A study of the characteristics of an atmospheric pressure microwave-induced argon plasma

r~ay.

1980

A STUDY OF THE CHARACTERISTICS OF AN

ATMOSPHERIC PRESSURE f'.1ICROWAVE-INDUCED ARGON PLASMA

by

S. K. Wong

TECHNISCHE HOGESCHOOL DELFT LUCHTVAART-EN RUIMTEVAARTTECHNIEK

BIBUOTHEEK

Kluyverweg

1 -

DELFT

UTIAS Technica1 Note No. 225

CN ISSN 0082-5263

..

.

.

A STUDY OF THE CHARACTERIST ICS OF AN

ATMOSPHERIC PRESSURE MICROWA~-IIIDUCED ARGON PLASMA

by

S. K. Wong

,

ACKNOWLEDGEMENTS

I am grateful to Professor J. B. French for his guidance and support throughout the course ofthis work. I am also thankful to Dr. C. C. Poon for his invaluable advice, and Mr. J. Leffers for his technical assistance. I would also like to thank.Dr. D. Douglas for many helpful discussions and his careful proofreading of this thesis.

This work has been partially supported under Grant No. A273l of the

·Natural Science and Engineering Council of Canada, and under Grant 3-250-182-10 from Sciex Inc. Toronto, Canada.

ABSTRACT

A microwave-induced argon plasma is produced at atmospheric pressure in a capacity-loaded resonant cavity operated at 2450 MHz and 100 to 300 watts. Plasma parameters such as electron temperature and electron density are measured

spectroscopically under various argon flow rates and power inputs to the cavity. Electron temperatures in the rapge 5400 - 59000_{K and electron densities in the}

range 3.60 x 1014 to 4.80 x 1014cm-3 are measured. The gas translational temperature of the plasma is estimated by measuring the rotational temperature of C2 molecules; this is done by mixing

5%

of'methane by volume with the argon gas. Gas translational temperatures from 3800 to 43000_{K are measured. Also,}

the general characteristics of the microwave-induced argon plasma as a possible ion source for mass spectroeopy are studied.

· ~ .. ' I II III IV VI VII TABLE OF CONTENTS Abstract Nomenclature List of Figures List of Tables Introduction

A Description of the Microwave Cavity A Description of the Microwave Frequency Excited Plasma

Theory

4.1

Measurement of Electron Temperature

and Electron Density

4.2

Measurement of Rotational Temperature

Experiment Equipment and Procedure Results and Discussion

Conclusions Reference Figures Tables Appendix A Appendix B Appendix C iv PAGE iii v vii vii 1 2 3

4

4

6 7 10

17

18

AUR. B BuR. BR.u c E(v, z) e E. ~ E u fR.u gi ge gu gR. h -h

I(v,

z) JuR. J

J(À)

k(V, z)

L(v)

k I,C r K' NOMENCLATURE

transition probabi1ity for spontaneous emission from u ~ R. rotationa1 constant of a diatomic molecule

transition probabi1ity for stimulated emission from u ~ R. transition probabi1ity for stimulated absorption from u ~ R. the speed of light

emission coefficient e1ectronic charge ionization potentia1 upper state energy

absorption oscillator strength

statistica1 weight of the ionized ground state statistical weight of the electron

statistica1 weight of upper state statistica1 weight of lower state p1anck's constant

h/21T

-2 -lQ-1 spectra1 radiance in microwatt cm str A

integrated spe ct ral 1ine intensity total angular momentum of a molecule

.

spectra1 radiance of the standard tungsten lamp absorption coefficient

1ine profile function

Bo1tzmann constant, propagation number Bo1tzmann constant

m e N u N e N(T) . N·. ~ N e N

~

.;.:

.

·

IR

UR.12 RuR.

C\)

R

0.)

.

sO.)

w

z

Z(~)" .. ). ... u

n

mass of an electron

upper state density of the transition u e lower state density of the transition

total part iele density over all energy states ion density

electron density

total numher of molecules rotational state sum

electronic dipole transit ion probability between stages u and R. signal measured by the photomultiplier in ampere

signalof the standard lamp measured by the photomultiplier sensitivity of the monochromator and photomultiplier

spectral slit width, A

geometrie depth of the plasma, cm partition function

wavelength ,

î

frequency, hz

solid angle, steradian

'

.

I

..

' Figure No. 1 2 3 4 5 6 7 8a-R. 9a-R. 10 11 12a-c 13a-c 14 Table No. 1 2 3

4

List of Figures

A sch~matic diagram of the resonant cavity. The equivalent circuit of the resonant cavity. ,A schematic ,diagram of' the experimental set-up.

Sensitivity of the manochromator-photomultiplier system as a function of wavelength.

A C2 rotational spectrum.

Pictures of the argon plasma jet. Energy level diagram for ArI.

Relative line intensities of the ArI lines. Upper state density of excited argon atoms. Electron temperature.

Electron density.

Peak line signalof ArI 4158R.

Relative line intensities of C2 rotational spectrum. Rotational tempe.rature for C2 molecules.

List of Tables _;

Atomic data of ArI spectral lines. Spectral slit width.

Rotational temperatures from the C2 Swan system emission. Measured values for the electron temperatures and electron densities.

I INTRODUCTION

Inductive-Coupled Plasmas (rcp) are becoming popular and are being used as

atomic emission spectroscop;:c:sour.ces for routine element al analysis of

'" 'almost any type of materials • They can provide high accuracy, high detection

limits, relative: ,freedom from matrix effects, and are capable of multi-element

analysis of solutions ahd solids (after disolution). However, Iep has two

,major drawbacks; -4:t has a' high power requirement, up to 5 KW, and high 'caprier

gas consumption (as plasma supporting gas). On the other hand,

Microwave-,Induced Plasma has a low power requirement (50W - 200W) and a low gas

consump-tion (about 0.5 litre/min). These two advantages make MIP potentially very

attracti ve economically,. 'inparticular the analYs'is of small amounts of sample.

Also, less bulky and lesseXpensive equipment is needed (i.e. power supply), and a small scale system is possible. Hence, MIP occupies a singular position

among various plasma 8'Ouree·s- (ICP included), and its application in

spectral-chemical analysi~s'·has beewn' extensively studied.

Atmosphe:ri'c pressure MIP plasma produced by various types of resonant .'

cavities is reported to háve- excellent analytical performance. C.I.M.

Beenakker (16), using acylîndrical wave guide type of cavity, reports that an

atmospheric pressure MIP wi~h applied power varying from 20 watts to 200 watts

has found application as a;'-versatile selective optical emission detector in gas

chromatography. K. Fallgàtter et al (27) reports

4

àn excitation temperature of

~9out 50000_{K and an} e1.~~!~QR density of about lOl cm-3 are obtained from an

'atmospheric pressure argon plasma produced by a tapered rectangular cavity at

100 watts. The encouraging and promising reports like these found in the

litera-ture provided'the' motivatiah 'for the work presented here. In subsequent work

this MIP sOurce will be studied as a candidate ion source for elemental analysis mass spectroscopy.

