Citation
Atomic fluorescence and absorption in an inductively coupled plasma with a continuum source

Material Information

Title:
Atomic fluorescence and absorption in an inductively coupled plasma with a continuum source
Creator:
Mignardi, Michael A., 1959-
Copyright Date:
1989
Language:
English

Subjects

Subjects / Keywords:
Absorption spectra ( jstor )
Atoms ( jstor )
Bonnets ( jstor )
Flames ( jstor )
Fluorescence ( jstor )
Nebulizers and vaporizers ( jstor )
Plasmas ( jstor )
Signals ( jstor )
Ultrasonics ( jstor )
Wavelengths ( jstor )

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright Michael A. Mignardi. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
21969357 ( OCLC )
AHC8148 ( LTUF )
0020758838 ( ALEPH )

Downloads

This item has the following downloads:


Full Text
ATOMIC FLUORESCENCE AND ABSORPTION
IN AN INDUCTIVELY COUPLED PLASMA
WITH A CONTINUUM SOURCE
By
MICHAEL A. MIGNARDI

A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
1989

I




Dedicated to my mother, father, family
and my wife, Marcy, without whose
love and support, this work
would not be possible.




ACKNOWLEDGEMENTS
I would first like to thank James D. Winefordner for having
such an incredible research group. Working with Jim has been a tremendous
experience. I will always try to maintain Jim's philosophy and attitude
towards people and research. My years within this group will always be
remembered. Next, I would like to extend special thanks to Benjamin W.
Smith. Ben is probably one of the best experimentalists I have met. I
have learned a great deal from Ben, and I truly appreciate the patience
he has given me--especially with all of my questions. Ben is a very
important asset to this group. I hope we can stay in touch throughout the
years.
The friends I have made in this group have made research and these
last four years very enjoyable. My first year as a graduate student was
very fruitful as I learned a great deal from Leigh Ann Files in molecular
spectroscopy. People like Brad Jones and Moi Leong will never be
forgotten, they were excellent group members and researchers. Thanks go
to Moi for teaching me all about the ICP. I wish Brad the best of luck
at Wake Forest. Mark Glick has done a tremendous amount for me and the
group concerning computer literacy. I know he will do well in academics.
Joe Simeonsson, Chris Stevenson, and Guiseppe Petrucci have all been great
guys to know. I wish I could continue working with them. I hope they
carry on the traditions of poker nights and PAH cookouts. Thanks go to




Guiseppe for helping me with the ultrasonic nebulizer, its operation and
design, and also with the ICP. Thanks also go to Nancy Szabo for
performing some of the work involving the ICP-AFS system--I wish her good
luck in her future endeavors.
My family has played a major role in my education throughout my
life. I have learned a great deal from my mother and father. Thanks go
to all of them for their love and support. And finally, special thanks
go to my wife, Marcy. Her understanding, patience, love, support, and
most of all, friendship have provided me with continuous encouragement
throughout these last couple of years.




TABLE OF CONTENTS
Pa Fe
ACKNOWLEDGEMENTS................. .................................... iii
ABSTRACT. ............................................................. vii
CHAPTERS
1 ATOMIC SPECTROSCOPY ............................................1
Historical Overview ........................................1
The Inductively Coupled Plasma. ............................10
Radiation Sources. .........................................13
2 A UNIQUE ULTRASONIC NEBULIZATION SYSTEM ...................... 15
Introduction ..............................................15
Design and Construction ...................................18
Operation .................................................25
3 ATOMIC FLUORESCENCE SPECTROMETRY IN AN ICP
WITH A PULSED-CONTINUUM SOURCE. ............................... 27
Introduction................................................27
Experimental. ..............................................28
Results and Discussion....................................34
Conclusion..................................................47
4 DOUBLE-RESONANCE ATOMIC FLUORESCENCE IN AN ICP
INDUCED BY A CONTINUUM SOURCE ................................. 49
Introduction.............................................. 49
Experimental..................................-.-.-................52
Results and Discussion ...................................55
Conclusion................................................60




5 HIGH-RESOLUTION ATOMIC ABSORPTION SPECTROMETRY
IN AN ICP WITH A CONTINUUM SOURCE .............................63
Introduction ..............................................63
Experimental. ..............................................66
Results and Discussion ....................................69
Conclusion ................................................78
6 ATOMIC ABSORPTION SPECTROMETRY IN AN ICP
MOUNTED WITH A "T-SHAPED" BONNET ..............................80
Introduction. ..............................................80
Experimental. ..............................................83
Results and Discussion ....................................91
Conclusion ...............................................101
7 SUMMARY .......................................... ...........102
APPENDICES
A PLASMA TEMPERATURE ...........................................104
3 ELECTRON DENSITY.............................................106
REFERENCES ...........................................................108
BIOGRAPHICAL SKETCH.................................................. 114




Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
ATOMIC FLUORESCENCE AND ABSORPTION
IN AN INDUCTIVELY COUPLED PLASMA
WITH A CONTINUUM SOURCE
By
Michael A. Mignardi
August 1989
Chairman: James D. Winefordner
Major Department: Chemistry
The most commonly used multielement atomic spectroscopy technique is
inductively coupled plasma atomic-emission spectrometry (ICP-AES). The
use of atomic fluorescence spectrometry, however, results in lower
detection limits, greater spectral selectivity, and a reduced emission
background compared to ICP-AES. In this study, the ability to capitalize
on the simplicity of AFS for all elements, as well as the possibility of
exciting all atoms (and/or ions) simultaneously by means of a spectral
continuum light source, seems very attractive when compared with the use
of many line sources (e.g., hollow cathode lamps, HCLs).
Double-resonance fluorescence refers to a process in which atoms (or
molecules) are excited into the fluorescence state in a stepwise manner
with two lasers tuned at appropriate atomic (or molecular) energy levels.
The analytical advantage of such excitation processes as compared to that




involving only one step (single-resonance fluorescence) lies in its
increased spectral selectivity without a significant loss of sensitivity
if both excited state transitions are saturated. In this work, double-
resonance fluorescence signals were studied with a conventional excitation
source.
The choice of atom reservoir and its atomization efficiency is
obviously one of the most important considerations in atomic absorption
spectrometry (AAS). An ideal atom reservoir should provide complete
atomization of all elements in an inert chemical environment with a low
emission background. The most common atom reservoirs used in AAS are
flames and graphite furnaces. There are, however, several advantages of
an ICP as an absorption cell as compared to flames. In this study, a
novel approach was taken to increase the atomic absorption pathlength in
an ICP by using a water cooled quartz "T-shaped" bonnet. This unique "T-
shaped" bonnet was constructed so as to increase the absorption path
length and the analyte residence times within this sample-cell. Atomic
absorption studies were performed using a continuum source. Plasma
diagnostics were also performed to study the plasma temperature as well
as the electron density of the plasma within the "T-shaped" bonnet.

viii




CHAPTER 1
ATOMIC SPECTROSCOPY
Historical Overview
In 1802, Wollaston discovered dark lines in the sun's continuum
spectrum. These lines were further investigated by Fraunhofer, an
optician and instrument maker who had no formal education.1,2 His use
of a spectrometer with a slit produced solar spectral lines. Some of
these spectral lines were dark and one of these dark lines was called the
"D" line. In all, Fraunhofer discovered some 500 dark lines (later to be
called absorption lines) in the solar spectrum.3 In 1820, Brewster
expressed his view that these dark lines were due to absorption by the
sun's atmosphere. Bunsen was able to show with his new burner that
certain salts exhibited colored flames and thus indicated a mean for
identifying elements within salts. Kirchhoff, along with Bunsen,
established the foundation of atomic spectrochemical analysis. Their
experiments involved the flame emission of sodium chloride onto the dark
band of the solar spectrum. The D line was discovered to equal the same
wavelength as the yellow sodium doublet as exhibited by its flame
emission.' Thus, the underlying principles of absorption were established
by Kirchhoff and Bunsen. By the end of the 19th century, the
accomplishments of Newton, Fraunhofer, Bunsen and Kirchhoff had laid a




2
strong foundation for the basis of chemical analysis by atomic spectra.'
In 1866, Angstrom published a table of some 1200 lines by wavelength of
the solar spectrum. Over 800 of those lines were identified as known
elements on earth.4
In 1900, Planck established the quantum law of absorption and emission
of light. An atom is surrounded by orbital electrons that are not
arbitrarily distributed around the nucleus. Each orbit corresponds to a
specific energy level of the atom. If an atom is not perturbed, the most
stable electron configuration is that of the lowest energy orbital, i.e.,
the ground state. In 1911, Rutherford proposed his atomic model where
atoms exist in certain fixed states. Upon absorbing a quantum of energy,
an atom is excited to an energy-enriched state. If an atom absorbs energy
(e.g., from a photon) under certain conditions, one or more of the outer
(valence) electrons becomes promoted to an unstable energy-enriched state
which is more distant from it nucleus. The excited state of this outer
(valence) electron is unstable and thus returns to the ground state after
ca. 107 109 s by one step or a series of steps. The energy of the
emitted radiation (in accordance to Bohr's quantum theory) is
AEatom = E2 El = hv (1)
where E2 and E, are the higher and lower orbital energies (J),
respectively, and are characteristic for each atomic species, h is
Planck's constant (6.626 x 10-" J-s) and v is the frequency of energy
(Hz). If an atom were to absorb photons, it can only absorb light of a




specific frequency of discrete energy such that
Ephoton = h v (2)
where Ephoton is exactly equal to the energy separation between a filled
energy level (E1) and a more energetic (unoccupied) atomic orbital energy
level (E2), i.e., AEatom.4 Atoms in the ground state are capable of
absorbing radiation, thermal energy, and/or electrical energy at discrete
energies that are proportional to particular excited states.
In general, the number of atoms, Ni, excited to energy level i, is
given by
Ni / No = (gi/go0) exp (-Ei / kT) (3)
where No is the number of atoms in the ground state, gi and go are the
statistical weights (dimensionless) of excited and ground states,
respectively (i.e., the statistical weight is the probability that a
particular transition will occur), Ei is the energy of level i (assuming
E0, the ground state energy is zero), k is Boltzmann's constant (1.381 X
10-23 J/K), and T is the temperature (K).3
The mechanisms for atomic emission spectroscopy (AES), atomic
absorption spectroscopy (AAS), and atomic fluorescence spectroscopy (AFS)
are shown in Figure 1. For simplicity, only one excited level is shown
in this figure. In AES, the atom is thermally or electrically promoted
to an excited state where upon it can radiationally decay to the ground

I




Col li si onal
Excitation

Energy level schemes for atomic spectroscopy.

AES

AAS

AFS

Figure 1.




5
state. In AAS, a fraction of the radiant flux from the light source is
absorbed and the attenuated absorbed light is detected. In AFS, a
fraction of the radiant flux from the light source is absorbed and
radiational decay is observed as the atom relaxes back to the ground
state.2 Associated with each excited state of the atom is a very specific
excitation energy. In AAS and AFS, special excitation sources are usually
employed to excite the atoms to their respective selected energy levels.
In contrast, AES is subject to a broad range of energies due to
collisional excitation. Thus, there is little control over the excitation
of atoms in a sample; i.e., many excited energy levels can become
populated in AES. Therefore, the spectra of AES typically is more
complicated than the much simpler spectra of AAS and AFS.
The main instrumental components for atomic spectroscopy and their
configuration are shown in Figure 2. For AES, AAS, and AFS, a sample
introduction system, an atomization cell (which atomizes the analyte), and
a signal detection system is employed. Only for AAS and AFS is a light
source utilized. In AFS and AAS, the light source is positioned at a 900
and 1800 angle, respectively, to the detection system.
Atomic Fluorescence Spectroscopy
AFS is based on the radiational activation of atoms and ions resulting
in a subsequent resulting radiation deactivation, called fluorescence.
AFS is attractive because of its analytical and diagnostic purposes.
Discussion of AFS will be emphasized with the use of an inductively
coupled plasma (ICP) as the atomization cell. The excitation of atoms for
AFS in an ICP has been utilized by electrodeless discharge lamps (EDLs),




Source AFS

AAS

T AES
At om
Cell
'p

Mono

Detect or

Sampl e

General instrumental setup for atomic spectroscopy.

0
0
CO

Figure 2.




7
hollow cathode lamps (HCLs), ICPs and dye lasers. The sample is
introduced into the atomizer and is excited by means of a suitable light
source (HCL, EDL, dye laser, or a second ICP). Considering only two
energy levels in an atom (and the low optical density case) the
fluorescence radiance, BF (WFluoresced m-2 sr-1), can be given by
BF = k/4, Y21 Ev,12 f' k(v) du (4)
where Z is the path length to the detector (m), Y21 is the fluorescence
power efficiency (WFluoresced/WAsorbded), E,12 is the spectral irradiance of
source radiation at the absorption line of frequency v,12 (W m-2 Hz-1), and
J0 k(u) du is the integrated absorption coefficient over the absorption
line (m1 Hz) (where k(u), m-1, is the atomic absorption coefficient at
frequency v).5 Thus, the fluorescence radiance increases linearly with
effective source radiance (if saturation of the upper level is not
achieved).
AFS was first used as an analytical tool by Winefordner and co-
workers.6 8 The use of an ICP and an atomization cell for AFS was first
reported by Montaser and Fassel.9 In AFS, the external radiation source
is focused into the atomization cell and produces excited atoms where a
fraction of the excited atoms radiationally decay (i.e., give off atomic
fluorescence). AFS is ideally suited for multielemental analysis for
several reasons: (i) the physical arrangement of the light source is
simple since the exciting light can be directed at any angle to the
detector (except 1800); (ii) fluorescence radiation is emitted in all
directions (4n steradians); and (iii) AFS spectra are considered simpler




than AAS and AES spectra. 10 ICP-AFS provides great spectral selectivity
compared to other spectroscopic techniques, e.g., ICP-AES and ICP-mass
spectroscopy (ICP-MS). Several comprehensive reviews have been written
for AFS.'-16 AFS is primarily useful for trace analysis.
Atomic Absorption Spectroscopy
In 1833, Brewster recorded the earliest observation of absorption
spectra. As discussed above, the dark lines from the sun's spectrum were
absorption lines. In the early 20th century, Wood performed definitive
experiments of AAS by resonance in gases.' In 1939, Woodson apparently had
the first publication of absorption for quantitative elemental
measurement.1 The birth of AAS as we know it today came from two
independent papers in 1955 by Walsh and by Alkemade and Milatz. 17',18
Walsh's paper was a discussion of how AAS could be a promising method of
chemical analysis with vital advantages over emission methods.17 In AAS,
the external radiation source is focused through the atomization cell and
into a monochromator. The attenuation of the incident radiation by the
analyte atoms is measured. The situation most commonly encountered in
practical AAS analyses is that of a narrow line source used for
excitation. The relationship between atomic absorption and atomic
concentration is given by the integrated absorption coefficient
fo k(v) dv (~re2)/(mc) n1 f12 (5)
where e is the electronic charge (C), m the electronic mass (kg), c the
v.elocity of light (m s-'), nj the number of atoms per m3 (capable of




9
absorbing within the frequency bandwidth, v + du), and f12 (dimensionless)
is the oscillator strength (the average number of electrons per atom which
can be excited by the incident radiation)."7 Considering a beam of
incident radiation, I0, the relationship of the intensity of the
transmitted beam of radiation, I, to I0v is
I, = I0, exp [-k(v)2 ] (6)
where i is the thickness of the atomic gas or vapor (m).17 The physical
conditions (i.e., temperature, pressure, and electric fields) to which the
atoms are subjected and the nature of the atomic transition involved
affect the shape of the absorption line (i.e., the dependence of k(v) on
v). Most treatments in AAS assume that the ground state energy level is
the most populated in analytical atomization cells.19 However, other
situations may exist where other nearby atomic levels are populated as
well. In these cases, the percent of atoms in level i, %Ni, can be
calculated as
%Ni = (gi/Z) exp (-Ei/kT) (7)
where Z, the electronic partition coefficient (dimensionless), corrects
for v-:arying populations in multiplet ground levels and lower excited
levels.
AAS is a good quantitative tool: it utilizes flame or graphite furnaces
that are easy to use, provides excellent sensitivity and limits of
detection (LODs) in the part-per-billion (ppb) range, and requires small




amounts of solution. The principle disadvantages of AAS are (i) only one
element at a time can be measured; (ii) its lack of sensitivity for
refractory elements because of formation of monoxides in low temperature
chemical flames; and (iii) chemical and ionization interferences are
sometimes observed.20
The Inductively Coupled Plasma
The development of ICPs began in 1942 when Babat published his first
paper on the properties of electrodeless discharges in a high frequency
field.21 In the early 1960s, Reed described his ingenious approach to the
stabilization and thermal isolation of these plasmas.22 Reed operated an
argon ICP as a heat source at atmospheric pressure with 4 MHz and 1 to 10
kW, to grow refractory crystals at temperatures from 10 to 20 kK.
Greenfield used a modified torch as that used by Reed and was first to
report the ICP as a light source for solution samples.23 Since then, much
work has been done especially by Fassel24-27 and by Boumans and De Boer28-
30 with an ICP as an emission source.
The ICP torch itself is made of fused quartz of three concentric tubes;
the outer (coolant) tube, the middle (auxiliary) tube, and the central
aerosol injection tube. Figure 3 is a diagram of a typical ICP torch.
The coolant argon gas is introduced tangentially at flows of 10 to 20 L
min-1 and results in cooling, vortex stabilization of the torch, and is
also the plasma gas. The auxiliary gas is used to keep the plasma in a
suitable position within or above the load coil at flows of 0.5 to 2 L
min-. The carrier argon gas flows into the central injection tube at 0.5
to 1 L min-1. The 2 to 4 turn copper water-cooled induction coil is




cool ant- o
auxi 1 iary r -

Diagram of an ICP torch.

r f .
coil

carri er

Figure 3.




12
coupled to a radio-frequency generator. A common oscillator frequency is
27.12 MHz, although 40.68 MHz is now becoming more common.3" The high
frequency currents generate oscillating magnetic fields where lines of
force are axially oriented inside the quartz. Since argon gas is a non-
electrical conductor, a seed of electrons is planted into the torch by a
tesla coil. The "tickled" argon gas then forms a plasma induced by the
axial magnetic fields. The electric field imparts kinetic energy to the
electrons in the plasma which then share this energy with the plasma atoms
and ions by colliding with them. There are, however, several proposed
excitation mechanisms for the ICP (e.g., the Penning ionization reaction,
charge transfer, electron collision, and atom collision). 31 Extensive
research on the mechanisms of operation of an ICP has resulted in the
acceptance of several theoretical models or simulations. The plasma is
an effective atomization source and a sample can be efficiently injected
into the plasma. Typical residence times of the sample in the plasma are
about 2.5 ms. The free atoms flow upstream in a narrow cylindrical
channel which can be easily and efficiently focused onto the slit of a
monochromator.
The r.f. coil, replacing the classical electrode of an arc/spark
discharge, does not come into direct contact with the plasma and thus
eliminates the problem of elemental contamination. 32 The high r.f. field,
however, must be adequately shielded, to avoid interference with other
electronic equipment.33 The number density of free atoms in the hot argon
sheath is very low. Free atoms or ions of the analyte tend to behave as
an optically thin emitting source. The excitation temperature and
electron number density are the most significant properties of an ICP




13
making it a useful excitation source for atomic spectrometry. (In
calculating the plasma temperatures, discrepancies exist due to lack of
a general agreement of published transition probabilities).20 The plasma
is not in a complete state of thermodynamic equilibrium since all energy
distributions can not be described by only one temperature. Instead, the
ICP is considered to be in local thermal equilibrium (LTE).3
The requirements of a good atomizer are (i) good atomization
efficiency; (ii) low radiational background and background flicker; (iii)
low concentration of quenchers; (iv) long residence time of analyte in the
optical path; (v) simplicity of operation; and (vi) low cost of initial
purchase and operation." Compared to flames and electrothermal atomizers,
the ICP is an excellent atomization and ionization cell for virtually all
elements.33 The ICP has a low spectral background compared to other
atomization cells and has remarkable freedom from chemical and physical
interferences.31,33 Argon ICPs are excellent vaporization, atomization,
excitation, and ionization sources for atomic spectroscopy. Several
attractive properties of the ICP are (i) high gas temperatures; (ii)
capability of sustaining the analyte in an inert gas environment ; and
(iii), freedom of contamination from electrodes since none are required.34
However, some disadvantages of the ICP are that is not easy to operate and
it is expensive to purchase and operate.
Radiation Sources
Requirements for good radiation sources are (i) high radiance over the
absorption of analyte atoms; (ii) long term and short term stability
(i.e., little drift and little flicker); (iii) simple tuning and focusing;




14
(iv) availability for all elements; (v) long lifetime; (vi) safety of
operation; and (vii) low cost.11 Compared to continuum sources, line
sources generally have higher radiance over the absorption line, fewer
spectral interferences, and little, if any, wavelength tuning is required.
Compared to line sources, continuum sources offer the utility of one
source for multielement excitation, have better stability, are simpler to
focus, and generally have longer lifetimes.11
Since fluorescence intensity is proportional to source intensity
(assuming no optical saturation), it is desirable to employ high
luminosity sources. Dye lasers appear to be the ideal excitation source
for AFS (except for their cost). The high intensity of lasers shows great
promise of extending the linear dynamic range (LDR) of AFS. EDLs are
commonly employed for AFS and have intensities more than one order of
magnitude greater than HCLs. The main drawback to EDLs is that some good
lamps do not exist or are difficult to make for some elements. ICPs have
also been utilized as an intense line (or continuum) source for AFS,
however, cost is also a factor here.5 HCLs are commonly used in AAS and
AFS. In order to increase their output intensity, these lamps can be
pulsed at high current levels (EDLs can also be pulsed). The xenon arc
lamp is a continuum source. A continuum source used with a monochromator
or interference filters can be readily used for multielemental analysis.
These lamps can also be pulsed to increase their overall peak power.
Their biggest disadvantage is their poor spectral output in the low UV
(i.e., < 250 nm).5




CHAPTER 2
A UNIQUE ULTRASONIC NEBULIZATION SYSTEM
Introduction
Sample introduction is one of the most troublesome aspects in
analytical atomic spectroscopy.35 The goal of sample introduction methods
is to efficiently and reproducibly transfer a representative amount of
sample to the atomization cell. The analytical sample can be a liquid,
solid, or gas. Because many analytical samples are dissolved to form
aqueous solutions, nebulizers are the most common sample introduction
methods investigated.36 The efficiency of nebulization of the sample has
a direct influence on both the sensitivity of the technique and the extent
of interference effects.37 The nature of the atomic population (and the
signals observed) depends on the type of atomizer and sample introduction
method employed.38 Methods of non-pneumatic nebulization include spinning
disks, high voltage sparks, jet-impact, fritted disks, and ultrasonics.36-38
The pneumatic nebulizers include those that are concentric (e.g., the
Meinhard nebulizer) and cross-flow (e.g., the Babington and MAGIC
nebulizer);3739-41 these are the most common nebulizer types utilized in
atomic spectroscopy. The ideal aerosol produced by a nebulizer should be
of high density and of small and homogeneous droplet size.37




16
Ultrasonic nebulizers contain a piezoelectric transducer, a lead-
zirconate-titanate (PZT) ceramic in most cases, which vibrates at
ultrasonic frequencies from 20 kHz to 5 MHz.37'38 The high frequency
voltages applied to the transducer crystal cause it to twist, bend and
shear which lead to mechanical vibration of the crystal.42 As the
analytical solution flows across the face of the vibrating PZT transducer,
standing waves are formed on the surface of the liquid. The wavelength,
A (m), of these waves on the surface of the liquid is
A = [(2 r a) / (p f2)]/3 (8)
where a is the liquid surface tension (N m-1), p is the liquid density (kg
m-3), and f is the ultrasonic frequency (Hz).42 The liquid surface obtains
a sufficient vibrational amplitude to become unstable, and aerosol
droplets of nearly equal diameters are ejected.37 The median droplet
diameters, D (m), are related to the liquid surface wavelength as42
D = 0.34 A (9)
Therefore, the higher the ultrasonic frequency, the smaller the aerosol
droplets.38
The face of the PZT transducer is protected from cavitation and pitting
by the sample with a glass plate that is acoustically matched (i.e., the
glass plate thickness is a half integral of the wavelength of the
ultrasonic vibration)42 to the transducer's resonance frequency. Even with




17
good acoustic matching of the glass plate to the PZT transducer, the
ultrasonic wave is attenuated to some extent.
In the mid-1960s, ultrasonic nebulizers were first suggested as a
replacement for pneumatic nebulizers.43-45 Ultrasonic nebulizers are
electrically driven and are not dependent on the aerosol carrier gas flows
which restrict the flexibility of optimization with pneumatic
nebulizers.38,42'46 Also, unlike pneumatic nebulizers, ultrasonic nebulizers
do not experience clogging or excessively long term memory effects
associated with high analyte salt concentrations. Another drawback of
pneumatic nebulization is their wide droplet size distributions, whereas
ultrasonic nebulizers produce smaller and more uniform particle size
distributions (1.5 to 2.5 Mm range)38 which allow for more efficient
transport and desolvation of the sample aerosol before atomization.4648
The sample transport efficiency-of most pneumatic nebulizers is normally
2 to 10%, whereas, the transport efficiency of ultrasonic nebulizers has
been reported to be as high as 90%.49 With the use of ultrasonic
nebulizers, the sensitivity can often be improved from 5 to 50 fold.
Thus, one can work below previous pneumatic detection limits,
preconcentration steps are not needed, and more dilute samples can be used
to reduce interelement effects.46
Some problems of ultrasonic nebulizers, however, are (i) a more complex
and expensive apparatus is required; (ii) excessive clean out times can
occur and thus lead to less analytical throughput; (iii) short and long
term drifts of the analytical signal can be observed; and (iv) cross
contamination and memory effects can occur within the ultrasonic spray
chamber.42 Many of these problems can be overcome by careful attention to




instrumental design of the ultrasonic system. Careful design
considerations were utilized in construction of an ultrasonic spray
chamber so as to reduce as many of the above mentioned problems as
possible.
Design and Construction
A detailed drawing of the ultrasonic nebulizer used is shown in Figures
4-9. All of the designed apparatus shown in the figures, except for the
PZT transducer, were constructed within our machine and glass shops. The
housing for the PZT transducer was constructed of brass (see Figure 5).
The top (and its screws) of the housing was made of KLF nylon and the
bottom nut was made of TLF teflon (see Figure 6). A 2-021 size 0-ring
(Parker Seal Co., Lexington, KY) was used to seal the PZT transducer with
the housing and its top (see Figure 7). All O-rings used in the housing
were made of N-butyl rubber. The nebulizer spray chamber and the entire
desolvation apparatus were made of borosilicate glass (see Figures 8 and
9). The piezoelectric transducer was purchased from Channel Products Inc.
(Chesterland, OH) and was completely assembled (i.e., with attached glass
plate).
Sample delivery was performed continuously at a constant rate of 3.3
mL min- by a peristaltic pump (Rainin Instrument Co., Inc., Boston, MA,
Model Rabbit). The sample was pumped to a sample delivery tube orifice
(stainless steel, 1 mm i.d.) positioned approximately 1 to 2 mm from the
surface of the PZT transducer and near the upper edge of the transducer
(see Figure 7). Proper sample delivery tube positioning was easily
manipulated by the ball and socket joint.




