A cylindrical helium capacitively coupled microwave plasma

MISSING IMAGE

Material Information

Title:
A cylindrical helium capacitively coupled microwave plasma diagnostics and determination of silicon
Physical Description:
viii, 184 leaves : ill., photos ; 29 cm.
Language:
English
Creator:
Spencer, Billy Mac, 1961-
Publication Date:

Subjects

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

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1993.
Bibliography:
Includes bibliographical references (leaves 170-183).
General Note:
Typescript.
General Note:
Vita.
Statement of Responsibility:
by Billy Mac Spencer II.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001950909
notis - AKC7451
oclc - 31200857
System ID:
AA00003260:00001

Full Text









A CYLINDRICAL HELIUM CAPACITIVELY COUPLED
MICROWAVE PLASMA: DIAGNOSTICS AND
DETERMINATION OF SILICON














By

BILLY MAC SPENCER, II


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

TTKTNTTI/C DTT T V f P3T fDTTh A












ACKNOWLEDGEMENTS


I would like to give special thanks to my research advisor, Professor James D.


Winefordner,


for his patience and advice during the past several years.


great help in the preparation of this manuscript.


I appreciate Dr.


He provided


Winefordner agreeing


to take on this research project in conjunction with Texaco Inc.


Their are many group members who have given much help.


I appreciate Dr. Ben


Smith


laboratory


answering


many


questions.


Acknowledgement is given to


plasma,


Denis Boudreau


which are exhibited in chapter


for taking the photographs of the


Special thanks must go to Dennis Hueber for


his help with the research and especially for dealing with the computers.


Others in the


group include, but is not limited too, Giuseppe Petrucci, Anil Raghani, Mike Wensing,


Ohorodnik,


Cheryl


Davis,


Stefanie


Pagano,


Denise


Imbroisi,


Andrea


Pless,


MaryAnn Gunshefski, Donna Robie, and Wellington Masamba.


My appreciation goes to


to school.


Texaco Inc. for allowing me the opportunity to return


Texaco has provided the financial support that made this research possible.


In addition, I thank Dr.


Cliff Mansfield who has given me exceptional advice and for


help in establishing the


Texaco fellowship which I received.


Additional help has been


provided by Dr.


Bhagwandas Patel and Dr. Abdalla Ali.







My love and gratitude go to my parents,


support for all my endeavors.


who have always given their time and


I want to thank my sister and her husband, Shonna and


Robert Gage, and their family, who have given their support and understanding.


I have missed my niece and nephew,


I know


Vanessa and Derrick, during the time I have been


away


at school.


In addition,


gratitude


in-laws,


Glenda


Ravanelli, for their guidance and assistance.


Most of all, I


understanding and su]

to return to school.


would like to thank my wife, Kimberly,


pport.


for her love, patience,


She has given up part of her life to allow me the opportunity


I am always grateful for her help during the past several years.















TABLE OF CONTENTS


ACKNOWLEDGEMENTS


ABSTRACT


. 1i


S. ~ S S S 6 S S S V11


CHAPTER 1 A HISTORICAL PERSPECTIVE OF PLASMAS


S* 1


Plasmas in Elemental Analysis


Plasma Sources
Plasma Gases


*


. S S S S S S S S S S S S
I S S S S S S S S S . S S S S S


* I 9 5 .
* S S 6
* S S S 5 9 .


History of Capacitively Coupled Plasmas . .
Capacitively Coupled High- Frequency Plasmas
Capacitively Coupled Microwave Plasmas ..


Microwave Radiation Safety
Goals of This Research


CHAPTER


. 5
. 5
. 7


* S S S S 5
* S S 9 5 S S S S S


CAPACITIVELY COUPLED MICROWAVE PLASMA


INSTRUMENTATION


Introduction


Instrumentation


Microwave Generators


Waveguides . . .
Instrumentation Used in This Research


S I S S S 15


Apparatus
Detector


Sample Introduction
Plasma Configuration .


* a .
f f S S


. 19
. 26
. 27
. 29


CHAPTER 3 DETERMINATION OF SILICON IN ORGANIC SOLUTION


USING PNEUMATIC NEBULIZATION


Introduction


Experim mental . . . * * * *
D1ata Ioints . * * * * .
Data Curves
Wavelength s .. ..... .... .......


Gases


14


14


14








Cylindrical and Spherical Plasmas . . . . .
Tandem Electrode .
Electrode Depth
Spatial Profiles
Electrodes
i tmit of Detection


Results and Discussion
Cylindrical and Sp
Tandem Electrode
Electrode Depth
Spatial Profiles
Effect of Power


Electrodes


Molecular Gas Addition
Figures of Merit


herical Plasmas


. S S S U 9 9 S 9 9 S S
U U U U S S U S 9 4 5 5 0 U 9 9


* 9 9 9 U 9 9 5 U S S 0 0 9 S S 9 6 2


S . . a a .. 67


S a a a S 9 *. 81


Conclusion


81


CHAPTER 4 DETERMINATION OF SILICON IN ORGANIC SOLUTION BY


THERMAL VAPORIZATION


Introduction


Experimental
Results and Discussion


SU U 9 9 S 83


.8
* U S S S S S S U S U S U U S U U U 9 5 5 U U U S S S L83
* U S S S S S U U U a U a S S S S S S U S 5 U 9 4 5 5 U 8 8J\-


S. S 5 9 9 S S S . S 91


Integration Time
Vertical Profile


. S 91
94


Temporal Nature of Silicon
Effect of Ashing Time .
Effect of Ashing Power
Effect of Atomization Power
Electrode Cup Materials .
Matrix Modifiers......
Figures of Merit .
sion


S . . . * 10 1

. .* . . . 106



* * . . . 101
* . * * . 1 16
. * *. . 107
........................113
........................116
-17


CHAPTER 5 DIAGNOSTICS IN THE CYLINDRICAL CMP
Introduction . . . . . .
Electronic Excitation Temperature, Tex . .
Rotational Temperature, Trot. . . .


S S S U 0 9
. a a U U


. 119
. 119
. 120
. 123
. 123


OH
Electron Number Density, n . . .
lonization-Recombination Temperature, Tion .
Experimental * * .
Electronic Excitation Temperature, Tc .


...124
. 125
126
. 127
. ..128
I-1 rf\


Conclu







Electron Number Density, n . .
Ionization Recombination Temperature,Tio.


Background Spectra
Limits of Detection, LODs


Results and Discussion


Electronic Excitation Temperature, Texc
Rotational Temperature, Trot . .
Electron Number Density, n . .
Ionization Recombination Temperature,


Background Spectra
Limits of Detection, LODs


133


S 5 S S S S S S S S S S 13 4


St ft S .. 5 135


.* . . 5 135
S t ft 142
. .. .. ........ 149
T io n S f S f f f f S f S S 1 5 1
. .. ... .152


S S S 5 ft ft S S 15 8


Conclusions


CHAPTER 6 CONCLUSIONS


Summary of This Work
Unexplored Marvels


161


. . 163

S. 166


APPENDIX ABBREVIATIONS AND SYMBOLS


REFERENCE LIST


. . . 170


BIOGRAPHICAL SKETCH


. 184













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

A CYLINDRICAL HELIUM CAPACITIVELY COUPLED MICROWAVE


PLASMA:


DIAGNOSTICS AND DETERMINATION OF SILICON


By

Billy Mac Spencer, II


December


1993


Chairperson:


James D.


Winefordner


Major Department:


Chemistry


A cylindrical


shaped


plasma


is produced


a helium


capacitively


coupled


microwave plasma (He-CMP) and is applied to the determination of silicon in organic


solutions.


Samples are introduced by nebulization and thermal vaporization.


Spatial


profiles of Si emission, microwave power, helium plasma gas flow rate, and the addition


of molecular gases are studied.


Electrodes are fabricated from graphite, titanium, and


tungsten.


Using


pneumatic


nebulization,


tungsten


electrode


gave


lower


background and noise levels, compared to the graphite electrode.


W electrode is visible after 150 hours of operation.


No deterioration of the


A limit of detection (LOD), 3cr, of


mL-1


nebulization


a relative


sample


standard


introduction


deviation


technique.


(RSD)


obtained


Electrode


cups,


using


sample


introduction by thermal vaporization, are made from graphite,


titanium, and tungsten.






The optimum ashing power and ashing time are


135 W and 10 s, respectively.


Higher


ashing power


levels or longer ashing


times decreased


the Si signal,


probably due to


formation of silicon carbide.


with a RSD of


The LOD (3cr) for thermal vaporization is 0.03 pg mL-'


-18%.


Diagnostics,


temperatures and electron number densities, are determined in the


cylindrical He- CMP.


Electronic excitation temperature (Tex) is 3400 K.


Rotational


temperatures (Trot) are


=1600 K and


electron number density (ne) is 4


= 1900 K, using OH and N1+
^-7 0


x1014 cm-3


Values for


, respectively.


Trot, and ne are similar


for both aqueous and organic solutions.


Ionization-recombination temperatures (Tion)


are 6200


K and


5600


using


aqueous


solutions


respectively.


Background spectra of the cylindrical He CMP, while introducing aqueous and organic

solutions, are presented.


Tc,
Aexc












CHAPTER 1
A HISTORICAL PERSPECTIVE OF PLASMAS


Plasmas in Elemental Analysis


The quantitative and qualitative analysis of elements is routinely performed by


atomic spectroscopy.


One of the common ways is observing emission of electromagnetic


radiation as elements are introduced into a plasma.


A plasma is a partially ionized gas


in which a portion of the atomic or molecular species are present as ions.

examples include neon lamps, argon welding arcs, and interstellar space.'


Common


The plasma


in the most popular analytical technique is an inductively coupled plasma (ICP).


ICP uses radio frequency to generate a fluctuating magnetic field in which the ions and


electrons interact.


Plasmas may also be formed using microwave radiation.


Microwave


plasmas may be divided into categories which depend on the way energy is transferred


to the plasma.


Microwave induced plasmas (MIP) and capacitively coupled microwave


plasmas (CMP) use different means of coupling the microwave energy.


In the MIP, the


plasma is generated inside a quartz discharge tube which is held in a resonant cavity.


The microwave energy is transferred


cable.


from the generator to the cavity using a coaxial


In the CMP, a magnetron generates the microwave energy which is directed


through a waveguide to an electrode.


The plasma is formed at the top of the electrode.


In the following sections, different plasmas (CMP, MIP, and ICP) will be compared.






2

The history and application of CMPs as emission sources will be presented along with

the instrumentation used in this research.


Plasma Sources


In comparison to ICP, microwave plasmas offer several advantages.


Microwave


plasmas


generally


have


lower initial


operating


costs.


to the


popularity


commercial


microwave ovens,


magnetrons are


low cost and widely available.


Since


microwave


ovens


operate


at 2450


Mhz,


frequency


is commonly


employed


microwave plasmas.


Most microwave plasmas operate at lower gas flow rates than the


ICP, thus decreasing operational costs.


There are several reviews covering MIPs.3 -7


In comparison to CMP, the MIP


has greater stability and lower background noise when operated at similar power levels.

MIPs do not have contamination or background problems associated with the electrode,


which has to be routinely replaced in CMPs.

cost system and may be operated with He or a


argon.


