The optogalvanic effect in a hollow cathode discharge

MISSING IMAGE

Material Information

Title:
The optogalvanic effect in a hollow cathode discharge a resonance detector for very weak light levels
Physical Description:
xiv, 202 leaves : ill. ; 28 cm.
Language:
English
Creator:
Petrucci, Giuseppe Antonio, 1963-
Publication Date:

Subjects

Subjects / Keywords:
Sodium   ( lcsh )
Lead   ( lcsh )
Uranium   ( lcsh )
Neon   ( lcsh )
Fluorescence spectroscopy   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre:
bibliography   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1990.
Bibliography:
Includes bibliographical references (leaves 196-201).
Statement of Responsibility:
by Giuseppe Antonio Petrucci.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001659163
oclc - 24529570
notis - AHX0890
sobekcm - AA00004769_00001
System ID:
AA00004769:00001

Full Text









THE OPTOGALVANIC EFFECT IN A HOLLOW CATHODE DISCHARGE:
A RESONANCE DETECTOR FOR VERY WEAK LIGHT LEVELS















By

GIUSEPPE ANTONIO PETRUCCI


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

UNIVERSITY OF FLORIDA


1990





























Copyright 1990

by

Giuseppe Antonio Petrucci































Per i miei geniton













ACKNOWLEDGEMENTS


I would like to thank many group members, past and present, who helped

bring about the completion of this work and who made my stay here an enjoyable

and often-times interesting and confusing one. There are, of course, some who had

a more pronounced influence on me, during the course of this work. Josef

Simensson, my "circus friend", who jumped through all the hoops with me; Chris

Stevenson, for his much appreciated help in the lab; and, Norma Ayala, for policing

my wardrobe. They all helped very much to make some very frustrating times

tolerable. Their friendship outside of the lab was priceless.

I would like to thank Jim Winefordner for his guidance throughout this work.

His example of perseverance and hard work will always be with me.

I would also like to thank Ben Smith. His contributions to my work, both

practically and emotionally, are incalculable. His advice often kept me going when

my spirits were down. I think that I can safely say that without his help, the

completion of this dissertation would have been a monstrous task.

On a more practical nature, I would like to thank Steve Miles who made

invaluable contributions in the electronics part of this work. His willingness to help,

discuss with (and teach) me some of the "black magic" of electronics is greatly








appreciated. I should give a special thanks to the secretaries, Jeanne Karably and

Susan Ciccarone, for putting up with me and my "unique" questions.

Finally, I would like to thank my (soon to be) wife, Nancy. Whether she

realizes it or not, her support of and confidence in me has always strengthened my

heart. At times when I strongly doubted myself and my abilities, she brought back

into my life a perspective and respect for myself that I often lost.















TABLE OF CONTENTS


ACKNOWLEDGEMENTS ..........

LIST OF TABLES .................

LIST OF FIGURES ................

ABSTRACT .....................

CHAPTER 1

INTRODUCTION ...........

Need for a High Sensitivity Photon
Photon Detectors .............
Intent of Dissertation ..........


........................... iv


ix


............................. x


Detector ........

.. ..... .. ... ... .


CHAPTER 2

CONCEPT OF RESONANCE MONOCHROMATOR

Historical Background .........................
Resonance Spectrometers ......................
Fluorescence Resonance Spectrometer ..
Resonance Ionization Detector ........
Line Broadening Mechanisms ...................
Natural Broadening .................
Collisional Broadening ..............
Doppler Broadening ............

CHAPTER 3

OPTOGALVANIC EFFECT ....................

Introduction ................ ...............
Anatomy of a Glow Discharge ...................








Electrical Maintenance of the Discharge ............


CHAPTER 4

CONCEPT OF RESONANCE MONOCHROMATOR .......... 38

Introduction ............................................ 38
Theory ............................................... 39
Phenomenological Description of Ionization Mechanisms .......... 39
Evolution of OG Signals .................................. 40
Negative Voltage Changes ....................... 41
Positive Voltage Changes ....................... 44
Mathematical Treatment of OG Signals ....................... 49
Double-resonance Optogalvanic (DROG) Effect ................ 54

CHAPTER 5

CONSIDERATIONS ON THE INTERACTION OF ATOMS
WITH LIGHT ................................... .... 58

Absorption of Radiation .................................. 58
Laser Excitation of Atomic Transitions ........................ 60
Single-Step Excitation ......................... 60
Optical Saturation of an Atomic Transition .......... 63
Two-step Excitation ............................ 65

CHAPTER 6

EXPERIMENTAL ................................... 69

General Experimental Configuration ......................... 69
Pick-off Circuitry for OG Signal ............................. 74
One-step Excitation Experiments ............................ 74
Two-step Excitation Experiments ............................ 81
Timing of Laser Beams in Two-step Excitation Experiments ........ 81
Absorption Experiments ................................... 86
Saturation Curves ....................................... 91
Measurement for Enhancement of Two-step Excitation of Na
and Pb ............................................... 91


vii








CHAPTER 7


RESULTS AND DISCUSSION ...........................

Sodium OG Effect ......................................
Determination of Lamp Impedance .........................
Evaluation of Collisional Ionization Rate Constants by the


OG Effect


Two-step OG effect of Na .....
Lead OG Effect in the HCL ...
Neon OG Effect in the HCL ...
Electronic Configuration of Ne .
One-step Neon and Uranium OG
Case 1...............
Case2 ...............
Case3 ...............


Effect


Case 4 ..........................................
Neon OG Signal Dependence on Lamp Current ................
Double-resonance OG Effect of Ne .........................
Optimization of Boxcar Gate Position .......................
Alignment of Laser Beam(s) in the Hollow Cathode .............
Ionization Within the Negative Glow ...................
Ionization Within the Dark Space .....................
Two-step Enhancement of Ne OG Effect .....................
Evaluation of Ne OG in the HCL as a Sensitive Photon Detector ...


