gray, Spectrochim. Acta 40B, 1525 (1985).
[8]
A. L. Gray, J. Anal. Atom Spectrom. 1,103 (1986).
[9]
D. J. Douglas, Can. J. Spectrosc. (1988), in press.
[10] G. M. Hieftje and G. H. Vickers, Anal. Chim. Acta 216, 1 (1989).
[11] B. Sansoni, W Brunner, G. Wolff, H. Ruppert, and R. Dittrich, Fresenius' Z.
Anal. Chem. 331, 154 (1988).
[12] D. J. Hall and P. K. Robinson, Am. Lab. 19, 74 (1987).
[13] D. J. Douglas and J. B. French, Anal. Chem. 53, 37 (1981).
[ 141 J. T. Creed, T. M. Davidson, W.-L. Shen, P. G. Brown and J. A. Caruso
Spectrochim. Acta 44B, 909 (1989).
[15] D. A. Wilson, G. H. Vickers, and G. M. Hieftje, Anal. Chem. 59, 1664 (1987).
[16] E. Poussel, J. M. Mermet, D. Deruaz, and Claude Beaugrand Anal. Chem. 60,923
(1988).
[17] R. S. Houk and J. J. Thompson, Mass Spectrometry Reviews 7,425 (1988).
[18] D. Beauchemin, J. W. McLaren, and S. S. Berman, Spectrochim. Acta, Part B,
42B, 895 (1987).
[19] D. C. Gregoire, Spectrochim. Acta, Part B 42B, 895 (1987).
[20] S. H. Tan and G. Horlick, J. Anal. At. Spectrom. 2, 745 (1987).
[21] J. M. Hayes, Chem. Rev. 87, 745 (1987).
Chapter 1
12
[22] R. N. Franklin, Plasma Phenomenon of Gas Discharges, Clarendon Press, Oxford
(1976).
[23] O. Klemperer and M. E. Barnett, Electron Optics, 3rd ed. Cambridge Univ. Press
(1971).
[24] J. B. Hasted and D. Phil, Physics of Atomic Collisions, American Elsevier,
NY(1972), p. 9.
[25] H. Ashkenas and F. S. Sherman, in Rarified Gas Dynamics, 4*h Symposium, Ed.
J. H. de Leeuw, Academic, New York (1966), p. 84.
[26] R. Campargue, Rarified Gas Dynamics, 6th Symposium, Ed. L Trilling and
H. Y. Wachnam, Academic Press, New York (1969), p. 1003.
[27] A. Hoglund and L.-G. Rosengren, Int. J. Mass Spectrom. Ion Processes
60, 173 (1984).
[28] J. E. Fulford and D. J. Douglas, Appl. Spectrosc. 40, 971 (1986).
[29] G. N. Spokes and B. E. Evans, in Tenth Symposium (International) on
Combustion, The Combustion Institute, Pittsburgh, Pa., 1965), p. 639.
[30] R. G. Wilson and G. R. Brewer, Ion Beams with Applications to Ion Implantation,
Wiley, NY (1973).
[31] M. A. Vaughan and G. Horlick, submitted to Spectrochim. Acta B, 1990.
[32] D. M. Manos and H. D. Dylla, Plasma Etching: An Introduction, Eds. D. M. Manos
and D. L. Flamm, Academic, New York (1989), p. 263.
[33] F. F. Chen, Plasma Diagnostic Techniques. Eds. R. H. Huddlestone and S. L.
Leonard, Academic, NY (1965), pp.177-183.
[34] M. A. Uman, Introduction to Plasma Physics, McGraw-Hill, NY (1964).
[35] D. J. Douglas and J. B. French, Spectrochim. Acta 41B, 197 (1986).
[36] K. Bier and B. Schmidt, Z. Angew. Phys. 11, 34 (1961).
[37] D. D. Neiswender and F. C. Kohout, Rev. Sci. Instrum. 43, 1475 (1972).
[38] K. Hiraoka and H. Kamada, Japanese J. Appl. Phys. 10, 339 (1971).
[39] J. L. Franklin, S. A. Studniarz, and P. K. Ghosh, J. Appl. Phys. 52, 3633 (1981).
[40] J. A. Simpson, Rev. Sci. Instrum. 32, 1283 (1961).
[41] J. A. Olivares and R. S. Houk, Appl. Spectrosc. 39, 1070 (1985).
Chapter 1
13
[42] J. E. Fulford and D. J. Douglas, Appl. Spectrosc. 40, 971 (1986).
[43] R. C. Hutton and A. N. Eaton, J. Anal. At. Spectrom. 2 , 595 (1987).
[44] J. S. Crain, R. S. Houk, and F. G. Smith Spectrochim. Acta 43B, 1355 (1988).
[45] H. B. Lim, R. S. Houk, and J. S. Crain, Spectrochim. Acta 44B, 989 (1989).
[46] R. D. Satzger, F. L. Fricke, P. G. Brown, and J. A. Caruso, Spectrochim. Acta
42B, 705 (1987).
[47] A. Montaser, S.-K. Chen, and D. W. Koppenaal, Anal. Chem. 59,1240 (1987).
[48] D. W. Koppenaal and L. F. Quinton, J. Anal. At. Spectrom. 3, 667 (1988).
[49] A. L. Gray, R. S. Houk, and J. G. Williams, J. Anal. At. Spectrom. 2 , 13 (1987).
[50] R. S. Houk, J. K. Schoer, and J. S. Crain, J. Anal. At. Spectrom. 2, 283 (1987).
[51] S. M. Tan and G. Horlick, J. Anal. At. Spectrom. 2, 745 (1987).
[52] G. R. Gillson, D. J. Douglas, J. E. Fulford, K. W. Halligan, and S. D. Tanner,
Anal. Chem. 60, 1472 (1988).
[53] K. G. Michlewicz and J. W. Carnahan, Anal. Chem. 58, 3122 (1986).
[54] A.T. Zander and G. M. Hieftje, Appl. Spectrosc. 39, 214 (1985).
i
14
2
L a n g m u ir Probe M ea su re m en ts
2.1. I n t r o d u c t i o n
Recent attention in inductively coupled plasma-mass spectrometry (ICP-MS) has been
directed toward understanding the mechanisms that influence such deficiencies as matrixinduced interferences, mass-bias effects, and the lack of long-term stability [1,2]. Because
ICP-MS has been found to be more susceptible to these sorts of difficulties than
inductively coupled plasma-optical emission spectrometry, they are perceived to be an
artifact of the ion-transport mechanisms that exist in the interface [3,4]. Such ion-transport
processes have already been examined in the physics field where plasma source-mass
spectrometry (PS-MS) has been used for plasma-flux analysis. In these studies, both gaskinetic and coulombic forces are taken into account in the ion-transport model [5], and it
has been shown that ion movement in low-pressure zones can be complicated by coulombic
interactions, specifically by ion space charge [6].
In analytical chemistry, the effect of gas-kinetic forces on ion movement has been the
topic of several fundamental papers and is a well understood aspect of the sampling process
in PS-MS [7-9]. Unfortunately, this is not the case for the effect of coulombic forces
among ions, where only a few studies exist [3-4]. The purpose of the present study is to
Chapter 2
15
investigate the fundamental processes, both gas-kinetic and electrostatic, that influence ion
movement and ion-ion interactions in an ICP-MS instrument. The method selected here to
monitor these processes involves Langmuir probe measurements on the partially ionized
gas that is sampled into the mass analyzer first stage. This type of analysis permits one to
examine the electrostatic features of the gas after it has passed through the first differentialpumping aperture of the interface. At this orifice the plasma constituents have their first
encounter with the walls of the differential-pumping aperture, and determination of the
space and floating potentials of the transmitted stream provides an indication of the extent
and influence of this interaction. In particular, space and floating potentials directly reflect
the ion-electron balance, the magnitude of any local space charge, and the degree of ionsheath formation. Knowledge of these parameters can be used for predicting ion
movement For example, an ion accelerating away from a positive space-charge field will
gain energy proportional to the potential of that field. Similarly, the floating potential is
indicative of the voltage drop across an ion sheath that can form over the aperture of the
differential pumping plate [6]. As will be shown in Chapters 3 and 4, the electrostatic
features of this expanding partially ionized gas become increasingly important in the lowerpressure zones of the interface. In fact, knowledge of how source and interface parameters
affect space and floating potentials has revealed methods to manipulate these potentials
advantageously.
In this chapter the effect of several source and interface conditions, including solvent
load, first-stage backing pressure, and inner-gas flow rate were examined for their effect on
space and floating potentials. In addition, floating potentials were spatially mapped
throughout the expansion zone to characterize the distribution of the expanding ion plume.
16
Chapter 2
2.2. E x p e r i m e n t a l
2.2.1. Modification of ICP-MS for Langmuir Probe Measurements
The ICP-MS system used in this study is similar to the one described by Wilson etal.
[10]. However, several modifications were made to perform the Langmuir probe
measurements. First, the skimmer cone and second-stage ion lenses were removed to
provide room for the Langmuir probe, and the skimmer cone was replaced with a flat
skimming plate having a 1.0-mm-diameter orifice. Next, a number of flanged vacuum
ports were opened to the mass spectrometer first stage that permitted the attachment of a
pressure transducer and the Langmuir probe assembly. Finally, a desolvation system was
incorporated to control the amount of aerosol and water vapor introduced into the plasma.
A diagram of the overall instrumental setup is shown in Figure 2.1. A more detailed
depiction of the first-stage pumping and Langmuir probe configurations is given in Figure
2 .2 .
An ICP with an inverted, asymmetrically grounded load coil was used, which is
standard for this ICP-MS instrument [10]. Operating conditions are listed in Table 2.1 and
are standard for this system [11].
2.2.2. Spatially Adjustable Langmuir Probe Assembly
The Langmuir probe was supported by a specially designed gimbal-type manipulator
and mounted into the first-stage vacuum region through a welded bellows (Huntington,
model 247-08) with a 1.0" Kwik-Flange® attached to each side (see Figure 2.2a). One
side of the bellows was attached to the interface and the other to a high-vacuum voltage
feedthrough (MDC, K100-BNC). A tungsten Langmuir probe was attached to the
feedthrough by a brass-rod connector. The probe was bent 90° so that the probe tip was
perpendicular to the flowing partially ionized gas. Dielectric materials shielded portions of
17
Chapter 2
Figure 2.1. Instrumental diagram of the ICP-MS configured for Langmuir probe
measurements.
Chapter 2
\ySU
Laboratory
PC
v__________OQ
.________J
Function
Generator
Power
Supply
Voltage Buffer
Circuit
r~
Impedance
Matcher
Langmuir Probe
Nebulizer
Ar
Peristaltic Pump
To Rotary-Vane
Pump (2000Umin)
Desolvator
Power Supply
00
19
Chapter 2
Figure 2.2. First-stage (a) Gimbal mount for three-dimensional positioning of the
Langmuir probe and (b) vacuum arrangement. To assemble this apparatus the welded
bellows was placed in the Gimbal-type manipulator and the brass probe assembly was
attached to the vacuum feedthrough. The complete probe assembly was then attached to the
mass spectrometer interface.
o
Springs
/A
m
\ikvvv
kVVVVlW M
El
Micrometers
Gimbal-Type Assembly
I
Interface
Welded Bellows
8248
nn
Tungsten Probe with
Quartz Shielding
I
"K
r
c=
Lr
L
Feedthrough
J
Brass Probe Assembly
Ni
O
Chapter 2
To
Capacitance
Manometer
Welded Bellows
To Rotary-Vane Pump
Skimming
Plate
jSrJ
Sampling /
To Rotary-Vane Pump
,
Butterfly Valve
CP
9
j—
Plate
To Rotary-Vane Pump
22
Chapter 2
Table 2.1. Operating Conditions for the ICP System
Forward Power
1250 W
Reflected Power
<10 W
Argon Gas Flow (L/min)
♦Inner
Intermediate
Outer
I.04
1.00
II.0
Sample Uptake (mL/min)
0.64
Sampling Depth (mm)
Above the Lend Coil
I U iU
in n
♦This parameter was altered for particular experiments. Any different values are cited in
text.
Chapter 2
23
the probe that were not to be exposed to the partially ionized gas. Quartz was used to
shield the tungsten portion and Teflon tape was used to shield the brass portion of the
probe.
2.2.3. Langmuir Probe Circuit
The voltage-buffer-amplifier circuits that were used to supply voltage to the probe and
to monitor the resulting current are shown in Figure 2.3. The first circuit offered a range of
applied voltages between ±19 V (Figure 2.3a) and the second circuit a range of 0 to +35 V
(Figure 2.3b). The voltage available at the Probe Current Monitor output was proportional
to the current flow to the probe. A dual power supply (Trygon Electronics, model DL40-1)
served to power the circuit while the input voltage ramp (0.175 V/s) was supplied by a
function generator (HP, model 3325A). Because the output range of the function generator
was limited to 10 V, the voltage to the probe was amplified appropriately to reach the
electron-saturation current The final voltage to the probe was ramped at 0.7 V/s.
2.2.4. First-Stage Vacuum System
The first-stage pumping configuration (see Figure 2.2b) was designed so a wide range
of backing pressures could be obtained. The previous interface layout [10] had only a
single 25-mm-diameter vacuum port; as a result, gas flow from the interface to the firststage pump was conductance-limited. For the present studies, an additional 25-mmdiameter vacuum port was opened to reduce the flow resistance to the rotary-vane pump.
Each of the two vacuum ports was connected by a 2-m length of 25-mm-diameter vacuum
tubing to a 40-mm-diameter cross piece (Balzers, Model BP 217 016-R). Placed between
the cross piece and the rotary-vane pump (Balzers, DUO 120) was a butterfly valve
(Huntington, BF 150-SF), which permitted the gas flow to the pump to be throttled so the
24
Chapter 2
Figure 2.3. Diagram of the voltage-buffer-amplifier circuit used to supply voltage to the
Langmuir probe and to monitor current from it. The amplifier circuit was added before the
output to extend the range of the probe voltage. Circuit (a) permitted probe voltages from ±
19 V to be applied, whereas circuit (b) provided voltages from 0 to + 35 V.
Chapter 2
R14
R17
25
R15
V+
C6
U3
0P27
V+
PROBE
CURRENT
MONITOR
V+
R18
R2
RAMP
INPUT
0P77
V+
PROBE
OUTPUT
VOLTAGE
U2
R19
VR3
V-
R13
C2
R7
C3
R20
R8
X2
PROBE
OUTPUT
VOIJAGE
V+ »
R11
A/SA-
U2 _
LM317
R5
R13
R19
X2
R20
a _________________________
v+
VC 1.C2.C 6.C 7
C3
R1.R13
R2.R3.R 15.R 16
R4.R6.R8
R5
R7
R9
+19
-1 9
.1 u F
27pF
1M
10
1K
9.1K
10K
750K
R10.R12
R11
R14
R17
R18
R19
R20
U1
U2
U3
1 5 .OK
6 .8
732K
100K
10K
20.0K
1 9 .6K
0P77G
L465
0P27G
b
V+
VC1.C 2.C 6.C 7
C3
01
R1.R13
R 2.R 3.R 15.R 16
R4
R5
R7
R8
+37
-2 .5
•1uF
27pF
1N4004
1M
10
1K
240
20K
19.6K
R9
R10.R12
R11
R14
R17
R18
R19
R20
U1
U2
U3
750K
15 .OK
6 .8
732K
50K
10K
5 1 .1K
16.2K
0P77G
LM317T
0P27G
Chapter 2
26
first-stage backing pressure could be controlled. The rotary-vane pump used in this study
had a greater pumping speed (2000 IVmin air at standard temperature and pressure) than is
commonly used in ICP-MS instruments. Pressure in the first stage was monitored with a
capacitance manometer (MKS Instruments, Model 122A) that was placed as close as
possible (approximately 25 cm) to the centerline axis of the expansion to obtain an accurate
backing-pressure reading.
2.2.5. Sample Introduction System
A desolvator was coupled to the sample introduction system to regulate the amount of
aerosol and solvent vapor introduced into the plasma. This desolvation system, similar to
the one described by Deutsch and Hieftje [12], was attached to the aerosol exit port of a
Scott-type spray chamber. Solution samples were delivered by a peristaltic pump (Gilson,
Model Minipuls 2) to a concentric nebulizer (constructed in house). To control the amount
of water introduced into the plasma, the heater column of the desolvation arrangement was
maintained with a heating tape at a constant temperature of approximately 200° C while the
condenser column temperature was regulated with a refrigerated-circulating bath (Neslab,
Model RTE-5B).
To gauge the amount of water contained in the inner-gas flow of the ICP, the output of
the desolvation system was delivered to silica-gel traps [13]. This calibration was
performed only once to illustrate the effect of condenser temperature on solvent load for
this particular desolvation system. The absolute values recorded for these solvent loads,
plotted in Figure 2.4 as a function of condenser temperature, are expected to vary
somewhat from run to run and with experimental conditions. For this reason, the values in
Figure 2.4 are only estimates of the true solvent load. Later references to solvent loading
are therefore stated in terms of condenser temperature.
27
Chapter 2
Figure 2.4. Solvent (water) load added to inner-gas flow as a function of condenser
column temperature of the desolvation system.
—
i-------------------- '---- 1-------------1-------------1
10
20
30
Condenser Temperature (°C)
Si
00
29
Chapter 2
2.3. B a c k g r o u n d
2.3.1. Current-Potential Measurements
Current-voltage (I-V) curves were obtained by using the single-probe method [14] in
which the probe potential was referenced to the metal wall of the first-stage vacuum
chamber. Space potentials were deduced by finding the point of discontinuity between the
regions of electron retardation and electron acceleration on each I-V curve. Because this
point of discontinuity is not always obvious, it was found by extrapolating the two linear
portions of the semilogrithmic plot of the I-V curves in the electron-accelerating region.
The potential corresponding to the point where the two extrapolated lines meet is
approximately equal to the space potential [14]. An example of this extrapolation method
used to determine space potential is illustrated in Figure 2.5. Shown in this figure is the
electron-accelerating region of the I-V curve for probe potentials from 0 to +35 V. As the
potential was increased in the positive direction the electron current flowed from the
partially ionized gas to the probe. For these measurements the probe length and diameter
were both 1.0 mm. This relatively large diameter might have contributed to the curvature
of the I-V plot because of a spread in work function over the surface of the probe [14]. A
smaller diameter was not employed because the hot gases were found to degrade the probe
surface over an unacceptably short period of time. As a result of this erosion problem, it
was preferable to perform measurements quickly. One method used to reduce analysis time
was to measure only the electron-accelerating region of the I-V curve.
Even though floating-potential measurements could be derived from individual I-V
plots (as the voltage at zero current), they also were taken from an electrically floating
probe. This latter method is preferred because the floating potential can be averaged over a
longer time interval, yet the analysis is quick compared to the time needed to record a full IV curve. The floating potentials that were measured with an electrically isolated probe were
30
Chapter 2
Figure 2.5. Semilogarithmic plot of the electron-accelerating region of a current (I)potential (V) curve obtained from a single Langmuir probe. Space potential is
approximated as the intersection point of the two extrapolated straight lines. Probe
diameter, 1.0 mm; Probe position, on axis and 10 mm downstream from the samplingplate orifice; Condenser temperature 10°C; Inner-gas flow rate, 1.04 L/min.
-
4-
-
6
-
8-
-
Ln (I)
-
10-
-12
0
5
10 15 20 2 5 30 35 4 0
Potential (V)
Chapter 2
Space Potential
-2 i
32
Chapter 2
performed in the same manner as that described by Lim, Houk, and Crain [15]. For these
measurements a smaller probe could be used (0.6 mm) because the experiments could be
carried out before erosion became troublesome.
2.3.2. Charge Separation at the Sampling Plate
As the plasma constituents (ions, electrons, and neutrals) approach the walls of the
sampling-plate orifice, their relative populations will be altered. An imbalance in ions and
electrons can be lost by collisions with the walls of the sampling-plate orifice and will
obviously change the electrostatic characteristics of the sampled gases throughout the
interface. Because elections have a far greater velocity than ions [16] they will strike the
walls of a sampling aperture at a higher frequency than ions and will be lost preferentially
from the plasma. As a result, the gas expanding through such a sampling aperture is likely
electron-poor. This charge-separation process is well documented in the physics field [5,6]
and is anticipated to be quite significant in an analytical ICP-MS system.
Charge separation can be treated in a semiquantitative fashion [6]. The critical factors
that determine the degree of charge separation are the mean free paths and velocities of the
charged species and the outcome of their collisions with the sampling-plate walls. Because
the electron-collisional frequency in a free-flowing plasma is much larger than that of an
ion, preferential electron extraction will occur as the plasma passes through the samplingplate aperture. The collisional frequency of electrons in a fiee-flowing plasma can be
calculated as the mean velocity (v) divided by the mean free path (A,).
The mean free path (A.^) for species in a two-component medium can be obtained from
the following expression [17]
-l
4f l Tin^i
+ 7t n2 ( ri + r2 ) 2 ( v i_l_v 2 ) 2
Vi
(2 . 1)
Chapter 2
33
where n is the number density, r is the cross-sectional particle radius, v is the average
particle velocity, and the subscripts 1 and 2 refer to the two types of particles. To
determine the mean free path of either the ion or electron, subscript 1 in equation (1) refers
to charged species (ions or electrons), whereas subscript 2 specifies neutral species.
Because collisional cross sections among charged particles are the same for any two singly
charged (positive or negative) species, collisions between ions and electrons can be counted
as an electron-electron collision. Thus, the first term in Equation 1 can be applied to
collisions among electrons and ions as well as among electrons; the second term pertains to
collisions between electrons and neutrals. The assumptions made in using Equation 1 are
that the plasma is quasi-neutral and that the velocity of the electrons is much greater than
that of the heavier gas atoms. The electron number density is assumed to range from 1014
to 1015 cm-3 [18], the argon atom collisional cross-section is 3.6 x 10'15 cm2 [19], and the
gas density is 1.3 x 1018 cm*3. The mean free path for ions should be similar to that of the
electrons in view of their similar collisional cross-sections, so the ion number density is
assumed to be equal to that of the electrons. The collisional cross-section between
electrons (Qee). between electrons and ions (Qei), and between ions (Q;;) is calculated to be
10‘12 cm2 by
where the charge number is 1, A (= 5) is the maximum impact parameter, which is
dependent on electron temperature (5500 K [18]) and electron number density (1015 cm*3)
in the plasma [20].
Chapter 2
34
From Equations 1 and 2 the mean free path of electrons or ions in a free-flowing ICP
should range from 1 to 10 pm. As the plasma is brought in contact with the sampling plate a
decrease in the density of ions and electrons from collisional loss will increase the mean free
path beyond these estimated values. Even though electrons and ions have similar mean free
paths, charge separation occurs because of the higher velocity of electrons and their
consequently favored diffusion to the orifice walls. At a temperature of 5500 K, the velocity
of an electron is expected to be approximately 4.6 x 106 m/s. Because Equations 1 and 2
assume thermodynamic equilibrium, the gas temperature can be set equal to the electron
temperature for this calculation. Thus, the ion velocity will be 1700 m/s (in actuality, the gas
temperature is approximately 500 K lower than the electron temperature in a free-flowing ICP
[18]). From these values, the collisional frequency for electrons will range from 4.6 x 108 to
4.6 x 109 collisions s_1, whereas the ion collisional frequency will be only 1.7 x 104 to 1.7 x
105 collisions s*1. This higher collisional frequency for electrons can be confidently expected
to yield charge separation.
Evidence for charge separation in an analytical ICP-MS instrument is found in the
experimental results of Gray, Houk, and Williams [21]. Floating potentials in their
asymmetrically grounded ICP were approximately -50 V. Yet, as the plasma came in
contact with a sampling plate, the floating potentials jumped to +20 V. Similar trends were
reported by Houk, Schoer, and Crain [22], who used a center-tapped load coil. In their
studies, the floating potential increased from -6 V to a range of +0.2 V to -2 V as the
plasma came in contact with the interface. The move toward positive floating potentials as
the plasma interacts with a sampling plate signals that the plasma is electron-deficient and is
no longer quasi-neutral; in short, charge separation has occurred.
As will be shown later, floating potentials in the first stage of this ICP-MS instrument
have been found to be consistently positive. Theoretically, when a Langmuir probe is
Chapter 2
35
placed in a low-pressure, magnetic-field-free region and is allowed to float electrically in a
quasi-neutral plasma, the probe attains a negative potential as a result of the greater mobility
of elections than ions [16]. The positive floating potential measured in the experiments
reported below therefore indicates convincingly that the partially ionized gas in the first
stage is not quasi-neutral but is electron-poor.
Faced with these facts one must remain open to the possibility that once the plasma
expands into the interface, it is no longer a plasma. In turn, the non-plasma state of the
expanding gas will complicate the interpretation of Langmuir probe measurements because
the theory used to interpret those measurements relies on the quasi-neutrality of the plasma
[23,24]. Although others have used this traditional treatment to quantitate electron
temperature and number density [25], the same treatment will not be used here because of
the questionable validity of the necessary assumptions. For the same reason, the sampled
discharge will not be identified below as a “plasma” but rather describe it as an expanding
“partially ionized gas”.
2 .3 .3 . Influence of Shockwave Structure on Electrostatic Features in the
Expansion Core
The expansion into the first stage of the ICP-MS consists of several regions (Figure
2.6) where local pressure, density, and flow velocity differ. In the continuum-flow region
of the expansion, where the gas first enters the vacuum chamber, the collisional frequency
between gas species is on the order of that found in the original source. As the gas
expands, the collisional frequency decreases to a point were it becomes negligible in a zone
termed the molecular-flow region of the expansion. Between the continuum- and
molecular-flow zones is the transition-flow regime. When a gas expands into a chamber at
low pressure, such as in the first stage of an ICP-MS, these flow zones are confined by the
background gases. As a result, there is a region of discontinuity in pressure, density, and
36
Chapter 2
Figure 2.6. Diagram of a free-jet expansion;
is the distance from sampling-plate
orifice to the Mach disk, Xmax is the distance from the sampling-plate orifice to the
"optimal" sampling location, YB is the maximum diameter of the isentiopic core,
diameter of the Mach disk, and YR is the outer diameter of the barrel shock [26].
is the
Dimensions for these parameters are given in Table 2.2 as a function of first-stage backing
pressure.
37
Chapter 2
Mach Disk
Molecular Flow
Barrel Shock
Sampling Plate
►
Transition Flow
Continuum Flow
max
Chapter 2
38
flow velocity between the background gas and the expanding species; this discontinuity is
referred to as the shockwave structure. The two regions of this structure include the Mach
disk (that region perpendicular to the expansion flow) and the barrel shock (the region
parallel to the expansion flow). Electrostatic potentials in these distinct regions can vary
because of the different pressures, densities, and flow velocities and might make it possible
to map the dimensions of this expansion.
The expected dimensions of this classical expansions can be derived from
experimentally determined values given by Bier and Schmidt [26]. The calculated values
compiled in Table 2.2 correspond to an argon expansion in which the source pressure is
760 Torr and the first-stage backing pressure is alternatively 1.1 or 3.0 Torr. When the
first-stage backing pressure is 1.1 Torr the maximum diameter of the isentropic core (Y3) is
expected to be 13.0 mm. Because there are no experimental Yg values available for an
argon expansion having the traits given above, this dimension was approximated by
subtracting 20% fiom the diameter of a nitrogen expansion in accordance with the results of
Bier and Schmidt [26]. Similarly, the Mach disk diameter (YM) is expected to be 10.0 mm
and the walls of the barrel shock ((Yr - Yg)/2) will be 1.0-mm thick. These values were
extrapolated from the ratios of Yg/YM (= 1.26) and Yr /Ym (= 1-49), also determined
experimentally by Bier and Schmidt [26]. Technically, these ratios pertain to conditions
only up to a pressure ratio (Pq/Pi) of 200, where P0 is the source pressure and Pj is the
first-stage backing pressure. For pressure ratios up to 700 (the maximum ratio found in
our system) the dimensions of the expansion are expected to change linearly with a
decrease in first-stage pressure. In making this assumption, one does not anticipate any
significant changes in viscosity or in the continuum nature of the expanding gas [27].
39
Chapter 2
Table 2.2. Calculated Geometrical Dimensions of a Pure Argon Expansion with a
Reservoir Pressure of 760 Torr.
First-Stage Backing Pressure (Torr)
Dimensions (see Fig. 2.6)
1.1
3.0
Yb (mm)
13.0
5.0
Ym (mm)
10.0
6.5
Yr (mm)
15.0
7.5
Xmax (mm)
10.6
7.1
Xm (mm)
17.6
10.7
40
Chapter 2
The calculation for the Mach disk location (XM), however, is valid for pressure ratios
from 15 to 17,000 and is calculated to be 17.6 mm behind the sampling-plate orifice by
[28]
(2.3)
where D0 is the sampling-plate orifice diameter (1.0 mm in the present case).
Upstream from the Mach disk is a position viewed by jet spectroscopists as the
"optimal skimming position", where the gas velocity distribution is the narrowest. This
position is located deep in the isentropic core of the expansion just before the onset of
background-gas penetration and can be calculated from the following expression [29]
12.4)
where Xmax is the distance from the sampling orifice to the optimal skimming position, Xq
is the point from which the gas stream tines appear to be originating, the sltielding
coefficient (Cs) is 0.125, and the Knudsen number (Kn) is 1.3 x 10'3. The Knudsen
number is calculated as the ratio of the bath-gas mean free path to the sampling-plate orifice
diameter (A/D0). The mean free path for argon neutral atoms is calculated to be 1.3 Jim,
which assumes a collisional cross section of 36 A2 [19] and a source temperature of 5000
K [18]. At a first-stage backing pressure of 1.1 Torr this position (Xmax) lies 10.6 mm
downstream from the sampling-plate aperture. When the first-stage backing pressure is
taken to be 3.0 Torr, the dimensions of the isentropic core of the expansion understandably
decrease as can be seen in Table 2.2.
41
Chapter 2
2.4.
RESULTS AND DISCUSSION
2.4.1. Stability and Characteristics of Measured Floating Potentials
Fluctuation in the space potential of the partially ionized gas that passes through a mass
spectrometer interface is one possible cause for instabilities in ICP-MS instruments.
Changes in these electrostatic potentials might not only alter the kinetic energy of an analyte
ion, but could also influence the density of the ion beam through space-charge interactions
(ion-ion repulsion). The stability of the electrostatic characteristics in the mass
spectrometer first stage can be deduced by tracking the floating potential of an electrically
isolated probe. Shown in Figure 2.7 is a trace of the first-stage floating potential as a
function of time, monitored for one hour after the plasma was ignited. As shown in the
figure, the potential jumps initially as the inner-gas flow is turned on; however, whether
the inner-gas flow is on or off the potential stabilizes quickly, in this case within the first 8
seconds after ignition. After this first 8 seconds the potential varied only slightly (from
3.25 to 3.35 V) during the first hour of operation. This slight instability might be caused
by erosion of the 0.6-mm-diameter probe or by a drift in plasma operating parameters.
These results are somewhat surprising because for the first hour of operation of the
ICP-MS as an analytical instrument, ion signals drift by over an order of magnitude. Of
course, the possibility exists that small changes in electrostatic properties in the interface
cause a large variation in the analyte signal. However, it will be shown in the following
section that parameters such as the first-stage backing pressure, solvent load, and inner-gas
flow can be chosen so that changes in these parameters influence space and floating
potentials only slightly. It would be difficult to reconcile the small changes in electrostatic
characteristics that arise from fluctuations in first-stage backing pressure, solvent load, or
inner-gas flow rate with the large signal deviations.