In comparison to ICPs, the MIP is a lower

ir, where the ICP is usually operated with


With these advantages, the MIP has been widely used as an analytical technique


in comparison


to the CMP.


The MIP has been primarily used as a detector for gas


chromatography8'9


supercritical


fluid chromatography'0


(SFC).


Currently, a gas


chromatograph with an MIP detector is commercially available (HP 5921 AED, Hewlett


Packard,


USA).


The MIP is also a source for a mass spectrometer.11


The CMP does offer several advantages over the MIP.


Higher power levels may








cable and coupling loop with power levels above 150 W.2"


The MIP requires additional


transformers to impart continuity of the impedance between the coaxial cable and the


solid state generators. 12


CMP.


to 30 mm.


A very small plasma is produced by the MIP in contrast to the


e plasma diameter in the MIP ranges from 0.5 to 4 mm with a length of 10

13,14 Dimensions of the plasma in the CMP are approximately 8-25 mm in


diameter with a length of


-70 mm.


Since the MIP is a small plasma and must be


operated at low power levels,


there is difficulty in introducing samples as solutions.


Plasma enthalpy is insufficient in the MIP to desolvate and vaporize aerosols effectively


from directly nebulized solutions.] 3,1 5


Several research groups have introduced solutions


into an


MIP,


usually employing an


ultrasonic nebulizer or desolvation- condensation


system


remove


water. 16,17


plasma


be extinguished


if aerosols are


introduced over a long time or if concentrated solutions (


>500 upg mL-1) are used.16


Small quantities


pg) of


material


can easily perturb


the stability and excitation


capability of the MIP.3


By operating CMPs at higher power, they are more tolerant to


molecular species and have an increased ability to desolvate, atomize, and excite species

that have been introduced.


Several


groups have compared the performance of an ICP with the CMP.18'19


Boumans et al.20 performed the analysis of solutions by optical emission spectrometry


(OES) using


ICP and


CMP.


The characteristics


studied included detection


limits, matrix effects, sensitivity, and precision.


Results were corrected for instrumental


differences


nebulizer introduction


rates,


those of


measuring


equipment








superior,


to better


overall


detection


limits,


higher


sensitivity,


lower


interelement interference.

worse in CMP-OES. Th


Relative standard deviations for the 12 elements studied were


e conclusion was the ICP demonstrated an overall supremacy


to the


CMP


as an excitation


source


simultaneous


multi- element


analysis


solutions.20


Plasma Gases


Several different gases are employed in producing plasmas.


In ICPs, argon is


generally used,


with helium and nitrogen also having been used.


The helium plasma in


an ICP, He-ICP, has problems in stability and arcing to the induction coil.


A thin,


needle plasma is formed in the He-ICP which makes injection of the sample difficult.


Nitrogen -


air- ICPs


exhibit


molecular


band


spectra,


which


cause


spectral


interferences.-- Helium and argon are the preferred gases since they give no molecular

background spectra.23 Tanabe et al.24 reported the background in a helium MIP was an


order of magnitude lower than with an argon MIP.


Comparing helium and argon, helium


has a higher electrical resistivity which causes a lower power transfer efficiency in the


plasma.25


The thermal conductivity of helium is greater than argon.


In helium plasmas,


the heat dissipates faster toward the outer tube of the torch, increasing the chances of the


torch melting.22


Minimization of helium "attacking" the torch is accomplished by adding


a small amount of H,


(or N-,)


to the plasma gas.26


Several


groups have researched


special torch designs for use with helium plasmas.14,15,27- 29 The ionization energy of He








should be more efficient as sources for emission analysis.25


Helium has the advantage


over argon


in populating the higher energy


levels of free atoms and ions.31


Helium


plasmas have been applied to the


analysis of halogens32 -34 and nonmetal


Microwave


plasmas have been successfully operated using argon, helium, nitrogen,


and air.36


In the past,


CMPs have not been researched


as extensively as ICP


or MIPs,


possibly owing to the


success and commercialization of ICP and MIP systems.


Still there


exist


the possibility that CMP has the potential to be an alternative source for OES.


This research centers on the further development of the CMP for OES.


discussion


The following


on CMP and the techniques of MIP and ICP will only be mentioned for


comparison purposes.


The history


application


instrumentation of the CMP will


be discussed in greater detail.


History of Capacitive


y Coupled Plasmas


Capacitively Coupled High


- Freauencv Plasmas


Plasmas have


been studied for over 60 years.


They may be classified according


to the frequency of electromagnetic radiation used to produce the plasma.


radio (Rf) and microwave frequency


These are


. Radio frequency plasmas use frequencies below


109 Hz.


This


frequency range


has also been


referred


to as


"high


- frequency"


Microwave plasmas use frequencies greater than 109 Hz (GHz).


Early research centered


around hf plasmas.


In 1933


, Rohde and Schwarz37 produced a "flammenbogen," or gas








first to examine the emission spectrum of this gas discharge.


They identified 02, NO,


, NO2


NH bands


spectrum.


Cristescu


Grigorovici39


developed a hf discharge. A high frequency oscillator output was applied to two plates,

separated vertically by up to 15 cm. The lower plate had a copper cone with a platinum


A discharge was formed at this tip by touching with an isolated conductor.


conductor caused electrons to be emitted from the resulting heat generated by the strong


electric


field.


collisions,


the gas


was


heated


enough


ionization occurred


sustain the discharge.


Based on this description, what they had was a plasma.


A plasma


is defined as being a partially ionized gas.


determined the temperature to be 3000


In the same year, Cristescu and Grigorovici40


-3600 K using a power level of 650 W


temperature was determined by the NZ and OH bands of the spectrum.


Later, the same


authors published some of the theoretical treatments of the discharge they described in


1941.41


mechanism


proposed


based


electrical,


optical


thermal


measurements.


Calculations included current


vs power characteristics of the plasma.


High frequency


plasmas


have


been


to analyze


aqueous


solutions


OES.42'

MHz.


43


Badarau et a


43 used a nickel tipped electrode operating at a frequency of 43


Solutions were introduced, by a nebulization process, into the air plasma.


sensitivity of 13 elements were comparable to those obtained in an air-acetylene flame.

Dunken et al.42 employed a capacitively coupled radio-frequency plasma (50 MHz) for


the analysis of aqueous solutions.


Calibration curves were made for Ca, Sr, Ba, Cr, and


The concentration of the elements ranged from 30 to 7000 ppm.


Relative standard








Instruments, LTD.,


Vancouver,


B.C.,


Canada) became available which utilizes an Rf


plasma inside a graphite furnace atomic absorption spectrometer (GFAAS).44


A graphite


electrode (rod) is placed in the center of the graphite tube (furnace) of the GFAAS.


27.12 MHz Rf generator is connected to the central electrode and used to generate the


plasma.


The furnace acts as the counter-electrode.


This instrument is referred to as


graphite


furnace


capacitively


coupled


plasma


atomic


emission


spectrometer


(GFCCP- AES).


acronym.


Furnace atomization plasma emission spectrometry (FAPES) is another


Sturgeon et al.45 applied this system to the analysis of nine elements (Ag, Cd,


Cu, and


Detection limits (3a) were between


164 pg.


Precision was


12% with a linear dynamic range of 2-4 orders of magnitude.


Capacitively Coupled Microwave Plasmas


Microwave plasmas were initially discovered during the development of radar


equipment.46


The formation of these plasmas caused problems in the design of radar


transmitting


receiving


switches.


Physicists


became


involved as


microwave plasmas


presented a new area of research.


During the 1940s, microwave plasma systems were


developed,


theoretical


aspects


undertaken,


fundamental


processes


(temperature,


electron density, etc.) studied.


In the


1960s, actual applications were reported.


Since


1981, Dahmen47 has given an annual review on microwave plasmas.


Cobine and Wilbur48 studied plasmas produced with an


"electron torch."


apparatus described


used a


1000 MHz magnetron to produce microwaves at








Gases employed were air, nitrogen, carbon dioxide, argon, and helium.


They found the


air, N1, and CO, plasmas would melt a tungsten rod (mp =


33700C).


An argon plasma


would barely ignite paper held axially in the plasma.


The explanation was argon did not


dissociate into any other species.

Future work with plasmas, using a central electrode, are based on the work on


Cobine and Wilbur.48


In these


systems,


microwaves generated


by a


source are


coupled,


via a waveguide,


to the electrode.


The technique is then termed capacitively


coupled microwave plasma (CMP).


The first application of a CMP, for the analysis of


solutions or solids,


was performed by Mavrodineanu and Hughes.


Mavrodineanu and


Hughes49 used a 2450 MHz magnetron to produce a helium plasma.


analyzed using a graphite cup electrode.


Solid samples were


The plasma formed on the top of the electrode


vaporized


sample.


In addition,


several


elements


aqueous


solution


were


determined using a pneumatic nebulizer.


The limits of detection were in the range of 10


(Mo,


Pb, and Zn) to 0.8 ppm (Na).


Temperature of the helium discharge was


estimated between 2900 and 3300 K by its ability to melt molybdenum but not tantalum

or tungsten.

Jecht and Kessler50 developed a similar CMP which used a molybdenum water-


cooled electrode.

and helium. The

2370 A. This CI


They investigated plasmas produced at 2400 MHz for air, nitrogen,


, temperature was estimated at 4000 K using the NO rotational band at


MIP-OES system was then used in several applications to real samples.


Limestone and dolomit were analyzed for the concentration of Ca, Mg, Al and Fe.51








batch materials.52


Murayama et al.53 introduced solutions by nebulization into a CMP.


This work employed a water-cooled aluminum electrode.


The detection limit, based on


two times the root-mean-square (rms) deviation of the background,


elements.


were determined


Inter-element effects of sodium on seven elements was also studied.


Work performed by various research


groups, especially


Kessler and coworkers,50 -52


Murayama and coworkers,53 -56 Goto et al.,


57 and Kitaqawa and Takeuchi58'59 led to the


commercialization of CMP-OES equipment by (1) Hitachi,


Tokyo, Japan, designated


ultra high


frequency


(UHF)


plasma,


Applied


Research


Laboratories,


Ecublens-


Lausanne, Switzerland, and (3) Erbe-Elektromediziu,


Tuebingen, Federal Republic of


Germany.60


Previous CMPs utilized a solid electrode. In this design, aerosol samples were

introduced by diffusion into the outer edges of the plasma. Hwang et al.61 improved on


this design by making a tubular electrode where the sample is introduced through the


center of the plasma.


elements C,


thermal vaporization device.


The CMP was applied to the analysis of several materials.


0, and Hg were determined in orchard leaves and in fish using a


The analysis of trace oxygen and hydrogen extracted from


metal was carried out using a chamber to heat the samples and allow the helium to carry

the oxygen and hydrogen to the plasma.63

During the past several years, applications and diagnostics have centered around


work


done


Winefordner and


coworkers.


Uchida


et al.64


measured


butlyltin


compounds by interfacing a


helium CMP with a gas chromatography


system.








plasma formed on top of the cup and vaporized the sample.


analysis of solutions by discrete sampling.65


This was applied to the


Masamba and coworkers67,68 investigated


the temperature, electron number density, as well as the influence of power, observation


position, solution uptake rate and carrier gas flow on the CMP.


developed an electrode made from tungsten wire.