0-r ing

PZT Transducer

I
I
I
I
I
I
I
I
I
I
I
r1
1
/
I

Wa ter
Out
<-- BNC

Li

Diagram of the complete PZT transducer housing.

a ter
In

Figure 4.




5, 2-56 TPI
0 {9.92
. 15- H

0.50'

Detailed drawing of the cross-section of brass housing.

1. 5'

1.B6
8.19"

1.42'

1. 7'

For 2-028
O-Ring

1/8' NPT

1/2 28 TPI

2.50"

Figure 5.




Hous i ng

op

Side

B. S6

0.135-I

I 0.09
.I I

1. 5"

Clearance for
#2 Screw

Bott

0.23"-

0 m

Hut

0. 50'
5" 18 32 TPI
1/2 20 TPI
1/2" radlus

.74 I
0,.9"

Diagram of the housing top and bottom nut.

Fron t

Bo t tom

Figure 6.




Water In
PZT Transducer
i .1

De I ivery
Tube

BNC
O r n g S
Wa ter ou t
Dra in
Argon

Diagram of the assembled PZT housing and spray chamber.

Figure 7.




1.30"
I

.75

3/8"

2.0"

3. 90"

Detailed drawing of the spray chamber.

18/9
Ba I I
Jo int

1. 0"

o.d.

1.0
1.0"

- 1/ f"

o d .

Figure 8.




1 3 mm
o. d.
14. 0"

ICP -

Demolvati on Tube
10. 0"

o. d.
13 mm
o.d. Double
Surf ace
Condenser
37 mm
o. d.
4 5 mm
o. d.

- 8 mm o d.

Dr al n

Detailed drawing of the desolvation system.

Figure 9.




The unique spray chamber was designed to utilize aerodynamic principles
that promote streamlining of the aerosol (see Figure 8). This spray
chamber should reduce aerosol concentration within the spray chamber,
thereby reducing turbulence and allowing for faster aerosol evaporation.
Streamlining should result in shorter clean-out times, fewer memory
effects, and minimize condensation within the spray chamber itself.
Operation
Power to the PZT transducer was supplied by a Plasma-Therm power supply
(Plasma-Therm, Inc., Kresson, NJ, Model UNPS-1). Maximum power to the
transducer could be obtained by tuning the frequency of the power supply
to achieve minimum reflected power. The measured forward and reflected
operating powers were 37 and 3 W, respectively, and the frequency of
operation was 1.35 MHz. One power lead was hard-wired to the center-most
contact electrode of the PZT transducer. The outer-most contact electrode
of the PZT transducer was bonded to the brass nebulizer housing with a
silver filled epoxy (Epoxy Technology, Inc., MA, EPO-TEK, H27D). The
connection between the power supply and transducer housing was made via
a BNC connector. The ultrasonic energy produced by the transducer is also
transferred to the backing medium, which acts as the coolant for the
piezoelectric. The backing medium (coolant) for the transducer was
flowing chilled deionized water (5 C). To promote smoother draining of
the un-nebulized solution, the nebulizer was operated in a vertical
orientation as described by Olson et al.46
Because of the high nebulization efficiency of the ultrasonic
nebulizer, the excessive solvent mass transport rate to the atomizer could




26
lower the temperature of the ICP and thus, the atomization efficiency of
the ICP. Also, the excess solvent vapor can cause a great loss in aerosol
drop size homogeneity due to aggregation of the droplets.37 This excessive
solvent loading was minimized by an efficient desolvation system (see
Figure 9). The desolvation tube was heated with standard heating tape
(Fisher Scientific, Pittsburgh, PA) to approximately 200 'C. The cooled
condenser (ca. 5 C) was a modified Davies' double surface condenser.
There is a drain following solvent desolvation to allow the condensed
liquid solvent to be removed while "dry" aerosol sample droplets are
transported to the ICP. The same chilled water to cool the PZT transducer
was also used for the desolvation system and ultimately allowed operation
of the entire ultrasonic system for long periods of time ( > 8 h). It
took approximately 65 s from introduction of a sample to stabilization of
the signal. The majority of this time is due to the length of peristaltic
tubing used for sample delivery.




CHAPTER 3
ATOMIC FLUORESCENCE SPECTROMETRY IN AN ICP
WITH A PULSED-CONTINUUM SOURCE
Introduction
The most commonly used multielement atomic spectroscopic technique is
inductively coupled plasma-atomic emission spectrometry (ICP-AES). ICP-
AES achieves nanogram per milliliter (ppb) detection limits, long linear
calibration curves, excellent precision and matrix free measurements for
most elements. 50-52 More recent techniques include the multielement hollow
cathode lamp inductively coupled plasma atomic fluorescence
spectrometer (HCL-ICP-AFS), the inductively coupled plasma mass
spectrometer (ICP-MS), and the continuum source furnace atomic
absorption spectrometer (CSF-AAS).5052 The fluorescence approach results
in lower detection limits, greater spectral selectivity, and a reduced
emission background;53,54 the furnace atomic absorption approach is
applicable to smaller sample amounts than the ICP-AES approach. In this
study, the ability to capitalize upon the simplicity of atomic
fluorescence spectra of virtually all elements as well as the possibility
of exciting all atoms (and/or ions) simultaneously by means of a spectral
continuum light source was investigated.55-57 In order to increase the
source spectral irradiance, 58 especially in the UV, a repetitively pulsed




28
xenon flashtube and a gated detector were used to increase the measured
signal-to-noise ratio. The type of fluorescence observed in this work is
resonance (one-wavelength or one-color) fluorescence. Resonance
fluorescence is where the wavelength of light used for atomic excitation
is the same wavelength as the atomic fluorescence. A representation of
resonance fluorescence is depicted in Figure 10.
The radiation sources typically used in commercial AFS are line sources
(e.g., mostly HCLs). HCLs exhibit strong atomic emission whereas they
provide insufficient intensity for the ionic emission of a specific
element. For some easily ionized elements, it would be advantageous to
optically probe their ionic transitions. The spectral continuum output
from the pulsed xenon light source should be able to probe both the atomic
and ionic transitions of an element. In this study, the experimental
system is described and some initial analytical figures of merit are
given. Also, LODs for some elements will be compared utilizing the
ultrasonic and pneumatic nebulizers.
Experimental
Instrumentation. A schematic diagram of the experimental system used
in this study is shown in Figure 11. The experimental components and
manufacturers are listed in Table I. Source radiation from the pulsed
flashtube was focused into the ICP using two lenses, L, and L2 (both with
diameter and focal lengths of 50 mm): L, collimated the radiant flux from
the flashtube and L2 focused the collimated radiant flux onto the center
of the ICP above the load coil. The diameter of the focused beam was




Cd

59516
59220
43692
31827
228. 8
nm
0

D3

PI

So

Figure 10. Atomic energy level diagram of cadmium showing resonance
fluorescence.




cope

Recorder

PD

L2L
I CP Fl ash
Torch Tube
Figure 11. Schematic diagram of the experimental setup. (

L 3




Equipment
Micropulser power
supply
Flashtube
SRS gated integrator
and boxcar averager
SRS computer interfa
Digital oscilloscopE
Monochromator
Photomultiplier
PMT high-voltage
supply
Trigger photodiode
Radio frequency
generator
ICP Plasma-Therm
torch
ICP Plasma-Therm
concentric nebulizer
Microcomputer
Peristaltic pump

Table I. Experimental
Model
457A
Novatron-722
SR 250
ice SR 245
2430A
EU-700-77
R 928
412A
FND 100
HFP 1500D
Standard
and long
Type A
PC-XT
Rabbit

components
Manufacturer
Xenon Corporation, Woburn MA
01801
Xenon Corporation
Stanford Research Systems, Inc.,
Palto Alto, CA 94306
Stanford Research Systems, Inc.
Tektronix, Inc., Beaverton, OR
97077
GCA/McPherson Co., Acton, MA
01720
Hamamatsu, Waltham, MA 02145
John Fluke Mfg., Co., Inc.,
Seattle, WA
EG&G Electro-Optics, Salem, MA
01970
Plasma-Therm Inc., Kresson, NJ
08053
Precision Glassblowing of
Colorado, Parker, CO 80134
Precision Glassblowing of
Colorado
International Business Machines
Corp., Boca Raton, FL
Rainin Instrument Co., Inc.,
Boston, MA




32
approximately 8 mm. Since resonance fluorescence was the signal detected,
several precautions were taken to reduce scattered light reaching the
=onochromator from the flashtube. This was accomplished with a light trap
placed behind the ICP directly in line with the flashtube (see Figure 11)
and with blackened baffles set up around the ICP. The collection lens,
1 (diameter 76 mm and focal length 178 mm), was also enclosed within a
blackened tube to further reduce scattered light. Despite these
precautions, the limiting noise of the system was still due to scattered
light from the excitation source.
The 300 W xenon flashtube was enclosed within a fan-cooled housing
which contained a front surface spherical mirror (diameter 50 mm and focal
length 31.5 mm) and collimating lens L,. The lamp was operated from a
pulsed power supply at 5 kV. A 0.2 pF discharge capacitor provided an
input energy of 5 J per flash, a flash half-width of ca. 680 ns, and a
lamp peak power of 4.2 kW. The lamp was pulsed at a repetition rate of
20 Hz. To help reduce radio frequency (r.f.) noise, the entire lamp
housing was surrounded and grounded with copper wire cloth acting like a
Faraday cage. The fluorescence radiation was collected at a 90* angle to
the excitation beam and a 1:1 image of the ICP was focused with a third
lens, L3 onto the entrance slit of the monochromator (focal length = 350
m, reciprocal dispersion = 20 A/mm, and aperture ratio f/6.8). To avoid
overfilling the monochromator collimator, an iris diaphragm was placed
between L3 and the monochromator. A small fraction of the exciting light
from the flashtube was reflected with the edge of a mirror to a photodiode
which triggered the boxcar averager. The photo-current pulse produced by
the photomultiplier tube (PMT) was terminated through a 1000 ohm load




33
resistance directly into the boxcar input. The resulting signal pulse has
a full width to half maximum (FWHM) of ca. 1.5 ps. The boxcar delay time
(the time between the trigger pulse and the start of the measurement) was
700 ns. The gate width (the time during which fluorescence was measured)
was 1.8 ps for all cases. Thirty signals (i.e., thirty lamp flashes) were
averaged for each output signal. The "busy out" of the boxcar averager
triggered a Stanford analog-to-digital (A/D) system in order to measure
the output signal.
Horizontal and vertical translation of the ICP torch was
accomplished with two single axis translation mounts. This allowed
horizontal translation of the ICP torch up to ca. 100 mm and vertical
translation of the ICP torch up to ca. 80 mm. The ICP concentric
pneumatic nebulizer was fed with a peristaltic pump to allow for lower
sample rate uptake and thus reduce salt encrustation in the torch. The
ultrasonic nebulizer was also fed with a peristaltic pump as described in
chapter 2.
Reagents and Procedure. All components of the experimental system
were operated according to the directions given in the manufacturers'
manuals. The chemicals used in preparation of the stock solutions were
all reagent grade. The preparation of standard solutions were made with
compounds best suited for the solutions as determined by Parsons et al.9
Distilled demineralized water (Barnstead Sybron Corportation, Boston, MA)
was used throughout this study. Standard solutions were obtained by
serial dilutions of the stock solutions.




34
The majority of this work involved the use of a pneumatic (Meinhard)
nebulizer and a conventional ICP torch. For each element studied, the
optimal ICP r.f. power, observation height above the ICP load coil,
monochromator slit width, and ICP gas flows were determined. Calibration
curves and limits of detection (LODs) were also determined for each
element and compared to some results obtained with the ultrasonic
nebulizer. In measuring the synthetic mixtures containing five elements,
compromise values of the above mentioned parameters were necessary. ICP-
AFS results were also compared using an extended ICP torch versus a
conventional torch. An extended ICP torch has an outer sleeve that is 40
mm longer than the conventional torch.
Results and Discussion
Molecular species of non-refractory elements typically have low
molecular dissociation energies and are easily atomized by the plasma at
lower r.f. powers, whereas those of the refractory elements typically have
high molecular dissociation energies and are atomized by the plasma at
higher r.f. powers. Also, the atomic and ionic populations of the sample
species in the plasma are often greatly affected by the choice of r.f.
power and observation height.58'9 For this study, the dependence of the
fluorescence on the r.f. power and observation height were examined
independently, as shown in Figures 12 and 13, respectively. In these two
figures, the plots for each element were arbitrarily shifted for clarity
in interpreting the graph.
Figure 14 shows the variation in fluorescence signal-to-noise with
monochromator slit width for each element (at optimal r.f. power and




Ba Cd(L)
- Na
E-
'- Cd
Ca
I I I I I I I I I I I
10 15 20 25 30 35 40 45 50 55 80 65
OBSERVATION HEIGHT (mm)
Figure 12. The effect of the observation height above the ICP load coil
on the fluorescence signal for each element.




400

Figure 13.

800

1200

R.F. POWER (W)
The effect of r.f. power on the fluorescence signal for each
element.




56do' 10'oo
SLIT WIDTH

i5'oo
(/j M)

Figure 14. The variation in fluorescence signal-to-noise ratio for each
element as a function of spectrometer slit width.

2000
C
1600-
N a
1200-
800.
8 0 0-
400- Cd
I

S 2000

I




38
observation height). Table II lists the optimal experimental conditions
(found by univariate search) for each element, based on Figures 12-14.
Typical calibration plots obtained under the optimal conditions are
shown in Figure 15. These plots were also arbitrarily shifted for
clarity. (Figures 12 15 were all obtained with a pneumatic nebulizer.)
Table III lists the analytical figures of merit using a pneumatic
nebulizer. The detection limit is defined as the concentration in ig mL-1
(ppm) of the element in pure aqueous solution resulting in a signal that
is three times the standard deviation of the blank measurements. The
electronic band-width for all measurements was ca. 1 Hz. Previously
reported ICP-AFS detection limits for the same elements are also listed
in Table III. All the log-log calibration plots have slopes between 0.96
and 1.06. Typical with atomic fluorescence, the calibration curves cover
a wide linear dynamic range. The deviation from linearity at the higher
concentrations in Figure 15 is due to pre- and post-filter effects.
The effect of using a long torch was examined for cadmium as a model
element. The results are also shown in Figures 12-15 and the
corresponding curves are labeled "Cd(L)". With the long torch, the
fluorescence signal was practically constant at observation heights of 45
to 60 mm above the load coil; whereas with the standard torch, the signal
rapidly decreased with increase in observation height (Figure 12). The
variation in signal with increase in r.f. power was similar in pattern for
both torches, but the signal was significantly larger with the standard
torch (Figure 13). A similar difference was observed for the effect of
monochromator slit width on the signal-to-noise ratio, and again the
standard torch gave much the better performance (Figure 14). Despite




Table II. Optimal experimental conditions

r.f. power

Observation
height (mm)

Slit width

(pm)

2000

1500

500

2000

1000

Ba(II)
Ca(II)
Cd(I)
Na(I)
V(II)
Cd(I)

Mixturett 200-800

* Other experimental conditions: sample uptake
pneumatic nebulizer pressure 31 psig; plasma Ar
auxiliary Ar flow rate 1-3 L min-1

rate 1.15 mL min ;
flow rate 15 L min-';

(I) indicates an atomic line, (II) an ionic line (singly ionized).
Analyzed with a long torch.
SA mixture of the 5 elements at concentrations of 20 pg mL-1 for each
element.

Elementt

455.4
393.4
228.6
589.6
292.4
228.6

AAFs,
(nm)

1000

600

1500




Na
Cd
Cd(L)
Ca

lissal ll u ni n anis|I l lli ..ll ..lli .. llillll I 1 111 1ie 11i1llillu s i[1111ill1 1 1111 I ff i
-2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
LOG [CONCENTRATION (ppm)]

Figure 15. Analytical calibration curves for each element.




Analytical figures of merit using a pneumatic nebulizer.

Sensitivity*
(mV mL pg-1)

14.1
18.4
26.3
10.7
1.88
14.9

LOD
(pg mL-1)

0.09
0.03
0.02
0.04
0.4
0.04

Literature LOD,6o
(pg mL-')

0.05
0.0004
0.0005
0.0003
0.1
ND

Referred to boxcar input.
See footnote t to Table II.
Measured with a long torch
Not done (no literature value available)

Element

Ba(II)

Ca(II)

Cd(I)

Na(I)

V(II)
Cd(I)t

Table III.




42
this, the detection limits for cadmium were essentially the same with both
torches (Table III).
Table IV shows the analytical figures of merit for several elements
using the ultrasonic nebulizer as the sample introduction device. The
LODs in this table are also compared to some literature values.1o In both
Tables III and IV, the literature values reported are values obtained by
HCL-ICP-AFS. Thus, since HCLs were used, only the atomic transitions of
a particular element was studied. As seen in Table IV, the LOD for Ba(II)
(Ba(II) designates an ionic transition whereas Ba(I) designates an atomic
transition) was approximately 20 times better than that obtained from the
literature value. Also, the LODs for Mg(I) and Ca(II) were an order of
magnitude poorer than those obtained in the literature. Reasons for this
can be attributed to poorer lamp intensity around the Mg(I) line. Also,
since the excitation energy of Ca(II) to an upper ionizational level is
higher than that of Ba(II), lower atomic populations within the ionization
levels for Ca(II) may have been achieved with the ICP. The LODs for Pb(I)
and V(II) were essentially the same as the literature values. Overall,
the LODs obtained with this setup (a continuum source in combination with
an ultrasonic nebulizer) should be comparable to line sources used with
pneumatic nebulizers in AFS. The most intense fluorescence line of each
element except vanadium was excited and observed (see Table II). The most
intense fluorescence signals for vanadium are at 309 to 310 nm, i.e., in
the middle of a strong OH band.61 Therefore, the next most intense
fluorescence line of vanadium at 292.4 nm was observed; which is also
attributable to a poorer LOD than expected. In general, with the
ultrasonic nebulizer and the continuum xenon flashtube source, LODs were




Table IV. Analytical figures of merit using an ultrasonic nebulizer.

Element AAFS
(nm)

Mg(I)
Ca(II)
Ba(II)
Pb(I)
V(II)

285.2
393.4
455.4
405.7
292.4

Log-Log
Slope

0.952
1.24
1.06
1.08
0.992

%RSD

9.
1.24
8.
7.
13.

LOD Literature LOD60
(pg mL- ) (pg mL-)

0.003
0.002
0.002
0.03
0.3

0.0005
0.0004
0.05
0.07
0.1

S See footnote t in Table II.




44
within one order of magnitude poorer or better than the reported
literature values. Calibration curves generated with the ultrasonic
nebulizer were similar to those generated with the pneumatic nebulizer as
shown in Figure 15. The analytical precision (% relative standard
deviation, %RSD) of this experimental system is rather poor (ca. 10%).
Although high, a large portion of this poor precision is due to
inefficient mixing of the aerosol within the spray chamber of the
ultrasonic nebulizer. The precision of the pneumatic nebulizer was less
than 5%.
Table V shows a comparison of three elements (Ba, Ca, and V) studied
using both the pneumatic (Meinhard) and ultrasonic nebulizer. The
ultrasonic nebulizer showed an improvement in the LODs for Ba and Ca by
a factor of 10. V, however, showed no improvement in using the ultrasonic
nebulizer. Since V is a refractory metal, its signal is more sensitive
to factors that affect the plasma temperature (e.g., r.f. power) than the
signals for Ba or Ca. In all cases, the log-log analytical curve slopes
were near unity and the linear dynamic range (LDR) for each element was
at least two orders of magnitude. In one instance, high salt
concentrations were aspirated into both the pneumatic and ultrasonic
nebulizer. The ultrasonic nebulizer did not suffer from a long clean-out
time nor memory effects as that experienced by the pneumatic nebulizer.
No pulsation of the carrier gas flow into the ICP was noticed with use of
this ultrasonic nebulization apparatus.
Figure 16 is a wavelength scan of 1 pg ml-1 magnesium showing atomic
and ionic fluorescence at 285.2 and 279.5 nm respectively. This figure
shows the ability of one source, the spectral continuum xenon flashtube,




Table V. Comparison of the analytical figures of merit
of a pneumatic and ultrasonic nebulizer.
Pneumatic/Ultrasonict
Nebulizer

Log-Log

Slope

LOD

LDR*

(pg mL-)

0.97 / 1.1
0.99 / 1.2
1.1 / 0.99

0.09 / 0.002
0.03 / .002
0.4 / 0.3

f See footnote f in Table II.
* LDR in orders of magnitude.
A value to the left of the diagonal is for the pneumatic nebulizer
and a value to the right of the diagonal is for the ultrasonic
nebulizer. All other experimental components and parameters are
identical.

Element

Ba(II)

Ca(II)

V(II)

3.0 / 2.0
2.5 / 2.0
3.0 / 2.5




Mg ( I )

Mg(II)

280 290
Wavel enet h

300
( nm)

Figure 16.

Wavelength scan of magnesium showing a) its atomic and ionic
fluorescence and b) a background (blank) spectrum.

270




47
to probe both the atomic and ionic transitions of magnesium. It would be
difficult to use a magnesium HCL to sufficiently probe the ionic
transition of magnesium.
Figure 17 shows a multielement scan of a synthetic mixture of the
five elements (20 pg mL-1 each), with a standard torch. Note the
simplicity of the spectrum in this figure. As seen in Figure 16, the
atomic and ionic fluorescence lines of calcium at 422.7 and 393.4 nm,
respectively, are also observed here.
Conclusion
The work presented thus far shows promising results for the use of
a pulsed xenon flashtube in ICP-AFS. The present system would appear to
have considerable use in multielement analysis. Although the detection
limits for each element were at least an order of magnitude poorer than
the best reported in the literature for the pneumatic nebulizer, the
results with the ultrasonic nebulizer seem promising. Together, the
ultrasonic nebulizer and the spectral continuum xenon flashtube are an
attractive combination. Future work on this project will involve studies
to improve the detection limits so as to be competitive with the present
commercial atomic spectrometry systems. Improvements could be made by
better r.f. shielding, decreased excitation source scatter, and better and
more efficient collection of the fluorescence signal. Since the limiting
noise in this system was due to scattered light from the flashtube,
efforts will be made to reduce scattered light by observing non-resonance
fluorescence and by using better light traps.




Ca(II)

Ca(I)
/B
Ba

Na

200 300 4
Fl uorescence

500 600
wavelength

Figure 17. A multielement analysis of a synthetic mixture of five
elements.