A capacity-loaded coaxial line type resonator is used in this work (more

detailed discussion on the resonator is presented in next section). Argon, as

a plasma supporting gas, is found to produce a very stabIe plasma at atmospheric pressure. Besides the low power requirement and low carrier gas consumption, the

other advantages of a microwave-induced plasma are:

a) it can be operated without internal electrodes; therefore

contamin-ation of the gas by met al vapor can be avoided and gas absorption is

reduced;

b) a high tempe~·s.t<ure.Yf-àame is acquired from an electrical di s charge , ·thus

eliminating the ,potential danger of.explosion generated in an open

flame by combustion process;

c) interference from dissociation equilibra, a major source of t rouble in

flame spectros'cöpy is largely reduced in high temperature plasma

sources;

d) nebulized liquids or finely ground powders can be introduced into the

excitation region easily.

The high temperature obtained in an argon plasma is also very attractive in

a mass spectrometric ion analysis system since a more complete ionization of the

detecting trace elements.

In a plasma, the state of the plasma is basically characterized by the

electron temperature and the electron density. Under local thermodynamic

. equilibrium conditions, the population of a particular atomic level is given

by a Boltzmann distribution. The fraction of a species which is ionized is

... gi ven by the Saha e quat ion , and the plasma is characterized on a microscopic

-level by the electron timperature.

Since the mass of an electron and the mass of an argon atom differ greatly, energy transfer between the two species upon collision is very

inefficient. The energy of the electron~ is dissipated mainly byelastic

collisions with the argon atoms. The amount of energy transfered from a free

electron to an argon atom upon a collision is proportional to the ratio of the

mass of an electron over the mass of an argon atom. A complete energy exchange

at atmospheric pressure and at a root mean square velocity of the electrons at

60000_{K takes about 5 x 10-7 second (I). At the same time, free electrons}

continuously acquire new energy in the electric field so that the system will

approach a state of equilibrium in which the electron temperature is far above

the gas temperature. Nevertheless, the gas temperature is an important qua.ntity

in determining the applicability of the microwave-induced plasma, i.e. a high temperature is needed for thermal dissociation and thermal ionization of

mole-cular trace analytes. To measure the gas temperature , a

5%

mixture by volume of

methane gas is added to the argon gas that passes through the cavity, and the

.. _{C2 rotational temperature} . is measured from the relative intensities of the

rotational lines. The rotational temperature would represent the true gas

temperature if either the excitation is strictly thermal or is of such a type that

it does not affect the thermal distribution.

To characterize the atmospheric pressure argon plasma, the following

. quantities are determined -by-spectroscopic methods:

a) electron temperá.~'ilre;

b) electron density;

c) rotational temperature C

2 molecules.

II A DESCRIPTION OF TEE MICROWAVE CAVITY

The re sonant cavity used in this work is a cylindrical coaxial-line type.

It is designed and developed by Dr. C. C. Poon of the Hypersensitive Trace

Analysis group at the UTIAS (unpublished work, UTIAS). The basic construction of

the cavity is shown in Figure I. The cavity is energized by a magnetron (a

microwave frequency oscillator) operated at 2450 MHz; the field inside the

cavity is excited by a magnetically coupled loop antenna. A st rong local

·· electric field is generated in the gap between the center post and the opposite

... ''conducting wall (2). This type of cavity is also referred to as a capacity-loaded coaxial-line resonator. To a first approximation, the cavity may be considered as a short length of coaxial line that is resonated by a large lumped

capacity at the end of the center post (3). An equivalent circuit of the cavity

is shown in Figure 2. The usefulness of this model will be discussed later.

2

The resonator conditions for the cavity are discussed in reference 2 and

3. In terms of the cavity's dimensions, the resonant wavelength,

ÀO

in the

cavity is given by,

,oo[

J

~

Ào = 2iT

R.~~1

R.n

~

where R.O is the spacing betweeri the two circular parallel walls, rl is the radius

of·the center post, r2 0is the radius of the cavity, and d is the spacing of the

small gap. However, this calculation neglects the additional equivalent capacity of the fringing field at the end of the center post and hence always

gives aresonant wavelength that is smaller than the true value. A "fudge

factor" by which the wavelength given by the above equation must be multiplied

to give the true value is of the order 1.25 to 1.75. A quick calculation of ÀO

using the dimensions of the cavity used in this work found that the ÀO obtained

multiplied by 1.25 would give about 12 cm., which is roughly equal to the free

space wavelength of the 2450 MHz microwave supplied to the cavity. Hence, a

fudge factor greater than 1.25 must be used to obtain the true resonant

wave-length since the wavewave-length of an electromagnetic wave inside a conductor

wave-guide or resonator is always greater than the free space wavelength. Moreover, there is a further complication in determining the resonant condition. The presence of the tuning screw and different types of dielectric media (i.e. air, quartz and partly ionized argon gas) in the cavity causes the resonant wavelength to deviate further from the above calculation. Thus, in practice, the cavity is made to resonate by keeping the reflected power from the cavity to a minimum.

That is to say, the loop impedance of the cavity is zero at resonance. As seen

in the equivalent circuit in Figure 2, this meens the short circuit impedance as

seen across the capacitor must be equal to the capacitance reactance, i,e.

1

jZO tan SR. + - -_jwc

=

0

o

as given in reference 2.

The minimum reflected power can be obtained by a combination of adjusting the spacing between the two parallel walls, adjusting the gap spacing between the center post and the opposi te wall and fine tuning by the tuning screw. The

reflected power can be tuned down to less than 10% for all operating conditions.

III A DESCRIPTION OF THE MICROWAVE FREQUENCY EXCITED PLASMA

A microwave frequency discharge is produced by applying a rapidly

alter-nating electromagnetic field to the argon gas in the quartz tube. The theory of

high-frequency or radio-frequency discharges was developed by ~olstein, Margenau

and Hartmann

14,

5, 6

J.

The mechanism can be roughly summarized as follows: an

alternating electrical field causes the charged particles to move in the field direction. The field and the particle velocity have different phases; the

difference is proportional to the mass of the particles. Ions can be treated as motionless and only the interaction between the field and the free electrons must

be considered. This interaction depends on the ratio of the

A high frequency, the field causes the electrons to perform an oscillatory motion superimposed on their random thermal motion. Electrons drift in the

field direct ion and frequently eollide with the gas particle during a single

field cycle. The oscillatory motion of the electrons and the frequent collisions between electrons and gas paTticles lead to ionization of the gas paTticles. Energy transfered from the electric field to the electrons is continuous as

long as there is gas present in the eleetric field. It is found that no effect of frequency on the spectra of the plasma is noted over an extremely large range of frequencies and pressures

[7].