42
Chapter 2
Figure 2.7. Floating potential as a function of time for a dry plasma. The floating
potential drifts whether the inner-gas flow is on or not for approximately the first 8 seconds
but then stabilizes. Probe diameter, 0.6 mm; Probe position, on axis and 10 mm
downstream from the sampling-plate orifice; Inner-gas flow rate, 1.04 L/min.
Chapter 2
Ignition of Plasma
Floating Potential (V)
7V
1-
6V
5V
4V
Inner-Gas Flow On
i
h
3V
70 sec
1 hr
Time
44
Chapter 2
2.4.2. Space and Floating Potentials
Space and floating potentials taken from measured I-V curves are plotted in Figures
2.8-2.10 and demonstrate the effect of first-stage backing pressure, solvent load, and
inner-gas flow rate. These parameters were all found to alter the space and floating
potential to some degree. In particular, an increase in first-stage backing pressure, solvent
load, and inner-gas flow rate all produced a rise in the measured potentials.
2.4.2.I.
Effect of Solvent Load and First-Stage Backing Pressure.
Each
pair of points for the space and floating potential plotted in Figure 2.8 was obtained from
the same I-V curve. Not surprisingly, the floating potential always fell below the space
potential, yet was generally within one volt. According to Langmuir probe theory, the
floating potential should always be more negative than the potential of the surrounding
partially ionized gas because the electron-diffusion rate is greater than the ion-diffusion
rate. Only when the probe is biased negatively will both types of species diffuse toward
the probe at an equal rate [14].
Solvent load had little influence on space or floating potentials until the condenser
temperature reached 15°C. Above this temperature, an increase in solvent load caused a
substantial rise in both kinds of potential. Similarly, as the first-stage backing pressure
was raised at a fixed condenser temperature, the potentials both increased.
These trends became even clearer when floating-potential measurements were taken
from an electrically isolated probe, rather than from the I-V curves (see Figure 2.9). This
improved discrimination probably arises because each plotted point in Figure 2.9 is derived
from a signal averaged over a longer time interval (approximately 10 sec.) than those in
Figure 2.8, which were deduced from an I-V curve. At first-stage pressures below 2.0
Tonr and at condenser temperatures below 15°C, floating potentials are level at a low value.
45
Chapter 2
Figure 2.8. Effect of first-stage backing pressure on space potentials and floating
potentials taken from I-V curves obtained at condenser temperatures of (a) 10, (b) 15, and
(c) 20° C. Probe diameter, 1.0 mm; Probe position, on axis and 10.0 mm downstream
from the sampling-plate orifice; Inner-gas flow rate, 1.04 L/min.
46
Chapter 2
>
(0
4c>
O
a
Space
Floating
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
First-Stage Backing Pressure (Torr)
6
5
>
.2
■
' 5
C
Space
4
o*
+
o
CL
3
Floating
2 Im
' n | " "I 11" I 111 ..............................I
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
First-Stage Backing Pressure (Torr)
6
Space
5
>
cCl
o
a.
4
Floating
3
2 1 1 ■1 ■
............................1 1 1 ■ ■ i ■ n
11 ■ ■ ■ ■ 11 ■ i ■ 1 1 1 1 1 1
1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
First-Stage Backing Pressure (Torr)
47
Chapter 2
Figure 2.9. Effect of first-stage backing pressure and condenser temperature on floating
potential. Floating potentials here were measured with an electrically isolated probe and
were not derived from I-V curves. Probe diameter, 0.6 mm; Probe position, on axis and
10.0 mm downstream from the sampling-plate orifice; Inner-gas flow rate, 1.04 L/min.
Potential (V)
Floating
6
Condenser Temperature (°C)
Chapter 2
7i
-
54-10
32 ~|"rT*n i"i r i i |
1.0
m
2.0
i
i
i ii iii ii i
3.0
|
ii iii ii ii
4.0
|
5.0
r i]
6.0
First-Stage Backing Pressure (Torr)
00
49
Chapter 2
Figure 2.10. Effect of inner-gas flow rate on space potentials and floating potentials
taken from I-V curves with condenser temperatures of (a) 10, (b) 15, and (c) 20° C. Probe
diameter, 1.0 mm; Probe position, on axis and 10.0 mm downstream from the samplingplate orifice.
50
Chapter 2
5.0*1
4.5 *
>
4. 0 *
*3
C
0)
o
D.
3. 5 *
Space
3. 0 *
2. 5 *
2.0 t —
0.80
Floating
0.90
1.00
1.20
1.10
Inner-Gas Flow Rate (L/min)
5. 0*1
4. 5 *
>
ra
C
<u
♦-»
o
Q.
Space
4. 0 *
3. 5 *
Floating
3. 0 *
2.5
0.90
1.00
1.10
1.20
Inner-Gas Flow Rate (L/min)
5. 0 i
Space
4. 5 *
>
4.0 *
c
3. 5 *
<u
o
Ql
Floating
3. 0 *
2. 5 *
2 . 0 *1—
0.80
0.90
1.00
1.10
Inner-Gas Flow Rate (L/min)
1.20
Chapter 2
51
Similarly, the floating potential are relatively constant, but at a high value, at first-stage
pressures above 3.5 Torr and condenser temperatures over 10°C.
This behavior is likely a result of several competing phenomena. At low first-stage
backing pressures the probe is embedded deep in the expansion core and is well shielded
from the surrounding background gases; as a result, small changes in pressure should
have little effect on the environment surrounding the probe. Nevertheless, large increases
in solvent load can still alter the gas by cooling and changing the composition of the plasma
before it is sampled. Similarly, a large increase in first-stage backing pressure would
reduce the dimensions of the expansion (see Sec. 4.3), alter the environment of the probe,
and increase the space and floating potentials. At higher first-stage backing pressures, near
3.4 Torr, the Mach disk moves in front of the probe (see Table 2.2), located 10.0 mm
downstream from the sampling orifice. Accordingly, the high-pressure plateau of the
floating potential in Figure 2.9 (beyond approximately 3.6 Torr) probably corresponds to
the location of the Mach disk (X^). Not surprisingly, however, the location of the plateau
is dependent also on solvent load.
Importantly, these variations in electrostatic potential (over 4 eV in some instances),
caused by a changing solvent load and first-stage backing pressure, are larger than the
kinetic energy an ion gains during gas-kinetic expansion (= 1.1 eV at a gas temperature of
5000 K [18,8]). It would therefore be expected that the resulting electrostatic effects would
dominate ion movement in the lower-pressure zones of the interface (second and third
stages) where ion movement is more susceptible to coulombic forces. In these zones,
changes in solvent load and first-stage backing pressure can indeed influence ion kinetic
energies and ion populations strongly. These effects are examined in detail in Chapters 3
and 4 of this series, in which ion kinetic energies and beam composition are monitored,
respectively.
Chapter 2
52
An important conclusion that can be deduced from the foregoing studies is that
fluctuations in either solvent load or first-stage backing pressure will have a smaller effect
on the electrostatic properties of the first-stage expansion if the solvent load and first-stage
backing pressure are both low. Because shifts in electrostatic potential have the ability to
alter ion kinetic energy and ion-beam density, such changes are best minimized.
2.4.2.2. Effect of Solvent Load and Inner-Gas Flow Rate. As shown in
Figure 2.10 space and floating potentials change only slightly with inner-gas flow rate or
solvent load at gas flows between 0.80 and 0.90 L/min. However, at flow rates above
0.90 L/min, the parameters appear to influence the potentials in a cooperative fashion. For
example, at a condenser temperature of 20°C (a relatively high solvent load; see Figure
2.4), potentials measured at the probe begin to increase when the inner-gas flow is raised
above 0.90 L/min. In contrast, at a lower condenser temperature of 10°C (a lighter solvent
load) space and floating potentials do not rise appreciably until the central-gas flow rate is
brought above 1.05 L/min.
The reason why changes in neither inner-gas flow nor solvent load have a strong
influence on potential at low inner-gas flow rates is probably related to the formation of a
well defined central channel in the ICP. At low inner-gas flows, the central channel is
visibly less distinct than at higher flows, suggesting that the solvent is not carried well into
the central channel. On the contrary, at higher inner-gas flows penetration of the solvent
will be more effective and can alter strongly the characteristics of the plasma being
sampled.
At relatively high condenser temperatures (15 to 20°C), both space and floating
potentials plateau at higher inner-gas flow rates (see Figures 2.10b and 2.10c). This
leveling might arise because further increases in inner-gas flow do not transport much
Chapter 2
53
additional aerosol to the plasma and because higher gas flows do not significantly alter the
plasma characteristics. In contrast, when large solvent loads are being introduced at low
inner-gas flow rates, small increases in flow rate may improve solvent transport to the
central channel and affect space and floating potentials accordingly (compare Figures 2.10a
and 2.10c).
At this time no unambiguous mechanism is obvious that accounts for the noted increase
in space and floating potential with first-stage backing pressure, solvent load, and inner-gas
flow. Clearly, changes in any of these parameters will affect the environment surrounding
the probe in some way. As either the solvent load or inner-gas flow rate is adjusted, the
composition, density, and temperature of the sampled gases will be altered. Changes such
as these will also influence the interaction of the plasma with the sampling interface (e.g.,
will influence charge separation and boundary layer formation) and will produce new
conditions at the probe. In a similar fashion, as the first-stage backing pressure is altered,
the gas density, temperature, and pressure that surrounds the probe will be changed.
2.4.3. Mapping Floating Potentials in the First Stage
In this section floating potentials that were measured throughout the expansion zone are
compared with the expected geometrical dimensions of the expansion calculated earlier and
compiled in Table 2.2. Spatial mapping was performed at two first-stage backing
pressures, 1.1 Torr and 3.0 Torr, with an electrically isolated probe having a 0.6-mm
diameter. Shown in Figure 2.11 are floating potentials measured at a first-stage backing
pressure of 1.1 Torr as the probe was moved horizontally and vertically in the expansion
from a dry plasma (with and without an inner-gas flow) and a wet plasma. As the probe
was moved horizontally (Figure 2.1 la), the floating potential remained relatively constant
until the probe reached an axial distance of 7.0 mm, at which point the floating potential
54
Chapter 2
Figure 2.11. Floating potential as a function of (a) radial and (b) axial position measured
when the inner-gas flow was off (labeled Off); on, but with no solvent introduction (labeled
Dry); and on with solvent being introduced (labeled Condenser, 10°C). Radial profiles
were taken 10.0 mm downstream from the sampling-plate orifice. Axial profiles were taken
on axis. Probe diameter, 0.6 mm; First-stage backing pressure, 1.1 Torr; Inner-gas flow
rate, 1.04 L/min.
55
Chapter 2
Floating
Potential (V)
3.41
Off
Probe
Expansion
2. 6 Sampling Plate
2.4-
□ Dry
» Condenser, 10°C
2.2 2.0
0
2
4
6
8
10 12 14
Axial Position (mm)
Floating
Potential (V)
3.51
3.0-
Probe
Off
Expansion
2.51
2.0 -
Dry
Sampling Plate
Condenser, 10°C
1.0
0
2
4
6
8
Lateral Position (mm)
10
Chapter 2
56
began to decrease. This decrease in floating potential probably corresponds to moving the
probe through the barrel shock, a distance ranging from Yg/2 to Yr/2 (see Figure 2.6),
which was calculated earlier to be from 6.5 to 7.5 mm (see Table 2.2). This decrease was
found to occur at approximately the same position regardless of plasma operating
conditions (dry, wet, and no inner-gas flow). The only difference seen between these three
plasma operating states is that with the inner-gas flow turned on the floating potential
typically dropped about 0.8 V.
As the probe was moved vertically downward from 12.0 mm to 2.0 mm behind the
sampling-plate orifice (Figure 2.11b) the floating potential increased slightly, but declined
as the probe was moved below 2.0 mm to within 1.0 mm of the sampling-plate orifice.
These lower values measured within 2.0 mm from the sampling-plate orifice might be the
result of moving the probe from the transition-flow region to the continuum-flow regime,
or to a place where magnetic fields begin to influence electron movement
Shown in Figure 2.12 are horizontal and vertical profiles measured in the expansion of
a wet plasma when the first-stage backing pressure was either 1.1 Torr or 3.0 Torr.
Because absolute potentials varied slightly on a daily basis (not more than 1 V), Figure
2.12 displays profiles taken at 1.1 Torr that were collected on the same day as those
mapped at a pressure of 3.0 Torr. The horizontal profiles in Figure 2.12a reveal that the
floating potential begins to decrease at a shorter lateral distance (4.0 mm) in the 3.0 Torr
expansion than in the 1.1 Torr expansion (where it is 6.0 mm). Again, the point at which
the floating potential rolls off is believed to correspond to the location of the barrel-shock
structure. For a first-stage backing pressure of 1.1 Torr, the shockwave boundary was
calculated to be from 6.5 to 7.5 mm off axis (Table 2.2), which is reasonably close to the
location in Figure 2.12 (6.0 mm) where the floating potential begins to decrease. At a firststage backing pressure of 3.0 Torr the maximum diameter of the isentropic core of the
57
Chapter 2
Figure 2.12. Floating potential as a function of (a) radial and (b) axial position measured
at a first-stage backing pressure of 1.1 and 3.0 Torr. Radial profiles were taken 10.0 mm
downstream from the sampling-plate orifice. Axial profiles were taken on axis. Probe
diameter, 0.6 mm; Inner-gas flow rate 1.04L/min; Condenser temperature, 10°C.
Chapter 2
58
a
Floating
Potential (V)
5 .0-1
4.0-
1.1 Torr
Probe
Expansion
3.02.0 -
3.0 Torr
Sampling Plate
0.0
5
0
15
10
Lateral Position (mm)
Floating
Potential (V)
3.6 1
3.0 Torr
3.4Probe
Expansion
1.1 Torr
3.0Sampling Plate
2 .8 2.6
0
5
10
15
Axial Position (mm)
20
Chapter 2
59
expansion (Yr) is calculated to be 7.5 mm. This calculated boundary of the barrel shock
likely corresponds to the decrease in floating potential at 4.0-mm off axis seen in the lateral
profile of Figure 2.12.
Because the probe movement was restricted to 12.0 mm in the vertical direction, it
could not be moved beyond the Mach disk when the first-stage backing pressure was 1.1
Torr (calculated to be 17.6 mm behind the sampling-plate orifice). However, floating
potentials could be examined behind the Mach disk with the first-stage backing pressure at
3.0 Torr (calculated to be 10.7 mm downstream from the sampling-plate orifice). As a
result, the structural features of the curves taken at first-stage backing pressures of 1.10
and 3.0 Torr are quite different in Figure 2.12b. At 3.0 Torr the floating potential achieved
a distinct maximum with the probe 6.0 to 8.0 mm downstream from the sampling-plate
orifice, whereas the potentials measured at a first-stage pressure of 1.1 Torr were nearly
constant with horizontal position. The maximum in the curve taken at 3.0 Torr possibly
corresponds to that region in the expansion just before the onset of background-gas
penetration. This region, often termed the optimal skimming position, Xmax, is calculated
to occur 7.1 mm downstream (Table 2.2). The difference in shape between the two
profiles taken at 1.1 and 3.0 Torr is probably due to the strength of the shockwaves in the
Mach disk region. At lower pressures the dimensions of the expansion core are larger and
the shockwave structure is more diffuse than at higher pressures.
2.4.4. Local Floating Potential Variation with First-Stage Backing Pressure
It can be deduced from the results in Sections 4.2 and 4.3 that floating potentials are
dependent on both position in the expansion and the dimensions of the expansion core.
For example, the floating potentials were found to change in different directions depending
on whether the probe was moved behind the Mach disk or the Mach disk displaced in front
60
Chapter 2
of the probe. This phenomenon can be seen by comparing Figures 2.9 and 2.12b. In
Figure 2.9 the floating potentials were found to rise as pressure was increased and the
Mach disk was moved in front of the Langmuir probe. Yet, as shown in Figure 2.12b the
floating potential decreases when the probe is translated beyond the Mach disk. This
comparison typifies the complicated electrostatic properties throughout the expansion
region.
Another example of these complicated electrostatic properties is shown in Figure 2.13.
Here, floating potentials measured as a function of first-stage backing pressure are
compared at two lateral positions (on axis and 9.0-mm off axis) in the expansion region.
For both measurements the probe was located 10.0 mm downstream from the samplingplate orifice. On the axis of the expansion the floating potential increased with first-stage
backing pressure, whereas 9.0 mm off axis the potential decreased. Although it is not clear
what mechanism is responsible for this opposing trend, a shift toward negative floating
potentials indicates that the electron diffusion rate to the probe is increasing relative to the
ion diffusion rate. These results demonstrate that electrostatic properties within the
expansion can vaiy at the probe surface and are dependent on whether the probe is moved
to a specific region or the first-stage backing pressure is adjusted. From a practical
standpoint, a similar behavior should occur when skimming in this region: The skimmer is
likely to sample ions from different electrostatic environments depending on if the
skimming position or first-stage backing pressure is adjusted.
2.5. C o n c l u s i o n
As an analytical plasma is sampled into a mass spectrometer interface, two major events
occur that alter the character of the plasma. The first is the preferential loss of electrons,
which disrupts the quasi-neutrality of the plasma. The second is a decrease in gas density,
61
Chapter 2
Figure 2.13. Floating potential as a function of first-stage backing pressure measured at
a lateral position (X) on axis and 9-mm off axis and at an axial position (Z) 10.0 mm
downstream from the sampling-plate orifice. Probe diameter, 0.6 mm; First-stage backing
pressure, 1.1 Torr; Inner-gas flow rate 1.04L/min; Condenser temperature, 10°C.
Chapter 2
Floating Potential (V)
X = 0 mm, Z = 10 mm
Probe
Expansion
Sampling Plate
X = 9 mm, Z = 10 mm
r i'i
1.0
m
m i | ■m i m i 111
2.0
in
3.0
i i ■i i i 11 i m i
4.0
hi
m 11 11 u i |
5.0
6.0
First-Stage Backing Pressure (Torr)
C\
N)
Chapter 2
63
caused by expansion of the bath gas into a low-pressure chamber, which reduces the
collisional frequency among species. As a result of these two processes, the importance of
ion-ion interactions increases. These electrostatic interactions cause ions to have different
velocities and trajectories than neutral-gas species and produce what is referred to as “ion
slip”, ion movement that deviates from what gas dynamic theory predicts [30]. However,
as these ions reach the shockwave structure, the collisional frequency between ions and
neutrals increases so the two kinds of particles once again share the same velocities and
trajectories [30].
Evidence is provided here that indicates how the shockwave structure can be located
with spatial profiles of probe potentials. According to these findings, the dimensions of the
free-jet expansion of an analytical ICP are similar to those of an expanding neutral gas. In
the continuum-flow region and shockwave structure of the expansion, where the collisional
frequency between ions and neutral bath-gas species is high, gas dynamic theory appears to
be useful for describing ion movement. However, in the transition- and molecular-flow
regions of the first-stage expansion and in the molecular-flow regions of the second and
third stages where space potentials are large, ion-ion interactions are likely to increase and
influence ion movement. The influence of these sorts of coulombic effects on ion
movement are covered in more detail in Chapters 2 and 3 of this study where ion kinetic
energies and ion beam composition are examined, respectively.
64
Chapter 2
2.6. L i t e r a t u r e C i t e d
[1]
G. M. Hieftje and G. H. Vickers, Anal. Chim. Acta 216, 1 (1989).
[2]
D. J. Douglas, Can. J. Spectrosc. 34, 38 (1989).
[3] S. M. Tan and G. Horlick, J. Anal. At. Spectrom. 2 , 745 (1987).
[4] G. R. Gillson, D. J. Douglas, J. E. Fulford, K. W. Halligan, and S. D. Tanner,
Anal. Chem. 60, 1472 (1988).
[5]
G. N. Spokes and B. E. Evans, Tenth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, Pa (1965), p. 639.
[6]
M. J. Vasile and H. F. Dylla, Plasma Diagnostics. Eds. O. Auciello and D. L.
Flamm, Vol 1. Academic Press, New York (1989), chapter 4.
[7] D. J. Douglas and J. B. French, J. Anal. At. Spectrom. 3,743 (1988).
[8] J. E. Fulford and D. J. Douglas, Appl. Spectrosc. 40, 971 (1986).
[9]
D. W. Koppenaal and L. F. Quinton, J. Anal. At. Spectrom. 3, 667 (1988).
[10] D. A. Wilson, G. H. Vickers, G. M. Hieftje, and A. T. Zander, Spectrochim. Acta
42B , 29, (1987).
[11] B. S. Ross, D. M. Chambers, G. H. Vickers, and G. M. Hieftje, Characterization of
a 9-mm Torch for Inductively Coupled Plasma-Mass Spectrometery, Appl. Spec.
(1990). in press.
[12] R. D. Deutsch and G. M. Hieftje, Appl. Spectrosc. 39, 214 (1985).
[13] R. F. Browner, Inductively Coupled Plasma Emission Spectroscopy. Ed. P. W. J.
M. Boumans, Part II. Wiley, New York (1987), chapter 8.
[14] J. D. Swift and M. J. P. Schwar, Electrical Probes for Plasma Diagnostics.
American Elsevier, New York (1969).
[15] H. B. Lim, R. S. Houk, and J. S. Crain, Spectrochim. Acta 44B, 989 (1989).
[16] D. M. Manos and H. F. Dylla, Plasma Etching: An Introduction. Eds. D. M. Manos
and D. L. Flamm, Academic, New York (1989), p. 263.
[17] J. B. Hasted, Physics of Atomic Collisions. American Elsevier, New York (1972),
p. 9.
Chapter 2
65
[18] M. Haung, D. S. Hanselman, P. Y. Yang, and, G. M. Hieftje, Isocontour Maps of
Electron Temperature, Electron Number Density and Gas Kinetic Temperature in the
Ar ICP Obtained by Laser-Light Thomson and Rayleigh Scattering, submitted to
Spectrochim. Acta B, 1990.
[19] P. W. Atkins, Physical Chemistry, Oxford University, Great Britain (1982),
p. 873.
[20] M. Mitchner and C. H. Kruger, Partially Ionized Gases. Wiley, New York (1973),
p. 58.
[21] A. L. Gray, R. S. Houk, and J. G. Williams, J. Anal. At. Spectrom. 2, 13 (1987).
[22] R. S. Houk, J. K. Schoer, and J. S. Crain, J. Anal. At. Spectrom. 2, 283 (1987).
[23] R. H. Kirchoff, E. W. Peterson and L. Talbot, AIAA J. 9, 1686 (1971).
[24] R. B. Fraser, F. Robben and L. Talbot, Phys. Fluids 14, 2317 (1971).
[25] H. B. Lim and R. S. Houk, Spectrochim. Acta 45B, 453 (1990).
[26] K. Bier and B. Schmidt, Z. Angew. Phys. 11, 34 (1961).
[27] H. W. Liepmann and A. Rosko, Elements of Gas Dynamics. Wiley, New York
(1957). Chapter 2.
[28] H. Ashkenas and F. S. Sherman, Rarefied Gas Dynamics, 4th Symposium. Ed.
J. H. de Leeuw, Vol. II, Adademic Press, New York (1966), p. 89.
[29] R. Campargue, Rarefied Gas Dynamics, 6* Symposium. Ed. L Trilling and
H. Y. Wachnam, Academic Press, New York (1969), p. 1003.
[30] M. Y. Jaffrin, Phys. Fluids 8, 606 (1965).
66
Chapter 3
3
ION KINETIC ENERGY
MEASUREMENTS
3.1. I n t r o d u c t i o n
Sampling an inductively coupled plasma (ICP) into the low-pressure environment of a
mass spectrometer (MS) has proven to be the least understood process that occurs in an
ICP-MS instrument. In this sampling process, ions from an atmospheric-pressure plasma
are shaped into an ion beam and focused toward a mass analyzer. The model currently
used to describe this system combines gas dynamic theory of the free-jet expansion [1]
with electrostatic-lens theory [2]. Unfortunately, this model fails to describe accurately ion
movement through the mass spectrometer interface [2]. The inaccuracy of these theories
stems from the fact that they do not take into account ion-ion collisions. Because these
theories have not been useful in guiding instrumental modifications, most of the
innovations in instrument design and operation have been achieved empirically or semiempirically.
Ion sampling in ICP-MS is derived from the skimmed free-jet expansion method used
in jet spectroscopy [3]. Although the skimmed free-jet expansion is a straightforward
Chapter 3
67
technique to sample a high-pressure gas into a low-pressure region, there are several
differences between sampling a plasma and a neutral rare gas. In jet spectroscopy the bath
gas is a pure, often noble gas that has been seeded with a low concentration of an analyte
species, whereas in ICP-MS, the bath gas is a plasma to which a large quantity of analyte
and solvent material has been added. Also, the gas dynamic theory of a free-jet expansion
describes the movement only of neutral species; it does not describe ion movement or take
into account coulombic forces that influence ion movement.
The fact that ion movement cannot be described accurately by gas dynamic theory alone
is evident from the "ion-slip" effect [4]; ions and atoms exhibit different velocities in a
ffee-jet expansion. The existence of electrostatic fields produced by the ion space charge
are responsible for this phenomenon. A related electrostatic effect has been reported by
physicists interested in plasma-source mass spectrometry for plasma-flux analysis [5,6,7].
In these plasma-diagnostic endeavors, physicists were concerned not only with the effects
of gas-kinetic processes on ion movement, but also with interferences from electrostatic
fields of the ion space charge. They demonstrated that as analyte ions are sampled from a
plasma into the low-pressure regions of a mass spectrometer, two important forces
influence ion movement: gas diffusion and ion-ion interactions [8]. These same forces
likely exist in the ICP-MS interface. Gas-kinetic processes are known to be dominant as
the plasma is initially sampled into the mass spectrometer interface because ion-neutral
collisions are high [1,8]. However, as analyte ions are transported to the lower pressure
regions of the interface, ion-ion interactions will increase.
Although it has not previously been determined where in the ICP-MS interface these
electrostatic forces begin to overpower gas-kinetic forces, the existence of electrostatic
forces has been deduced by several groups through ion kinetic energy measurements. In
these measurements it was found that ions had velocities greater than could be rationalized
Chapter 3
68
by gas-kinetic processes alone. The first of these studies was reported by Olivares and
Houk [9] who measured ion kinetic energies at the mass analyzer. They attributed the high
measured kinetic energies to a residual “pinch discharge” that forms between the plasma
and the sampling plate. Following this work was a similar study by Fulford and Douglas
[10] who used a center-tapped load coil as a modification to the ICP source. Lower ion
kinetic energies were reported and were attributed to a weakening of the residual pinch
discharge [11]. The most probable energy (MPE) measured by Fulford and Douglas for
the argon ion (40 amu) was low, at approximately 2 eV, compared to the 14 eV determined
by Olivares and Houk [9]. Although Fulford and Douglas [10] reported much lower ion
kinetic energies, their results still could not be reconciled with gas-kinetic processes alone.
Fulford and Douglas [10] supported the hypothesis of Olivares and Houk [9] for the
existence of an electrostatic force. They described this electrostatic force as being derived
from the plasma space potential or possibly from the ion-beam space charge. However, the
specific region in the interface where the ions gain energy from that force was remains
uncertain.
In this chapter those forces that influence ion movement (i.e., gas kinetic and
electrostatic) are studied in different regions of the interface through the measurement of ion
kinetic energies. In addition, the effect of several instrumental parameters (i.e., first-stage
backing pressure, and solvent load) on these forces is examined. The object of these
experiments is to guide the development of a qualitative model that describes ion movement
through each vacuum chamber of the mass spectrometer interface and to ascertain how
these forces might be controlled.
69
Chapter 3
3.2.
Ex p e r im e n t a l
3.2.1. Conversion of the ICP-MS into a Retarding-Plate Energy Analyzer
The ICP-MS system used for these experiments is described in Chapter 2; however,
several modifications were made to covert the instrument into a retarding-plate energy
analyzer that is similar to those used in the physics field [6,7,12].
In the first stage of the interface the Langmuir probe (refer to Chapter 2) was removed
and the skimmer cone was restored at a distance of 10.0 mm from the sampling-plate
orifice. In the second and third stages, ion-optic lenses were removed and a retarding plate
was placed in each stage as shown in Figure 3.1. Removal of the ion-optic lenses not only
made room for the retarding plates, but also eliminated the possibility of complicating
electric-field interactions [5,6,7] among the retarding plates, the optics, and any spacecharge fields of the ion beam.
An ion-optic plate in the second stage, which served also as a differential-pumping
aperture between the second and third stages, was replaced with a retarding plate 62.5 mm
in diameter and 1.0 mm in thickness. This second-stage retarding plate was positioned
61.5 mm behind the skimmer-cone orifice. A third-stage retarding plate, which was 27.6
mm in diameter and 0.5 mm in thickness, was located 20.0 mm in front of the quadrupole
housing and 114.0 mm behind the second-stage retarding plate. Apertures of 0.5 mm were
chosen for both these plates to achieve high spatial resolution and to produce equipotential
retarding fields near the sampling orifice in each plate [12]; plate thicknesses were dictated
by convenience of machining.
The absence of ion-optic lenses reduced ion-transport efficiency through the interface
and required the use of ample analyte concentrations to produce usable ion signals.
Nevertheless, solution concentrations were kept as low as possible so ion kinetic energies
would not be biased by analyte concentration. It was reported by Fulford and Douglas that
70
Chapter 3
Figure 3.1. Geometry of the ICP-MS first-, second-, and third-stage interface regions
and the location of second- and third-stage retarding plates.
71
Chapter 3
To
Electron
Multiplier
6
Quadrupole\
\
/Quadrupole Guard
/
Third-Stage
Retarding Plate
Ceramic Spacer^
t
J
20 mm
1
114 mm
Second-Stage
Retarding Plate
•Skimmer Cone
^ 6 1 5 mm
Sampling Plate
Vacuum Port
Chapter 3
72
high analyte concentrations might alter ion kinetic energies [10]. Because solution
concentrations greater than 200 fig/mL were found to alter the shape of the retardingpotendal curves, analyte and background ions were kept below this level.
Elements were chosen that were essentially monoisotopic. Solutions, which were
prepared from 1000 ppm stock solutions, were 10 (ig/mL Li (Fisher), Co (Fisher), Zn
(Morton Thiokol), Mn (made from Mn metal [13]), Sr (Morton Thiokol), and Ce (Aldrich)
orCs(madefromCsCl[13]); 200 (ig/mLBi (made from Bimetal [13]); and 100 (ig/mL
U (Aldrich). In most cases the use of these solution concentrations provided easily
measurable signals. However, the signal level was sometimes inadequate for 200 pg/mL
bismuth, which is why bismuth does not appear in all ion kinetic energy curves shown
later.
3.2.2. First-Stage Vacuum System
The first-stage pumping configuration used for these studies is the same one described
in Chapter 2. As discussed in Chapter 2, the vacuum system was specially configured so a
wide range of first-stage backing pressures could be obtained. For the present experiments
an additional vacuum port was opened to the first-stage pump, giving a total of three
vacuum lines leading to the pump.
3.2.3. Sample Introduction System
The sample introduction system used in these experiments is the same one described in
Chapter 2. The desolvation system described in Chapter 2 was employed to regulate the
amount of aerosol and solvent vapor introduced into the plasma. This unit was calibrated
only once to illustrate the effect of condenser temperature on solvent load for this particular
desolvation system (see to Figure 2.4). Although a rise in condenser temperature
Chapter 3
73
corresponds to an increase in solvent load, the absolute solvent load is expected to vary
somewhat from run to run and with experimental conditions. For this reason, the solvent
load is later indicated in terms of condenser temperature.