Ali and Winefordner69'70


Having a smaller mass, the filament


gives a high rate of sample vaporization while being inexpensive and easy to construct.

Rotational and excitation temperatures were determined along with the effect of power


and plasma gas flow rate.69


The dependence of Cu and He emission on plasma flow rate


power,


along


radial


axial


emission


distributions,


were


studied.70


Hueber et al.


71 developed a hydride generation system for the introduction of arsenic into


a CMP.


Microwave Radiation Safety


Exposure to microwaves has been related to a variety of health problems including


denaturation


of human


proteins,


eye cataracts,


sterility,


headaches,


and disruption of


neural


cardiovascular


systems.


72-75


operation


microwave


instrumentation


is of vital


concern.


One problem exists is the establishment of safe


exposure limit


In the United States, the standard safe exposure level is


mW cm3


at 2.450 GHz.


The standard in Russia


78 is 0.01 mW cm-3


while that in Australia7 is


mW cm-3


The measurement of stray radiation is made at a distance of 5 cm from the


source of emission. Microwave radiation from the plasma is easily contained using wire
I I 6 i I V V arrh/^ ^ ... I B* 1* 6. r* in^ a








radiation is much more complicated.


Van Dalen et al.


78 reported that an aluminum box,


around an MIP, reduced the radiation leakage to a safe level.


The system used in the


research


for this


dissertation


was contained


inside an


aluminum


box.


Copper wire


screen,


which


prevents radiation


leakage,79


was placed


around openings in


the box.


Measurement


microwave


leakage


was


performed


using


a leak


detector


(Narda


Microwave Corp., Hauppauge, New


York).


Goals of This Research


The control of trace elements,


< 20 ng mL)1


, in high purity solvents is critical


to the semiconductor industry. 8-s 8


As the integration density of integrated circuits in


microchips increases, the manufacturing yields of microchips are limited by impurities

in the processing chemicals.8 Currently, the quantitation of 35 elements, at a level of


20 ng mL-1


, is required for chemicals such as amines, alcohols, and hydrocarbons.


element required is silicon.


Antifoaments, such as poly-dimethylsiloxanes, are a source


of silicon contamination in high purity solvents.


In the chemical industry, antifoaments


are added to the crude oil during the refining process.


minimum


level


at which


an element


is quantitated


is called


limit


quantitation (LOQ).


The LOQ is found by multiplying the limit of detection (LOD) by


Multiplying the standard deviation of a blank solution by three, 3a, and dividing


by the slope of the calibration curve gives the LOD.83


Quantitation at a level of 20 ng


mL-1 requires an LOD of 2 ng mL1






12

The techniques used to determine silicon include inductively coupled plasma-


optical


emission


spectroscopy


(ICP-OES),


inductively


coupled


plasma- mass


spectrometry (ICP-MS), and graphite furnace atomic absorption spectrometry (GFAAS).

Each of the above techniques have disadvantages that limit its ability in the analysis of


In ICP-OES,


the LOD for Si is


16 ng mL-'.84


This technique does not give the


required LOD of


ng mL-'


ICP-MS is often considered the ultimate method for trace


element analysis.


Typical LOD values for ICP-MS are in the parts per trillion range, pg


mL-'


The determination of Si, in organic solutions, by ICP-MS i


difficult.


Problems


are encountered due to spectral interference.


The isotopes and natural abundances (in


parenthesis) of Si are


(92.2%), 29 (4.


%), and 30 (3.1


Interferences at isotope


include


14N14N+


12C160+


The interference at isotope 29 include 14N15N+


14N14NH+


12C160H+


and '3C160+


, while at isotope 30,


14N160+ interferes.85


The reported


LOD for Si,


in organic solutions,


is above


in ICP-


86,87


In aqueous solutions, the LOD for Si is 10 ng mL]' by ICP-MS.


Smith et


89 added 37 mL min-' of xenon to the argon carrier gas.


Adding xenon gas decreased


the spectral interference of '4N4+


, by a factor of 300,


versus the background when no


xenon is present.


This technique can prove expensive because xenon is a rare gas and


costs $8-$16 per liter.


An LOD of 0.6 ng mL-U


, using isotope


is reported for Si in


aqueous solution.89


The high LODs exhibited by Si, Ca, and Fe in ICP-MS are due to


the additional polyatomic peaks, formed by organic compounds, which produce spectral

interference. 86









Another technique often employed for Si analysis is GFAAS.


mL'1 is reported for GFAAS.8'

analysis in high purity solvents.

analyzed per sample determination

require 30 determinations. It wo


An LOD of 1 ng


This gives an LOD that meets the requirements of Si

A disadvantage of GFAAS is that only one element is

n. The quantitation of 30 elements in a sample would

uld be beneficial to have a technique that could detect


mL-'


which


could


be developed


simultaneous


multielement analysis


(SMA).


The technique of optical emission spectroscopy is frequently used for SMA.












CHAPTER


CAPACITIVELY COUPLED MICROWAVE PLASMA INSTRUMENTATION


Introduction


Microwave energy is produced in CMP using a generator;


directed down a waveguide.


the energy is then


An electrode is placed at the opposite end of the waveguide.


The electrode capacitively couples with the microwaves.


A plasma may be formed at the


tip of the electrode,


where the microwave energy is emitted from the electrode.


If high


enough power is applied, the gas surrounding the electrode is partially ionized and auto-


ignition of the plasma occurs.


If lower power is utilized,


"seeding"


of electrons


required using a tesla coil or by touching a piece of wire, held in an insulator, to the

electrode tip.


Instrumentation


Microwave Generators


Initial microwave plasma sources were medical diathermy units.5'91


of 100 to 250 W produced microwaves at 2450 MHz.


These units


Development of microwave ovens


has decreased the cost while increasing the availability of magnetron power sources.

most common magnetrons are operated at 2450 MHz.







15

Waveguides


Waveguides, or resonant cavities, are commonly used to transmit electromagnetic


radiation of the microwave frequency.


The first idea of propagating electromagnetic


waves


down


a hollow


waveguidee)


was


1893.


Thomson92


examined


mathematics of what might occur if an electric charge was released inside a closed metal


cylinder.


It was not until the years 1913 to 1920 that Zahn93 and Schriever94 performed


the experimental work.

Generally, the waveguide is made of a hollow cavity that may be rectangular or


cylindrical in shape. Coaxial

comparison to coaxial cables,


cables may also be used for microwave transmission.


waveguides have a higher power handling capability, a


lower loss per unit length, and a lower cost mechanical structure.95 The coaxial cables

have a tendency to become hot and radiate heat at power levels above 100 W, causing


a loss of power.'


The reflections caused by the flanges used in connecting wavequide


sections is usually less than that associated with coaxial connectors.95


In waveguides, the


electric and magnetic fields are maintained on


the inside of the walls.


No power is


therefore


lost by


radiation.


Dielectric


loss is also negligible,


since the guides are


normally filled with air.96

Waveguides are fabricated from highly conductive metals with dimensions similar


to the


wavelength.


cavity


can sustain


microwave


oscillations


which


form


constructive interference pattern from superimposed microwaves multiply reflected from


Th;c nhnnrnmonnn kc ryfrrpA ,n QC Q


" .. 4:. ..n..... "97


4. Vl v., 6.1, .w /* I 1?,,


%I V.. III I I |I I J"l' l~ .-"K 1 rI. 1 1-


^intL 44,n


l lr- a 'i


I


I








This power loss is related to the skin depth (5) of the material.


Skin depth is defined as


the depth at which the current decays to 1/e (0.37) of its value at the surface.97


Equation


defines the skin depth.95


Here pr


= relative permeability dimensionlesss), /.o


= permeability in free space (4r


7 Henry m-'), a = conductivity of the medium mhoss m'), and f


= frequency (Hz).


The skin depth is inversely proportional to the conductivity of the waveguide material


and frequency of the electromagnetic radiation.

will have less power lost to heat. At microwa


Materials which have small skin depths


ve frequencies, the skin depth of a metal


is in


micrometer


range.


Therefore,


the current


flows


along


surface of


conductor which increases its resistive effect.


This is referred to as the


"skin effect,"


because the current is concentrated in the outer surface of the conductor.98


Values for


several metals at 2450 MHz are Ag,


1.3 pm; Cu,


1.3 pm; Au, 1.6 pm; Al,


1.6 pm and


brass, 2.7 pm.99


The metals Ag,


Cu, and Au are difficult to machine,


while Ag and Cu


also corrode easily in air.


Au is expensive.


Al and brass are the materials generally


used to manufacture waveguides.


Electromagnetic waves


be propagated


three different ways,


which are


described as transverse electric magnetic (TEM), transverse electric (TE), and transverse


magnetic


(TM).96


In the


TEM


mode,


electric


magnetic


components


/Zr -o o"


__


I


_








waves


which


are perpendicular


to the


direction


propagation.


is where


transverse


magnetic


wave exists.


order


for the electromagnetic


wave


to travel


through


the waveguide,


two conditions must be met.95


Electrical


flux lines must be


perpendicular to the waveguide walls and magnetic lines must run parallel to the surface


of the walls.95


Because of these two points, the


TEM


mode is not propagated in the


waveguide. 00

Describing the mode which is propagated, two subscripts are added to the TM and


TE designations,


to give


TMmn and TEmn.


The m and n are integers that define the


quantity of half wavelengths that are in the A and B dimensions of the waveguide where


A represents the width and B the length.


A


in order for power currents to flow


prevents the flow of power currents.


The A dimension must satisfy the inequality


>_X (2)
2


. The value where


2


This frequency is designated the cutoff frequency


waveguide


properties


a high


filter


since


power


transmission


is possible only when


f > f


or X

The cutoff wavelength


is then


=2A.


Since f~~


= V, the cutoff frequency for the waveguide is


(fc)." 95











~r~r


where Lr is the relative permeability dimensionlesss) and Er is the relative permittivity


dimensionlesss).


The value 1r equals the magnetic permeability (or inductivity) of the


medium, k [Henry (H) m-i], divided by the permeability of vacuum or free space, po (H


Similarly, e, equals the dielectric permittivity (or capacitivity) of the medium, e


[farad (F) m-'] divided by the dielectric permittivity of vacuum or free space,


The wavelength in the guide is represented by


X


o (F m-').


It is defined as96


/f)2


where X (= v,/f) is the wavelength in an unbounded dielectric, and


f is the operating


frequency


(usually


2450


MHz).


phase


velocity


an unbounded


dielectric


represented by v, and is equal to (Le)-'-


, where u and e


are the permeability (H m-1) and


permittivity (F m-')


respectively.


The phase velocity in the positive z direction inside


the waveguide is


/f/)


Because the cutoff frequency is a function of the modes and waveguide dimensions, the


actual size of the waveguide governs the propagation of the modes.

lowest cutoff frequency in a given guide is the dominant mode. In


The mode with the


a rectangular guide


1-( fc


1-( f


m-l).