IV THEORY

4.1 Measurements of Eleetron·~~emperature and Electron Density

The spectral radiance of a luminous plasma can be derived from the one-dimensional continuity equation for the energy density of a radiation field. At steady state, it is given by

dI~U2 zl

₌

_{e:(u, z) - k(U, z)I(u, z)}

₍₁₎

dz

where e:(u, z)

=

~N

A R.L(u) is the emission coefficient 1T u u

I(u, z) is the spectral radiance

k(u, z) is the absorption coeffieient

z is the geometrie plasma depth

u is the spectral line frequency L(u) is the spectral line profile

In a homogeneous, selt' luminous and an optically thin medium, where self absorption is negligible, the solution to equation (1) yields (see Appendix A),

I(u, z)

=

~

N A nL(u)z

Ll-1T U u ... (2)

where h is the Planckis ,~o.ns~ant

N is the upper state density of the transit ion

u

AuR. is the transition probability for the transition u-+-R. L(u) is the line profile of the spectral line

The total integrated line intensity over the line profile,

S b t Ot tu s ~ u ~ng O N u

=

N~T~

Z T gu -E /kT u

~nto

• equa t~on O (3) ,

where N(T) is Z(T) is

J - hc N(T)

uR. -

4TI

zrrr

z

the total particle

e

density the partition function

-E /kT

u

over all energy levels

(4)

gu is the statistical weight of the upper energy level of à transit ion

0°

À is the wave~eng&h -of the spectral line E is the energy of the upper energy level

u

In a LTE plasma, it is assumed that the free electrons and the population of the high lying levels of the argon atoms are in a state of complete

thermodynamic equilibrium. The distribution of the population of the high lying levels of the argon atoms is determined exclusively by collisional processes with the free electrons. The three conditions that describe the

state of the electrons in a LTE model plasma are as follows

(8):

1. the velocities of the free electrons have a Maxwellian distribution;

2. for the bound levels of the argon atoms, the population densities have

a Boltzmann distribution described by the electron temperature, i.e.,

N g -(E - E )/kT

n n n m e

- = - e

Nm gm

3. the degree of ionization is given by the Saha equation.

Thus, if the plasma is in LTE, a straight line will be obtained from the

1 JuR. ÀuR.

plot of og A vs. E using equation

(4);

the slope of the line is -I/kT.

gu uR. u e

The electron temperature can be deduced by measuring the slope from the plot. The electron density can be found from the Saha equation (see Appendix B),

NoN ~ e 3 M kT - N -(Eo - E )/kT e e 2 u _{- e} ~ u h g u

where Ni is the ion density, N is the electron density, Ei is the ionization potential, gi and gare the statistical weights of the ground state ion and of the electron respectively. M, k, h retain their conventional meanings (see

Nomenclature ) •

Upon substituting the relation,

2 2

A = 81T

e uR, M CÀ2 e ge f -R,u g , u

-'i~to equation (3), it becomes

N N

u

- =

(6)

A plot of g u, on a loga,;-i thmic" scale, vs. (Ei - Eu) by measuring the absolute

u

intensities of the spectra! lines should also give a straight line, where Ei is

N

u -(E - E

)/kT

'the ionization potential:"Q.f. neutral argon. The quantity - e i u e at

~ ~ - ~ .

the ionization limit i-s found:by extrapolation to (Ei -

Eu)

=

O. Now, setting

N,i-.=

Ne at the ionizati~n limi~, the electron density is given by

[

.

'

[21TM kj

t t

N

~t

N = g.u' e T ~i)

e ,. ~e- '~f!2 e gu

4.2

Measurement of Rotationaa~' Temperature

. ~ . -. j$ .

The thermal distrlbution óf the rotitional levels of molecules is not simply given by the Boltzmann factor e-E kT. The number of molecules, NJ in the rotational level, J of the lowest vibrational state at temperature, T is

... "given by (9),

(8)

where (2J + 1) is the number· of degenerate levels for a state which has a total

'-angular momentum, J. F( J )hC is the rotational energy of state Jo For the simplest case where a diatomic molecule is assumed to be a rigid rotor,

F(J)

=

BJ(J + 1)

(9)

where B is a rotational constant. The actual number of molecules in the rotational states is given by,

N

=

1L

(2J + l)e -BJ(J + l)hC/kT

J Qr

Qr is the rotational state sum,

Q = 1 + 3e-2BhC/kT + te-6BhC/kT

r + 0 • •

From equation (3), the intensity of an emission spectral line is given by,

hu

J n = or- N A nZ UA. Lj.1f U UA.

A ,the Einstein transiti"on' 'Probability can also beo '-written as,

·U

(10)

(11)

(12)

where

IR

U1/,12 is the probability of an electric dipole transit ion between state u

and 1/,. Substituting equations.(lO) and (12) into (3),

C u

4

J ct em u1/, Q <.

r

e -BJ"(J + l)hC/kT (13)

,.here Cem is a constant depending on the change of dipole moment and the total number of molecules in the initial vibrational level. Cemu4/Qr is very nearly

-1

constant within a moleeular-.baI':l:d, and u is measured in cm

The formula employed by Alder and Mermet

(10)

is used in measuring the

relative intensities of the rotational lines in the P branch of the u(O, 0)

A31f - X31f Swan System of the C

2 molecules. g u C U

4

(K' + 1)2_ - 1 em -BhCK' (K' + l)/kT J = ~~----~~----~e' (14) . ~ (K' + 1)

. where K', the rotational quantum number of the upper rotational state is used

. J(K' +

1)

instead of J. A plot of log . . _ '2 vs. K' (K' + 1) yields two parallel

K'

