Electrode heating sample vaporization in capacitively coupled microwave plasma atomic emission spectrometry

MISSING IMAGE

Material Information

Title:
Electrode heating sample vaporization in capacitively coupled microwave plasma atomic emission spectrometry
Physical Description:
vi, 155 leaves : ill. ; 28 cm.
Language:
English
Creator:
Ali, Abdalla H., 1957-
Publication Date:

Subjects

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

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 147-154).
Statement of Responsibility:
by Abdalla H. Ali.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001693343
notis - AJA5422
oclc - 25221992
System ID:
AA00003289:00001

Full Text









ELECTRODE HEATING SAMPLE VAPORIZATION IN
CAPACITIVELY COUPLED MICROWAVE PLASMA
ATOMIC EMISSION SPECTROMETRY














By

ABDALLA H. ALI


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

UNIVERSITY OF FLORIDA


1991













ACKNOWLEDGEMENT

It has been a great pleasure being in Winefordner's

group. Every member has been very helpful, but I will extend

special thanks to few who have made my life at the University

of Florida a little easier. First of all, I would like to

thank Dr. Winefordner for his invaluable guidance which was

always vital for our success, his patience and above all his

financial support. I am very grateful to Ben Smith to whom I

turned for answers to the technical problems. I would like to

thank Wellington Masamba for his help in the final days of

dissertation, Chris Stevenson and Dennis Hueber whom we always

bother whenever we have any problems with the computers. For

their friendship and wholehearted support, I am indebted to

Joe Simeonson, Ramee Indralingam, Giuseppe and Nancy Petrucci.

Finally, my deepest gratitude goes to my wife, Insaf, for

her understanding, loving and caring, particularly during the

final days of this work.








TABLE OF CONTENTS


ACKNOWLEDGEMENT..... ........................................ii

ABSTRACT................. ....................................iii

CHAPTER
ONE MICROWAVE PLASMAS ............................ 1

Introduction................................. 1
Microwave Induced Plasmas.................... 1
Resonant Cavities.........................2
Surface Wave Launchers.................... 11
Capacitively CoupledMicrowave Plasmas...... 15

TWO SOLID ANALYSIS IN GRAPHITE ELECTRODE
CAPACITIVELY COUPLED MICROWAVE PLASMA ......... 21

Introduction...............................21
Experimental................................ 24
Results and Discussion......................31

THREE MICROSAMPLING OF LIQUIDS IN GRAPHITE ELECTRODE
CAPACITIVELY COUPLED MICROWAVE PLASMA ......... 39

Introduction...............................39
Experimental.................................42
Results and Discussion......................49
Electrode length optimization............49
Electrode/cup design.....................49
Cup Temperature..........................54
Plasma Temperature.......................57
Analytical Results........................61
Conclusion. ............................. 67

FOUR DIAGNOSTICS IN TUNGSTEN FILAMENT ELECTRODE
CAPACITIVELY COUPLED MICROWAVE PLASMA ......... 68

Introduction............................... 68
Theoretical Considerations..................70
Experimental...............................72
Results and Discussion....................... 75
Precision................................75
Effect of Gas Flow Rate ................. 82
Effect of Power...........................91
Radial Profiles..........................107
Axial Profiles..........................120







FIVE ANALYTICAL FEATURES OF TUNGSTEN FILAMENT
ELECTRODE CAPACITIVELY COUPLED MICROWAVE
PLASMA........................................130

Introduction............................... 130
Experimental.................................131
Results and Discussion.....................132

SIX CONCLUSIONS AND FUTURE WORK...................142

REFERENCES............................. ......................147

BIOGRAPHICAL SKETCH..... ..................................155













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

ELECTRODE HEATING SAMPLE VAPORIZATION IN CAPACITIVELY COUPLED
MICROWAVE PLASMA ATOMIC EMISSION SPECTROMETRY

By

ABDALLA H. ALI

AUGUST 1991

Chairman: James D. Winefordner
Major Department: Chemistry

This work is divided into two major portions, work that

was done with a graphite-cup electrode system and work with a

tungsten filament electrode. The system took advantage of

heating of the electrode to vaporize samples into the plasma.

The analysis of solids, by directly introducing them into the

microwave plasma, was evaluated using the graphite electrode.

Samples used were standard reference materials of coal fly ash

(SRM 1633a) and Tomato Leaves (SRM 1750).

Microsampling of liquids in our plasma was also done.

Detection limits obtained were comparable to the other

microsampling emission techniques and were better than those

of graphite furnace atomic absorption spectrometry in the

visible spectral region. Precision was better than 12% and

linear dynamic range was 3-4 orders of magnitude. The cup

temperature increased proportionally with both the power and







the mole fraction of the dopant nitrogen gas. Temperatures as

high as 2400K can be achieved. The excitation temperature

decreased with power, which showed that the plasma tends

towards thermal equilibrium. Optimization of the length of the

graphite electrode indicated that it should penetrate into the

electric field by 3 cm for optimal coupling of microwave

energy.

Diagnostic studies were made on the tungsten-filament

electrode supported plasma. The effect of power, plasma gas

flow rate, radial and axial positions on excitation and

rotational temperatures, and emission intensities of He and Cu

were investigated. The results indicated that metals and non-

metals have the same optimal conditions and that the spatial

emission was independent of the element type. The effect of

hydrogen dopant gas on the background was examined. The

analytical results were better than those with the graphite

electrode supported plasma.










CHAPTER ONE

MICROWAVE PLASMAS

Introduction

Most of the early research on microwave plasmas utilized

a medical diathermy as a microwave source supplying 100-200 W

and operating at 2.45 GHz. Current microwave plasmas use the

same frequency. The analytical utility of microwave plasmas is

influenced by the instrumentation and the means of coupling

energy into the plasma gas, and is often limited due to

inadequate understanding of various parameters that determine

the performance and the efficiency of the plasma.

Microwave plasmas are classified, according to the means in

which the energy is transferred into the plasma, into

Microwave Induced Plasma (MIP) and Capacitively Coupled

Microwave Plasma (CMP). Each type will be discussed in the

following sections.

Microwave Induced Plasmas

MIPs are generated in resonant cavities or surface wave

launchers. In both cases, the microwave energy is inductively

transferred to the plasma gas. This is realized by inserting

a discharge tube in the resonant device.









Resonant Cavities

The resonant cavities used in spectrochemical analysis

have been reviewed by Fehsenfeld [1] and Hieftje [2] and

coworkers. These cavities have been basically operated at low

pressures. Very significant improvements in design were made

by Beenakker [3] whose resonator was easily operated at

atmospheric pressure. It is illustrated in Figure 1-1. Further

modification on coupling mechanism was made by van Dalen [4].

The role of the cavity is to set up a standing wave within the

cavity structure and, therefore, to concentrate the microwave

field in the vicinity of the plasma. The cavities serve as

reservoirs of microwave energy. The most commonly employed

geometrical shape is the cylindrical cavity. Rectangular

resonators have also been used but do not compare favorably

with the cylindrical cavities because of their low Q factor.

The Q factor is the ratio of stored energy in the resonant

cavity to energy loss within it. In cylindrical cavities, the

height does not play a role in establishing a resonance, hence

it can be made as small as possible to achieve high power

density. However, good coupling is maintained when power is

raised either by increasing the operating frequency of the

microwave, which would require a costly tunable RF source, or

increasing the length of the cavity. The construction of

variable length cylindrical cavity to achieve tunability has

been discussed [5]. Rait et al. [6] obtained an optimal height

of 2 cm for iodine emission intensity in their 100 W, helium




























Figure 1-1. Top (a) and side (b) views of Beenakker cavity.
(1) Cylindrical wall (2) Fixed bottom wall (3)
Removable lid (4) Discharge tube (5) Discharge
tube holder (6) Coupling loop (7) Connector (8)
Vacuum sealing (9,10) Tuning screws (11) Holes for
air cooling (12) Mount. (Adapted from ref. 3)










































5cm


(a)







5

plasma. The maximum emission intensity may have been observed

at different height if the power was greater than 100 W since

tuning with height adjustments is dependent on power.

The critical dimension in cylindrical cavities is the

diameter, which is dictated by the microwave frequency of the

source. The modes of wave propagation in the cavities are

classified according to the orientation of the electric or the

magnetic field with respect to the axis of propagation as TE

(transverse electric) and TM (transverse magnetic); three

subscripts are adapted to denote the orders. For example,

subscripts m, n and p in TMp stand for the number of full-

wavelength patterns around the circumference, the number of

half-wavelength patterns across the radius, and the number of

half-wavelength patterns along the axis, respectively. The

dominant order is TMo01 and source frequency, f (Hz),

determines the diameter, d (cm), of the resonant cavity filled

with air [7]:


d = 2.405-



where c is the velocity of light (cm/s). The electric field

distribution and current patterns of this mode are illustrated

in Figure 1-2 [8].