CHAPTER 8

FINAL COMMENTS ........

Summary ..................
Future Work ...............

REFERENCE LIST .........

BIOGRAPHICAL SKETCH ...


viii


92

92
100

103
104
110
117
117
122
125
131
136
146
151
154
157
161
162
167
167
175


192


202


.................


...........................
...........................









LIST OF TABLES

Table 1. Values used for calculating ni/nt in equations (26) and (27) ..... 68

Table 2. Listing of experimental components ...................... 73

Table 3. Experimental values for determining ni in equation (32) ....... 105

Table 4. Values used to calculate vi and v. in equation (44) ........... 163









LIST OF FIGURES

Figure 1. Resonance monochromator proposed by Sullivan and Walsh [23] 9

Figure 2. Resonance ionization detector .......................... 15

Figure 3. Normalized Lorentzian line profile ....................... 23

Figure 4. Normalized Gaussian line profile ........................ 25

Figure 5. Electrical regions of a dc discharge [53] ................... 30

Figure 6. Voltage distribution across a dc glow discharge .............. 32

Figure 7. Electron energy distribution across a dc glow discharge ........ 34

Figure 8. Oscilloscope trace of the negative OG signal for the
3s1 2S1/2 --> 3p 2P3/2 transition of Na ..................... 43

Figure 9. Oscilloscope trace of the positive OG signal for the
3P2 (Is5) --> 3D3 (2p9) transition of Ne .................... 46

Figure 10. Partial energy level diagram of Ne ....................... 48

Figure 11. Electric field distribution across the dc glow discharge ......... 52

Figure 12. Partial energy level diagram of Na ....................... 56

Figure 13. Possible excitation/deexcitation process in an atom ........... 62

Figure 14. General experimental configuration ........ .............. 71

Figure 15. Diagram of a common HCL (lamp 1) ..................... 76

Figure 16. Diagram of a Galvatron (see-through HCL) (lamp 2) ......... 78

Figure 17. a) Pick-off circuit for measuring OG signals
b) Diagram of housing for pick-off electronics and lamp holder .. 80

Figure 18. Alignment of laser(s) through lamp 1 ..................... 83

Figure 19. Alignment of laser(s) through lamp 2 ..................... 85









Figure 20.


Figure 21.

Figure 22.

Figure 23.

Figure 24.

Figure 25.


Figure 26. Two-step enhancement of Na OG signal


Oscilloscope trace of temporal coincidence of the two
laser beams at the hollow cathode ......................

Experimental configuration for absorption experiments .......

OG signals for ground state transitions of Na ..............

Example of signal-to-noise attainable using the OG effect ......

OG signal for excited state transition of Na ................

Plot of (Vout)1 vs R, for determination of lamp impedance ....


Collisional coupling of Na energy levels in the HCL ......... 109

Partial energy level diagram for Pb ...................... 112

Two-step enhancement for Pb OG signal ................. 114

Pb two-step OG signal vs lamp current ................... 116

OG spectrum of U/Ne lamp from 580 601 nm ............ 119

Expanded U OG spectrum from 580 601 nm ............. 121

OG spectrum of U/Ne lamp from 594 607 nm ............ 124

Saturation curve for 597.55 nm absorption of Ne ............ 128

Log-log plot of Figure 34 ............................. 130

Saturation curve for 598.80 nm absorption of Ne ............ 133

Log-log plot of Figure 36 ............................. 135

Proposed excitation/ionization scheme for 598.80 nm
absorption of Ne ................................... 138

Excitation scheme through a virtual level ................. 140


Figure 40. Saturation curve for 599.56 nm absorption of Ne ....


88

90

94

96

99

102

107


Figure 27.

Figure 28.

Figure 29.

Figure 30.

Figure 31.

Figure 32.

Figure 33.

Figure 34.

Figure 35.

Figure 36.

Figure 37.

Figure 38.


Figure 39.








Figure 41.

Figure 42.

Figure 43.

Figure 44.


Figure 45.


Figure 46.

Figure 47.

Figure 48.


Figure 49.







Figure 50.


xii


Log-log plot of Figure 40 .............................

Saturation curve for 599.24 nm absorption of U ............

Log-log plot of Figure 42 .............................

Proposed excitation/ionization scheme for 599.24 nm
absorption of U ....................................

Plot of relative laser induced impedance change vs.
lamp current.......................................

Boxcar gate position considerations ......................

One-step ionization signal observed in cathode dark space ....

Convolution of ionization signals from cathode dark space
and negative glow region .............................

Two-step enhancement of Ne OG signal in HCL
a) Oscilloscope trace of one-step excitation OG signal of
Ne (ground state) ...................................
b) Oscilloscope trace of one-step excitation OG signal of
Ne (excited state) ...................................
c) Oscilloscope trace of two-step excitation enhancement of
OG signal of Ne ....................................

Scan showing simultaneous recording of absorption of X 12
and OG signal for determination of a ....................

Calibration curve for Ne RID in the HCL .................

Log-log plot of Figure 51 .............................

Plot of V(rms) vs lamp current ............. .........

Plot of signal-to-noise of Ne RID vs. lamp current ..........

Summary figure of results .............................