+

1) - 1

straight lines with slopes

~~~C

0 from which the temperature is calculated. The

reason for two parallel

.

stra~gRt

lines instead of just one will be discussed

later.

V EXPERIMENTAL EQUIPMENT AND PROCEDURE

A schematic outline of the microwave-induced plasma and the emission measurement system is shown in Figure 3. Microwave power is coupled to the

cavity through a coaxial cable-waveguide system. The microwave generator is capable of delivering a power output up to 600 watts. However, in order to avoid overheating the generator and putting too much strain on the magnetron in continuous operation, a maximum power output of 300 watts was used. A quartz tube with inner diameter-of

4

mm and out diameter of 6 mm is inserted through the center of the cavityo Argon is used as the test gas, and the argon plasma is :":,

initiated by a Tesla coi1o A rotatab1e plane mirror with a micrometer drive is used so that any part of'the plasma could be observedo The radiation emitted

~rom the plasma is focused'onto the entrance slit of the monochromator by a

spherical mirror; the mirrbr has a diameter of 10.0 cm and a focal length of 20.0 cm. The image of the plasma is focused at the sagital focal plane of the

'spherical mirror where a -hor-i-Eontal image is in focus. The spherical mirror is located 38.0 cm in front ö1'"-the entrance slit. The optical path length between the spherical mirror and the plasma source is approximate1y 47 cm. This set up

~nsures that the grating inside the monochromator is ful1y il1uminated by the

-~ncoming radiation, thus-giving the best possib1e resolution. The monochromator

us~d is a Heath model EU-700-70. Czerny-Turner with dispersion of approximate1y 20Ä per mmo The entranee -and exit slit widths are set at 100 microns for the argon plasma emission. Tlie grating is driven by a stepping motor, scanning at a 'rate of 0.2X/sec. Behin~the' exit slit, light is focused by a simp1e quartz lens

anto the cathode of a RCk l.P28A photomul tiplier 0 The signal output from the

photomultiplier is amp1if~ed 'by a Keithley model 640 electrometer and recorded bn'a Gould chart recorder.

The electron temperatHTe is deduced from the relative intensities of twelve ArI 1ines in the 4100-4500A range. The argon 1ines in this region are found to be the strongest inthe whole argon spectrum. The particular ArI lines used are shown in Tab1e 1 a10ng with their relevant atomic data taken from W. L.

Wiese [11]. These 1ines are optica11y thin, i.e. there is no self-absorption. This is confirmed in work by'Malone et al and Knopp et al [12, 13]; absorption measurements were made on·the intensities of these lines, and almost no

absorption was found within an experiment al accuracy of ~2%.

If the signal from the photomultip1ier of the monochromator, when set at wavelength À to measure radiation from states u + R" is called RuR, (À), then the

integrated 1ine intensity, JuR, as in equation (3) is given by, RUR,(À)W

JuR, = S(À)

as:shown in Appendix C. RuR,.(À) is the measured signa1 in ampere, S(À) is the sensitivity of the monochromatGr and photomultip1ier iy unit of amp/~ watt cm-2nm-l str-1 , W is called the spectra1 slit width in A and it is independent of À within experimental error.

The sensitivity, S(À) is obtained by placing a standard tungsten lamp of known ,spectral radiance where the plasma source is 10cated. A 4mm mask is placed in

~r.ont of the entrance slit_~~. the monochromator so that the size of the lamp's radiation fal1ing on the entrance slit is the same as that of the plasma's. S(À) is defined to be,

where RO _{(À) is the measured signal from the lamp in ampere and J}O

( À) is the'

known spectral radianee of the tungsten lamp. A graph of S(À) vs. À is shown

in Figure

4.

'

,

,The spectral slit,W±dth, W is determined by using narrow Hel singlet lines

from a discharge tube in several regions of the spectrum. ThS width of these

-lines is about O.07A. Each line is scanned at a rate of 0.05A/sec. Upon

producing a curve of Ru~(À) over a helium singlet line, the resulting trace is

integrated with a planimeter. W, the spectral slit width is given by,

Where Ru~(À0) is the signal measured in ampere at the center line wavelength.

The value for W obtained from each of the five Hel singlet lines is shown in

Table 2. An average value is taken and it is used as the value for the spectral

slit width.

Rut (À)À

Thus, for a LTE plasma a plot of log S(À)gA vs. Eu will result in a

straight line, having a slopè' of -0.625/Te if

,

Eu

is in unit of cm-I. The

electron t emperat ure is gi ven by, I I : : ' ..• : '.

T (OK)

=

0.625/SLOPE

e

The measured absolute intensity of the ArI lines is given by equation (15).

'The upper state density Nu/~ in a radiative transition is obtained by

substituting equation (15) into equation (6),

À3 1

g f _n Z

e JVU

(18)

where Nu is in cm-3 , gl+ is dimensionless, À is in microns Rut(À) is an ampere,

W is in

:X,

S is in amp/ll watt cm-2mm-l str-l -,and A is in cm.

N

The quantity

~

e -(Ei - Eu)/kTe at ionization limit is obtained by

gu

extrapolation to (E. - E )

=

0 in the plot

Nu/~

Vs. (Ei -

~).

ge' the

statistical weight for tRe electron is 2. gi' the stat~stical weight of the

argon ion ground state is

4

since the ground state of the Ar ion is 8;',_2P3/2

state. Hence the electron density from the Saha equation, equation (7), is given by,

N

=

[1.9

28 x 1016 T

i

~i)]2:

e e g u

(20)

The C2 rotational temperature is deduced from the relative intensities of the rotational lines in the P brace of the u(O, 0) A3ng - X3nu Swan System.

Rotational lines with upper rotational quantum number, K', from 31 to 47 are measured. oThe particular band of the Swan system meas~ed is located in the 5120-5170 A region. This band has a bandhead at 5165.2A and it is theostrongest band in the Swan system (14). The lines are scanned at a rate of 0.05A per sec; the slits on the monochromator are set at 25 micron for better resolution. The method of measuring the relative intensities of the rotational lines is exactly the same as that for then<ArI'·lines.

Applying the formula--given in equation (4), the plot of log J(À) (K' + 1)/

~(K' + 1)2 - k] vs, K~(K' + 1) will give two parallel straight lines. The slope, BhC .,...-=~--

-klnlOT _ro 't 1.093 T rot

where B, the rotational constant of the upper vibrational state is calculated to

"-~e 1.7443 cm-l(lO). .:-.• ----

-The two parallel lines obtained in the plot are due to the homonuclear characteristics of the C2 mo~ecules

(9).

The homonuclear molecules display

alternate statistical weights between even number and odd number rotational levels. The alternate statistical wèights give rise to an alternation of intensities

which show up as 2 parallel lines in the relative intensity plot described above. Figure 5 shows the alternation of intensities in the P branch rotational lines.

VI RESULTS AND DISCUSSION

The argon plasma can be, ignited very easily using a tesla coil when the cavity is tuned properly. The plasma is very stabIe and can operate for a long period of time, for exam.p'l~.,f'our to five hours continuously. Moderate

overheating of the cavity seems to be the only problem. It is observed that the plasma is composed of three to four filaments of h~ gas located around the circumference near the inner wallof the quartz tube.

At a high argon flow rate, the argon gas tends to pierce the plasma; no emission is observed at the center of the tube. The filaments of the argon plasma have fairly weIl-'dëf'ined boundaries, and they combined to form a long slender tail flame at the exit of the quartz tube. At a lower argon flow rate, the boundaries of the plasma' filaments become less wêll defined and tend to

spread inward toward the' c-enter; but the main bulk of the plasma is still located near the inner wallof the quartz tube. A shorter tail flame is