Cd

OH

(nm)

0 0




CHAPTER 4
DOUBLE-RESONANCE FLUORESCENCE IN AN ICP
INDUCED BY A CONTINUUM SOURCE
Introduction
When an atom (ion) absorbs a photon and subsequently re-emits a
photon of the same spectral frequency as that absorbed, one refers to this
as single-resonance (single-step) fluorescence. The types of atomic
(ionic) fluorescence not considered as single-resonance fluorescence are
referred to as non-resonance fluorescence: double-resonance (two-step or
two-color) fluorescence is an example of this type. Double-resonance
fluorescence refers to a process in which atoms (or molecules) are excited
into the fluorescence state in a stepwise manner with two lasers tuned at
appropriate atomic (or molecular) energy levels.62 Figure 18 is an energy
level diagram of cadmium showing a representation of double-resonance
fluorescence. Here, the excitation of a selected atomic level is achieved
with a single laser tuned to a particular transition frequency, while a
second laser, coincident in time and space to the first laser, is tuned
to a higher excitation level starting from the level reached by the first
laser (or from a collisionally populated nearby level).63 The analytical
advantage of such excitation processes as compared to that involving only
one step (single-resonance fluorescence) lies in its increased spectral
selectivity without a significant loss of sensitivity if both excited
49




Cd

59516 X I.

59220-
64
43692-
31827-
22E

3
D3
1
- Dg
361
C.

P2
So

Figure 18. Atomic energy level diagram for cadmium showing double-
resonance fluorescence and the relevant transitions studied
in this work.

hi'1




51
state transitions are saturated. The CV and visible spectra of atoms
(ions) in a multielement mixture can show a high degree of spectral
overlap within this spectral region. In order to obtain simple spectra,
either the spectral resolution can be increased or the number of spectral
lines within the spectral window can be reduced. Double-resonance
fluorescence can reduce the number of spectral lines within a spectrum
because of its selectivity. Several analytical studies performed with
tunable lasers in atmospheric pressure atomizers have been reported in the
literature.64-68
In principle, when using spectrally unfiltered excitation sources,
several observed fluorescence lines might be a result of double-resonance
excitation processes. This possibility seems to have been overlooked so
far in most analytical studies performed in flames and other atomizers,
especially when continuum sources (e.g., xenon arc lamps) are used for
excitation. It is also clear, however, that the requirements needed for
a successful use of the double-resonance fluorescence approach are not met
by conventional excitation sources; i.e., non-laser sources. On the
contrary, such processes will increase the possibility of observing
spectral interferences in the analysis of complex sample matrices. Since,
to the best of my knowledge, double-resonance fluorescence signals have
not been studied with conventional excitation sources, I felt it would be
interesting to investigate if, and to what extent, such processes could
be observed.
The aim of this study was to describe the observations made with
cadmium atoms in an inductively-coupled argon plasma illuminated with a
spectral continuum xenon pulsed-flashtube. Cadmium was chosen because the




52
pertinent transitions could easily be isolated from one another with the
proper choice of optical filters.
Experimental
The experimental system is similar to the one described previously.69
As discussed in chapter 3 an ultrasonic nebulizer was used as the aerosol
generation system for sample introduction. The unique nebulizer system
was described in chapter 2. The experimental operating conditions are
listed in Table VI. The source ICP nebulizer was fed with a peristaltic
pump (Rainin Instrument Co., Inc., Model Rabbit).
A stock solution of cadmium was prepared as described by Smith and
Parsons. 70 A working standard solution of cadmium was made by serial
dilutions of the stock solution. All dilutions were made with distilled,
demineralized water (Barnstead Sybron Corp.). A 100 pg mL-1 solution of
cadmium was used throughout the study. All solutions were stored in
opaque polyethylene bottles.
The optical filters used throughout the study are commercially
available (Corion Corp., Holliston, MA) and are listed in Table VII. The
detection system was a Stanford analog-to-digital (A/D) system (Stanford
Research Systems, Inc.). The lamp radiance and monochromator-detector
assembly were not calibrated.




Table VI. Experimental operating conditions.

ICP

Forward power
Reflected power
Ar coolant flow rate
Ar auxiliary flow rate
Ar carrier flow rate
Observation height *
Ultrasonic nebulizer
Forward power
Reflected power
Sample uptake rate
Pulsed-Flashtube
High voltage
Pulse rate
Pulse width

450 W
0 W
11.5 L min-
1.1 L min-1
0.8 L min-1
35 mm
37 W
3W
3.3 mL min-I
4.5 kV
20 Hz
700 ns

Miscellaneous

Spectrometer bandpass
PMT high voltage

2 nm
-800 V dc

Above the ICP load coil.




Table VII. Percent transmittance of the optical filters (%).
Corion filter number

Monochromator
Wavelength (nm)

LG 595

0.04

0.06

WG 320

0.05
91

BG 25

0.03
78

92 0.02

644 83




Results and Discussion
The partial energy level diagram for cadmium pertinent to this work
is shown in Figure 18. As already discussed by Omenetto et al., for laser
excitation, this scheme involves a connected double-resonance absorption
process when the first and the second excitation steps share a common
level (e.g., the 1P1 level).64 Moreover, because of collisional coupling
between the 1P, and 3P2 levels, the second excitation step might also be
chosen at Aex = 361.0 rnm (i.e. starting from the 3P2 level), in which case
one can name the process "disconnected" double-resonance excitation. In
this case, since the source is a spectral continuum, both processes can
be effective. In either case, fluorescence emission should be observed at
643.8, 361.0 rnm and at any other allowed transition origination from the
ID2, D3 or nearby levels. As an example, Figure 19 shows a fluorescence
spectral scan in the region of 330-370 nm. The spectrum was obtained by
aspirating a solution of 100 fg mL1 Cd into the plasma. As seen in Figure
19, several lines are observed, the most intense one at 361.0 nm. The
other two lines which can be identified are at 340.3 and 346.6 nm. Since
the 3P2 and 1P, levels might be "thermally" populated in the plasma, it
remains to be proven that the observed signals are due to a double-
resonance excitation mechanism rather than to a single-resonance
excitation mechanism proceeding from a thermally populated level. We have
restricted our observation to the two fluorescence transitions as
indicated in Figure 19 (i.e., AFL = 361.0 and 643.8 nm). It could be
easily proven that these fluorescence transitions were not a result of
ICP-excited Cd emission, flashtube scatter, nor blank background. The
fluorescence intensities of these lines (i.e., AFL 361.0 and 643.8 nm)




Cd Scan

340
Wavel

I
350
engt h

360
(nm)

Figure 19.

Partial fluorescence spectrum of cadmium (100 pg mL-1). No
optical filters were used between the flashtube and the ICP.

I
330

370
370




57
are listed in Table VIII with relative units of 3.33 and 1.08,
respectively. These signals were measured without any type of optical
filter placed between the flashtube and ICP. From here on, these measured
signals will be referred to as the "observed non-filtered fluorescence."
To prove that double-resonance fluorescence was indeed being
observed, a series of optical filter were placed between the flashtube and
ICP. UV and visible wavelength transmission scans were performed on each
filter and the percent transmissions of each filter at A = 228, 361, and
644 nm are recorded in Table VII. Figure 20 shows the wavelength
transmission for each filter within the Cd energy level diagram and the
observed fluorescence signal. When a long wave pass filter (Corion # LG
595) was used, neither the AFL = 361.0 nor 643.8 rnm spectral lines were
observed (Table VIII). If the 'P, level had been significantly thermally
populated, then we should have expected (based on the filter transmission
characteristics) approximately 83% of the observed non-filtered
fluorescence spectral line at AFL = 643.8 nm. Therefore, the observation
of the AFL = 643.8 nm spectral line in the other cases is a result of some
type of two-step photon excitation (whether of the connected or
disconnected type).
When another long wave pass filter (Corion # WG 320) was used, both
the A, = 361.0 and 643.8 nm spectral lines were observed (Table VIII).
Based on the filter transmission characteristics, we expected 91% and 92%
of the observed non-filtered fluorescence signal for AFL = 361.0 and 643.8
rm, respectively. However, only 50% and 13% of the expected signals at
-. = 361.0 and 643.8 nm, respectively, were observed. This indicated that




Table VIII. Double-resonance results.
Fluorescence signals
(Relative units)

Experimental
conditions

AFL = 361.0 nm
Expected' Observed

AFL = 643.8 rnm
Expected' Observed

No Filter NA2 3.33 NA2 1.08
LG 595 0.0 0.0 0.89 0.00 A
WG 320 3.0 1.6 0.99 0.13 B,C
BG 25 2.6 1.2 0.00 0.16 D,E
1 Based strictly upon resonance fluorescence processes (involving a
thermal population of a lower absorbing level and taking into account
the filter transmission characteristics, i.e., expected values were
determined by multiplying the percent transmittance of each filter
by the "observed non-filtered fluorescence" for each wavelength).
2 Not Applicable, since lamp radiance and monochromator-detection
assembly were not calibrated.
A Therefore, negligible thermal population of the 'P, level exists.
B Approximately 50% of the signal observed at AFL = 361 nm is a result
of thermal population of the 3P2 level.
C The observation of the signal at AFL = 643.8 nm (ca. 13%) is a
result of the thermal population of the 3P2 level, photon excitation
(AFL = 361 nm) to the 3D3 level, then collision quenching to the 1D2
level resulting in stepwise fluorescence of the AFL = 643.8 nm line.
D Again, approximately 50% of the signal observed at AFL = 361 nm is
a result of the thermal population of the 3P2 level.
E Although no signal at AFt = 643.8 rnm was expected, approximately 16%
of the observed non-filtered fluorescence at this wavelength was
observed due to the same explanation given in comment C.

Comments




No Fi 1 t er

X-
43. 81
888. 8

3
D 3
Do
P1
- Ps

1
so

Observed
Fl uorescence
381 > 643.8
WG 320
Dx
643.8 361
P1
a
P a
1
so
30
Observed
Fluorescence
361 >> 643.8

LG 595
3
Da
843. 8
1
P1
Pa
so
Observed
Fl uorescence
NONE

BG 25
8-
361

D8
1
- P
- Ps

1
so
Observed
Fl uorescence
361 >> 643. 8

Figure 20. Atomic energy level schemes indicating the different excitation
transitions isolated by the different optical filters used.
The transmittance of the three filters is given in Table VII.




60
about 50% of the AFL = 361.0 nm spectral line is a result of thermal
population of the 3P2 level, followed by single-resonance fluorescence.
The observation of the AFL = 643.8 rnm is also a result of thermal
population of the 3P2 level, followed by photon excitation to the 3D3
level, then collisional quenching to the 'D2 level and fluorescence at
643.8 runm.
When a bandpass filter (Corion # BG 25) was used, both the AFL =
361.0 and 643.8 rnm spectral lines were observed (Table VIII). Based on
the filter transmission characteristics, we expected 78% and 0% of the
observed non-filtered fluorescence signal at AFL = 361.0 and 643.8,
respectively. However, only approximately 50% of expected signal at AFL
= 361.0 nm was observed. Also, approximately 16% of the observed non-
filtered signal at AFL = 643.8 rnm was observed. The observation of
fluorescence at 361.0 and 643.8 nm follows the same explanation as that
given for the Corion # WG 320 long wave pass filter.
Conclusion
From the series of experiments described in this paper, the following
conclusions can be given: (i) there is negligible thermal population of
the 1P, level; (ii) approximately 13% of the observed fluorescence signal
for AFL = 643.8 nm is due to the stepwise fluorescence resulting from
thermal population of the 3P2 level, photon excitation of Aex 361.0 rnm to
the 3D3 level, followed by collisional quenching to the 'D2 level resulting
in fluorescence at AFL = 643.8 run; (iii) approximately 87% of the observed
fluorescence signal at AFL = 643.8 rnm is due to connected double-resonance
fluorescence from photon excitation (Aex = 228.8 nm) to the 'P, level




61
followed by photon excitation (Aex = 643.8 nm) to the 3D3 level; (iv) there
is "thermal" population of the 3P2 since approximately 50% of the observed
non-filtered fluorescence signal at AFL = 361.0 nm was observed when both
the Corion # WG 320 and BG 25 filters were used (thereby blocking the A
= 228.8 nm spectral line); and (v) the other 50% of the observed
fluorescence signal at AFL = 361.0 nm is a result of double-resonance
fluorescence either as a) photon excitation (Aex = 228.8 nm) to the 1P,
level, photon excitation (Aex = 643.8 nm) to the 1D2 level, collisional
coupling to the 3D3 level, followed by photon emission of AFL = 361.0 nm
(connected scheme), or as b) photon excitation (Aex = 228.8 rnm) to the 1P,
level, collisional quenching to the 3P2 level, followed by photon
excitation and emission of A = 361 nm (disconnected scheme). However, the
explanation as described in (va) is more likely to occur since the
collisional quenching from P1 3P2 as described in (vb) should be less
favored in the argon ICP.
In summary, double-resonance fluorescence with our experimental setup
utilizing a xenon pulsed-flashtube was observed. The analytical
advantages of double-resonance fluorescence as discussed earlier with line
sources (Ref. 64-68) will not be utilized with our continuum source. On
the contrary, as already stated, the ability to observe double-resonance
fluorescence with a continuum source will actually be a disadvantage as
this can present a great potential for spectral interferences using a
multi-element matrix. Since a continuum source is a non-selective
excitation source, some of the selectivity advantages as seen with line
sources can be gained only if the proper choice of optical filters is
utilized, which would be practical only in a limited number of cases. For




62
example, in this work, the 228.8 nm line could not be solely isolated
because of the lack of a suitable optical filter.




CHAPTER 5
HIGH-RESOLUTION ATOMIC ABSORPTION SPECTROMETRY
IN AN ICP WITH A CONTINUUM SOURCE
Introduction
The choice of atom reservoir and its atomization efficiency is
obviously one of the most important considerations in atomic absorption
spectrometry (AAS). An ideal atom reservoir should provide complete
atomization of all elements, in an inert chemical environment with a low
emission background. The most common atom reservoirs used in AAS are
flames and graphite furnaces. There are, however, several advantages of
an ICP as an absorption cell as compared to flames:7173 (i) the higher
temperature and longer residence times of a species in the plasma leads
to a more efficient atomization; (ii) the chemical environment can be
better controlled than that of a flame (e.g., free atoms have longer
lifetimes in the chemically inert argon environment of an ICP); (iii) the
formation of chemical compounds (e.g., stable refractory compounds) is
reduced in the inert argon environment of the ICP; and (iv) refractory
solid samples are more efficiently vaporized. There are, however, several
disadvantages: 747 (i) there is a larger dilution of gaseous analyte
atoms within the plasma of the ICP; (ii) the absorption path length is
relatively small; (iii) relatively high ion and atom excited states can




64
be populated leading to lower ground state atom or ion populations and
more intense and noisy emission background; and (iv) an ICP may be more
expensive and more costly to operate than either flame systems or
electrothermal atomizers.
Conventional radiation sources in AAS have been line sources (e.g.,
HCLs, EDLs, and vapor discharge lamps VDL). The ideal radiation source
for AAS must meet several requirements: 76 (i) have a high spectral
irradiance at all wavelengths; (ii) exhibit little short term fluctuations
and long term drift in emission intensity; and (iii) require a minimum
amount of operator maintenance for optimum performance. A major
disadvantage of most line sources is that only one element per source can
usually be investigated. While the use of a continuum source offers the
possibility of multielement determinations, until the availability of high
pressure xenon arc lamps, continuum sources (such as tungsten filament
lamps) had little use in AAS because of their relatively low spectral
output in the ultraviolet spectral region. 76 Moreover, the requirement of
a high resolution spectrometer has made the continuum source less
attractive for AAS. Several excellent reviews have appeared in the
literature discussing continuum source atomic absorption spectrometry
(CSAAS).77-80 One of the substantial benefits of CSAAS is the ability to
perform good background correction at any wavelength.38,81,82 Several
authors have had success using wavelength modulation for background
correction and this approach has since become the method of choice for
background correction in CSAAS.8387
Analytical work in the field of ICP-AAS has been extremely limited.
Despite the short absorption path length of the ICP viewed normal to the




gas flow (never greater than ca. 0.5 cm), ICP atomization and ionization
in CSAAS experiments offers the same advantages as flame atomizers; easy
sample introduction, high sample throughput, minimal sample pretreatment
requirements, and the ability to directly introduce liquid samples as a
continuous stream. The advantages of CSAAS with flame atomization are
well-documented: simple optical configuration, single source requirement,
simultaneous multielement detection, excellent background correction, and
low detection limits (above wavelengths of 250 nm),77-0 which should also
hold true for CSAAS with ICP atomization.
Although low source spectral irradiance below 250 nm is the biggest
disadvantage for CSAAS, other disadvantages include the need for a high
resolution monochromator to fully resolve the absorption line, and the
need for multiple detectors (e.g., photomultiplier tubes) for simultaneous
multielement analysis. This work presents an ICP atomization (and
ionization) CSAAS system with photodiode array detection. As opposed to
conventional scanning spectrometry, a photodiode array allows for fast
data processing and storage of spectra and the ability to make
simultaneous multiwavelength measurements.88 The photodiode array also
allows absorbance versus wavelength spectra to be obtained in real time
and any background absorption or spectral interferences could be directly
observed and corrected for by using the manufacturer's computer software.
This eliminates the need for wavelength modulation or other background
correction techniques.




Experimental
Instrumentation. Figure 21 is a schematic diagram of the optical
arrangement used in this work. The light source was a 300 W Cermax
compact xenon arc lamp (ILC Technology, Sunnyvale, CA). A portion of the
collimated radiation from the lamp was directed through the ICP by a
series of diaphragms. A fused silica lens, L, (5 cm focal length and
diameter) focused the collimated radiation after the ICP at the entrance
slit of a J-Y H10 0.1 m focal length monochromator (Instruments SA, Inc.,
Metuchen, NJ) acting as a pre-disperser. Optical apertures were
positioned between the lens and the ICP to spatially discriminate the
transmitted source beam from ICP emission. The 16 nm bandpass of radiant
flux at the exit slit of the H10 was focused as a one-to-one image by a
second fused silica lens, L2 (7.6 cm focal length and 5 cm diameter) onto
the entrance slit of an HR1000 1.0 m focal length monochromator (ISA).
The reciprocal linear dispersion of the HR1000 is 0.5 nm mm- in first
order. The photodiode array positioned at the exit plane of the HR1000 was
an Optical Spectrometric Multichannel Analyzer (OSMA, Model IRY-1024,
Princeton Instruments, Princeton, NJ). With the HRI000 monochromator
slit width set between 0.075 and 0.100 mm, the spectral bandpass in first
order was less than 0.05 rnm. Since this resolution was not sufficient to
fully resolve atomic absorption lines (the full width at half maximum,
FLM, for atomic lines are typically 0.01 rn or less), spectral
measurements were obtained with the monochromator set to pass radiation
in the fourth or fifth order. In the fifth order, the spectral bandpass
was less than 0.01 nm. Because of the limited length of the photodiode
array (ca. 2.54 cm), the spectral window in fifth order was limited to




"O -I-t = I LaempXe
~LiT-----------ILamp

Figure 21.

Experimental system for CSAAS-ICP.




68
2.5 nm. The pre-disperser simply eliminated lower order source radiation
and plasma emission. The experimental operating conditions of the ICP and
ultrasonic nebulizer are the same as listed in Table VI (chapter 4). The
ICP was operated at a low forward power to reduce the analyte's atomic and
ionic emission. Horizontal and vertical translation of the ICP torch was
accomplished with two single axis translation mounts. This allowed
horizontal translation of the ICP torch up to 100 mm and vertical
translation of the ICP torch of up to 80 mm. A laboratory-constructed
ultrasonic nebulization system was used for sample introduction.
Ultrasonic nebulization improved the nebulization efficiency of the
analyte and decreased the dilution of gaseous analyte atoms within the
ICP. The experimental setup of the ICP is further described elsewhere.69
The operating current for the 300 W xenon arc lamp was 17 A.
Sample Preparation. Stock solutions (1000 mg L-1 in 2 % HNO3) were
obtained from Inorganic Ventures, Inc. (Brock, NJ) for the 14 elements
tested. Several synthetic mixtures were also made consisting of 2 to 3
elements. Standard solutions were obtained by serial dilutions of the
stock solutions with distilled, demineralized water (Barnstead Sybron
Corp.).
Procedure. All components of the experimental system were operated
according to the directions given by the manufacturer. Aqueous samples
were nebulized into the ICP by the ultrasonic nebulizer. As mentioned
previously, the fifth order of an atomic absorption line was usually
selected on the HR1000. The upper wavelength limit for the HR1000 was
1500 nm; therefore, for elements with atomic absorption peaks above 300
nm, detection with fourth order dispersion was necessary. One complete




scan of the photodiode array required 33 ms: 300 data acquisition spectra
were averaged to achieve a total acquisition time of 10 s.
For each atomic absorption spectrum obtained, a blank spectrum (water
aspirated into the ICP) of the incident radiation (10) and a dark spectrum
(D) (all radiation blocked at the entrance slit of the HR1000
monochromator) was also obtained. The emission observed (Ie), if any, with
the source radiation blocked was subtracted from the transmittance signal
(I'a) showing the absorption of an atomic or ionic line such that the true
transmittance signal (I) was calculated as I = Ia Ie. The absorption
spectrum was calculated for each spectrum using the computer software
supplied with the OSMA and the equation A = log (I0-D)/(I-D). Analytical
calibration curves were plotted as absorbance peak height versus
concentration. The LOD for each analyte was calculated with a confidence
level of 98.3% as that absorbance equal to three times the standard
deviation of the blank spectrum (IO).
Results and Discussion
Absorption Spectra. Both ionic as well as atomic absorption
transitions may be observed in the ICP. The atomic and ionic populations
are greatly affected by choice of the r.f. power and observation height
of the plasma.73,89 The r.f. power was maintained at 500 W to reduce atomic
or ionic emission. The observation height within the plasma was also
optimized for the best atomic or ionic absorption. Observation heights
(above the r.f. load coil) of 27 mm and 37 mm were used to study the ionic
and atomic transitions, respectively. Unlike conventional line sources,
which typically have strong radiation output at only the atomic




70
transitions, continuum sources allow one to optically probe ionic
transitions as well. This may prove useful since many elements are
efficiently ionized within an ICP. Molecular and atomic absorption
measurements have been reported for CSAAS with photodiode array detection;
however, poor atomic line resolution was observed.90-91 As previously
mentioned, the observation of fifth and fourth order spectra was necessary
to achieve sufficient atomic resolution. Figure 22 is an absorption
spectrum obtained for a 10 ppm magnesium solution. Unlike work performed
previously with a flame,92 molecular OH absorption bands in the vicinity
of the Mg atomic absorption line were not observed. Figures 23 and 24 are
absorption spectra for 10 ppm and 100 ppm solutions of manganese and
nickel, respectively. The simultaneous observation of the manganese
triplet absorption lines and the many nickel absorption lines demonstrates
the greater informing power of the technique compared to single wavelength
conventional AAS. The observation of more than one element in a mixture
is shown in Figure 25, an absorption spectrum of a synthetic mixture of
100 ppm each of copper and silver. This absorption spectrum shows the
prominent silver absorption line at 328.07 nm and a strong, but not
prominent line for copper at 327.40 nm. Even though copper has a stronger
absorption line elsewhere, it can still be detected at the mg L-1 (ppm)
level in this limited spectral window. Several spectral windows can be
judiciously chosen to observe several different elements simultaneously
and quickly.
The ability to observe both atomic and ionic absorption lines within
a single spectral window is shown in Figure 26, a three-dimensional
absorption spectrum of a synthetic mixture of 100 ppm of both magnesium




0.45
S0.35
C
0
L.
0 0.25
()
0.15
0.05

Wavelength (nm

Figure 22. Atomic absorption spectrum of 10 pg mL-1 Mg.




0.23
Q,
O
E 0.18
0
L..
0
CO
0.13
0.08 -
279.2

Wavelength

Atomic absorption spectrum of 10 mg mL- Mn.

Figure 23.




0.21
0.19
Q)
C-
C
U
00.17
-o3
0
-0.15
0.13
0.11 -
231.00

Ni Ni
S Ni I

231.40

Atomic absorption spectrum of 100 mg mL-1 Ni.

Wavelength

Figure 24.




0.15 -
0.13
G.)
C0.11
0
0
0.09
0.07

0.05 1 I I I I I I
327.00 327.50

Wavelength

. i 1 ,
328.00 328.50
(nm)

Figure 25. Atomic absorption spectrum of a synthetic mixture of 100 pg
mL- each of Cu and Ag.

I .




V
.16-
;4
0
M 11--
.0
279.
Figure 26.

2

50
45 4
4o
40
35
25
isl I II ~ st l gI III I an ui gs ug en ss ue ss ge 1 5 C )
279. 4 279. 8 279. 8 2B0. 0 280. a 880. 4
W avel engt h ( n m)
A three dimensional atomic absorption spectrum of 10 jg mL-1
each of Mg and Mn showing relative absorbance as a function of
observation height above the plasma load coil.