19

may also be present, but only the dominant mode will propagate, and the higher modes

near the sources or discontinuities will decay very fast.96

Earlier it was noted that microwave generators are used to generate and introduce


microwaves into a waveguide.

energy from the waveguide. I


distance of mX,/4,


An electrode, or coaxial cable, is used to withdraw the


Both the generator probe and the electrode are placed at a


where m is an odd integer, from opposite ends of the waveguide.


Here the microwaves traveling to the edges return in phase with the waves going in the


opposite direction.


This phase shift occurs since the wave shifts 900 by travelling the


Xg/4 distance, then a


1800


shift by reflection at the wall, and another 900 by traveling


back the X./4 distance.'6


A maximum coupling efficiency


and a maximum density in


the electric fields occur when the probe and the electrode are placed a distance of mXg/4


from


the edges.'o'


Ali 0'


reported


the optimum


depth,


inside a


waveguide,


for the


electrode was 3 cm.


This also corresponds to X\/4.


Instrumentation Used in


This Research


Apparatus


Diagrams


CMP


electrodes


are given


Figures


2-2,


respectively.


Table 2 1 lists the instrumentation and manufacturers.


The materials used


to fabricate the electrodes are listed in Table 2 -2.


The dimensions of the waveguide are


given by Hwang et al.61


The torch is made from two concentric quartz tubes.


The inner


tiheP hnl d the peletrnde and cdirepet the va mnl1 vannr thrnnirh the center nf the electrnce
























































cA

(/i
C-
0-
03




Ci-)


0


* t
r:







c-
C.)




Id








CL








21








a>
0 a. 0 0
ff L 3. C3O

.9= o~ cm
I: 1 .c

,O
4'- N



Ut~
S)- 0)



____._Eccu 0




1E 0I



<-E
3 cu














a, IIl 4 I
ra I
____\__ L.^






V 0








OI
---------~~~ I--- m-^ ----m mof
SI-




EE/
oN







O O
( t




L ..
s0) Co
00 \






--a
0) -^- i C,




+- -

o \





.-
p
0) 00//\1--1S<







aa
E
O S '' ''
0D '







CU 0
0) 1L.*'0)



00















0~*-

SI-
u-
-a c




~0
3 H
u rt



c E

ed C



Q)
0^

N
3C





+ -~


0 0
L'






4 C)
(^ X
E 4,
3 Cl
bU 0

















,C
*S E



O 5


o -c
u. +-
cn <
QJ*r
_~. E-






rt F

21
^











































\I/ I


T


ri








Table 2-1


: CMP Instrumentation and Manufacturers


Instrument


Manufacturer


Photodiode array:
model IRY-1024G


OSMA


Princeton Instruments,


Princeton


USA


OSMA Detector Controller

Photodiode Array Software,
Spectrometric Multichannel


Analysis, ST-120


Princeton Instruments

Princeton Instruments


Ver. 2.00


Spectrometer:
HR 1000. 1 I


Jobin-Yvon


v, 2400 grooves


Instruments SA, Inc.,
Metuchen, NJ


mm'


, linear dispersion


0.5 nm mm-


Computer:


80286,


MHz


PC's Limited


Austin,


High voltage d.c.


power supply:


mode


Hipotronics,
Brewster, NY


805-1A (maximum power
output of 1.6 kW)


Magnetron:


Model


National Laboratories


NL 10251-2 (frequency
2.45 GHz, maximum


power output


Orlando, FL


1.6 kW)


Nebulizer


J. E.


Meinhard Assoc.


Santa Ana,


Aluminum Waveguide


Quartz torch


USA


Laboratory made


Laboratory made









Table


Electrode and Electrode Cup Materials


Material


Manufacturer


Graphite:


National Carbide


Laboratory made


Corp., Grade L3810 AGKSP,


Si contamination


= 1.9 ppm.


Titanium:


99.7%


Laboratory made


Johnson Matthey
Ward, Hill, MA.


Tungsten, 99.97%


Thermo-Electron/Tecomet


Si contamination


= 20 ug/mg


Wilmington, MA








The outer tube directs the plasma gas, in a tangential flow, around the electrode.


dimensions of the outer torch are 15.5 mm i.d. and 17.5 mm o.d.

and covering the top of the waveguide are cooling coils. Water


the coils to cool the torch.


Surrounding the torch


is recirculated through


A coaxial waveguide is place above the cooling coils.


coaxial waveguide propagates the microwaves out of the waveguide, toward the top of


the electrode.


This configuration increases the stability of the plasma.


A quartz chimney


( 4 cm diameter) surrounds the coaxial waveguide.


This chimney is used to keep air


currents


from


disturbing


plasma.


Masamba'6


found


current


from


magnetron to the plasma fluctuated randomly.

random fluctuations in the emission from the (


This fluctuation in current contributed to

CMP. Masamba26 did not determine the


cause of


fluctuations.


current


regulator and


stabilizing


circuit were built (D.


Hueber, unpublished results).


This system improved the stability of the CMP.


Detector


The detector used in this work is a photodiode array (PDA).


of silicon photodiodes in an integrated -circuit form.83


The PDA is made


A total of 1024 diodes (or pixels)


are arranged in a linear manner.


The PDA covers


= 20nm.


Each diode corresponds


to 0.0198 nm (wavelength range around 300 nm). The PDA sequentially reads the signal

from each diode after a specified integration time. The integration time is the interval


the diode is allowed to collect the incident electromagnetic radiation.


from 0.033 s to greater than 2


This time ranges


Longer integration times usually lead to overexposure







27

integrated, the signal from each diode is added to the signal, from the same diode, from


the previous scan(s).


the PDA.


This allows for larger signals to be collected without overexposing


A disadvantage of the PDA includes the necessity to cool the detector in order


to decrease


the dark current.


Also,


the diodes must be analyzed sequentially.


system does not allow for random access of any diode(s).


to collect and analyze data.


The PDA software was used


Additional software programs were written to improve data


handling capabilities.


program,


COMPARE


Hueber,


unpublished results),


allows for the average and standard deviation of a specified diode to be obtained from


a given number of spectra.


data (SIDE,


A program was also developed to display three dimensional


1.1, D. Hueber, unpublished results).


Samole Introduction


There are several ways to introduce samples into a plasma.

sample introduction technique is pneumatic nebulization.102 In pne


a capillary tube carries the nebulized sample to the spray chamber.


tube, gas flows in a concentric manner.


The most common


,umatic nebulization,


Surrounding this


This causes reduced pressure at the tip of the


capillary.


fine aerosol


is produced by the Bernoulli affect.83


The aerosol passes


through a spray chamber,


which allows removal of the larger droplets.


The smaller


droplets, typically less than 10 pcm in diameter, proceed to the plasma.l03


primarily used because of their ease of operation.


Nebulizers are


The disadvantages include high sample


solution consumption, low sample transport efficiency, and clogging due to particulates








as the weakest link.'04


Recently,


nebulizers have been developed specifically for use


with helium gas.


Previous researchers


have noted that nebulizers do not operate very well


helium.23'25


Nebulizers, primarily used with ICPs, are operated with argon gas.


In these


nebulizers, helium produces larger droplets and a nonuniform mist.


Helium is a smaller


atom than argon (1.22 A radius for He


1.91 A radius for Ar).30


Helium nebulizers


have a smaller outlet orifice and operate at higher pressure (45-60 psi He vs 30-35 psi


The efficiencies of an argon nebulizer operated with helium and a helium nebulizer


were


calculated


aspirating


a fixed


amount


solvent


weighing


amount


received


in the drain.


The difference in weights,


or efficiency,


was attributed to the


amount of vapor reaching the plasma.


The nebulizers were optimized for pressure, and


gas and solution flow rates using the signal and signal-to-noise ratio of silicon.


efficiency of the argon nebulizer,


with helium gas,


was


The helium nebulizer had


an efficiency of 11


of two.


Using the helium nebulizer, the LOD for Si decreased by a factor


The helium nebulizer had a higher efficiency, but lower solution flow rate (0.6


mL min-'


vs 2.2


mL min-' for Ar).


However,


visually it was observed that the helium


nebulizer


produced


a finer


versus


the argon


nebulizer.


Currently,


there


is no


research in the literature on nebulizers operated with helium gas.


Another technique for introduction of samples is by thermal vaporization.


In this


research,


an electrode cup is used to hold a fixed amount of sample.


The plasma is


produced on top of the electrode cup.


The sample is then vaporized by the plasma.








nebulization,


since


the entire


sample reaches


plasma.


sample is


vaporized


rapidly, leading to a larger signal than by nebulization of the same amount of solution.

Another advantage of this technique is that the sample may be heated or ashed to remove


volatile components,


organic material,


or water from


the analyte.


the CMP,


plasma is operated at low power to remove the sample matrix; a higher power level is


used to vaporize the analyte.


ectrode and electrode cup were


fabricated from


graphite, titanium, and tungsten.


Plasma Configuration


Figure

are formed.


is a stability diagram showing where three different plasma shapes


e


The operating stability diagram was obtained following a procedure similar


to that of Rezaaiyaan


et al.'o05


Forbes


et al. 13


The data for the diagram


were


obtained by igniting the discharge and allowing a stable plasma to form.


The flow rates


were 4 L min


1 helium and 150 cm3 min'1 hydrogen with a power level of 450 W


. The


helium plasma gas was increased or decreased and visual observations performed.


experiments were repeated by keeping the helium gas flow rate constant and varying the


power level.


To determine the points where a plasma was started, an insulated wire was


touched to the electrode at each power and flow rate.


The power or gas flow was varied


until an appropriate setting was reached where the plasma formed.


Power levels of 0-


1400 W


helium


flows of 0-17 L min'


were investigated.


In general,


emission intensity increases


increasing


microwave power and


the plasma length










































;-4
E

e
a

Ia






cn
rJ

I-j







b1
















N,0



CO




E
-i
o 0
rcu


o

()
(0


0 0
co
(D


(M) JOMOd






32

gas flow above 2 L min'1 and below 17 L min-' and power levels above 150-300 W and


below


1400 W


Regions specified


the diagram are semi-quantitative and do not


define exact values.


The addition of a molecular gas (H2, N2, or 0-) or a change in the


diameter of the quartz torch alters the size and shape of each region.


Colors exhibited


by the plasma range from pinkish-blue to very bright pink.


The stability diagram has four regions:


no plasma,


filament plasma, spherical


plasma, and cylindrical plasma.


A photograph of the spherical plasma is given in Figure


2-4.


The cylindrical plasma is shown in Figure


2-5.


The filament plasma is bright


pink in col

in length.

electrode.


or.


It is a thin plasma that is approximately 2-3 mm in diameter and 10 mm


This plasma is very unstable, as it frequently moves around the surface of the

Focussing of this plasma onto the entrance slit of the spectrometer was very


difficult.


This


plasma


also eroded


the electrode


much


faster


spherical


cylindrical plasmas.


Flow rates below 4 L min' typically cause the filament plasma to


attack and melt the quartz torch.

The spherical plasma generated with the CMP has previously been used to analyze


steel samples, 106 aqueous solutions,68 and arsenic71 by hydride generation.


this plasma is


The size of


1- 1.5 cm in diameter with a length of 2-3 cm depending on the power


level and flow rates of helium and hydrogen gas.


a pinkish-blue tint.


between 6 and


This plasma is very stable and exhibits


The spherical plasma changes to the cylindrical plasma at flow rates


7 L min-'


At flow rates above 6 -


L min-1


, the plasma is cylindrical.