formed at the tube's exit. In other words, the plasma produced in this work is very non-symmetrical in the discharge region but becomes much more symmetric in the tail flame region.

A plume or after-glow is also observed to follow right after the plasma tail flame as shown in Figure 6. As the applied power is increased, th~s after-glow grows longer. The plasma tail and the glow tail curve upward as power is increased. A spectroscopie study of this after-glow tail reveals that it is mainly composed of N₂ and ~ molecular bands. The curvature of the plasma tail

~lame and the after-glow-traces out a volume of hot gases which have such low density that they suffer a buoyancy effect. J.D. Cabine [15] suggested that the mean gas temperature can be estimated from the flame's buoyance as

,

.

attempt had been made to estimate the gas temperature of the plasma tail flame;

~ut the results are found to be inconsistent and much lower than expected on the

basis of the spectroscopic measurement. Also, a relative large measurement error is resulted using this methode

A few authors [16, 17] suggested that argon metastable atoms are

res-ponsible for the excitation of the _N2 and ~ species upon collisions. As shown

in Figure 7, the argon metastables are formed when the excited atoms, making transition by spontaneous emission, terminated in the metastable levels. These tnetastable levels haveenergies of 11.50eV and 11.67eV respectivelyo Although no measurement has been made on the concentration of the argon metastable atoms, their population is believed to be fairly large. From the observation of the argon spectrum, many of the strongest lines are terminated in these metastable levels; thus quantitatively, a significant fraction of the excited atoms are

terminated in these meta~table levels.

The second positive"bands of the _N2 emission observed have an electronic

excitation energy of 11.leV, and the first negative bands of the ~ have an

electronic excitation energy of 9.9geV. Therefore, the excitation of these

nitrogen bands is weIl within the capability of the argon metastable atoms •

. Excitation of the _N2 bands by collisions with high energy electrons is also

very probable. Energy transfer to the vibrational levels of the _N2 molecules

by collisions with electrons is very efficient (23); thus the electronic levels

can be excited thermall:y--." --But at the tail flame, the electron density is

expected to be low, an'd collisions with the argon metastables, which have a

relatively long life time, would be more dominanto

It is found that the plasma is best tuned when the end of the quartz tube is just about flush with the wallof the cavity. When the quartz tube is

extended out from the cavity, say 3mm or more, the degree of excitation of the

plasma decreases significantly. The effect is dramatic and can be observed

visually from the phys-ical appearance of the plasmao This phenomenon may be

explained as follows: Consider the short gap region of the cavity as a parallel

plate capacitor, and the quartz tube and the argon plasma within as dielectric

media. (see Figure 2). As,the quartz tube is extended, the quartz, which acts

as a dielectric, causespart' of the electric field to fringe outside the short

gap region; thus the electric field is not as concentrated and intense in the

short gap region, leading to a less excited plasma. The reflected power goes

up also, indicating the resonance condition is affected. However, it is found

that even by tuning the cavity in an effort to reduce the reflected power, the degree of excitation of the plasma is still not as high as when the end of the

quartz tube is about flush with the cavity wallo Thus, the fringing effect of

the electric field in the parallel plate capacitor model seems to be a plausible

explanation.

Viewing the plasma fx~e spectroscopically from the outside of the cavity

turned out to be difficult as there is a lot of _N2 and N~ molecular bands

superimposed on the ArI lines. Interference from these molecular bands makes

the measurement of the ArI line intensities unreliable. As a result, a smal 1

rectangular hole of 8mm x 6mm is drilled on the side of the cavity sö that the emission in the shortgap excitation region can be observed. It is found that the

-argon spectrum observed 'in -the short-gap region is virtually free of

inter-ference from the nitrogen -bands. There is a noticeable background signal in

4278A is clearly detected even though the signal is very small as compared to the ArI lines. The ni trogen emission can be due to small air leak along the argon gas feeding line or a slight impurity in the argon gas.

Only lines belonging to ArI are found; no ionic argon line is observed. This is presumably due to the fact the more highly ionized excited levels of the ArII ions cause them to recombine very rapidly with electrons in

a

three body recombination process, and that the ions which are populated in the ground level do not radiate.

The line intensities of the ArI lines are measured by having the tull

height of a thin slab of plasma laterally focused onto the entrance sli t of

· the monochromator. It is assumed that the plasma along the line of sight has a uniform temperature; therefore, the electron temperature and the electron

densities deduced are average values. It is not possible to obtain a radial temperature profile us~ng~he Abel inversion method because of the

non-, symmetrical nature of the plasma. That is, the plasma is striated, and the Abel inversion method requrres a cylindrically symmetrie plasma.

The plots of log (RÀ/SgA) vs.

Eu

_{and Nu / gu vs. (Ei} - ~) (Figures 8a-l and 9a-l) for various argon· flow rates and power settings g1ve a very good straight line fit with·the uata. The results suggest that LTE is likely to exist in the plasma and show that the population of the upper states of the

. argon atoms have a Bolt~·distribution. Additignal ArI lines at 5l62.29î, 5l87.75î, 522lo27î, 5495.87î, 5606.73Î and 6032.13A are measured for the

Nu/~ vs. (Ei -

Eu)

plots. These lines have upper state energies closer to

the 1onization limit thus giving a better extrapolation for the quantity, Nu / gu ( i) at the ionization' limit.

The electron temperatures and the electron densities deduced are shown in Table 4. Each of the values is an average value from three trials.

The measurement error for the line intensity signal, R(À) is ± 2% and the error for the monochromator-photomultiplier sensitivity. S(À) is ± 3%. However, the temperature deduced is very sensitive to the slope of the plot log RVSgA vs.

Eu;

it has an uncertainty of ±400oK. As shown in Figure 10, the electron temperatures for various argon flow rates and power settings are more or less constant within that 4000_{K uncertainty. It is observed that the}

relative line intensities of the ArI lines remain invariant under various conditions at atmospheric pressure; thus the electron temperature appears to be strongly pressure dependent.

N N

The quantity ~i) extrapolated from the plot ~vs. (E. - E ) has a

gu gu 1 u

rather large error; it is due mostly to the uncertainty of the oscillator

strength, f~u. While the relative values of the oscillator strength for various spectral line transitions are accurate to less than ± 5% (11), the absolute value for each of the spectral line transitions has an accuracy or ± 25% (12). Furthermore, the spectral sli t width, W has an error of ± 7% and the error for the plasma depth, Z is about 10%. Since the plasma is composed of mainly three to four filaments of emitting gases, Z is an estimated sum of the cross-sectional

widths from all the filament S; also, Z varie s wi th flow rates and power sett ings •

N

Hence, ~ i) can have an error as much as ± 30%.