Figure

Figure

Figure

Figure

Figure


145

148

150


153


156

160

166


169



171

172

173


179

182

184

186

189

191













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

OPTOGALVANIC EFFECT IN A HOLLOW CATHODE LAMP:
A SENSITIVE DETECTOR FOR VERY WEAK LIGHT LEVELS

By

GIUSEPPE ANTONIO PETRUCCI

DECEMBER 1990


Chairperson: James D. Winefordner
Major Department: Chemistry

Both one- and two-step excitation optogalvanic (OG) effects in two

commercially available hollow cathode lamps are studied. Elements studied are

sodium, lead, uranium and neon. Four different excitation/ionization mechanisms

were observed by the OG effect and are discussed. These include: 1) one-step

excitation followed by collisional ionization; 2) one-step excitation followed by

photoionization; 3) two-step excitation through a virtual level, followed by collisional

ionization; and, 4) two-step excitation through a virtual level, followed by

photoionization. The two-step excitation to a real level followed by collision and

photoionization were also observed for Ne. Two-step excitation enhancements in the

OG effect of Na, Pb and Ne, relative to the one-step excitation cases were

determined.


xiii








The use of a coupled, step-wise, two-step excitation of Ne in the hollow

cathode lamp was evaluated as a sensitive detector for very low light levels. The first

transition studied was the 3P2(ls5) --> 3D3(2p,), at a wavelength of 640.225 nm. The

excited state transition, coupled to the first transition, was the 3D3(2p9) --> 4d4'

transition, at 576.442 nm. The limiting experimental noise was determined to be the

shot noise of the hollow cathode lamp. Under optimized conditions, the

experimentally attained minimum detectable energy and number of photons were 6

x 1015 J and 2 x 104 photons, respectively. The limits of detection corresponding to

the theoretical shot noise limit in the experimental system were 1 x 1016 J and 4 x

102 photons.


xiv













CHAPTER 1
INTRODUCTION



Need for a high sensitivity photon detector


The search for and development of methods for determining traces of

elements is one of the most important trends in analytical chemistry [1].

Development of these methods is determined to a large extent by the needs of

various branches of science, technology and industry relating to the production and

use of ultrapure materials. Two examples of the need for ultra-trace analysis include

the study of the effect of impurities during the production of silicon solar-cells [2]

and the production of optical glass [3]. In meeting these needs, the problem of

measuring light intensities and their spectral distribution at very low levels is

becoming considerably important.

In contemporary optical spectroscopy, complex and expensive monochromators

and photon detectors are used for discriminating and detecting an extremely narrow

wavelength range in a light train with an arbitrary spectral composition. Such

instruments must possess high sensitivity to the photons to be detected and high

spectral resolution for reducing background noises. At present, two kinds of

photoelectric measurements are being used to accomplish the measurement of very

low light levels. These are referred to as analog [4,5] and digital [6,7] schemes.








2

With the digital method, contributions from photoelectrons are resolved in time so

that the signals in the form of electron pulses are detected by means of a pulse

counting system. This technique is the one of choice when concerned with the

measurement of ultra-low light levels.


Photon detectors


A detector of optical radiation is defined by the Commission Internationale

de l'Eclairage [8] as a device in which incident optical radiation produces a

measurable physical effect. Two classes of detectors of optical radiation produce a

measurable signal by one of two primary detection mechanisms: the photoelectric

effect (photon detectors) and the thermal effect (thermal detectors). Photon

detectors are the most common in optical spectroscopy for the measurement of low

light levels. Photomultipliers (PMTs) are the standard photon detector in

commercial spectroscopic instruments. Their basis of operation is the generation or

change of an electric signal by an external photoelectric effect in which a

photoelectron is emitted by a cathode and captured by a second electrode (or

dynode). The important feature of the PMT is the dynode system, which consists of

electrodes covered with special materials which emit various "secondary electrons"

per incident primary electron. The number of dynodes in PMTs ranges from 3-6,

resulting in a total gain or amplification of the cathodic photo-induced current of up

to 107 or more.








3

The detection of light has undergone a great metamorphosis since 1800 when

Sir William Herschel [9] used an ordinary thermometer to measure the intensity of

light passing through different filters. In 1831, Nobili and Melloni [10] developed

a thermocouple for the quantitation of light intensity. Samuel Pierpont Langley

[11], in 1881, developed the bolometer which was 30 times more sensitive than

Melloni's thermocouple. The bolometer is based on the change in resistance of a

wire as it is heated. By incorporating the wire into a Wheatstone bridge, high

sensitivities were possible. In 1884, W.N. Hartley [12] applied photographic

detection of light intensity to quantitative chemical analysis. The key development

in the use of photography as a means of detection for quantitative chemical analysis

was the concept of using an internal standard. The next step in the development of

detection systems for spectroscopy was the development of photoelectric detection.

The origins of photoelectric detection date back to Hertz [13,14,15]

who observed that a spark produced by one circuit was somehow transmitted through

space so as to induce a spark in a secondary circuit even though the two circuits were

not connected directly. By placing a slit and a quartz prism between the primary

spark and the induced spark gap, he was able to show that the effect occurred only

when the prism was arranged to transmit light in the ultraviolet region of the

spectrum. The next year, A. Righi [16] demonstrated that a small current could

be made to flow between a mesh grid and a polished metal plate when the plate was

irradiated. The study of the photoemissiveness of different materials, mostly

amalgams of alkali metals, was undertaken by Elster and Geitel [17] and the first








4

modern photoemissive surface was described by LR. Koller [18,19]. Even with

the great strides made in improving the sensitivity of photocathode materials, the

most sensitive surface could not pass a large amount of current without destroying

the photoemissive layer. In 1940, Rajchman and Snyder developed a nine-stage

electrostatically focused multiplier phototube that has served as a model for

"modern" photomultiplier tubes [20].

In general, the spectral responsiveness of photon detectors is selective, being

determined largely by the chemical-physical composition of the materials used to

absorb photons. This spectral responsivity of individual photocathode materials is

normally "limited" to a wavelength range of approximately 600-700 nm in width

[21]. This dictates that for higher spectral resolution of the detected line, a primary

disperser must be used to separate the incident light into its component wavelengths,

before it is made incident on the photocathode of the PMT. Monochromators are

the method of choice for wavelength dispersion of the incident light.

Monochromators with several dispersing stages in series are used for very high

resolution of the wavelength of interest and reduction of stray light.