In practice, the diameter should be less than the

theoretical value to take into account the presence of the

quartz discharge tube and the plasma itself, which could alter

the resonance frequency. When a dielectric material such as






























Figure 1-2. The electric (a) and current (b) patterns in TMo10
mode in cylindrical cavity. (Adapted from Ref. 8).









































(b)







8

quartz, is inserted in the cavity, the resonance frequency

shifts to a lower value. The magnitude of the shift depends

upon the volume of the dielectric and its dielectric constant.

A further frequency shift may occur when the plasma is ignited

and/or the sample aerosol introduced. Thus, it is essential to

construct a cavity with a smaller diameter than theory

predicts.

Bollo-kamara et al. established a relationship between

the frequency shift and the volume of quartz inserted [9]:
-x 4 -3
Af = -1.45x10 GHz mm for quartz.

Frequency tuning mainly is done with a slug and/or a screw

[3]. The screw method involves placing a screw into the top of

the cavity, which makes a small reduction in dielectric

thickness between the top and the bottom. This provokes an

increase of capacitance and hence a reduction of resonance

frequency. Note that frequency is inversely proportional to

the square root of the product of inductance and capacitance.

Frequency shifts due to small perturbations in volume of a

cavity operating at TM010 can be predicted if the disturbing

device is of the same material as the cavity [7]:

Af = -1.85frAV/V

Where fr is the resonance frequency, V and AV are the cavity

volume and the variations caused by the perturbant,

respectively. Slug tuning is accomplished by inserting a slug

into the side of the cavity. The slug causes a decrease in







9

effective inductance, resulting in an increase in resonant

frequency.

The microwave energy is transmitted from the generators

to the cavities with a coaxial cable of 50n impedance. It is,

therefore, necessary that the lines be terminated to a 50n

load, otherwise some of the energy may be radiated or

reflected back to the generator. This causes instability in

the power and may damage the power supply. Reflections occur

where there are impedance changes such as generator-line and

line-MIP interfaces. The process of multiple reflections at

the end of the lines gives rise to standing wave which is

dissipated in the form of heat or radiation. Hubert et al.

[10] recently reviewed the impedance changes and measurement

of power in MIPs. Impedance matching is performed with a stub

line arrangement. The stub which is a section of a short-

circuited transmission line is connected parallel to the

feeder line.

There are two principal methods of coupling a coaxial

line to a resonant cavity. The first involves inserting a

probe, which is an extension of the center conductor, into the

cavity, parallel to the electric field, that is to say, from

top or bottom. The coupling efficiency may be altered by

raising or lowering the probe into the cavity. The second

method utilizes a loop which is placed where the magnetic

field will be located, i.e from the side of the cylindrical

cavity. The degree of coupling can be altered by rotating the







10

loop relative to the direction of the magnetic lines. In

cylindrical resonant cavities, the dominant mode TMo10 is

generally excited with the coupling procedures mentioned

above. This mode is preferred because of its electric field

distribution which has a maximum at the center and a parabolic

decay of the field towards the walls. The microwave energy is

concentrated at the center of the cylinder. By inserting a

quartz tube and using the proper gas flow rate and power, a

breakdown of the gas results producing a plasma.

Microwave plasmas generated with resonant cavities have

been widely evaluated for spectrochemical analysis. The first

successful analytical application was reported by Broida and

Chapman [11]. These plasmas generally employ low power sources

(<200 W) and do not tolerate solvent from solution

nebulization. They have been very successful for gaseous

samples, such as GC effluent and chemically or thermally

generated vapors, particularly electrothermal vaporization.

However, Beenakker [3] was the first to introduce solutions

into an MIP. His plasma operated at 150 W. Solution

nebulization was also made possible in an even lower power

plasma (75 W) in highly efficient He-MIP [12]. Despite the

successes of solution nebulization, few studies have been

devoted to the evaluation and characterization of microwave

plasmas for solution analyses. Plasmas have also been

generated with air, N2, Ar, H2 and their mixtures [13].









Surface wave Launchers

Surface waves propagate along the interface of two media

without energy loss by radiation. The surface waves are

characterized by an exponential decay of the electric field in

the direction perpendicular to the boundary sustaining the

propagation. If the electric field is sufficiently intense, it

can break down gaseous media, thus producing plasma. Unlike

resonant cavities, the cut-off condition are independent of

the dimensions of the surface wave launcher (also called

surfatron). Surface waves can be expressed as combination of

TM and TE waves [14]. However, the mode that is appreciably

excited in surfatron is the azimuthally symmetric mode with

m = 0 (where m is the wavenumber corresponding to the

azimuthal angle 0) which is purely TM mode. The electric field

of the surfatron differs remarkably from that of resonant

cavity. In a cylindrical resonant cavity operating in TM010,

the electric field has a central maximum, whereas in

surfatron, the fields have a minimum along the central axis.

Excitation in surfatrons is relatively easy. However, to

launch only one particular mode is rather difficult. A general

rule is that the design of a wave launcher be as much similar

as possible to the azimuthal configuration of the mode to be

excited. There are two major parameters that characterize the

efficiency of the surfatron, namely coupling and launching

efficiencies. The coupling efficiency is related to the

impedance characteristics of the source, line and the load.







12

Proper impedance matching between the launcher and the

transmission line should improve it. The launching efficiency,

defined as the ratio of the power carried away by surface

waves to the total power emitted including radiation, depends

on the purity of the selected mode. Simultaneous excitation of

more than one mode would render the surfatron inefficient.

This was extensively discussed by Moisan et al. [14].

A schematic diagram of a surfatron is illustrated in

Figure 1-3 [15]. It consists of a coaxial structure which

shapes and orients the electric field and the coupler to match

the impedance to that of the generator for efficient power

transfer. The plasma contained in a glass tube is generated in

the coaxial field shaping structure, which is shorted at one

end and terminated by a capacitative gap at the other end. The

electric field is initially perpendicular to the axis of the

torch but changes direction as it exits through the gap and

becomes parallel to the tube. It has been demonstrated that

the best coupling is achieved when the coupler plate is made

as thin as possible and very close to the torch (few tenths of

mm) [15,16]. For a given diameter, the length of the plasma

increases linearly with power instead of inflating, which is

the case with the classical microwave plasmas. The performance

and design considerations were examined by Selby [15] and

Moisan [16] and coworkers.

Surface waves were also generated in waveguide surfatrons

which utilized a rectangular wave guide and a coaxial line






























Figure 1-3.


Schematic drawing of surfatron. The insert shows
important dimensions (in mm) and three tuning
adjustments, which are: vertical positioning of
the coupling plate, the gap length and the
internal chamber length.(adapted from Ref. 15).




















POWER
INPUT
I.


N-TYPE
CONNECTOR








COUPLER
ADJUSTMENT



\K


FUSED SILICA
TUBE


TUNING
ADJUSTMENTS


COUPLING
PLATE 3
2
I cm
0







15

elements [17]. The surfguide shares the excellent tuning

capabilities of surfatron and additionally has high power

handling capability as well as the ability to operate higher

frequencies than in surfatrons. The microwave power is

supplied from a generator to a rectangular waveguide section

terminated by a movable short-circuiting plunger. The coaxial

part is attached perpendicularly to the wide wall of the guide

and its inner conductor extends into the waveguide as a sleeve

around the discharge tube, forming a circular gap in the

immediate vicinity of the opposite wall. This system is

actually a waveguide-based counterpart of surfatron. No

evaluation of the waveguide surfatron for spectrochemical

analysis has been made.

Surface wave induced plasmas have found application in

laser systems, spectral lamps, small plasma-jet welding and

thin film coating devices [18,19]. No significant work on

characterization of these plasmas as an excitation source was

performed. The first analytical application of a surfatron

plasma to liquid solutions was reported by Abdallah et al.

[20]. Desolvation system was employed in their system to

remove 80% of solvent. No investigation, concerning solvent

introduction tolerance, was made. Plasmas doped with CO2 were

evaluated for supercritical fluid chromatography [21].