'.

It must be emphasized that the straight line obtained in the plot: _{Nu / gu vs.}

(Ei - Eu) is fitted over the actual data points obtained from experiment. The relative error between all the points taking into account the errors from the relative oscillator strengths, the spectral slit width, and the plasma depth, is only about 12%. On a logarithmic plot, this 12% error does not scatter the ,data points much. Henc&,·the data points falling in a straight line establish

the existence of a Boltzmann distribution of the population in the upper

~nergy levels. What the~O~ accuracy means is that the straight line with a

·f'ixed slope can be shiffted-up and down such that the extrapo~ated value for

~u/~ (i) has an accurae·y. of ± 30%. Therefore, the electron density calculated

. from equation (20) has an--·error a bit more tItan ± 30%. Hence, the values for -;the electron density are of the order of 101cm-3 as shown in Figure Ilo

Figure 11 also displays-oer "tI-rend of increasing electron density with increasing

. power and flow rate.~' . '

AI! interesting obser~tion is made on the ArI line intensities. The line intensity .signals meas~d·~ the electrometer increase very little as the applied power is increased ~r a given argon flow rate; but the line intensity signals increase noticeably with increasing flow rate for a given applied power. Figure l2a-12c show how ·the line signal for a particular line at 4l58Ä varies with different power settings and flow rates. The set of fiY-e data points on each .

power setting indicates thespread of the line signal recorded on fiv.e

different sets of expe1"iment collected over a period of two months. The spread 01' the line signal at a; given power could be (and is likely) due to small

·variation in the argon :f!l:oW"Tate over the period of two months. The observed behavior of the line signal-in response to different power inputs and flow rates can be interpreted as fol~~: As the power is increased, the electron density increased. As the elect-ron"density increased, the degree of collision

domination increased; hence the population of the excited levels approaches an equilibrium value. Ohce collision domination is achieved, the population does not change with power; hence the line intensities t.end to change very little as power is increased. As the argon flow rate increased, more argon atoms are being excited in a given period of time; thus the population of the excited levels increased, leading to greater line emission signal. A rather important deduction can be made from the line signal data presented in figures l2a-12c; they show that the plasma produced in the capacity-loaded coaxial line cavity has a stabIe characteristic and is very reproducible over a long period of time.

With a 5% mixture of methane seeded into the argon plasma, it i·s observed that the CH4-Ar mixture produces a blueish flame, forming an annulus around a typical argon plasma flame. _{C2 emission},·is ~observed only inside the quartz tube where the blue annulus appears. At the exit of the quartz tube, the blue

annulus disappears abruptly. A spectroscopic study of the flame away from the quartz tube's exit reveals no C emission. Furthermore, by increasing the methane concentration to ab out 7-8%, the annulus gives off an intense blue light substantially brighter when observed byeyes; the C₂ rotational line intensities are about 1.2 times as strong. A further increase in methane concentration would extinguish the plasma. When the methane concentration is decreased to approximately 2-3%, the blue annulus becomes very soft in color, and the C2 rotational line intensities are ab out 0.7 times as strong. Thus, it :a,ppears that the C2 emi.ssion is coming from the blue annulus.

According to Gaydon and Wolfhard [18], C2 is formed from the breaking up of tbme of the relatively complex molecules, such as methane; the formation of these molecules depends on a rapid chain of polymerization process. It is very

difficult to suggest a plausible mechanism for the formation and liberation of

. energy using only simple molecules or radicals. The mechanism for the formation

of the blue annulus wher_{e C2 emission is thought to occur is not understood.}

Apparently, the formation o_{f C2 mole}cules is favorable only near the wallof

. the tube; further inwarà i;-eward the argon plasma, the C2 molecules could be

dissociated upon collisions with the argon metastables where their

concen-tration could be much higher. C2 has a dissociation energy of 6.25eV;

whereas, the argon metastables have an energy of 11.50 and 11.67eV

respect-ively. At the exit of the quartz _{tube, the C2 molecules} could react readily

with oxygen which is available in abundance, thus quenching the radiation (19).

The relative intensities of the rotational lines are measured from the short

gap excitation region •. With the entrance and exit slit widths set at 25

microns, the resolution between the P lines is very good. However, there is

s~me overlapping between ~he P branch and the R branch of the rotational lines.

The interference is noticeab1e especially in the region where the P branch lines

have high rotational quantum m.nnbers; the intensities of the R branch lines .'

-a-re comparable to those irr--the P branch as seen in Figure 5. For those P

branch rotational lines with quantum numbers 44 to 47, the pe~ to peak

·separation between a P·line and an R line is about 0.6 to 0.8Ä. Having the

slit widths setoat 25 microns, the half base width of a rotational line can

be as much as lA. Hence overlapping could cause some measurement errors in

the line intensities. In the region of lower quantum numbers, the intensities

uf the R branch lines are considerably less than those of the P branch; the

degree of overlapping between the P and R lines is not known. The separation

between a pair of Pand R lines is not uniform throughout the band. Thus,

there can be an error of 5-10% in the measured relative intensities of the P

branch lines used in the calculation of the rotational temperature.

Despite the errors in the measurements of the relative intensities, two

parallel straight lines are obtained in the plot of log [J(K' + l)/(K' + 1)2 - 1]

vs. K'(K' + 1) as shown in Figures 13a-d. The interference from overlapping of

the R branch lines is probably very smallor the amount of overlapping is about

the same for all the lines. This claim is justifiable from the display of

alternating intênsities in the P lines which is clearly shown in Figure 5; it

shows that the amount of overlapping has not been so serious as to destroy the

alternating intensity characteristic. The two parallel straight lines obtained

could not be a coincidence; many trials with various flow rates and power

setting are tried, an~ the· r esults are consistent.

It is found that the argon flow rates do not affect the rotational

temperature deduced within experiment al errors. The rotational temperatures of

the C_{2 molecules at various} 'power settings are shown in Table 3 anër are

graphed in Figure 14. Each of these temperatures is an average value from three

trials. It appears the rotational temperature at 130 watts is a little

higher than the one at 185 watts from figure 14. One would expect that the

temperature will increase in increasing power. The measurement error can offer

an explanation here. Since the measurement error really amounts to the error

of the slope of the straight line drawn through data points in the plots as

shown in figures l3a-d. A slight change in the slope of those straight lines

corresponds to a large change in temperature derived (approx. ± 2000_{K). Now,}

the two temperatures at 130 watts and 185 watts differ by only 23°K; thus the

measurement error can change the lower power end of figure 14 easily.