Two main criteria used to evaluate monochromators are optical throughput

and resolution. The optical throughput of a monochromator is defined as the

amount of radiation reaching the detector for a given amount leaving the source.

The effective throughput of the entire monochromator is limited by the portion of

the instrument with the smallest throughput. In conventional monochromators, the

limiting throughput is either the entrance or exit slit of the monochromator.








5

The resolution, on the other hand, is a measure of the ability of the

monochromator to disperse incident light composed of many different wavelengths

into its component wavelengths. It is given by the product of the width of the

entrance or exit slit (assumed to be equal in this case) and the reciprocal linear

dispersion of the dispersing element in the monochromator. The product of the

throughput and resolution is constant for a given monochromator. However, since

resolution improves with narrower slit widths and throughput increases with wider

slit widths, the spectroscopist is constantly faced with a choice of compromise

conditions involving resolution and throughput.

For a single-stage monochromator the throughput is on the order of 10-103.

For multiple-stage monochromators, the overall throughput is a product of the

throughput of each stage. So, for a triple-stage monochromator, the fraction of

incident light that reaches the PMT is only 107-109. This inverse relation between

throughput and resolution is one of the major limitations of using conventional

detectors. The other major limitation is a result of the low quantum efficiency of the

photocathodes of PMTs.

The quantum efficiency is defined [8] as the ratio of the number of elementary

events (e.g., photoelectrons) contributing to the detector output to the number of

incident photons. For most cathode materials, the quantum efficiency is very low;

on the best sensitized commercial photosurfaces, the maximum yield reported is as

high as one electron for three light quanta incident on the photocathode. An ideal

photodetector has a quantum efficiency of 1; i.e. every incident photon produces one








6

photoelectron. All practical photocathode materials have quantum efficiencies of less

than 1. Therefore, a PMT could never detect a single photon from the source.

As described below, resonance spectrometers offer the advantages over

monochromator/PMT detection systems, of high spectral resolution (103 nm) in

conjunction with a high optical throughput (>0.5).


Intent of dissertation


The present work was intended to evaluate and characterize a photon detector

based on the optogalvanic effect in a commercial hollow cathode lamp. Optical

transitions of sodium, lead and neon, the inert filler gas were considered for the

detector. Both single- and double-resonance excitation schemes were used. Also,

several single-step transitions of Ne and U were studied in terms of excitation-

ionization mechanisms.













CHAPTER 2
CONCEPT OF RESONANCE MONOCHROMATOR



Historical Background


The requirements of high sensitivity and wavelength selectivity were already

met in 1905 through the simple and effective techniques for discriminating resonance

radiation of specific elements which were based on the resonance absorption of

photons and their subsequent re-emission in atomic vapors of the element [22].

Such a device is termed a resonance monochromator (RM) and was first proposed

in 1968 as an accessory for modern spectroscopy by Sullivan and Walsh [23]. The

basic concept of the RM, as a detector for atomic absorption spectroscopy, is shown

in Figure 1.

The light from several spectral lamps is collimated and passed through the

flame. The radiation transmitted is then passed into several RMs. Each spectral

lamp (or element of interest) has a corresponding RM. In Sullivan and Walsh's case,

the RMs were simply high current hollow cathode lamps to provide the requisite

atomic vapor density with minimal emission. Upon absorption of resonance radiation

by the RM, ground state analyte atoms would be excited and subsequently reemit the

radiation, which was observed at right angles to the incoming radiation. Each RM
















0

o


0
to


0

a















o







0
cd

0
53















I-
0
E















0
I-











0
o











-4
I-
0d
^3
0~
'u















-4
























LLI I








10

would then respond selectively to the transmitted intensity of the corresponding

analysis lamp.

Since 1968, resonance spectrometers based on several optical and electrical

principles have been studied, including fluorescence, ionization and optoacoustic

techniques. A variety of atom cells have also been studied for resonance detection.

Among these are flames [24] (laser enhanced ionization, LEI), hollow cathode

lamps [25,26] (Optogalvanic effect, OGE), high-frequency spectral lamps [27]

and furnaces [28]. Resonance monochromators, however, have been used mainly

for absorption experiments.


Resonance Spectrometers


Resonance spectrometers offer several important advantages over

conventional spectrometers. They possess an inherently high spectral resolution (10-3

nm), a large solid angle of collection of the incoming light (up to 2n sr), low intrinsic

noise, and high quantum efficiency, namely, the capability of single photon detection.

This is in contrast to photomultiplier tubes, which have maximum cathode efficiencies

of about 30%, meaning that at best, 3 photons are required on the average for the

detector to record an event. Also, as stated above, several different spectroscopic

techniques may be used as the basis for a RM. Two of the most studied are

fluorescence (fluorescence resonance spectrometer, FRS) and atomic ionization

(resonance ionization detector, RID). Atomic ionization is used here as a generic








11

term for all spectroscopies which are based upon the galvanic measurement of

ionization rate enhancements induced by the absorption of resonance radiation.

Fluorescence Resonance Spectrometer

The fluorescence-based resonance spectrometer has been the most widely used

by far. The resonance spectrometer proposed by Sullivan and Walsh [23] was of the

fluorescence type. The principle of operation is based on the absorption of the

resonance radiation of interest by atoms in the optical cell and the recording of the

emitted (resonance or non-resonance fluorescence) light. Such a resonance

spectrometer was used by Bolger [29] for the recording of weak Raman scattering.

In this experiment, a 10 ns, 1 kW pulsed tunable dye laser, with a bandwidth of 0.1

cm71 (S6 =0.002 nm) was used to illuminate the sample. The wavelength of the laser

was chosen so that the Raman scattered light was around 455 nm. This light was

used to excite Cs atoms. The fluorescence of the Cs atoms at 850 nm was then

measured as the analytical signal. The detector sensitivity was found to be

approximately 3 orders of magnitude higher than a dispersion system with a

conventional detector.