Capacitively Coupled Microwave Plasma

These plasmas require electrodes to transfer energy to

the plasma gas. A schematic diagram of the setup is shown in







16

Figure 1-4 [22]. The magnetron radiates microwaves through its

probe coupler into the rectangular waveguide short-circuited

at both ends. The interference between the forward and the

reflected waves from the opposite wall generate a standing

wave. The stored energy is then transferred to a coaxial

waveguide whose central conductor is the electrode and

eventually coupled to plasma gas. These guides operate in TDMp

and TE, modes discussed previously, where m, n and p

represent, in this case, the half-wavelength variations in the

wide and narrow dimensions, and the direction of propagation,

respectively. The dominant mode in rectangular guides is TE10p

where p is determined by the length of the guide and should be

a multiple of Ag/2. In a rectangular waveguide filled with

air, the guide wavelength, Ag, is related to the wavelength in

free space and is given by








where AX is the cut-off wavelength where radiation of higher

wavelengths will not be propagated without exciting higher

orders. The cut-off wavelength, Xc, equals twice the wide

dimension of the guide, and A is the operating wavelength.

Consequently, the apparent wavelength of an electromagnetic

wave of known frequency is greater when measured in a

waveguide than in free space. The maximum electric and minimum

magnetic fields occur at A\/4. The microwave energy is
































Figure 1-4. Schematic diagram of microwave single electrode
plasma torch assembly. (Adapted from Ref. 22).









18































Coaxol Wave Guide





Magneiron Observaion Winloe
(2450 MHz)

S---Electrode
Cooling Woter Tube


Bass Tube
Oulet

.Teron Tube
. Rectongulo Woveguide Siecon Tube






Cooling Woaer Tube

-- Sheaoh gos

-- Plasmo gos




Simple \pel .ilh Cxrer gas






19

inserted or extracted at those positions without drawing any

current. Therefore, the magnetron and the electrodes are

placed at A /4 from the ends.

Some loss of energy is associated with the finite

conductivity of metallic walls. The Q factor (defined as the

ratio of stored energy to loss energy) of an unloaded

rectangular resonator is proportional to the volume-to-surface

area ratio because the energy is stored volumetrically, while

energy loss is due to conduction of current on inner surface

walls. The inner walls are usually silver-plated to lower

surface resistivity. Surface irregularities must also be

reduced for uniform impedance. Irregularities cause changes in

capacitance and, hence, an overall change of the impedance

which results in a mismatch.

The slug and screw tuning methods discussed previously

are limited to very small changes in frequency; plungers are

useful when large changes are required. Plunger tuning is

frequently used in CMPs since the power is high and the

frequency detuning is large. Insertion of the plunger serves

to reduce the volume of the waveguide. Since the resonance

frequency is directly proportional to the dimensions of the

cavity, size reduction results in an increase of the

frequency.

Most of the CMP designs are derived from that of Cobine

and Wilbur [23]. The CMPs are operated at moderate and high

powers (200-1000 W) and, therefore, tolerate solvent







20

introduction. However, these plasmas do not compare favorably

with Inductively Coupled Plasmas (ICP). It has been shown that

CMPs suffer from matrix effects [24]. The electrodes used

initially in CMPs were constructed from aluminum [22]. Those

electrodes presented two major setbacks. First, the electrode

melting point was low, and relatively high plasma gas flow

rates had to be used to extrude the plasma from the electrode

tip. Second, mixing the aerosol with the plasma gas caused

dilution and drift of particles owing to the temperature

gradient and the resulting gas expansion. Better results were

obtained with a tantalum tubular electrode held in position by

an aluminum frame [25]. With this design, the sample aerosol

was mixed with the carrier gases and introduced into the

central core of the plasma for efficient excitation. Those

aluminum electrodes or supports were water-cooled. Hwang et

al. [26] reported a graphite tubular electrode which operated

similarly to that of tantalum electrode [25] discussed

earlier. The graphite electrode was supported by the central

quartz tube of the torch. The graphite electrode did not

require cooling since its melting point was much higher than

that of aluminum. With that electrode, they were able to

operate the plasma at powers up to 1 kW. Ali et al. [27]

adopted a microsampling technique for solution samples. They

used a graphite electrode and cup from which microvolume

samples were vaporized, atomized and excited in the plasma.

The system was also applied to solid sampling [28].













CHAPTER TWO

ANALYSIS OF SOLIDS IN GRAPHITE ELECTRODE
CAPACITIVELY COUPLED MICROWAVE PLASMA


Introduction

In atomic spectroscopy, direct analysis of solids is

important for several reasons. Dissolution of solids may

require the use of hazardous chemical reagents and, sometimes,

is time-consuming. Even though this problem may be eased by

the recently introduced microwave digestion, still

deterioration of sensitivity due to dilution of the analyte

and degradation of the plasma performance as excitation and

atomization source owing to the solvent introduced (with the

analyte) could occur. Contamination and losses may also occur

in the process of dissolution. Besides, the common methods of

liquid sample introduction, particularly by pneumatic

nebulization, is known to be inefficient (<10% sample

throughput). The resultant solution after sample dissolution

may be of high salt content, potentially clogging the

nebulizer. These problems are not encountered in direct solid

sample introduction approach. However, direct solid sampling

poses other problems. Poor precision results from sampling

errors which depend on the distribution pattern and

concentration of the analyte, particle and sample sizes [29].







22

Moreover, physical and chemical interference of matrices in

solid samples resulted in unreliable calibration curves using

aqueous standards. In some cases, matrix modification in the

condensed or gaseous phase may minimize chemical

interference. Application of chemical modification to solid

samples has been successfully done in direct sample insertion

inductively coupled plasma (DSI-ICP) [30-33], and graphite

furnace atomic absorption spectrometry (GF-AAS) [34].

Other methods such as sparks, arcs and laser ablation

have also been used for sample vaporization into plasmas

[35,36]. In order to minimize chemical interference, these

methods subject samples to harsh thermal conditions for

decomposition. In sparks, the sample must be electrically

conductive or must be made conductive by mixing it with

conductive matrix such as graphite. Laser ablation offers the

possibility of handling both conductive and non-conductive

samples. However, its use is limited by lack of precision and

poor linearity. Arcs do not erode the sample uniformly and,

therefore, suffer from poor reproducibility of sample amount

vaporized.

Attempts were made to nebulize solids suspended in

slurries into the plasmas [37,38], particularly the ICP in a

manner similar to that used in flame AAS. This method is

plagued by several problems. First, the suspensions should be

freshly prepared to avoid any aggregation of particles.

Second, even though a number of different nebulizers were







23

used, clogging and particle size limitations were still the

major problems. An ideal technique for direct solid sampling

should fulfill the following requirements [39]: the rate of

volatilization of samples should be high and independent of

the nature of the matrix; the analyte transport efficiency

must be high and memory effects minimal; the composition of

the volatilized material must be the same as that of the

sample; and sample material must have small particle size. The

methods of direct solid sample introduction into plasmas have

recently been reviewed [35,36].

Microwave induced plasmas (MIP), which are usually

operated at low powers (<200 W), have been mainly successful

for gaseous samples, particularly gas chromatographic

effluents [40,41]. Capacitively coupled microwave plasmas

(CMP), with which high power can be obtained, have the

capability to vaporize efficiently liquid aerosols and atomize

and excite analyte atoms [25,26,42,43]. Hence, most previous

research has been focused on the analysis of liquid samples.

For both types of microwave plasmas, very few studies have

involved solid sample introduction methods [22,44-51]. In most

cases, electrothermal vaporization devices have been used to

facilitate introduction of solid samples into the plasma

[22,44,46,50,52,53]. However, no work on direct solid sampling

methods without analyte vapor transport have been developed

for microwave plasmas.







24

In this chapter, we will describe a new method for rapid

analysis of solids which involves direct sample introduction.

This technique exploits the need in the CMP for an electrode

for plasma generation and the subsequent heating of the

electrode. This heating effect is used to advantage for

sampling. For this purpose, a graphite electrode with a cup

end was constructed into which solid powder was placed. The

heating of the electrode vaporized the sample into the plasma

for emission measurements. The rapid heating of the electrode

caused a rapid vaporization of the sample, thus producing a

high momentary concentration of the analyte. Coal fly ash (NBS

SRM 1633a) and tomato leaves (NBS SRM 1735) were chosen for

evaluation of the technique.

Experimental

Instruments

The experimental setup is shown in Figure 2-1 and the

components are listed in Table 2-1. The electrode/cup system

used is illustrated in Figure 2-2. The torch employed was

similar to the conventional ICP torch except that the central

tube was larger in diameter to accommodate the graphite

electrode. The cavity for the generation of microwave plasma

is described elsewhere (22,43,50).