Before retarding-plate analyses were performed, the analyte signal was maximized by
selecting the proper first-stage backing pressure and condenser temperature (related to
solvent load). As a general trend higher solvent loads required lower first-stage backing
pressures. The procedure used here involved maximizing manganese signal when the
second- and third-stage retarding plates were grounded. Manganese was chosen because it
has a mass close to argon, the bath gas. In the ion-flux monitoring experiments to be
presented in Chapter 4, it will be shown that these optimal conditions are dictated by ionbeam space charge in the low-pressure zones of the interface.
3.2.4. Retarding-Plate Analyses
A positive voltage ramp (0 to +10 V) was applied to the selected retarding plate from
the output of a function generator (Hewlett Packard, Model 3325A). To extend the range
of the ramp, a voltage amplifier with a gain of two was placed between the function
generator and the selected retarding plate. The retarding-plate voltage could then be
scanned either from 0 to +10 V at 0.2 V/s or 0 to +20 V at 0.4 V/s. Ions having sufficient
energy to pass through the retarding-plate aperture were focused through the quadrupole
(Balzers, Model QMG 511) and were detected with a discrete-dynode secondary electron
multiplier [16]. Ion signal was recorded in counts per second (cps). Both the analyte
signal and retarding voltage were sampled at 10 Hz by a personal computer (IBM XT,
model 5160) with a laboratory data-collection program (Asystant+, Version 1.0, Macmillan
Software).
Chapter 3
74
Retarding-plate analyses were performed in two ways: The first involved performing
the analyses sequentially in the second and third stages; an analysis was performed in the
third stage while the second-stage retarding plate was grounded and then performed in the
second stage while the third-stage plate was grounded. The second group of analyses was
carried out in the second stage with the third-stage retarding plate removed. Analyte flux
could be increased to the quadrupole by removing the third-stage retarding plate because the
aperture diameter of the quadrupole housing was 2.5 mm.
The time allotted to collect each ion kinetic energy curve was kept within a 50-minute
window to minimize drift. This narrow time window limited the qumber of elements and
sample repetitions that could be performed. Retarding-potential curves were collected for
seven elements, with one to three repetitions for each element. The elements were ordered
randomly so experimental drift did not bias the measurements. Likely causes for such
drift, which were discussed in Chapter 2, include fluctuations in first-stage backing
pressure, solvent load, inner-gas flow rate, or variations in the plasma characteristics.
Mean ion kinetic energies were determined from the experimental curves after they had
been smoothed with a Blackman window (0.03 Hz cut-off frequency), which accompanied
the Asystant+ software package. This smoothing routine was chosen because it was
quicker than a polynomial curve fit and did not require prior knowledge of the functional
curve shape. The smoothed curves were compared to a fifth-order polynomial fit and were
found to be in close agreement (usually less than 3% deviation). The standard deviation of
the mean is included with some of the results presented later to illustrate the magnitude of
the curve-to-curve variance. The retarding-potential curves presented later in this paper are
unsmoothed to provide an indication of typical signal-to-noise ratios.
The retarding-potential curves recorded in this study exhibited a classical quasi­
exponential character, as a result, it was judged to be more reliable to determine a mean
75
Chapter 3
energy rather than MPE. Mean energies were determined by adding the "fall-off potential"
to the average retarding potential, which was calculated from the curve area starting at the
fall-off potential. The fall-off potential is that potential at which the plate begins to retard
ions as illustrated in Figure 3.2.
3.2.5. Plasma-Source and Operating Parameters
The asymmetrically grounded, inverted load-coil configuration is standard for the ICPMS system used here [14]. However, a center-tapped load coil ICP [15] was interfaced to
the mass spectrometer so retarding-potential and ion kinetic energy curves for the two
instruments could be compared. For this comparison, operating parameters for both the
asymmetrically grounded and the center-tapped load-coil ICP-MS systems were matched
as closely as possible, as seen in Table 3.1. The inner-gas flow and the sampling depth
were the only two parameters that were slightly different between the two systems; these
two parameters were adjusted to maximize signal levels [16].
3.3. B a c k g r o u n d
3.3.1. Use of Retarding Plates to Determine Ion Kinetic Energies
One of the most convenient methods to measure ion-kinetic energies in a mass
spectrometer interface is by retarding-plate analysis. The concept is straightforward;
analyte ions with kinetic energies higher than the potential applied to a given retarding plate
will pass through the plate aperture and proceed to the detector. In contrast, those ions
with lower energy will be reflected and not detected. This method has been borrowed from
similar experiments performed by physicists involved in plasma-flux analysis [5,17-20].
Although others have attempted to determine ion kinetic energies by using the quadrupole
76
Chapter 3
Figure 3.2. Illustration of the method used to calculate mean ion kinetic energy from a
retarding-potential curve. See text for discussion.
Chapter 3
Fall-Off Potential (Vf )
Intensity
AV
i(ma:
AV
Voltage
dV
Mean =
[di
/ di
Area
i(max)
Mean Kinetic Energy = Fall-Off Potential + Mean
78
Chapter 3
Table 3.1. ICP-MS Operating Conditions for the Aasymmetrically Grounded (Inverted)
and Center-Tapped Load-Coil ICP-MS
Inverted Ground
Center taDoed
Forward Power (W)
1250
1300
Reflected Power (W)
<10
<10
Argon Gas Flow (L/min)
Inner
Intermediate
Outer
1.02
1.00
11.0
1.30
1.00
11.0
Sample Uptake (mL/min)
0.64
0.64
Sampling Depth (mm)
10.0
13.0
First-Stage Pressure (Torr)
2.3
2.1
Second-Stage Pressure (Torr)
2.2 x 10-3
2.1 x 10-3
Third-Stage Pressure (Torr)
5.8 x 10’6
6.0 x 10'6
Desolvation System
Condenser Column (°C)
Heater Column (°C)
5
200
5
200
Chapter 3
79
bias potential as a retarding field [9,10,21-24], the retarding-plate method was chosen here
for several reasons. First, a retarding plate can be positioned in different regions of the
mass spectrometer so ion velocities can be determined in spatially selected zones. Second,
information concerning ion-beam space charge can be deduced from distortions that appear
in retarding-potential curves. Because a retarding plate operates in the same manner as an
ion-optic lens, such distortions would alert one to space-charge artifacts that might affect
the ion-optic lenses. Finally, because the quadrupole bias potential is superimposed upon a
hyperbolic field, the retarding field it generates may be somewhat distorted. In addition,
for a quadrupole with a low mass range (0 to 511 amu) such as the one used in this study,
ramping the bias potential can compromise mass-analysis capabilities by changing the ionfocusing rf and dc fields. In turn, these field changes can lead to mass-bias effects that
cause heavy ions to have less stable trajectories than lighter ones.
These problems are illustrated in Figure 3.3, which shows retarding-potential curves
for lithium, argon, manganese, and uranium obtained by using the quadrupole pole-bias
potential as a retarding field. The drastic mass dependence of the curve shapes is believed
to be the result of quadrupole mass biasing. The curves for heavy analyte ions appear to be
more exponential in shape than those for lighter ions. This trend biases the calculated mean
energies of heavy ions toward lower values. Ordinarily, heavy ions would be expected to
have higher kinetic energies than lighter species; however, because of this bias, the
opposite trend appears. Similar behavior can be seen in the results of Gray and Williams
[24].
Because quadrupole mass filters can accept ions over a fairly broad range of kinetic
energies, serious biasing effects are not anticipated when retarding plates are employed.
The ion kinetic energy distributions measured in this study did not exceed 20 eV, well
within the acceptance limits for a quadrupole. In addition, the occurrence of ion kinetic
80
Chapter 3
Figure 3.3. Retarding-potential curves obtained using the quadrupole pole bias as the
retarding field for (a) lithium, (b) argon, (c) manganese, and (d) uranium. Condenser
temperature, 5°C; First-stage backing pressure, 2.0 T oit; The energy (in eV) cited below
each element designation is the mean ion energy for the particular curve.
1.0e+6i
sl3nal <CPS>
Signal (cps)
8.0e+5 *
Lithium (7 amu)
7.1 eV
1 .Oe+3 ■
5.0e+2
O.Oe+O
0
5
10
15
20
Retarding Potential (V)
Argon (40 am u)
5.3 eV
6.0e+5 ■
4.0e+5 *
0.0e+0
25
0
5
10
15
20
R etarding P otential (V)
25
1.0e+3i
2.0e+3i
Signal (cps)
Signal (cps)
8.0e+2M anganese (55 am u)
6.3 eV
1.0e+3
U ranium (238 amu)
6.3 eV
6.0e+2
4.0e+22.0e+20.0e+0
0.0e+0
0
5
20
10
15
Retarding Potential (V)
25
0
5
10
20
15
R etarding P o ten tial (V)
25
00
Chapter 3
82
energy biasing was not evident from the curve shapes. Elements of different mass and
with different mean ion kinetic energies exhibited similar quasi-exponential curve shapes,
the expected result in retarding-plate analysis (exceptions caused by "potential effects", are
discussed in Section 3.4.1) [5, 6,7].
3.3.2.
Gas Dynamic Acceleration
As ICP gases are sampled into the first stage of a mass spectrometer interface, an ion
in that gas stream can gain kinetic energy from either gas-kinetic or electrostatic forces.
Although there is a substantial body of knowledge on gas-kinetic processes, there exists
little information that describes the effect of ion-beam space charge on ion movement.
However, because the influence of gas-kinetic forces on ion movement can be predicted
from gas dynamic theory and the accompanying empirical relationships, the contribution of
electrostatic forces can be approximated as the difference between those calculated values
and experimental results.
A detailed description of gas dynamic theory as it applies to ICP-MS instruments has
been provided by Douglas and French [1]. In their treatment, the kinetic energy of the
expanding bath gas (in this case neutral argon) is predicted from standard gas dynamic
relationships. All other species in the expansion are then assumed to share the velocity
distribution of the argon bath gas because of the high collisional frequency in the
continuum-flow region of the expansion. This process by which species of different mass
are brought to the bath-gas velocity is termed "gas dynamic acceleration". As a result of
this similar velocity distribution, the kinetic energy of any heavy species will be greater
than that of any lighter species.
In such a situation, the energy of a seed ion of any desired mass can be deduced by
first determining the velocity of the bath-gas species and then calculating the energy for the
83
Chapter 3
desired seed particle, which is assumed to attain the same mean velocity. The ion kinetic
energy of bath-gas species that expand isentropically in a free jet can be described by [25]
Cp(gas) dT
(3.1)
where Egas is the kinetic energy of the bath-gas species, C p^y is the specific heat of the
bath gas at constant pressure, T0 is the gas temperature of the source, and T is the
temperature at a specified point in the expansion. If heavy species that are mixed with the
expanding bath gas sustain a sufficient number of collisions to achieve the velocity of the
bath gas [1,10], their kinetic energy can then be described by [25]
tWsas)<i'r
(3.2)
where m refers to atomic mass and the subscript h denotes heavy species. The heavy
species will therefore have kinetic energies greater than the bath-gas species by a factor of
mh/mgas- This type of gas dynamic acceleration has been noted by Fulford and Douglas to
occur in an ICP-MS instrument [10].
It is important to note that the behavior of species lighter than the bath-gas atoms is not
described accurately by Equation 3.2. Lighter species are more likely to be accelerated in
random directions upon collisions with heavier species, an effect which would broaden
their ion kinetic energy distributions, whereas heavier species remain more on axis. In
some instances, the results presented in this study show lithium and aluminum to have
84
Chapter 3
higher ion kinetic energies than that of argon. Inefficient energy transfer between these
lighter elements and the bath-gas atoms is suspected to be the cause for this phenomenon.
3.4.
R e su l t s and d i s c u s s i o n
3.4.1. Retarding-Potential Curves and "Potential Effects"
Interpretation of retarding-potential curves can be complicated by “potential effects”
(also termed “focusing effects”) [5,6,7], which alter the shape of the curves by electrostatic
focusing of the ion beam. This focusing can occur through the formation of a virtual ion
lens between the retarding plate and the electrostatic fields produced by the ion beam itself.
Accordingly, the phenomenon is significant only in rather dense ion beams. Focusing
effects have been well documented in both the mass spectrometry and physics fields
[5,6,7,26]. In mass spectrometry the effect is derived from the electrostatic field of the ion
beam (referred to as ion-beam space charge) and compromises the ability of ion-optic
lenses and a mass analyzer to focus ions predictably [26]. In the physics field an identical
effect is reported in retarding-plate analyses on PS-MS instruments [5,6,7].
An example of these potential effects can be seen in Figure 3.4 where the retardingpotential curves for lithium (7 amu) measured in the second and third stages are compared.
Here the distortion is most prominent in the measurement performed in the second stage.
However, potential effects can appear also in the third stage when the orifice diameter in the
second-stage retarding plate is enlarged from 0.5 to 1.0 mm and when high solution
concentrations are employed (for example, see Figure 3.5). The orifice in the retarding
plate has this effect because it serves also as a differential pumping aperture; a larger orifice
increases ion flux to the third stage sufficiently to produce a high space charge which, in
turn, perturbs the retarding-plate potential fields. Not surprisingly, the parameters that alter
85
Chapter 3
Figure 3.4. Second-and third-stage retarding-potential curves for lithium. An
electrostatic ("potential") effect is evident in the second-stage curve. Condenser
temperature, 5°C; First-stage backing pressure, 2.0 Torr, Second-third-stage differential
pumping aperture, 0.5 mm.
Chapter 3
Signal (cps)
3.0e+31
2.0e+3
1
10 ppm Lithium
Third Stage
/
1.Oe+3:
y 1 0 ppm Lithium
L . 1/
Second Stage
O.Oe+O
t —1 —
r—1 -j
10
15
20
Retarding Potential (V)
25
00
o\
87
Chapter 3
Figure 3.5. The effect of high solution concentrations on third-stage retarding-potential
curves measured for (a) lithium and (b) uranium. Potential effects increase in severity at
high analyte concentrations. Condenser temperature, -5°C; First-stage backing pressure,
3.2 Torr, Second-third-stage differential pumping aperture, 1.0 mm.
Chapter 3
88
5.0e+4
Net Signal (cps)
4.0e+4 ■
3.0e+42.0e+4-
1
1000 ppm Lithium
1.Oe+4
10 ppm Lithium (87.1 X)
O.Oe+O
20
Retarding Potential (V)
25
Net Signal (cps)
3.0e+2i
2.0e+2
I [*li
\m.i
i
mu
1000 ppm Uranium
10 ppm Uranium (6.43X)
1.0e+2-
0.0e+0
Retarding Potential (V)
Chapter 3
89
the magnitude of the potential effects are those that influence ion flux, such as analyte
concentration and first-stage backing pressure.
Figure 3.5 reveals that highly concentrated analyte solutions can induce potential
effects in the third-stage retarding-plate analysis regardless of the mass of an analyte ion.
As seen in this figure both Li and U retarding-potential curves broaden and exhibit an
anomalous shoulder at approximately 8 eV when the analyte concentration is high. The
same behavior was observed with the other elements examined in this study.
Potential effects were evident also in argon-ion retarding-potential curves when high
analyte concentrations were being introduced into the plasma (see Figure 3.6). The curve
distortion is minimal when distilled-deionized water (Figure 3.6a) is introduced; however,
the presence of 1000 ppm Li (Figure 3.6b) or U (Figure 3.6c) produces a pronounced
potential effect Notice also that high analyte concentrations lower the Ar1" signal (compare
vertical scale in Figure 3.6a with those in Figures 3.6b and 3.6c). This response is likely
cause by the addition of highly concentrated solutions that increase the ion-beam density to
a point where it is dispersed spatially and transmitted less efficiently. This dispersion
process is covered in more detail later in this section.
As suggested above, first-stage backing pressure also can influence the magnitude of
these potential effects. Second- and third-stage retarding-potential curves for the argon ion
(40 amu) at different first-stage pressures are compared in Figure 3.7. In general, potential
effects were far more evident in the second stage than in the third stage and became greater
as the first-stage backing pressure was decreased. In particular, the second-stage curve
measured at a first-stage backing pressure of 1.1 Torr exhibited severe distortions. In
contrast, the curves obtained at 3.6 and 5.0 Torr appeared quasi-exponential and yielded
the same mean energy of 2.1 eV.
90
Chapter 3
Figure 3.6. Third-stage retarding-potential curves for argon (40 amu) while (a) water,
(b) 1000 ppm lithium, and (c) 1000 ppm uranium are nebulized into the plasma. The argon
retarding-potential curves become distorted as the result of potential effects, which increase
with sample-solution concentration. Condenser temperature, -5°C; First-stage backing
pressure, 3.2 Torr, Second-third-stage differential pumping aperture, 1.0 mm.
Chapter 3
2.0e+5
Argon
No concomitant
Signal (cps)
1.56+5
1.06+5
0.0e+0
0
5
10
15
20
25
Retarding Potential (V)
8.0e+3
Signal (cps)
Argon
Li concomitant
O.Oe+O
0
5
10
15
20
25
Signal (cps)
Retarding Potential (V)
Argon
U concomitant
1.Oe+4
5.0e+3:
0.0e+0
0
5
10
15
20
Retarding Potential (V)
25
92
Chapter 3
Figure 3.7. Retarding-potential curves for argon (40 amu) measured in the (a) second
and (b) third stage. Cited pressures pertain to those in the first stage. Curves have been
normalized by factors that appear in parentheses. Potential effects are more evident in the
second stage with a first-stage backing pressure of 1.1 Torr (nominally the "optimal"
skimming condition) although the signal is highest at 3.6 Torr. Condenser temperature, 5°C; Second-third-stage differential pumping aperture, 0.5 mm.
Chapter 3
1.2e+5l
Signal (cps)
1.0e+5i
Argon
8.0e+4i
5.0 torr (1.77X)
3.6 torr
4.0e+4 i
1.1 torr (10.7X)
2.0e+4i
0.0e+0
0
5
10
15
20
25
Retarding Potential (V)
1.2e+5 1
Signal (cps)
1.0e+5
Argon
8.0e+4
6.0e+4
5.0 torr (1.39X)
4.0e+4
2 . 00+4 *■
3.6 torr
If s W
1.1 torr (42.8X)
0.0e+0
* * 1 * ~ y i ~ - 1 *-j-* 'I — ^
5
10
15
20
Retarding Potential (V)
■i " I - 1
25
Chapter 3
94
In the third stage, potential effects were absent when the orifice diameter leading to that
stage was held at 0.5 mm, as shown in Figure 3.7b. The mean ion kinetic energies that
were determined from these third-stage measurements are 1.3,1.8, and 2.3 eV for firststage backing pressures of 1.1, 3.6, and 5.0 Torr, respectively.
It is somewhat surprising that the largest ion signal in Figures 3.7a and 3.7b occurred
at a backing pressure of 3.6 Torr. Ordinarily, skimming should be most efficient at 1.1
Torr among all the pressures studied, because the skimmer cone is then located close to the
optimal skimming position, which is described by Equation 2.4. The optimal skimming
distance at 1.1 Torr is calculated to be at 10.6 mm (the skimmer cone is placed 10 mm
behind the sampling-plate orifice in the present system). This calculation assumes a source
temperature of 5000 K [27], a shielding coefficient of 0.125 [28], and a Knudsen number
of 1.3 x 10‘3 (refer to Chapter 2).
Contrary to this prediction, a first-stage backing pressure of 1.1 Torr produced the
lowest ion sensitivities and the most severe potential effects of all the pressures
investigated. At this low first-stage backing pressure, the skimmer is embedded deep in the
isentropic core of the expansion, background-gas penetration is minimal, and the flux of
bath-gas ions passing through the skimmer orifice should be the greatest.
To explain this apparent anomaly, let us consider briefly the events that occur when a
plasma expands into the first stage of a vacuum system and how first-stage backing
pressure might influence the properties of the sampled ion beam. First, as the plasma
passes through the minute sampling aperture, electrons will be preferentially lost [29].
This phenomenon is referred to as charge separation. In turn, ions will be less shielded
from each other and ion-ion interactions will increase [30]. Second, as the expanding ion
beam proceeds, there is a drop in the density of all species. As a result, a decrease in ionneutral collisions occurs, and ions can more freely respond to external fields, particularly
Chapter 3
95
those generated by other charged species. Finally, the penetration of background gases
into the expansion will increase at higher backing pressures and can cause scattering of the
expanding beam before it reaches the skimmer. As a consequence, ion velocities will
become somewhat more randomly directed and a greater fraction of ions will be neutralized
by collision with the walls of the skimmer cone [29].
From these observations, it is possible to set down the events that would be expected
to occur under extreme conditions of first-stage pressure. At a pressure corresponding to
"optimal skimming" (1.1 Torr in the present situation), a high flux will be transmitted to the
second stage. As this ionized gas passes through the skimmer orifice, both charge
separation and a fall in gas density will take place and ion-ion interactions will increase. In
turn, these interactions will disperse the ion beam more greatly, induce potential effects like
those seen in Figure 3.7, and lower ion flux to the third stage (and ultimately to the
detector).
In contrast, at rather high first-stage backing pressures (5.0 T oit here), penetration of
background gas into the expansion core becomes severe. Consequently, sampled ions are
scattered extensively and the ion beam is diluted by the background species. Although
these events diminish electrostatically induced ion-beam dispersion and remove the
anomalous potential effects, transmission efficiency of the ion beam is lower.
Between these extremes must exist a first-stage pressure at which the density of the ion
beam is low enough and the concentration of bath-gas ions high enough that potential
effects are eliminated and electrostatic dispersion is minimal, but where ion transmission is
still quite efficient This compromise pressure was determined to be 3.6 Torr under the
conditions given in Figure 3.7.
Aside from the foregoing considerations, the existence of shoulders (potential effects)
on the curves in Figure 3.7a indicates the presence of electrostatic forces in the ion beam
Chapter 3
96
that can overpower gas dynamic forces. Unfortunately, these "potential effects" also
complicate the interpretation of retarding-plate measurements. Even though valuable
information concerning ion-beam space charge can be extracted from these anomalies,
potential effects have been minimized in the following experiments to simplify
interpretation of the curves.
3.4.2. Retarding-Plate Analyses
Results from retarding-plate experiments that were performed sequentially in the
second and third stages are plotted in Figures 3.8 and 3.9, respectively. The displayed
retarding-potential curves for Li+, Ar+, Mn+, and U+ reveal ion sensitivities and curve
shapes that are representative of all those measured in the mass spectrometer second and
third stages. Ion kinetic energy plots extracted from these retarding-potential curves
(Figures 3.8 and 3.9) are compared in Figure 3.10.
In the third stage a difference in mean kinetic energy of approximately 6 eV was
measured between light (Li) and heavy (U) species (see Figure 3.10). A similar range was
reported by Fulford and Douglas [10] and is likely attributable to gas dynamic acceleration
(defined in Section 3.3.2). In contrast, no significant change in mean energy was found
for ions of different mass in the second stage; as a result, the ion kinetic energy plot has a
slope of approximately zero. Such a slope signifies that gas dynamic acceleration is being
suppressed; the mechanism explained below and portrayed in Figure 3.11 attributes this
phenomenon to coulombic forces.
In this model, electrons have been largely stripped from the expanding partially ionized
gases to produce a positively charged ion beam. Additionally, in the latter parts of the firststage expansion and in the second and third stages of the mass spectrometer, gas density
decreases to the point where the collisional frequency between ions and neutrals becomes
97
Chapter 3
Figure 3.8. Second-stage retarding-potential curves for (a) lithium, (b) argon,
(c) manganese, and (d) uranium. The energy (eV) associated with each curve is the mean
ion energy calculated from the curve. The similarity among these curves demonstrates the
suppression of gas dynamic acceleration. Condenser temperature, 5°C; First-stage backing
pressure, 2.0 Torr.
Lithium (7 amu)
2.8 eV
.
Signal (cps)
a
o>
Argon (40 amu)
2.6 eV
3.0e+4
1.0e+4
O.Oe+O
0
0.0e+0
5
10
15
20
Retarding Potential (V)
25
0
5
15
20
Retarding Potential (V)
10
25
1.0e+2
l.0e+3
8.0e+2
Manganese (55 amu)
3.2 eV
Signal (cps)
a
6.0e+2
a
Uranium (238 amu)
3.7 eV
6.0e+1
4.0e+1:
2.0e+1
0.0e+0
0
5
10
15
20
Retarding Potential (V)
25
0
5
10
15
20
Retarding Potential (V)
25
99
Chapter 3
Figure 3.9. Third-stage retarding-potential curves for (a) lithium, (b) argon,
(c) manganese, and (d) uranium. Mean energies increase with mass as a result of gas
dynamic acceleration. The energy (eV) associated with each curve is the mean ion energy.
Condenser temperature, 5°C; First-stage backing pressure, 2.0 Torr, Second-third-stage
differential pumping aperture, 0.5 mm.
Signal (cps)
3.0e+3
2.0e+51
Lithium (7 am u)
2.8 eV
O
ST
o
T3
"I
Argon (40 am u)
3.6 eV
CL
1.Oe+5
o>
1 .Oe+3
5.0e+4
O.Oe+O
0
5
10
20
15
Retarding Potential (V)
0.0e+0
25
0
5
10
20
15
Retarding Potential (V)
25
2.0e+2n
Signal (cps)
1.5e+2Uranium (238 amu)
8.9 eV
<
/>
a.
M anganese (55 am u)
4.6 eV
o
1.0e+3
<0
1-0e+2‘
c
o>
55
5.0e+1 -
0.0e+0
0.0e+0
0
5
10
20
15
Retarding Potential (V)
25
0
5
10
20
15
Retarding Potential (V)
25
101
Chapter 3
Figure 3.10. Mean ion kinetic energies deduced from retarding-potential measurements
performed in the second and third stages (see Figures 3.8 and 3.9). Condenser
temperature, 5°C; First-stage backing pressure, 2.0 Torr, Second-third-stage differential
pumping aperture, 0.5 mm.
Mean Ion Kinetic Energy (eV)
Chapter 3
First-Stage Pressure
a 2.6 Torr
7-
Li Al
3.4 Torr
Ar
50
150
200
250
Ion Mass (amu)
102
103
Chapter 3
Figure 3.11 Illustration of how the effect of gas dynamic acceleration can be suppressed
in the second stage. As the retarding potential is increased, bath-gas ions (Ai*),
represented by small circled + signs, are reflected. In turn, the Ar*-traveling away from the
retarding plate prevents other ions (large circled + sign) from passing. Similar events are
hypothesized to occur in the third stage when the ion flux to that stage is high.
104
Chapter 3
Second-Stage
Retarding Plate
Bath-Gas (Ar)
Ions
©
Anayte Ion
Skimmer Cone
Chapter 3
105
relatively insignificant. At the same time, ion-ion collisions increase in importance because
of diminished shielding by electrons and a lower ion-neutral collisional frequency. These
inter-ion collisions restrict ion mobility and cause ions to respond to the retarding-potential
field in a collective manner. That is, as the potential of the second-stage retarding plate is
increased, collisions with reflected bath-gas (argon) ions will impede the movement of
heavier ions that would otherwise not be reflected by the plate potential. As a consequence,
the kinetic energies of those ions that are measured do not reflect their true energies but
rather the kinetic energy of the predominant bath-gas ion (Ar+). Although such curves do
not reveal the initial energies of heavy (analyte) ions, they are useful in describing overall
ion movement under the influence of an electrostatic field.
Another interesting feature in the kinetic energy plots of Figure 3.10 is that the argon
ion (40 amu) generally has a lower kinetic energy than all the other ions. Although a
mechanism that explains this phenomenon has not been proven, this discrepancy likely
exist because the potential effects are less prevalent in the argon retarding-potential curves
than in those for other analyte ions. As will be shown in Section 3.4.5, the presence of
potential effects at high retarding potentials biases the calculated mean ion kinetic energies
of low mass ions to higher values; the absence of these anomalies in argon curves will
yield a slightly lower computed mean energy.
3.4.3.
Second>Stage Retarding-Plate Analysis
Additional ion kinetic energy curves measured in the second stage were obtained at
various condenser temperatures and first-stage backing pressures while the third-stage
retarding plate was grounded (Figure 3.12). Despite changes in first-stage backing
pressure and solvent load, the retarding-potential curves remained much the same. If
electrostatic forces are primarily responsible for influencing ion movement in this stage, as
106
Chapter 3
Figure 3.12. Second-stage mean ion kinetic energy curves taken at various first-stage
backing pressures and condenser temperatures. Changes in these parameters had little
effect on mean ion energies.
Mean Ion Kinetic Energy (eV)
Condenser Temperature (°C)
5 i
4-
Bi
•—
— ■—
— «—
U — *—
Al Mn
3-
■>-------- 1-------- •--------r
50
100
i-------- •--------1
150
Ion Mass (amu)
20 0
250
-5 (2.6 Torr)
0(1.4 Torr)
5 (2.0 Torr)
10 (1.4 Torr)
Chapter 3
108
discussed above, this independence suggests that the ion-beam space charge in the second
stage is not affected significantly by adjustments in first-stage backing pressure or solvent
load.
Just as in the results presented in Figure 3.10, the mean ion kinetic energy for argon
was found to be lower than that for analyte ions. Mean kinetic energies presented in Figure
3.12 range from 2.3 to 2.6 eV for argon and 2.5 to 3.7 eV for analyte species.
3.4.4. Third-Stage Retarding-Plate Analysis
Mean ion kinetic energies were evaluated also in the third stage at different condenser
temperatures (Figure 3.13) and first-stage backing pressures (Figure 3.14). The apparent
trend is that larger solvent loads increase the offset potential of the ion kinetic energy
curves, whereas higher first-stage backing pressures reduce the effect of gas dynamic
acceleration. The offset potential is the kinetic energy gained from electrostatic forces and
can be deduced by extrapolating the curves such as those in Figure 3.13 to the kinetic
energy at zero mass [10]. For example, the offset potential in Figure 3.13 rose from 1.3 to
2.7 eV when the condenser temperature was increased from 0 to 10°C. This increase in
offset potential with solvent load can be attributed to broadening of the ion kinetic energy
distributions. A greater number of ion-ion interactions among ions or an increase in ion
space charge before the skimmer cone are likely causes for this broadening phenomenon.
A mechanism for this phenomenon is presented in Section 3.4.5.
The third-stage ion kinetic energy curves in Figure 3.14 illustrate the effect of firststage backing pressure at a fixed condenser temperature. For this comparison a condenser
temperature of -5°C was chosen to minimize the influence of solvent loading on offset
potential, as described above. Interestingly, as the first-stage pressure was incremented
from 2.6 to 3.4 Torr the ion kinetic energy curve became virtually flat This change in
109
Chapter 3
Figure 3.13. Third-stage mean ion kinetic energy curves obtained at condenser
temperatures of 0 and 10°C. Increases in solvent load do not have an appreciable effect on
gas dynamic acceleration but do influence offset energy. First-stage backing pressure, 1.4
Torr; Second-third-stage differential pumping aperture, 0.5 mm.
Mean Ion Kinetic Energy (eV)
.
Chapter 3
Condenser Temperature
10°C
Cs
Mn
Ar
0
110
Ion Mhss (amu)
I ll
Chapter 3
Figure 3.14. Third-stage mean ion kinetic energy curves at first-stage backing pressures
of 2.6 Torr and 3.4 Torr. An elevated first-stage backing pressure appears to suppress gas
dynamic acceleration. Condenser temperature, -5°C; Second-third-stage differential
pumping aperture, 0.5 mm.