The first fluorescence resonance spectrometer (FRS) working with transitions

from excited states was proposed and studied by Gelbwachs et aL [30]. They

excited Na atoms in the FRS to the 4P3/2 state (1 =330.3 nm) with a dye laser.

Radiation from a xenon arc lamp, 3.42, 2.34, and 1.48 P m, was then absorbed by the

excited Na, promoting it to the 5S1/2, 4D5/2, and 5D5/2 levels, respectively. They then

observed fluorescence of the Na in the visible region at wavelengths of 616, 569, and








12

548 nm, respectively, for the three transitions. The same workers also proposed and

studied the only tunable FRS. This RM, based on potassium, was made tunable over

a narrow spectral range by shifting of the highly excited energy levels of K by the

quadratic Stark effect [31]. One of the problems of fluorescence-based resonance

spectrometers is the need still for a separate monochromator and photon detector.

With an FRS some means of isolating the fluorescent light from the FRS from other

wavelengths and a photoelectric detector are still required to measure the intensity

of fluorescent photons from the FRS. Therefore, the throughput and resolution

limitations of a monochromator/PMT detector are still applicable, albeit to a lesser

extent than in the simple case of detection with a monochromator and PMT. Also,

the less than unity quantum efficiency of the photocathode material of the PMT

places a fundamental limit on the minimum detectable number of photons.

Resonance Ionization Detector

Atomic ionization (AI) methods have been proven to be very sensitive

spectroscopic techniques [32,33]. They are based on the recording of charged

particles which are formed in the atom cell upon absorption of resonance radiation

followed by subsequent ionization of the excited atoms by various processes. In the

extreme case of detecting single charges, proportional gas counters are used. For

bulk measurement of a large number of charges formed, the impedance or current

change across the atom cell is measured. Limits of detection in normal atmospheric

pressure atomizers are in the low ng/mL to pg/mL (or pg to fg absolute detection

limits) [34]. V.S. Letokhov [35] was the first to draw attention to the fact that








13

it is possible to detect even single atoms by the AI method. This limit of single atom

detection, or SAD, was experimentally attained by Hurst et at, [36,37] for Cs and

by Bekov et at [38] for Na and Y. Detection of single atoms is possible by AI

because of the ability to detect charges with unity efficiency.

Matveev et at [39] first proposed the use of laser-enhanced ionization in a

flame as a photon detector. Work to date has been mainly on the theoretical

feasibility of such a detector [40,41,42]. Okada et at [43] were the first to

report quantitative results of a selective laser ionization photodetector. They

reported a minimum detectable energy, in the first wavelength, of 10" J, while Smith

et at [44] are the only workers to record a Raman spectrum with a RID. Their

minimum detectable energy can be estimated as approximately 1016 J at 285 nm.

The case of the RID (Figure 2) is very similar to that of the FRS. Radiation

(hv )emitted from the sample (SA) is conveyed by an optical system into the RID,

which contains a high concentration of atomic vapor. In the simple case of detecting

atomic emission or resonance fluorescence, the atomic vapor in the RID is the same

as the analyte of interest in the sample. Absorption of the radiation upon passing

into the RID causes a measurable change in the electrical properties of the RID.

For detection of scattered radiation (eg. Raman scatter), the RD becomes

much more favorable and versatile to use than conventional dispersive instruments

with photomultiplier detection. In Raman spectroscopy, for example, one is

interested in measuring the intensity of scattered radiation shifted in energy by a

characteristic amount from the central frequency of the exciting light. The












N

0






'I.



4-1


4,)
ob







4( 0
4)
5 o













0 03
0






eo









-S
co0
UN


0.-



04)
0



e,







*-







E4
















1<


-I r


tr


C~2


0,


2
~ P










frequency shift is characteristic for a particular scatterer and is given by


Av (1)



where Av-R is the characteristic Raman scatter energy, and 1/1a. and 1/.stok are

the wavelengths of the exciting light and Raman shifted light, respectively.

It is important to emphasize the inverted nature of the RID approach of

Raman scatter compared to conventional detection. In conventional Raman

spectroscopy, the spectrum of the Raman scattered light is obtained by imaging the

scattered light onto the entrance slit of a double monochromator. The

monochromator is then scanned from wavelengths longer than the exciting

wavelength to the central exciting wavelength, thereby observing the different A"'s

for the scatterer. In "RID" Raman spectroscopy, the detector operates at a fixed

wavelength, dictated by the energy level scheme of the "detector element." The

exciting light must therefore be scanned to shorter wavelengths than the expected

Raman scattered wavelengths. Any element can be chosen for the RID which gives

the best sensitivity and selectivity, and then the frequency of excitation for the

Raman scatter is chosen such that the scattered radiation will be at a frequency

detectable by the RID.

One of the major advantages of such a detector is that it will respond only to

wavelengths of radiation that are within the absorption bandwidths of the various








17

transitions of the RID element. Anything outside of the absorption bandwidth of the

RID element will not be absorbed and will therefore not produce a signal.


Line-Broadening Mechanisms


The absorption (response) bandwidth of the RID element will depend on the

operative broadening mechanism of the atomic line in the RID. This makes the RID

inherently insensitive to problematic scattered light. Atomic lines are not truly

monochromatic and do possess a finite frequency (or wavelength) distribution. The

finite widths obtained are a result of a variety of line-broadening phenomena. The

width of the line is normally characterized by the width at one-half the maximum

height, or full-width at half maximum, FWHM. All of the different line-broadening

phenomena result in a frequency distribution of the atomic transition which can be

described by either a Lorentzian or Gaussian expression. However, since more than

one line-broadening process can be operative at any one time, the overall profile is

neither purely Lorentzian or Gaussian, but a combination of the two, known as a

Voigt profile. The predominance of one or more of the line-broadening mechanisms

is a function of the chemical and physical environment of the absorbing or radiating

atom. The three most commonly discussed broadening mechanisms are natural

broadening, collisional or pressure broadening and Doppler broadening. Each will

be discussed briefly below.