76
and manganese. Three manganese atomic lines at 279.48, 279.83 and 280.11
nm and two magnesium ionic lines at 279.55 and 280.27 nm are observed.
Atomic and ionic absorption is shown as a function of plasma observation
height. At increasing plasma observation heights, the ground state ionic
population of magnesium and thus the ionic absorbance decreases; however,
the ground state atomic population and absorption of manganese appears to
remain fairly constant over the range of absorption heights shown in
Figure 26.
Background Correction and Spectral Interferences. In this work, the
multiwavelength capabilities of the photodiode array were utilized to
acquire background corrected spectra without the need for special
background correction techniques.9394 When measuring the absorbance peak
height, the spectral baseline was automatically subtracted in the final
spectrum using the supplied software. Both Figures 24 and 25 exhibit a
great deal of noise in their spectral baselines, which is attributable to
a combination of signal and source shot noise and detector noise.
Analytical Figures of Merit. Detection limits for the 14 elements
studied are reported in Table IX. Except for the atomization cell, the
experimental system used was identical to one used previously. 92 For most
elements, the detection limits by CSAAS in the ICP were at least an order
of magnitude poorer than our previous results in a flame. The detection
limits found for this work are approximately two orders of magnitude worse
than with the best previously reported detection limits for CSAAS in an
air-acetylene flame.77 Several causes for this are (i) the photodiode
array has a lower sensitivity than more conventional detectors such as




T
Elementa Ags
(rnm)
Ag(I) 328.1
Ca(I) 422.7
Cd(I) 228.8
Co(I) 240.7
Cr(I) 357.9
Cu(I) 324.7
Fe(I) 248.3
Mg(I) 285.2
Mn(I) 279.5
Na(I) 589.0
Ni(I) 232.0
Sr(I) 460.7
Zn(I) 213.9
Ba(II) 455.4
Ca(II) 393.7
Mg(II) 279.5
Mn(II) 257.6
Sr(II) 407.8

able IX. Limits
Limits
This Work

5.
3.
3.
10.
10.
4.
8.
0.7
4.
3.
10.
6.
20.
10.
5.
2.
10.
6.

iousd
ork
.2

of Detection
of Detection
Literatureb
Value
0.007
0.003
0.03
0.07
0.02
0.01
0.07
0.001
0.01
0.003
0.07e
0.02
0.07

(I) = atomic transitions, (II) = ionic transitions.
Taken from Reference 77 (CSAAS-Flame).
Taken from reference 95 (HCL-AAS-Flame).
Taken from reference 92 (CSAAS-Flame).
Measured nickel line at 352.4 nm for only the literature value.

Observed for AAS.
(mg L-)
Linee Prey
Source W
0.001 0
0.002
0.001
0.004 1
0.003 2
0.002 0.
0.005 1
0.0003 0.
0.002 0.
0.0004 0.
0.005 10
0.006
0.001

2
3
2
05




78
photomultiplier tubes; (ii) the shorter absorption pathlength of the ICP,
namely < 0.5 cm versus 5 10 cm (as seen with flames and graphite
furnaces); and (iii) the optical throughput of our system was lower
because of the use of two monochromators (especially with the HR1000
monochromator in the fifth order). The best detection limits reported for
commercially available line-source flame AAS instruments are still an
order of magnitude better than the best CSAAS results, especially within
the UV spectral region. Despite the relatively poor detection limits of
the ion absorption lines (see Table IX), the ability to probe the ionic
absorption transitions is advantageous especially with regard to the
possibilities it offers for diagnostic studies in the ICP. The linear
dynamic range for each element was between one and two orders of
magnitude, with the calibration curves deviating significantly from
linearity above 0.1 absorbance units. This range can be greatly
increased, however, by selecting less sensitive absorption lines or by
measuring the absorbance off the absorption line peak when the major lines
reach absorbances greater than 0.1.
Conclusions
The ICP has several advantages as an atom reservoir for atomic and
ionic absorption spectrometry. The ability to obtain real time,
simultaneous multiwavelength measurements makes the photodiode array much
more attractive than conventional scanning spectrometers with conventional
detectors for the measurement of CSAAS spectra. The optical throughput
of our HRI000 monochromator should significantly improve with a grating
blazed to enhance wavelengths from 200-300 rnm in fifth order. Also, the




79
ability to measure several elemental atomic or ionic absorption lines
simultaneously, improves the informing power of the entire system.
In flame-AAS, one is able to probe sufficiently only the atomic
transitions. In an ICP, many elements are readily ionized and therefore
the possibility of obtaining absorption spectra of virtually all atoms and
ions simultaneously with a continuum light source is very attractive. The
serious disadvantage of a short absorption pathlength should be overcome
with a specially designed ICP torch.




CHAPTER 6
ATOMIC ABSORPTION SPECTROMETRY IN AN ICP
MOUNTED WITH A "T-SHAPED" BONNET
Introduction
Since the beginning of flame AAS, workers have tried to enhance the
analytical sensitivity by increasing the absorption path length.96 Fuwa
and Vallee performed a study of the Beer-Lambert law for absorption of
molecules in solution as a model for the investigation of atomic
absorption. 97 They discovered that the absorption sensitivity is a
function of the length of the atomic absorption path length. In their
work, a long path absorption cell (90 to 250 cm) was made of Vycor for
flame AAS. The combustion flame burner was positioned such that the tail
flame was directed into the absorption cell. Source radiation from an HCL
passed through the absorption cell and then through focusing optics onto
a spectrograph. Compared to conventional flame AAS, they were able to
increase the sensitivity for some elements by two orders of magnitude with
this long path absorption cell.9798 Other work using this long path
absorption cell showed that the diameter, and reflection from the inner
walls of the absorption tube affect the absorbance.99 The smaller the
diameter (up to 1 cm) the higher will be the concentration of absorbing
atoms and hence the higher the absorbance.97'99 Earlier stages of the work




81
by Fuwa and Vallee involved placing the tube in a T-shape fashion over the
flame burner and thus deflecting the flame in two direction;97 however,
this system failed to yield satisfactory results with AAS. Rubeska also
used Fuwa long absorption cells in a T-shaped fashion over a flame. In
this setup, the absorption cell was electrically heated so as to increase
the mean lifetime of free atoms with the absorption cell.96 With the long
path absorption cell, good sensitivities were obtained and the formation
of oxides for some elements was reduced since the tube shields the flames
gases from the oxidizing atmosphere.
In 1966, Wendt and Fassel utilized a multipass system for ICP-AAS.72
In their setup, a collimated beam from an HCL made three passes through
an ICP with mirrors placed around the plasma and then into a spectrometer.
Veillon and Margoshes used a modified Wendt and Fassel ICP torch (without
the most central concentric tube) using a single pass system of a
modulated source and phase sensitive amplifier.73 They did not observe the
chemical interferences commonly observed in flames. Veillon and Margoshes
concluded that except for a few refractory elements, the ICP does not
appear to be a suitable replacement for the chemical flame in AAS. Like
Fuwa and Vallee, Greenfield conducted work involving a physical increase
in the absorption path length, however, in an ICP.100 Greenfield was able
to extend the outer sheath of the ICP torch to 24 inches, and later found
he was able to bend the extended plasma "tail-flame" at right angles.
With further work along this discovery, the outer sheath of the ICP took
on the shape of a "T". When one end of the "T" was closed with an optical
flat, the plasma "tail-flame" was forced to turn away from the optical
flat. A modulated HCL was collimated and sent through the optical flat




82
and into a spectrometer near the open end of the "T". Temperatures within
the "T" were recorded from 3000 to 8000 K. The high temperature in the
"T" should minimizes chemical interference due to the formation of stable
refractory compounds of the element under investigation. Also, a lower
background radiation was emitted by the tail-flame, thus indicating an
advantage over combustion flames in AAS.
Almost a decade later (1977), Mermet and Trassy performed work on
increasing the absorption path length in a ICP.75 Here, the absorption
path length was increased by having the light beam pass through the plasma
along the axis symmetry. The plasma torch was turned on its side and a
beam of radiation, focused with a lens at the base of the torch, was sent
down the torch axially. The outer sleeve of the torch was extended to
further enhance the lifetime of neutral atoms in the plasma. Other work
by Magyar and Aeschbach, involved removing the atomization cell from a
commercial instrument and replacing it with an ICP.71 The enhanced atomic
emission signal from the ICP made it necessary to use higher intensity
HCLs than is sufficient for flame AAS. Their results indicated the
possible use of ICP-AAS for the determination of metals in complex
compounds, which do not efficiently dissociate in combustion flames.
Downey and Nogar placed an ICP within the cavity of a pulsed, flashlamp-
pumped dye laser for AAS and achieved and enhancement factor of
approximately 170 relative to single-pass absorption in an ICP.74 This
technique is known as intracavity dye laser absorption spectroscopy
(IDLAS). An HCL was used as an optogalvanic detector in this system.
LODs in the parts-per-million range were obtained for sodium and barium.
Barium ion absorption was also studied because it is easily ionized.




83
Other authors have performed some successful fundamental studies on ICP
argon chemistry using AAS.101-103
In this work, further investigation has been performed utilizing an
ICP for AAS studies. An attempt to increase the absorption path length
of the ICP was accomplished with the design of a cooled quartz "T-shaped"
bonnet for the ICP. Unlike Greenfield's "T" shaped torch, this bonnet
was detachable from the torch and was positioned closer to the load coil.
An evaluation of this "T-shaped" bonnet was performed for ICP-AAS studies
for pure and multielement solutions with a continuum source.
Experimental
Design and Construction of the "T-shaped" bonnet. A detailed
drawing of the "T-shaped" bonnet is shown in Figure 27. The bonnet was
constructed entirely of quartz tubing (Thermal American Fused Quartz,
Monteville, NJ). The bonnet was constructed within our glass shop.
Tubing for the "T" portion consisted of two concentric tubes. The tubing
used for the outside portion was 25 mm i.d. and 28 mm o.d., and the tubing
used for the inside portion of the "T" was 17 mm i.d. and 19 mm o.d..
These two tubes were fused together at their ends so as to create a water
jacket between them. The length of the "T" was 115 mm long. The tubing
for the stem of the "T" was 22 mm i.d. and 25 mm o.d., and was 25 mm long.
The tubing used for the intake and return of chilled water to the "T-
shaped" bonnet was 6 mm i.d. and 8 mm o.d.. Chilled water to the bonnet
was kept at < 5 *C (Neslab Instruments, Inc., Portsmouth, NH). The flow




115 mm
I

7 mm
.d.

0 d.

2 5 mm
o. d

Schematic diagram of the "T-shaped" bonnet.

mm

128
0.

Figure 27.




85
rate of water to the bonnet was approximately 2-3 gallons per hour. The
"T-shaped" bonnet as mounted in two nylon clamps. The clamps were
constructed within our machine shop and were attached to an X, Y, Z-
translation axis mount such that the "T-shaped" bonnet could carefully be
positioned on top of the ICP. Figure 28 is a diagram of the bonnet within
the clamps. This translation also allowed the ability to optically probe
different regions within the "T-shaped" bonnet. The ICP torch was also
mounted on X, Y, Z-translation axis mount such that the torch could be
properly position within the "T-shaped" bonnet. The torch translation was
very useful during plasma ignition.
Operation. Before plasma ignition, chilled water was allowed to
flow through the "T-shaped" bonnet and ICP load coil for approximately 5
min. This ensured that the bonnet was properly cooled. Figure 29 is a
schematic diagram of the ICP torch positioned within the "T-shaped"
bonnet. Ignition of the plasma with the torch inside the "T-shaped"
bonnet was extremely difficult; therefore, before plasma ignition the ICP
torch had to be lowered such that the bottom of the bonnet stem was at
least 20 25 mm above the top of the outer quartz sheath of the torch.
Once the plasma was ignited in this lowered position, the torch was then
slowly raised into the "T-shaped" bonnet until the center of the
longitudinal "T" portion of the bonnet was approximately 27 mm above the
torch load coil. Access to all translation axis mounts was well away from
the r.f. load coil of the ICP and thus could be safely manipulated during
torch operation. Once in position, the torch was operated as usual. The
experimental operating conditions are listed in Table X. As expected, the




00Figure 28. Diagram of the bonnet within nylon clamps.
Figure 28. Diagram of the bonnet within nylon clamps. 7




Water Out

T-shaped
Bonnet

Water In
ICP
Torch

Figure 29. Diagram of the ICP torch within the bonnet.




Table X. Experimental operating conditions.

I. ICP

Forward r.f. power
Reflected r.f. power
Ar coolant flow rate
Ar auxiliary flow rate
Ar carrier flow rate
Observation height*
II. Ultrasonic nebulizert
III. Xenon arc lamp
Power
Current

400-600 W
0-20 W
17 L min-1
0.5 L min-1
0.9 L min-1
27 mm
300 W
17 A

* Measured above the load coil.
See Table VI




89
torch was operated at low powers (i.e., 400 to 600 W) so as not to
overheat and possibly crack the "T-shaped" bonnet. At these low ICP
powers, this experimental setup could be run continuously (the maximum
continuous operation at one time used was 8 h).
Instrumentation. Figure 30 is a schematic diagram of the optical
arrangement used in this work. The optical probing of atomic and ionic
species within the center of the "T-shaped" bonnet was 27 mm above the ICP
load coil. As discussed in chapter 5, the instrumentation (except for the
"T-shaped" bonnet), sample preparation, and experimental procedure were
identical.
Plasma Temperature and Electron Density. Spectroscopic measures are
one of the best means for obtaining spatially and temporally resolved
measurements of temperature and species number densities without
perturbing the mechanisms of the plasma or influencing the
temperature.104,105 An accurate knowledge of the plasma temperature can
lead to a better understanding of analyte vaporization, dissociation,
atomization and ionization processes.106 As discussed by Browner and
Winefordner, a two-line atomic absorption method was employed.104 The
elements thallium and indium were chosen based on their working ranges of
temperatures. The advantage of using this two-line atomic absorption
method are that the temperatures are averaged through the optical path in
the plasma. A spectroscopic emission method was also performed as a
comparison to the two-line absorption method as mentioned above. Neutral
iron was selected as the thermometric species. A factor in the line
selection process for iron was based on (i) maximal spread in excitation
levels; (ii) freedom from plasma spectral interferences; (iii)




ICP

Figure 30. Schematic diagram of the experimental system.

I o2




91
availability of accurate transition probabilities; and (iv), wavelength
proximity, which eliminates the need for calibrating detector
instrumentation response with respect to wavelength.105 The wavelengths,
excitation energies, statistical weights and transition probabilities were
taken from a reference by Fuhr, Martin, and Wiese.107
The method used for determination of relative electron (number)
density was based on the measurement of the Stark broadening of an atomic
hydrogen line. Stark broadening is due to the interaction of charged
particles in a dense plasma. The electron density is proportional to the
Stark broadening and the electric field strength. The Stark half-width
is the most reliable and convenient method for the determination of
electron densities.108 The H. (486.1 nm) line was chosen because it is
generally free from plasma spectral interferences, has a sufficient
intensity for measurement, a small half-width (1 5 A), and extensive
Stark data is available for this line.109.110 Both Doppler broadening and
instrumental broadening are usually negligible compared to Stark
broadening. 09
Results and Discussion
After extended use of this experimental system, and especially after
aspiration of high salt concentration into the plasma, an oxide formation
was observed within the inner wall of the "T-shaped" bonnet. Despite this
residue formation, there were no analyte memory effects observed within
the system. After periodic use, the "T-shaped" bonnet was removed and
immersed in a 10% HNO3 solution to clean the inner walls of the bonnet.
Absorption spectra, background correction and analytical figures of merit




were obtained as described in chapter 5. Detection limits for some
elements studied are reported in Table XI. These results are compared to
early work done to improve the absorption path-length in an ICP. In all
cases, except the work done by Mermet, the LODs obtained with this setup
were superior. Mermet and co-workers were able to achieve excellent
results by viewing the ICP axially.111 Detection limits for some elements
are also reported in Table XII. These results are compared to absorption
data taken from previous work (see chapter 5) as well as some literature
values. As expected, better LODs (by 1 to 2 orders of magnitude) were
achieved with this work as compared to AAS results with a conventional
torch due to the increase in atomic (ionic) absorption path length and
longer analyte residence times within the "T-shaped" bonnet. However,
these results were still poorer compared to the best continuum source and
line source in flame AAS (LODs of 1 and 1 to 2 orders of magnitude poorer,
respectively).
As discussed by Browner and Winefordner, the working temperature
ranges for thallium and indium are restricted: thallium--3650 to 7750 K
and indium--1220 to 3500 K.'04 Atomic absorption for thallium (the 535.1
nm line) was not observed; therefore, the plasma more than likely has a
temperature less than 3650 K since it does not sufficiently populate an
upper degenerate ground state of thallium. Atomic absorption results were
thus obtained using indium. Based on this two-line absorption method, the
plasma temperature at an r.f. power of 500 W was found to be 2165 K. As
a comparison, the atomic emission of iron was also used to determine the
plasma temperature within the "T-shaped" bonnet. As described by Kalnicky
and Kniseley, the temperature calculation was performed with the aid of




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EQGXD4Y0T_RXJI76 INGEST_TIME 2017-10-31T17:42:00Z PACKAGE UF00101023_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES


xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EY3ZR3UWZ_TE1NTV INGEST_TIME 2017-07-11T21:29:04Z PACKAGE UF00101023_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES



PAGE 1

ATOMIC FLUORESCENCE AND ABSORPTION IN AN INDUCTIVELY COUPLED PLASMA WITH A CONTINUUM SOURCE By MICHAEL A. MIGNARDI A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 1989

PAGE 2

Dedicated to my mother, father, family and my wife, Marcy, without whose love and support, this work would not be possible.

PAGE 3

ACKNOWLEDGEMENTS I would first like to thank James D. Winefordner for having such an incredible research group. Working with Jim has been a tremendous experience. I will always try to maintain Jim's philosophy and attitude towards people and research. My years within this group will always be remembered. Next, I would like to extend special thanks to Benjamin W. Smith. Ben is probably one of the best experimentalists I have met. I have learned a great deal from Ben, and I truly appreciate the patience he has given me-especially with all of my questions. Ben is a very important asset to this group. I hope we can stay in touch throughout the years The friends I have made in this group have made research and these last four years very enjoyable. My first year as a graduate student was very fruitful as I learned a great deal from Leigh Ann Files in molecular spectroscopy. People like Brad Jones and Moi Leong will never be forgotten, they were excellent group members and researchers. Thanks go to Moi for teaching me all about the ICP. I wish Brad the best of luck at Wake Forest. Mark Glick has done a tremendous amount for me and the group concerning computer literacy. I know he will do well in academics. Joe Simeonsson, Chris Stevenson, and Guiseppe Petrucci have all been great guys to know. I wish I could continue working with them. I hope they carry on the traditions of poker nights and PAH cookouts Thanks go to

PAGE 4

Guiseppe for helping me with the ultrasonic nebulizer, its operation and design, and also with the ICP. Thanks also go to Nancy Szabo for performing some of the work involving the ICP-AFS system-I wish her good luck in her future endeavors My family has played a major role in my education throughout my life. I have learned a great deal from my mother and father. Thanks go to all of them for their love and support. And finally, special thanks go to my wife, Marcy. Her understanding, patience, love, support, and most of all, friendship have provided me with continuous encouragement throughout these last couple of years.

PAGE 5

TABLE OF CONTENTS Page ACKNOWLEDGEMENTS m ABSTRACT vii CHAPTERS 1 ATOMIC SPECTROSCOPY 1 Historical Overview 1 The Inductively Coupled Plasma 10 Radiation Sources 13 2 A UNIQUE ULTRASONIC NEBULIZATION SYSTEM 15 Introduction 15 Design and Construction 18 Operation 25 3 ATOMIC FLUORESCENCE SPECTROMETRY IN AN ICP WITH A PULSED -CONTINUUM SOURCE 27 Introduction 27 Experimental 28 Results and Discussion 34 Conclusion 47 4 DOUBLE -RESONANCE ATOMIC FLUORESCENCE IN AN ICP INDUCED BY A CONTINUUM SOURCE 49 Introduction 49 Experimental 52 Results and Discussion 55 Conclusion 60

PAGE 6

5 HIGH-RESOLUTION ATOMIC ABSORPTION SPECTROMETRY IN AN ICP WITH A CONTINUUM SOURCE 63 Introduction 63 Experimental 66 Results and Discussion 69 Conclusion 78 5 ATOMIC ABSORPTION SPECTROMETRY IN AN ICP MOUNTED WITH A "T-SHAPED" BONNET 80 Introduction 80 Experimental 83 Results and Discussion 91 Conclusion 101 7 SUMMARY 102 APPENDICES A PLASMA TEMPERATURE 104 3 ELECTRON DENSITY 106 REFERENCES 108 BIOGRAPHICAL SKETCH 114

PAGE 7

Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Doctor of Philosophy ATOMIC FLUORESCENCE AND ABSORPTION IN AN INDUCTIVELY COUPLED PLASMA WITH A CONTINUUM SOURCE By Michael A. Mignardi August 1989 Chairman: James D. Winefordner Major Department: Chemistry The most commonly used multielement atomic spectroscopy technique is inductively coupled plasma atomic-emission spectrometry (ICP-AES) The use of atomic fluorescence spectrometry, however, results in lower detection limits, greater spectral selectivity, and a reduced emission background compared to ICP-AES. In this study, the ability to capitalize on the simplicity of AFS for all elements, as well as the possibility of exciting all atoms (and/or ions) simultaneously by means of a spectral continuum light source, seems very attractive when compared with the use of many line sources (e.g., hollow cathode lamps, HCLs) Double-resonance fluorescence refers to a process in which atoms (or molecules) are excited into the fluorescence state in a stepwise manner with two lasers tuned at appropriate atomic (or molecular) energy levels. The analytical advantage of such excitation processes as compared to that

PAGE 8

involving only one step (single-resonance fluorescence) lies in its increased spectral selectivity without a significant loss of sensitivity if both excited state transitions are saturated. In this work, doubleresonance fluorescence signals were studied with a conventional excitation source The choice of atom reservoir and its atomization efficiency is obviously one of the most important considerations in atomic absorption spectrometry (AAS) An ideal atom reservoir should provide complete atomization of all elements in an inert chemical environment with a low emission background. The most common atom reservoirs used in AAS are flames and graphite furnaces. There are, however, several advantages of an ICP as an absorption cell as compared to flames. In this study, a novel approach was taken to increase the atomic absorption pathlength in an ICP by using a water cooled quartz "T-shaped" bonnet. This unique "Tshaped" bonnet was constructed so as to increase the absorption path length and the analyte residence times within this sample-cell. Atomic absorption studies were performed using a continuum source. Plasma diagnostics were also performed to study the plasma temperature as well as the electron density of the plasma within the "T-shaped" bonnet.

PAGE 9

CHAPTER 1 ATOMIC SPECTROSCOPY Historical Overview In 1802, Wollaston discovered dark lines in the sun's continuum spectrum. These lines were further investigated by Fraunhofer, an optician and instrument maker who had no formal education. 1 2 His use of a spectrometer with a slit produced solar spectral lines. Some of these spectral lines were dark and one of these dark lines was called the "D" line. In all, Fraunhofer discovered some 500 dark lines (later to be called absorption lines) in the solar spectrum. 3 In 1820, Brewster expressed his view that these dark lines were due to absorption by the sun's atmosphere. Bunsen was able to show with his new burner that certain salts exhibited colored flames and thus indicated a mean for identifying elements within salts. Kirchhoff, along with Bunsen, established the foundation of atomic spectrochemical analysis. Their experiments involved the flame emission of sodium chloride onto the dark band of the solar spectrum. The D line was discovered to equal the same wavelength as the yellow sodium doublet as exhibited by its flame emission. 1 Thus, the underlying principles of absorption were established by Kirchhoff and Bunsen. By the end of the 19th century, the accomplishments of Newton, Fraunhofer, Bunsen and Kirchhoff had laid a

PAGE 10

2 strong foundation for the basis of chemical analysis by atomic spectra. 1 In 1866, Angstrom published a table of some 1200 lines by wavelength of the solar spectrum. Over 800 of those lines were identified as known elements on earth. 4 In 1900, Planck established the quantum law of absorption and emission of light. An atom is surrounded by orbital electrons that are not arbitrarily distributed around the nucleus. Each orbit corresponds to a specific energy level of the atom. If an atom is not perturbed, the most stable electron configuration is that of the lowest energy orbital, i.e., the ground state. In 1911, Rutherford proposed his atomic model where atoms exist In certain fixed states. Upon absorbing a quantum of energy, an atom is excited to an energy-enriched state. If an atom absorbs energy (e.g., from a photon) under certain conditions, one or more of the outer (valence) electrons becomes promoted to an unstable energy-enriched state which is rnore distant from it nucleus. The excited state of this outer (valence) electron is unstable and thus returns to the ground state after ca. 10~ 7 10~ 9 s by one step or a series of steps. The energy of the emitted radiation (in accordance to Bohr's quantum theory) is AE atom E 2 E 1 = hu (1) where E 2 and E : are the higher and lower orbital energies (J), respective^/, and are characteristic for each atomic species, h is Planck's constant (6.626 x lO -34 J-s) and u is the frequency of energy (Hz). If an atom were to absorb photons, it can only absorb light of a

PAGE 11

specific frequency of discrete energy such that -photon h v (2) where E photon is exactly equal to the energy separation between a filled energy level (E x ) and a more energetic (unoccupied) atomic orbital energy level (E 2 ), i.e., AE atom 4 Atoms in the ground state are capable of absorbing radiation, thermal energy, and/or electrical energy at discrete energies that are proportional to particular excited states. In general, the number of atoms, N i excited to energy level i, is given by N i / N = (gi/go) exp (-Ei / kT) ( 3 ) where N is the number of atoms in the ground state, g i and g are the statistical weights (dimensionless) of excited and ground states, respectively (i.e., the statistical weight is the probability that a particular transition will occur), E L is the energy of level i (assuming E the ground state energy is zero), k is Boltzmann's constant (1.381 X 10" 23 J/K) and T is the temperature (K). 3 The mechanisms for atomic emission spectroscopy (AES) atomic absorption spectroscopy (AAS) and atomic fluorescence spectroscopy (AFS) are shown in Figure 1. For simplicity, only one excited level is shown in this figure. In AES, the atom is thermally or electrically promoted to an excited state where upon it can radiationally decay to the ground

PAGE 12

^ Col 1 i si onal Exci tati o hi/ hi/ A E S A A S A F S Figure 1. Energy level schemes for atomic spectrosco py.