Sample preparation

For most elements determined, the sample was used without

further treatment. For coal fly ash, the concentrations of Mg,

Ca and Zn were so high that dilution of the sample was
















-H

0
o




En
m



0


0






o0






0





0)
"-4-















0-)
r-1













0,







0)


00)
-4r-1


4J











-4H
-r1
a

































rz4
I-P
^ *

0 >
u -
c-




















01m




ELr

*i-i

rl
































. 0
g,
C4
U


'00k
-p4e


ft
< \
CQ
ze










Table 2-1. Instrumentation for the capacitively coupled
microwave plasma atomic emission spectrometry.

Instrument/components Manufacturer

Diode array
OSMA model IR4-1024) Princeton Instruments, Princeton, NJ.

Spectrometer Princeton Instruments, Princeton, NJ.
Jobin-Yvon HR1000,
1 m, 2400 grooves/mm
linear dispersion
0.5 nm/mm

Software (1-120) Princeton Instruments, Princeton, NJ.

OSMA detector Princeton instruments, Princeton, NJ.
controller

PC computer IBM.

HV DC power supply Hipotronics Inc., Brewster, NY.
model 805-1A
(max. power output
5 kW)

Magnetron Hitachi Ltd, Des Plaines, II.
model 2M131
(Freq 2.45 GHz, max.
power ouput 1.6 kW)

Electrode/cup system Laboratory made.
(Spex graphite rod;
grade HPND)

Torch
(3 concentric quartz Laboratory made.
tubes)































Figure 2-2. The graphite electrode/cup system for the solid
sampling. Dimensions are in mm.











5. O0
-> 4. 0 K-
-- 3 0K-


- g r aphi
1 cup


Sgraphi


el ect r ode


\1


te


te


3. 5







30

necessary to determine the detection limits; 1% coal fly ash

in spectroscopic graphite powder (Union Carbide Corporation,

New York, NY) was prepared for that purpose.

Procedure

Quantities (5-10 mg) of standard reference material (SRM)

of tomato leaves or coal fly ash were deposited in the cup.

The cup was placed on the top of the electrode and inserted

into the central tube of the torch. There was a tight fit

between the cup and the electrode for good thermal and

electrical contact. The plasma (intermediate) mixture gas flow

rates (1 L/min for N2 and 3 L/min for He) were then adjusted.

The coolant He gas flow rate was 6 L/min. The plasma was

initially ignited at a low power (about 100 W) for the ashing

step of the sample. Without ashing, the sample popped out of

the cup causing flares in the plasma during the atomization

step, thus causing plasma instability. During the ashing step,

no analyte emission was observed. After about 15 s, the power

was raised to a preselected value, 400 W for tomato leaves and

700 W for coal fly ash. Lower powers were used for liquid

samples which were employed solely for analyte emission line

identification. The observation height was 2 mm above the cup.

The emitted radiation was focused on the entrance slit of the

spectrometer by a two matched quartz lens (f=4") system. The

emission was monitored with the diode array. For better

resolution the 2nd-5th orders were used and a single scan (33

ms) spectrum was obtained for all the elements.








Results and Discussion

Before the solid was introduced, the graphite cup and the

electrode were checked for impurities of the elements of

interest by monitoring their emission lines at powers between

400 700 W. No measurable signal was observed for any of the

elements. Moreover, blanks were run before every measurement

to insure the absence of memories from previous runs and

atmospheric contaminants.

The detection limits of some elements in tomato leaves

and coal fly ash are listed in Table 2-2. The low detection

limits obtained for Mg, Ca and Zn in coal fly ash arose from

an increased vaporization rate of the analytes when the sample

was diluted with graphite powder. Dilution of the sample in

the graphite powder reduced the matrix background, prevented

formation of glassy globules in the atomization step and

improved vaporization and atomization efficiencies resulting

in higher signal to background ratio. Similar increases in

sensitivity were observed in ICP by Brenner and coworkers for

Cu and Zn in silicate materials when mixed with graphite [54].

The addition of graphite resulted in greater analyte

evolution, reducing environment and complete consumption of

solid samples [55]. A positive influence on the vaporization

has also been achieved by adding organic halides into the

plasma gas and/or matrix modifiers to minimize differential

volatilization of the sample [56,57]. The precision of the

measurement of the analytical signals were between 12 and 18











Table 2-2. Detection limits (LOD) of some elements in Coal Fly
Ash (SRM 1633a) and Tomato Leaves ( SRM 1573).

Tomato leaves (NBS SRM 1573)


Element A(nm)


228.8
324.8
258.6
403.1
285.2
283.3
780.0
460.7
213.9


LOD(ng)

0.3
8
134
69
1
2
4


Coal fly ash (NBS SRM 1633a)
Element A(nm) LOD


422.7
279.5
403.1
780.0
213.9
324.8
193.7
283.3


(ng)


46
80
0.1
22
51
133






33

per cent. Considering the small sample size used, this degree

of precision is quite acceptable.

The accuracy of analytical determinations using the

graphite cup CMP-AES is obviously matrix-dependent as can be

seen from the results in Table 2.2. Therefore, either matrix-

matched standards or standard additions must be used to

quantitate unknown samples. Certainly, standard addition is

easier and faster than the matrix-matched approach. No samples

were analyzed to determine accuracy of the method, although

the accuracy should be comparable to the precision as long as

the standard addition is used.

The addition of nitrogen to the plasma was essential

since molecular gases are generally known to increase gas

temperature. The capability of the cup to vaporize samples is

influenced by the plasma kinetic temperature which is

primarily responsible for heating the cup. The construction

material of the cup was very important at powers used to

operate the CMP. Graphite is a suitable electrode material

since its melting point is high and its spectral emission is

simple. In the UV-VIS, it has only two major emission lines

(193.1 and 247.9 nm). Hence, the spectral interference are

minimal. In this work, the size and geometry was not

optimized. In DSI-ICP, it was observed that the most intense

and temporally narrowest peaks were achieved with electrodes

of the smallest cup depth and of smallest size [58]. Other

workers [59] had obtained low detection limits with a pellet







34

(flat-top) cup. Therefore, adapting the cup designs for DSI-

ICP may have improved the sensitivity of our CMP.

The presence of elements were confirmed at multiple

wavelengths, by identifying them with hollow cathode lamps

(HCL). Solution nebulization and aqueous solution deposition

in the cup were also used where HCLs were not available or the

intensity was too low for definitive identification. In the

HCL, the nearby neon emission lines made identification

difficult while, with the deposition of liquid solutions,

there were analyte residuals requiring a cleaning step. As a

result, solution nebulization was frequently used for line

identification. A graphite electrode with central hollow

channel and ultrasonic nebulizer were used for this purpose.

This electrode was also constructed in our laboratory.

Figure 2-3 illustrates the emission signals for manganese

in coal fly ash and aqueous solution recorded in the third

order. In the elements investigated, higher power was needed

for coal fly ash than for tomato leaves to observe a

sufficient signal to noise ratio indicating that the former is

more resistant to thermal decomposition than the latter. It

should be mentioned here that the cup temperature, which is

primarily responsible for sample vaporization, increased with

power. The duration of the emission signals depended on the

power, the analyte and the matrix. Manganese, which is present

in similar concentrations in the two samples, gave an emission

signal in coal fly ash which lasted twice as long as that in
































Figure 2-3. Emission spectra of Mn in (a) NBS-SRM 1633a and
(b) aqueous solution.

























Cl)


C




















-4-J
c
(D
























C
0)











(-
(D




















-171
'-4-


Diode


r 0

I c^
c
_r







37

tomato leaves for the same amount of analyte with the same

operating conditions; thus requiring high operating power or

longer integration time. Both were disadvantageous because of

the nearby high intensity emission signals from the

concomitants which easily saturated the diode array. The

difference in duration of the manganese emissions in the two

samples may be a result of the different chemical forms in

which the analyte exists in the sample. The appearance time of

different elemental emissions from the same sample varied,

indicating the possibility of temporal discrimination of

spectrally interfering elements. Thus a selective

volatilization procedure could be employed for the elements of

interest.

Our system is superior to DSI-ICP in terms of cost and

operation. Unlike DSI-ICPs, the electrode in the CMP is held

at a fixed position by the central tube of the discharge torch

and can only be changed by displacing the torch or the

rectangular waveguide. Therefore, it is not prone to poor

reproducibility of positioning. In DSI-ICP, studies of

analytical signals as a function of cup position showed that

variation of 1 mm produced signal intensity changes as large

as 10% [60,61]. In DSI-ICP, drying is generally done with

auxiliary heating devices [62]. In our CMP, both drying and

ashing are done in situ. Compared to electrothermal

vaporization (ETV) sample introduction, our system performs

both vaporization and excitation. Furthermore, dilution and







38

analyte loss by plating-out during the sample vapor transport

inherent in ETV is non-existent in our system. Besides, its

sensitivity depends on the design of the ETV cell.