Herzberg [9] stated that the rotational temperature obtained in this method (Le. J a exp(-B~(~' + l)hC/kT)N_J represents the true equilibrium temperature only if either the excitation is strictly thermal or of such a type that it dóes not affect the thermal distrîbution. The two parallel

straight lines obtained in each plot suggest that thermal distribution probably exists. The process of thermal excitation of molecules is due to collisions with energetic particles (18). In this case, the _{excitation of C2 molecules} is

most likely due to collisions with energetic electrons. The radiative life time of an excited C2 m~lecple is approximately 8 x 10-9sec ; gas kinetic calculations reveal that a molecule with normal cross section at atmospheric pressure and at a temperature of 33000_{K will make ab out 3 x 109 collisions}

per sec (18). Therefore, a C2 molecule with a radiative life time of 8 x 10-9 sec. will make 24 collisions before radiating. This number of collisions is

.. sufficient to cause equilibration of the translational and rotational energy, since an average of ab out 10 collisions is enough to establish a Boltzmann distribution. The collisional cross section of argon atoms is calculated to be comparable to that of the C2 molecules, using the viscosity coefficeint method (20, 21). Hence, thermal equilibrium upon collisions between argon atom and

C2 molecules is likely to have been established, and the C2 rotational temperature should be very close to the argon gas translational temperature.

There are reports that the effective C2 rotational temperature obtained in argon diluted flames at atmospheric pressure is abnormally high (22).

Reabsorption of the emitted radiation within the radiating gases at atmospheric pressure can lead to anomalous intensity distributions, thus resulting in high effective temperature. However, the rotational temperatures obtained here are believed to be quite rea:sonable. First of all, the rotational temperature is below the electron temperature for various power settings. This is expected

since electrons and C2 molecules would expect to have different thermal

distributions. Secondly, from the results of a tungsten wire heating

experiment, there is a good indication that _{the C2} rotational temperatures are in the right range. This experiment involves heating of a piece of tungsten wire in a pure argon plasma flame. At 130 and 185 watts of input power, the tip of a piece of tungsten wire evaporated into a fine sharp needle point. At 240 watts and 300 watts of input power, a piece of tungsten wire simply

melted into two pieces; a molten balI of tungsten can be seen deposited on the melted end of the wire. Tungsten has a melting temperature at 3650oK.

As a general conclusion from the experimental results, it is seen that the atmospheric pressure argon excitation spectrum is relatively invariant under various flow rates and power inputs. The relative intensities of the observed ArI lines remain more or less constant; consequently, the electron temperature

deduced from the relative argon line intensities remains constant for various conditions within experimental errors. On the other hand, the rotational temperature of the C_{2 molecules} is found to increase with increasing power input. Since the electron temperature is affectively a measure of kinetic energy of the free electrons, it seems that the excess increased energy

received by the electrons in the electric field from an increase in power input is transferred to the molecular species. In an articie by Gurevich and

Podinashenskii [23], the energy balance for the electrons has been considered and estimates have been made of the proportion of the energy dissipated

through elastic and inelastic collisions with heavy particles, and through

diffusion and conduct ion to the colder regions of the plasma. The authors show that the energy lost by electrons to the excitation of vibrational level in

molecular gas contributes by far the large st fraction; it exceeds that of the energy transferred in elastic collisions by about two orders of magnitude.

According to their findings, only a small amount, about 1%, of molecular impurities, such as N2 and CO is sufficient to aid the electrons in giving up their excess energy by inelastic impact leading to excitation of the vibra-tional levels of the molecules. Since the plasma source here is an open

system, a 1% impurity of N2 would certainly be expected to present; fUrthermore, . _{the background of the argon spectrum confirms that N2 impurity does exist in}

the argon plasma.

Inert gas, such as argon, which does not have vibrational degrees of freedom, collides with electrons mainly through elastic. collision.; energy transfer between electrons and argon atoms is very inefficient. When a higher concentration of molecular impurities is introduced into the argon plasma, -Le. the addition of methane, the electroIl3would give up their excess energy

gained in the electric field more readily to the molecular species by inelastic collisions. This process may explain why the rotational temperature of the C2 molecules is not affecteà by different argon flow rates at a given power

since electrons transferatheir energy mainly to the molecular species rather

-than to the argon atoms-ö .

The tendency for elèc~ons to dissipate their excess energy through 'inelastic collisions with·c·molecular species and heat conduction is consistent ~th the results that the rotational temperature increases with increasing power. Since the electron temperature is constant, this means that the kinetic energy of the electrons is constant; so the extra energy from an increase in power input must have gone-to the molecular species.

In the case of a ~:argon plasma, even though energy transfer is mainly byelastic collisions and-'"i-s very inefficient, inelastic collisions and heat ~C'Onduction between electl'''otrs> and argon atoms should not be ruled out and can contribute to some degree in the energy transfer process. Mors importantly, molecular impurity, suchTas N2 present in argon, though small in quantity, '_{acts as athermal sink for the electrons, and the N2 molecules effectively}

transfers the excess energy of the electrons to the argon atoms by inelastic collisions. Gurevich and Podmoshenskii point out that even if the molecular impurity is only 1%, the~Tgon gas behaves like a molecular gas. Comparing the values of Te and Tr t for-_{C2 shown in Tables 3 and}

4,

it is seen that the

difference between ~he two values is about lOOOoK, which is in the order of what Gurevich and Podmoshen~kii have found between the electron temperature and the gas temperature in an'~a.tlnospheric pressure argon arc. Hence, if equili-bration of the translational energy of the argon atoms and the rotational energy of the C2 molecules is attained, or if the two species are very close in achieving energy equil·ibration, _{then the rotational temperature of the C2}

molecules makes a very good estimate on the gas translational temperature in an ALMOST pure argon plasma. The tungsten wire heating experiment mentioned earlier seems to support this estimation.

It is noted that the results obtained in this work are from the excitation region inside the cavity. The conditions of the tail flame are not known. The gas temperature might be somewhat higher and the electron temperature might be somewhat lower as the two species have a chance to come to thermal equilibrium. Due to interference of the N2 molecular bands and lack of C2 emission at the plasma's tail flame, a study of the plasma parameters and the gas temperature there is much more difficulto

As noted in figure 14, an extrapolation from the graph indicates that higher temperature can be obtained by fUrther increasing the power. Given that the coupling of elect ri cal field energy to the plasma is reasonably efficient at