PAGE 13

5 state. In AAS a fraction of the radiant flux from the light source is absorbed and the attenuated absorbed light is detected. In AFS a fraction of the radiant flux from the light source is absorbed and radiational decay is observed as the atom relaxes back to the ground state. Associated with each excited state of the atom is a very specific excitation energy. In AAS and AFS, special excitation sources are usually employed to excite the atoms to their respective selected energy levels. In contrast, AES is subject to a broad range of energies due to collisional excitation. Thus, there is little control over the excitation of atoms in a sample; i.e., many excited energy levels can become populated in AES. Therefore, the spectra of AES typically is more complicated than the much simpler spectra of AAS and AFS. The main instrumental components for atomic spectroscopy and their configuration are shown in Figure 2. For AES, AAS, and AFS, a sample introduction system, an atomization cell (which atomizes the analyte) and a signal detection system is employed. Only for AAS and AFS is a light source utilized. In AFS and AAS, the light source is positioned at a 90 and 180 angle, respectively, to the detection system. Atomic Fluorescence Spectroscopy AFS is based on the radiational activation of atoms and ions resulting in a subsequent resulting radiation deactivation, called fluorescence. AFS is attractive because of its analytical and diagnostic purposes. Discussion of AFS will be emphasized with the use of an inductively coupled plasma (ICP) as the atomization cell. The excitation of atoms for AFS in an ICP has been utilized by electrodeless discharge lamps (EDLs),

PAGE 14

o o 0) CD Q O o 2 -> 8

PAGE 15

7 hollow cathode lamps (HCLs), ICPs and dye lasers. The sample is introduced into the atomizer and is excited by means of a suitable light source (HCL, EDL, dye laser, or a second ICP) Considering only two energy levels in an atom (and the low optical density case) the fluorescence radiance, B F (W Fluoresced nf 2 sr" 1 ), can be given by B F = £/4tt Y 21 E„ 12 ft k(i/) du (4) where I is the path length to the detector (m) Y 21 is the fluorescence power efficiency (W Fluoresced /W Absorbded ) E vl2 is the spectral irradiance of source radiation at the absorption line of frequency u 12 (W m~ 2 Hz" 1 ) and JiJ k(y) du is the integrated absorption coefficient over the absorption line (nf 1 Hz) (where k(i/) nf 1 is the atomic absorption coefficient at frequency v) 5 Thus, the fluorescence radiance increases linearly with effective source radiance (if saturation of the upper level is not achieved) AFS was first used as an analytical tool by Winefordner and coworkers. 6 8 The use of an ICP and an atomization cell for AFS was first reported by Montaser and Fassel. 9 In AFS, the external radiation source is focused into the atomization cell and produces excited atoms where a fraction of the excited atoms radiationally decay (i.e., give off atomic fluorescence) AFS is ideally suited for multielemental analysis for several reasons: (i) the physical arrangement of the light source is simple since the exciting light can be directed at any angle to the detector (except 180); (ii) fluorescence radiation is emitted in all directions (4* steradians) ; and (iii) AFS spectra are considered simpler

PAGE 16

than AAS and AES spectra. 10 ICP-AFS provides great spectral selectivity compared to other spectroscopic techniques, e.g., ICP-AES and ICP-mass spectroscopy (ICP-MS) Several comprehensive reviews have been written for AFS.* 1 15 AFS is primarily useful for trace analysis. Atomic Absorption Spectroscopy In 1833, Brewster recorded the earliest observation of absorption spectra. As discussed above, the dark lines from the sun's spectrum were absorption lines. In the early 20th century, Wood performed definitive experiments of AAS by resonance in gases. 1 In 1939, Woodson apparently had the first publication of absorption for quantitative elemental measurement. 1 The birth of AAS as we know it today came from two independent papers in 1955 by Walsh and by Alkemade and Milatz. 17,18 Walsh's paper was a discussion of how AAS could be a promising method of chemical analysis with vital advantages over emission methods. 17 In AAS, the external radiation source is focused through the atomization cell and into a monochromator The attenuation of the incident radiation by the analyte atoms is measured. The situation most commonly encountered in practical AAS analyses is that of a narrow line source used for excitation. The relationship between atomic absorption and atomic concentration is given by the integrated absorption coefficient Jo Mi/) dv Ue 2 )/(mc) n x f 12 (5) where e is the electronic charge (C) m the electronic mass (kg) c the velocity of light (m s" 1 ) n x the number of atoms per m 3 (capable of

PAGE 17

9 absorbing within the frequency bandwidth, u + du) and f 12 (dimensionless) is the oscillator strength (the average number of electrons per atom which can be excited by the incident radiation). 17 Considering a beam of incident radiation, I 0l/I the relationship of the intensity of the transmitted beam of radiation, I„, to l Qy is I„ = I „ exp [-k(.y) I ] (6) where I is the thickness of the atomic gas or vapor (m) 17 The physical conditions (i.e., temperature, pressure, and electric fields) to which the ato~s are subjected and the nature of the atomic transition involved affect the shape of the absorption line (i.e., the dependence of k(i/) on u) Most treatments in AAS assume that the ground state energy level is the nost populated in analytical atomization cells. 19 However, other situations may exist where other nearby atomic levels are populated as well. In these cases, the percent of atoms in level i, %N can be calculated as %Ni ( gi /Z) exp (-Ei/kT) (7) where Z, the electronic partition coefficient (dimensionless), corrects for varying populations in multiplet ground levels and lower excited levels AAS is a good quantitative tool: it utilizes flame or graphite furnaces that are easy to use, provides excellent sensitivity and limits of detection (LODs) in the part-per-billion (ppb) range, and requires small

PAGE 18

10 amounts of solution. The principle disadvantages of AAS are (i) only one element at a time can be measured; (ii) its lack of sensitivity for refractory elements because of formation of monoxides in low temperature chemical flames; and (iii) chemical and ionization interferences are sometimes observed. 20 The Inductively Coupled Plasma The development of ICPs began in 1942 when Babat published his first paper on the properties of electrodeless discharges in a high frequency field. 21 In the early 1960s, Reed described his ingenious approach to the stabilization and thermal isolation of these plasmas. 22 Reed operated an argon ICP as a heat source at atmospheric pressure with 4 MHz and 1 to 10 kW, to grow refractory crystals at temperatures from 10 to 20 kK. Greenfield used a modified torch as that used by Reed and was first to report the ICP as a light source for solution samples. 23 Since then, much work has been done especially by Fassel 2A ~ 27 and by Boumans and De Boer 28 with an ICP as an emission source. The ICP torch itself is made of fused quartz of three concentric tubes; the outer (coolant) tube, the middle (auxiliary) tube, and the central aerosol injection tube. Figure 3 is a diagram of a typical ICP torch. The coolant argon gas is introduced tangentially at flows of 10 to 20 L min" 1 and results in cooling, vortex stabilization of the torch, and is also the plasma gas. The auxiliary gas is used to keep the plasma in a suitable position within or above the load coil at flows of 0.5 to 2 L min The carrier argon gas flows into the central injection tube at 0.5 to 1 L min The 2 to 4 turn copper water-cooled induction coil is

PAGE 19

11 coil cool ant auxi 1 i ary c a r r 1 e r Figure 3. Diagram of an ICP torch.

PAGE 20

12 coupled to a radiofrequency generator. A common oscillator frequency is 27.12 MHz, although 40.68 MHz is now becoming more common. 31 The high frequency currents generate oscillating magnetic fields where lines of force are axially oriented inside the quartz. Since argon gas is a nonelectrical conductor, a seed of electrons is planted into the torch by a tesla coil. The "tickled" argon gas then forms a plasma induced by the axial magnetic fields. The electric field imparts kinetic energy to the electrons in the plasma which then share this energy with the plasma atoms and ions by colliding with them. There are, however, several proposed excitation mechanisms for the ICP (e.g., the Penning ionization reaction, charge transfer, electron collision, and atom collision). 31 Extensive research on the mechanisms of operation of an ICP has resulted in the acceptance of several theoretical models or simulations. The plasma is an effective atomization source and a sample can be efficiently injected into the plasma. Typical residence times of the sample in the plasma are about 2.5 ms The free atoms flow upstream in a narrow cylindrical channel which can be easily and efficiently focused onto the slit of a monochromator The r.f. coil, replacing the classical electrode of an arc/spark discharge, does not come into direct contact with the plasma and thus eliminates the problem of elemental contamination. 32 The high r.f. field, however, must be adequately shielded, to avoid interference with other electronic equipment. 33 The number density of free atoms in the hot argon sheath is very low. Free atoms or ions of the analyte tend to behave as an optically thin emitting source. The excitation temperature and electron number density are the most significant properties of an ICP

PAGE 21

13 making it a useful excitation source for atomic spectrometry. (In calculating the plasma temperatures, discrepancies exist due to lack of a general agreement of published transition probabilities). 20 The plasma is not in a complete state of thermodynamic equilibrium since all energy distributions can not be described by only one temperature. Instead, the IC? is considered to be in local thermal equilibrium (LTE). 31 The requirements of a good atomizer are (i) good atomization efficiency; (ii) low radiational background and background flicker; (iii) low concentration of quenchers; (iv) long residence time of analyte in the optical path; (v) simplicity of operation; and (vi) low cost of initial purchase and operation. 11 Compared to flames and electrothermal atomizers, the ICP is an excellent atomization and ionization cell for virtually all elements." The ICP has a low spectral background compared to other atomization cells and has remarkable freedom from chemical and physical interferences. 31,33 Argon ICPs are excellent vaporization, atomization, excitation, and ionization sources for atomic spectroscopy. Several attractive properties of the ICP are (i) high gas temperatures; (ii) capability of sustaining the analyte in an inert gas environment ; and (iii), freedom of contamination from electrodes since none are required. 34 However, some disadvantages of the ICP are that is not easy to operate and it is expensive to purchase and operate. Radiation Sources Requirements for good radiation sources are (i) high radiance over the absorption of analyte atoms; (ii) long term and short term stability (i.e., little drift and little flicker); (iii) simple tuning and focusing;

PAGE 22

14 (iv) availability for all elements; (v) long lifetime; (vi) safety of operation; and (vii) low cost. 11 Compared to continuum sources, line sources generally have higher radiance over the absorption line, fewer spectral interferences, and little, if any, wavelength tuning is required. Compared to line sources, continuum sources offer the utility of one source for multielement excitation, have better stability, are simpler to focus, and generally have longer lifetimes. 11 Since fluorescence intensity is proportional to source intensity (assuming no optical saturation), it is desirable to employ high luminosity sources. Dye lasers appear to be the ideal excitation source for AFS (except for their cost). The high intensity of lasers shows great promise of extending the linear dynamic range (LDR) of AFS. EDLs are commonly employed for AFS and have intensities more than one order of magnitude greater than HCLs The main drawback to EDLs is that some good lamps do not exist or are difficult to make for some elements. ICPs have also been utilized as an intense line (or continuum) source for AFS, however, cost is also a factor here. 5 HCLs are commonly used in AAS and AFS. In order to increase their output intensity, these lamps can be pulsed at high current levels (EDLs can also be pulsed). The xenon arc lamp is a continuum source. A continuum source used with a monochromator or interference filters can be readily used for multielemental analysis. These lamps can also be pulsed to increase their overall peak power. Their biggest disadvantage is their poor spectral output in the low UV (i.e. < 250 run) 5

PAGE 23

CHAPTER 2 A UNIQUE ULTRASONIC NEBULIZATION SYSTEM Introduction Sample introduction is one of the most troublesome aspects in analytical atomic spectroscopy. 35 The goal of sample introduction methods is to efficiently and reproducibly transfer a representative amount of sample to the atomization cell. The analytical sample can be a liquid, solid, or gas. Because many analytical samples are dissolved to form aqueous solutions, nebulizers are the most common sample introduction methods investigated. 36 The efficiency of nebulization of the sample has a direct influence on both the sensitivity of the technique and the extent of interference effects. 37 The nature of the atomic population (and the signals observed) depends on the type of atomizer and sample introduction method employed. 38 Methods of non-pneumatic nebulization include spinning disks, high voltage sparks, jetimpact, fritted disks and ultrasonics. 36 38 The pneumatic nebulizers include those that are concentric (e.g., the Meinhard nebulizer) and cross-flow (e.g., the Babington and MAGIC nebulizer) ; 37,39 ~'' 1 these are the most common nebulizer types utilized in atomic spectroscopy. The ideal aerosol produced by a nebulizer should be of high density and of small and homogeneous droplet size. 37 15

PAGE 24

16 Ultrasonic nebulizers contain a piezoelectric transducer, a leadzirconate-titanate (PZT) ceramic in most cases, which vibrates at ultrasonic frequencies from 20 kHz to 5 MHz. 37 38 The high frequency voltages applied to the transducer crystal cause it to twist, bend and shear which lead to mechanical vibration of the crystal.'' 2 As the analytical solution flows across the face of the vibrating PZT transducer, standing waves are formed on the surface of the liquid. The wavelength, A (m) of these waves on the surface of the liquid is A = [(2 n a) / (p f 2 )] 1/3 (8) where a is the liquid surface tension (N m' 1 ) p is the liquid density (kg m~ 3 ) and f is the ultrasonic frequency (Hz). 42 The liquid surface obtains a sufficient vibrational amplitude to become unstable, and aerosol droplets of nearly equal diameters are ejected. 37 The median droplet diameters, D (m) are related to the liquid surface wavelength as 42 D = 0.34 A (9) Therefore, the higher the ultrasonic frequency, the smaller the aerosol droplets. 38 The face of the PZT transducer is protected from cavitation and pitting by the sample with a glass plate that is acoustically matched (i.e. the glass plate thickness is a half integral of the wavelength of the ultrasonic vibration) 42 to the transducer's resonance frequency. Even with

PAGE 25

17 good acoustic matching of the glass plate to the PZT transducer, the ultrasonic wave is attenuated to some extent. In the mid-1960s, ultrasonic nebulizers were first suggested as a replacement for pneumatic nebulizers.'' 3 "'' 5 Ultrasonic nebulizers are electrically driven and are not dependent on the aerosol carrier gas flows which restrict the flexibility of optimization with pneumatic nebulizers. Also, unlike pneumatic nebulizers ultrasonic nebulizers do not experience clogging or excessively long term memory effects associated with high analyte salt concentrations. Another drawback of pneumatic nebulization is their wide droplet size distributions, whereas ultrasonic nebulizers produce smaller and more uniform particle size distributions (1.5 to 2.5 ^m range) 38 which allow for more efficient transport and desolvation of the sample aerosol before atomization. t,6 ~' ,a The sample transport efficiency of most pneumatic nebulizers is normally 2 to 10%, whereas, the transport efficiency of ultrasonic nebulizers has been reported to be as high as 90%. 49 With the use of ultrasonic nebulizers, the sensitivity can often be improved from 5 to 50 fold. Thus, one can work below previous pneumatic detection limits, preconcentration steps are not needed, and more dilute samples can be used to reduce interelement effects.'' 5 Some problems of ultrasonic nebulizers, however, are (i) a more complex and expensive apparatus is required; (ii) excessive clean out times can occur and thus lead to less analytical throughput; (iii) short and long term drifts of the analytical signal can be observed; and (iv) cross contamination and memory effects can occur within the ultrasonic spray chamber.'' Many of these problems can be overcome by careful attention to

PAGE 26

18 instrumental design of the ultrasonic system. Careful design considerations were utilized in construction of an ultrasonic spray chamber so as to reduce as many of the above mentioned problems as possible. Design and Construction A detailed drawing of the ultrasonic nebulizer used is shown in Figures 4-9. All of the designed apparatus shown in the figures, except for the PZT transducer, were constructed within our machine and glass shops. The housing for the PZT transducer was constructed of brass (see Figure 5) The top (and its screws) of the housing was made of KLF nylon and the bottom nut was made of TLF teflon (see Figure 6). A 2-021 size 0-ring (Parker Seal Co., Lexington, KY) was used to seal the PZT transducer with the housing and its top (see Figure 7). All 0-rings used in the housing were made of N-butyl rubber. The nebulizer spray chamber and the entire desolvation apparatus were made of borosilicate glass (see Figures 8 and 9). The piezoelectric transducer was purchased from Channel Products Inc. (Chesterland, OH) and was completely assembled (i.e., with attached glass plate) Sample delivery was performed continuously at a constant rate of 3.3 mL min -1 by a peristaltic pump (Rainin Instrument Co., Inc., Boston, MA, Model Rabbit) The sample was pumped to a sample delivery tube orifice (stainless steel, 1 mm i.d.) positioned approximately 1 to 2 mm from the surface of the PZT transducer and near the upper edge of the transducer (see Figure 7). Proper sample delivery tube positioning was easily manipulated by the ball and socket joint.

PAGE 27

19 Ula t e r In — r i n g PZT Transduce II P1 ^ ^ _p-^J Water > Ou t u _n BNC TJ 'igure 4. Diagram of the complete PZT transducer housing.

PAGE 28

20 1*4.1 1.865* 0.13* H y 2-5S TPI 125* I £1 l/a se TPI / For 2-028 O-Rina Figure 5. Detailed drawing of the cross section of brass housing.

PAGE 29

21 Housing Top Front S i d Bo t t o ica J_ a. ea* Clraranci ft #2 S(MH Bottom Nut e.23"- IB 32 TPI (mm 1/2 20 TPI 1/2* radius Figure 6. Diagram of the housing top and bottom nut.

PAGE 30

22 c o CO c \o Q

PAGE 31

23 S3 CD CO

PAGE 32

24 o H bO C •r-l C3 H T3

PAGE 33

25 The unique spray chamber was designed to utilize aerodynamic principles that promote streamlining of the aerosol (see Figure 8) This spray chamber should reduce aerosol concentration within the spray chamber, thereby reducing turbulence and allowing for faster aerosol evaporation. Streamlining should result in shorter clean-out times, fewer memory effects, and minimize condensation within the spray chamber itself. Operation Power to the PZT transducer was supplied by a Plasma-Therm power supply (Plasma -Therm, Inc., Kresson, NJ Model UNPS-1). Maximum power to the transducer could be obtained by tuning the frequency of the power supply to achieve minimum reflected power. The measured forward and reflected operating powers were 37 and 3 W, respectively, and the frequency of operation was 1.35 MHz. One power lead was hard-wired to the center-most contact electrode of the PZT transducer. The outer-most contact electrode of the PZT transducer was bonded to the brass nebulizer housing with a silver filled epoxy (Epoxy Technology, Inc., MA, EPO-TEK, H27D) The connection between the power supply and transducer housing was made via a BNC connector. The ultrasonic energy produced by the transducer is also transferred to the backing medium, which acts as the coolant for the piezoelectric. The backing medium (coolant) for the transducer was flowing chilled deionized water (5 C) To promote smoother draining of the un-nebulized solution, the nebulizer was operated in a vertical orientation as described by Olson et al. 46 Because of the high nebulization efficiency of the ultrasonic nebulizer, the excessive solvent mass transport rate to the atomizer could

PAGE 34

26 lover the temperature of the ICP and thus, the atomization efficiency of the ICP. Also, the excess solvent vapor can cause a great loss in aerosol drop size homogeneity due to aggregation of the droplets. 37 This excessive solvent loading was minimized by an efficient desolvation system (see Figure 9) The desolvation tube was heated with standard heating tape (Fisher Scientific, Pittsburgh, PA) to approximately 200 C. The cooled condenser (ca. 5 C) was a modified Davies' double surface condenser. There is a drain following solvent desolvation to allow the condensed liquid solvent to be removed while "dry" aerosol sample droplets are transported to the ICP. The same chilled water to cool the PZT transducer was also used for the desolvation system and ultimately allowed operation of the entire ultrasonic system for long periods of time ( > 8 h) It took approximately 65 s from introduction of a sample to stabilization of the signal. The majority of this time is due to the length of peristaltic tubing used for sample delivery.

PAGE 35

CHAPTER 3 ATOMIC FLUORESCENCE SPECTROMETRY IN AN ICP WITH A PULSED -CONTINUUM SOURCE Introduction The most commonly used multielement atomic spectroscopic technique is inductively coupled plasma-atomic emission spectrometry (ICP-AES) ICPAES achieves nanogram per milliliter (ppb) detection limits, long linear calibration curves, excellent precision and matrix free measurements for most elements. 50 52 More recent techniques include the multielement hollow cathode lamp inductively coupled plasma atomic fluorescence spectrometer (HCL-ICP-AFS) the inductively coupled plasma mass spectrometer (ICP-MS), and the continuum source furnace atomic absorption spectrometer (CSF-AAS) 50 ~ 52 The fluorescence approach results in lower detection limits, greater spectral selectivity, and a reduced emission background; 53,5i| the furnace atomic absorption approach is applicable to smaller sample amounts than the ICP-AES approach. In this study, the ability to capitalize upon the simplicity of atomic fluorescence spectra of virtually all elements as well as the possibility of exciting all atoms (and/or ions) simultaneously by means of a spectral continuum light source was investigated. In order to increase the source spectral irradiance 58 especially in the UV, a repetitively pulsed 27

PAGE 36

23 xenon flashtube and a gated detector were used to increase the measured signal-to-noise ratio. The type of fluorescence observed in this work is resonance (one -wavelength or one-color) fluorescence. Resonance fluorescence is where the wavelength of light used for atomic excitation is the same wavelength as the atomic fluorescence. A representation of resonance fluorescence is depicted in Figure 10. The radiation sources typically used in commercial AFS are line sources (e.g., mostly HCLs) HCLs exhibit strong atomic emission whereas they provide insufficient intensity for the ionic emission of a specific element. For some easily ionized elements, it would be advantageous to optically probe their ionic transitions. The spectral continuum output from the pulsed xenon light source should be able to probe both the atomic and ionic transitions of an element. In this study, the experimental system is described and some initial analytical figures of merit are given. Also, LODs for some elements will be compared utilizing the ultrasonic and pneumatic nebulizers. Experimental Instrumentation A schematic diagram of the experimental system used in this study is shown in Figure 11. The experimental components and manufacturers are listed in Table I. Source radiation from the pulsed flashtube was focused into the ICP using two lenses, L x and L 7 (both with diameter and focal lengths of 50 mm): L : collimated the radiant flux from the flashtube and L 2 focused the collimated radiant flux onto the center of the ICP above the load coil. The diameter of the focused beam was

PAGE 37

u C d 29 5 9 5 16 5 9 2 2 D D M 4 3 6 9 2 CD 3 18 2 7 X = hi/ 2 2 8.8 n m S Figure 10. Atomic energy level diagram of cadmium showing resonance fluorescence

PAGE 38

30

PAGE 39

31 Table I. Experimental components Equipment Model Manufacturer Micropulser power 457A supply Flashtube Novatron-722 SRS gated integrator SR 250 and boxcar averager SRS computer interface SR 245 Digital oscilloscope 2430A Xenon Corporation, Woburn MA 01801 Xenon Corporation Stanford Research Systems, Inc. Palto Alto, CA 94306 Stanford Research Systems, Inc. Tektronix, Inc., Beaverton, OR 97077 Monochromator

PAGE 40

32 approximately 8 mm. Since resonance fluorescence was the signal detected, several precautions were taken to reduce scattered light reaching the zonochromator from the flashtube. This was accomplished with a light trap placed behind the ICP directly in line with the flashtube (see Figure 11) and with blackened baffles set up around the ICP. The collection lens, L, (diameter 76 mm and focal length 178 mm), was also enclosed within a blackened tube to further reduce scattered light. Despite these precautions, the limiting noise of the system was still due to scattered light from the excitation source. The 300 W xenon flashtube was enclosed within a fan-cooled housing which contained a front surface spherical mirror (diameter 50 mm and focal length 31.5 mm) and collimating lens L : The lamp was operated from a pulsed power supply at 5 kV. A 0.2 fiF discharge capacitor provided an input energy of 5 J per flash, a flash half -width of ca. 680 ns and a lamp peak power of 4.2 kW. The lamp was pulsed at a repetition rate of 20 Hz. To help reduce radio frequency (r.f.) noise, the entire lamp housing was surrounded and grounded with copper wire cloth acting like a Faraday cage. The fluorescence radiation was collected at a 90 angle to the excitation beam and a 1:1 image of the ICP was focused with a third lens, L 3 onto the entrance slit of the monochromator (focal length = 350 rm, reciprocal dispersion = 20 A/mm, and aperture ratio f/6.8). To avoid overfilling the monochromator collimator, an iris diaphragm was placed between L3 and the monochromator. A small fraction of the exciting light from the flashtube was reflected with the edge of a mirror to a photodiode which triggered the boxcar averager. The photo-current pulse produced bv the photomultiplier tube (PMT) was terminated through a 1000 ohm load

PAGE 41

33 resistance directly into the boxcar input. The resulting signal pulse has a full width to half maximum (FWHM) of ca. 1.5 /is. The boxcar delay time (the time between the trigger pulse and the start of the measurement) was 700 ns The gate width (the time during which fluorescence was measured) was 1.8 us for all cases. Thirty signals (i.e., thirty lamp flashes) were averaged for each output signal. The "busy out" of the boxcar averager triggered a Stanford analog-to-digital (A/D) system in order to measure the output signal. Horizontal and vertical translation of the ICP torch was accomplished with two single axis translation mounts. This allowed horizontal translation of the ICP torch up to ca. 100 mm and vertical translation of the ICP torch up to ca. 80 mm. The ICP concentric pneumatic nebulizer was fed with a peristaltic pump to allow for lower sample rate uptake and thus reduce salt encrustation in the torch. The ultrasonic nebulizer was also fed with a peristaltic pump as described in chapter 2. Reagents and Procedure All components of the experimental system were operated according to the directions given in the manufacturers' manuals. The chemicals used in preparation of the stock solutions were all reagent grade. The preparation of standard solutions were made with compounds best suited for the solutions as determined by Parsons et al. 5S Distilled demineralized water (Barnstead Sybron Corportation, Boston, MA) was used throughout this study. Standard solutions were obtained by serial dilutions of the stock solutions.