CHAPTER THREE

MICROSAMPLING OF LIQUIDS IN GRAPHITE ELECTRODE CMP

Introduction

Microsampling in plasmas for atomic emission spectroscopy

has become a popular technique in trace element analysis. The

inductively coupled plasma (ICP) is the most common excitation

source for atomic emission spectrometry. The modes of sample

introduction into the ICP have been summarized by Fassel [63]

and Browner and Boorn [64]. In cases when sample size is

limited for cost or availability reasons, electrothermal

vaporization (ETV) is the method of choice for introduction

into a plasma [65,66]. Direct insertion of microvolumes of

liquid samples in the ICP has been developed [62,67-69]. The

sample is deposited on a tungsten or tantalum loop or a

graphite cup, dried prior to vaporization by the plasma heat,

either internally by inductively heating it with the

radiofrequency energy inside the plasma torch [67] or

externally with IR radiation or a heat gun [70] or with an

electrothermal heating device [71]. The problem encountered

with this system is that the sample holder should be removed

each time the sample is deposited and the entrained air should

be purged before the plasma is reignited. The emission signals

are significantly dependent on the cup position. Variation of







40

1 mm of the cup position was found to incur errors as large as

10% [60,61].

A radiofrequency arc in which the plasma was brought to

the sample on an electrode has been described (72). This

method is based on the concept that the ICP is attracted to

electrically grounded electrode. The electrode is positioned

below the plasma inside the central tube of a modified torch.

The sample vapor generated by the arc is then transported into

the ICP. This device is rather complex and dangerous to use.

Integrated vaporization devices for microsampling and

emission sources have also been used; these include furnace

atomic non-thermal spectroscopy (FANES) [73], and furnace

atomic plasma excitation spectroscopy (FAPES) [74]. Unlike

graphite furnace ETV, the furnace in these two techniques is

an integral part of the excitation source. Although FANES has

been shown to be a powerful analytical tool, it is operated at

low pressures making sample introduction difficult, if not

impossible, without breaking vacuum. FAPES, operated at

atmospheric pressure, is in an embryonic stage and extensive

work is required to explore its potential for spectrochemical

analysis.

The microwave plasma is another important excitation

source. Because of the low powers used in microwave induced

plasma (MIP), the plasma is easily disturbed by the sample

solvent. Therefore, it is imperative that the liquid sample be

dried before introduction into the plasma. ETV has been proven







41

to be successful in this regard; this subject has been

recently reviewed by Matusiewicz [75]. The design of the

vaporization chamber can have large effects on the sensitivity

and reproducibilty of the emission.

Hieftje et al. [76] developed a computer-controlled

micro-arc consisting of a high-voltage, low-current pulsating

DC arc to vaporize efficiently microvolumes of liquid samples

from a tungsten wire. Other methods include the nebulization

of small sample volumes directly with a conventional pneumatic

nebulizer [77-79]. Typically 25-100 gL solution volumes are

introduced from micropipettes or syringes into the sample

uptake tube of the nebulizer.

Unlike the MIP, very few researchers have attempted

microsampling of liquids into capacitively coupled microwave

plasma (CMPs). Hanamura et al. [22], used an RF inductively

heated graphite cup for volatilizing solution samples into a

microwave plasma for diagnostic studies.

ETV sample introduction devices require connection tubing

for sample vapor confinement and, therefore, suffer from

sample loss, dilution by the carrier gas and shorter residence

time of excited atoms for probing. The current work involves

development of a system for liquid microsample introduction

without the carrier gas and the extra cost of an

electrothermal vaporization device. The system takes advantage

of heating the electrode in the capacitively coupled microwave

plasma. A graphite electrode and a graphite cup, as a sample







42

holder, were constructed. Ignition of the plasma resulted in

sample vaporization into the plasma and analyte excitation.

Experimental

Optimization of the Electrode

Two aluminum waveguides with cross sections of 109x65 mm2

and 86x43 mm2 were used for this purpose. The former operates

at frequency range of 1.72-2.61 GHz and the latter at 2.17-

3.30 GHz. The guides were constructed in our laboratory. The

experimental setup is shown in Figure 3-1. The microwave

radiation emitted through the antenna of the magnetron

propagates forward along the rectangular waveguide and is

reflected from the opposite wall to generate a standing wave.

A resonator cavity is formed which serves as reservoir of

energy. It is known that any means of inserting energy into a

cavity also serves as a means of extracting it. A probe was

constructed by silver-soldering a 2 mm diameter copper

filament on a panel receptacle (UG-58A/U, Newark). The probe

was placed at \A/4. The microwave is then transmitted in a

coaxial cable (RG-214/U, Newark) and terminated into a 50Q oil

resistor. The forward and the reflected power was measured

simultaneously with a microwave power-meter (Opthos

Instruments Inc., Rockville, MD).

Apparatus. The instruments and optics used in this work

are the same as those employed in the previous experiment. A

high voltage power supply feeds a cw magnetron operating at a

frequency of 2.45 GHz. Microwave energy propagates in the






























-p
,C

a)

r-c







Q)

a)
0







0


r-
(U


N
*-
4-)





0

4a





0
*H





0








r4

4->



-1




I
3(1












a,)




en



a\
CP
*F-
































SC






00 .



OC

aEr oo






[ *







45

rectangular waveguide described previously [22,25,26] and

couples to the coaxial waveguide of which the graphite

electrode served as the central conductor. The electrode and

the sample cup were constructed in our laboratory and their

dimensions are detailed in Figure 3-2. The graphite cup which

serves as sample containment has a capacity of 20 pL.

The plasma torch consists of two concentric quartz tubes

and is similar in design to that of Hwang et al. [26]. A

mixture of helium and nitrogen is used as the plasma gas. The

emission from the CMP is collected with a 4" focal length

quartz lens. The emission radiation passes through an aperture

of 1/2" in diameter to match the f-number of the spectrometer

to that of the optical system so as to minimize stray light.

Another quartz lens (f=4") focuses a 1:1 image onto the

entrance slit of the spectrometer. The spectrometer slit

widths used are between 5 and 10 pm. The radiation at the exit

plane is detected with an intensified linear diode array. The

data is acquired with a PC computer and is processed by the

software.

Reagents. All the solutions were prepared by successive

dilution of 1000 Ag/mL standard stock solutions (Inorganic

Ventures Inc, Toms River, NJ).

Procedure. The flow rates of helium and nitrogen, which

make the plasma gas, are adjusted to optimized settings of

4 L/min and 1 L/min, respectively. Since the stability of the

nitrogen flow rate was found to affect the analytical signal




























*-'-







0


c








0

4(J
E





















4-I

0
-H















--I
.l4
I:



c,
















Q)














CP
*Hl








6 O .-
1K-4 0
K- 3.


Graphi t e
-cup








Graphi t e
- electrode


3. 5


\%







48

significantly, two valves in tandem were used for each

flowmeter to minimize day-to-day flow rate adjustment errors.

Five iL volumes of aqueous samples were deposited into the

graphite cup. The power was adjusted to 60 W to capacitively

heat the cup. A drying period of 60 s was allowed. The power

was then raised rapidly to 1.0 kW while, at the same time,

starting the computer to acquire data. The plasma auto-ignited

at about 80 W. The emission signal was integrated for 1.65 s.

Five such spectra (1.65 s each) were recorded and generally

the spectrum with the best signal-to-noise ratio was used for

analytical purposes. In the cases where the signal was shared

by more than one recorded spectrum, the emission signals in

individual spectrum were added. The noise was obtained from

five measurements of the blank. The observation height for

analytical signals was 5 mm above the cup. It was optimized

for good signal to noise ratio and signal to background ratio.

At lower observation heights the continuum emission from the

glowing was enormous.

Temperature measurements

The temperature of the graphite cup was measured with an

optical pyrometer (Pyrometer Instrument Co., Bergenfield, NJ).

The spectroscopic excitation temperature of the plasma was

determined using the two-line ratio method [80]. Cu was the

thermometric species and the lines used were 510.6 nm and

528.1 nm.









Results and Discussion

Electrode Length Optimization

The conducting copper probe (Figure 3-1), positioned in

place of the electrode, was trimmed from 100% penetration

depth (corresponding to full depth) to 36% and 42% for the

large and small waveguides, respectively. The applied power

was always maintained at 100 W. The reflected and the forward

powers were recorded at various lengths of the copper wire.