Figure


shows































Figure 2-4:


The Spherical Plasma in the He-CMP.







34



















































r































Figure 2-5:


The Cylindrical Plasma in the He-CMP.





36








decreased as the depth of the electrode, inside the torch,


increased.


Lowering of the


electrode


increased


the signal


signal to- noise ratio


for the analysis of silicon,


which will be discussed in chapter 3.

the electrode (0.6-0.8 cm). The le


This plasma has a diameter very similar to that of


ngth of the plasma ranges from 2 cm at 6 L min-1


to 4.5 cm at


17 L minl'


, with a characteristic very bright pink color.


Increasing the


power above 1000 W causes the emission intensity to increase and the plasma changes


from pink to almost white in color.


Like the spherical plasma, the cylindrical plasma is


very stable. The main differences include the color, size, and background due to plasma

emission. The cylindrical plasma has a background level three times higher than that of


the spherical plasma; however, the signal-to-noise for Si in the cylindrical is higher.


The cylindrical plasma will be studied in this work.


The analysis of silicon, in organic


solutions, will be performed by solution nebulization (chapter 3) and thermal vaporization


(chapter


diagnostics,


temperatures


electron


number


density,


cylindrical plasma will be presented in chapter












CHAPTER 3
DETERMINATION OF SILICON IN ORGANIC SOLUTION USING
PNEUMATIC NEBULIZATION



Introduction


Although most of the work performed with plasmas deal with aqueous solutions,


organic solutions


may also be analyzed.


Previous research


CMPs utilized


introduction of aqueous solutions by pneumatic nebulization.61'68


organic solutions into the CMP,


reported.


The introduction of


with pneumatic nebulization, has not previously been


Introducing solutions into an MIP has proven difficult because small amounts


of sample material (3


ug) degrade the stability and extinguishes the plasma.3,'15


Since


the MIP is operated at low powers (


200 W), the plasma does not have enough energy


to vaporize or evaporate solid or liquid samples, or to atomize the analyte species.3'79

Methods for the introduction of solutions into the MIP include heating the sample


vapor and


passing


it through


a cooled condenser to remove the


liquid. 107


Thermal


vaporization (TV) techniques are also used for sample introduction in MIPs.23'108


In this


technique, the sample is placed on a graphite cup,08os or metal wire,23 which is heated to


remove the solvent, followed by vaporization of the analyte into the plasma.


Different


MIP torches and cavities are currently being utilized to increase the operating power

(40-500 W) and facilitate the introduction of aerosols. 109, 10








Researchers have used ICPs to analyze organic samples.


lubricating oils with xylene to determine 21 elements.


Brown11' diluted used


The analysis of metals in used oils


is called


"wear metal analysis."


Blades and Hauser'1" analyzed nonmetals in xylene


using an Ar-ICP with a photodiode array detector.


Nygaard and Sotera'13 examined the


concentration of Cd, Mn, and Fe in organic solvents such as acetone, and tetrahydrofuran

using an Ar-ICP.

The research reported here will concern the analysis of Si, in organic solution,


with a He-CMI

analysis are desc

effect of power,


ribc


Development and optimization of the cylindrical He-CMP for Si

ed. Items presented will include the spatial profiles of Si, and the


plasma gas flow rate, and molecular gas addition.


Experimental


The instrumentation is discussed in chapter


The organic silicon standard (5000


pg mL') was purchased


from Conostan Division of Conoco, Inc.


(Ponca City,


OK).


Kerosene from Fisher Scientific was used to dilute the organic Si standard to the desired


concentration.


Slit height and width were


2 mm and 20 pm, respectively.


A UG5 filter


was employed to decrease the background emission continuum.


A pneumatic nebulizer,


designed for helium gas,


was obtained


from J


Meinhard Assoc.


(Santa Ana,


CA).


Solutions were introduced at a flow rate of 1.2 mL mini- using a peristaltic pump


(model:


Rabbit, Rainin Instrument Co.


Inc., Boston, MA).


A power level of 800 W


used


the experiments.


Parameters


for the photodiode array


detector (PDA)








integration time and number of accumulations were based on S/N measurements.


use of the PDA is described in greater detail in chapter


The above conditions were


used unless stated otherwise.


Data Points


Graphs


are generated


using


from


signal,


background,


noise


measurements.


The analytical signal is obtained by measuring the total signal from the


analytical sample and subtracting the signal from a blank solution.


is identical to the analytical sample except the analyte,


The blank solution


which in this research is silicon,


is absent.

for the sigr


A signal due to the blank sample is called the background signal.


are the average of five measurements.


Data points


Error bars in the graphs represent


one standard deviation (la), based on five measurements.


Blank noise levels (N) were


determined


measuring


standard


deviation


measurements


blank


solution.


Data Curves


Curves presented


the graphs


were generated


using a curve


fitting program


(Origin,


MicroCa


Software,


Inc.,


Northampton,


MA).


curves


were


determined using a polynomial fit with the following exceptions.


Data for plasma gas


flow rate (Figures 3


and 3


-3) were divided into two groups according to the type of


plasma (spherical or cylindrical).


Data for signal, background, and S/N for flow rates


la









Signal and background data for the cylindrical plasma, 7- 11 L minl1


, was also subjected


to least squares fit.


The S/N data, representing the cylindrical plasma,


was plotted as


the average of the five data points.


to emphasize the transition which occurs with


Lines connected the data points at 6 and 7 L minr1


increasing plasma gas flow rate.


graph


of S/N


versus


power,


Figure


3-10,


a Gaussian


fitting function.


Lines


connecting the data points were employed on


the graph of S/N versus molecular gas


addition, Figure


Wavelengths


Silicon emission is observed at either 251.61


nm or


288.16 nm.


Initially, the


spherical and cylindrical plasmas were compared using an aqueous Si standard.


samples interfere with the Si


a bandhead at 281.1 nm.


Aqueous


16 nm line because of the OH molecular band that has


The Si line at 251.61 nm was observed in the experiment with


the tandem


electrode.


In all


other experiments,


the 288.16 nm


line was


measured.


When a graphite electrode was use

emission was observed at 247.9 nm.


nm line, especially at power leve


or organic samples introduced, a strong carbon


This carbon line would interfere with the Si 251.61

above 700 W.


Gases


Gases


employed


research


were


commercial


grade.


helium


hydrogen gas pressure were 45 psi.


Helium gas flow rates for the plasma and nebulizer


*d,








tee was installed


in the plasma gas line,


between the plasma gas flow meter and the


quartz torch.


This allowed the introduction of molecular gases (H2, N2, or 02) to the


helium


plasma gas.


Hydrogen


was


added


to the


helium


plasma


to prevent


melting of the quartz


torch.


hydrogen


flow rate


150 cm3


min-1


experiments except the molecular gas addition studies.


Cylindrical and SDherical Plasmas


In comparing the spherical and cylindrical plasmas, the helium gas flow rate was


increased from 4 L min-' to 10 L minm'


An observation height of 5 mm was employed.


This height was the optimum for the spherical plasma, based on silicon signal and signal


-to-noise ratio (S/N) measurements.


was


The depth of the electrode inside the quartz torch


mm.


Tandem Electrode


In the experiment involving a tandem electrode, a graphite rod was employed as


the tandem electrode.


The graphite rod was mounted on an x-y translator to allow


positioning of the rod above the CMP electrode.


An insulated copper wire was attached


to the graphite rod and connected to an electrical


ground.


The distance between the


tandem


electrode


bottom


electrode


was


varied


from


mm to 30


mm.


emission


signals


were


observed


a vertical


distance,


respect


to the


bottom


electrode, ranging from 4 mm to


mm.






43

Electrode Depth


The depth of the electrode inside the quartz torch was investigated for values


between 0 mm and 16 mm.


kept constant.


Positioning of the electrode in relation to the waveguide was


Alil'0 found an optimum length of the electrode inside the waveguide


should be approximately


3 cm.


Placing the electrode 3 cm into the waveguide reduced


the plasma acoustic noise and resulted in increased plasma excitation.


This distance


corresponds to '4 the wavelength of the microwave radiation in free space.65 The bottom


of the tungsten electrode was manufactured


to fit inside the inner tube of the quartz


torch.


This is shown in Figure


A piece of teflon tape was wrapped around the


bottom of the electrode to prevent the electrode from descending completely into the


inner tube of the torch.


Various depths could be studied by changing the position of the


teflon tape.


The teflon tape required replacement after 15


-30 minutes, due to melting.


Soatial Profiles


Determining the spatial profiles was aided by having the waveguide, magnetron


torch


mounted


components in


on an apparatus


x, y,


z directions.


positioner


Metric


allowed


rulers on


maneuvering


the positioner provided


measurement scales.


Measurements for the vertical profile were performed after aligning


the top of the electrode with the center of the spectrometer's entrance slit.

height is the distance the electrode was lowered below the entrance slit.


profiles were taken using two different methods.


The vertical

Horizontal


Initially. the nlasma image was nlaced








Another


method


involved


centering


plasma


on the


entrance


spectrometer and moving the plasma sideways in either direction.


obtained using both methods,


plasma is reported.


Similar results were


therefore, only the result involving the centering of the


In the horizontal profile, a vertical height of 12 mm was used.


Electrodes


Electrodes were


fabricated


from


graphite,


titanium, and tungsten.


Table


2-2


gives the materials and manufacturers for the electrodes.


A graphite electrode was used


in comparing the spherical and cylindrical plasmas, examining spatial profiles, and in the


effect of a tandem electrode.


A tungsten electrode was used in examining the influence


of electrode depth, effects of power and molecular gas, and the limit of detection.

experiment compared electrodes fabricated from graphite, tungsten, and titanium.


electrodes


were


compared


observing


the emission


from


a blank


solution,


while


operating the plasma at 500 W.


In comparing the electrode materials, the blank noise


level (N) was determined from the standard deviation of 10 diodes occurring between 253


nm and 254 nm.


A silicon impurity in the titanium prevented the measurement of the


background at 251.61 nm.


Limit of Detection


The limit of detection for Si was based on 3-a.


The standard deviation, (a),


calculated


using


a blank


sample and


measuring


signal


eleven


times at


the same


was








along


a blank,


calibration


curve.


correlation


coefficient


calibration curve was 0.997


Results and Discussion


Cylindrical and Spherical Plasmas


Initially


spherical


shaped


plasma,


described


chapter


used


determine Si.


This plasma gave an LOD of 17


pg mLU1


for Si in aqueous solution.


Since this was an unacceptable LOD,


the cylindrical plasma was investigated.


Figure


-1 shows the signal obtained for Si using the spherical and cylindrical plasmas.


strongest Si emission line in the region from 240-260 nm is at 251.6 nm.


at 250.69, 251.43, 251.92, 252.41


Silicon peaks


and 252.85 nm are also observed in the spectra.


the spectrum

observed.


obtained


The sample


with

was


the cylindrical

an aqueous s(


plasma,


3lution.


a strong ca

The carbon


trbon


emission line is


is from


graphite


electrode.