PAGE 42

34 The majority of this work involved the use of a pneumatic (Meinhard) nebulizer and a conventional ICP torch. For each element studied, the optimal ICP r.f. power, observation height above the ICP load coil, monochromator slit width, and ICP gas flows were determined. Calibration curves and limits of detection (LODs) were also determined for each element and compared to some results obtained with the ultrasonic nebulizer. In measuring the synthetic mixtures containing five elements, compromise values of the above mentioned parameters were necessary. ICPAFS results were also compared using an extended ICP torch versus a conventional torch. An extended ICP torch has an outer sleeve that is 40 mm longer than the conventional torch. Results and Discussion Molecular species of non-refractory elements typically have low molecular dissociation energies and are easily atomized by the plasma at lower r.f. powers, whereas those of the refractory elements typically have high molecular dissociation energies and are atomized by the plasma at higher r.f. powers. Also, the atomic and ionic populations of the sample species in the plasma are often greatly affected by the choice of r.f. power and observation height. 58,59 For this study, the dependence of the fluorescence on the r.f. power and observation height were examined independently, as shown in Figures 12 and 13, respectively. In these two figures, the plots for each element were arbitrarily shifted for clarity in interpreting the graph. Figure 14 shows the variation in fluorescence signal-to-noise with monochromator slit width for each element (at optimal r.f. power and

PAGE 43

35 = >• era £

PAGE 44

36 >Ezn E-h > E-i < TIIIirillirilllirrilllTIITIfTITIIirTTTTTrTTIT'T' TTTriTlll 400 800 1200 R. F. POWER ( W) Figure 13. The effect of r.f. power on the fluorescence signal for each element

PAGE 45

37 2000w 16 0o i 12 00-^ o EI -* 800^
PAGE 46

38 observation height). Table II lists the optimal experimental conditions (found by univariate search) for each element, based on Figures 12-14. Typical calibration plots obtained under the optimal conditions are shown in Figure 15. These plots were also arbitrarily shifted for clarity. (Figures 12 15 were all obtained with a pneumatic nebulizer.) Table III lists the analytical figures of merit using a pneumatic nebulizer. The detection limit is defined as the concentration in y.g mL (ppm) of the element in pure aqueous solution resulting in a signal that is three times the standard deviation of the blank measurements. The electronic band-width for all measurements was ca. 1 Hz. Previously reported ICP-AFS detection limits for the same elements are also listed in Table III. All the log-log calibration plots have slopes between 0.96 and 1.06. Typical with atomic fluorescence, the calibration curves cover a wide linear dynamic range. The deviation from linearity at the higher concentrations in Figure 15 is due to preand postfilter effects. The effect of using a long torch was examined for cadmium as a model element. The results are also shown in Figures 12-15 and the corresponding curves are labeled "Cd(L)" With the long torch, the fluorescence signal was practically constant at observation heights of 45 to 60 mm above the load coil; whereas with the standard torch, the signal rapidly decreased with increase in observation height (Figure 12) The variation in signal with increase in r.f. power was similar in pattern for both torches, but the signal was significantly larger with the standard torch (Figure 13) A similar difference was observed for the effect of monochromator slit width on the signal-to-noise ratio, and again the standard torch gave much the better performance (Figure 14) Despite

PAGE 47

39 Table II. Optimal experimental conditions* element' -*afs> r -fpower Observation Slit width (nm) (W) height (mm) (pm) Mixture tf 200-800 600 15 3a(II) 455.4 500 15 2000 Ca(II) 393.4 800 27 1500 Cd(I) 228.6 500 15 900 -\'a(I) 589.6 500 15 2000 V(II) 292.4 700 15 1000 Cd(I)* 228.6 700 45 1000 1500 Other experimental conditions: sample uptake rate 1.15 mL min" 1 ; pneumatic nebulizer pressure 31 psig; plasma Ar flow rate 15 L min" 1 ; auxiliary Ar flow rate 1-3 L min" 1 x (I) indicates an atomic line, (II) an ionic line (singly ionized). i Analyzed with a long torch. A f A mixture of the 5 elements at concentrations of 20 y.g mL" 1 for each element

PAGE 48

40 E'en £5 £> i— i E-h -< O o -J 1 1 1 1 t 1 1 1 1 r i t i r 1 1 1 1 t r 1 1 1 r I [ 1 1 1 1 j rt 1 1 1 1 1 1 1 j 1 1 1 1 1 1 1 1 1 1 1 1 1 r 1 1 1 1 r 1 1 1 1 1 1 1 1 1 1 [ 1 1 ; i ) 1 1 1 1 1 1 t 1 1 1 1 1 1 1 j 1 1 1 1 1 1 1 1 1 1 -2.5 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5 LOG [CONCENTRATION ( p p m) ] Figure 15. Analytical calibration curves for each element.

PAGE 49

41 Table III. Analytical figures of merit using a pneumatic nebulizer. Element' Sensitivity* LOD Literature LOD (mV mL ^g" 1 ) Og ml/ 1 ) Og ml/ 1 ) Ba(II) 14.1 0.09 0.05 Ca(II) 18.4 0.03 0.0004 Cd(I) 26.3 0.02 0.0005 Na(I) 10.7 0.04 0.0003 V(II) 1.88 0.4 0.1 Cd(I)* 14.9 0.04 ND Referred to boxcar input. f See footnote f to Table II. $ Measured with a long torch ND Not done (no literature value available)

PAGE 50

42 this, the detection limits for cadmium were essentially the same with both torches (Table III) Table IV shows the analytical figures of merit for several elements using the ultrasonic nebulizer as the sample introduction device. The LODs in this table are also compared to some literature values. 00 In both Tables III and IV, the literature values reported are values obtained by HCL-ICP-AFS. Thus, since HCLs were used, only the atomic transitions of a particular element was studied. As seen in Table IV, the LOD for Ba(II) (Ba(II) designates an ionic transition whereas Ba(I) designates an atomic transition) was approximately 20 times better than that obtained from the literature value. Also, the LODs for Mg(I) and Ca(II) were an order of magnitude poorer than those obtained in the literature. Reasons for this can be attributed to poorer lamp intensity around the Mg(I) line. Also, since the excitation energy of Ca(II) to an upper ionizational level is higher than that of Ba(II) lower atomic populations within the ionization levels for Ca(II) may have been achieved with the ICP. The LODs for Pb(I) and V(II) were essentially the same as the literature values. Overall, the LODs obtained with this setup (a continuum source in combination with an ultrasonic nebulizer) should be comparable to line sources used with pneumatic nebulizers in AFS The most intense fluorescence line of each element except vanadium was excited and observed (see Table II) The most intense fluorescence signals for vanadium are at 309 to 310 nm i.e., in the middle of a strong OH band. 61 Therefore, the next most intense fluorescence line of vanadium at 292.4 nm was observed; which is also attributable to a poorer LOD than expected. In general, with the ultrasonic nebulizer and the continuum xenon flashtube source, LODs were

PAGE 51

43 Table IV. Analytical figures of merit using an ultrasonic nebulizer. Log-Log %RSD LOD Literature LOD e (nm) Slope Og niL -1 ) Og mL" 1 ) Mg(I) 285.2 0.952 9. 0.003 0.0005 Ca(II) 393.4 1.24 1.24 0.002 0.0004 Ba(II) 455.4 1.06 8. 0.002 0.05 Pb(I) 405.7 1.08 7. 0.03 0.07 V(II) 292.4 0.992 13. 0.3 0.1 | See footnote f in Table II.

PAGE 52

44 within one order of magnitude poorer or better than the reported literature values. Calibration curves generated with the ultrasonic nebulizer were similar to those generated with the pneumatic nebulizer as shown in Figure 15. The analytical precision (% relative standard deviation, %RSD) of this experimental system is rather poor (ca. 10%). Although high, a large portion of this poor precision is due to inefficient mixing of the aerosol within the spray chamber of the ultrasonic nebulizer. The precision of the pneumatic nebulizer was less than 51. Table V shows a comparison of three elements (Ba, Ca, and V) studied using both the pneumatic (Meinhard) and ultrasonic nebulizer. The ultrasonic nebulizer showed an improvement in the LODs for Ba and Ca by a factor of 10. V, however, showed no improvement in using the ultrasonic nebulizer. Since V is a refractory metal, its signal is more sensitive to factors that affect the plasma temperature (e.g., r.f. power) than the signals for Ba or Ca. In all cases, the loglog analytical curve slopes were near unity and the linear dynamic range (LDR) for each element was at least two orders of magnitude. In one instance, high salt concentrations were aspirated into both the pneumatic and ultrasonic nebulizer. The ultrasonic nebulizer did not suffer from a long clean-out time nor memory effects as that experienced by the pneumatic nebulizer. No pulsation of the carrier gas flow into the ICP was noticed with use of this ultrasonic nebulization apparatus. Figure 16 is a wavelength scan of 1 ng ml" 1 magnesium showing atomic and ionic fluorescence at 285.2 and 279.5 run respectively. This figure shows the ability of one source, the spectral continuum xenon flashtube,

PAGE 53

45 Table V. Comparison of the analytical figures of merit of a pneumatic and ultrasonic nebulizer. Pneumatic/Ultrasonic* Nebulizer Element* LogLog LOD LDR* Slope (/ig mL" 1 ) Ba(II) 0.97 / 1.1 0.09 / 0.002 3.0 / 2.0 Ca(II) 0.99 / 1.2 0.03 / .002 2.5 / 2.0 V(II) 1.1 / 0.99 0.4 / 0.3 3.0 / 2.5 j See footnote f in Table II. LDR in orders of magnitude. £ A value to the left of the diagonal is for the pneumatic nebulizer and a value to the right of the diagonal is for the ultrasonic nebulizer. All other experimental components and parameters are identical.

PAGE 54

.'t6 • i— I 00 CD CD > cd r—i CD Mg ( I ) Mg ( I I ) 270 280 290 300 "Wavelength ( n m) Figure 16. Wavelength scan of magnesium showing a) its atomic and ionic fluorescence and b) a background (blank) spectrum.

PAGE 55

47 to probe both the atomic and ionic transitions of magnesium. It would be difficult to use a magnesium HCL to sufficiently probe the ionic transition of magnesium. Figure 17 shows a multielement scan of a synthetic mixture of the five elements (20 ^g ml/ 1 each), with a standard torch. Note the simplicity of the spectrum in this figure. As seen in Figure 16, the atomic and ionic fluorescence lines of calcium at 422.7 and 393.4 run, respectively, are also observed here. Conclusion The work presented thus far shows promising results for the use of a pulsed xenon flashtube in ICP-AFS. The present system would appear to have considerable use in multielement analysis. Although the detection limits for each element were at least an order of magnitude poorer than the best reported in the literature for the pneumatic nebulizer, the results with the ultrasonic nebulizer seem promising. Together, the ultrasonic nebulizer and the spectral continuum xenon flashtube are an attractive combination. Future work on this project will involve studies to improve the detection limits so as to be competitive with the present commercial atomic spectrometry systems. Improvements could be made by better r.f. shielding, decreased excitation source scatter, and better and more efficient collection of the fluorescence signal. Since the limiting noise in this system was due to scattered light from the flashtube, efforts will be made to reduce scattered light by observing non-resonance fluorescence and by using better light traps.

PAGE 56

48 C/2 CD > CD C a ( I I ) 200 300 400 500 600 Fluorescence wavelength ( n m) Figure 17. A multielement analysis of a synthetic mixture of five elements

PAGE 57

CHAPTER 4 DOUBLE-RESONANCE FLUORESCENCE IN AN ICP INDUCED BY A CONTINUUM SOURCE Introduction When an atom (ion) absorbs a photon and subsequently re -emits a photon of the same spectral frequency as that absorbed, one refers to this as single-resonance (single-step) fluorescence. The types of atomic (ionic) fluorescence not considered as single-resonance fluorescence are referred to as non-resonance fluorescence: double-resonance (two-step or two-color) fluorescence is an example of this type. Double-resonance fluorescence refers to a process in which atoms (or molecules) are excited into the fluorescence state in a stepwise manner with two lasers tuned at appropriate atomic (or molecular) energy levels. 52 Figure 18 is an energy level diagram of cadmium showing a representation of double -resonance fluorescence. Here, the excitation of a selected atomic level is achieved with a single laser tuned to a particular transition frequency, while a second laser, coincident in time and space to the first laser, is tuned to a higher excitation level starting from the level reached by the first laser (or from a collisionally populated nearby level). 53 The analytical advantage of such excitation processes as compared to that involving only one step (single -resonance fluorescence) lies in its increased spectral selectivity without a significant loss of sensitivity if both excited 49

PAGE 58

C d 50 hi/ hi/ 5 9 5 16 5 9 2 2 tax) U Figure 18. Atomic energy level diagram for cadmium showing doubleresonance fluorescence and the relevant transitions studied in this work.

PAGE 59

51 state transitions are saturated. The UV and visible spectra of atoms (ions) in a multielement mixture can show a high degree of spectral overlap within this spectral region. In order to obtain simple spectra, either the spectral resolution can be increased or the number of spectral lines within the spectral window can be reduced. Double-resonance fluorescence can reduce the number of spectral lines within a spectrum because of its selectivity. Several analytical studies performed with tunable lasers in atmospheric pressure atomizers have been reported in the literature. 6 ''" 68 In principle, when using spectrally unfiltered excitation sources, several observed fluorescence lines might be a result of double-resonance excitation processes. This possibility seems to have been overlooked so far in most analytical studies performed in flames and other atomizers, especially when continuum sources (e.g., xenon arc lamps) are used for excitation. It is also clear, however, that the requirements needed for a successful use of the double-resonance fluorescence approach are not met by conventional excitation sources; i.e., non-laser sources. On the contrary, such processes will increase the possibility of observing spectral interferences in the analysis of complex sample matrices. Since, to the best of my knowledge, doubleresonance fluorescence signals have not been studied with conventional excitation sources, I felt it would be interesting to investigate if, and to what extent, such processes could be observed. The aim of this study was to describe the observations made with cadmium atoms in an inductively-coupled argon plasma illuminated with a spectral continuum xenon pulsedflashtube Cadmium was chosen because the

PAGE 60

52 pertinent transitions could easily be isolated from one another with the proper choice of optical filters. Experimental The experimental system is similar to the one described previously. 69 As discussed in chapter 3 an ultrasonic nebulizer was used as the aerosol generation system for sample introduction. The unique nebulizer system was described in chapter 2. The experimental operating conditions are listed in Table VI. The source ICP nebulizer was fed with a peristaltic pump (Rainin Instrument Co., Inc., Model Rabbit). A stock solution of cadmium was prepared as described by Smith and Parsons. A working standard solution of cadmium was made by serial dilutions of the stock solution. All dilutions were made with distilled, demineralized water (Barnstead Sybron Corp.). A 100 /jg ml/ 1 solution of cadmium was used throughout the study. All solutions were stored in opaque polyethylene bottles. The optical filters used throughout the study are commercially available (Corion Corp., Holliston, MA) and are listed in Table VII. The detection system was a Stanford analog-to-digital (A/D) system (Stanford Research Systems, Inc.). The lamp radiance and monochromator-detector assembly were not calibrated.

PAGE 61

53 Table VI. Experimental operating conditions. ICP Forward power 450 W Reflected power W Ar coolant flow rate 11.5 L min" 1 Ar auxiliary flow rate 1.1 L min" 1 Ar carrier flow rate 0.8 L min" 1 Observation height* 35 mm Ultrasonic nebulizer Forward power 37 W Reflected power 3 W Sample uptake rate 3.3 mL min" 1 PulsedFlash tube High voltage 4 5 kV Pulse rate 20 Hz Pulse width 700 ns Miscellaneous Spectrometer bandpass 2 run PMT high voltage -800 V dc Above the ICP load coil.

PAGE 62

54 Table VII. Percent transmittance of the optical filters (%). Corion filter number Monochromator LG 595 WG 320 BG 25 Wavelength (nm) 228 0.04 0.05 0.03 361 0.06 91 78 644 83 92 0.02

PAGE 63

55 Results and Discussion The partial energy level diagram for cadmium pertinent to this work is shown in Figure 18. As already discussed by Omenetto et al for laser excitation, this scheme involves a connected double -resonance absorption process when the first and the second excitation steps share a common level (e.g., the 1 P 1 level). 6 Moreover, because of collisional coupling between the 1 P 1 and 3 P 2 levels, the second excitation step might also be chosen at A ex = 361.0 run (i.e. starting from the 3 P 2 level), in which case one can name the process "disconnected" double-resonance excitation. In this case, since the source is a spectral continuum, both processes can be effective. In either case, fluorescence emission should be observed at 643.8, 361.0 run and at any other allowed transition origination from the D 2 D 3 or nearby levels. As an example, Figure 19 shows a fluorescence spectral scan in the region of 330-370 nm. The spectrum was obtained by aspirating a solution of 100 ^g ml/ 1 Cd into the plasma. As seen in Figure 19, several lines are observed, the most intense one at 361.0 nm. The other two lines which can be identified are at 340.3 and 346.6 nm. Since the 3 P 2 and 1 P 1 levels might be "thermally" populated in the plasma, it remains to be proven that the observed signals are due to a doubleresonance excitation mechanism rather than to a single-resonance excitation mechanism proceeding from a thermally populated level. We have restricted our observation to the two fluorescence transitions as indicated in Figure 19 (i.e., A FL = 361.0 and 643.8 nm) It could be easily proven that these fluorescence transitions were not a result of ICP-excited Cd emission, flashtube scatter, nor blank background. The fluorescence intensities of these lines (i.e., A FL = 361.0 and 643.8 nm)

PAGE 64

56 ^ > i — i CD Cd Scan 330 340 350 360 "Wavelength (n m) 370 Figure 19. Partial fluorescence spectrum of cadmium (100 ng ir.L l ) No optical filters were used between the flashtube and the 1CP.

PAGE 65

57 are listed in Table VIII with relative units of 3.33 and 1.08, respectively. These signals were measured without any type of optical filter placed between the flashtube and ICP. From here on, these measured signals will be referred to as the "observed nonfiltered fluorescence." To prove that double-resonance fluorescence was indeed being observed, a series of optical filter were placed between the flashtube and ICP. UV and visible wavelength transmission scans were performed on each filter and the percent transmissions of each filter at A = 228, 361, and 644 run are recorded in Table VII. Figure 20 shows the wavelength transmission for each filter within the Cd energy level diagram and the observed fluorescence signal. When a long wave pass filter (Corion # LG 595) was used, neither the A FL = 361.0 nor 643.8 nm spectral lines were observed (Table VIII). If the : P : level had been significantly thermally populated, then we should have expected (based on the filter transmission characteristics) approximately 83% of the observed non-filtered fluorescence spectral line at A FL = 643.8 run. Therefore, the observation of the A FL 643.8 nm spectral line in the other cases is a result of some type of two-step photon excitation (whether of the connected or disconnected type) When another long wave pass filter (Corion // WG 320) was used, both the A FL = 361.0 and 643.8 nm spectral lines were observed (Table VIII). 3ased on the filter transmission characteristics, we expected 91% and 92% of the observed non-filtered fluorescence signal for A FL 361.0 and 643.8 nm, respectively. However, only 50% and 13% of the expected signals at A r = 361.0 and 643.8 nm, respectively, were observed. This indicated that

PAGE 66

58 Table VIII. Double-resonance results. Fluorescence signals (Relative units) Experimental A FL = 361.0 nm A FL = 643.8 nm Comments conditions Expected 1 Observed Expected 1 Observed No Filter

PAGE 67

5'.' No Filter Observed Fl uorescence 361 > 643.8 L G 5 9 5 X 8 4 3.8 3, Oba erved Fl uorescence NONE WG 3 2 Observed Fl uorescence 361 >> 843.8 B G 2 5

PAGE 68

60 about 50% of the A FL = 361.0 run spectral line is a result of thermal population of the P 2 level, followed by single-resonance fluorescence. The observation of the A FL = 643.8 rju is also a result of thermal population of the 3 P 2 level, followed by photon excitation to the 3 D 3 level, then collisional quenching to the : D 2 level and fluorescence at 643.8 run. When a bandpass filter (Corion # EG 25) was used, both the A FL = 361.0 and 643.8 run spectral lines were observed (Table VIII). Based on the filter transmission characteristics, we expected 78% and 0% of the observed non-filtered fluorescence signal at A FL = 361.0 and 643.8, respectively. However, only approximately 50% of expected signal at A FL = 361.0 nm was observed. Also, approximately 16% of the observed nonfiltered signal at A FL = 643.8 nm was observed. The observation of fluorescence at 361.0 and 643.8 nm follows the same explanation as that given for the Corion # WG 320 long wave pass filter. Conclusion From the series of experiments described in this paper, the following conclusions can be given: (i) there is negligible thermal population of the 1 P 1 level; (ii) approximately 13% of the observed fluorescence signal for A FL = 643.8 nm is due to the stepwise fluorescence resulting from thermal population of the 3 P 2 level, photon excitation of A ex • 361.0 nm to the 3 D 3 level, followed by collisional quenching to the l D 2 level resulting in fluorescence at A FL = 643.8 nm; (iii) approximately 87% of the observed fluorescence signal at A FL = 643.8 nm is due to connected double -resonance fluorescence from photon excitation (A ex = 228.8 nm) to the X P, level

PAGE 69

61 followed by photon excitation (A ex = 643.8 nm) to the 3 D 3 level; (iv) there is "thermal" population of the 3 P 2 since approximately 50% of the observed non-filtered fluorescence signal at A FL = 361.0 nm was observed when both the Corion # WG 320 and EG 25 filters were used (thereby blocking the A = 228.8 nm spectral line); and (v) the other 50% of the observed fluorescence signal at A FL 361.0 nm is a result of double -resonance fluorescence either as a) photon excitation (A ex = 228.8 nm) to the 1 P 1 level, photon excitation (A ex = 643.8 nm) to the : D 2 level, collisional coupling to the 3 D 3 level, followed by photon emission of A FL = 361.0 nm (connected scheme), or as b) photon excitation (A ex = 228.8 nm) to the 1 P 1 level, collisional quenching to the 3 P 2 level, followed by photon excitation and emission of A = 361 nm (disconnected scheme). However, the explanation as described in (va) is more likely to occur since the collisional quenching from 1 P 1 3 P 2 as described in (vb) should be less favored in the argon ICP. In summary, double -resonance fluorescence with our experimental setup utilizing a xenon pulsedflashtube was observed. The analytical advantages of double-resonance fluorescence as discussed earlier with line sources (Ref. 64-68) will not be utilized with our continuum source. On the contrary, as already stated, the ability to observe double-resonance fluorescence with a continuum source will actually be a disadvantage as this can present a great potential for spectral interferences using a multi-element matrix. Since a continuum source is a non-selective excitation source, some of the selectivity advantages as seen with line sources can be gained only if the proper choice of optical filters is utilized, which would be practical only in a limited number of cases. For

PAGE 70

62 example, in this work, the 228.8 run line could not be solely isolated because of the lack of a suitable optical filter.