still higher power input than 300 watts, a higher plasma temperature is very possible. Commercial microwave power supplies providing power up to 600 watts are recently available at areasonabIe price. A fUrther study of the plasma at higher microwave power input would benefit the evaluation of the performance and applicability of the plasma source since higher temperature (if it is needed) can ionize trace samples more effectively. A more detailed study of the

electric field generated in the cavity and how to couple the electric field to

the plasma more efficiently is also invaluable in improving the performance of

the cavity. Dr. C.C. Poon has pointed out that using aluminium oxide tubing with the orifice of the tubing shaped properly, the dielectric property of aluminium oxide can help to couple the electric field energy to the plasma more efficiently. This might be an important area to study further since it is noted that the dielectrics of the tubing and how the tubing is placed through the cavity have a pronounced effect on the plasma as discussed earlier.

VII CONCLUSION:

The characteristics of the atmospheric microwave induced argon plasma

produced in the capacity-loaded coaxial line cavity compare favourably with other MIP sources reported in the literature. The electron temperature

4is in

the 5000 - 60000_{K range and the electron density is of the order of lOl cm-3 •}

The high gas translational temperature of the order of 40000_{K is encouraging,}

since thermal ionization of trac.e samples can be very effective. The cavity

is rugged; no difficulty or breakdown has been experienced during the course of

the experiment. The cavity was often operated for a period of three to four

hours continuously at 100 - 300 watts. No cooling system is incorporated to the cavity. The plasma is extremely stabIe even for long periods of operation. The characteristics of the plasma are very reproducible. This can be seen from the argon line emission characteristics as illustrated in figures l2a-12c and the related discussion presented earlier. In short, the mièrowave-induce

argon plasma produced in the capacity-load coaxial line cavity has a good

potential as athermal source in hypersensitive trace analysis.

REFERENCES

1) P.W.J.M. Boumans, Theory of Spectrochemica1 Excitation. Hi1ger and

Watts, London, 1966.

2) Curtis Johnson, Field and Wave Electrodynamics. McGraw-Hi11, New York,

1965. . . A ;

3) Theodore Moreno, microwave Transmission Design Data. McGraw-Hill,

New York, 1948.

4) T. Holstein, Phys; Rev: , 70,367 (1946).

"5) . H. Margenau, Phys. Rev7, 73,297;326 (1948).

6) L.M. Hartmann, Phys. Rev., 73,316 (1948).

7) O.P. Bochkova and E.-Y·ó Shreyder, Spectroscopie Ana1ysis of Gas Mixtures.

Academie, New York, 1965.

8) R.W.P. McWhirter, PTasma Diagnastic Techniques. Edited by R.H. Huddlestone

and S.L. Leonard, Academie, New York 1965.

9) G. Herzherg, Molecular Spectra and Molecular Structure. Van Nostrand, 1950.

10) J.F. Alder and J.M. Mermit, Spectrochimica Acta, Vol. 28B 421 (1973).

11) W.L. Wiese, M. W. Smith, B.M. Miles, Atomie Transition Probabilities.

Nationa1 Bureau ofStandard, 1969.

12) B.S. Ma10ne and W.H. Corcoran, J. Quant. Spectrosc. Radiat. Transfer,

Vol. 6, 443 (1966).

13) C.F. Knopp, C.F. Cottschlich and A.B. Campbel1, J. Quant. Spectrosc.

Radiat. Transfer, Vol. 2, 297 (1962).

1:4) R. W .B. Pearse and A.G. Gaydon, The Identification of Molecular Spectra.

Chapman and Hill Ltd., 1963.

--.

15) J .D. Cobine and D.A~ Wi1bur, J. of Appl. Phys., Vol. 22 no. 6, 835 (1951).

16-) C.IoM. Beenakker, Spe'!:!t:rochimica Acta, Vol. 32B, 173 (1977) •

.

rn

H. E. Tay1or; J. H. Gi bson and R. K. Skogerboe, Anal. Chem., 42, 876 (1970).

· 18) A.G. Gaydon and H.G. ·Wo1fhard, Proc. of Royal Soc. of London, Series A,

Vol. 201, 570 (1950).

19) E. L. Grove, Ana1yt ïëa~ < Emi ssion Spectrascopy, Part IL Marcel Dekker Inc.,

New York, 1972.

21) F.W. Sears and G.L. Salinger, Thermodynamies, Kinetic Theory and Statistical Thermodynamics. Addison-Wesley, N.Y., 1975.

22) H.P. Broida and H.J. Kastkowski, J. Chem. Phys., 25,676 (1956). 23) D.B. Gurevich and I.V. Podmashenskii, Opties and Spectrascopy, 15,

319 (1963).

24) Lecture notes from AER 2044Y Physics of Radiating Cases. Dr. R.M. Measures, 1978-1979.

25) H.R. Griem, PLASMA PHYSICS. McGraw Hill, N.Y., 1964.

26) F.A. Robben, Ph.D. Thesis. Technical Report HE-150-211, 1963. University of California, Institute of Engineering Research, Berkeley.

27) K. Fallgatter, V. Svoboda and J.D. Winefordner, Applied Spectroscopy,

short gap region antenn a

llllllllllllllllill

c

j L

-~

-~ tuning screw Quartz Tube

-=

Argon 4i microwave input

FIG. 1 A SCHEMATIC DIAGRAM OF THE RESONANT CAVITY.

~---(:~

_________ -1 __

\

,

'-FIG. 2 THE EQUIVALENT CIRCUIT OF THE RESONANr CAVITY REPRESENTED BY A SHORTED COAXIAL CABLE WITH A CAPACrrOR ON ONE END.

MONOCHROMATOR

SPHERICAL~

~

MIRROR ~

Î

FLOWMETER ARGON reflected power coupIer MICROWAVE REFLECTED POWER MONITOR MOVABL~ ,PLANE MIRROR MAGNETRON

-POWER SUPPLY

FIG.

3

A SCHEMATIC DIAGRAM OF THE EXPERIMENTAL SETUP.

HIGH VOLTAGE O.C.

POWER

SUPPLY PHOTOMULTIPLIER

'---11

ELECTROl METER CHART RECORDE

... I _...

...

_..

_1.2 ... I ~ N I Il 0

...

1.1

...

_«l ~ ... 0 .... ~ _1.0 ~ co I 0 ... ~

'"

0.9

..

0.8 400 0.7 ... I ... ...

..

0.5 ... I ~ N I Q

...

0.4

...

_«l ;. 0 ... u .... Il "- _0.3

'"

Il

'"

co I 0 ... ~ '" 0.2 510 410 420 520 530 540 550 ~ (run) 560 ~ (run) 570 580 590 600 610

4

0

•

In .." P. o o N o

.

~

.

. ~ p::; ~ 0

re

~ ti) ~ lf'I

~

H

~

p::; C\I 0 U"\

.

Ö H 0 I%i lf'I ~ lf'I

0.8 SCHF 300 HATTS 0.8 SCHF 240 UATTS 1. 4 SCHF 185 UATTS 1.4 SCHF 130 UATTS 1 SCHF '= 0.39 LITRE/rUN. OF ARGON

Energy (ev) 15. 755r---15 14 13 12 11.67 11.50 11

o

IONIZATION LIMIT 4 -~-I 3p5 Sp 3p5 Ss 3 9 8 7 5 3p5 4p

4 3

₂

-r

(GROUND STATE) .1=0 J=O J=1 J=2 J=O J=1 J=2 J=3

EVEN ODD EVEN

3.6 ;( 3.' Ö' til "-.-<

e

Ö' 0 3.2 .... 3.0 3.8 3.6 ;( Ö' til "-;ij 3.4 Ö' 0 ..-i 3.2 116400 T = 5608°K 130 11 0.8 SCf/F 117000 118000 119000 -1,

FIG. 8(a) RELATIVE LINE INTENSrrIES OF THE Ar I LINES.

T = 5406°K

130 U 1.1 SCHF

3.8 ;( 3.6 t» (/l "-~ t» 0 .... 3.4 3.2 3.6 ~ 3.4 <C t» (/l "-_-< p:: t» .8 3.2 3.0 T = 5363°K 130 \I 1.4 SCHF 116400 116400 T = 5643°K 185 ~1. 0.8 SCHF 117000 117000 118000 119000

FIG.

8(c) Figure 8(d). 118000 119000

FIG. 8(d)

3.8 3.6 3.2 3.8 ~ 3.6 0-til "-.< p:: 0-0

...

_3.4 3.2 116400 T = 5378°K 185 U 1.1 SCflF T = 5301°K 185 \I 1.4 SCHF 116400 117000 117000 118000 119000 FIG. 8(e) 118000 119000 8(f)

3.6 :< 0\ U) "-_-< 3.4 ~ 8' ... 3.2 3.8 3.6 3.2 T = 5416°K 240 11 0.8 SCHF 116400 116400 T = 5409°K 240 ~J 1.1 SCHF 117000 117000 118000 119000

FIG.

8(g) Figure 8(h). 118000 119000

FIG. 8(h)

3.8 ~ _3.6 0> lil

"

~ 0> 0 ~ 3.4 3.2 3.6 ~ 3.4 0> lil

"

.< ~ 0> 0 ~ 3.2 3.0 116400 T = 54030_K 240 1/ 1.4 SCHF T = 5892°K 300 U 0.8 SCHF 116400 117000 117000 118000 119000 FIG. 8(i) Figure B(j). 11BOOO 119000

3.8 3.6 :< 0-<IJ "--< _3.4 D<: 0-0 .... 3.2 3.8 ~ 3.6 <IJ "--< D<: 0-o .... 3.4 3.2 116400 116400 T = 5474°K 300 IJ 1.1 SCHF T = 5475°K 300 U 1.4 SCHF 117000 117000 Figure 8 (k) • 118000 119000 FIG. 8(k) Figure 8(1). 118000 119000 8U)

130W

0.8 SCHP

FIG.

9{a)

_{UPPER STATE DENSITY OF EXCITED ARGON ATOm.}

13 ow

1.1 SCHP

0.5 1.0

•

130W 1. 4 SCHF 0.5 1.0 Ei - Eu (eV) FIG. 9(c) 185W 0.85 SCHF 0.5 1.0 Ei - Eu (eV)

•

•

•

185W 1.1 .sCRF o 0.5 1.0 Ei - Eu (eV) FIG. 9( e) 185W 1.4 SCRF o 0.5 1.0 Ei - Eu (eV) 9(f)

r Î '~ o:? "-" z 0.5 Ei - Eu (eV)

FIG.

9(g) 0.5 Ei - Eu (eV)

FIG. 9(h)

240W 0.8 SCRF 1.0 240W 1.1 SCHF 1.0

240W 1.4 SCRF 0.5 1.0 Ei - Eu (eV) FIG. 9(i) • • 300W 0.8 SCRF 0.5 1.0 Ei - Eu (eV)

300W 1.1 SCHF 0.5 1.0 FIG. 9(k)

•

300W 1.4 SCHF 0.5 1.0 Ei - Eu (eV) 9U)