Figure 3-3 shows the dependence of the reflection factor on

the reduced length (1/A)of the probe, 1 being the length of

the probe, A the wavelength of microwave radiation in free

space and the reflection factor defined as the ratio of the

reflected to the forward power. Similar work was carried out

on the efficiency conversion of DC power to microwave energy

(Figure 3-4). In both guides, we found that the optimum length

of the probe was A/4. As a result, we designed a new electrode

(Figure 3-2) which was a little longer than the one previously

described [27]. This electrode increased the stability of the

plasma, reduced its acoustic noise, and correspondingly

enhanced its excitation capability.

Electrode/cup Design

After the liquid sample was vaporized from the cup, a

memory effect was observed for each of the elements. The

memory was attributed to the porosity of the cup; aqueous

samples seeped into the graphite structure. This problem was

overcome by coating the cup with tantalum carbide in the












-'l




0
0


4Q

4P
0






a-










,4





U
s-I
pt







a)







0

0

4-4,

r.-I
0
3


















4-,
a)
r-P













0




0
a4












a4

0

c*

--






fL4
M


tPl






QJ

n,



^i






-Hl





































6O 6 6
OJ- d Ui d I
jopDj^ uo!ipalQQl













4J
>-1














4
-4
C
Cr


















0
-4

4Q)








0








>1


Q)
C
o



























*4
-r-1










c4-
0










-1





00
ci)












40
X4




00
*HQ

MO
c()





















*





1:
*i-la













CD
r-0




--N
-d

-V)









-0
cx
-0






-0
e<





d




to
-0




N

d



-0
~-o
I IIIII IIII III II II I i i I III III II II II I I II I I I II IIIIII I I I II II I II I 0
00'L 06"0 0890 0/O'0 09'0 09c0 OVO 0'O
UOISJaAUOD "laJI







54

procedure described by Zatka [81]. Comparison was made between

the emission signals of 5 jL of 5 ppm Cu solution obtained

with and without a tantalum carbide coating. Higher emission

signals, observed with the tantalum carbide coated cup,

clearly demonstrated a higher analyte vaporization rate by

comparison with the uncoated cup. Even though major emission

lines of Ta were observed throughout the use of the cup, no

deterioration of the inside surface of the cup was noticed as

evidenced by the consistency of Cu emission line intensities.

The cup had to be changed after 30-40 firings because of the

plasma etching of the graphite cup rim. The nitrogen gas has

a greater etching effect on the graphite cup than the helium

gas. The lifetime of the cup shortens as more nitrogen was

added to the plasma gas. Lighter dopant gases may have less

etching effect. However, the products of the etching and the

Ta emission lines did not interfere with the analytical

signals.

Cup temperature

Upon ignition of the plasma, the cup was rapidly heated.

However, the temperature was not uniform throughout the

electrode/cup system. The cup, which glowed white, was hotter

than the electrode which did not glow. The cup temperature

depended on the applied power as well as the plasma gas

composition. Figure 3-5 shows the temperature dependence of

the cup on the power at three different gas compositions (0.0,

0.1 and 0.2 mole fraction of N2). The temperature was the









Q)

0




4 0





-H
C





0 4









-H
o






U)

S0











41.
O
O


























4-0
O) >
a)o








f 0






4





-C,-
Oa



0 4J
U)






4J 0
(a








U) 4
C10




-0
4Q)-






Q) a
Cb>
()0D






d>
^1
Q,
&*
-H














































0 0 0 0 0
0 0 0 0 0
0 r cOC


(0o) eiJneeJedcuw dnD







57

average of three measurements. The total gas flow rate was

maintained at 5 L/min in all cases. The dependence of cup

temperature on power is linear, which was similar to that

observed by Alandari et al. [82] for the surface temperature

of a composite material immersed in a low pressure He, Ar and

02 plasmas. Plotting the slopes of the curves in Figure 3-5

versus plasma gas composition also showed a linear dependence.

Therefore, it was concluded that the temperature of the cup

was directly proportional to the power and the mole fraction

of nitrogen in the plasma gas, at least within the ranges

studied. It may be speculated that the plasma thermalizes at

high powers especially with increasing molecular gas.

Molecular species absorb a large amount of energy in their

various degrees of freedom before they undergo ionization

and/or dissociation [83] and therefore raise the kinetic

temperature of the plasma. A desired temperature can be

achieved by the proper combination of power and mole fraction

of the dopant gas. The temperatures, reached with applied

power and plasma gas compositions, are high enough to vaporize

a large number of elements and/or compounds. This

automatically eliminated the need for an external vaporization

device.

Plasma temperature

It was impossible to measure the excitation temperatures

from helium emission lines because nitrogen quenches those

lines. Therefore, copper was introduced as a thermometric







58

species to study the relationship between the excitation

temperature and the power. An uncoated cup was used in this

experiment. As shown in Figure 3-6, the excitation temperature

decreased with increasing power, contrary to the trend that is

generally observed for single gas MIPs [84]. This strange

relationship may be attributed, in part, to the great

difference of vaporization rate of the Cu into the plasma

between low and high powers, and/or change in the

characteristics of the plasma. Murayama [85] also found that

the excitation temperature determined from the slope method

using hydrogen Balmer series increased with power while that

obtained from absolute intensity measurement of the argon line

at 415.9 nm decreased with power. The discrepancy decreased

with power. Murayama [85] concluded that the results indicated

that the plasma approaches thermal equilibrium with rising

power. In previous works in which temperature vs power studies

were made, single monatomic gases were used. For binary gases

(atomic and molecular), the molecular gases seemed to dictate

the asymptotic temperature. Kirsch et al. [86] calculated the

excitation temperatures of 7800 K for He and 4970 K for N2 at

operating power of 480 W. In fact, the temperatures of our

He/N2 plasma approximate those of He at low power and those of

N2 at high power. At both observation heights, the temperature

plateaus between 700 and 800 W. Such a large fall of

excitation temperature between 450 and 750 W was not observed

previously. Shimizu et al. [87] also found a decrease of












AUl


Wed
WO
0 4


X) (a
4C

Q) to





Q)ed


4
C






0m




















04
ul
































4J -
I
2rod




















-
Q)



















-H
En












a>

4-)

0 c
-,4 Q) 0











Z)



Pnl


























































0
0
LO


0
0
C\J
1-


(>1) eJn.edwua: uoie1!ox3


I I I


. I I







61

excitation temperature in CMP with power even though

analytical emission signals increased. They used solution

nebulization followed by desolvation. Since excitation

temperatures are higher at low powers, it may be useful to

operate microwave plasmas at low power for high energy level

transitions. This applies only to samples containing no

solvent, such as gases; whereas in solution nebulization high

power is required for better tolerance of the plasma to

solvent loading.

The excitation temperature of this plasma at an

observation height of 5 mm and 1.0 kW power, was 6100K. The

comparison of this temperature to those of mixed gas ICP was

difficult because of the various conditions and plasma gases

in which the plasmas were operated. Nevertheless, the

temperature in the analytical zone was similar to that of Ar-

ICP with molecular gas in outer gas flow [88] and higher than

that of Ar-ICP with molecular injector gas [89] at similar

powers.

Analytical results

Neutral density filters were used to extend the linearity

of response of the diode array. The filters were calibrated

for the wavelengths of interest before use. The log-log

calibration curves of Li, Mn and Cu had slopes of 0.997, 1.106

and 0.971, respectively. Table 3-1 lists the detection limits

based on three times the standard deviations of the blank, the

linear dynamic ranges and the relative standard deviations











Table 3-1. Analytical figures of
AES system.


Element

Ag

Ba

Cd

Cu

Ga

Ge

In

Li

Mg

Mn

Rb

Zn


A(nm)

328.1

553.5

228.8

324.8

294.4

265.1

303.9

670.8

279.5

257.6

780.1

213.9


LOD(pg)

210

50

30

15

45

100

65

10

15

85

30

65


%RSD

12

12

8

8

11

12

7

10

7

10

12

9


merit for graphite cup CMP-



LDR

3.1

3.7

>3.3

4.3

>3.5

>3.0

>3.2

4.0

3.8

3.1

>3.5

>3.7







63

obtained with our CMP system. The ">" indicates that the

linear range was greater than the value listed since the upper

limit was not determined. The linear range is 3 to 4 orders of

magnitude. A typical vaporization time at low concentrations

is shown in Figure 3-7. The non-linearity at high

concentration was partially due to incomplete vaporization

during the measurement time. This problem could be minimized

by increasing the power or the mole fraction of nitrogen, but

this had an adverse effect on the cup lifetime. Faster

vaporization for more concentrated solutions could be achieved

with smaller cup sizes.