Graphite electrodes require replacement after 1-2 hours of operation, due to


erosion caused by the cylindrical plasma.


If the spherical plasma is formed, the graphite


electrode required replacement after 6-8 hours.


Results obtained with


the cylindrical


plasma,


versus the spherical


plasma, are


superior for the analysis of Si.


Based on Figure


, the signal increased by


350%


using


cylindrical


plasma.


Background


noise


increased


200%,


based


on the


standard deviation of 11


measurements of a blank solution.


The background emission


level increased by


300%,


with the signal-to-noise ratio (S/N) increasing 140%.











































I-r"
o.t


O
cc

- '

NOC










47


















E

\ ^?
E -Ce) --- --
\ I '



a)
1

-c
a .





I n






i >
I














S
I


















icoo
\ I0

IOO





o-- _0_ 0 0- i
00-- '.0 &

(s---q ^~rq~ ps01u uoss








cylindrical plasma was formed by increasing the plasma gas flow rate from 4 L min-


10 L min'


Hydrogen


150 cm3


min -


, was added to the helium plasma gas to


prevent the spherical and cylindrical plasmas from melting the torch.


At the time of this


experiment, conditions were not optimized for the cylindrical plasma.


The Si signal, as a function of plasma gas flow rate, is shown in Figures


3-3.


3-2


Plasma gas flow rates between 3.5 and 6 L min-' produced the spherical plasma,


while plasma


rates above


L min-'


produced


the cylindrical


plasma.


increase in the signal and background occurred at 7 L min-'


At this plasma gas flow


the plasma changed


from spherical


to cylindrical in shape.


The change from a


spherical plasma to cylindrical plasma repeatedly occurred in the region of 6-8 L min'1


Several additional observations were noted.


Background levels between 200 and 700 nm


were higher in the cylindrical plasma than in the spherical plasma.


Standard deviations,


measured


spherical


a blank,


plasma,


were


respectively.


10 counts


silicon


counts


signal


in the cylindrical


increased


plasma and


with an increase


plasma gas flow rate.


Previous researchers have reported an increase in the silicon response,


chromatograph-microwave induced plasma (GC-MIP) system,


in a gas


with increasing plasma


gas flow rate.


114,115


This behavior is not unique to Si.


Estes et al.114 reported that 19


elements exhibited this response in a GC- MIP with a plasma gas flow rate between 40


and 800 mL mint'


The increase in Si signa


in the MIP,


was also accompanied by a


decrease in the background emission from the quartz torch and the carbon emission from













S0
00o
d -O
OO cv


"N li


-3
C cti
"d
Se-




C.)
'- O

Es
C^) D.


*a-





oc,


cRo
*-
^ *:



















-=I-o
OC
-0-c







0 c



ci
(4 '-
-<- ( dL


^ 5 '

3-
r3 C -
u, E
E ` '
S ^3 -

















































---4
I
I
I


(sl!un fj JJ!qJv) rl!suoluI uoIsstIIU
















D E
d


co
\-0
* r
co
oo a
No


c: 3^

0a ,


3td E

-&-S
0 *
00 4
O k
O .C




































































































I I I


N/S


---


U





'U





.
I
5
I .


-'^0


I I









in the background signal with increasing plasma gas flow rate.


Increasing the plasma gas


flow rate decreased


the residence time of the Si atoms in the plasma.


A decrease in


residence time limited the time Si atoms have to react with 02,,


plasma.


which is present in the


Oxygen may be present as a contaminant in the helium or hydrogen gases, or


from entrained air that surrounds


the plasma and


CMP.


Figure


-3 gives the S/N


versus plasma gas flow rate.


The results with the spherical plasma show an increase in


the S/N with increasing plasma gas flow rate.


The cylindrical plasma gives a constant


S/N with plasma gas flow rates between


7 and 11 L min-


Future sections will discuss


the optimization of power, height, and flow rates.


A problem with the cylindrical plasma was flickering or instability


. The top of


the plasma, 5 mm above the electrode, constantly moved sideways over a 1.5 cm range.


Two experiments


were


performed


to decrease


flickering.


A "tandem"


electrode


arrangement is initially examined, followed by changes in the quartz torch configuration.

The results reported in the rest of this work will deal with the cylindrical plasma, the

spherical plasma will only be mentioned for comparison purposes.


Tandem Electrode


Previously,


electrodes


have


been


placed


above


plasmas


increase


stability. 116,117


In direct current plasmas (DCPs), initial instrument designs consisted of


producing a plasma by passing a de current


between two electrodes,


placed at a 300


angle.60


The plasma appeared as an inverted "v."60


Dekker117 improved the stability,









electrode. 17


The third electrode was placed above the two electrodes,


producing an


inverted


shaped


plasma. 60


Chan et al.


116 positioned a


water cooled,


grounded


electrode above a helium-ICP (He-ICP).


This tandem electrode localized the helium


discharge and indicated a better energy transfer through capacitive coupling.116


LODs for ]

electrode.

respectively,


Br,


Similar


Cl, I, and S were reported for the He-ICP with and without the tandem


Power levels were 1500 W and 500 W for the He-ICP and tandem He-ICP,

y. No explanation was given why the LODs did not decrease or why higher


power was not used in the tandem system.


In the research


for this work, a graphite


electrode was positioned above the cylindrical He- CMP to investigate the effects on the

silicon signal and S/N.

Figure 3-4 gives the spectra obtained with and without the tandem electrode.


In comparing the results,


the tandem electrode increased the Si signal by 200


problem


is the noise increased


over


increase


in the S/N


or LOD


observed.


The distance between the bottom electrode and the top, tandem electrode, is


varied


from


mm


to 30 mm.


observation


height,


position


where emission


observed, is varied from 4 mm to


mm.


Visually, the stability of the plasma increased


with the tandem electrode.


If the tandem electrode is moved horizontally,


off-center, the plasma remained coupled to the electrode.


The signal increased because


of increased


energy


transfer as


indicated


through


the coupling of the plasma


to the


tandem electrode.


At this time, there is no explanation why the noise increased.


LODs with and without the tandem electrode are


ug mLU' and 0.8 zg mL-U
















S
C)
"a



03
C:

0
S-
C)



4-
r:





*-c
'C







C)

-e-


'C

C)
1-
4-.-
h















C'..


0
a.


U



















coo

'-4
bO
C
b)



cr



4-1
-C



9''
V1 b


n

be-
LL









56






















.m--
in








ins






In
Crl



0





Cl

f~lt
m






E


N-


dC
icr



lCi

0

In










o 0 0 0 o 0 g
So o 3 3 o
~ : : ^ = : = O^-























on 0 0 -
00 N en C


(stiiuq~~ Jr q^)&iUlU ospJ








respectively.


Since the addition of a tandem electrode did not benefit the LOD, it was


not investigated further.


Electrode Depth


The effect of increasing the depth of the electrode, inside the quartz torch,


investigated.


The "depth"


of the electrode is the distance between the top of the torch


and the top of the electrode.


A depth of zero mm indicates the electrode is level with


the top of the torch.


A depth of 10 mm has the electrode


10 mm below the top of the


torch.


The results for the Si signal and background versus depth are shown in Figure

The Si signal is measured at a height of 12 mm above the electrode, corresponding


to the optimum


height


emission.


Results


of Si


signal


versus


height will


discussed in the following section.


Figure 3-6 gives the S/N versus depth.


show an optimum signal and S/N at a depth of 6 mm to


10 mm.


The results


A decrease in signal


and S/N is observed at 13 mm and 15 mm, because the signal is transmitted through the


quartz torch.


The transmission of the quartz is less than 100%, due to the characteristics


of the quartz and previous etching from the plasma. When the depth is greater than 12

mm, the signal is also observed 1 mm above the torch. Here, increasing the depth from


12 mm to 16 mm did not improve the signal or S/N over the previous results.

of the cylindrical He-CMP, with a depth of 10 mm, is given in Figure 2-4.


A picture

Visually,


as the depth is increased to


10 mm, the diameter of the plasma decreased by


1-4 mm.


An increase in the intensity is also noticed, which appeared to indicate an increase in the


was














C)


a)
E -0

a.C
1O

tn 2
. *
Or


O I
do
00 X.
NO0









59


































-- O0
C.l















c-











(uI 1 -
/ E,



I C























So o o 0
o 0 0
0 0 0 0


(sIUn XIinsurcuv uoSssSIiW























c4-1
-c
C

c-i
P






*
C
(U



C


U,

* -
E







N
1r
0~



S-J
1d,
c3
C
(U




4-1



$-U -n
O -
-C)




o -5





clo

*0 *
e)



V O
;dt+






=3
b0
'










61









Uo



























-00
\--
-h


















cooob
NC
/ -o




























0n I3 n C,
*l Cl
\I> N/






62




Spatial Profiles


Spatial profiles of the Si emission distribution are measured to locate the regions


with maximum signal and S/N.


Spatial mapping has been performed on the DCP'18 and


ICP.119'120 Blades and Horlick'12 studied the emission profiles of twelve elements, while


Kawaguchi et al.119 examined Fe, Cr, and Mn emission profiles, in the Ar-ICP.


Figure


3-7


gives the signal and S/N versus height above the electrode.


The optimum signal


and S/N are located approximately 11 mm to 14 mm above the electrode.


The curve for


S/N, Figure 3


is obtained using a polynomial fit.


Data for the S/N, in the region of


- 14 mm, is constant.


Above


12 mm the signal decreased rapidly.


The S/N did not


decrease as rapidly as the signal, due to lower noise levels exhibited at greater heights


plasma.


the depth


the electrode


is varied


from


mm


to 10


significant change is observed in the vertical profile of Si.


horizontal


distribution


signal


was


obtained


for the


cylindrical


He- CMP.


The vertical height was 12 mm above the electrode.


Figure 3


shows the


horizontal


distribution.


x -axis


gives


displacement


from


the center


plasma.


The value of zero mm is obtained by visually observing the plasma silhouette,


on the entrance slit of the spectrometer, and aligning the center of the plasma with the


A relatively constant Si signal is observed at +


1.5 mm.


Occasionally, the data


gives skewed results, giving a greater Si emission (20%) on one side of the electrode.
















































4.*3
r: /3









Signal-to-Noise Ratio


(poz!IUwJON) leug!s uon!!S
















































cn)









Signal-to-Noise Ratio


(poZIIEtUJON) It3uI!S uoZITS












Effect of Power


A plot of signal and background, versus power (W)


is found in Figure 3-9.


signal increased from 450 W to


000 W.


Above 1000 W


, the signal appeared to remain


constant.


Power levels greater than


the filament plasma.


Figure


1200 W are not investigated due to the formation of


The background level increased between


10 is a graph of S/N versus power.


700 W and 1000 W


An optimum S/N is obtained in the region


of 700-975 W


. The decrease in the S/N above 1000 W is due to the signal remaining


constant while the noise level


increased.


The results are in contradiction to those of


Masamba and Winefordner.68


Masamba and Winefordner68 used the spherical He-CMP


operating at 6.5 L min~l He and 300 cm3 min-1 H,.


Their results showed the signal for


Al and Mn


increased


power levels between 600 W


1000 W


. Power levels


above 1000 W were not investigated.


The S/N for Al also increased in this power range.