Natural Broadening

Because of interaction of atoms with radiation fields and collisional processes,








18
the lower, i, and upper, j, energy states participating in the atomic excitation have a

finite lifetime, and this gives rise to uncertainties in the energy of both states

according to the Heisenberg uncertainty relation. Usually, the extent of the line-

broadening is determined by the much shorter lifetime of the upper state of the

transition. This is called lifetime or natural broadening. The FWHM due to natural

broadening is smaller than that due to other broadening mechanisms except in

extreme cases (eg. transitions to autoionizing levels of the atom). The spectral

profile of a naturally broadened line is described by a Lorentzian. The normalized

spectral profile due to natural broadening, SN, is given by


S 2/( Av, )
S 2( v ) (2)
AvN



where

AvN = FWHM of the naturally broadened line (Hz)

vm = frequency at the line center (Hz)

The width of a naturally broadened line, between two real energy states of the

atom (see below), is on the order of 107 Hz. In wavelength units, the FWHM, A~N

(m), is expressed as


AlN Av, (3)
C










where

s8 wavelength at the line center (m)

c speed of light (m s1)



For example for Na, at 589.0 nm, ALrP2x1014 m or 0.02 pm.

Collisional Broadening

The line profile is also influenced by collisions of the atom of interest with

other atoms of the same kind (Holtzmark broadening) or with foreign species

(Lorentz broadening). The amount of broadening caused by collisions increases with

the concentration of collision partners and hence, collisional broadening is sometimes

called pressure broadening. Collisions which leave the atom in the same energy state

(adiabatic collisions) have a more pronounced broadening effect than collisions that

leave the atom in a different energy state diabeticc collisions). The FWHM of the

spectral distribution resulting from adiabatic collisions, Av, (Hz), is given by


Aa oan (4)




where

a, optical cross section for adiabatic collisional broadening (cm2)

n. = density of collision partners (cm"3)

A = reduced mass of collision partners (g)








20
The total, normalized spectral profile, S,, for collisional broadening is also

Lorentzian and is given by equation (2) by replacing the natural FWHM, Av N, with

the collisional FWHM, A v. The typical FWHM of a collisionally broadened line is

on the order of 3 x 10"12 m or 3 pm.

If natural and collisional broadening are assumed to be mutually independent,

the resulting FWHM of the Lorentzian profile is




Av, Avc + AvN (5)



where Av is the FWHM from both adiabatic and diabetic collisions.

Doppler Broadening

A third major source of line-broadening is a result of the statistical

distribution of the velocities of the atoms absorbing radiation along the observation

path, called Doppler broadening. Because atoms are in motion with respect to the

observer, the Doppler effect causes a statistical distribution of frequencies absorbed

that is directly related to the velocity distribution. Briefly, the FWHM for Doppler

broadening, AvD (Hz), is given by



Av 2- 2(n 2)kT c (6)
m I c (6)









where

k = Boltzmann constant (J K-1)

m = atomic weight (g)

vm = central frequency (Hz)



The normalized spectral distribution, SvD, is described in this case by a

Gaussian relation and is given by



SVD J exp (n2)(v )2 (7)
AV/ L Av2



Typical Doppler broadened linewidths are on the order of 4 x 10T12 m or 4 pm. The

overall profiles for most atomic spectral lines are neither purely Gaussian nor purely

Lorentzian, but rather a combination of the two known as a Voigt profile.

The predominance of one type of line over another is largely dependent on

the environment of the atoms. For example, in an atmospheric pressure flame, one

might expect that collisional broadening would dominate over Doppler broadening,

whereas in a low pressure environment the opposite is expected to be true. The

importance of which broadening mechanism predominates is clear from Figures 3

and 4 which show the line profiles for purely collisionally and Doppler broadened

lines, respectively. These line profiles were plotted using equations (2) and (7) and

assuming a FWHM of each line of 1011 m. The selectivity of the RID under each

















0



















o
*1
I-1




cb.





0
C



.)








3













N
ct







g-





'5


























I
O t


T"
o.






0






0 0 0 0 0 0 0




uo dilosqB a!7ruIaa












0
0
O
0
0


0
0
o
d
O
O





o
d


0
0
0
c0

0
C C
0


O
0
0O


Sd T 0 q0 0 0 0 0 0
'-" -"- T -r 0" T


uo!ldjosqB 9A!ji'6gI









26

of these conditions can be readily obtained from these plots. For example in the

case of a Doppler broadened line, the fraction of the incident light falling within the

absorption linewidth of the detector element falls to 10'12, comparable to the stray

light rejection of a good doublemonochromator, at a wavelength only 3 x 1011 m or

30 pm from the central line (evaluated at 640.2 nm). The collisionally broadened

line on the other hand, has long tails, giving a fractional absorption of 107 as far out

as 6 nm away from the central line (evaluated at 640.2 nm). Therefore, a low

pressure cell, in which collisional broadening is not the dominant broadening

mechanism, is essential for highest wavelength selectivity.

One other advantage of the RID is that there are no limiting apertures,

limiting the optical throughput of the detector. This allows for a greater fraction of

the sample light to be collected and imaged into the RID, up to 50%. A good two-

stage monochromator will have an optical throughput of 105-106%. The product of

throughput and selectivity is constant for any one system. Clearly, the higher the

value of this product, the more selective and sensitive is the detector. The RM is the

only photon detector for which the detection efficiency need not be compromised for

better resolution or selectivity.














CHAPTER 3
OPTOGALVANIC EFFECT



Introduction


The optogalvanic (OG) effect is a change in the properties of an electrical

discharge caused by illuminating the discharge with radiation having a wavelength

corresponding to an atomic or molecular transition of a species in that discharge.