>
Mean Ion Kinetic Energy
a>
First-Stage Pressure
8
-
Sr
Cs
a 2.6 torr
3.4 torr
Ar
0
50
100 150 200
Ion Mass (amu)
250
112
Chapter 3
113
slope is likely caused by the same electrostatic phenomenon that exists in the second stage
and which was described in Section 3.4.2. In particular, as the first-stage backing pressure
is raised to 3.4 Torr the transmission efficiency in the second stage should increase as
described in Section 3.4.1. As a result, the ion beam will become more concentrated and
coulombic interactions will be enhanced in the third stage. Ions will then respond to the
retarding-potential field collectively (see Figure 3.11), and the effects of gas dynamic
acceleration will be suppressed.
The retarding-potential curves for lithium and uranium that were used to calculate the
data presented in Figure 3.14 yield additional insight into this type of ion-beam dispersion.
In these curves, shown in Figure 3.15, the analyte signal for lithium decreases slightly (by
a factor of two) as the first-stage backing pressure is increased from 2.6 to 3.4 Torr
(compare Figures 3.15a and 3.15c), whereas the uranium signal increases dramatically
(compare Figures 3.15b and 3.15d). This different response is likely attributable to massdependent coulombic dispersion. It would be expected that light ions such as lithium are
ordinarily dispersed evenly throughout the interface because of their poor ion kinetic energy
transfer with the heavier bath-gas (Ar) species. As a result, a larger ion-beam space charge
might disperse these species further but not dramatically. On the contrary, when the ion
beam contains heavy elements such as uranium that would remain more on axis, dispersion
of the ion beam could dramatically lower ion signals at the detector. Beam dispersion of
this type is expected to occur when the first-stage backing pressure is decreased from 3.4 to
2.6 Torr (see Section 3.4.1).
3.4.5. Induced Ion-Ion Interaction in the Mass Spectrometer Third Stage
In the foregoing discussion, it was suggested that ion movement in the second stage is
dominated by coulombic interaction among ions. This interaction is believed to swamp the
114
Chapter 3
Figure 3.15. Third-stage retarding-potential curves taken at first-stage backing pressures
of (a and b) 2.6 and (c and d) 3.4 Torr. These plots were used to construct the mean ion
kinetic energy curves in Figure 3.14. The lithium and uranium curves taken at 2.6 Torr (a
and b) have different shapes, unlike those taken at and 3.4 Torr, (c and d) which have
similar shapes. Condenser temperature, -5°C; Second-third-stage differential pumping
aperture, 0.5 mm.
8.0e+2n
1.Oe+2
O
sr
Lithium (7 amu)
2.6 eV
Signal (cps)
Signal (cps)
6.0e+2-
4.0e+2 -
2.0e+2-
Uranium (238 am u)
7.3 eV
6.0e+1 4.0e+1 2.0e+1 -
O.Oe+O
0
65
■o
“J
8.0e+1 -
m
O.Oe+O
5
10
15
20
Retarding Potential (V)
25
5
m
10
15
20
R etarding Potential (V)
4.0e+2i
3.0e+3
Lithium (7 amu)
2.7 eV
Signal (cps)
Signal (cps)
3.0e+2
2.0e+2
1.0e+2
U ranium (238 am u)
2.5 eV
2.0e+3
1.0e+3
0.0e+0
0.0e+0
0
5
20
Retarding Potential (V)
25
0
5
10
15
20
Retarding Potential (V)
25
in
Chapter 3
116
effect of gas dynamic acceleration in the second- and third-stage ion kinetic energy plots of
Figures 3.10,3.12, and 3.14. To support this hypothesis, the ion flux to the third stage
was intentionally increased by enlarging the diameter of the orifice in the second-stage
retarding plate from 0.5 to 1.0 mm. Of course, this larger orifice also raised the third-stage
backing pressure. For example, for a dry plasma the third-stage pressure was 5.0 x 10-6
Torr with the 1.0-mm-diameter orifice, compared to 5.2 x lO’7 Torr with the 0.5-mm
orifice. Mean energies were deduced from the retarding-potential curves shown in Figures
3.16-3.18 and are compiled along with the mean energies derived from other curves in
Figure 3.19.
A result of intentionally increasing the ion flux through the differential pumping plate
is that the third-stage ion kinetic energy curves resembled the second-stage curves
measured earlier (see Section 3.4.3). Yet, the influence of condenser temperature and firststage backing pressure remained similar to that seen in Section 3.4.4 in the third stage
when the plate with the original 0.5-mm-diameter orifice was used. With the higher thirdstage flux, potential effects such as those noted in the second stage (Figures 3.4 and 3.7a)
became evident in the third stage (see Figure 3.16). These effects signal the interaction of
the ion-beam space charge with the potential fields of the third-stage retarding plate. The
most pronounced example occurs (Figure 3.16) when the condenser temperature is 10°C
and the first-stage backing pressure is 2.3 Torr. The potential effects, which are most
prominent at a retarding-plate voltage of 8 V, biases the mean ion kinetic energy to lower
values for heavy elements and to larger values for lighter elements.
Just as in the measurements reported earlier (Figure 3.7a), third-stage potential effects
are lowest in retarding-potential curves taken at high first-stage backing pressures or high
condenser temperatures (see Figures 3.17 and 3.18). Results from the curves obtained at a
relatively high first-stage pressure (3.2 torr), displayed in Figure 3.17, suggest a collective
117
Chapter 3
Figure 3.16. Third-stage retarding-potential curves for (a) lithium, (b) argon, (c)
manganese, and (d) uranium at a moderate first-stage pressure (2.3 Torr) and condenser
temperature (10°C) with a 1.0-mm orifice leading to the third stage. Measurements were
found to be susceptible to potential effects; however, these effects could be eliminated by
increasing the first-stage backing pressure (see Figure 3.17).
2.0e+3q
1.5e+3
3.0e+5
Lithium (7 amu)
5.1 eV
to
o.
o
Q
O.
(0
c
o>
(0
c
o>
c/5
iio
0
5
15
20
Retarding Potential (V)
10
Argon (40 amu)
3.9 eV
CO
25
1.Oe+5
0.0e+0
0
5
15
20
Retarding Potential (V)
10
25
5.0e+2
1.5e+3
M anganese (55 amu)
5.0 eV
Q.
Uranium (238 amu)
5.5 eV
(0
O)
CO
5.0e+2
0.0e+0
0.0e+0
0
5
15
10
20
Retarding Potential (V)
25
0
5
10
15
20
Retarding P otential (V)
25
00
119
Chapter 3
Figure 3.17. Third-stage retarding-potential curves obtained at a high first-stage
pressure (3.2 Torr) and low condenser temperature (-5°C) for (a) lithium, (b) argon, (c)
manganese, and (d) uranium with a 1.0-mm orifice leading to the third stage. Energy
distributions are similar to those seen in the second stage and indicate that gas dynamic
acceleration is being suppressed.
2.0e+5
1.5e+5
Signal (cps)
Signal (cps)
4.0e+2
Lithium (7 amu)
4.0 eV
3.0e+2
Argon (40 am u)
4.4 eV
1.0e+5
1.0e+2
0.0e+0
0.0e+0
0
5
10
20
15
R etarding Potential (V)
0
25
5
10
15
20
R etarding Potential (V)
25
5.0e+1
2.0e+2 1
Signal (cps)
Signal (cps)
4.0e+1
M anganese (55 am u)
4.2 eV
1.0e+2
2.0e+1
5.0e+1
1.0e+1
0.0e+0
0.0e+0
0
5
10
20
15
Retarding Potential (V)
25
U ranium (238 am u)
4.5 eV
3.0e+1
0
5
10
20
15
Retarding Potential (V)
25
121
Chapter 3
Figure 3.18. Third-stage retarding-potential curves at a low first-stage pressure (1.1
Torr) and a high condenser temperature (20°C) for (a) lithium, (b) argon, (c) manganese,
and (d) uranium with a 1.0-mm orifice leading to the third stage. Rise in off-set potential is
likely caused by a positive ion sheath that forms in front of the skimmer orifice or an
increase in collisional frequency between ions.
Signal (cps)
Signal (cps)
1 .Oe+3 n
Lithium (7 amu)
7.5 eV
6.0e+2
Argon (40 amu)
5.8 eV
5.0e+4'
2.0e+20.0e+0
O.Oe+O
0
5
10
15
20
Retarding Potential (V)
0
25
M anganese (55 am u)
8.4 eV
Signal (cps)
Signal (cps)
10
15
20
R etarding Potential (V)
25
1.0e+3i
1.0e+3n
6.0e+2
5
Uranium (238 am u)
9.8 eV
4.0e+2-
2.0e+20.0e+0
0.0e+0
0
5
10
15
Retarding Potential (V)
20
0
5
10
15
20
R etarding Potential (V)
25
^
to
to
123
Chapter 3
Figure 3.19. Mean ion kinetic energy plots measured in the third stage. The orifice
leading to the third stage was widened from 0.5 to 1.0 mm to increase ion flux to the third
stage. Mean energies resemble those measured in the mass spectrometer second stage yet
seem to be affected more by changes in condenser temperature and first-stage pressure.
Values were taken from retarding-potential plots in Figures 3.16-3.18.
Mean Ion Kinetic Energy (eV)
10
Chapter 3
20°C (1.1
-
M n
10°C (2.3
?V
Ar
200
250
Ion Mass (amu)
124
Chapter 3
125
response among ions to the potential fields of the third-stage retarding plate and follow the
model illustrated in Figure 3.11 for second-stage events (see Section 3.4.2). In contrast,
the main effect of an increase in condenser temperature (to 20°C) is a rise in ion kinetic
energy offset, revealed more clearly in the kinetic energy plots of Figure 3.19. This offset
appears to be caused by broadening of the retarding-potential curves in Figure 3.18
(compare with Figures 3.16 and 3.17).
This broadening might be derived from either of two processes: The first is an
increase in ion-ion collisions, which will randomize ion directions and cause the energy
distributions to be broader than those created by gas-kinetic processes alone. The second is
an increase in ion-beam space charge in the interface. It was reported in Chapter 2 that
higher solvent loads elevated both space and floating potentials in the first stage. This
higher potential might be sufficient to accelerate ions as they pass from the ion space charge
region toward the skimmer cone (held at ground potential) [5]. If ion space charge is lower
in the second stage than the first, these ions will maintain a higher mean velocity and will
yield broader retarding potential curves.
3.4.6. Effect of Load-Coil Configuration on Mean Ion Kinetic Energy
The plasma space potential, which is enhanced by the interaction of the plasma with the
sampling plate, is believed to accelerate ions as they are sampled into the mass spectrometer
first stage [31,32], Earlier studies of this phenomenon have linked the magnitude of this
space potential to the load-coil geometry. Specifically, higher ion kinetic energies have
been measured in ICP-MS systems that employ the asymmetrically grounded, inverted
load coil [9] than in systems that use the center-tapped load coil [10]. However, in recent
studies by Hutton and Eaton [21] and by Lim, Houk, and Crain [23], mean ion kinetic
Chapter 3
126
energies measured in systems that use the inverted-type load coil were found to be similar
to those measured by Fulford and Douglas with a center-tapped load coil [10].
It has been inferred from floating-potential measurements that the asymmetrically
grounded, inverted load coil produces a higher space potential than the center-tapped load
coil [32]. Floating potentials in the asymmetrically grounded ICP were approximately -50
V; as the plasma came in contact with a sampling plate, the potentials increased to +20 V
[31]. Similar trends were found by Houk, Schoer, and Crain with the center-tapped load
coil [32]; however, the jump in floating potential was smaller, from -6V to a range of -2
V to +Q.2V.
This move toward positive floating potentials as the plasma comes in contact with a
sampling plate implies that charge separation is occurring [29]. Space-potential
measurements performed in the mass spectrometer first stage, presented in Chapter 2
confirms this charge separation process. In these Langmuir probe measurements, it was
determined that changes in source and first-stage conditions can significantly alter space
potential in the mass spectrometer, and can thus overcome any effect load-coil geometry
has on space potential. Inadvertent or indirect manipulation of the space potential could
account for the results reported by Hutton and Eaton [21], who were able to alter ion
kinetic energies from 3-5 to 10-12 eV across the mass range simply by increasing the
solvent load. Their results indicate that plasma parameters such as solvent load might be a
more significant influence on ion kinetic energies than load-coil geometry.
To substantiate this hypothesis, ion kinetic energy measurements were performed with
different ICPs coupled to the ICP-MS instrument; one ICP had a center-tapped load coil
and the other an asymmetrically grounded load coil. When similar source and first-stage
conditions were chosen, the mean ion kinetic curves were found not to vary appreciably
(see Figure 3.20). The ion kinetic energy curves in Figure 3.20 were measured in the mass
127
Chapter 3
Figure 3.20. Mean ion kinetic energies measured in the mass spectrometer third stage
for ICPs equipped with the asymmetrically grounded and the center-tapped load coils. The
orifice to the third-stage was 1.0 mm so ion flux to that stage was intentionally high. The
optimal first-stage backing pressure was 2.3 Torr for the inverted-type and 2.1 Torr for the
center-tapped load coil. Condenser temperature, 5°C.
Mean Ion Kinetic Energy (eV)
Chapter 3
Inv erted (2.3 torr)
C en te r-T ap p ed (2.1 torr)
50
100
150
200
250
300
Ion Mass (amu)
128
Chapter 3
129
spectrometer third stage with a 1.0-mm orifice leading to that stage. Curves were not
collected in the mass spectrometer second stage as a result of the high ion-space charge that
was shown above to interfere with the ion kinetic energy measurements.
Under similar source and first-stage conditions the mean ion kinetic energy plots
obtained with the center-tapped and inverted load-coil ICP systems did not differ
appreciably (see Figure 3.20). This similarity suggests that neither gas-kinetic nor
electrostatic forces in the interface is greatly affected by the load-coil geometry. It must
therefore be concluded that the two ICP arrangements yield similar gas temperatures or that
the difference in plasma gas temperature between the two discharges does not influence the
expansion strongly. Moreover, if the two plasmas initially have different densities of
charge carriers (ions and electrons), they must reach a similar stage of charge equilibrium
as they pass through the sampling-plate orifice; that is, a plasma that is initially electron
rich will lose a greater percentage of electrons than a plasma that is electron poor.
Additional support for this hypothesis lies in the fact that ion signals for both systems
maximized at nearly the same first-stage backing pressure when the same condenser
temperature was used. At a condenser temperature of 5°C the ion signal was strongest at a
first-stage backing pressure of 2.1 Torr for the inverted load-coil system and at 2.3 Torr for
the center-tapped load coil system. The fact that ion signal (ion-beam density) was greatest
at similar first-stage backing pressures (implies that ion space charge throughout the ICPMS interface was similar for the two systems. As will be shown in Chapter 4, where the
composition of the ion beam is monitored, ion transmission depends heavily upon the ionbeam space charge. These electrostatic fields are believed to influence ion flux by forming
a virtual ion-optic lens in conjunction with the mass spectrometer interface. In Chapter 4 it
will be shown that first-stage backing pressure in combination with a predetermined solvent
load can maximize ion flux to the third stage.
130
Chapter 3
3.5. CONCLUSION
The theory currently used to describe ion movement through an ICP-MS interface is
covered by Vaughan and Horlick [2]. In this model, the kinedc energy and trajectory that
an ion is expected to gain from gas-kinetic processes is entered into the appropriate lensmodeling functions. The calculated ion-lens fields are then used to describe the focusing of
ions through each stage of the interface. Commonly, this model is modified to take into
account electrostatic forces that might alter ion kinetic energy [2,10,23]. For example, the
offset potential is added to the kinetic energy that an ion gains from gas diffusion forces,
which are determined experimentally (see Sections 3.4.4). This offset potential is the
energy that an ion gains from electrostatic forces and has been suggested by others to be
induced by the plasma space potential [10]. Even though space charge of the ion beam is
ordinarily neglected in these treatments, the merits of taking space charge into account has
been investigated by Gillson et al. [33]. In their study they calculated that the ion fluxthrough a skimmer-cone orifice could conceivably be as high as 1 x 1016 ions s_1, which
likely results in an ion space charge that disperses the ion beam.
Tne approach that has been taken io modify this ion-transport model is to retain the
basic principles of gas dynamic theory but to include specifically any electrostatic forces
produced by ion space charge; especially in the transition- and molecular-flow regions of
the expansion. In truth, there exist two processes that might limit the importance of such
ion-ion interactions: ion-neutral collisions and the shielding of ions by electrons. As the
plasma first passes into the mass spectrometer first stage, both charge separation and a
decrease in gas density occurs (refer to Chapter 2). Although the degree of charge
separation has not been determined, it is expected to be quite large because floating
potentials measured in the first stage range from +2 to +6 V. Shielding between ions is
Chapter 3
131
expected to decrease further as the plasma passes through each successive differentialpumping aperture; eventually an ion beam will form.
The modified ion-transport model proposed in Figure 3.21 compares skimming at the
“optimal” position (Figure 3.21a) and skimming near the Mach disk (Figure 3.21b). When
one skims at the "optimal" position, the gas flux to the second stage has the largest on-axis
directional component; as a result, ion flux to that stage will also be high. As the ionized
gas passes into the second stage, there will be a consequent drop in the concentration of
neutral species and most likely a further preferential loss of electrons over ions at the walls
of the skimmer. Both of these processes reduce shielding among ions and cause ion-ion
interactions to become more significant The ions transported to the second stage will
therefore be dispersed more greatly and a larger fraction of them will be intercepted at the
surface of the differential pumping plate. Downstream in the third stage and at the mass
analyzer, ion flux will consequently be low.
As shown in Section 3.4.1 skimming under the optimal conditions produced the
lowest detected ion signals and the most severe potential effects. However, raising the
first-stage pressure above this optimum (1.1 Torr) allows background gas to penetrate into
the expansion core, a process that can promote neutralization and scattering of the ion
beam. The resulting decrease in ion flux to the second stage, depicted in Figure 3.21b, can
reduce ion-beam dispersion and thereby maintain the on-axis directional component of ions
to the third stage. This behavior agrees with the discussion in Section 3.4.1 in which the
argon signal was found to be greatest at 3.6 Torr. As the pressure was increased further
(to 5.0 Torr) the signal decreased, presumably caused by further attenuation in ion
population because of background scattering and neutralization of the ion beam. The
conclusion from these considerations is that a compromise needs to be achieved for ion and
neutral throughput into the lower pressure stages. This optimum will depend not only on
132
Chapter 3
Figure 3.21. Qualitative ion transport model, (a) Skimming at the gas dynamic optimal-
skimming position and (b) near the Mach disk region. The continuum flow region is
located below the dotted line, where ion movement will be dominated by the flow of neutral
species. Above the dotted line, in the transition- and molecular-flow regions, ion
movement is greatly influenced by coulombic interactions.
133
Chapter 3
0
®
Third Stage
Low Throughput to the Third Stage
®
©
Second Stage
\ Dispersion by lon-lon
/
Interaction
Dense Ion Beam
‘.-S 'S'S
-v*s»s»*
First Stage
cftft#
4ftftft
%'S'S'S
•bWs*s
vyy*.
.s%9“«a\
^■Vy;i
■S"%«V
•s»s«s«
yy>
* /y y .
/y y *
•/y»
i*VV
n
©
~o~
Space-Charge
Dominated Region
■' ■
?■” "
•as*
■ ft.
Collision Dominated Free-Jet
Expansion
134
Chapter 3
b
0
High Throughput to Third Stage
®
Third Stage
©
©
Second Stage
Limited Dispersion
Moderate-Density Ion Beam
First Stage
i
,S
■S■H■S•'
■s/•
■s<
••
^•s■%■s ■s/•
yy
y*/
•&
/V
•O
/ is
s
*
v.
»i*S«\
^y
^y»»
•/yV s* ;,
Space-Charge
Dominated Region
Collision Dominated Free-Jet
Expansion
Chapter 3
135
ion-ion interactions, but also on ion-beam space charge interactions with the interface.
Thus, interface geometry (i.e., the dimensions and orifice diameters of the differentialpumping plates and the ion-optic lens geometry and potentials) will influence the conditions
under which such an optimum occurs.
The qualitative ion transport model presented above not only stands as a reasonable
interpretation of these experimental results but also points the way to experimental
improvements in practical ICP-MS instrumentation. This understanding is being used
currently to eliminate mass-bias effects through suppression of gas dynamic acceleration.
This task is being accomplished by increasing ion flux to the mass spectrometer second and
third stages; an action that should obviate the need for adjustment of ion-optic lens voltages
as a function of ion mass. Additional innovations aimed at the reduction of interelemental
matrix effects have also been investigated [34].
Chapter 3
136
3.6. L i t e r a t u r e C i t e d
[1] D. J. Douglas and J. B. French, J. Anal. At. Spectrom. 3,743 (1988).
[2] M. A. Vaughan and G. Horlick, Ion Lens Simulations o f the Inductively Coupled
Plasma-Mass Spectrometer, submitted to Spectrochim. Acta B, 1990.
[3] J. M. Hayes, Chem. Rev. 87, 745 (1987).
[4] M. Y. Jaffrin, Phys. Fluids 8, 606 (1965).
[5] M. J. Vasile and H. F. Dylla, Plasma Diagnostics, Eds. O. Auciello and D. L.
Flamm, Academic, NY (1989).
[6]
P. S. Wei and A. Kupperman, Rev. Sci. Instrum. 40, 783 (1969).
[7]
B. Rowe, Int. J. Mass Spectrom. Ion Phys. 16, 209 (1975).
[8]
G. N. Spokes and B. E. Evans, Tenth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, PA (1965), p. 639.
[9]
J. A. Olivares and R. S. Houk, Appl. Spectrosc. 39, 1070 (1985).
[10] J. E. Fulford and D. J. Douglas, Appl. Spectrosc. 40, 971 (1986).
[11] D. J. Douglas and J. B. French, Spectrochim Acta 41B, 197 (1985).
[12] J. A. Simpson, Rev. Sci. Instrum. 32, 1283 (1961).
[13] J. A. Dean and T. C. Rains, Atomic Absorption Spectrometry, Ed. J. A. Dean and
J. C. Rains. Dekker, New York (1971), p. 327.
[14] D. A. Wilson, G. H. Vickers, G. M. Hieftje, and A. T. Zander, Spectrochim. Acta
42B , 29 (1987).
[15] P. Yang, B. S. Ross, and G. M. Hieftje, Appl. Spectrosc. 43, 1093 (1989).
[16] B. S. Ross, P. Y. Yang, D. M. Chambers, G. H. Vickers, and G. M. Hieftje,
Comparison of a Center-Tapped and Inverted Load Coilfor Inductively Coupled
Plasma-Mass Spectrometry, submitted to Spectrochim. Acta B, 1990.
[17] D. D. Neiswender and F. C. Kohout, Rev. Sci. Instrum. 43, 1475 (1972).
[18] K. Hiraoka and H. Kamada, Japanese J. Appl. Phys. 10, 339 (1971).
[19] J. L. Franklin, S. A. Studniarz, and P. K. Ghosh, J. Appl. Phys. 52, 3633 (1981).
Chapter 3
137
[20] J. G. Sdguin, C. H. Dugan and J. M. Goodings, Int. J. Mass Spectrom. Ion Phys.
9, 203 (1972).
[21] R. C. Hutton and A. N. Eaton, J. Anal. At. Spectrom. 2 , 595 (1987).
[22] J. S. Crain, R. S. Houk, and F. G. Smith, Spectrochim. Acta 43B, 1355 (1988).
[23] H. B. Lim, R. S. Houk, and J. S. Crain, Spectrochim. Acta 44B, 989 (1989).
[24] A. L. Gray and J. G. Williams, J. Anal. At. Spectrom. 2, 599 (1987).
[25] N. Abuaf, J. B. Anderson, R. P. Andres, J. B. Fenn, D. G. H. Marsden, Science
156, 997 (1967).
[26] P. H. Dawson and J. E. Fulford, Int. J. Mass Spectrom. Ion Phys. 42, 195 (1982).
[27] R. Campargue, Rarefied Gas Dynamics, 6 ^ Symposium. Ed. L Trilling and
H. Y. Wachnam, Academic Press, New York (1969), p. 1003.
[28] M. Haung, D. S. Hanselman, P. Y. Yang, and, G. M. Hieftje, Isocontour Maps of
Electron Temperature, Electron Number Density and Gas Kinetic Temperature in the
Ar ICP Obtained by Laser-Light Thomson and Rayleigh Scattering, submitted to
Spectrochim. Acta B, 1990.
[29] D. M. Manos and H. F. Dylla, Plasma Etching: An Introduction, Eds. D. M. Manos
and D. L. Flamm, Academic, New York (1989).
[30] R. G. Wilson and G. R. Brewer, Ion Beams with Applications to Ion Implantation.
Wiley, New York (1973), chapter 3.
[31] A. L. Gray, R. S. Houk, and J. G. Williams, J. Anal. At. Spectrom. 2, 13 (1987).
[32] R. S. Houk, J. K. Schocr, and J. S. Crain, J. Anal. At. Spectrom. 2, 283 (1987).
[33] G. R. Gillson, D. J. Douglas, J. E. Fulford, K. W. Halligan, and S. D. Tanner,
Anal. Chem. 60, 1472 (1988).
[34] B. S. Ross, D. M. Chambers, and G. M. Hieftje, The Reduction of Spectral and
Non-Spectroscopic Interferences in Inductively Coupled Plasma-Mass Spectrometry,
submitted to J. Am. Soc. Mass Spectrom., 1990.
138
Chapter 4
4
MONITORING THE ION BEAM
4.1. I n t r o d u c t i o n
Although inductively coupled plasma-mass spectrometry (ICP-MS) has become an
effective tool for trace multielemental analysis, it suffers from several shortcomings that
include mass-bias effects, a lack of long-term stability, and matrix-induced interferences
[1,2]. These shortcomings might arise in the discharge or might be the result of sampling
the plasma into the low-pressure environment of a mass spectrometer. However, because
these deficiencies are associated more with ICP-MS than with inductively coupled plasmaoptimal emission spectrometry, they have been perceived to be an artifact of the iontransport mechanisms that exist in the mass-spectrometer interface. Unfortunately, this
perception has not been experimentally confirmed. To determine the cause of these
shortcomings, a fundamental study of ion-transport mechanisms in the ICP-MS interface is
warranted. At this time only a few such studies exist; most of the mechanistic information
that is available has been derived from gas dynamic theory of the skimmed ffee-jet
expansion or from experimental studies that characterize the response of an ICP-MS
system to adjustable parameters [3-5].
Chapter 4
139
The model that is now commonly used to describe ion movement in the interface
consists of two parts; the first uses gas dynamic theory of the skimmed free-jet expansion
to approximate initial ion movement [6] and the second uses ion-optic-lens theory to
describe the focusing of ions toward the quadrupole [7]. Researchers have used these
theories both to guide the design of the interface-sampling configuration [3] and to help
explain results obtained during examination of the sampling process [4,8-13]. In taking
this approach, one assumes inherently that coulombic interaction among charged species
does not interfere with ion movement. This is because gas dynamic theory describes the
movement of only neutral species and not ions. If charged-particle interactions were
significant, a new theory would need to be developed that describes the effects of both gaskinetic and coulombic forces on ion movement Furthermore, ion-optic-lens theory should
take into account ion-beam space charge if such fields are significant
Recent studies have indicated that such coulombic forces indeed play a significant role
in the ion-transport process. A mechanism for ion-beam dispersion in the second stage of a
two-stage system was suggested by Tan and Horlick [4] and based on experiments aimed
at characterizing matrix-induced interference effects. They proposed that ion-beam
dispersion increases with ion number density in the mass spectrometer second stage and
that the space charge created by heavy species forces lighter species off axis. These ionbeam dispersion processes were later supported by ion-trajectory calculations of Gillson et
al. [5] who took into account space-charge dispersion of the ion beam.
The primary focus of this chapter is to identify ion-transport mechanisms in the
interface and to relate these mechanisms to what is currently understood about forces that
influence ion movement. Specifically, gas dynamic relationships that describe a skimmed
ffee-jet expansion in the first stage are initially verified. Results from this study are
combined with those from Chapters 2 and 3 to confirm the ion-transport model that was
140
Chapter 4
presented tentatively in Chapter 3. Results given here support that model and are derived
from experiments where the density and composition of the ion beam were monitored.
These measurements have been found to be a useful diagnostic method to elucidate the role
of electrostatic forces on ion movement.
Specific experiments included measuring the ion-beam density and composition (by
means of mass-spectral scans) as a function of first-stage backing pressure under different
plasma conditions, skimming distances. (6.0 mm and 10.0 mm), and ion-optic-lens
configurations. Results from these experiments serve as an example of how gas dynamic
forces can be overpowered by electrostatic forces in the molecular-flow regions of the
interface.
4 .2 .
Ex p e r im
ental
The ICP-MS system used here is described in connection with the Langmuir probe studies
in Chapter 2. However, the instrument was modified several times for different portions of
the current study. Figure 4.1 depicts one of the instrumental configurations in which the
photon stop was removed and the ion-optic lenses were grounded. In other experiments
the ion-optic lenses were removed and the skimming distance was adjusted. For all
measurements the first-stage pumping configuration and sample introduction system were
as described below.
4.2.1. First-Stage Vacuum System
The first-stage pumping arrangement was configured so a wide range of pressures
could be obtained to intentionally vary the dimensions of the expansion core. The
experimental configuration is described in detail in Chapter 2.
141
Chapter 4
Figure 4.1. One of several instrumental configurations used in this study. In this
particular configuration, the ion optics are grounded and the photon stop is removed. Other
configurations are described in the text.
Chapter 4
Ratemeter
Pre-Amp
Quadrupole
Control Unit
Quadrupole
Power Supply
To Rotary-Vane
Pump (2000 L/min)
Impedance
Matcher
Nebulizer
Peristaltic
Pump
Desolvator
Power Supply
27.12 MHz
Chapter 4
143
4.2.2. Samples and Sample Introduction System
The sample introduction system included a desolvation apparatus that regulated the
amount of aerosol and solvent vapor introduced into the plasma (see Chapter 2). The
solvent load produced by the system was calibrated only once as shown in Figure 2.4.
Although a rise in condenser temperature corresponds to an increase in solvent load, the
absolute values recorded for these solvent loads are expected to vary from run to run and
with experimental conditions. For this reason, solvent load is later described in terms of
condenser temperature.
The chosen elements were essentially monoisotopic and possessed atomic masses that
spanned a broad range. Stock solutions were 1000 p.g/mL Li (Fisher), A1 (Aesar/Johnson
Matthey), Mn (made from Mn metal [14])), Ce (Aldrich), and U (Aldrich), and 2000
|ig/mL Bi (made from Bi metal [14]). Solution concentrations varied but did not exceed 10
ppm, which provided adequate signal magnitudes. A decrease in solution concentration
below 10 ppm did not change any of the results presented below.
4.2.3. First-Stage Skimming
Two separate skimmer cones, each with a 1.0-mm-diameter orifice, were used to adjust
the skimming distance from 10.0 to 6.0 mm. The cone that provided the 10.0-mm
skimming distance had an inner-cone angle of 59.66° and an outer-cone angle of 73.24°.
The 6.0-mm skimming distance was achieved with a longer skimmer cone that had an
inner-cone angle of 68.41° and an outer-cone angle of 62.83°.
4.2.4. Ion-Optic Lens Configurations
Most of the experiments were performed with the second- and third-stage ion-optic
lenses (Figure 4.2) removed. This step eliminated the possibility of any interactions of the
144
Figure 4.2. Second- and third-stage ion-optic lenses.