PAGE 71

CHAPTER 5 HIGH-RESOLUTION ATOMIC ABSORPTION SPECTROMETRY IN AN ICP WITH A CONTINUUM SOURCE Introduction The choice of atom reservoir and its atomization efficiency is obviously one of the most important considerations in atomic absorption spectrometry (AAS) An ideal atom reservoir should provide complete atomization of all elements, in an inert chemical environment with a low emission background. The most common atom reservoirs used in AAS are flames and graphite furnaces. There are, however, several advantages of an ICP as an absorption cell as compared to flames: 71-73 (i) the higher temperature and longer residence times of a species in the plasma leads to a more efficient atomization; (ii) the chemical environment can be better controlled than that of a flame (e.g., free atoms have longer lifetimes in the chemically inert argon environment of an ICP); (iii) the formation of chemical compounds (e.g., stable refractory compounds) is reduced in the inert argon environment of the ICP; and (iv) refractory solid samples are more efficiently vaporized. There are, however, several disadvantages: 7 '*' 75 (i) there is a larger dilution of gaseous analyte atoms within the plasma of the ICP; (ii) the absorption path length is relatively small; (iii) relatively high ion and atom excited states can 63

PAGE 72

64 be populated leading to lower ground state atom or ion populations and more intense and noisy emission background; and (iv) an ICP may be more expensive and more costly to operate than either flame systems or electrothermal atomizers. Conventional radiation sources in AAS have been line sources (e.g., HCLs, EDLs, and vapor discharge lamps VDL) The ideal radiation source for AAS must meet several requirements: 76 (i) have a high spectral irradiance at all wavelengths; (ii) exhibit little short term fluctuations and long term drift in emission intensity; and (iii) require a minimum amount of operator maintenance for optimum performance. A major disadvantage of most line sources is that only one element per source can usually be investigated. While the use of a continuum source offers the possibility of multielement determinations, until the availability of high pressure xenon arc lamps, continuum sources (such as tungsten filament lamps) had little use in AAS because of their relatively low spectral output in the ultraviolet spectral region. 76 Moreover, the requirement of a high resolution spectrometer has made the continuum source less attractive for AAS. Several excellent reviews have appeared in the literature discussing continuum source atomic absorption spectrometry (CSAAS). 77 80 One of the substantial benefits of CSAAS is the ability to perform good background correction at any wavelength. 38,81 92 Several authors have had success using wavelength modulation for background correction and this approach has since become the method of choice for background correction in CSAAS. 83 87 Analytical work in the field of ICP-AAS has been extremely limited. Despite the short absorption path length of the ICP viewed normal to the

PAGE 73

65 gas flow (never greater than ca. 0.5 cm), ICP atomization and ionization in CSAAS experiments offers the same advantages as flame atomizers; easy sample introduction, high sample throughput, minimal sample pretreatment requirements, and the ability to directly introduce liquid samples as a continuous stream. The advantages of CSAAS with flame atomization are well-documented: simple optical configuration, single source requirement, simultaneous multielement detection, excellent background correction, and low detection limits (above wavelengths of 250 run), 77 80 which should also hold true for CSAAS with ICP atomization. Although low source spectral irradiance below 250 nm is the biggest disadvantage for CSAAS, other disadvantages include the need for a high resolution monochromator to fully resolve the absorption line, and the need for multiple detectors (e.g. photomultiplier tubes) for simultaneous multielement analysis. This work presents an ICP atomization (and ionization) CSAAS system with photodiode array detection. As opposed to conventional scanning spectrometry, a photodiode array allows for fast data processing and storage of spectra and the ability to make simultaneous multiwavelength measurements. 88 The photodiode array also allows absorbance versus wavelength spectra to be obtained in real time and any background absorption or spectral interferences could be directly observed and corrected for by using the manufacturer's computer software. This eliminates the need for wavelength modulation or other background correction techniques.

PAGE 74

66 Experimental Instrumentation Figure 21 is a schematic diagram of the optical arrangement used in this work. The light source was a 300 W Cermax compact xenon arc lamp (ILC Technology, Sunnyvale, CA) A portion of the collimated radiation from the lamp was directed through the ICP by a series of diaphragms. A fused silica lens, L, (5 cm focal length and diameter) focused the collimated radiation after the ICP at the entrance slit of a J-Y H10 0.1 m focal length monochromator (Instruments SA, Inc., Metuchen, NJ) acting as a pre-disperser Optical apertures were positioned between the lens and the ICP to spatially discriminate the transmitted source beam from ICP emission. The 16 run bandpass of radiant flux at the exit slit of the H10 was focused as a one-to-one image by a second fused silica lens, L 2 (7.6 cm focal length and 5 cm diameter) onto the entrance slit of an HR1000 1.0 m focal length monochromator (ISA). The reciprocal linear dispersion of the HR1000 is 0.5 nm mm" 1 in first order. The photodiode array positioned at the exit plane of the HR1000 was an Optical Spectrometric Multichannel Analyzer (OSMA, Model IRY-1024, Princeton Instruments, Princeton, NJ). With the HR1000 monochromator slit width set between 0.075 and 0.100 mm, the spectral bandpass in first order was less than 0.05 run. Since this resolution was not sufficient to fully resolve atomic absorption lines (the full width at half maximum, PaHM, for atomic lines are typically 0.01 run or less), spectral measurements were obtained with the monochromator set to pass radiation in the fourth or fifth order. In the fifth order, the spectral bandpass was less than 0.01 run. Because of the limited length of the photodiode array (ca. 2.54 cm), the spectral window in fifth order was limited to

PAGE 75

67 -h r-

PAGE 76

2.5 run. The pre-disperser simply eliminated lower order source radiation and plasma emission. The experimental operating conditions of the ICP and ultrasonic nebulizer are the same as listed in Table VI (chapter 4) The ICP was operated at a low forward power to reduce the analyte's atomic and ionic emission. Horizontal and vertical translation of the ICP torch was accomplished with two single axis translation mounts. This allowed horizontal translation of the ICP torch up to 100 mm and vertical translation of the ICP torch of up to 80 mm. A laboratory-constructed ultrasonic nebulization system was used for sample introduction. Ultrasonic nebulization improved the nebulization efficiency of the analyte and decreased the dilution of gaseous analyte atoms within the ICP. The experimental setup of the ICP is further described elsewhere. 69 The operating current for the 300 W xenon arc lamp was 17 A. Sample Preparation Stock solutions (1000 mg L" 1 in 2 % HN0 3 ) were obtained from Inorganic Ventures, Inc. (Brock, NJ) for the 14 elements tested. Several synthetic mixtures were also made consisting of 2 to 3 elements. Standard solutions were obtained by serial dilutions of the stock solutions with distilled, demineralized water (Barnstead Sybron Corp. ) Procedure All components of the experimental system were operated according to the directions given by the manufacturer. Aqueous samples were nebulized into the ICP by the ultrasonic nebulizer. As mentioned previously, the fifth order of an atomic absorption line was usually selected on the HR1000. The upper wavelength limit for the HR1000 was 1500 nm; therefore, for elements with atomic absorption peaks above 300 an, detection with fourth order dispersion was necessary. One complete

PAGE 77

69 scan of the photodiode array required 33 ms : 300 data acquisition spectra were averaged to achieve a total acquisition time of 10 s. For each atomic absorption spectrum obtained, a blank spectrum (water aspirated into the ICP) of the incident radiation (10) and a dark spectrum (D) (all radiation blocked at the entrance slit of the HR1000 =onochromator) was also obtained. The emission observed (I e ) if any, with the source radiation blocked was subtracted from the transmittance signal (IJ showing the absorption of an atomic or ionic line such that the true transmittance signal (I) was calculated as I = I a I e The absorption spectrum was calculated for each spectrum using the computer software supplied with the OSMA and the equation A = log (I0-D)/(I -D) Analytical calibration curves were plotted as absorbance peak height versus concentration. The LOD for each analyte was calculated with a confidence level of 98.3% as that absorbance equal to three times the standard deviation of the blank spectrum (10). Results and Discussion Absorption Spectra. Both ionic as well as atomic absorption transitions may be observed in the ICP. The atomic and ionic populations are greatly affected by choice of the r.f. power and observation height of the plasma. 73 89 The r.f. power was maintained at 500 W to reduce atomic or ionic emission. The observation height within the plasma was also optimized for the best atomic or ionic absorption. Observation heights (above the r.f. load coil) of 27 mm and 37 mm were used to study the ionic and atomic transitions, respectively. Unlike conventional line sources, vhich typically have strong radiation output at only the atomic

PAGE 78

70 transitions, continuum sources allow one to optically probe ionic transitions as well. This may prove useful since many elements are efficiently ionized within an ICP. Molecular and atomic absorption measurements have been reported for CSAAS with photodiode array detection; however, poor atomic line resolution was observed. 90 91 As previously mentioned, the observation of fifth and fourth order spectra was necessary to achieve sufficient atomic resolution. Figure 22 is an absorption spectrum obtained for a 10 ppm magnesium solution. Unlike work performed previously with a flame, 92 molecular OH absorption bands in the vicinity of the Mg atomic absorption line were not observed. Figures 23 and 24 are absorption spectra for 10 ppm and 100 ppm solutions of manganese and nickel, respectively. The simultaneous observation of the manganese triplet absorption lines and the many nickel absorption lines demonstrates the greater informing power of the technique compared to single wavelength conventional AAS The observation of more than one element in a mixture is shown in Figure 25, an absorption spectrum of a synthetic mixture of 100 ppm each of copper and silver. This absorption spectrum shows the prominent silver absorption line at 328.07 nm and a strong, but not prominent line for copper at 327.40 nm. Even though copper has a stronger absorption line elsewhere, it can still be detected at the mg I/ 1 (ppm) level in this limited spectral window. Several spectral windows can be judiciously chosen to observe several different elements simultaneously and quickly. The ability to observe both atomic and ionic absorption lines within a single spectral window is shown in Figure 26, a three-dimensional absorption spectrum of a synthetic mixture of 100 ppm of both magnesium

PAGE 79

71 Wavelength (nm) 285.60 Figure 22. Atomic absorption spectrum of 10 fig ml/ 1 Mg.

PAGE 80

72 Mn W^^vMH^ W^^wuj^U Mn ^*W^ „_, „ i i i i i i i i i i i i | i i i i i i i i i i i i i i i i i i 279.2 279.4 279.6 279.8 280.0 280.2 Wavelength (nm) Figure 23. Atomic absorption spectrum of 10 /jg ml/ 1 Mn.

PAGE 81

73 I I I I I I | I I I | | I I | | | | | | | | | | | | | | | | | | 231.00 231.40 231.80 232.20 232'.60 Wavelength (nm) Figure 24. Atomic absorption spectrum of 100 ng ml/ 1 Ni

PAGE 82

74 0.15 -i Uw^V^vV^ Wavelength (nm) — i — i — i — i — i — i — i — i — i — i — i — i 328.00 328.50 Figure 25. Atomic absorption spectrum of a synthetic mixture of LOO jig mL" 1 each of Cu and Ag.

PAGE 83

75 Mn 3 378.4 878.8 378. B 380.0 380.3 880. Wavelength (nm) 'igure 26. A three dimensional atomic absorption spectrum of 10 /ig ml/ 1 each of Mg and Mn showing relative absorbance as a function of observation height above the plasma load coil.

PAGE 84

76 and manganese. Three manganese atomic lines at 279.48, 279.83 and 280.11 nm and two magnesium ionic lines at 279.55 and 280.27 nra are observed. Atomic and ionic absorption is shown as a function of plasma observation height. At increasing plasma observation heights, the ground state ionic population of magnesium and thus the ionic absorbance decreases; however, the ground state atomic population and absorption of manganese appears to remain fairly constant over the range of absorption heights shown in Figure 26. Background Correction and Spectral Interferences In this work, the multiwave length capabilities of the photodiode array were utilized to acquire background corrected spectra without the need for special background correction techniques. 93 9 '' When measuring the absorbance peak height, the spectral baseline was automatically subtracted in the final spectrum using the supplied software. Both Figures 24 and 25 exhibit a great deal of noise in their spectral baselines, which is attributable to a combination of signal and source shot noise and detector noise. Analytical Figures of Merit Detection limits for the 14 elements studied are reported in Table IX. Except for the atomization cell, the experimental system used was identical to one used previously. 92 For most elements, the detection limits by CSAAS in the ICP were at least an order of magnitude poorer than our previous results in a flame. The detection limits found for this work are approximately two orders of magnitude worse than with the best previously reported detection limits for CSAAS in an air-acetylene flame. 77 Several causes for this are (i) the photodiode array has a lower sensitivity than more conventional detectors such as

PAGE 85

77 Table IX. Limits of Detection Observed for AAS Limits of Detection (nig L" : ) Element

PAGE 86

78 photomultiplier tubes; (ii) the shorter absorption pathlength of the ICP, namely < 0.5 cm versus 5 10 cm (as seen with flames and graphite furnaces) ; and (iii) the optical throughput of our system was lower because of the use of two monochromators (especially with the HR1000 conochromator in the fifth order) The best detection limits reported for commercially available line-source flame AAS instruments are still an order of magnitude better than the best CSAAS results, especially within the UV spectral region. Despite the relatively poor detection limits of the ion absorption lines (see Table IX) the ability to probe the ionic absorption transitions is advantageous especially with regard to the possibilities it offers for diagnostic studies in the ICP. The linear dynamic range for each element was between one and two orders of magnitude, with the calibration curves deviating significantly from linearity above 0.1 absorbance units. This range can be greatly increased, however, by selecting less sensitive absorption lines or by measuring the absorbance off the absorption line peak when the major lines reach absorbances greater than 0.1. Conclusions The ICP has several advantages as an atom reservoir for atomic and ionic absorption spectrometry. The ability to obtain real time, simultaneous multiwavelength measurements makes the photodiode array much sore attractive than conventional scanning spectrometers with conventional detectors for the measurement of CSAAS spectra. The optical throughput of our HR1000 monochromator should significantly improve with a grating blazed to enhance wavelengths from 200-300 nm in fifth order. Also, the

PAGE 87

79 ability to measure several elemental atomic or ionic absorption lines simultaneously, improves the informing power of the entire system. In flame-AAS, one is able to probe sufficiently only the atomic transitions. In an ICP, many elements are readily ionized and therefore the possibility of obtaining absorption spectra of virtually all atoms and Ions simultaneously with a continuum light source is very attractive. The serious disadvantage of a short absorption pathlength should be overcome with a specially designed ICP torch.

PAGE 88

CHAPTER 6 ATOMIC ABSORPTION SPECTROMETRY IN AN ICP MOUNTED WITH A "T-SHAPED" BONNET Introduction Since the beginning of flame AAS workers have tried to enhance the analytical sensitivity by increasing the absorption path length. 96 Fuwa and Vallee performed a study of the Beer-Lambert law for absorption of molecules in solution as a model for the investigation of atomic absorption. 97 They discovered that the absorption sensitivity is a function of the length of the atomic absorption path length. In their work, a long path absorption cell (90 to 250 cm) was made of Vycor for flame AAS. The combustion flame burner was positioned such that the tail flame was directed into the absorption cell. Source radiation from an HCL passed through the absorption cell and then through focusing optics onto a spectrograph. Compared to conventional flame AAS, they were able to increase the sensitivity for some elements by two orders of magnitude with this long path absorption cell. 97 98 Other work using this long path absorption cell showed that the diameter, and reflection from the inner walls of the absorption tube affect the absorbance 99 The smaller the diameter (up to 1 cm) the higher will be the concentration of absorbing atoms and hence the higher the absorbance. 97,99 Earlier stages of the work 80

PAGE 89

81 by Fuwa and Vallee involved placing the tube in a Tshape fashion over the flame burner and thus deflecting the flame in two direction; 97 however, this system failed to yield satisfactory results with AAS Rubeska also used Fuwa long absorption cells in a T-shaped fashion over a flame. In this setup, the absorption cell was electrically heated so as to increase the mean lifetime of free atoms with the absorption cell. 95 With the long path absorption cell, good sensitivities were obtained and the formation of oxides for some elements was reduced since the tube shields the flames gases from the oxidizing atmosphere. In 1966, Wendt and Fassel utilized a multipass system for ICP-AAS. 72 In their setup, a collimated beam from an HCL made three passes through an ICP with mirrors placed around the plasma and then into a spectrometer. Veillon and Margoshes used a modified Wendt and Fassel ICP torch (without the most central concentric tube) using a single pass system of a modulated source and phase sensitive amplifier. 73 They did not observe the chemical interferences commonly observed in flames. Veillon and Margoshes concluded that except for a few refractory elements, the ICP does not appear to be a suitable replacement for the chemical flame in AAS. Like Fuwa and Vallee, Greenfield conducted work involving a physical increase in the absorption path length, however, in an ICP. 200 Greenfield was able to extend the outer sheath of the ICP torch to 24 inches, and later found he was able to bend the extended plasma "tail-flame" at right angles. With further work along this discovery, the outer sheath of the ICP took on the shape of a "T" When one end of the "T" was closed with an optical flat, the plasma "tail-flame" was forced to turn away from the optical flat. A modulated HCL was collimated and sent through the optical flat

PAGE 90

82 and into a spectrometer near the open end of the "T" Temperatures within the "T" were recorded from 3000 to 8000 K. The high temperature in the "T" should minimizes chemical interference due to the formation of stable refractory compounds of the element under investigation. Also, a lower background radiation was emitted by the tail -flame, thus indicating an advantage over combustion flames in AAS Almost a decade later (1977), Mermet and Trassy performed work on increasing the absorption path length in a ICP. 75 Here, the absorption path length was increased by having the light beam pass through the plasma along the axis symmetry. The plasma torch was turned on its side and a beam of radiation, focused with a lens at the base of the torch, was sent down the torch axially. The outer sleeve of the torch was extended to further enhance the lifetime of neutral atoms in the plasma. Other work by Magyar and Aeschbach, involved removing the atomization cell from a commercial instrument and replacing it with an ICP. 71 The enhanced atomic emission signal from the ICP made it necessary to use higher intensity HCLs than is sufficient for flame AAS. Their results indicated the possible use of ICP-AAS for the determination of metals in complex compounds, which do not efficiently dissociate in combustion flames. Downey and Nogar placed an ICP within the cavity of a pulsed, flashlamppumped dye laser for AAS and achieved and enhancement factor of approximately 170 relative to single-pass absorption in an ICP. 7 '' This technique is known as intracavity dye laser absorption spectroscopy (IDLAS). An HCL was used as an optogalvanic detector in this system. LODs in the parts-per-million range were obtained for sodium and barium. Barium ion absorption was also studied because it is easily ionized.

PAGE 91

S3 Other authors have performed some successful fundamental studies on ICP argon chemistry using AAS. 101 103 In this work, further investigation has been performed utilizing an ICP for AAS studies. An attempt to increase the absorption path length of the ICP was accomplished with the design of a cooled quartz "T-shaped" bonnet for the ICP. Unlike Greenfield's "T" shaped torch, this bonnet was detachable from the torch and was positioned closer to the load coil. An evaluation of this "T-shaped" bonnet was performed for ICP-AAS studies for pure and multielement solutions with a continuum source. Experimental Design and Construction of the "T-shaped" bonnet A detailed drawing of the "T-shaped" bonnet is shown in Figure 27. The bonnet was constructed entirely of quartz tubing (Thermal American Fused Quartz, Monteville, NJ). The bonnet was constructed within our glass shop. Tubing for the "T" portion consisted of two concentric tubes. The tubing used for the outside portion was 25 mm i.d. and 28 mm o.d. and the tubing used for the inside portion of the "T" was 17 mm i.d. and 19 nun o.d.. These two tubes were fused together at their ends so as to create a water jacket between them. The length of the "T" was 115 mm long. The tubing for the stem of the "T" was 22 mm i.d. and 25 mm o.d. and was 25 mm long. The tubing used for the intake and return of chilled water to the "Tshaped" bonnet was 6 mm i.d. and 8 mm o.d.. Chilled water to the bonnet was kept at < 5 C (Neslab Instruments, Inc., Portsmouth, N'H) The flow

PAGE 92

84 T3 CO CV2 a CV2 a a CM o a to

PAGE 93

85 rate of water to the bonnet was approximately 2-3 gallons per hour. The "T-shaped" bonnet as mounted in two nylon clamps. The clamps were constructed within our machine shop and were attached to an X, Y, Ztranslation axis mount such that the "T-shaped" bonnet could carefully be positioned on top of the ICP. Figure 2S is a diagram of the bonnet within the clamps. This translation also allowed the ability to optically probe different regions within the "T-shaped" bonnet. The ICP torch was also mounted on X, Y, Ztranslation axis mount such that the torch could be properly position within the "T-shaped" bonnet. The torch translation was very useful during plasma ignition. Operation. Before plasma ignition, chilled water was allowed to flow through the "T-shaped" bonnet and ICP load coil for approximately 5 min. This ensured that the bonnet was properly cooled. Figure 29 is a schematic diagram of the ICP torch positioned within the "T-shaped" bonnet. Ignition of the plasma with the torch inside the "T-shaped" bonnet was extremely difficult; therefore, before plasma ignition the ICP torch had to be lowered such that the bottom of the bonnet stem was at least 20 25 mm above the top of the outer quartz sheath of the torch. Once the plasma was ignited in this lowered position, the torch was then slowly raised into the "T-shaped" bonnet until the center of the longitudinal "T" portion of the bonnet was approximately 27 mm above the torch load coil. Access to all translation axis mounts was well away frc the r.f. load coil of the ICP and thus could be safely manipulated durint torch operation. Once in position, the torch was operated as usual. The experimental operating conditions are listed in Table X. As expected, the :om

PAGE 94

86

PAGE 95

87 Water Out A ft T-shaped Bonnet Figure 29. Diagram of the ICP torch within the bonnet.

PAGE 96

88 Table X. Experimental operating conditions, I. ICP Forward r.f. power 400-600 W Reflected r.f. power 0-20 W Ar coolant flow rate 17 L min" 1 Ar auxiliary flow rate 0.5 L min" 1 Ar carrier flow rate 0.9 L min" 1 Observation height* 27 mm II. Ultrasonic nebulizer' III. Xenon arc lamp Power 300 W Current 17 A Measured above the load coil. See Table VI

PAGE 97

89 torch was operated at low powers (i.e., 400 to 600 W) so as not to overheat and possibly crack the "T-shaped" bonnet. At these low ICP powers, this experimental setup could be run continuously (the maximum continuous operation at one time used was 8 h) Instrumentation Figure 30 is a schematic diagram of the optical arrangement used in this work. The optical probing of atomic and ionic species within the center of the "T-shaped" bonnet was 27 mm above the ICP load coil. As discussed in chapter 5, the instrumentation (except for the "T-shaped" bonnet) sample preparation, and experimental procedure were identical Plasma Temperature and Electron Density Spectroscopic measures are one of the best means for obtaining spatially and temporally resolved measurements of temperature and species number densities without perturbing the mechanisms of the plasma or influencing the temperature. 104,105 An accurate knowledge of the plasma temperature can lead to a better understanding of analyte vaporization, dissociation, atomization and ionization processes. 106 As discussed by Browner and Winefordner, a two-line atomic absorption method was employed. 10A The elements thallium and indium were chosen based on their working ranges of temperatures. The advantage of using this twoline atomic absorption method are that the temperatures are averaged through the optical path in the plasma. A spectroscopic emission method was also performed as a comparison to the twoline absorption method as mentioned above. Neutral iron was selected as the thermometric species. A factor in the line selection process for iron was based on (i) maximal spread in excitation levels; (ii) freedom from plasma spectral interferences; (iii)

PAGE 98

90

PAGE 99

91 availability of accurate transition probabilities; and (iv), wavelength proximity, which eliminates the need for calibrating detector instrumentation response with respect to wavelength. 106 The wavelengths, excitation energies, statistical weights and transition probabilities were taken from a reference by Fuhr Martin, and Wiese. 107 The method used for determination of relative electron (number) density was based on the measurement of the Stark broadening of an atomic hydrogen line. Stark broadening is due to the interaction of charged particles in a dense plasma. The electron density is proportional to the Stark broadening and the electric field strength. The Stark half-width is the most reliable and convenient method for the determination of electron densities. 103 The H B (486.1 nm) line was chosen because it is generally free from plasma spectral interferences, has a sufficient intensity for measurement, a small half-width (1 5 A) and extensive Stark data is available for this line. 109 110 Both Doppler broadening and instrumental broadening are usually negligible compared to Stark broadening. 109 Results and Discussion After extended use of this experimental system, and especially after aspiration of high salt concentration into the plasma, an oxide formation was observed within the inner wall of the "T-shaped" bonnet. Despite this residue formation, there were no analyte memory effects observed within the system. After periodic use, the "T-shaped" bonnet was removed and immersed in a 10% HN0 3 solution to clean the inner walls of the bonnet. Absorption spectra, background correction and analytical figures of merit

PAGE 100

92 were obtained as described in chapter 5. Detection limits for some elements studied are reported in Table XI. These results are compared to early work done to improve the absorption pathlength in an ICP. In all cases, except the work done by Mermet, the LODs obtained with this setup were superior. Mermet and co-workers were able to achieve excellent results by viewing the ICP axially. 111 Detection limits for some elements are also reported in Table XII. These results are compared to absorption data taken from previous work (see chapter 5) as well as some literature values. As expected, better LODs (by 1 to 2 orders of magnitude) were achieved with this work as compared to AAS results with a conventional torch due to the increase in atomic (ionic) absorption path length and longer analyte residence times within the "T-shaped" bonnet. However, these results were still poorer compared to the best continuum source and line source in flame AAS (LODs of 1 and 1 to 2 orders of magnitude poorer, respectively) As discussed by Browner and Winefordner, the working temperature ranges for thallium and indium are restricted: thallium-3650 to 7750 K and indium-1220 to 3500 K. 10 '' Atomic absorption for thallium (the 535.1 nm line) was not observed; therefore, the plasma more than likely has a temperature less than 3650 K since it does not sufficiently populate an upper degenerate ground state of thallium. Atomic absorption results were thus obtained using indium. Based on this two-line absorption method, the plasma temperature at an r.f. power of 500 W was found to be 2165 K. As a comparison, the atomic emission of iron was also used to determine the plasma temperature within the "T-shaped" bonnet. As described by Kalnicky and Kniseley, the temperature calculation was performed with the aid of

PAGE 101

93 Table XI. Limits of Detection Compared to Literature Values for ICP-AAS. Limits of Detection (/ig mL" 1 ) Element This MulitAxial ICP ICP IntraWork Pass 3 View 15 HCL C HCL d Cavity 8 Ag(I)

PAGE 102

94 Table XII. Limits of Detection Compared to Best AAS Results. Element

PAGE 103

95 the slope method for several iron (Fe I) emission lines. 112 The temperature determined from this method was 3,974 K. The difference in the calculated temperatures from the two methods (absorption and emission) supports the fact that the plasma within the "T-shaped" bonnet is not in thermodynamic equilibrium. The electron density was determined by the calculated half-width of the H Q line as a function of the electron density. 108 From this function, the experimentally determined value of the electron density can be read. Based on a H half -width of 0.7 A, the electron density corresponds to 2 X 10 1A cm" 3 The measured electron density has also been used to calculate the electron temperature, T e LIE (K) assuming local thermodynamic equilibrium. T e LTE values have been evaluated for the corresponding electron densities. 113 According to Caughlin and Blades, 113 the LTE temperature based on the measured electron density corresponds to 6827 K. The difference in this temperature from the two temperatures mentioned above, further proves nonthermodynamic equilibrium. A multielement synthetic mixture containing several elements (at 100 /ig ml/ 1 concentrations) was prepared and analyzed. Figures 31 to 35 show absorption spectra of this synthetic multielement mixture at judiciously chosen wavelengths. Some spectra have several absorption lines for a particular element demonstrating the greater informing power of the system. Figure 35 shows not only a multielement scan for magnesium and manganese but an atomic (Mn) and ionic (Mg) absorption scan as well. LODs for many of the ionic absorption lines would probably improve if the plasma r.f. power could be increased. However, increasing the r.f. power may jeopardizes the structure of the "T-shaped" bonnet. Low r.f.