Table 3-2 compares our LODs with other emission

techniques; it should be stressed that the LODs were obtained

under compromise conditions. The LODs with the graphite cup

CMP are comparable and in some cases better than those

obtained by graphite cup/furnace DSI-ICP [68,57], ETV-MIP [90-

94], ETV-ICP [95-98]. Compared to GF-AAS [99], our LODs are

better for spectral lines in the visible region; the converse

is true for spectral lines in the UV.

The relative standard deviation, based on 3 to 5

measurements, was better than 12 %. Typical RSD for GF-AAS is

5-10%. Overall RSDs of 1-20% and 1-15% have been reported for

ETV-MIP [75] and ETV-ICP [65], respectively. The imprecision

in the current work was mainly a result of poor

reproducibility of power stepping which was done manually.

This was confirmed by integrating a He emission line



























0
0
4J



0



0
0

,--








,--I




0
O


















,-1



0





0
C




Or-













4-1
,--













r1l

r1-

















1^-








k& .fC)


um u 9 J J




mu 6 'OL





UIc T
UI U T C Q P ,


X. IsU1 a Ju I











Table 3-2. Comparison of detection limits in CMP-AES, DSI-ICP,
ETV-MIP, ETV -ICP and GF-AAS.


This Work#


ETV-MIP

(cal


DSI-ICP

(Pg)

165[68]



13[57]

140[57]



705[68]

3750[68]

90[68]

150[57]

43[57]


ETV-ICP GFAAS[99]

(pbb) (ppb)

300[95] 6

700[95] 240

500[95] 2

200[95] 45

10[96]

60[97]

20[98] 240

6[97] 45

1[97] 1

20[95] 10


element

Ag

Ba

Cd

Cu

Ga

Ge

In

Li

Mg

Mn

Rb

Zn


A(nm)

328.1

553.5

228.8

324.8

294.4

265.1

303.9

670.8

279.5

257.6

324.8

213.9


(pg)

210

50

30

15

45

100

65

10

15

85

30

65


# 5 jL volume samples
* Graphite sample holder
ETV-MIP: electrothermal vaporization microwave induced plasma.
DSI-ICP: direct sample insertion inductively coupled plasma.
ETV-ICP: electrothermal vaporization inductively coupled
plasma
GFAAS: graphite furnace atomic absorption spectrometry.


100[97]


2000[90]

150[91]

250[91]

3[92]

5000[93]

5[92]

1[94]

180[91]

200[91]


12[57] 500[91]






67

(A=512.8 nm) over the same period of time as the analytical

signals starting from initiation of the plasma. The

intensities of He showed an RSD of 6-10%. Automatic power

change to pre-set levels will improve the precision of the

technique.

Conclusion

The operation of this capacitively heated graphite

furnace microwave plasma is similar to that of the popular

electrically heated graphite furnace for atomic absorption

spectrometry (GF-AAS). Like the graphite furnace, our CMP

provides drying, ashing and atomization stages for micro-

volume samples. In the microwave plasma, the samples should be

atomized completely and, therefore, matrix chemical effects

should be as low as those in GF-AAS. Furthermore, the GF-CMP

is used in the atomic emission mode, providing multi-element

detection capability; whereas, the GFAAS is generally a

single-element technique. Our CMP has some similarities to

FANES and FAPES in terms of operation and construction. In all

three techniques, emission measurements are made on initiated

plasmas and the sample holders are indispensable parts of the

excitation system. Further optimization of the design and

operation of the present system will certainly reduce the

present level of detection limits.

Our system lacks the problems of dead volume, memory

effects and plating out of the analyte on cold surfaces of

transport tubes common in ETV-ICP and ETV-MIP.














CHAPTER FOUR

DIAGNOSTICS IN TUNGSTEN FILAMENT ELECTRODE
CAPACITIVELY COUPLED MICROWAVE PLASMA

Introduction

To understand the physical characteristics of the microwave

plasma, certain plasma parameters such as temperatures,

electron density and other quantities like the current density

must be measured. If the state of the plasma is known, all

parameters of interest can be derived. Methods which give

information on the state or parameters of the plasma are known

as diagnostic techniques. Spectroscopic methods are often the

preferred diagnostical tools. There are two main reasons for

that. First, spectroscopic methods are non-intrusive and hence

pose no disturbing effects on the measurement, as probe

methods do. Second, the information content of the spectrum is

very large.

There have not been as extensive diagnostic work in

microwave plasmas (MP) as in inductively coupled plasmas

(ICP); the reasons being that MPs are not as powerful

excitation sources as ICPs. Rotational temperature

measurements in plasmas have been made using the (0,0) band of

the first negative system, BZ2u X2Z of N2 [89], (0,0) band







69

of OH, A2+ X2H [100]; (0,0) band P-branch swan system of C2,

A3nH X3Hg [101]; and the R4 CN band (A= 421.6-419.7 nm) [102].

Several authors have examined the rotational temperature

of low and moderate power microwave plasmas using exclusively

N2 and OH. However, there are major discrepancies between the

temperatures determined; differences are sometimes greater

than 5-fold [103]. Fallgator et al. [104] measured

spectroscopic temperatures and electron number densities in an

Ar-MIP with and without nebulization of water. They found that

the introduction of water into the plasma resulted in an

increase of electron number density (ne) due to the ionization

of water molecules. Excitation temperatures of low pressure of

MIPs were also studied. Seravallo et al. [105] obtained higher

excitation temperatures with Cr than with He. Busch and

Vickers [106] performed a rigorous diagnostic work in He-MIP

and Ar-MIPs. They reported small changes in spectroscopic

temperatures with pressure and incident power. The electron

density remained unchanged with pressure and was slightly

affected by changes in power. Goode et al. [107] investigated

the influence of pressure on the properties of MIPs.

Experiments were done in Ar and He-MIPs. They found that the

rotational and excitation temperatures were essentially

constant with power and gas flow rates. The electron density

was modestly influenced by the applied power. Tanabe et al.

[84] observed an increase of electron density and excitation

temperature as the power was raised. The electron density







70

increase was steeper in smaller diameter discharge torches.

Besner et al.[108] studied the dependence of the plasma

parameters on the microwave and radio-frequencies. They

concluded that fundamental properties such as excitation and

rotational temperatures, and electron densities are

practically independent of frequency.

In this chapter, several characteristics of He-CMP with

tungsten filament electrode are evaluated and discussed. These

characteristics include the dependence of Cu and He emission

intensities, rotational and excitation temperatures on power,

flow rate, radial and axial distances.

Theoretical consideration

Rotational Temperature Measurements

The rotational temperature was determined from the (0,0)

first negative band of N2 The rotational temperatures are

related to the experimental line intensities assuming the

existence of Boltzmann distribution of rotational levels. For

the R branch, we have,

In aI/(K"+l) = -Bhc/kT(K"+l) (K"+2)

where K is the rotational quantum number, a=l for lines with

even quantum numbers and a=2 for odd-numbered lines owing to

the odd-even intensity alternation, k (J.K ) is Boltzmann

constant, h (J.s) Planck's constant, c (cm.s ) velocity of

light, I emission intensity of the molecular bands, T (K )

temperature and B dimensionlesss) the rotational constant of







71

the upper vibrational level. The temperature is determined

from the value of the slope, Bhc/kT = 2.983.

For OH, the structure is more complex. Five main branches

(O, P, Q, R, S) with a total of 12 branches are observed for

the A2 U-X2H transition.

Excitation Temperature Measurements

In case of an injected thermometric species, certain

requirements should be met; namely wide energy range, good

knowledge of transition probabilities or oscillator strengths,

and closely spaced wavelengths to avoid calibration of the

detector for wavelength responsivity. Ti and Fe are widely

used since they fulfill the above requirements.

Assuming that population densities have a Boltzmann

distribution, the excitation temperature, T (K) may be

obtained from the relationship,

log IA /gf E Ex

where I is the intensity of the emission line with wavelength,

A, oscillator strength f, statistical weight g, and excitation

energy Eex. The slope is related to the temperature and is

given by -0.625/T when Eex is in cm ,; and -5040/T when Eex is

in eV.

Electron Number Density Measurements

The method used for electron density measurements was

based on the Stark broadening of Hg (A=486.1 nm). Stark

broadening on this line is much larger than all of the other

line broadening effects. The Stark full-width at half-maximum







72

(FWHM), AA, and the electron density (ne) are related by the

following equation:

ne = 7.97x1012 (A/a/2)3/2 cm-3

where a1/2 is the semi-half-width of the reduced Stark profile.