Chapter 4
145
Chapter 4
Third-Stage Optics
Second-Third-Stage
Differential Pumping
Plate
Wire Mesh
/
Skimmer Cone
Second-Stage Optics
Chapter 4
146
ion-beam space charge with the lenses and resulted in a response that was unbiased by ionoptic lens fields. The interaction of the ground ion-optic lenses with the ion-beam space
charge is demonstrated later in Sec. 4.4.3 of the present paper. Results from experiments
with the ion-optic lenses removed were compared to those with the grounded optics in the
second stage only and to those with the grounded optics in both the second and third
stages. For all the instrumental configurations the photon stop was removed so the ion
beam could be observed on axis.
The second- and third-stage optics shown in Figure 4.2 are cylindrically symmetrical.
The second-stage optics have wire walls fabricated from 0.006-mm-diameter nichrome
wire, whereas the walls of the third-stage optics are solid. A nichrome wire mesh (100
lines per inch with a wire diameter of 0.001 inches) covers the front of the initial third-stage
optical element (Vj).
4.2.5. Plasma-Source and System Operating Parameters
The asymmetrically grounded, inverted load-coil configuration described in Chapters 2
and 3 was used in this study. Given in Table 4.1 are typical operating conditions for the
ICP-MS and sample introduction system. Although the first-stage backing pressure, innergas flow, and solvent load were changed throughout this study, values listed in Table 4.1
exemplify the magnitude of the second- and third-stage pressures, which vary as a function
of these parameters. Other source conditions are as indicated in the table and were not
changed.
147
Chapter 4
Table 4.1. Operating Conditions for the ICP System and Corresponding Pressures in the
Mass Spectrometer Interface
Forward Power
1250 W
Reflected Power
<10 W
Argon Gas Flow (L/min)
Inner
Intermediate
Outer
1.02
1.00
11.0
♦Sample Uptake (mL/min)
0.64
♦Sampling Depth (mm)
10.0
♦First-Stage Pressure (Torr)
1.1
♦Second-Stage Pressure (Torr)
2.0 x 10-3
♦Third-Stage Pressure (Torr)
6.9 x lO'6
Nebulizer
♦Condenser Column (°C)
Heater Column (°C)
200
♦ These parameters were altered for particular experiments. Any different values are cited
in the text
Chapter 4
148
4.3. B a c k g r o u n d
4.3.1. Skimming of the Free-Jet Expansion
The classical skimmed free-jet expansion, similar to the one found in the ICP-MS
interface, has a geometry as drawn in Figure 4.3. As the source gas diffuses into a low
pressure chamber, the directed motion of the expanding gases drives the background
species downstream to create a region known as an isentropic core. This name is
appropriately given to the expansion core because the gas behaves as though it were
expanding into a perfect vacuum. As the gas expands downstream toward the skimmer
orifice, collisions with other expanding species increase the velocity and narrow the
velocity distribution of all the species. However, when the number density of the
expanding gas is insufficient to drive back the background-gas atoms (illustrated as small
open circles in Figure 4.3), these atoms penetrate into the isentropic core and broaden the
velocity distribution of the expanding gas atoms or molecules. Scattering collisions
-
between the expanding and background species produce a shockwave structure (indicated
by the crescent-shaped zone in Figure 4.3) that surrounds the expansion. The dimensions
of the isentropic core are dependent on several parameters, including source pressure (P0),
first-stage backing pressure (Pj), and sampling-plate orifice diameter (D0). The terminal
region of the barrel shock, where the on-axis expanding species meet the shockwave
structure, is referred to as the Mach disk, whose location can be calculated using Equation
2.3 [15].
Upstream from the Mach disk at the back of the isentropic core just before the onset of
background-gas penetration, lies the optimal location to skim the expansion. At this
position, which is described by Equation 2.4 [16], the distribution of velocity vectors for
the expanding species is the narrowest, gas movement along the axis is the most ordered,
and flux at some downstream position will be the largest.
149
Chapter 4
Figure 4.3. Diagram of skimmed ffee-jet expansion in the first stage of an ICP-MS.
Small circles represent background-gas species. The barrel-shock region is confined by the
crescent-shaped section. The area below the dotted line signifies the continuum- and
transition-flow regions of the expansion where ion movement is dominated by neutral
flow. Above the dotted line, coulombic interactions among ions are postulated to be of
increasing significance. Abbreviations; P0, initial source pressure; Pj, first-stage backing
pressure; xs, distance between sampling and skimming orifices; D0, sampling-plate orifice
( tin m p tp r
150
Chapter 4
o
o
°
<fe
Mach Disk
B ackground 3
P en etra tio n
C o u lo m b ic E ffe c ts
In crease
A
x s = 10 or 6 mm
First S tage
Sam pling Plate
D0 = 1.0 mm
C ollision-D om inated
F ree-Jet E xpansion
151
Chapter 4
In ICP-MS one ordinarily tries to situate the skimmer orifice at the optimal skimming
position described by Equation 2.4 to direct the largest on-axis flux of ions to the
downstream mass analyzer. However, this approach might not be the best one as show in
Chapter 3 because ion-ion interactions, which can disperse the ion beam, increase as the
partially ionized gas passes into the lower pressure regions of the interface.
4 .4 .
4 .4 .1 .
RESULTS AND D IS C U S S IO N
Experimental and Calculated Optimal Skimming Conditions
Ion signals obtained from wet and dry plasmas were examined as a function of firststage backing pressure and skimming distance and were compared with results calculated
from gas dynamic theory. In particular, a dry plasma with no inner-gas flow, a dry plasma
with a dry inner-gas flow, and a plasma with different quantities of aerosol combined with
the inner gas were studied to determine if gas dynamic theories could predict the skimming
conditions that yielded the densest ion beam (and the largest resulting signal). The argon
ion (40 amu) was monitored because argon is the predominant bath-gas species and has a
high concentration in all the plasmas studied. Two skimming distances, 10.0 and 6.0 mm,
were used and the first-stage backing pressure was adjusted so the skimming conditions
that produced the largest downstream ion flux could be determined indirectly by moving the
Mach disk back and forth in front of the skimmer orifice.
According to gas dynamic theory, at low first-stage backing pressures the skimmer
orifice is located deep in the isentropic core of the expansion. At this position signal levels
should be affected only slightly by small adjustments in pressure. As the first-stage
backing pressure is then raised, background-gas penetration increases, the ion flux is
scattered, and ion signals should drop. The optimal first-stage backing pressure will then
be defined as that pressure just before the decrease in signal.
152
Chapter 4
To compare experimental and calculated optimal first-stage backing pressures,
knowledge of the Knudsen number (Kn) and the shielding coefficient (Cs) is required. An
accurate value for the shielding coefficient is more critical than is the Knudsen number
because of the cubed relationship between Cs and Pj (see Equation 2.4). Here, the
common literature value for Cs of 0.125 was used [3,16]. However, it must be
remembered that the shielding coefficient depends upon the skimmer-cone and samplingplate arrangement and geometry (e.g., cone angles, skimming distance) and that the correct
value for this ICP-MS might be sightly different.
4.4.I.I.
Dry Plasma. Dry plasmas were examined first because they are closer in
composition to an ideal gas than is a plasma to which aerosol, water vapor, and analyte
have been added. Because gas dynamic theory is technically applicable to the skimming of
only a neutral and pure-gas expansion, the first-stage skimming should therefore be closest
to calculated results when a dry plasma source is sampled. In this section, a dry plasma is
studied with the inner-gas flow turned on and off at two different skimming positions, 6.0
and 10.0 mm. The effect of first-stage backing pressure on argon-ion signal under each set
of conditions is shown in Figures 4.4 and 4.5. If gas dynamic processes solely are
responsible for ion-beam formation and propagation, the first-stage backing pressure
should control ion flux through the interface and ultimately the shape of the curves in
Figures 4.4 and 4.5.
From Equation 2, the largest on-axis ion flux should pass through the skimmer at 5.3
Toit when the skimming distance is 6.0 mm (Figure 4.4); at pressures below this optimum
the ion flux should remain nearly the same. Similarly, at a 10.0-mm skimming distance
(Figure 4.5), the largest on-axis ion flux through the skimmer should occur at first-stage
153
Chapter 4
Figure 4.4. Argon (40 amu) signal versus first-stage backing pressure taken at a 6.0mm skimming distance (a) with no inner-gas flow and (b) with a dry inner-gas flow of
1.02 L/min.
154
Chapter 4
Signal (cps)
5.0e+41
4.0e+4 ’
3.0e+4 -
Flow, Off
2.0e+4
0.0
2.0
4.0
6.0
8.0 10.0 12.0
First-Stage Backing P ressure (Torr)
Signal (cps)
4.0e+6 1
2.0e+6"
1.0e+6Flow, On
0.0e+0
0.0
2.0
4.0
6.0
8.0 10.0 12.0
First-Stage Backing Pressure (Torr)
155
Chapter 4
Figure 4.5. Argon (40 amu) signal versus first-stage backing pressure taken at a 10.0mm skimming distance (a) with no inner-gas flow and (b) with a dry inner-gas flow of
1.02 L/min.
156
Chapter 4
(sdo) teuBis
2.0e+5-
1.0e+5'
Flow, Off
2.0
3.0
4.0
6.0
5.0
First-Stage Backing P ressure (Torr)
(sd0) |Bu6!S
1.0e+7 -
5.0e+6 ~
Flow, On
0.0e+0 +n
1.0
2.0
3.0
4.0
5.0
6.0
First-Stage Backing Pressure (Torr)
Chapter 4
157
backing pressures of 1.1 Tort or less. These values were calculated with an assumed gas
temperature of 5000 K [17] and an argon collisional cross-section of 36.0 A2 [18].
As the first-stage pressure is increased above these theoretical optima, background-gas
penetration moves the Mach disk upstream and in front of the skimmer cone. At pressures
that exceed 9.5 Torr for the 6.0-mm skimming distance and 3.4 Torr for the 10.0-mm
distance (see Equation 2.3), the Mach disk portion of the shockwave structure will be
located at the skimmer-cone orifice position. From these considerations, the analyte signal
should remain constant at a high level at the low-pressure end of the curves shown in
Figures 4.4 and 4.5 until the backing pressure exceeds 5.3 T oit at the 6.0-mm skimming
distance and 1.1 Torr at the 10.0-mm distance. Above these pressures the ion signal
should drop, especially when the skimmer orifice is located behind the Mach disk.
Switching on the inner-gas flow, which should lower the gas temperature of the
source, ought to have little effect on gas-kinetic processes in the interface. A slightly lower
gas temperature would be expected to raise first-stage backing pressure somewhat, but not
to a sufficient degree to alter the dimensions of the expansion (i.e., Mach disk location and
optimal skimming position). For example, the first-stage backing pressure increased by
only 0.1 Torr (from 1.0 to 1.1 Torr) when the inner-gas flow was switched on. The effect
of inner-gas flow on first-stage backing pressure is covered in more detail in Sec. 4.4.4.
At both the 6.0- and 10.0-mm skimming distances, for a dry plasma with no inner-gas
flow, the ion signal appears to behave roughly according to what gas dynamic theory
predicts (Figures 4.4a and 4.5a). At the 6.0-mm skimming distance, shown in Figure
4.4a, the ion signal rises slightly (approximately 19%) from 1.1 to 6.5 Torr, beyond which
point the signal noticeably drops. This drop in signal might correspond to background-gas
penetration at the skimming position, which was calculated above to occur when the firststage backing pressure exceeds 5.3 Torr. At the 10.0-mm skimming distance the argon
Chapter 4
158
signal decreased throughout the observed range as a function of the first-stage backing
pressure (Figure 4.5a). Figure 4.5a is again similar to what one would predict from gas
dynamic theory; a continuous drop in signal as the first-stage backing pressure is raised
above 1.1 Torr. Unfortunately, first-stage backing pressures lower than 1.0 Torr could not
be obtained; it was therefore not possible to determine if the ion signal would remained
constant as the skimmer is moved deeper into the isentropic core of the expansion.
When the inner-gas flow was activated the shapes of the curves in Figures 4.4a and
4.5a changed dramatically, unlike what theory would predict. At the 6.0-mm skimming
distance (Figure 4.4b) the ion signal increased with first-stage backing pressure up to 8.0
Torr, then sharply declined after 10.0 Torr. Similarly, at the 10.0-mm skimming distance
(Figure 4.5b) the signal peaked at 3.6 Torr then decreased as the pressure was raised.
From gas dynamic theory one would expect the low-pressure end of the ion-signal curves
to remain steady until higher pressures forced the onset of background-gas penetration at
the skimming position. Clearly, the rise in signal at the low-pressure end of the curves in
Figures 4.4b and 4.5b is too large to be the result of minor deviations from ideal expansion
behavior such as those caused by viscous effects [19]. In the present case, electrostatic
forces, the only other forces known to influence ion movement in these instruments, are
likely responsible for the observed deviations.
In contrast, the loss in signal seen at the high-pressure end of Figures 4.4b and 4.5b
might be the result of gas-kinetic processes. For example, at a first-stage pressure of 9.5
Torr the Mach disk is calculated to be 6.0 mm downstream from the sampling-plate orifice.
The decrease in signal above 9.0 Torr for the 6.0-mm skimming distance (Figure 4.4b)
could be caused by penetration of background species into the expansion core as the Mach
disk is moved ahead of the skimmer. Similarly, at the 10.0-mm skimming distance the
Chapter 4
159
signal decreased at first-stage backing pressures above 3.6 Torr (Figure 4.5b); the Mach
disk is expected to be in front of the skimmer at pressures over 3.4 Torr.
Another interesting phenomenon is that the argon-ion (40 amu) signal was larger with
the inner-gas flow on than with it off (note vertical scales in Figures 4.4 and 4.5). This
behavior is opposite of what one would initially expect. In a free-flowing ICP, electron or
argon-ion density in the central channel is larger when there is no inner-gas flow [20] (this
assumes that the local ion and electron densities are equal). One explanation for this
apparent discrepancy is that a higher initial ion density leads to greater dispersion of ions
off axis in the low-pressure zones of the instrument. As a plasma or partially ionized gas
passes through a differential pumping aperture to a lower pressure chamber, ion-ion
coulombic interactions will be enhanced by both a decrease in gas density and the
preferential loss of electrons at the orifice (refer to Chapter 2). These interactions will
normally be limited by the presence of electrons, which shield ions from one another [21],
and by the presence of neutrals, which through ion-neutral collisions can inhibit ions from
responding to electrostatic forces [22]. If a gas contains a high percentage of ions before it
passes through a differential pumping aperture, dispersion will begin to occur immediately.
In the mass spectrometer first stage this dispersion could begin in the transition-flow region
where ion-neutral collisional frequency falls off. The resulting strong coulombic
repulsions can cause ions to have unstable on-axis trajectories, so ion flux downstream (at
the skimming position) will drop. Conversely, if a more moderate flux of ions is being
sampled, ion-ion interactions will be less severe, and ions will be able to pass directly from
the continuum- to the molecular-flow regions of the interface and remain directed on axis.
This sort of ion-dispersion process would account for low ion signals when an extremely
high ion population enters a particular vacuum stage and how lowering the initial ion
density might paradoxically increase on-axis ion flux downstream.
Chapter 4
160
4.4.I.2.
Wet Plasma. The effect of first-stage backing pressure and skimming
distance (10.0 and 6.0 mm) on Ar+ signal was measured under a range of solvent loads.
When the solvent was introduced into the plasma the response of the Ar+ signal with firststage backing pressure changed from that seen for the dry plasma at the 6.0- (see Figure
4.4b) and 10.0-mm (see Figure 4.5b) skimming distances. In the wet-plasma behavior
shown in Figure 4.6 for the 10.0-mm skimming distance, ion signals maximized at specific
combinations of solvent load and first-stage backing pressure. The general trend is that
higher solvent loads produced an ion signal that peaked at lower pressures. The
combination of parameters that gave the largest ion signal overall was a first-stage backing
pressure of 1.6 Tore and a condenser temperature of 0°C. Over the course of these studies
these optimal conditions drifted somewhat The best signals were achieved at temperatures
that ranged from 0 to 10°C and first-stage pressures from 1.6 to 2.2 Tore.
A similar relationship among ion signal, first-stage backing pressure, and solvent load
was observed when the 6.0-mm skimming distance was employed (Figure 4.7). The
maximum signal at the 6.0-mm skimming distance appears to fall below a solvent load that
corresponds to a -5°C condenser temperature. (The condenser temperature could not be
brought any lower because the solvent would freeze in the condenser column and would
clog the passageway to the ICP.) Another similarity that can be seen between the results
presented in Figures 4.6 and 4.7 is that the optimal first-stage backing pressures at the 6.0and 10.0-mm skimming distances are within 1.2 Tore of one another when the condenser
temperature is the same. The fact that ion signals could be maximized at nearly the same
pressures for the two skimming distances is surprising because the optimal skimming
distance has a cubed relationship with the first-stage backing pressure (see Equation 2.4).
161
Chapter 4
Figure 4.6. Argon (40 amu) signal versus first-stage backing pressure with a wet innergas flow taken at a 10.0-mm skimming distance. The condenser temperature was adjusted
from (a) -5, (b) 0, (c) 5, (d) 10, and to (e) 15°C to vary the solvent load. The pressures
listed in parentheses pertain to the maximum in the corresponding plot. Note variation in
vertical axis. Inner-gas flow, 1.04L/min.
Chapter 4
162
4.0»t-5
(max. 2.0 Torr)
Condonaar, -5°C
Signal (cpa)
3.004-5
(max. 1.6 Torr)
Condonaar, 0°C
2.004-5
1.004-5
0.00+0
.0
2.0
3.0
4.0
5.0
First-SUgo Backing Prasauro (Torr)
2.0
3.0
4.0
5.0
Firat-Stago Backing Proaaura (Torr)
4.004-5
5.0e+4
Signal (cps)
4.00+4
(max. 1.5 Torr)
Condanaar, 5°C
3.004-5
R- 3.0e+4
2.004-5
§,
1.00+5
2-0e+4
1.0O+4
-n
O.Oo+O
1 0
2.0
3.0
4.0
5.0
Firat-Stago Backing Pressure (Torr)
e
2.00+31
(max. <1.0 Torr)
Condanaar, 15°C
1.50+3
Signal (cps)
(max. 1.6 Torr)
Condanaar, 10°C
1.00+3
5.06+2
i>>i "
O.Oo+O
1 0
2.0
3.0
4.0
5.0
Firat-Stago Backing Proaauro (Torr)
0.00+0
■■i■'
•-1■■
■*T—
■
1.0
2.0
3.0
4.0
5.0
Firat-Stago Backing Proasuro (Torr)
163
Chapter 4
Figure 4.7. Argon (40 amu) signal versus first-stage backing pressure with a wet innergas flow taken at a 6.0-mm skimming distance. The condenser temperature was adjusted
from (a) -5, (b) 0, (c) 5, (d) 10, (e) 20, and to (f) 30°C to vary the solvent load. The
pressures listed in parentheses pertain to the maximum in the corresponding plot. Note
variation in vertical axis. Inner-gas flow, 1.04 L/min.
164
Chapter 4
(max. 3.2 Torr)
C ondanser, -5°C
(max. 1.9 Torr)
C ondsanser, 0°C
Signal (cps)
ao•>
O.O04O .......................................................................
0.0
2.0
4.0
6.0
8.0
10.0
First-Stage Backing Pressure (Torr)
0.0
2.0
4.0
6.0
8.0 10.0
First-Stage Backing Pressure (Torr)
1.0e448.0043-
Signal (cps)
(max. 1.7 Torr)
Condanaar, S°C
(max. 1.3 Torr)
Condanaar, 10°C
g- 6.00431.0e+4
|
35
4.00432.06430.0640 ................................................ ...........
0.0
2.0
4.0
6.0
8.0
10.0
First-Stage Backing Pressure (Torr)
0.0e+0 ................ ......................................................
0.0
2.0
4.0
6.0
8.0 10.0
First-Stage Backing Pressure (Torr)
2.4642
2.00431
Signal (cps)
1.5043-
(max. 1.5 Torr)
Condanaar, 20°C
1.0043-
2.2042
-7
2.0042
-
1.8042
ac
ai
(max. <1.1 Torr)
C ondenser, 30°C
OT 1.6042
5.00421.4042
.
0.0O40
0.0
2.0
4.0
6.0
8.0
10.0
First-Stage Backing Pressure (Torr)
1.2042
0.0
2.0
4.0
6.0
8.0
10.0
First-Stage Backing Pressure (Torr)
Chapter 4
165
Again it must be concluded that gas-kinetic process do not dictate the shapes of these
optimization curves. Instead, it seems likely that the optimal first-stage backing pressure is
dictated by the effect instrumental parameters have on the electrostatic properties of the ion
beam. Langmuir probe experiments presented in Chapter 2 support this hypothesis. In
these earlier investigations it was found that an increase in condenser temperature or firststage backing pressure raised the space potential of the extracted ion beam. To maintain a
constant space potential in the first stage, the backing pressure had to be lowered as the
condenser temperature was raised and vice versa. In turn, the largest ion flux is probably
achieved through the establishment of the proper ion-beam space charge. Although those
conditions may compromise skimming efficiency in the first stage, they favor on-axis ion
movement through the lower pressure zones of the interface.
4.4.2.
Analyte Ion Beam
The first-stage pressures and condenser temperatures that favorably create a dense
argon-ion beam also were found to maximize analyte-ion signals. Because the ion-optic
lenses were not in place to focus the ion beam, sensitivities for analyte ions were
expectedly low. The results presented in this section were collected at a 10.0 mm
skimming distance. Shown in Figure 4.8 is the effect of first-stage pressure and solvent
load on analyte (Mn+) signal. These data were collected in the same day as those for Ar+
presented in Figure 4.6. At a set condenser temperature, the Ar*- and Mn+ curves have
relatively similar shapes and maxima. This similarity is clear evidence that analyte ions
constitute an integral part of the ion beam.
However, elements with low atomic masses were found to behave differently than
heavier species. The term “light elements” here pertains to those that have atomic masses
lower than the predominant bath-gas species, argon (40 amu), whereas “heavy elements”
166
Chapter 4
Figure 4.8. Manganese ion signal versus first-stage backing pressure as the solvent load
is increased. Curves were collected on the same day as those presented in Figure 4.6. The
condenser temperature was adjusted from (a) -5, (b) 0, (c) 5, and to (d) 10°C to vary the
solvent load. The pressures listed in parentheses pertain to the maximum in the
corresponding plot. Skimming distance, 10.0 mm; Inner-gas flow, 1.04 L/min.
(max. 2.0 Torr)
Condenser, -5°C
4.0e+2-
0Q.)
U
3.0e+2cco
o
(max. 1.8 Torr)
Condenser, 0°C
Chapter 4
5.0e+2~i
(0
e
CO
2.0e+2l.0e+2
3.0
4.0
5.0
First-Stage Backing Pressure (Torr)
1.0
3.0
4.0
5.0
First-Stage Backing Pressure (Torr)
2.0
1.0
2.0
5.0e+2'
-CO
ac
jyj
(max. 1.5 Torir)
Condenser, 5°C
3.0e+2‘
2.0e+2'
(max. 1.6 Torr)
C ondenser, 10°C
1.5e+2‘
2.00
3.00
4.00
5.00
First-Stage Backing Pressure (Torr)
167
0.0e+0
1.0
3.0
2.0
4.0
5.0
First-Stage Backing Pressure (Torr)
Signal (cps)
3.0e+2~|
168
Chapter 4
Figure 4.9. Effect of first-stage backing pressure on elemental ion signal for (a) light
elements, Li+ (7 amu) and Al+ (27 amu), and (b) heavy elements, Mn+ (55 amu) and U+
(238 amu), with the ion-optic lenses removed. Condenser temperature, -5°C; Skimming
distance, 10.0 mm; Inner-gas flow, 1.04L/min.
169
Chapter 4
7.0e+3 1
6.0e+3Signal (cps)
5.0e+3 Li (8.2X)
Al
3.0e+3~
2.0e+31.0e+3O.Oe+O -rr
1.0
2.0
3.0
4.0
5.0
First-Stage Backing P ressure (Torr)
Signal (cps)
3.0e+3 1
Mn
U (4.8X)
2.0
3.0
4.0
5.0
First-Stage Backing P ressure (Torr)
Chapter 4
170
will apply to those with atomic masses above argon. Light elements (see Figure 4.9a)
appear not to remain in the ion beam as do heavy elements (see Figure 4.9b). In particular,
signals from light elements were suppressed at first-stage backing pressures where heavy
elements and Ar4- exhibited maxima. The suppression of the light-ion signals (Li+ and Al+)
likely is caused by space-charge-induced scattering of a dense Arf ion beam. Although
such light species will be surrounded and partially constrained by the abundant Ar ions
about them, they will still possess higher translational velocities than the heavier ions. As a
result, they will tend to "rattle around" among the argon ions and migrate to the edges of
the ion beam where they can be lost. Conditions that favor the highest Ar4- density will
therefore produce lower light-ion signals. In contrast, ions heavier than argon do not
migrate to the edges of the ion beam as effectively as do lighter species because they
possess a larger ion kinetic energy than Ar+ and are not readily knocked off axis. In this
situation, conditions that favor the bath-gas ion (Ar4-) will also favor the analyte ion.
Despite the fact that light-ion signals are greatest at conditions different from those at which
Ar+ and heavy-ion signals are maximal, the trends produced by varying solvent load are the
same (Figure 4.10). In particular, the maxima and minima occur at successively lower
first-stage backing pressures as the solvent load is increased.
The behavior shown in Figures 4.9 and 4.10 leads to optimal first-stage pressures that
vary with analyte mass (see Figure 4.11). In Figure 4.11, error bars mark first-stage
backing pressures that brought the ion signal to within 5% of the maximum. Light
elements produce response curves such as those in Figures 4.9a and 4.10 that are bimodal,
with maxima that fall to either side of the optima for heavy elements (Figure 4.9b). As a
result the “optimal” first-stage backing pressures for light elements can fall either above or
below that which is optimal for the heavy elements.
171
Chapter 4
Figure 4.10. Effect of first-stage backing pressure and solvent load on lithium (7 amu)
ion signal. No optics were present in the instrument. Curves were collected on the same
day as those presented in Figure 4.9. The condenser temperature was adjusted from (a) 0,
(b) 5, (c) 10, and to (d) 15°C to vary the solvent load. The pressures listed in parentheses
pertain to the maximum in the corresponding plot. Skimming distance, 10.0 mm;
Sampling depth, 10.0 mm; Inner-gas flow, 1.04L/min.
8.0e+2 n
8.0e+2-i
5.0e+2-
(max. 2.5 Torr)
Condenser, 5°C
6.0e+2
Signal (cps)
g,
7.0e+2
(max. 2.6 Torr)
Condenser, 0°C
7.0e+2
4.0e+2
3.0e+2
4.0e+2-
2.0
3.0
4.0
5.0
First-Stage Backing Pressure (Torr)
2.0
3.0
4.0
5.0
First-Stage Backing Pressure (Torr)
1.0e+3
6.0e+2
(max. 1.9 Torr)
Condenser, 10°C
ft
6.0e+2
2.0e+2
o
4.0e+2
c
3.0e+2
(max. <1.1 Torr)
Condenser, 15°C
1.0e+22.0
3.0
4.0
5.0
First-Stage Backing Pressure (torr)
2.0
3.0
4.0
5.0
First-Stage Backing Pressure (Torr)
173
Chapter 4
Figure 4.11. Dependence of optimal first-stage pressure on ion mass. Ion optics have
been removed. Data obtained from curves such as those in Figures 4.9 and 4.10. The
condenser temperature was adjusted to vary the solvent load. Error bars mark first-stage
backing pressure that brings ion signal to within 5% of the maximum. Skimming distance,
10.0 mm; Inner-gas flow, 1.04 L/min.
o
CO
o
Condenser Temperature (°C)
a>
3
CO 2.5 "
CO
<1)
Q_
Q)
O) 2.0CO
0)
C
1-O
U. 1-5
75
E
Q.
i
B ^ H
"
I
I
I
I
I
1
I
I
I
I
■
-t- o
o
I
o
50
100
I “I 11 I
l
150
200
Ion M ass (amu)
|
I
‘I " I 1 i " 1
250
Chapter 4
175
Best sensitivities for all elements were obtained at condenser temperatures between 0
and 10°C. Shown in Figure 4.12 are the peak signals derived from curves such as those in
Figures 4.6-4.8, for eight elements (50 pM) that span the mass range when the condenser
temperature was 10°C. Listed in parentheses are the first-stage backing pressures at which
the peak signals were obtained. The signal, level for aluminum was nearly seven times
greater than that for most other species. At this time no reason can be given for this
enhancement
4.4.3.
Interaction of Optics with Ion Beam
When ion-optic lenses (but without a photon stop) were placed back into the interface
and held at ground potential, light elements were found to be more effectively retained
within the ion beam. Signal-pressure curves for lithium and uranium taken with the
grounded optics in place are shown in Figure 4.13 and were found to have similar shapes,
unlike those presented in Figure 4.9. Presumably the ion-optic lenses overcame the
suppression of the lithium signal by decreasing the influence of ion-beam space charge and
by reducing the loss of light ions at the edges of the beam.
There are two ways in which the grounded ion optics might negate the effect of the
ion-beam space charge. The cylindrically shaped lens elements in the mass spectrometer
second stage have walls that are constructed of a wire grid to let species pass to the region
outside the lenses. Consequently, off-axis ions that pass beyond the walls of the optics
are isolated from the ion beam by a barrier at ground potential. This isolation of a portion
of the ion beam certainly helps to overcome the influence of ion-beam space charge. In
the mass spectrometer third stage, the lenses are fabricated from solid cylindrical elements.
Ions confined by the solid cylinders can strike their surfaces and become neutralized [23].
176
Chapter 4
Figure 4.12. Maximum signal for elements of different mass. Solution concentrations
were 50 (iM. Data derived from the peak values plotted in Figures 4.9-4.11. No ion-optic
lenses were present in the instrument. Condenser temperature 10°C; Skimming distance,
10.0 mm; Inner-gas flow, 1.04 L/min.
Chapter 4
8.0e+31
Maximum
Signal (cps)
Al (1.90 Torr)
6.0e+3'
4.0e+3 Mn (2.00 Torr)
Bi (2.10 Torr)
2.0e+3 -
Ba, Ce (1.70 Torr)
Li (2.60 Torr)
U (2.00 Torr)
0.0e+0
0
50
100
150
200
250
Ion Mass (amu)
vj
178
Chapter 4
Figure 4.13. Effect of grounded ion-optic lenses on response to light and heavy
elements. Signals are shown forLi+ (7 amu) and U+ (238 amu) with grounded optics in
place as depicted in Figure 4.1. Compare to Figure 4.9 for which ion optics were
removed. Condenser temperature, -5°C; Skimming distance, 10.0 mm; Inner-gas flow,
1.04 L/min.
n
sr
BS
fB
“1
4.0e+2 1
Signal (cps)
3.0e+2
2.0e+2 -
—-
Li (1.4X)
♦—
u
1 .Oe+2 -
O.Oe+O ~i 1 *******1 1 1 1 1 1 1 1 1 1 1 1 *i
1.0
2.0
3.0
I
I
»
I
I "" I
I
\
I
4 .0
I I
1 I T " V '1
■ |
5.0
First-Stage Backing P re ssu re (Torr)
i-*
'
v o
Chapter 4
180
The ion-beam dispersion process was investigated further by placing grounded optics
in only the third stage. Signals were then monitored for both light and heavy species as a
function of first-stage backing pressure. The lithium and uranium curves were again
found to have similar shapes (optima at 1.7 Torr, see Figure 4.14), which suggests that
dispersion of light species is likely most serious in the mass spectrometer third stage. The
ion-transport model presented in Chapter 3 illustrates a reasonable mechanism: In the
mass spectrometer second stage, electron and neutral-atom density might be sufficient to
prevent the preferential migration of light ions off axis and to force ions to respond in a
collective manner. However, in the mass spectrometer third stage, and in the absence of
grounded ion-optic lenses, gas density should be low enough to permit ions to move
independently of each other. In this sort of environment the migration of light species off
axis is more likely when a high-density ion beam is formed. In contrast, when the thirdstage optics are present, both the effect of ion space charge and the preferential migration
of light analyte species would be suppressed.