PAGE 104

0) u c D _Q L_ O CO _Q < 230.50 96 1 I i i | i i i i i i i i | i i i i i i i i i | i i | i i 231.00 231.50 232.00 232.50 Wavelength (nm) Figure 31. Atomic absorption spectrum of Ni and Co (100 /jg ml/ 1 )

PAGE 105

97 CD O c D _Q k_ O CO < Fe Fe Fe Fe Fe Fe 247.70 1 I i i i i i i [ i i i i i i i i i i i i i i i i i i TT 1 248.10 248.50 248.90 249.30 Wavelength (nm) Figure 32. Atomic absorption spectrum of Fe and Cu (100 pg ml/ 1 )

PAGE 106

93 Cr CD U c D _Q l_ O (n _Q < 357.25 357.45 357.65 357.85 358.05 358.25 Wavelength (nm) Figure 33. Atomic absorption spectrum of Cr, Fe and Co (100 pg ml/ 1 )

PAGE 107

99 CD U C o o ID _Q < i ''" 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 1 1 1 1 1 1 i 1 1 1 1 1 1 327.00 327.20 327.40 327.60 327.80 328.00 328.20 328.40 Wavelength (nm) Figure 34. Atomic absorption spectrum of Cu and Ag (100 ^g ml/ 1 ).

PAGE 108

100 u c D jQ o w _Q < i i i i i i i i i i i i i i i i i | i i i i i i i i i i i i i i i i i i i | i i i i i i i i i | i i i i i i i i i I i i i 279.10 279.30 279.50 279.70 279.90 280.10 280.30 Wavelength (nm) Figure 35. Atomic and ionic absorption spectrum of Mg and Mn (10 (iz ml/ 1 )

PAGE 109

101 operating power allowed prudent operation of the system and prevented the potential of cracking the bonnet. Since the "T-shaped" bonnet is made of quartz, the bonnet acted like a "light-pipe"; i.e., the plasma continuum emission was reflected from the ends of the bonnet. Most of this plume emission was spatially discriminated against by using appropriate apertures. There was very little atomic emission observed within the center of the "T" of the bonnet. The greatest advantage of this "T-shaped" bonnet is the ability to optically probe atoms within a fairly dark region. Compared to flames, this is attractive since the background emission is very low. Conclusion These preliminary results with the "T-shaped" bonnet seem promising. The ability to increase the atomic absorption path length and the utility of an excellent atomization cell, such as the ICP seems very attractive. The bonnet can be used continuously on top of the ICP torch making it attractive for commercial use. Future work will employ the use of an HCL and possibly other line sources as radiation sources with this setup. With the relatively low plasma temperatures determined in this study, the plasma within the middle of the "T" portion of the bonnet appears to be a very hot gas. Future work will also involve the increase of r.f. powers to obtain higher plasma temperatures and electron densities. The maximum r.f. power that the "T-shaped" bonnet can withstand without cracking or melting will be determined.

PAGE 110

CHAPTER 7 SUMMARY Compared to conventional line sources for AFS and AAS the ability to use one source to simultaneously excite many elements is very attractive. The use of a continuum source is an elegant and optically simple approach. Despite the need for a high resolution monochromator and background correction, the convincing advantage for use of a continuum source is a single light source to achieve simultaneous multielemental analysis with good sensitivity. Thus, the continuum source does away with the need to use one source per element. As discussed in chapters 3 and 4, the continuum source is also a non-selective excitation source. Therefore, some of the selectivity advantages as seen with line sources can be gained only if the proper choice of optical filters is utilized. Compared to other atomization cells, the ICP is an attractive alternative. Despite the initial capital outlay, operating costs, and difficulty of use, its high sample throughput, good sample delivery, small sample volumes, lack of chemical interferences, and ability to excite atomic and ionic transitions makes the ICP the atomization of choice when used with continuum sources. As the atomic and ionic transitions are excited in an ICP, the continuum source can optically probe these transitions. With the use of a photodiode array, real time, simultaneous 102

PAGE 111

103 multiwavelength measurements can be obtained. This makes the photodiode array much more attractive than conventional scanning spectrometers with conventional detectors, especially with a continuum source and an ICP. The construction of the ultrasonic nebulizer provided better and more efficient sample introduction compared to pneumatic nebulization. More work is needed to improve the sample introduction efficiency, memory effects, and spray chamber mixing of the ultrasonic nebulizer system. The majority of this work will be focused on instrumental design. In an attempt to increase the atomic absorption path-length in an ICP, the preliminary results with the "T-shaped" bonnet seem promising. Future work in this area will involve further diagnostic studies on the plasma. Also, adjustments to various parameters such as bonnet size, ICP r.f. power, and height above the load coil will be tried. Work will also be performed in an attempt to evaluate the system utilizing line sources. The combination of the ultrasonic nebulizer, continuum source, ICP, and photodiode array should prove itself as a very useful method and capable of solving many analytical problems with convenience and speed. This approach should play a role in the future of analytical atomic spectroscopy.

PAGE 112

APPENDIX A PLASMA TEMPERATURE Two-Line Atomic Absorption Method 10A The relative absorption, a (dimensionless) of radiation from a continuum light source in the limiting case of low optical density is =
PAGE 113

105 where E : and E are the energy levels of the upper and lower energy level states (J), respectively, g 1 and g are the statistical weights of the upper and lower states, respectively (dimensionless) and k is the Boltzmann constant (J/K) Substituting the constants for indium at A MS = 410.2 and 451.1 nm, the absolute temperature equation can be written as T el (K) = 3146 / [In (2.637 (a / ai ) ) ] (12) Once a and a x are determined, T el can be determined for low optical densities. The twoline absorption method should be less susceptible to errors due to chemical, excitational and radiational disequilibrium than the emission method. Iron Emission Slope Method 112 The slope temperature method is acquired by the following equation: In [(giAjQj/K,) / I 10 )] = Z l / (kT) + In [ (47rg ) / (hn ) ] (13) where I 10 is the emission intensity of the transition 1 (J m" 2 Hz), T is the temperature (K) g lt and g are the statistical weights of the emitting and ground level, respectively, u w is the frequency of emission (Hz), A 10 is the transition probability for spontaneous emission (s _1 ) E x is the energy of the emitting level (J), k is Boltzmann' s constant (J/K), h is Planck's constant (J s) and n is the number density of the ground level per m 3 For emission lines originating from the same ionization level, a plot of In [(gAi/)/I] versus E : should yield a straight lino, with a slope equal to l/(kT) where T is the slope temperature.

PAGE 114

APPENDIX B ELECTRON DENSITY The first choice in the determination of electron density in dense plasma should be the measurement of the HA (486.1 nm) line. This line is strong, it is sufficiently broadened for precise measurements, and the possibility of self absorption is relatively small. Also, their is very good agreement with theoretical and experimental values (experimental values agree within 7% of the theoretical values) For most diagnostic applications, the relationship between the full width at half maximum (FrtTiM) AA 1/2 (A), and electron density, n e is AA 1/2 (A) 2.5 X 1CT 9 a 1/2 nf 3 (14) where a is the reduced wavelength distance (A) (i.e., a 1/2 is the value of a at one-half the maximum intensity of the profile). A linear plot is obtained as electron density is plotted as a function of FWHM. 108 Assuming local thermodynamic equilibrium (LTE) the electron number density can be used to derive an equilibrium temperature by the following equation n? [(n + n e )/g + ] [h 3 /( 2 (27rmkT e ) 2/3 ) exp [ (I 1 -E 1 )/(kT e ) ] (15) 106

PAGE 115

107 where nf refers to the Saha population (nf 3 ) n* is the density of the argon ion ground state (nf 3 ) g + is the statistical weight of the argon ion ground state (dimensionless) h is Planck's constant, m is the electron mass, k is Boltzmann's constant, T e is the electron temperature (K) I : is the ionization potential of argon (J), and E 4 is the excitation energy of level i (J). From this equation and the fact that the ICP is an atmospheric plasma with a low ionization degree, T e LTE values have been evaluated for corresponding n e values. 113

PAGE 116

REFERENCES 1. M.Slavin, "Atomic Absorption Spectroscopy", John Wiley & Sons, New York, 1978. 2. B.Welz, "Atomic Absorption Spectroscopy", Verlag Chemie New York 1976. 3. W. Slavin, "Atomic Absorption Spectroscopy" Interscience Publishers New York, 1968. 4. W.G. Schrenk, "Analytical Atomic Spectroscopy", Plenum Press, New York, 1975. 5. J.V. Loon, Anal. Chem. 53, 332A (1981). 6J-D. Winefordner and T.J. Vickers, Anal. Chem., 36, 161 (1964). 7. J.D. Winefordner and R.A. Staab, Anal. Chem., 36, 165 (1964). S. J.D. Winefordner and R.A. Staab, Anal. Chem., 36, 1367 (1964). 9. A. Montaser andV.A. Fassel, Anal. Chem., 48, 1490 (1976). 10. A.H. Ullman, Prog. Analyt. Atom. Spectrosc, 3, 87 (1980). 11. J.D. Winefordner, R.C. Elser, Anal. Chem., 43, 24A (1971). 12. J.D. Winefordner, V. Svoboda, and L.J. Cline, CRC Crit. Rev., Anal Chem. 1, 233 (1970) 13. W.J. Price, "Spectroscopy," D.R. Browning, Ed., McGraw-Hill New York, 1969. 14. D.W. Ellis and D.R. Demers Advan. Chem. Ser. 73, 326 (1968). 15. R. Smith, "Spectrochemical Methods of Analysis," J.D. Winefordner, Ed., Interscience, New York, 1971. 16. J.D. Winefordner and R. Smith, "Analytical Flame Spectrometry, Selected Topics," R. Maurodineanu, Ed., Centres, Eidhouen, The Netherlands, 1969, Chapter 11. 17. A. Walsh, Spectrochim. Acta, 7, 108 (1955). 108

PAGE 117

19. 20. 27. 30. 109 C.T.J. Alkemaae and J.M.W. Milatz, J. Opt. Soc, Amer. 45 583 (1955). M.L. Parsons, B.W. Smith, and P.M. McElfresh, Appl. Spectrosc, 27 471 (1973). H. Uchida, Ph.D. Thesis, Industrial Research Institute of Kanagawa Prefecture, Showa-machi, Yokohama, Japan, 1981. 21. G.I. Babat, J. Inst. Elect. Eng., 94, Part III, 27 (1947). 22. T.B. Reed, J. Appl. Phys 32, 2534, 1961. 23. S. Greenfield, I.L. Jones, and C.T. Berry, Analyst, 89, 713 (1964). 24. V.A. Fassel and R.N. Kniseley, Anal. Chem., 46, 1110A (1974). 25. V.A. Fassel and R.N. Kniseley, Anal. Chem., 46, 1155A (1974). 26. R.H. Scott, V.A. Fassel, R.N. Kniseley, and D.E. Nixon, Anal. Chem 46, 75 (1974). C.F. Larson, V.A. Fassel, R.H. Scott, and R.N. Kniseley, Anal Chem. 47, 238 (1975). 28. P.W.J.M. Boumans and F.J. De Boer, Spectrochim. Acta 27B 391 (1972). 29. P.W.J.M. Boumans and F.J. De Boer, Spectrochim. Acta, 30B 309 (1975). P.W.J.M. Boumans and F.J. De Boer, Spectrochim. Acta 31B 355 (1976). 31. T. Hasegawa and H. Haraguchi, "Inductively Coupled Plasmas in Analytical Atomic Spectrometry", A. Montaser and D.W. Golightly, Eds., VCH Publishers, Inc., New York, 1987. p. 267. 32. G.A.Meyer, Anal. Chem., 59, 1345A (1987). 33. N. Omenetto and J.D. Winefordner, "Inductively Coupled Plasmas in Analytical Atomic Spectrometry", A. Montaser and D.W. Golightly, Eds., VCH Publishers, Inc., New York, 1987. p. 323. 34. V.A. Fassel, Science, 202, 183 (1978). 35. J.R. Garbarino and H.E. Taylor, Appl. Spectrosc, 34, 584 (1980). 36. M.P. Douherty and G.M. Hieftje, Appl. Spectrosc, 38, 405 (1984). 37. J. Stupar and J.B. Dawson, Appl. Optics, 7, 1351 (1968).

PAGE 118

110 33. J.D. Ingle, Jr., and S.P. Crouch, "Spectrocheraical Analysis", Prentice Hall, Englewood CLiffs, NJ 1988; Chapter 7. J9. R.L. Willoughby and R.F. Browner, Anal. Chem. 56, 2626 (1984). P.C Winkler, D.D. Perkin, W.K. Williams, and R.F. Browner Anal Chem. 60, 489 (1988) -0. 41. C.T. Apel, T.M. Bieniewski, L.E. Cox, and D.W. Steinhaus ICP Inf Newsl. 3(1), 1 (1977). 42. V.A. Fassel and B.R. Bear, Spectrochim. Acta, 41B, 1089 (1986). 43. CD. West and D.N. Hume, Anal. Chem., 36, 412 (1964). 44. R.H. Wendt and V.A. Fassel, Anal. Chem., 37, 920 (1965). 45. J. Spitz and G. Uny, Appl. Optics, 7, 1345 (1968). 46. K.W. Olson, W.J. Haas, Jr., and V.A. Fassel, Anal. Chem 49 632 (1977). 47. J. A. Borowiec, A.Q. Boorn, J.K. Dillard, H.S. Cresser, R.F. Browner, and M.T. Matteson, Anal. Chem., 52, 1054 (1980). 48. E.D. Prudnikov and Y.S. Shapkina, Anal. Chim. Acta, 175, 329 (1985). 49. H.J. Issaq and L.P. Morgenthaler Anal. Chem., 47, 1661 (1975). 50. G. Tolg, Analyst, 112, 365 (1987). 51. J.A.C. Broekaert, Anal. Chim. Acta, 196, 1 (1987). 52. J.A.C. Broekaert and G. Tolg, Z. Anal. Chem., 326, 495 (1987). 53. A. iMontaser and V.A Fassel, Anal. Chem., 48, 1940 (1976). 54. L.A. Davis, R.J.Krupa, and J.D. Winefordner, Anal. Chim Acta 173 512 (1985). 55. N. Omenetto and J.D. Winefordner, in "Inductively Coupled Plasmas in Analytical Atomic Spectrometry", A. Montaser and D W Goliehtly Eds., VCH, New York, 1987, p. 323. 56. D.J. Johnson, W.K. Fowler, and J.D. Winefordner, Talanta 24 227 (1977). 57. D.J. Johnson, F.W. Plankey, and J.D. Winefordner, Anal Chem 46 1898 (1974). 53. G. Beck, Rev. Sci. Instrum. 45, 318 (1974).

PAGE 119

Ill 59. M.L. Parsons, B.W. Smith, and G.E. Bentley, "Handbook of Flame Spectroscopy", Plenum Press, New York, 1975, p. 16. 60. 61. 62 D.R. Demers, D.A. Busch, and CD. Allemand, Am. Lab., 22 no 3 167 (1982). R.J. Krupa, G.L. Long, andJ.D. Winefordner, Spectrochim. Acta 40B 1485 (1985). W. Demtroder, "Laser Spectroscopy-Basic Concepts and Instrumentation", Springer-Verlag, New York, 1982, p. 434. 63. N. Omenetto and H.G.C. Human, Spectrochim. Acta, 39B, 1333 (1984) 64. N. Omenetto, B.W. Smith, L.P. Hart, P. Cavalli, and G. Rossi, Spectrochim. Acta, 40B, 1411 (1985). 65. M.E. Tremblay, J.B. Simeonsson, B.W.Smith, and J.D. Winefordner Appl. Spectrosc, 42, 281 (1988). 66. A.W. Miziolek and R.J. Willis, Optics Lett., 6, 528 (1981). 67. N. Omenetto, P. Cavalli, M. Broglia, P. Qi and G Rossi, J Anal Atom. Spec. 3, 231 (1988) 68. M. Leong, J. Vera, B.W. Smith, N. Omenetto, and J.D. Winefordner Anal. Chem. 60, 1605 (1988). 69. M.A. Mignardi, B.W. Smith, B.T. Jones, R.J. Krupa, and J.D. Winefordner, Talanta, 36, 311 (1989). 70. B.W. Smith and M.L. Parsons, J. Chem. Ed., 50, 679 (1973). 71. B. Magyar and F. Aeschbach, Spectrochim. Acta, 35B, 839 (1980). 72. R.H. Wendt and V.A. Fassel, Anal. Chem., 38, 337 (1966). 73. C. Veillon and M. Margoshes Spectrochim. Acta, 23B, 503 (1968). 74. S.W. Downey and N.S. Nogar, Appl. Spectrosc, 38, 876 (1984). 75. J.M. Mermet and C. Trassy, Appl. Spectrosc, 31, 237 (1977). 76. G.F. Kirkbright and M. Sargent, "Atomic Absorption and Fluorescence Spectroscopy", Academic Press, New York, 1974, Ch 5. 77. T.C. O'Haver, Analyst, 109, 211 (1984). 78 79 J. Marshall, B.J. Ottaway, J.M. Ottaway, and D. Littlejobn Anal Chim. Acta, 180, 357 (1986). J.M. Harnly, Anal. Chem., 58, 933A, (1986).

PAGE 120

112 80. T.C. O'Haver and J.D. Messman, Prog. Analyt. Spectrosc, 9, 483 (1986). 81. W. Snelleman, Spectrochim. Acta, 23B, 403 (1968). 82. A.T. Zander, T.C. O'Haver, and P.N. Keliher, Anal. Chem. 48, 1166 (1976). 83. J.D. Messman, M.S. Epstein, and T.C. Rains, Anal. Chem., 55, 1055 (1983). 84. A.T. Zander, T.C. O'Haver, and P.N. Keliher, Anal. Chem., 49, 838 (1977). 85. T.C. O'Haver, Anal. Chem., 51, 91A (1979). 86. 87 89. N.J. Miller-Ihli, T.C. O'Haver, andJ.M. Harnly, Anal. Chem., 54 799 (1982). T.C. O'Haver, J.M. Harnly, J. Marshall, J. Carroll, D. Littlejohn, and M. Ottaway, Analyst, 110, 451 (1985). A.J. Owen, "The Diode-Array Advantage in UV/Visible Spectroscopy", Hewlett-Packard Co. Publication No. 12-5954-8912, Federal Republic of Germany, 1988, Ch. 6. G. Gillson and G. Horlick, Spectrochim. Acta, 41B, 431 (1986). 90. P. Tittarelli, R. Lancia, and T. Zerlia, Anal. Chem 57 2002 (1985). 91. J.M. Shekiro, Jr., R.K. Skogerboe, and H.E. Taylor, Anal Chem 60 2578 (1988). 92. B.T Jones, M.A. Mignardi, B.W. Smith, and J.D. Winefordner, J. Anal. Atomic Spectrosc, Submitted (1989). 93. P.W.J.M. Boumans "Inductively Coupled Plasma Emission Spectroscopy, Part 1", John Wiley & Sons, New York, 1987, Ch. 7. 94. A. Zander, "Inductively Coupled Plasmas in Analytical Atomic Spectroscopy", A. Montasser and D.W. Golightly, Eds., VCH Publishers, Inc., New York, 1987, Ch 6 95. Thermo Jarrell Ash, Technical Report, 1/88, 5K, Franklin, MA. 96. Rubeska and B. Moldan, Appl. Opt., 7, 1341 (1968). 97. K. Fuwa and B.L. Vallee, Anal. Chem., 35, 942 (1963). 98. A. Ando, K. Fuwa, and B.L. Vallee, Anal. Chem., 42, 818 (1970).

PAGE 121

113 99. E.J. Agazzi, Anal. Chem. 37, 364 (1965). 100. S. Greenfield, P.B. Smith, A.E. Breeze, and N.M.D. Chilton, Anal. Chim. Acta, 41, 385 (1968). 101. ?. Hart, B.W. Smith, and N. Omenetto, Spectrochim. Acta, 41B, 1367 (1986). 102. H. Uchida, K. Tanabe Y. Nojiri, H. Haraguchi, and K. Fuwa, Spectrochim. Acta, 35B, 881 (1980). 103. M.W. Blades and G.M.Hieftje, Spectrochim. Acta, 37B, 191 (1982). 104. R.F. Browmer and J.D. Winefordner, Anal. Chem., 44, 247 (1972). 105. D.Robinson and P.D. Lenn, Appl. Optics, 6, 983 (1967). 106. D.J. Kalnicky, R.N. Kniseley, and V.A. Fassel, Spectrochim. Acta, 30B, 511 (1975). 107. J. P. Fuhr, G.A. Martin, and W.L. Wiese, in "Atomic Transition Probabilities, Iron Through Nickel", J. Phys Chem. Ref. Data, 17(4), American Institute of Physics, New York, 1988, p 13. 108. W.L. Wiese, in "Plasma Diagnostics Techniques," R.H. Huddlestone and S.L. Leonard, Eds., Academic Press, New York, 1965, Ch 6. 109. D.J. Kalnicky, V.A. Fassel, and R.N. Kniseley, Appl. Spectrosc, 31, 137 (1977). 110. B. Kirsch, S. Hanamura, and J.D. Winefordner, Spectrochim. Acta, 39B, 955 (1984). 111. M.H. Abdallah, R. Diemiaszonek, J. Jarosz, J.M. Mermet, J. Robin, and C. Trassy, Anal. Chim. Acta, 84, 271 (1976). 112. D.J. Kalnicky and R.N. Kniseley, Spectrochim. Acta, 30B, 511 (1975). 113. 3.L. Caughlin and M.W. Blades, Spectrochim. Acta, 39B, 1583 (1984).

PAGE 122

BIOGRAPHICAL SKETCH Michael A. Mignardi was in born in West Babylon, New York, on July 22, 1959, to Estelle L. and Ferruccio Mignardi, Jr. There were three more siblings to follow: they were William P., Laura A., and Linda A. The entire family moved to Winter Park, Florida, in 1965, where the family is presently. Michael obtained his B.S. degree from the University of Florida in June 1981. He was accepted into the post-baccalaureate program at UF (an excuse to stay in school another year). He then obtained a j ob in December 1982 at Southern Research Institute in Birmingham, Alabama. It was during this time that Michael met his wife-to-be, Marcy M. Webster. It was also during this time that he was promoted to a supervisory level and decided to attend graduate school when he realized he would be a subservient to the Ph.D.s (aka "Phuds") in the world if he did not get an advanced degree. A letter from Jim Winefordner stating to "get the hell out of Alabama" further convinced him to attend graduate school. Michael began graduate school at the University of Florida in August, 1985. He then joined Jim Winefordner s research group. Michael was married to Marcy on July 18, 1987, in Vicksburg, Mississippi. He is presently a candidate for a Doctor of Philosophy degree from the University of Florida. 114

PAGE 123

I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. {for**, F^mes D. Winefordner, Chair Graduate Research Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. rsey \ Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the degree of Doctor of Philosophy. \i Vaneica Y. YoiMg Associate Professor of Chemistry I certify that I have read this study and that in my opinion it conforms to acceptable standards of scholarly presentation and is fully adequate, in scope and quality, as a dissertation for the^d^gree of Doctor of Philosophy. Stephen <£/ Schulman/ Professor of Pharmacy This dissertation was submitted to the Graduate Faculty of the Department of Chemistry in the College of Liberal Arts and Sciences and to the Graduate School and was accepted as partial fulfillment of the requirements for the degree of Doctor of Philosophy. August, 1989 Dean, Graduate School

PAGE 124

UNIVERSITY OF FLORIDA 3 1262 08553 4575