The results for Mn also showed the S/N increased with power up to 900 W


. Values for


the Mn S/N at 900 W and


000 W are similar,


which suggests a plateau.


This would


give results similar to those of Si above


000 W


Electrodes


Research was also performed on materials for electrode fabrication.


In CMPs,


the plasma


is formed on


top of the electrode.


One disadvantage of the CMP is the



























C
I-
C)
0
aM

0
0
*-

P3


S


0
0
0



U,

i+,
o
cU






*
'S
c








I)
C
r-






0
-n







V




CO
U,/




Ca
*-
o
ij
-a
c
13

0\
Cc)




ba
r-









69














C
10











\-o
CC
0
3 .*
05
S



\ )
6





CC
00
\^ **



fk*
tZ

-0 0





S0
aa
















O0











0
o a 0 0





0 \ 0 0
















Inv 0 In0 n 0 In
C~J tt
H^ *



-0












0
() io-







0> 0 0 0 0 0 0
a /) r o on

























































to
Sa

SE

0
ze




=3
b1-
&0
LL











71

















0O





aa








-.
0 0







o
0

U
((A














rct
W0
4





0*












C)
O



00~












U* 0

OO





O





O .0
rU










N S
f
fB
e






0
























0









o 0 0 0 0 0
0 In 0 0 In
--

N/









carbon emission from the graphite electrode is visible.


Electrodes were fabricated from


graphite (G),


titanium (Ti)


and tungsten (W).


Previous electrodes used in CMPs include


raphite,61,'68 Ta,64,121 and a tungsten wire.70


Graphite electrodes offer the advantages


of being cheap, easily fabricated, and able to operate at power levels up to 1200 W


. The


emission from the He-CMP, while nebulizing kerosene, is given in Figure 3


-11.


This


figure shows the background being decreased with Ti and W electrodes.


A larger carbon


(247.9 nm) emission is observed with the G electrode. Using the G electrode, carbon

emission is due to both the kerosene and the graphite electrode. The G electrode only


lasted


hours,


as the electrode changed


from


a smooth,


flat surface


to a rough,


rounded surface.


A rounded surface produced a filament shaped plasma.


During the


operation


cylindrical


plasma,


above


graphite


particles


are visually


observed flowing through the plasma.


electrode


has a


lower


background


compared


to the


electrode.


disadvantage of the


Ti electrode included melting above 500 W


. In Figure


Ti electrode is starting to melt and Si emission is observed; the Si is an impurity in the


In comparison to G, the


Ti decreased the background and noise by 40% and 50%,


respectively.

The W electrode also produced a lower background and noise, compared to the


graphite electrode.


The background and noise decreased by 30% and 50%, respectively.


Power levels up to 1400 W are used with the tungsten electrode.


At 1400 W


, tungsten


emission is observed, suggesting that the electrode is being eroded by the plasma.


Power

















t-
C3

O
c:
O
0

* -
I)

rC
CS
U, 1
c
^- p-
-^ I











74








0






<'
*
/





/ I




















**
II
'*











I
I
*I
/















I
I


XS i


















































(sl!U li !qsv 1!suol UOss
------r
0- 0 00
0 0 0 0-0

0 0 00 04
r~l ooo \ ^ 4

(snun~~~~~~ *-~j)^soU osiu








operation, the W electrode was used for over


150 hours at 700-1000 W


. No erosion


or pitting of the electrode was visible.


initial cost ($635.00).


The disadvantage of the W electrode is the high


However, since the electrode lasted longer than the G electrode,


it should be cost effective in the long term.


The advantages of the W electrode,


versus


G, are a longer lifetime and decreased background and noise levels.


Molecular Gas Addition


Molecular


gases


have been used in ICPs'22"3 and CMPs68,70 for spectrochemical


analysis.


The number of studies employing molecular gases (Oz, N2) in an Ar-ICP has


increased over the past several years."'


The addition of a molecular gas to an Ar-ICP


offered the advantage of greater heat transfer to analyte aerosol particles.3'


Mixed gas


Ar-ICPs decompose refractory particles and operate with higher solvent and analyte


loads. 124


In a He-CMP, hydrogen gas was added to prevent the plasma from adhering


to the torch walls.68


Ali and Winefordner70 reported the addition of H2 gas at 100 mL


min1' into a He-CMP decreased the background level and noise.


In the cylindrical He-


CMP, increasing the H, gas flow rate from 0 cm3 min-' to 250 cm3 min-1 produced a


decrease in the plasma diameter and height by


14 mm to 6 mm and 20 mm to


15 mm,


respectively.


ICPs,


adding


a molecular


produced a


reduction


the plasma


size. "'"122,125


The reduction in plasma


has been attributed to the additional absorption


of energy required in dissociating the molecular species. 25


Introducing


molecular


gases


a plasma


altered


emission


signals.








and Hg in a He-CMP.


The addition of H, decreased the S/N for As and Hg, while the


S/N for Cr and Zn remained relatively constant."26


If organic samples are introduced into


the He-CMP,


the dominant spectral


features in the background are from CN and C,
^"^ jhr


molecular emission (see chapter 5).


The presence of 02 reduced molecular emission due


to CN and C,.


In our research,


the effects of H2, N1, and 02 on the Si signal and


S/N are investigated.

Figure 3- 12 gives the effect of the signal for various amounts of molecular gas.


results


are normalized


to the


signal


obtained


no molecular


addition.


Nitrogen, at a flow rate between zero cm3 mint' and 200 cm3 minT1


signal.


, slowly doubles the


The (N,)-He-CMP produced a relatively constant signal in the range from


200-1000 cm3


mrin


. Both 0, and H, produced a curved response in the signal.


Figure


are the


results


ratio


Adding


produced


little


enhancement (180%) in the S/N. Oxygen increased the S/N by a factor greater than two,

although the signal increased over four times. This was because increasing the 02 flow


rate, from 0 cm3 min-l to 300 cm3 min-'


, doubled the noise in the background.


Green


C, emission,


visually observed


when organic solutions are introduced


into a plasma,


decreased as 0, was added.


The possibility exists that adding increased amounts of 02


promote


formation


of SiO,,


therefore,


decreasing


the amount


of Si


atoms


available to produce an emission signal.


The largest increase in the S/N was observed for HI.


Upon the addition of 100


cm3 min-1 of H,, the S/N increased over five times versus the S/N without H,.


The S/N































[ d
C C
C cn
.- OD
0C


a, ~8
J-g
s_ S-r
O rD
v, O
0 -)
cd 0

r3 C
~a
E- ) D
U pC























































(paz!oIuUJoN)


louD!S













-C



1-j
0

0
6d
pi
U
D,


0
Csl

u


I-
a

o
'iC







^0>
'n


S"o








5-7




82
m u

*I S
V] o







0 3







11
on



rC
bDen


^-,
=3
*c r-f
0.











80
















-o











CI.
4














*










*
Im
*
*: 0











U

*v
GO












0
0 0


















CC
0









Q*
60.


I-I-
















-vj C

x 0)
5(5


















o .0 0


0' -
s 1 o rt
















o o 0 o 00
rJ 7 o \D












0G00 0 0 0
0 00*
-
O > N/S -: '
^ *-'
>^ r u ,
S / oo
/ >-.
/ x *
/ 0 < -








"^^0. Oi







^1- ~ ~~~ CN00 \ t(



60- *- *--

Opy /








250 cm3 minl'


, none or very little Si emission was observed.


The H2 makes a stronger


reducing environment, decreasing the reaction of Si with traces of O,.


as an impurity in the He plasma gas and as entrained air.


The 0, is present


Based on the increased S/N,


the addition of H, is preferred over 02 (and N,) for the analysis of Si.


Figures of Merit


The LOD for Si, in kerosene,


was measured using the spherical and cylindrical


plasmas.


LOD


values of


mL-


were


obtained


for the spherical


cylindrical He-CMP, respectively. These values are based on three times the standard

deviation of the blank, also represented as 3o. Using the cylindrical plasma, the LOD

for Si was decreased by two orders of magnitude. In comparison, the LODs by ICP-


OES84 and GFAAS8' are 0.016 p


mL-' and 0.001


pg mL


, respectively.


ICP-MS


gives an LOD greater than


pg mL


-I 86,87


The linear dynamic range of the cylindrical


He-CMP is three orders of magnitude.


Relative standard deviations in the signal were


approximately


Conclusion


The cylindrical He-CMP was applied to the analysis of Si in organic solution.


The depth of the electrode, inside the torch


of the plasma and increased the Si signa


were examined.


was increased.


and S/N.


This decreased the flickering


Spatial profiles of the Si emission


The Si reached a maximum signal and S/N between 700-1000 W


. The








An electrode made from tungsten exhibited lower background and noise levels,


a graphite electrode.

titanium electrodes.


versus


The tungsten electrode also has a longer lifetime than graphite or

The LODs for the He- CMP are over an order of magnitude greater


than those of ICP-OES and GFAAS.


To lower the LOD in the cylindrical He- CMP,


sample


introduction


technique


thermal


vaporization


was


investigated


discussed in the next chapter.












CHAPTER 4
DETERMINATION OF SILICON IN ORGANIC


SOLUTION BY THERMAL


VAPORIZATION


Introduction


The sample introduction technique of thermal


vaporization (TV) is often used


when lower limits of detection (LODs) are required than can be obtained by pneumatic


nebulization.'26,127


Thermal


vaporization


term


whenever


thermal,


electrothermal,


electrical,


or direct


vaporization


is the


source


used


to produce


appropriate free species for spectrometric detection. 128


Ng and Caruso'27


compared an


electrothermal carbon cup vaporizer with a pneumatic nebulizer for the introduction of


samples into an ICP.


The LODs for the vaporizer were an order of magnitude lower


than those of the pneumatic nebulizer. 127


Improved


LODs


compared


to nebulization, are a result of a larger


percentage of the sample being transported into the plasma.


sample reaches the plasma,


In TV, up to


while nebulizer efficiencies are usually less than


100% of the


10%.


nebulizers, the sample is diluted in a carrier gas before reaching the plasma.


are diluted less in


plasma.


Samples


, which increases the atomic number density of the analyte in the


Disadvantages of TV include poorer precision, greater interference effects,


and a lower throughput of samples.83'129






84

In TV, a discrete amount of a sample is deposited inside or on top of an atomizer.


The atomizer may be a cup, rod,


three steps.83


platform, or wire.129


The atomizer is then heated in


Initially, the first step is the drying or desolvation step,


where sufficient


energy is applied to evaporate the solvent and leave a solid residue.


then performed,


An ashing step is


where the power is increased to convert organic material to H20 and


CO,, and vaporize the volatile inorganic components.


step.


The third step is the atomization


Here the power is increased and the sample is vaporized and atomized, producing


an atomic vapor.


The vapor is probed by a source, as in atomic absorption spectroscopy,


or carried into a plasma or flame,


spectrosocpy


where emission is observed using atomic emission


Since the atomic vapor is formed rapidly, a transient, peak-shaped signal


is produced.


Additionally, a fourth "cleaning" step may be used.


Here a higher power


is employed to remove any residue from the atomizer.


Each step is optimized for power


time.