From the foregoing argument, it might be surmised that ion-beam dispersion would be
significant in analytical ICP-MS instruments where ion-optical lenses are employed.
However, in such instruments, ions are purposely brought to one or more focal points
where ion-beam density is particularly high. In such zones, ion beam dispersion and
mass-selective migration is likely to occur as described above. In turn, such events
promote mass-dependent matrix-induced interferences. This mechanism has been
investigated further in an ICP-MS configured for sample analysis and will be reported
elsewhere [24].
18 1
C h a p te r 4
Figure 4.14. Effect of third-stage lenses on response for light and heavy elements.
Curves are shown for Li+ (7 amu) and U+ (238 amu) with only the grounded third-stage
ion optics in the system; the second-stage optics were removed. Condenser temperature,
5°C; Skimming distance, 10.0 mm; Inner-gas flow, 1.04 L/min.
5.0e+2 ~i
4.0e+2 3.0e+2 2.0e+2 1 .Oe+2 O.Oe+O
1.0
2.0
3.0
4.0
5.
First-Stage Backing P re ssu re (Torr)
Chapter 4
4.4.4.
183
Effect of Inner-Gas Flow Rate on Analyte Signal
Because most ICP-MS instruments have a fixed first-stage backing pressure, it is
attractive to consider using the inner-gas flow rate to optimize the number density in the ion
beam and thereby to minimize coulombic dispersion and interelemental effects. Changes in
the inner-gas flow rate are expected to have several effects; the most obvious is to alter the
source temperature, which can change the gas dynamic characteristics of the expansion. In
addition, changes in inner-gas flow will alter other plasma characteristics (e.g., ionelectron concentrations, spatial distributions) and thus the way in which the plasma
interacts with the metal interface (e.g., to affect charge separation). Because electrostatic
fields were postulated above to be partially responsible for the composition and density of
the extracted ion beam, changes in the inner-gas flow rate might be a straightforward
method to control the ion beam advantageously. Through the following experiments it will
be shown that first-stage backing pressure and inner-gas flow rate can be used
interchangeably to influence analyte-ion signals.
The first experiment was performed to determine the direct effect of central-gas flow on
first-stage backing pressure (see Figure 4.15). For this experiment the inner-gas flow was
initially set to 1.00 L/min and the backing pressure adjusted to a desired value. The
pressure was then monitored as the inner-gas flow was raised from 0.80 to 1.40 L/min. In
compliance with gas dynamic principles, the first-stage backing pressure increased with
inner-gas flow; this trend is caused by lowering of the source temperature, which increases
gas density and consequently the first-stage pressure. The first-stage pressure was
influenced most by the central flow when the initial pressure was set at 4.0 Torr, even
then, the first-stage backing pressure could be raised only from 3.8 to 4.4 Torr when the
inner-gas flow was adjusted from 0.80 to 1.40 L/min. From Equation 2.3, this increase
should cause the Mach disk to move merely 0.70 mm upstream. Clearly, inner-gas flow
184
Chapter 4
Figure 4.15. Direct effect of inner-gas flow rate on first-stage backing pressure. The
first-stage backing pressure was initially set to 1.1,2.0, 3.0, or 4.0 Torr at an inner-gas
flow of 1.00 L/min. Inner-gas flow was then adjusted and the resulting backing pressures
monitored. Condenser temperature, 10°C, Skimming distance, 10.0 mm.
5.001
4.00 Torr
4.00
3.00 Torr
3.00~i rft AAAi c
2.00 Torr
2 .0 0 :
1.10 Torr
| . 0 0 T 1 1 1 1 1 1 1 1"»~"i" ■ » » ■ » i i i i i i i i i i i i i i | i i i i i
0.8
1.0
1.2
1.4
Inner-Gas Flow Rate (L/min)
Chapter 4
186
has only a small effect on first-stage backing pressure and therefore on gas-kinetic
processes.
The effect of inner-gas flow on an analyte (Mn+) signal was first examined at the 10.0mm skimming distance. Figure 4.16a shows how the Mn+ signal varies as a function of
inner-gas flow at first-stage backing pressures of 1.1,2.0, 3.0, and 4.0 Torr, which were
preset when the inner-gas flow was 1.00 LNmin. After each of these pressures was preset
the inner-gas flow was adjusted from 0.90 to 1.10 L/min or beyond. The combination of
conditions that yielded the highest signal was at a first-stage backing pressure of 2.0 Torr
and an inner-gas flow of 1.04 L/min.
The general trend in Figure 4.16a is that as the first-stage backing pressure is raised the
ion signal peaks at lower inner-gas flow rates. This shift is probably not related to changes
in gas-kinetic processes; as discussed above, changes in inner-gas flow alter the
dimensions of the expansion by only a few percent. In addition, the rise in ion signal that
occurs at the low-pressure end of the curves shown in Figure 4.16a cannot be reconciled
with gas-kinetic processes (see Sec. 4.4.1). One must therefore conclude that electrostatic
influences dictate the curve shapes.
Because higher solvent loads lowered the first-stage pressure at which ion signals
peaked (see Sec. 4.4.1.2), it might be expected that the above behavior is caused by the
influence of central-gas flow on solvent load. However, no significant change in solvent
load could be measured over the range of inner-gas flows used here. This lack of effect
might be due to the presence of the desolvation system, which perhaps buffers changes in
solvent load delivered by the nebulizer. Faced with these results, one must conclude that
inner-gas flow influences ion flux independently from the solvent load.
The effect of first-stage backing pressure and inner-gas flow rate observed at the 6.0mm skimming distance (Figure 4.16b) is similar to that seen at the 10.0-mm distance
187
Chapter 4
Figure 4.16. Manganese signal versus inner-gas flow rate taken at a (a) 10.0- and (b) 6mm skimming distance. Each curve was collected at the first-stage backing pressure
indicated in the figure. Condenser Temperature, 10°C.
Chapter 4
188
Signal (cps)
First-Stage Pressure (Torr)
at 1.0 L/min
1.1
2.0
3.0
4.0
1.5e+5
1.Oe+5
0.90
1.00
1.10
Inner-Gas Flow Rate (L/min)
1.20
Signal (cps)
First-Stage Pressure (Torr)
at 1.0 L/min
2.0
3.0
4.0
4.0e+2‘
0.90
1.00
1.10
Inner-Gas Flow Rate (L/min)
1.20
Chapter 4
189
(Figure 4.16a); at higher first-stage backing pressures the ion signal peaked at lower innergas flow rates. The best sensitivity was obtained by first setting the first-stage backing
pressure to 2.0 Torr and then adjusting the inner-gas flow to 1.02 L/min; however, the ion
signal was approximately as large at a first-stage backing pressure of 3.0 Torr and an innergas flow of 1.04 L/min. (Fig 16b).
The fact that inner-gas flow has such a marked effect on ion signals supports the
hypothesis that electrostatic forces such as ion-beam space charge are a key component in
the ion-transport mechanism. It can be safely concluded that parameters such as solvent
load, inner-gas flow rate, and first-stage backing pressure affect ion movement more by
influencing ion-beam formation than through gas-kinetic processes. The relatively small
effect of instrument geometry supports the same hypothesis: Gas-kinetic processes should
be very different at skimming distances of 6.0 and 10.0 mm; yet, optimal inner-gas flow
and first-stage backing pressure combinations were found to be close for the two distances
(see above). A reasonable explanation is that electrostatic fields produced by the ion-beam
space charge, which dominate ion movement, are not significantly altered when the two
different skimmer cones are used.
4.4.5.
Oxides and Doubly Charged Species
Oxide and doubly charged ions were monitored at the 10.0-mm skimming distance to
determine if they constitute an integral part of the ion beam as do their singly charged
atomic-ion counterparts. Oxides were not observable at a 10.0-mm sampling depth (the
"standard" depth for this system [25]) with the ion-optic lenses removed. As a result, the
sampling depth was reduced to 7.0 mm to raise the metal-oxide signal. The effect of firststage backing pressure on Ce+ (Figure 4.17a) and U+ (Figure 4.17b) signals is compared
to their associated metal-oxide ions in Figures 4.17c and 4.17d, respectively. The optimal
190
Chapter 4
Figure 4.17. Comparison of (a) cerium, (b) uranium, (c) cerium oxide, (d) uranium
oxide, and (e) argon (bath gas) ion signal versus first-stage pressure. Condenser
temperature, 5°C; Skimming distance, 10.0 mm; Sampling depth, 7.0 mm; Inner-gas
flow, 1.02 L/min.
Chapter 4
191
a
2.0e+3
1.5e+3
Signal (cps)
U (238 amu)
(max. 2.3 Torr)
Ca (140 amu)
(max. 2.1 Torr)
M
au
1.0e+3
a
ea
0)
5.0e+2
0.0e+0
4.0
2.0
3.0
First-Stage Backing Pressure (Torr)
1 0
2.0
3.0
4.0
First-Staga Backing Pressure (Torr)
6.0e+3
CaO (156 amu)
(max. 2.1 Torr)
Signal (cps)
w
oa .
4.0
2.0
3.0
First-Stage Backing Pressure (Torr)
e
8.0e+3
Argon (40 amu)
(max. 2.2 Torr)
6.0e+3
Signal (cps)
UO (254 amu)
(max. 2.6 Torr)
4.0e+3
2.0e+3
0.0e+0
■•i ■>
’-n
4.0
3.0
2.0
First-Stage Backing Pressure (Torr)
1
3.0
4.0
2.0
First-Stage Backing Pressure (Torr)
192
Chapter 4
first-stage backing pressure, which can be deduced also from the argon curve shown in
Figure 4.17e, is approximately 2.2 Torr. The maxima for the Ce+ and CeO+ signals were
both at 2.1 Torr, whereas those for uranium and uranium oxide were at 2.3 Torr and 2.6
Torr, respectively. At the 10.0-mm skimming distance, the penetration of background gas
is predicted by Equation 2.4 to become significant at pressures above 1.1 Torr and the
Mach disk should form in front of the skimmer at pressures over 3.4 Torr. The similarity
in signal maxima and curve shapes as a function of first-stage backing pressure is evidence
that the penetration of background gas into the expansion core does not noticeably alter the
oxide population.
Signals from doubly charged ions were found to respond in the same manner to firststage backing pressure as those from singly charged species. Plotted in Figure 4.18 are
curves for Ba+ and Ba++ taken at the 10.0-mm sampling depth; the curves have similar
shapes and peak at nearly the same first-stage pressure. Again, the doubly charged ions
appear to constitute an integral part of the ion beam, just as do the oxide ions.
The signals from both oxide and doubly charged species were greatest at those firststage backing pressures that were also most favorable for atomic-ion signals. These
findings imply that the oxide and doubly charged ion populations formed in the plasma
remain relatively unaffected by the sampling process in the interface. Consequently, any
attempt to minimize the incidence of oxide- and doubly charged ions is probably best
directed at the plasma as it passes into the mass spectrometer first stage.
4.5. C o n c l u s i o n
The results presented above suggest that operating parameters such as solvent load,
inner-gas flow rate, and first-stage backing pressure greatly influence coulombic fields in
an ICP-MS interface and therefore ultimately guide ion transport. Gas-kinetic forces, in
193
Chapter 4
Figure 4.18. Comparison of barium and barium doubly charged ion signal as the firststage backing pressure is adjusted. These curves were collected on the same day as the
results presented in Figures 4.6 and 4.8. Condenser temperature,-5°C; Skimming
distance, 10.0 mm; Sampling depth, 10.0 mm; Inner-gas flow, 1.04 L/min.
Chapter 4
3.0e+21
Signal (cps)
Ba
2.0e+2 Ba
++
1.Oe+2 -
O.Oe+O
1.0
2.0
3.0
4.0
5.0
194
First-Stage1 Backing Pressure (Torr)
Chapter 4
195
contrast, appear to play a lesser role in ion-transport processes in the transition- and
molecular-flow regions of the interface. "Optimal" skimming conditions chosen according
to gas dynamic theory (see Equation 2.4) were found in a lens-free ICP-MS system not to
produce the maximum ion signal. Rather, the largest ion signals were achieved under
conditions where ion-beam dispersion is reduced, so the flux that remains on axis is
greatest Variables that can be profitably employed to control ion-beam dispersion are
solvent load, inner-gas flow rate, first-stage backing pressure, and the geometrical features
of the interface (i.e., placement of grounded optics in the interface).
These observations discourage the use of gas dynamic theory to predict ion movement
in the transition- and molecular-flow regions of an ICP-MS interface. However, such
relationships remain useful for describing the continuum flow regions of the expansion and
in the shockwave structure where ion movement is dominated by neutral flow as shown in
Chapter 2.
196
Chapter 4
4.6. L i t e r a t u r e C i t e d
[1] G. M. Hieftje and G. H. Vickers, Anal. Chim. Acta 216, 1 (1989).
[2] D. J. Douglas, Can. J. Spectrosc. 34, 38 (1989).
[3] D. J. Douglas and J. B. French, J. Anal. At. Spectrom. 3,743 (1988).
[4] S. M. Tan and G. Horlick, J. Anal. At. Spectrom. 2,745 (1987).
[5] G. R. Gillson, D. J. Douglas, J. E. Fulfond, K. W. Halligan, and S. D. Tanner,
Anal. Chem. 6 0 , 1472 (1988).
[6]
J. M. Hayes, Chem. Rev. 87, 745 (1987).
[7]
M. A. Vaughan and G. Horlick, Ion Lens Simulations of an Inductively Coupled
Plasma-Mass Spectrometer, submitted to Spectrochim. Acta B, 1990.
[8]
A. L. Gray and A. R. Date, Analyst 108, 1033 (1983).
[9]
J. E. Fulford and D. J. Douglas, Appl. Spectrosc. 40, 971, (1986).
[10] J. A. Olivares and R. S. Houk, Anal. Chem. 57, 2674 (1985).
[11] D. W. Koppenaal and L. F. Quinton, J. Anal. At. Spectrom. 3, 667 (1988).
[12] H. B. Lim, R. S. Houk, and J. S. Crain, Spectrochim. Acta 44B, 989 (1989).
[13] H. B. Lim. and R. S. Houk, Spectrochim. Acta 45B, 453 (1990).
[14] J. A. Dean and T. C. Rains, Atomic Absorption Spectrometry, Eds. J. A. Dean and
J. C. Rains. Dekker, New York (1971), p. 327.
[15] H. Ashkenas and F. S. Sherman, Rarified Gas Dynamics, 4th Symposium. Ed.
J. H. de Leeuw, Vol. II, Academic, New York (1966), p. 89.
[16] R. Campargue, Rarefied Gas Dynamics, 6th Symposium. Ed. L Trilling and
H. Y. Wachnam, Academic, New York (1969), p. 1003.
[17] M. Haung, D. S. Hanselman, P. Y. Yang, and, G. M. Hieftje, Isocontour Maps of
Electron Temperature, Electron Number Density and Gas Kinetic Temperature in the
Ar ICP Obtained by Laser-Light Thomson and Rayleigh Scattering, Spectrochim.
Acta B, submitted in 1990.
[18] P. W. Atkins, Physical Chemistry, Oxford University, Great Britain (1982), p. 873.
[19] H. W. Liepmann and A. Roshko, Elements of Gas Dynamics. Wiley, New York
(1957), chapter 2.
Chapter 4
197
[20] B. L. Caughlin and M. W. Blades, Spectrochim. Acta 42B, 353 (1987).
[21] R. G. Wilson and G. R. Brewer, Ion Beams with Applications to Ion Implantation,
Wiley, New York (1973), chapter 3.
[22] G. N. Spokes and B. E. Evans, Tenth Symposium (International) on Combustion,
Combustion Institute, Pittsburgh (1965), p. 639.
[23] S. A. Cohen, Plasma Etching: An Introduction, Eds. D. M. Manos and D. L.
Flamm, Academic, New York (1989), chapter 3.
[24] B. S. Ross, D. M. Chambers, and G. M. Hieftje, The Reduction o f Spectral and
Non-Spectroscopic Interferences in Inductively Coupled Plasma-Mass Spectrometry,
submitted to J. Am. Soc. Mass Spectrom., 1990.
[25] B. S. Ross, P. Y. Yang, D. M. Chambers, G. H. Vickers, and G. M. Hieftje,
Comparison of a Center-Tapped and Inverted Load Coilfor Inductively Coupled
Plasma-Mass Spectrometry, submitted to Spectrochim. Acta B, 1990.
198
Chapter 5
5
HELIVM PLASMA S O U R C E MASS SPECTROMETRY
5.1. I n t r o d u c t i o n
Argon inductively coupled plasma-mass spectrometry (ICP-MS) is currently one of
the most effective tools for simultaneous elemental analysis [1]. However, in some cases
the method is limited by the use of argon as the plasma gas. For example, the massspectral background produced by the argon plasma is rather complex [2], and hinders the
quantitation of several elements (K, Ca, V, Fe, As, and Se) [3]. Also, detection limits for
elements with high first-ionization potentials (Ru, P, Hg, S, As, Se, Be, Au, Pt, Cd, Zn,
Os, and the halogens) are relatively poor, presumably because the ionization potential of
argon (15.8 eV) does not gready exceed that of these elements [3].
Utilization of a helium plasma might alleviate some of these shortcomings. For
example, helium-related background ions likely will not interfere with the detection of
many analyte ions. Additionally, the high first-ionization potential of helium (24.5 eV),
which is over twice that of most halogens, should lead to greater ionization efficiency and
improved detection limits for elements with large ionization potentials. Recognizing these
advantages, numerous researchers have coupled helium plasmas to mass spectrometers.
Chapter 5
199
Unfortunately, development of such systems has been slow and success has been mixed
[4-9].
In 1987 Satzger et al. [5] and Montaser et al. [4] were among the first to couple a
helium plasma source to a mass spectrometer (a microwave-induced plasma (MIP) and a
helium inductively coupled plasma, respectively) for elemental analysis. Both groups
used mass spectrometer interfaces that were originally designed to sample an argon bath
gas. Although the detection limits were promising, the analyses were impaired by a
complex mass-spectral background, presumably from air entrainment.
Koppenaal and Quinton [3] were the first to address the necessity for modification of
the mass spectrometer interface for sampling a helium plasma. They reported the
difficulties encountered when a helium ICP was coupled to a conventional ICP-MS
interface, the most important of which was the adaptation of the mass spectrometer for
sampling a helium bath gas. Although no modifications were made, they concluded that
the sampling-plate orifice, skimming distance, and first-stage backing pressure had to be
altered to sample the helium plasma correctly.
The most recent work in this area was by Creed et al. [8], who described the
determination of selected metals and nonmetals by helium MIP-MS. Despite the high
first-ionization potential of helium, the detection limits reported for helium MIP-MS were
not significantly better than those obtained with an argon ICP-MS. For the most part, the
disappointing performance of the helium MIP-MS instrument was attributed to
interference by an intense and complex mass-spectral background.
In our laboratory similar attempts have been made to couple a helium MIP to an
interface designed originally for sampling an argon bath gas [9]. Results from these
studies were similar to those previously reported by Creed et al. [8]. In our preliminary
investigations, background interferences were found to hinder the detection of many
Chapter 5
200
elements. Also, mass-bias effects were found to be much greater in the helium MIP-MS
than in the argon ICP-MS; as a result, the MIP-MS required large adjustments in ionoptic-lens potentials to maximize ion signals over the mass range.
In the present study, ion-transport processes will be examined in a helium MIP-MS
by using many of the same diagnostic techniques for the argon ICP-MS described earlier
in Chapters 2-4. Because many of the processes that result in ion-beam formation should
be similar between the argon ICP-MS and the helium MIP-MS, it is possible to use the
ion-transport model described in Chapter 3 to explain the experimental results presented
here.
To sample the helium plasma, the first stage of a three-stage argon ICP-MS was re­
configured. Gas dynamic relationships that describe the sampling and free-jet skimming
of a gas served as an initial reference for the first-stage modifications [10,11]. Even
though this theory was meant to describe the movement of neutral species, it is well
established that in the continuum-flow region and in the shockwave structure of the
expansion, ion movement can be described by neutral flow [12,13]. However, in the
transition- and molecular-flow regions of the expansion and in the lower pressure regions
in the interface, ion-neutral collisions decrease. In these regions, ion-ion interactions
become increasingly important, and cause ion flow to deviate from that of the neutral
species [14]. The extent to which gas-kinetic processes and coulombic interactions
influence ion movement will be determined in this study. The techniques that will be
employed to carry out this fundamental investigation include ion kinetic energy
measurements in the mass spectrometer second stage (see Chapter 3) and the examination
of the density and composition of the ion beam (see Chapter 4). Results from these
201
Chapter 5
experiments will clarify what limitations can be expected when helium is used instead of
argon as the expansion gas in plasma source-mass spectrometry (PS-MS).
5.2. E x p e r i m e n t a l
5.2.1. Ionization Source
The helium MIP-MS instrument used in these experiments was adapted from an argon
ICP-MS, that has been described in Chapter 3. As part of the conversion, the argon ICP
was replaced by a Beenakker-type MIP, which was similar to that described by
Michlewicz and Carnahan [15]. The only modification made to this MIP arrangement was
to place the tuning stubs on the back of the microwave cavity to provide clear access to the
plasma. A Micro-Now generator (Model 420B1) was used to power the MIP. Operating
parameters for both the source and mass spectrometer interface, listed in Table 5.1, were
chosen to minimize analyte signal levels.
5.2.2. First-Stage Vacuum System
The first-stage pumping arrangement was configured to achieve the lowest possible
backing pressure. To accomplish this goal both pumping speed and vacuum-line
conductance in the first-stage were maximized; additional vacuum lines were linked from
the first stage to the first-stage rotary-vane pump, which had a pumping speed of 2000
L/min (air at standard temperature and pressure(STP)). The pumping capacity of this
system is over twice that commonly found in commercial instruments. A full description
of the first-stage vacuum system used here given in Chapter 3.
Chapter 5
202
5.2.3. Sampling-Plate Material
In preliminary experiments, a copper sampling plate was used. However, the
discharge between the sampling plate and MIP, described in more detail later, was found
to erode the orifice quickly; an orifice with a 0.4-mm diameter eroded to 0.8 mm in
approximately nine hours. To avoid this problem, the sampling-plate material was
switched to aluminum, which did not erode. The durability of the aluminum sampling
plate might be the result of a buildup of aluminum oxide on the surface of the sampling
plate. This oxide layer likely shields the walls of the sampling aperture from the plasma
constituents.
5.2.4. Sample Introduction System
The sample introduction system consisted of a MAK nebulizer and spray chamber
designed for helium operation (Sherritt Gordon) and was coupled to a desolvation system.
A peristaltic pump was used to regulate the amount of solvent introduced into the
nebulizer. The analyte-containing aerosol produced by the nebulizer was then passed
through a desolvation system similar to the one described in Chapter 2 and directed into
the central channel of the torch. Operating conditions for the condenser and heater column
of the desolvator are listed in Table 5.1. The solvent load delivered by this sample
introduction system was determined by trapping and weighing the aerosol and water vapor
in a silica gel tube (refer to Chapter 2).
Because the absence of ion-optic lenses reduced ion-transport efficiency through the
interface, high analyte concentrations were used to boost ion signals. Stock solutions
were of 1000 ppm Li (Fisher), Co (Fisher), Zn (Morton Thiokol), Mn (made from Mn
metal [16]), Sr (Morton Thiokol), Ba (Morton Thiokol), Ce (Aldrich), and U (Aldrich).
The most abundant isotope for each of the elemental ions was monitored.
Chapter 5
203
Table 5.1.
Typical Operating Conditions for the Beenakker Type Cavity [15] MIP-MS
Forward Power (W)
400
Gas Flow Rates (L/min)
Inner
0.4
Outer
20.7
Sampling Depth (mm)
4*
Sampling-Plate Orifice Diameter (mm)
0.4
Skimmer-Cone Orifice Diameter (mm)
1.0
Sample Uptake Rate (mL/min)
0.64
Pressure
First Stage (Torr)
Second Stage (Torr)
Third Stage (Torr)
0 . 66**
1.1 x 10'3**
6.0 x 10’5**
Desolvator
Heater Column (°C)
Condenser Column (°C)
Solvent Load (mg/min)
Nebulizer Type
20
7
MAK
* Measured from the top of the MIP cavity to the sampling-plate orifice.
** Pressure calibration assumes pure helium.
Chapter 5
204
5.2.5. Retarding-Plate Analyses
The instrumental configuration used to measure ion kinetic energies here is similar to
the one described in Chapter 3 where ion kinetic energies were measured in an argon ICPMS. All the ion-optic lenses were removed from the interface both to make room for the
retarding plate and to reduce the likelihood of “potential effects” (see Section 5.2.6) that
can complicate the measurements [17,18]. Kinetic energies were determined by way of
coulombic repulsion using the retaiding-plate potential. Ions that had sufficient energy to
overcome the potential barrier of the retarding plate passed directly through the plate
orifice toward the quadrupole. Those ions able to pass through the quadrupole were
detected with a discrete-dynode secondary electron multiplier in the ion-counting mode
(refer to Chapter 3). Ion signals were recorded in counts per second (cps).
Kinetic energy analyses were performed only in the mass spectrometer second stage
because placement of the third-stage retarding plate attenuated the ion flux below a
detectable limit The second-stage retarding plate had a 0.5-mm-diameter orifice and was
positioned 61.5 mm behind the skimmer-cone orifice as described in Chapter 3. The plate
voltage was ramped from 0 to +20 V at 0.2 V/s. As the potential of the retarding-plate
was increased, the ion signal was recorded. Plots of ion signal versus retarding-plate
potential will be referred to here as “retarding-potential curves”. Mean ion kinetic energies
were deduced from these retarding-potential curves for various elements in the same
manner as described Chapter 3 and were plotted as "ion kinetic energy curves". The
retarding-potential curves shown in this chapter have not been smoothed or altered
intentionally to demonstrate signal magnitude and quality.
As a result of plasma instabilities, which are discussed in more detail in Section
5.3.4.1, all retarding-plate analyses were obtained within a 30-minute time window. This
narrow window required the number of elements and sample repetitions to be limited.
Chapter 5
205
Retarding-potential curves were collected for six to nine elements with one-to-three
repetitions of each element In addition, the elements were ordered randomly so that
experimental drift did not influence the results. For the entire set of runs the best standard
deviation was no smaller than 0.2 V and the worst was no greater than 0.6 V.
5.2.6. Potential Effects
In the ion kinetic energy experiments performed on the argon ICP-MS (see Chapter
3), it was found that solution concentrations over 200 |ig/mL produced "potential effects".
These effects arise when the space potential generated by a dense ion beam distorts the
retarding-plate potential fields [17,18]; the result is a somewhat misshapen retardingpotential curve and a skew in the calculated mean ion energy. Understandably, this
phenomenon can be induced when high solution concentrations are used. Conveniently,
potential effects appear to be absent in the helium PS-MS; as shown in Figure 5.1,
solution concentrations up to 1000 ppm did not have a significant effect on the retardingpotential curves.
5.2.7. First-Stage Pressure Measurements
Pressure in the first stage was monitored with a Pirani gauge (Balzers, Model TPR
035). A capacitance manometer would have been preferred because it requires little
calibration when one measures the pressure of different gases. Unfortunately, a
capacitance manometer could not be used because its controller displayed unstable and
erratic pressure readings as the MIP discharge came in contact with the sampling plate.
The Pirani gauge system, in contrast, responded normally and indicated stable pressures
that were within the range predicted by gas-kinetic theory [19]. The Pirani pressure gauge
was placed approximately 25 cm from the centerline axis of the expansion to insure an
206
Chapter 5
Figure 5.1. Second-stage retarding-potential curves for helium while (a) water, (b) 1000
ppm lithium (7 amu), (c) 1000 ppm barium (138 amu), and (d) 1000 ppm uranium (238
amu) were introduced. High solution concentrations have little effect on retarding-potential
curves. First-stage backing pressure, 0.80 Torr.
Q.
Signal (cps)
2.0e+5n
Helium
1.2 eV
No Analyte
1.0e+5O)
0.0e+0
0
5
10
15
1.Oe+5 ■
0.0e+0
20
0
Retarding Potential (V)
2.0e+5*
Helium
1.0 eV
Li Analyte
1.0e+5‘
0.0e+0
0
10
15
20
1.2e+5i
Signal (cps)
o>
5
Retarding Potential (V)
3.0e+5-i
o.
Helium
1.3 eV
Ba Analyte
5
10
15
Retarding Potential (V)
20
8.0e+4Helium
1.2 eV
U Analyte
4.0e+4-
0.0e+0
0
5
10
15
20
Retarding Potential (V)
K>
O
208
Chapter 5
accurate backing-pressure reading. Because the pressure controller was calibrated for air,
readings were converted to the corresponding helium value with the conversion
information provided by the manufacturer.
5 .3 .
Results
and
D is c u s sio n
A fiee-jet expansion such as the one found in an ICP-MS interface has a geometry
(see Figure 2.6) that has been verified experimentally [20]. In the isentropic core of the
expansion exist the continuum-, intermediate-, and molecular-flow regions in which the
collisional frequency among gas species is described as being high, low, and negligible,
respectively. In the continuum- and intermediate-flow zones, collisions among
expanding species increase the gas-stream velocity along the centerline axis. These
collisions convert the random kinetic energy of gas species into directed movement on
axis. At the same time, there occurs a progressive decrease in gas density. However, at
some point in the expansion a shockwave structure forms where the background species
begin to penetrate significantly into the expansion core. In this structure, collisions
between the expanding gas and background species increase and cause scattering of the
neutral flow. The shockwave structure that forms perpendicular to the expansion stream
is called the Mach disk, whereas the portion that forms parallel to the expansion stream is
known as the barrel shock.
Inside the barrel shock is the isentropic part of the expansion in which gas species
expand as though they are moving into a perfect vacuum. The length and diameter of this
expansion core is dependent on the first-stage backing pressure. At a location in the
isentropic core of the expansion, just before the onset of background-gas penetration, is
situated the "optimal skimming position". At this position the largest portion of the gas
Chapter 5
209
flux that passes through the skimmer-cone orifice remains on axis. Properly, the
empirical relationships that specify this position apply to the movement of only neutral gas
species. To adapt this theory for ion skimming, it has to be assumed that ion movement
can be approximated by the motion of neutral bath-gas atoms.
In Chapter 2 the electrostatic characteristics of the first-stage expansion were mapped
and it was determined that the expansion of a partially ionized gas could be described
adequately by gas dynamic theory. In those regions where ion-neutral collisions are
high—in the continuum flow region and the shockwave structure—gas dynamic forces
likely influence ion movement to a greater extent than electrostatic ones. Strong
electrostatic fields resulting from sheath formation [13], an orifice-linked discharge [1], or
plasma potential [21-23] should therefore have a minimal effect on ion movement in these
regions. For this reason ion movement near the sampling-plate orifice is best described by
gas dynamic theory.