Experimental

The instrumentation and the optical layout used in this

experiment are the same as in previous works. A tungsten

filament electrode was utilized in this work instead of

graphite electrode and its dimensions are detailed in Figure

4-1. Its length was the same as that of the graphite electrode

in Chapter 3 which was found to be optimal. The top of the

filament electrode is coil-shaped to hold samples principally

by adhesion. The shoulders prevented the electrode from going

down the quartz tube. We noted that a small plasma formed on

the shoulders if their curvature is sharp, which wasted energy

and consequently reduced the efficiency of plasma to heat the

electrode. The plasma gas consisted of He gas only at a flow

rate of 4 L/min.

Procedure

Two microliters of Cu or Fe solutions of 0.10 and 1 ppm

concentrations, respectively, were deposited on the loop with

a micropipette (Eppendorf, mod. 4700). The power was set to 40

W and the sample dried capacitively for 120 s. The power was

then raised to 135 W. Ignition occurred at 50 W. The sample

was vaporized, atomized and excited. Wavelengths 501.6 nm and

324.8 nm lines were monitored for He and Cu, respectively.




















-H







0
-U
a,









0
-r1



Q






Ia
O







4-)
a)




r-










4-i

44



ro






-H




U
u





-4
3a
-l4
-H
IM
.-t

-pl




















-i-l






74

3.









no







;j ;J
-4-..










C \/
4"~ e ^ \-
M^~ = =1\







75

In this work, Fe was chosen as the thermometric species.

The mean gf-values of four different groups [109] shown in

Table 4-1 were used. The emissions of Fe and Cu were

integrated for a period of 3 s, sufficient time for the sample

to be vaporized completely. The emission spectrum of Fe is

shown in Figure 4-2. The emission signals of N2 were

integrated for 0.5 s, and 60 of such spectra were averaged. He

emissions were very intense, so no integration was necessary

but 10 scans were averaged. N2 was used for rotational

temperature measurements because of its better signal-to-noise

ratio than OH emission lines in our plasma. The emission

spectrum of N2 is shown in Figure 4-3; the wavelength and

rotational quantum numbers are listed in Table 4-2. The

assignment of wavelengths is based on the work of Childs

[110]. The spectra of Fe and nitrogen were both recorded in

second and third orders, respectively.

For electron density measurements, the semi-half-width of

the reduced Stark profile were those of Griem [111].

Results and Discussion

Precision

Measurements of relative excitation and rotational

temperatures were generally precise to less than 10%. There

were only three points that were slightly worse than 10%. The

major source of uncertainty lies in gf values. For He emission

measurements (A=501.6nm), the standard deviation was small

because of the averaging. A typical plot for determination of











Table 4-1. Log of gf-values of Fe.


A(nm) log(gf) log(gf) log(gf) log(gf)


371.99 -0.43 -0.43 -0.43 -0.43

373.49 0.3 0.31 0.31

373.71 -0.57 -0.57 -0.58 -0.58

374.83 -0.98 -1.01 -1.0

374.95 0.18 0.17 0.16

375.82 0.00 0.00 -0.03

376.38 -0.18 -0.19 -0.24

376.72 -0.35 -0.34 -0.39


* g is the statistical weight of the upper
emission oscillator strength.


level and f the












































































-.H




a,
Lo



cJ


a,



i4
4-
fr
0s
*f-
in







78

















(-Z



-
-








3-d









Af!sua9ul 9A!1DI|d





































































+
N


4-i
0





cn
M


as



0







*r
-4
e]


nr
E^
0)




cu
tP

*iL(








80







aD












-0-
00









-0C






C


0)



00




0-
9f'9SUGU aH G 7 i0











Table 4-2. The bands used for rotational temperature
measurements and their rotational quantum number.


x(nm) Quantum number


390.49 6

390.30 8

390.19 9

390.08 10

389.97 11

389.85 12

389.60 14

389.47 15

389.33 16

389.19 17







82

excitation and rotational temperatures are illustrated in

Figure 4-4 and Figure 4-5, respectively. The measurement of

electron density was more precise. The reproducibility of FWHM

hydrogen line was better than 5%.

Effect of flow rate

The influence of the He plasma gas flow rate on Cu and He

emissions, rotational and excitation temperatures was studied.

The excitation temperature, shown in Figure 4-6, decreased

slightly with increasing flow rate. The largest change

occurred between 3 and 4 L/min. Goode et al. [108] observed a

similar trend of plasma temperature with flow rate in

atmospheric pressure Ar and He-MIPs. It has also been reported

that the dependence of excitation temperature on flow rate in

three different torch diameters was insignificant [84]. The

findings of Uchida and coworkers [112] in their CMP were much

different from the above trends. They observed a linear

decrease of the temperature with carrier gas flow rate.

The rotational temperatures were practically constant

over the range of flow rates investigated (see Figure 4-7).

However, the emission intensity of nitrogen decreased with

increasing flow rate. The decrease of intensity can be

explained on the basis of short residence time which does not

allow enough time for the equilibration of excitation

processes of the entrained nitrogen to occur or may result

from a decrease of mole fraction of entrained nitrogen. Our

results disagree with the radial temperature measurements of


































UJ







41















r.
U)






0

-1
4)
























0
44























4J
0
in



























0



0








4)
0)























4-d
0x
-H
4-)
(0



-P(

-'-I








cP









Cp













-c(













+ 0
(0

+ O
F-



C7
(T)




E



CC)
O

C\j


+ -
m+ C)O


CC
1 0



O ( 0 cu
(j b/I O
(~/ n ^





































0


En





r-
Q)












(a
*4









P-
41
co
--i


0





4-
0
4-,


r-
0
-I











-I
P4






4-

1-1

4












0
0







0
0






+

-0
++


+ 0




-0
+
+ O











4 0
-'( +)IdU d d

( l-+.'>l/. l]U-I



































a,
0)
4-1
0









x
0






U)
0







4-
a
c4
o
















4-4









U,

-4
0E



4)








88









-O








E

LO
H-



...
'4-

I













0 0 0 0 0 0 0 0
0 0 0 0 0 0 0
0 0 0 0 0 0 0
O O O O O O O( -


(>0) ajnlejeduJeI uo!0e1.!oxl
































a)


(0

a)


a



+J
r4-
(0





0
*-H

+a









4-)
0









4-4
(E


0





I)
r-4

t-I


4-
0


(1)
4-4
4-1


a)

E-4








-r





















-







\(0 -
I--






(I
(0


(-5
H-
(/)














0 0 0 0 0
0 0 0 0
0 0 0 0
Iqt 03 04 ,-


(>1) eJnljeeduwai


leuo!Ie.OHl







91

Workman et al. [113] who measured a 2.5-fold rise in

rotational temperature when the carrier gas flow rate was

changed from 120 to 20 mL/min.

He and Cu emissions were not affected by changes of the

flow rate as shown in Figure 4-8 and Figure 4-9, respectively.

Our results disagree with other works. Boudreau et al. [114]

reported a tremendous decrease of N(I) emissions with plasma

gas flow rate in an open configuration surfatron. Workman et

al. [113] noted that the He excited state intensity

measurements were highly flow-dependent.

Effect of power

Figure 4-10 shows that rotational temperature increased

slightly with applied power. These results confirm previous

observations with the graphite electrode and can be explained

as follows; molecules absorb energy in a quantity dependent on

applied power until they are ionized or dissociated. This

effect aids the thermalization of the plasma. Our results are

in agreement with those of Workman et al. [113] but contradict

those of Goode et al. [108] who reported a constant rotational

temperature over the range 25-120 W.

Contrary to rotational temperatures, the excitation

temperatures remained essentially constant with increasing

power, as illustrated in Figure 4-11. This is not in agreement

with the temperature measurements made with graphite

electrode, and depicts a power dependence much different from

that of Murayama [85], who reported a decrease of excitation























Q)

(0
.4


0
r-4
4-1







(0



0
r1
t4\








r-












0

Ut
-H


Ir-
0









a)
-04
U













a)
H
















3)















-H1
-rl



U,

C-
Q)


Q)
at
0)













U,)





00











_J
CI


(d
L1
\ 'O 0
O
LL















) 0 9 O( Y O

AIsuG U i OAileAIG























a)
4J
(0


0
r--
4-4








,--I

0


-H
(0










C-)



0

-U
0
1-,-






a
c











E-1




r)










0)
c






























aI-
-i
c14
CI








a
u,