The effect was first observed by Foote and Mohler [45] in cesium-vapor filled hot

cathode diodes and Meissner et al. [46] and Penning [47] in rare gas glow

discharges. It was Penning who first discovered that an increase in breakdown

voltage occurred in a mixture of neon and argon when discharge tubes were

irradiated by emission from an identical tube.

Although the OG effect was first observed over 60 years ago, little work has

been done exploring its analytical capabilities. It was not until the development of

tunable lasers that the analytical potential of the OG effect began to emerge. Since

that time, the OG effect has found applications in optical spectroscopy [48],

analytical chemistry [49,50], laser stabilization [51] and wavelength

measurements and calibration [52], to name a few.











Anatomy of a Glow Discharge


The typical schematic representation of a dc glow discharge is shown in Figure

5 (taken from Nasser [53]). The normal glow discharge consists of 5 distinct

regions: 1) cathode dark space; 2) negative glow; 3) Faraday dark space; 4) positive

column; and 5) anode dark space. The corresponding diagnostic plots are given

below the figure. The positive column has by far, been the most probed and

modelled region. As the electrode separation is decreased, the negative glow and

cathode dark space remain unaffected, while the positive column shrinks, alluding to

the importance of these two regions in the maintenance of the discharge. When the

electrode separation is just a few times the cathode dark space length, only the dark

space and negative glow regions remain [54]. This is the normal operating mode of

hollow cathode lamps and is the experimental mode of operation in this work.

Note that almost the entire applied voltage is dropped across the cathode dark

space, while the negative glow remains essentially field-free. As such, its boundary

acts as a diffuse anode.

To maintain electrical balance in the system, the currents at the two

electrodes must be equal. Experimentally, however, it was observed [54] that the

current density at the anode was smaller than the random current density and so

there must be a decelerating field for the electrons at the anode. Figure 6 shows the

voltage distribution in a de glow discharge. The values given are for typical

experimental parameters used in this work. Figure 7, redrawn from Figure 6, is an































Figure 5. Electrical regions of a dc discharge [53].



























1st, 2nd ...
CATHODE LAYERS.


R SUPPLY
-- .-0


POSITIVE COLUMN
NEGATIVE GLOW\ ANODE GLOW


I FARADAY DARK SPACE
CATHODE
DARK SPACE (CROOKES. H




K!=


I
ITT
-A(




-I


> I i









t I I




4 ----


ANODE DARK SPACE
ORF)

LIGHT INTENSITh'




LLCTW.,IC E i i,


POTENTIAL




POSITIVE L3 A,
I:HARGE DE';SI T

NEGATIVE SPACf
CHARGE DNSI T Y
- IOTAL





I ,.S TELMP(f[ A I U F i!


ASTON
DARK
SPACE


i\


























































O
I-










v,
rU


I-

0




-o













CL























AC
5, M


S o



















5 0




















+ aB


- -I















































-d




0
U

-o

0


0



' i







0

U
c4





































a)a









35

electron energy diagram for the discharge and shows the potential barriers to

electron movement. The negative glow does not take a potential intermediate to

those of the electrodes, as might be intuitively expected, resulting in a potential well

between the cathode and anode. The electric fields in the system are restricted to

sheaths at each of the electrodes and are such as to repel electrons trying to reach

either electrode.


Electrical Maintenance of the Discharge


In the text by Chapman [54], ion pair contributions from possible ionization

mechanisms in the three regions of interest are calculated. He concludes that the

two most probable discharge maintenance mechanisms, both occurring in the

negative glow region, are ionization by impact with thermal electrons [54, p.117]


A* + e- A' + 2e- (8)



and associative ionization of metastables [54, p.122].


A' + A' A + A + e- (9)



Collisional ionization of ground state atoms of the inert gas by electron impact is not

very probable due to the high energy ( >21.6 eV for Ne) required.

Ionization of sputtered cathode atoms can occur by any of the above

mechanisms, as well as Penning ionization (10)












X* +M-X + M + e- (10)



where X represents a buffer gas atom, and direct ionization of ground state sputtered

atoms, since much lower electron energies are required ( < 12 eV).

It is interesting to examine how photo-induced transitions can affect these

ionization processes if A is Ne, for example. Under normal glow discharge

conditions, the electron energy distribution approximates a Maxwell-Boltzmann

distribution and falls rapidly above 16 eV [55], where inelastic collisions dominate

(excitation of excited electronic states and ionization of Ne). Several points should

be noted. First, the neon metastable states, Ne'm, will be efficiently populated by

electron-neon ground state collisions, since singlet-triplet excitation functions peak

near the threshold voltage for discharge maintenance [56]. The subscript m is used

to denote neon in the metastable state. Second, collisional ionization of metastable

neon by electrons [process (8) above], whose cross-section peaks at about 15 eV

[57] can be expected to become increasingly important at high discharge currents.

Finally, the electron-neon collision frequency (- 1010 s-1) exceeds both the neon-neon

collision frequency (107-108 s"1) at normal HCL pressures (typically 1-5 torr) and the

reciprocal of the shortest radiative lifetimes for the excited Ne atoms

[58,59,60]. One can conclude from the above that, regardless of the precise

mechanism of ionization, metastable Ne atoms play a central role in the maintenance

of the discharge, as discussed by Chapman [54]. This was also supported

experimentally by the work of Smyth et aL [61] and Hess and Harrison [62]. In









37

these experiments, the Ne' concentration was monitored with a mass spectrometer.

Substantial changes in the Ne+ density were detected and were found to correlate

qualitatively with the photon-induced voltage changes. One can therefore expect that

the dynamics, voltage and current characteristics of the discharge would be sensitive

to even small perturbations of the Ne'm population.