By contrast, in the transition- and molecular-flow regions of the expansion and at
pressures where the collisional frequency among ions and neutrals is low, ion mobility is
expected to be influenced more by coulombic forces than gas-kinetic forces. In the argon
ICP-MS the sampling process was found to be complicated by ion-ion interactions,
which altered ion movement from what gas dynamic theory would predict (refer to
Chapters 3 and 4). In turn, these interactions were induced by processes such as charge
separation and the reduction in gas density that occurs as the partially ionized gas is
sampled through the interface (see Chapter 2).
In Sections 5.3.1 and 5.3.2 the classical description of gas dynamic theory will be
applied to two aspects of the first-stage sampling process: regulating gas flow through the
sampling-plate orifice and skimming of the free-jet expansion. This theory has served as a
guide for designing an improved mass spectrometer interface for the helium MIP-MS; in
Chapter 5
210
truth, however, the adopted design has been compromised practically by the difficulty of
pumping helium as a bath gas.
5.3.1. Selecting the Sampling-Plate Orifice Diameter
The most critical dimension in the mass spectrometer interface is the size of the
sampling-plate orifice. Because this diameter influences the degree of boundary layer
formation [24], the amount of air entrainment, and the required pumping speed, a
compromise must be achieved for proper sampling. For example, a small sampling-plate
orifice can reduce pumping requirements in the interface and the degree of air entrainment;
however, boundary layer formation will become more significant In turn, a thick
boundary layer over the sampling-plate orifice will promote both ion neutralization and
cluster formation [24]. Optimally, a sampling-orifice diameter could be determined that
would simultaneously minimize boundary layer formation, eliminate air entrainment, and
maintain pumping speeds at a moderate level. To judge whether such a compromise
exists, let us calculate the influence of the orifice diameter on these variables.
The boundary layer thickness is dependent only on the size of the orifice and the mean
free path of the gas atoms. A measure of the boundary layer thickness—the Knudsen
number (Kn)—is defined as the ratio of the mean free path of the gas atoms (k) to the
orifice diameter (Do). A generally accepted compromise involves setting this ratio to a
value less than 0.01 [24]. If this Knudsen number is to be maintained for a 2000 K [25]
helium MIP, the orifice diameter should not fall below 0.13 mm. This value was derived
from a helium-atom mean free path of 1.30 x 10*3 mm and a collisional cross-section of
0.148 nm2 [26].
211
Chapter 5
Air entrainment can be minimized by insuring that the plasma-gas flow rate exceeds
the gas flow through the sampling-plate orifice. The flow rate (Uv) through this orifice
(of diameter D0) can be calculated from
„
_ ltf(7)NAD§P0
K ,,
UV -
( 5 ' 1)
where Na is the Avogadro number, P0 is the source pressure (one atmosphere here), mm
is the molar mass of the sampled gas, R is the gas constant, T0 is the source temperature,
and the function f(y) is calculated by
f(Y)=Y1/2( ^ _ j 2(y-l)
(5.2)
where y is the ratio of the heat capacities (Cp/Cv) for the gas [24].
It is apparent from this relationship that gas flow through the sampling plate is higher
for helium than argon. Gas flow rates through sampling-plate orifices that range in
diameter from 0.1 to 2.0 mm were calculated for a 2000 K helium source and a 5000 K
argon source and are displayed in Figure 5.2 to illustrate the great difference between the
two gases. The reported gas flux has been converted to STP.
A high plasma-gas flow rate (21.1 L/min total) was adopted for the helium MIP in an
effort to reduce air penetration into the plasma and thereby enable the largest possible
sampling-plate aperture to be used. According to Equation 5.1 and Figure 5.2, orifice
diameters greater than 1.4 mm will transmit more than the plasma-gas flow rate;
accordingly, aperture diameters of 0.1 to 1.0 mm were tested. Not surprisingly, because
helium is much lighter than atmospheric gases, air entrainment occurs readily as the
212
Chapter 5
Figure 5.2. Calculated helium and argon gas flow rate through the sampling-plate orifice
as a function of orifice diameter at a gas temperature of 2000 K and 5000 K, respectively.
Gas flux has been converted to STP. Source Pressure, 760 Torr. Values computed from
Equation 5.1.
213
Chapter 5
o
o
o
CM
O
o■
o°
r“
co
o
o
cm
(ujiu/i) ajeu m o |j
Orifice
o
o
o
Diameter (mm)
O
Chapter 5
214
plasma gas passes from the torch to the sampling aperture. As a result, diameters near 0.1
mm reduced air entrainment most effectively. Unfortunately, such small orifices tended to
clog when high analyte concentrations were introduced into the system. A diameter of
0.4-mm was found to offer an acceptable compromise to eliminate both air entrainment
and orifice clogging. For this orifice diameter the calculated helium gas flow rate is 1.7
L/min, well below the total gas flow for the torch. Background scans for both a dry and
wet plasma will be presented in Section 5.3.4.1 and exhibit no signs of air entrainment.
5.3.2. Skimming the Free-Jet Expansion
The ideal skimming position is located as far back from the sampling-plate orifice as
possible, yet at a position where background gas has not penetrated significantly into the
expansion core. At this position one is assured of the highest Mach number so the largest
flux of gas species remains on axis. These optimal skimming conditions can be
determined from Equation 2.4 [10].
A plot of first-stage backing pressure needed to meet the requirements for optimal
skimming are plotted as a function of skimming distance in Figure 5.3. For this
calculation a sampling-plate with a 0.4-mm diameter orifice and a Knudsen number of 3.2
x 10‘3 were assumed. From Equation 2.4 and Figure 5.3, skimming distances of 12.0 to
4.0 mm would require first-stage backing pressures of 0.03 to 0.46 Torr, respectively. A
4.0-mm skimming distance was chosen as a lower limit because this distance
approximately marks the beginning of the transition-flow region of the expansion.
Unfortunately, it was not possible to obtain these low pressures despite the
incorporation of an improved vacuum system, which included a high-capacity pump and
multiple vacuum lines. In addition, attempts to move the skimmer cone closer to the
215
Chapter 5
Figure 5.3. First-stage backing pressure calculated as a function of skimming distance
needed to optimize skimming for a helium bath gas. Sampling-plate orifice diameter, 0.4
mm; Source Temperature, 2000 K; Knudsen number, 3.2 x 10-3; Source pressure, 760
Torr. Values derived from Equation 2.4.
First-Stage Backing
Pressure (Torr)
Chapter 5
4
5
6
7
8
9
10
11
12
Skimming Distance (mm)
216
217
Chapter 5
sampling-plate orifice than 4 mm failed because pressures then became too high to operate
the turbomolecular pumps in the second and third stages. Under the best vacuum
conditions, the pressures in both the second and third stages were barely within an
acceptable operating range, 1.3 x 10 3 and 6.6 x 10'5 Torr, respectively. These pressures
were achieved with a 10.0-mm skimming distance and a first-stage backing pressure of
0.66 Torr (the lowest that was obtained).
At this practically limited first-stage backing pressure, it is clearly impractical to skim
at the optimal distance (see Figure 5.3). More importantly, it is equally unrealistic to skim
in front of the Mach disk, which is calculated to be 9.1 mm downstream from the
sampling-plate orifice as determined from Equation 2.3 [11]. This Mach disk location is
p
independent of y for 15 <
> 17,000 [11]. To move the Mach disk to the 10.0-mm
skimming position the first-stage backing pressure would need to be 0.55 Torr.
Skimming behind the Mach disk generates a few repercussions. First, any species
skimmed behind the Mach disk will have more randomized trajectories, broader ion kinetic
energy distributions, and lower mean energies than those that exist at the optimal
skimming position. The drawback of sampling such species is that they will have a
greater off-axis trajectory component, which will decrease the flux directed toward the
mass analyzer. Second, collisions between ions and background gases might either
increase the degree of cluster-ion formation and thereby clutter the background spectrum
or promote dissociation of polyatomics.
The severity of these anticipated problems was assessed by intentionally introducing
air into the helium plasma while the interface was held at two relatively high backing
pressures (0.77 and 1.32 Torr). The degree of cluster-ion formation and the loss in signal
at the different backing pressures could then be determined from the resulting background
spectra. Ratios of the ion signals measured at the two pressures are compiled in Table
Chapter 5
218
5.2. As the first-stage backing pressure was raised, all the ion signals decreased as
expected; generally by a factor that ranged from 30 to 40. However, the influence of
first-stage pressure on polyatomic ion signals depended on the amount of air that was
introduced. At the low air flow rate there occurred a preferential loss in ion signal for
lighter species (i.e., 4He+, 8He2+, and 14N+). A drop in ion signal occurred also at the
higher air flow for 4He+, which was lost altogether, and for 8He2+. This loss of lighter
species is likely the result of collisions with background-gas species, which can deflect
light ions off-axis. The argon-ion (40 amu) signal was the least affected by the increase in
first-stage backing pressure. The relatively large atomic mass of argon probably permits it
to remain on axis after sustaining a collision with a bath-gas (He) particle. In addition, the
argon atom has a smaller diameter than the polyatomic ions and will have a greater chance
of passing to the skimmer cone without having undergone a collision with a neutral atom
or molecule.
Interestingly, signals for polyatomic species behaved differently at the two air flow
rates. At low air flow rates the loss in signal for polyatomic ions—28N2+, 30NO+, and
3202+—was no greater than for most of the other species. However, at high flow rates,
signals for these ions decreased substantially (see Table 5.2). This difference is probably
related to the proportion of air in the background gas. At high air flow rates, a greater
fraction of the collisions that are sustained by polyatomic ions are with the relatively
massive constituents of air (N2, N, 02,0), which encourage fragmentation and
neutralization. In contrast, when the fraction of air in the background gas is low, the
principal collisions will be with He, which will be less efficient at promoting
fragmentation. One initially might think that this fragmentation of polyatomic ions would
reduce the loss seen for the monatomic and smaller polyatomic ions; however, ion
219
Chapter 5
Table 5.2. Ratio of Background Ion Signals Collected at a First-Stage Backing Pressure of
0.71 Torr to those Collected at 1.32 Torr
Inner-Gas Flow
Presumed
Species
vajz
22 L/min Air
60 L/min Air
He+
4
120
not observable
He2+
N+
NH+
0+
OH+
h 2o +
h 3o +
n 2+
NO+
o 2+
Ar*
8
14
15
16
17
18
19
28
30
32
40
77
71
43
53
43
40
not observable
87
56
46
39
43
51
40
74
133
253
19
56
43
37
29
Chapter 5
220
fragments formed by collisions with background species will generally be directed off axis
and not detected.
From these results it would appear that the ion-beam composition traveling out of the
expansion core would remain unchanged even beyond the Mach disk if the background
gas consisted mainly of helium (similar to the low air flow case above). Under such
conditions, only those ions that have masses near that of the bath gas (i.e., 4He+, 8He2+,
and 14N+) would be affected. It would therefore seem to be all the more important to
minimize air entrainment into the He plasma and, to the greatest extent possible, remove
solvent from the introduced sample material.
5.3.3. Ion Kinetic Energy Measurements in the Second Stage
Ion kinetic energy measurements serve as a useful tool to deduce the relative influence
of gas-kinetic and electrostatic forces on ion movement. Energy not attributable to gaskinetic processes is presumably gained from electrostatic forces. Previous studies, which
involved measuring ion kinetic energies in the argon ICP-MS, revealed that a substantial
fraction of the total energy of an ion arises from electrostatic forces that emanate from the
ion beam itself (refer to Chapter 3). In those experiments, kinetic energies were
monitored in both the second and third stages of the interface. Mean ion kinetic energies
in the second stage were similar for elements of different masses and resulted in an ion
kinetic energy curve with a slope of approximately zero. This zero slope is believed to be
caused by a strong space-charge field (created by the ion beam) that overpowers the
influence of gas-kinetic forces on ion movement. In reality, ions in such a dense beam
might have different kinetic energies because they are traveling at the same velocity;
however, coulombic interactions among the ions force them to respond in a collective
Chapter 5
221
manner to the retarding-plate field. In essence, the mean ion kinetic energies are
representative of the predominant bath-gas ions and not of the seed (analyte) ions.
In the mass spectrometer third stage the ion-transport mechanism stemming from the
argon plasma was found to be slightly different (see Chapter 3). Because ion flux to the
third stage was attenuated by a small orifice in the differential pumping plate that separates
the second and third stages, ion beam space charge and therefore ion-ion interactions were
weaker than those in the second stage. As a result ions responded independently of each
other and kinetic energies were found to vary linearly with atomic mass; with heavier
elements having higher energies than lighter species.
Ion kinetic energies measured in the helium PS-MS system exhibited different trends
from those determined in the argon ICP-MS. In the second stage of the helium MIP-MS,
ions were found to possess mean kinetic energies that increased with atomic mass. This
finding suggests that coulombic interactions among ions are less severe in the helium
MIP-MS than in the argon ICP-MS, although they appear to be significant in both.
Ion kinetic energy curves taken at two first-stage backing pressures, 0.80 and 1.32
Torr, are plotted in Figure 5.4 along with a theoretical curve corresponding to a source
temperature of 2000 K and derived from relationships to be discussed below. Data used
to construct the experimental plots in Figure 5.4a were deduced from the retardingpotential curves for He+, Mn+, Ba+, and U+ shown in Figure 5.5 (collected at 0.80 Torr)
and Figure 5.6 (collected at 1.32 Torr) and from the curves for other elements (He2+,
Co+, Ba++, Sr+, and Ce+) not reproduced here. The retarding-potential curves shown in
Figures 5.5 and 5.6 are representative of those obtained throughout this study where
heavier ions are detected.
The theoretical plot in Figure 5.4b was deduced solely from gas-kinetic relationships.
Equations 5.3 and 5.4 were used to determine the velocity of the expanding bath gas, in
222
Chapter 5
Figure 5.4. Mean ion kinetic energy curves determined (a) experimentally at first-stage
backing pressures of 0.80 Torr (CD and 1.32 Torr (♦) and deduced by (b) calculation as
described in the text (see Section 5.3.3). Note different vertical axes on plots (a) and (b).
223
Chapter 5
Ion Kinetic Energy (eV)
1.32 Torr
0.80 Torr
432
-
50
100
150
200
250
150
m/z
200
250
m/z
Ion Kinetic Energy (eV)
301
25on -I
15:
10:
50
100
224
Chapter 5
Figure 5.5. Retarding-potential curves for (a) helium (4 amu), (b) manganese (55 amu),
(c) barium (138), and (d) uranium (238) taken when the first-stage backing pressure was
0.80 Torr. Mean energies deduced from these curves are included with the data plotted in
Figure 5.4a.
5.0e+2 *
3.0e+3 n
Chapter 5
3.0e+2
Signal (cps)
Signal (cps)
4.0e+2-
Helium
1.0 eV
2.0e+2
M a n g a n e se
2.1 eV
1.0e+3-
1.0e+2O.Oe+0-»
0.0e+0
5
10
15
0
20
10
15
20
R etarding Potential (V)
Signal (cps)
Signal (cps)
R etarding Potential (V)
5
Barium
3.1 eV
0.0e+0
U ranium
4.4 eV
1.0e+3-
0.0e+0
0
5
10
0
5
10
15
Retarding Potential (V)
20
225
Retarding Potential (V)
20
226
Chapter 5
Figure 5.6. Retarding-potential curves for (a) helium (4 amu), (b) manganese (55 amu),
(c) barium (138 amu), and (d) uranium (238) taken when the first-stage backing pressure
was 1.32 Torr. Mean energies deduced from these curves are included with the data
plotted in Figure 5.4a.
a
Signal (cps)
1.0e+5Helium
0.9 eV
5.0e+4 ■
0.0e+0
0
5
10
19
1.0e+3
M an g a n e se
2.0 eV
5.0e+2-
0.0e+0
0
20
R etarding Potential (V)
Barium
2.1 eV
1.0e+3-
UN »LVUl
5
10
15
20
2.0e+3
2.0e+3-
0
10
R etarding Potential (V)
C
0.0e+0
5
Signal (cps)
Signal (cps)
3.0e+3n
Chapter 5
s '9 nal (CPS>
1.5e+5n
15
R etarding Potential (V)
20
U ranium
4.0 eV
0.0e+0
0
5
10
15
20
R etarding Potential (V)
227
228
Chapter 5
this case helium. With the assumption that analyte ions achieve the same mean velocity as
the bath gas, Equation 5.5 was then used to determine the kinetic energy of the analyte
ions. Of course, for this treatment to be valid, ions are assumed to be skimmed from the
isentropic portion of the expansion core and to be influenced only by gas-kinetic processes
(i.e., ion-ion interactions are neglected). Finally, both ions and neutrals are presumed to
achieve the same velocity, which is determined by [27]
(5.3)
where k is the Boltzmann constant, m is the mass of the gas atom, and Mx is the Mach
number (the velocity, vx, of the atom divided by the local speed of sound), which is a
useful quantity that describes the state of the species (the pressure, density, or
temperature) along the expansion centerline under specified source conditions and is
approximated by
y+1
/xx - X p vy-l
\ D0 /
2(y-l)
a
/Xx - x0\y-1
I Dq /
where xx is the skimming distance downstream from the sampling-plate orifice, x0 is the
point at which the gas streamlines appear to be originating, and A is a constant dependent
on y (here equal to 3.26) [11]. A Mach number of 28 is calculated when the skimming
distance is 10.0 mm and the sampling plate has a 0.4-mm-diameter orifice. At a source
temperature of 2000 K, the velocity of the bath gas is calculated to be 4530 m/s.
229
Chapter 5
Heavy species that are included in the expansion can achieve the velocity of the
expanding bath gas if they sustain a sufficient number of collisions in the continuum flow
region of the expansion. Their kinetic energy will be higher than the bath-gas species by a
factor of m^/nigas, in accordance with the following
Eh = 5 ^
Egas
(5.5)
where the subscript h denotes heavy species and Egas is the mean kinetic energy of the
bath gas [28]. This type of gas dynamic acceleration has been reported also by Fulford
and Douglas for the argon ICP-MS [29].
On the basis of this calculation the ion kinetic energy difference between the helium
bath gas and uranium is nearly 25 eV. This energy spread is particularly large when
compared to the 6 eV difference expected between argon and uranium in an argon
expansion [29]. These high ion kinetic energies that are expected in the helium expansion
are the result of a higher stream velocity for helium than argon. Stream velocities derived
from Equation 5.3 for a Mach number of 28 are compared for helium and argon as a
function of source temperature in Figure 5.7. From this figure the gas-stream velocity
anticipated in an expansion originating from a 2000 K helium source is 4500 m/s whereas
that from a 5000 K argon source is only 2300 m/s.
The slope of the calculated ion kinetic-energy curve in Figure 5.4b is a factor of
approximately 10 greater than that of the experimental curves in Figure 5.4a. Although
species skimmed behind the Mach disk should have lower velocities and correspondingly
lower kinetic energies than those in the expansion core, this difference in slope is so great
that it must also be caused in part by coulombic interactions among ions. This hypothesis
is supported by the mean ion kinetic energies measured for helium at the two first-stage
230
Chapter 5
Figure 5.7. Calculated helium and argon velocities in their respective expansions 10.0
mm downstream from a sampling-plate orifice assumed to be 0.4 mm in diameter (Mx =
28). See Equations 5.3 and 5.4. The helium discharge is assumed to possess a gas
temperature of 2000 K and the argon plasma a temperature of 5000 K.
100001
800060004000'
Argon
--------
2000
-
t
i—
r
■«— j— i— i
2000
i
i— i
|
i
4000
i— i— «— «— | — i— i — i
6000
i— r
| —i
8000
Source Temperature (K)
i— i— i
i
|
10000
Chapter 5
232
backing pressures (see Figures 5.5a and 6a). Both values are far greater than 0.4 eV
(deduced from Equation 5.3), the energy that can be gained from gas-kinetic processes.
In addition, the similarity in slopes of the experimental curves (see Figure 5.4a) taken at
different backing pressures suggests that electrostatic processes are involved. If gaskinetic processes were responsible for the slopes, an increase in first-stage backing
pressure should reduce the kinetic energy difference among the ions. The curves in
Figure 5.4a show no such trend.
Additional evidence for the influence of coulombic forces on ion movement was seen
in the form of potential effects when the first-stage backing pressure was lowered to 0.66
Torr. Shown in Figure 5.8 are retarding-potential curves for He+, Mn+, Ba+, and U+ that
exhibit an anomalous drop in signal at approximately 1 V. Potential effects, which are
induced by the interaction of the ion-beam space charge with the retarding-plate
electrostatic field, are almost certainly responsible for these distortions.
These potential effects are no doubt a result of increased ion flux to the second stage.
As the first-stage backing pressure is lowered, on-axis gas flux is expected to increase
because the skimmer will be located closer to the expansion core. If ion flux also remains
directed toward the skimmer, then ion density in the second stage will rise. This increase
in ion density can promote potential effects if the ion space charge begins to interfere with
the retarding-plate electrostatic fields. Ion-beam space charge will be increased also by
two other phenomena that promote ion-ion interactions: a decrease in gas density and the
loss of electrons caused by charge separation (refer to Chapter 2). Both these processes
are known to occur whenever a partially ionized gas passes through a differential-pumping
plate aperture (refer to Chapter 2).
The retarding-potential curves taken at higher first-stage backing pressures (0.80 and
1.32 Torr, shown in Figures 5.5 and 5.6, respectively) appear to be less influenced by
233
Chapter 5
Figure 5.8. Retarding-potential curves for (a) helium (4 amu), (b) manganese (55 amu),
(c) barium (138 amu), and (d) uranium (238) taken at a first-stage backing pressure of 0.66
Torr. The signal depressions seen at approximately 1 V are presumed to be caused by
potential effects.
4.0e+3 n
Signal (cps)
Helium
1.4 eV
Chapter 5
Signal (cps)
1.2e+4
8.0e+3
M an g a n e se
3.2 eV
6.0e+3
l.0e+3H
2.0e+3
O.Oe+O
0
0.0e+0
5
10
1£i
20
0
Retarding Potential (V)
10
1
20
Retarding Potential (V)
6.0e+3
5.0e+3n
5.0e+3i
4.0e+3H
4.0e+3
Signal (cps)
Signal (cps)
5
Barium
4.9 eV
3.0e+3
2.0e+3
3.0e+3
U ranium
4.0 eV
2.0e+3
1.0e+ 3i
1.0e+3
0.0e+0
0
5
10
15
0
5
10
Retarding Potential (V)
20
234
Retarding Potential (V)
0.0e+0
20
235
Chapter 5
potential effects. These effects are probably suppressed by a lower ion flux to the second
stage. As the first-stage backing pressure is raised, background-gas penetration will
randomize ion movement and increase the likelihood of ion neutralization that can result
from collisions with the walls of the skimmer. At the same time, ion-electron
recombination, which occurs in the shockwave structure, will increase with higher firststage backing pressures.
5.3.4. Effect of Air Entrainment on Ion Kinetic Energy
5.3.4.I.
Influence of off-axis sampling on background spectrum.
Elemental analysis performed with the helium MIP-MS combination required that
sampling be performed slightly off axis because of the manner in which the discharge
attached itself to the sampling plate. When the sampling plate was moved to within a few
millimeters of the discharge, a filamentary plasma formed between the sampling plate and
the wall of the plasma torch. The greatest analyte signals were achieved when the wall of
the torch was positioned below the sampling orifice (approximately 3.0-mm off axis). At
this position the plasma filament arced directly to the sampling-plate orifice. Because this
filamentary discharge shifted spatially with time, determinations had to be performed as
quickly as possible. In addition, sampling at this position permitted a small degree of air
entrainment, unlike sampling in the center of the plasma. This off-center sampling
manifested itself in the background spectrum and in ion kinetic energies (see Figures 5.95.11).
The background spectrum (0 to 50 m/z) shown in Figure 5.9 was taken at the 3.0-mm
off-axis position in a wet plasma. The existence of nitrogen ions in the spectrum serves as
evidence of air entrainment. The presence of
is due mainly to the introduction of
H2O. No ions could be detected above 17 amu. In contrast, the background spectra
236
Chapter 5
Figure 5.9. Background scan taken at the analyte-sampling position (3.0-mm off axis)
for a helium discharge with aerosol introduced. First-stage backing pressure, 0.77 Torr.
Chapter 5
237
©
CO
o
in
o
CO
X
o
o
CM
CM
o
X
6.0e+3 n
0
CO
+
0
o
in
CO
+
0
o
CO
+
0
o
CO
+
0
o
CO
cm
(sdo) |eu6|S
CO
+
0
©
r-
o
+
0
Oa
o
m/z
o
238
Chapter 5
displayed in Figures 5.10a and 5.10b, obtained on-axis in a dry and solvent-fed
discharge, respectively, show no evidence of atmospheric gas entrainment For the dry
helium plasma (Figure 5.10a), the helium dimer ion prevailed in the background. When
water was introduced into the plasma, the oxygen ion became predominant species (Figure
5.10b).
5.3.4.2.
Influence of off-axis sampling on helium ion kinetic energies.
Helium retarding-potential curves were measured for the on-axis sampling configuration
and are shown in Figures 5.1 la and 5.1 lb for dry and wet discharges, respectively.
Interestingly, the addition of water had little effect on the helium ion kinetic energy
distribution. The mean energy measured for the dry plasma was 0.4 eV (Figure 5.1 la),
whereas that for the wet plasma was 0.3 eV (Figure 5.1 lb). These mean energies are
very close to the values predicted by gas dynamic theory. From Equation 5.3 (Mx set
equal to infinity [29]), these energies correspond to source temperatures of 1700 and 1600
K, respectively, values that are not unreasonable.
This correlation with gas dynamic theory is especially surprising when it is contrasted
to the behavior noted earlier when the MIP was sampled off axis. In that situation, the
mean ion kinetic energy of helium was found to be far greater than could be reconciled by
gas-kinetic processes (see Figures 5.1, 5.5a, 5.6a, and 5.8a). The reason for this
apparent discrepancy must lie in the relative concentrations of bath-gas (He) and "seed"
ions in the sampled regions of the plasma. For example, it would seem reasonable that the
correlation observed on-axis between the experimentally determined and calculated mean
ion energies is a result of the fact that the ions extracted from that zone are almost
exclusively He+. In effect, the helium ions respond collectively to the potential field of
the retarding plate and the measured kinetic energy is of the predominant bath-gas, He+.
239
Chapter 5
Figure 5.10. Background spectral scans for a (a) dry plasma and a (b) wet plasma
sampled on axis. First-stage backing pressures, 0.76 and 0.71 Torr, respectively.
o
<D
+
cn
+
__ L_
__ L_
I
(D
ro
o
X
x
(0 ,
ro
01
(D
+
Ol
.j
L_
2.0e+5 1
U1
o
<D
o
<D
+
O
o
o
o
<D
+
O
o
a>
+
4*
4*
o
<D
+
4*
p
o
(D
+
CO
o
(D
+
4*
o
+
o
o
o>
o
o>
CO
ro
Chapter 5
Signal (cps)
Signal (cps)
o
00
o
o
o
o
240
241
Chapter 5
Figure 5.11. Helium retarding-potential curves for a (a) dry plasma and a (b) wet
plasma sampled on axis. First-stage backing pressures, 0.76 and 0.71 Torr, respectively.
Chapter 5
242
4.0e+4 1
Signal (cps)
3.0e+4 H
Helium
0.4 eV
2.0e+41
1.0e+4i
0.0e+0
0
5
10
15
20
Retarding Potential (V)
Signal (cps)
6.0e+3 1
Helium
0.3 eV
2.0e+3 -
0.0e+0
0
5
10
15
Retarding Potential (V)
20
243
Chapter 5
5.3.4.3.
Effect of air intentionally added to the MIP. The suggestions
above axe supported by measurements performed on an MIP into which selected quantities
of air were intentionally added. Figures 5.12a and 5.12b show the background spectra
when air is added at a rate of 22 and 60 mL/min, respectively, to the inner-gas flow of the
torch. This introduced air increased the prominence of nitrogen and nitrogen-containing
polyatomic species in the background spectra (Figures 5.12a and 5.12b). Also, upon the
addition of air to the plasma, both the helium monomer and dimer signals decreased and
the He+ kinetic energy distributions broadened (see Figures 5.13a and 5.13b). The
retarding-potential curves seen in Figures 5.1,5.6a, 5.7a, and 5.9a in which air
entrainment occurred, appear to be similar to those in Figures 5.13a and 5.13b. Even
though the method in which air was introduced into the plasma (air entrainment versus
being combined with the plasma gas) differed, it is reasonable to assume that air
entrainment is partly responsible for the broadening of the ion kinetic energy distributions
in the earlier studies.
It is not clear why the mean ion kinetic energy of helium increases when air is
intentionally or inadvertently introduced into the MIP. One possibility is that ions gain
energy from an electrostatic field as they pass through the skimmer orifice. Air
entrainment could promote the formation of an ion sheath in front of the skimmer-cone
orifice as the partially ionized expansion gas comes in contact with the surface of that
cone. Ions that pass into the ion sheath will then be accelerated toward the skimmer cone,
which is at ground potential [30]. If ion space charge is low in the mass spectrometer
second stage, these accelerated ions will maintain their newly acquired velocity.
244
Chapter 5
Figure 5.12. Background spectral scans for a (a) dry plasma and a (b) wet plasma
sampled on axis in which an additional 22 and 60 mL/min of air was introduced. Firststage backing pressures, 0.71 and 0.72 Torr, respectively.
o
o
ro
0
0
+
+
+
■
■
■
<D
<D
(D
01
o
01
__ L_
__ L_
ro
o
O
o>
o
00
o
Signal (cps)
Lg+eo’8
Signal (cps)
O
0
<D
+
01
_L_
z
+
X O \ zror
ro
+
o
+
246
Chapter 5
Figure 5.13. Helium retarding-potential curve for a (a) dry plasma and a (b) wet plasma
sampled on axis in which an additional 22 and 60 mL/min of air was introduced. Firststage backing pressures, 0.71 and 0.72 Torr, respectively.
247
Chapter 5
6.06+2"
(/)
4.0e+2
CL
o
Helium
0.8 eV
CO
c
o>
«
c/)
2.0e+2-
0.0e+0
i
f
i
i
'r
i
f — 1‘
5
15
10
20
Retarding Potential (V)
3.0e+2-i
o npa.9
a.
a
Helium
0.7 eV
co
c
o>
jyj
1.Oe+2'
0.0e+0 ■
r
i
"i
i
|
5
i ” t
i
»
| 1i
10
i”
i
i
|
1
1
15
Retarding Potential (V)
1
1
I
20
Chapter 5
248
5.4.
Co n c l u s io n
Sampling a helium plasma properly is challenging, in part, because it is difficult to
pump helium effectively and to reduce air entrainment. Air entrainment is especially
troublesome because of its effect on the spectral background. In an attempt to overcome
these limitations, both pumping capacity and plasma gas flow rate were increased.
Despite this course of action, it was still necessary to use a small sampling-plate aperture
(0.4-mm diameter) and to skim behind the Mach disk. The use of the small samplingplate orifice decreased ion flux through the interface and compromised sensitivity.
Compared with similar experiments presented in Chapter 3 for the argon ICP-MS, ion
flux to the mass analyzer was lower by a factor of approximately 100 in the helium MIPMS system. Of course, a reduced ion flux does not necessarily translate into a low
sensitivity because ion-optic lenses can be used to collect and focus ions on axis and to
concentrate the ion beam.
In PS-MS, gas dynamic acceleration is an undesirable phenomenon because it can
ultimately lead to mass discrimination by the ion-optic lenses. As demonstrated in Chapter
3, gas dynamic acceleration (and the consequent mass-bias effects) was completely
suppressed in the second stage and could be equally overpowered in the third stage if the
ion flux was maximized. However, in the helium MIP-MS these coulombic interactions
were found to be lower than those determined for the argon ICP-MS, so gas-dynamic
acceleration remained prominent Therefore, if helium PS-MS is to be a viable alternative
to argon ICP-MS, steps will need to be taken in ion-optic lens or sampling-skimming
design.