It has


been


established


separation


sampling/vaporization


processes


from


atomization/excitation


process


improves


general


analytical


capability of the system. '30- 3

Thermal vaporization devices have been used with CMP-OES, MIP-OES, and


ICP-OES.


Matusiewicz reviewed sample introduction techniques for MIP-OES and


states


"sample


introduction


to excitation


sources


is one of


most


fertile


fields of


research


analytical


atomic emission


spectrometry."


Hanamura et al.62 used a


furnace vaporizer where the sample is held in a quartz crucible.


heated,


As the sample was


the volatile components were swept by a carrier gas flow into the CMP.








plasma.


Detection limits for H,


for 1 g of solid sample.


O, N, C, Hg and As were in the high nanogram range


Ali et al.65'66 developed an electrode cup system.


The plasma


formed on top of the sample cup.


This offered an advantage in that the sample was


vaporized


generally


directly


require


plasma.


a carrier


Techniques


sweep


such


sample


as MIP-OES


from


ICP-OES


atomizer


plasma. 133- 136


Figure


2-2


gives a diagram of an electrode cup system.


Limits of


detection (3a) were between


10 and 210 pg for nineteen elements in aqueous solution.65


This system was applied to the analysis of coal fly ash and tomato leaves.66


MIPs are also used with


134,137- 140


The characteristics of MIP, such as low


operating


power


small


plasma


diameter,


limit


the ability


plasma


vaporize and atomize solid or liquid samples.139


Sample atomizers include a tantalum


filament,138


tungsten


boat,141


graphite


furnace. 134


Several


reviews


have


been


published on


sample introduction


139,142


Heltai et al.134


combined a graphite


furnace vaporization system with argon


and 100 ng mL-'


and helium-MIPs.


, using 50 uL aliquots, for 13 elements.


LODs were between 0.1


The MIP had a gas circulation


system


that allowed venting of the solvent vapor during the drying and ashing steps,


preventing extinguishing of the plasma.


carried into the MIP.


In the atomization step, the analyte aerosol is


Problems in this type of arrangement included unstable plasma


operation during manipulation of the valves and plating of the samples onto the walls of


the transfer tubes.3'134


Operation of the TV MIP- OES system required a narrow range


of parameters for stable plasma operation.3'134








Atomizers in TV


- ICP- OES include filaments, boats, rods, and tubes made from


metal or graphite. 29


McLeod et al.


129 have reviewed TV -ICP- OES,


with particular


attention to direct sample insertion (DSI) probes and electrothermal vaporization (ETV).


use of ETV-ICP-OES


is generally


performed


modifying


ETV


from a


graphite furnace atomic absorption spectrometer.136 Samples are dried and vaporized into


the ICP.


The ICP then uses its energy more efficiently in dissociating, atomizing, and


exciting the small amount of the sample that was previously desolvated.


ETV connected


ICP-OES


magnitude


is extremely


lower


those


sensitive


and is reported


pneumatic


to give


nebulization.


LODs


two orders


There


several


disadvantages


ETV


-ICP- OES


including


variable


incomplete


sample


vaporization,


formation of


refractory


compounds


carbidess) due


to contact of the


analyte with graphite containers, 3) change in the rate of transport between the ETV and


ICP,


losses


analyte


during


transportation


between


ETV


irreproducible.'13 Even with the limitations of ETV -ICP-OES, it has been successfully

applied to the analysis of sea water,143 biological materials,135 and ceramic powders.130


Another method


TV sample introduction is using direct sample insertion-


inductively coupled plasma-optical emission


spectrometry (DSI-ICP-OES).


Direct


sample insertion is performed by inserting a probe, containing the sample, directly into


the ICP. 129


The probe is similar to CMP electrodes.


In DSI-ICP-OES, the plasma is


"on"


during manipulation of the probe.'29


to within a few mm of the plasma. 129


The probe is raised axially, or transversely,


Sample desolvation and ashing occurs due to the








vaporization


excitation


analyte.


Development


DSI-ICP-OES


performed by Salin and Horlick'44 and Sommer and Ohls.145


Currently,


the research


involves automating the sample introduction assembly. 146


Disadvantages of DSI-ICP-


OES included a shift in the background level following insertion of the probe into the


Precise positioning


of the probe was required to give reproducible vaporization


and excitation conditions. 129


A difference in probe positioning by


mm produced a


change in the signal greater than 10%


automatic insertion process is about 0.5


Reproducibility with computer control of the

mm.146 DSI-ICP-OES has been applied to


the analysis of aqueous solutions, 146 aluminum oxide,148 and nickel alloys.149

The technique of TV-CMP-OES has several advantages over TV-MIP-OES


and TV-ICP-OES.


In TV-CMP-OES, the plasma is formed directly on top of the


electrode cup,


which is holding the sample.


Analyte is not lost due to transportation


between


the atomizer and


the plasma.


Positioning of the electrode does not change


between sample determinations, negating the positioning errors that occur in DSI-ICP-


OES.


After the sample is deposited inside the cup, the plasma is ignited and used to dry,


ash, and vaporize the sample.

to TV- MIP- OES. Disadva


No manipulation of the plasma gas is required, in contrast


stages of TV CMP OES include possible contamination


due to electrode materials and changes in background levels when the microwave power

is manipulated.

In this research, the technique of TV- CMP-OES is applied to the determination


of Si in organic solution.


A comparison is made between electrode cups fabricated from


ICP.'129









ashing


power,


atomization


power on


the silicon


signal.


The addition of matrix


modifiers, such as Mg(N03),, are investigated.


Finally, the LOD for Si by


TV CMP -


OES is determined and compared to those reported by GFAAS, GF-OES,


TV-MIP-


OES, and TV-ICP-OES.


ExDeri mental


The instrumentation is discussed in chapter


The electrode cup system, shown


Figure


-2B,


is employed


research.


In general,


same parameters as


reported for pneumatic nebulization, chapter


are used.


Standards were prepared in


methyl isobutyl


ketone (MIBK),


obtained from


Burdick & Jackson


Laboratories Inc.,


Muskegon, MI.


Kerosene,


used in chapter


was determined to contain silicon in the


100-500 ng mL-' range, using the technique of TV CMP- OES.


Blank MIBK samples


did not


give silicon emission at


6 nm.


Electrode cups are


fabricated using the


materials


given


Table


2-2.


raphite


titanium


cups


are made


laboratory. Thermo Electron/Tecomet (W


mington, MA) manufactured the tungsten cup.


graphite


electrode,


which


supports


cups,


is made


laboratory.


This


electrode has a depression that gives a snug fit and good electrical contact with the cup.

Dimensions of the cup are given in Figure 2 -2B and are similar to those reported in the

literature.66


The optimum helium and hydrogen


flow rates are


10 L min' and


150 cm3


min '


, respectively.


These are the same as reported in chapter 3.


The vertical spatial









-7 and


The optimum height is between


7 and 9 mm.


Increasing the "depth"


the electrode gave similar results to those obtained by pneumatic nebulization, as shown


Figures


3-5


and 3-6.


An optimum


height of


6-8


mm is


found


using thermal


vaporization.


The vertical height and electrode depth used in the experiments reported


in this chapter are


mm and 6 mm, respectively.


Parameters for the photodiode array


include an


integration


time of


also discussed


the results


section,


and an


accumulation of zero.


Data points for the signals are the average of three measurements.


Error


represent


one standard


deviation


(lcr),


based


three


measurements.


Background noise levels (N) are determined by measuring ten consecutive diodes at 2 nm


from


silicon


signal.


Using


nm line,


background


signal


background noise is measured at


= 290 nm.


Due to the transient nature of the Si signal,


the background and background noise are determined using the same spectrum that gives

the maximum Si signal.


general


procedure


for performing


the analysis by


TV CMP OES


is as


follows.


The sample,


10-30 tL,


is added to the sample cup.


Microwave power is


increased to the level where the ashing step is performed.


ashing is performed in one step.

electrode, initiating the plasma.


In our system, the drying and


An insulated wire is briefly touched to the side of the

The ashing time starts from the point where the plasma


is ignited.


Before the ashing step is over, data acquisition is started.


When the ashing


step is finished, the power is manually increased to the atomization power level.


After


data acquisition is over, the power is raised to


1200 W for


10 s, to clean the sample








The parameters


usually


employed


are as


follows:


ashing


time- 10


s, ashing


power- 135 W, atomization power-500 W, and sample size-20 pL.


are used unless otherwise noted.


These parameters


Investigations into the vertical profile, integration time,


ashing time, and ashing power are performed using a titanium cup.


A tungsten cup is


employed


while


studying


effects


atomization


power,


matrix


modifiers,


determination of the LOD.

integration time is 67 ms.


In examining the temporal nature of silicon emission,


Curves drawn to show the effect of atomization power on the


signal


are obtained


using


a computer


program


(Origin,


ver.2.8,


MicroCal


Software,


Inc. ,


Northampton,


MA).


points


the atomization


power


submitted to a polynomial fitting function.


Determination of the Si LOD is performed


using a sample size of 30 ,L, the maximum capacity of the cup.


effect


adding


tantalum


niobium


solutions


to the


graphite cup


investigated.


Impregnation


process


similar


to Miiller-Vogt


Wendl150


This involved pipetting solutions of 1000 pg mL1' of Ta or Nb into the cup,


followed by drying the solution in situ with the plasma.

cup and the analysis performed as previously described.


16 nm, except in the study involving matrix modifiers.


is employed during matrix


285.2


Silicon samples are added to the

Silicon emission is observed at


The Si line at 251.61 nm


modification studies due to interference caused by the Mg


nm line.


Matrix


modifiers


containing


trichloroethylene


are investigated.


Sensitivity is increased with the addition of modifiers.,s


In GFAAS, the elements Mg


employed.








carbide forming elements.' 15,15


In our research,


the modifier is added to the sample


cup, followed by an ashing step.


of 135 W for 15


The ashing step is performed by applying a power level


After a time of 10 s, no solution was visually observed in the cup.


Ashing times of longer than 20 s only removed the Mg modifier.


measuring the Mg 285.2 nm line.


This was observed by


Sample is then pipetted into the cup, followed by


another ashing step and finally, the atomization step.


A cleaning step is performed to


remove the matrix modifier residue. The cleaning step included the addition of 10%

HNO3, followed by igniting the plasma at 135 W. The power is then increased to 1200


for approximately


The cleaning


is performed


three


times,


in order to


remove the previous matrix modifier.

check the cleaning process. An aque'


Magnesium emission at 285.2 nm is observed to


ous solution containing 10,000 ug mL'1 Mg in


HNO3


is prepared


from


Mg(NO3),


*6H,O (Aldrich


Chemical


Co.,


Inc.).


Palladium


solution,


1000


mL-


HCL


was


obtained


from


Inorganic


Ventures,


Fisher Scientific supplied the trichloroethylene.


Results and Discussion


Integration


Time


optimum


integration


photodiode


array


detector


(PDA)


investigated.


Integration times are varied from 0.033


s to 0.33 s.


The integration time


is the time the PDA collects the signal for each scan.


Results for the signal and S/N are


presented in Fiure 4 1.


the integration time increases, the signal increases.


A J


Figure







































&n Z5
c U
bD
rw
* --