Another issue that should be considered is whether skimming can be performed
behind the Mach disk without compromising the operation of a PS-MS instrument. In
Chapter 5
249
Chapter 4 it was concluded that skimming under compromised conditions (i.e., behind the
Mach disk) could actually increase ion flux to the mass analyzer. The reason for this
behavior is that a moderate ion flux to the second and third stage, where ion-ion
interactions are significant, can moderate ion-beam dispersion. As a result, a greater
number of species maintain on-axis trajectories and can ultimately be detected. As shown
in this study, there are several potential drawbacks to skimming behind the Mach disk
including the formation of cluster ions from collisions of background species with ions
traveling from the expansion core, the negative effects of gas dynamic acceleration, and a
reduction in ion flux. Practically, it was found that the only disadvantage to skimming
behind the Mach disk appears to be that on-axis ion flux is low.
Overall, it is concluded that if helium PS-MS is to become a viable alternative to argon
ICP-MS, two obstacles need to be overcome: The first is to eliminate air entrainment and
the second is to maintain a high ion flux through the interface. As has been shown here, a
small sampling-plate aperture can be used to control air entrainment. Unfortunately, this
approach limits ion flux through the interface. Although ion flux could be restored by
skimming closer to the expansion core, a higher pumping speed in the first stage would be
needed. Similarly, pumping speed in the mass spectrometer second and third stages
should be raised so the skimmer can be positioned farther upstream were ion and gas
densities are directed more on axis.
250
C h ap ter 5
5 .5 . L i t e r a t u r e C i t e d
[1]
G. M. Hieftje and G. H. Vickers, Anal. Chim. Acta 216, 1 (1989).
[2]
S. H. Tan and G. Horlick, Appl. Spectrosc. 4 0 , 445 (1986).
[3]
D. W. Koppenaal and L.F. Quinton, J. Anal. At. Spectrom. 3, 667 (1988).
[4]
A. Montaser, S.-K. Chen, and D. W. Koppenaal, Anal. Chem. 5 9 , 1240 (1987).
[5]
R. D. Satzger, F. L. Fricke, P. G. Brown, and J. A. Caruso, Spectrochim. Acta
42B , 705 (1987).
[6]
P. G. Brown, T. M. Davidson, and J. A. Caruso, J. Anal. At. Spectrom. 3, 763
(1988).
[7]
J. T. Creed, A. H. Mohamad, T. M. Davidson, G. Ataman, and J. A. Caruso,
J. Anal. At. Spectrom. 3,763 (1988).
[8]
J. T. Creed, T. M. Davidson, W.-L. Shen, P. G. Brown, and J. A. Caruso,
Spectrochim. Acta 44B, 909 (1989).
[9]
D. M. Chambers, W.-H. Hsu, G. H. Vickers, and G. M. Hieftje, 1988 Pittsburgh
Conference and Exposition, New Orleans Louisiana, February 1988 Abstract
No. 096.
[10] R. Campargue, Rarefied Gas Dynamics, 6 ^ Symposium. Eds. L. Trilling and
H. Y. Wachnam, Academic Press, New York (1969), p. 1003.
[11] H. Ashkenas and F. S. Sherman, Rarefied Gas Dynamics, 4 * Symposium. Ed.
J. H. de Leeuw, Vol. n, Academic Press, New York (1966), p. 89.
[12] H. Helm, T. D. Mark, and Lindinger, Pure Appl. Chem. 5 2 , 1739 (1980).
[13] G. N. Spokes and B. E. Evans, Tenth Symposium (International) on Combustion,
The Combustion Institute, Pittsburgh, Pa (1965), p. 639.
[14] M. Y. Jaffrin, Phys. Fluids 8, 606 (1965).
[15] K. G. Michlewicz and J. W. Carnahan, Anal. Chem. 58, 3122 (1986).
[16] J. A. Dean and T. C. Rains, Atomic Absorption Spectrometry, Vol. 2, Eds. J. A.
Dean and J. C. Rains, Dekker, New York (1971), p. 327.
[17] B. Rowe, Int. J. Mass Spectrom. Ion Phys. 16, 209 (1975).
[18] P. S. Wei and A. Kupperman, Rev. Sci. Instrum. 4 0 , 783 (1969).
Chapter 5
251
[19] S. Dushman, Scientific Foundations of Vacuum Technique, Ed. J. M. Lafferty,
Wiley, NY (1962).
[20] K. Bier and B. Schmidt, Z. Angew. Phys. 11, 34 (1961).
[21] D. J. Douglas and J. B. French, Spectrochim. Acta 41B, 197 (1986).
[22] A. L Gray, R. S. Houk, and J. G. Williams, J. Anal. At. Spectrom. 2, 13 (1987).
[23] R. S. Houk, J. K. Schoer, and J. S. Crain, J. Anal. At. Spectrom. 2, 283 (1987).
[24] A. Hoglund and L-G. Rosengren, Int. J. Mass Spectrom. Ion Processes
60, 173 (1984).
[25] M. H. Abdallah and J. M. Mermet, Spectrochim. Acta 37B, 391 (1982).
[26] P. W. Atkins, Physical Chemistry, Oxford University, Great Britain (1982), p.873.
[27] J. B. Anderson and J. B. Fenn, Phys. Fluids 8, 780 (1965).
[28] N. Abuaf, J. B. Anderson, R. P. Andres, J. B. Fenn, D. G. H. Marsden, Science
156, 997 (1967).
[29] J. E. Fulford and D. J. Douglas, Appl. Spectrosc. 40, 971 (1986).
[30] M. J. Vasile and H. F. Dylla, Plasma Diagnostics, Eds. O. Auciello and D. L.
Flamm, Academic, NY (1989), chapter 4.
252
6
FUTURE WORK
6.1. I n t r o d u c t i o n
The qualitative ion-transport model presented in this work not only is useful for
elucidating those processes that influence ion movement, but also serves as a guide to
innovations in plasma source-mass spectrometry (PS-MS). On the basis of this model it is
apparent that a next-generation instrument design should take into account both gas-kinetic
and coulombic forces. In the designs of current inductively coupled plasma-mass
spectrometer (ICP-MS) instruments dispersion of the ion beam by ion space charge is
neglected. Ion-optic-lens designs do not take into account the magnitude of the ion-beam
space charge; as a result, the lenses are not able to focus ions properly as theory would
predict [1]. At this time no research group has been able to model successfully ion
trajectories through an ICP-MS [1,2]. Consequently, many of the innovations and
developmental trends seen in this field have been achieved empirically or have been driven
by conventions that are not fundamentally proven.
A recent trend in commercial ICP-MS instruments has been the utilization of a threestage differentially pumped instrument. The primary advantage to a three-stage instrument
is that pumping requirements for each vacuum stage are less stringent than for a one- or
Chapter 6
253
two-stage system. For example, a one-stage system would require the pressure to drop by
a factor of 108 to sample an atmospheric pressure plasma to 10'5 mbar, the pressure of the
mass analyzer chamber,. The addition of a second vacuum chamber can reduce the
pressure drop between the stages to a factor of 104 if the pressure is divided evenly.
Further, the common pressure differential between the successive stages in a three-stage
instrument is generally less than a factor of 103. However, with the addition of each stage
to the mass spectrometer interface, the mass analyzer is moved farther downstream from
the source; as a result, ion flux to the mass analyzer decreases. Even though this loss can
be controlled by the use of ion-optic lenses, the ion-beam density and composition can be
alter by ion focusing. A good example of this phenomenon in current instruments is the
matrix effects, which are presumed to be an artifact of sampling the plasma ions through
the interface (refer to Chapter 3) [2,3]. The physical processes responsible for these matrix
interferences were examined in Chapters 3 and 4. As discussed in Chapter 3, a sampled
argon-ion flux was dispersed when high concentrations of heavy or light analyte elements
were added to the ion beam. Moreover, light ions (i.e., Li+ and Al+) were found to diffuse
preferentially to the edges of the ion beam presumably because they possess a higher
collisional frequency than heavier bath-gas ions (Arf) (see Chapters 3 and 4). From these
studies it was concluded that ion-beam dispersion is one cause for the observed matrix
effects. This sort of mechanism is believed to be responsible for matrix interference effects
in ICP-MS where analyte ion signals are affected greatly by the presence of other matrix
elements [2,3].
6.1.1.
Single-Stage Interface
A potentially attractive approach for the development of a PS-MS instrument is to
employ a one-stage extraction system. This design would place the mass analyzer close to
254
Chapter 6
the ionization source and would diminish the need for ion-optic lenses (see Figure 6.1). In
this instrument the plasma flux would be extracted through the aperture of the first-stage
sampling plate according to the appropriate kinetic-gas laws (see Chapter 5). In addition to
these gas-kinetic processes, electron loss at the sampling orifice and a decrease in gas
density will increase coulombic interactions among ions (refer to Chapters 3 and 4). As
these coulombic interactions increase, the appropriate steps would need to be taken to
ensure that a representative ion sample reaches the mass analyzer. This goal may be
accomplished by moving the mass analyzer upstream before beam dispersion alters the ionbeam composition.
Unfortunately, there exists a few design constraints that force the modification of this
initial plan. The first is that the mass analyzer is limited by the number of ions it can
accept As the mass analyzer is moved upstream, ion density may increase beyond a
tolerable level. Fields from the ion-beam space charge can hinder electrostatic focusing of
in mass analyzer. The most direct approach to overcome this problem is to place a beam
collimator that has a small aperture before the entrance of the mass analyzer.
Another critical factor is the need for high pumping speed so the proper pressure can be
maintained in the mass analyzer vacuum chamber. If a sampling plate with a 1.0-mmdiameter orifice were used in this one-stage system, the necessary pumping speed to
maintain a first-stage backing pressure of 10'5 mbar would be approximately 9.1 x 109 L/s
(argon converted to standard temperature and pressure (STP)). This theoretical pumping
speed (S) was determined by combining Equation 2.4,which describes viscous flow
through a differential-pumping-plate aperture, with the following expression [4]
S = 0<Pl-P2)
(6 . 1)
255
Chapter 6
Figure 6.1. Single-stage ICP-MS instrument. The atmospheric-pressure plasma is
sampled directly into the mass analyzer chamber held at 10-5 mbar.
n
QUADRUPOLE
RATEMETER
PRE-AMP
to
r s
t i ­
nt
"i
PUMP
QUADRUPOLE
CONTROL UNIT
° l-J - - - - i—
1 J1 = °
QUADRUPOLE
POWER SUPPLY
DESOLVATOR
IMPEDANCE
MATCHER
NEBULIZER
p\
PERISTALTIC
PUMP
POWER SUPPLY
K>
l/l
0\
Chapter 6
257
where G is the gas flow velocity (conductance in L/s) through the sampling-plate orifice,
Pi is the source pressure (atmospheric), and P2 is the first-stage backing pressure. For this
calculation the source temperature was assumed to be 5000 K [5],
Because such a high pumping speed is not feasible, the more obvious approach is to
reduce the size of the sampling aperture. Unfortunately, a small sampling-plate orifice
induces boundary layer formation [6] and charge separation (Chapter 2). Both of these
processes can interfere with ion extraction in the following manner: Boundary layer
formation results in the creation of a stagnant gas layer over the sampling aperture. In this
stagnant region there is a high incidence of cluster formation as a result of collisions among
the different gas species [6J. However, the effect of charge separation at the differential
pumping aperture, specifically the preferential loss of electrons that shield ions from one
another, will be to promote dispersion of the ion beam (refer to Chapter 2).
To reduce the undesirable effect of boundary layer formation the sampling-plate
aperture is generally required to be over two orders of magnitude larger than the mean free
path of the bath-gas species [6], calculated to be 1.3 (im in Chapter 2. This guideline
would require the sampling-plate orifice to be larger than 0.13 mm in diameter. Along
these same lines the mean free path for ions and electrons in a free flowing ICP was
calculated to range from 1 to 10 (im (refer to Chapter 2). (Because the mean free path of
the charged species will be altered as the plasma comes in contact with the sampling plate, it
must be noted that this range of values is only an estimation.) If the same rule for bulk
sampling of a gas is applied to the charged species of a plasma, the sampling-plate orifice
would need to be between 0.1 and 1.0 mm in diameter. The pumping speed needed to
maintain a 10'5 mbar pressure within this range of orifice diameters is graphed in Figure
6.2. The argon gas temperature is assumed to be 5000 K [5].
258
Chapter 6
Figure 6.2. Pumping speed required to maintain a vacuum-chamber pressure of 10'5
mbar with various aperture diameters and with a reservoir pressure of 1013 mbar (760
Torr). Source gas, argon; Source temperature, 5000 K.
800000 600000 400000 -
200000
-
0.0
0.1
0.2
0.3
0.4
0.5
0.6
Sampling-Plate Orifice Diameter (mm)
Chapter 6
260
Turbomolecular pumps are not made generally to exceed a pumping speed of 9000 L/s.
Such a pump is very expensive, approximately $70,000, and quite bulky, as large as 1.5 m
in diameter. At this available pumping speed the orifice diameter should be no larger than
0.05 mm so a pressure of 10~s mbar can be maintained in the mass analyzer chamber
(Equation 6.1). The cost for the required vacuum system can be lowered if the sampling
aperture is reduced further in size. An example of this cost relationship is graphed in
Figure 6.3 as a function of pumping speed for the turbomolecular pump. As the required
pumping speed goes up the cost of a one-stage vacuum system becomes prohibitively
expensive and quickly exceeds the cost of a three-stage instrument, which is approximately
$25,000.
Clearly, a compromise has to be achieved among pumping speed, instrument cost, and
the aperture size of the differential pumping plate so vacuum requirements are realistic and
ion flux is sufficient. Even though the number of vacuum components needed for a onestage system would simplify the instrument design, pumping requirements would certainly
drive the cost of such a system beyond that of a three-stage instrument One way to offset
these pumping requirements is to reduce the dimensions of the sampling-plate orifice. This
approach is risky because boundary layer formation and charge separation will be
increased; however, the extent to which these processes attenuate and alter ion population
in an ICP-MS still has not been investigated. For this reason, studies that examine the
effects of boundary layer formation and charge separation need to be performed before a
one-stage instrument should be considered.
6.1.2.
Multi-Stage Interface
The stringent pumping requirements of a one-stage system can be reduced by adding
another stage. A two-stage system would be most cost effective if the pressure differential
261
Chapter 6
Figure 6.3. Approximate cost of a turbomolecular pumping system in U.S. Dollars as a
function of pumping speed (nitrogen, STP).
50000i
40000'
30000 -
1000
2000
3000
4000
5000
Pumping Speed (L/s nitrogen)
6000
Chapter 6
263
between the stages were evenly distributed. With this formulation the first-stage pressure
should be approximately 0.1 mbar and the second stage 10‘5 mbar.
One benefit of having the first-stage backing pressure at 0.1 mbar is that the bath-gas
species would behave as through they were expanding into a perfect vacuum. As a result,
the shockwave structure would be diffuse and the expansion core enlarged. The benefit of
this type of expansion is that skimming can be performed at greater downstream distances
without complications caused by the penetration of background gases. The gas species
skimmed from a downstream position have large on-axis velocity components, which
permit bath-gas species to remain on-axis. Of course there will be space-charge dispersion
among ions, which may disrupt this ordered distribution and lead to matrix effects;
however, it is uncertain whether these effects will be significant in the initial stage of the
interface, which is to be held at 0.1 mbar.
Although the pumping requirements for a two-stage system are demanding, sufficiently
low pressures could be achieved with the proper vacuum-system design. This particular
instrumental configuration would require a high-speed pump for both the first and second
stages. One device that would be able to maintain the first-stage backing pressure at 0.1
mbar is a roots vacuum pump. Conventionally, the first-stage backing pump in a threestage ICP-MS is a rotaty-vane pump with a pumping speed below 20 L/s. A roots pump
that is comparable in size to a rotary-vane pump generally will have a greater pumping
speed by a factor of approximately 10. In the second stage, a high-speed turbomolecular
pump can be used to reduce the pressure to 10‘5 mbar. A pumping speed of 3000 L/s
(argon converted to STP) in the first stage and 5000 L/s (argon converted to STP) in the
second stage would be needed to maintain each of these pressures with the present
sampler-skimmer arrangement (i.e., orifice diameters of 1.0 mm, separated by a 10.0 mm
distance). This first-stage pumping speed was calculated by combining Equations 2.4 and
264
Chapter 6
6.1. The second-stage pumping speed was determined from the following relationship,
which describes the rate of molecular flow (Q) in molecules s'1 through an orifice [4]
Q = ^ v a As (P2 -P i)
(6.2)
where va is the mean molecular speed, which has been determined as the mean velocity of
the stream passing through the skimmer refer to Chapter 5; As is the area of the skimmer
orifice; and Pi and P2 are the first- and second-stage pressures, respectively. For this
calculation the source is at atmospheric pressure and 5000 K [5].
Even though these pumping requirements are stringent, they may be diminished by
reducing the aperture size in the first-second-stage differential pumping plate. The best
approach is to reduce the size of the skimmer-cone orifice and to maintain a large aperture
in the sampling plate. Because skimming is usually performed in the molecular-flow
region of the expansion, the size of the skimmer-cone orifice can be reduced without the
onset of significant boundary layer formation. This action probably will decrease ion flux
through the interface, but might be offset by the use of ion optics or through movement of
the quadrupole upstream.
Another approach to compensate for a low ion flux is to reduce ion-beam dispersion at
the skimmer orifice. Ion flux can be increased by applying a negative potential to the
skimmer cone [7]. The maximum flow of ions through the skimmer-cone orifice can be
deduced from the following relationship [2]
W = 0 .9 (m )'« v 3 « (^ )2
(6.3)
265
Chapter 6
where Imax is the maximum current that can flow through the metal orifice in pA, m/z is the
mass-to-charge ratio of the ion, V is the ion kinetic energy, Ds is the orifice diameter of the
skimmer, and L is the length of the skimmer channel. Essentially, the ion flow through the
skimmer orifice can be raised by increasing the ion velocity, which decreases ion space
charge dispersion. The maximum current through the skimmer-cone orifice in a typical
ICP-MS instrument is calculated to be approximately 33 pA; this assumes that m/z is 40,
V is 1.3 eV (see Chapter 2), Ds is 1.0 mm, and L is 0.5 mm. In analytical ICP-MS
instruments, far more ions exist in the expansion core than can pass through the 1.0-mmdiameter orifice in the skimmer cone. The theoretical ion flux can be deduced from the gas
flow that passes through an area the same size as the skimmer-cone orifice by [8]
Gs= nx vx As
(6.4)
where nx and vx are the gas density and gas velocity at a distance x downstream from the
sampling-plate orifice and As is the area of the skimmer-cone orifice. For a sampling-plateorifice diameter of 1.0 mm, a skimming distance (x) of 10.0 mm, and a source temperature
of 5000 K [5], the gas flow is calculated to be 1.41 x 1018 atoms s*1. If the expanding
argon gas is 0.1% ionized, the ion density passing through a 1.0-mm-diameter area is 1.4 x
1015 ions s-1 or 2300 pA. Ion dispersion at the skimmer-cone orifice may account for a
loss in ion throughput of approximately a hundred fold in typical ICP-MS instruments.
Fortunately, a portion of this loss could be recovered by accelerating the ions toward the
skimmer orifice with a negative potential as described above.
An additional advantage of a negatively biased skimmer cone is that it will retard
electron migration toward the orifice walls. As a result, this process may sustain a large
electron population that can shield ions from one another and reduce dispersion of the ion
266
Chapter 6
beam. If beam dispersion persists, muldple-apertured ion lenses may be placed after the
skimmer to help reduce the ion-beam space charge [9]. This ion-optic-lens system would
shield ions from one another to reduce ion-ion dispersion and block electron loss.
6.2. C o n c l u s i o n
The main advantage of a multistage vacuum system is that pumping requirements are
reduced in each stage. However, with each addition vacuum stage the mass analyzer is
positioned further from the ionizadon source and, as a result, ion signal decreases. The
conventional approach used to counterbalance the loss of ion flux is to employ ion focusing
lenses that collect and direct these species toward the mass analyzer. However, another
approach is to move the mass analyzer toward the ionization source. The mass analyzer
can be placed closest to the ionization source in a single-stage instrument. Unfortunately,
the pumping requirements for such an instrument are prohibitive. A two-stage instrument
would be the next system of choice and might be realized with the development of a novel
vacuum system similar to the one suggested in Section 6.1.2.
An illustration of one possible instrument configuration is shown in Figure 6.4. The
first stage is backed by a roots pump and kept within a 0.1 mbar range and the secondstage pressure is maintained below 10"5 mbar with a high-speed turbomolecular pump.
Because the mass analyzer will be near the ionization source, ion flux might not require
collection and focusing, which are processes that tend to bias the composition of the ion
beam (refer to Chapter 4). If possible focusing optics are to be avoided, instead
accelerating-decelerating extraction optics that simply transport ion flux without
concentrating the ion beam are a better choice [9]. This ion-optic arrangement should
reduce matrix-induced interferences by decreasing preferential ion migration and dispersion
267
Chapter 6
Figure 6.4. Two-stage ICP-MS instrument. The atmospheric-pressure plasma is
sampled in the the mass spectrometer first stage maintained at the 0.1 mbar range and the
mass analyzer chamber is kept below 10'5 mbar.
n
sr
69
RATEMETER
PRE-AMP
QUADRUPOLE
QUADRUPOLE
CONTROL UNIT
QUADUPOLE
POWER SUPPLY
TO
TURBOMOLECULAR
r—
PUMP
o
J—
o
H
'
■o
r*
n
n
EXTRACTION
OPTIC
ROOTS PUMP
IMPEDANCE
MATCHER
NEBULIZER
PERISTALTIC
PUMP
DESOLVATOR
rf
POWER SUPPLY
K>
Os
00
Chapter 6
269
within the ion beam (refer to Chapters 3 and 4). Additional dispersion can be reduced by
biasing the first-second-stage differential pumping plate to a negative potential as discussed
in Section 6.1.2. Several benefits are likely to arise from the development of a two-stage
instrument as compared with the three-stage instruments. First, the number and
complexity of the ion-optic lenses can be reduced and as a result instrument operation and
response will not only be improved but simplified. Second, because of the upstream
positioning of the mass analyzer and efficient ion collection, ion flux could be increased if a
higher analyte throughput were desired. At this time the capacity for ion throughput has
not been investigated in ICP-MS, which typically employs the quadrupole mass analyzer.
Increases in ion flux to the quadrupole could push detection limits below ppt levels. In
addition, a high ion flux was shown to reduce mass bias effects caused by gas dynamic
acceleration (Chapter 3).
Chapter 6
270
6.3. L i t e r a t u r e C i t e d
[1]
M. A. Vaughan and G. Horlick, Ion Lens Simulations of an Inductively Coupled
Plasma-Mass Spectrometer, submitted to Spectrochim. Acta B, 1990.
[2]
G. R. Gillson, D. J. Douglas, J. E. Fulford, K. W. Halligan, and S. D. Tanner,
Anal. Chem. 60, 1472 (1988).
[3]
S. M. Tan and G. Horlick, J. Anal. At. Spectrom. 2, 745 (1987).
[4]
S. Dushman, Scientific Foundations o f Vacuum Technique. Ed. J. M. Lafferty,
Wiley, N. Y. (1962), chapter 2.
[5]
M. Haung, D. S. Hanselman, P. Y. Yang, and, G. M. Hieftje, Isocontour Maps o f
Electron Temperature, Electron Number Density and Gas Kinetic Temperature in the
Ar ICP Obtained by Laser-Light Thomson and Rayleigh Scattering, submitted to
Spectrochim. Acta B, 1990.
[6]
A. Hoglund and L-G. Rosengren, Int. J. Mass Spectrom. Ion Processes
60, 173 (1984).
[7]
R. G. Wilson and G. R. Brewer, Ion Beams with Applications to Ion Implantation,
Wiley, NY (1973).
[8]
D. J. Douglas and J. B. French, J. Anal. At. Spectrom. 3,743 (1988).
[9]
Y. Okamoto and H. Tamagawa, Rev. Sci. Instrum. 43,1193 (1972).
D a v id M i c h a e l C h a m b e r s
W ork Address
Chemistry Department
Indiana University
Bloomington, IN 47405
(812) 855-7905
Home Address
1287 Eigenmann Hall
Bloomington, IN 47406
(812) 857-5252
P e r s o n a l In f o r m a t io n
Date of Birth: March 26,1964
Marital Status: Single
SSN: 246-33-9001
E d u c a t io n
Ph.D., Analytical Chemistry, 1990
Minor, Physical Chemistry
Indiana University, Bloomington, IN 47405
Adviser: Distinguished Prof. Gary M. Hieftje
Dissertation title: "Fundamental Studies of the Sampling Process in
the Argon Inductively Coupled Plasma- and Helium MicrowaveInduced-Plasma Mass Spectrometer."
B. S., Chemistry, May 1986
University of North Carolina
Chapel Hill, NC 27514
G r a d e P o in t A v e r a g e
Undergraduate - 3.5/4.0
Graduate - 3.8/4.0
E x p e r ie n c e
1/88Graduate Research Assistant, Indiana University
7/86-12/88 Associate Instructor, Indiana University
1/85-5/85 Undergraduate Research Assistant
O r g a n iz a t io n s
Member, American Chemical Society
Member, Society for Applied Spectroscopy
P u b l ic a t io n s
B. S. Ross, D. M. Chambers, G. H. Vickers, P. Yang, and G. M. Hieftje;
"Characterization of a 9-mm Torch for Inductively Coupled Plasma Mass Spectrometry";
submitted to J. Anal. At. Spectrosc., 1990.
B. S. Ross, D. M. Chambers, and G. M. Hieftje; “Fundamental and Applied
Investigations in Plasma-Source Mass Spectrometry for Elemental Analysis”, submitted to
MicrochimicaActa, 1990.
B. S. Ross, D. M. Chambers, and G. M. Hieftje; “The Use of a Center-Tapped Load Coil
with a Non-Commercial Instrument for Inductively Coupled Plasma-Mass Spectrometry”,
submitted to Spectrochim. Acta B, 1990.
B. S. Ross, D. M. Chambers, and G. M. Hieftje: “The Reduction Interference Effects in
Inductively Coupled Plasma-Mass Spectrometry”, submitted to J. Am. Soc. Mass
Spectrom., 1990.
B. S. Ross, D. M. Chambers, and G. M. Hieftje, “The Use of Ion-Optic Lens
Configuration to Eliminate Matrix-lnterferences Effects In Inductively Coupled PlasmaMass Spectrometry”, submitted to Spectrochim. Acta B, 1990.
D. M. Chambers, J. W. Carnahan, Q. Jin, and G. M. Hieftje; "Fundamental Studies of
the Sampling Process in an Inductively Coupled Plasma Mass Spectrometer. Part IV:
Replacement of the Inductively Coupled Plasma with a Helium Microwave-Induced
Plasma", submitted to Spectrochim. Acta B, 1990.
D. M. Chambers, B. S. Ross, and G. M. Hieftje; "Fundamental Studies of the Sampling
Process in an Inductively Coupled Plasma Mass Spectrometer. Part IE: Monitoring the
Ion Beam", submitted to Spectrochim. Acta B, 1990.
D. M. Chambers and G. M. Hieftje; "Fundamental Studies of the Sampling Process in an
Inductively Coupled Plasma Mass Spectrometer. Part II: Ion Kinetic Energy
Measurements", submitted to Spectrochim. Acta B, 1990.
D. M. Chambers, J. F. Poehlman, P. Yang and G. M. Hieftje; "Fundamental Studies of
the Sampling Process in an Inductively Coupled Plasma Mass Spectrometer. Part I:
Langmuir Probe Measurements", submitted to Spectrochim. Acta B, 1990.
Q. Jin, F. Wang, C. Zhu, D. M. Chambers, and G. M. Hieftje; "A New Atomic Emission
Detector for Gas Chromatography and Supercritical Fluid Chromatography"; accepted in
Appl. Spectrosc. (1990).
Presentations
12- M- Chambers. B. S. Ross, and G. M. Hieftje; “Design Considerations for the Next
Generation ICP-MS Instruments” Paper # 200 presented at the 17th Annual Meeting of the
Federation of Analytical Chemistry and Spectroscopy Societies, Cleveland OH, October
1990.
2 . M- Chambers. P. Yang, and G. M. Hieftje; "Characterization of the First-Stage
Expansion in an ICP-MS Interface"; Paper# 1051 presented at the Pittsburgh Conference
and Exposition, New York NY, March 1990.
2- M- Chambers. P. Y. Yang, A. Verbeek and G. M. Hieftje; "Formation of a Positive
Space-Charge Over the Skimmer Cone of an Inductively Coupled Plasma Mass
Spectrometer and Its Effects on Analyte Ion Kinetic Energy"; Paper #740 presented at the
lb1*1Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy
Societies, Chicago IL, October 1989.
Q. lin, D. M. Chambers, C. Zhu, and G. M. Hiefqe; "A Comparison of Two
Atmospheric Pressure Microwave Plasmas used in Spectroscopic Detection of Supercritical
Fluid Effluents"; Paper #732 presented at the 16^ Annual Meeting of the Federation of
Analytical Chemistry and Spectroscopy Societies, Chicago DL, October 1989.
2- M- Chambers. B. S. Ross, and G. M. Hieftje; "Location of the Optimal Skimming
Position in an Inductively Coupled Plasma Mass Spectrometer"; Paper #595 presented at
the 16^ Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy
Societies, Chicago IL, October 1989.
D. M- Chambers and G. M. Hieftje; "Ion Kinetic Energies in the Low-Pressure Zones of
an Inductively Coupled Plasma Mass Spectrometer", Poster presented at the 37® ASMS
Conference on Mass Spectrometry and Allied Topics, Miami FL, May 1989.
D. M- Chambers. P. Yang, and G. M. Hieftje; "The Role of the Plasma Support Gas on
Ion Kinetic Energy and Mass Bias in a Plasma Source Mass Spectrometer"; Paper #1091
presented at the Pittsburgh Conference and Exposition, Atlanta GA, March 1989.
M- W- Borer. D. Chambers, and G. M. Hieftje; "An ICP/MIP Tandem Source for Atomic
Emission Spectroscopy"; Paper #1295 presented at the Pittsburgh Conference and
Exposition, Atlanta GA, March 1989.
P. Yang. D. M. Chambers, and G. M. Hieftje; "Matrix Effects on Ion Redistribution and
Separation Near the Sample Orifice in an ICP-MS"; Paper #732 presented at the 15®
Annual Meeting of the Federation of Analytical Chemistry and Spectroscopy Societies,
Chicago IL, October 1988.
2- M- Chambers. W-H. Hsu, G. H. Vickers, and G. M. Hieftje; "The Surface-Wave
Induced Microwave Plasma as an Ionization Source for Mass Spectrometry"; Paper #96
presented at the Pittsburgh Conference and Exposition, New Orleans LA, March 1988.
£)• M- Chambers. G. H. Vickers, and G. M. Hieftje; "Plasma Source-Mass Spectrometry
Employing a Surface-Wave-Induced Microwave Plasma"; Paper #370 presented at the
14*" Annual Meeting of the Federation of Analytical Chemistiy and Spectroscopy
Societies, Detroit MI, October 1987.
Документ
Категория
Без категории
Просмотров
0
Размер файла
7 725 Кб
Теги
sdewsdweddes
1/--страниц
Пожаловаться на содержимое документа