An immediate realization gained from contemporary literature on glow

discharges in general and more specifically hollow cathode discharges is that such

discharges are dynamic environments which do not readily lend themselves to precise,

detailed modeling. In fact, a unified and generally agreed upon (quantitative) theory

of the mechanisms which are operative in the glow discharge has yet to be

presented.














CHAPTER 4
OPTOGALVANIC EFFECT IN THE HOLLOW CATHODE DISCHARGE



Introduction


A great deal of experimental work has been performed in a variety of

discharges, many of which were "home-made" [63,64,65] and contained an

assortment of vapors, such as 12 [66,67], rare gases, such as Kr [68], He [69],

Ar [70] and Ne [71] and numerous mixtures thereof. Of most practical interest

to the analytical chemist, however, is the use of commercially available hollow

cathode discharges (HCDs) or lamps (HCLs) in OG experiments. Unfortunately,

knowledge of the operating characteristics and parameters of HCDs is largely

empirical. General conclusions, however, can been drawn from previous work about

the OG effects in glow discharges. Firstly, the impedance changes normally induced

by the introduction of a collimated beam of light that is resonant with an absorption

transition of a discharge species into a discharge are large (as high as 10% of the

quiescent voltage) and are easily measured. Secondly, it is well known that these

impedance changes can be positive or negative.














Theory


Numerous theories and mathematical expressions have been postulated to

quantify and more fully understand the OG effect in HCDs. These range from the

relatively simplistic theories of Erez et aL [72] and Smyth and Schenck [61] to the

more arduous and pedantic theories of Ben-Amar et aL [73] and Broglia et aL

[74]. As stated above however, no agreement has been made on any one, complete,

unified theory to describe the processes in an HCD.

A complete description of proposed theories and mathematical treatments of

this dynamic system are beyond the scope of this dissertation: a phenomenological

description of the HCD characteristics and mechanisms and a simplified

mathematical treatment of the OG process shall be presented herein as pertinent to

this work.


Phenomenological Description of Ionization Mechanisms


There are two principal mechanisms suggested to explain the impedance

change of a hollow cathode discharge upon illumination by resonant photons. The

first mechanism, analogous to the OG effect in flames [34] is based on the increase

in cross-sections for electron-collision induced ionization as the final state energy

approaches the ionization limit. This mechanism is believed to be the dominant one

in the cathode dark space due to the high concentration of energetic electrons. The









40

second mechanism, which is proposed to be dominant in the negative glow region,

is an increase in the electron temperature of the discharge [75] upon irradiation

with resonant photons. The basis for this mechanism, proposed by Keller et aL

[71,76], is the establishment of an equilibrium in a hollow cathode discharge

between thermal electrons and the atomic excitation through inelastic and

superelastic collisions between excited atoms and thermal electrons, such that, to a

first approximation, the electron temperature and the atomic excitation temperature

are equal. This results in an increase in the electron temperature and a

corresponding shift of the electron energy distribution toward higher energies,

producing more electrons with sufficient energy for ionization of discharge species.

These two ionization mechanisms are not exclusive to the particular regions of the

glow discharge. The overall formation of charges is certainly a convolution of the

two processes.


Evolution of OG Signals


Upon irradiation of a HCD, the voltage change across the discharge may be

positive or negative. The polarity of these voltage changes is a complex function of

the originating level of the transition, the lamp voltage or current, the region of the

discharge illuminated, the electrode geometries and the pressure of the discharge.

Most observed OG signals are negative in polarity. This includes some signals due

to the buffer gas and all signals due to sputtered cathode atoms.













Negative Voltage Changes

Signals which are due to transitions originating in non-metastable levels

generally lead to negative voltage changes across the discharge. The only case where

this might not be true is that of excitation from a non-metastable level which is

important in maintaining the discharge, to a level from which the probability of

ionization decreases. This is not very common, however. Upon absorption of

resonant photons, atoms are promoted to higher lying energy levels, thereby directly

or indirectly increasing the ionization rate in the discharge. Depending on the region

of the discharge irradiated, the mechanism leading to an increased ionization rate is

different as discussed above. In the negative glow, excitation of atoms in the

discharge leads to an increase of the electron temperature (energy) through inelastic

collisions between slow electrons and the excited atoms74. This shifts the electron

energy distribution to higher energies. In either the cathode dark space or negative

glow region, the end result is that is that as a result of atomic excitation there will

be more electrons with sufficient energy to ionize discharge species, resulting in an

increase in the conductivity of the discharge. Since the lamp is made to operate at

constant current, assuming simple Ohmic behavior, an increase of the conductivity,

or correspondingly a decrease in the resistance of the lamp, will result in a lower

voltage required to maintain the constant current. An example of such a negative

voltage change can be seen in Figure 8. This is an oscilloscope trace, obtained in

this case for the 3s1 2S, --> 3p1 2P3/2 transition of Na upon laser irradiation of the























z


































0
OD
c0











4-
cl
^



4-1
0





















0m
o












IS
o
1














o
r:-










60














43
































I-









A!P/AL 03









44
hollow cathode at 588.995 nm with a laser beam of 10 0pJ. This signal corresponds

to an approximate -2% change in the voltage between the lamp anode and cathode

while the laser is on.

Positive Voltage Changes

The case of positive voltage changes is much more involved and less

understood. All transitions originating from metastable states lead to positive voltage

changes upon irradiation by resonant photonsif those metastable states are important

in the maintenance of the discharge. An example of such a positive signal is given

in Figure 9 for the 3(P2(s) --> 3D3(2p9) transition of Ne. A partial energy level

diagram for Ne is given in Figure 10. Neither of these two states is radiatively

coupled to the Ne ground state. The upper level however, is coupled to other 2p,

levels through electron collisions. These other 2p levels are radiatively coupled to

the ground state indirectly through the non-metastable levels of the Is quartet, 3P1

(ls4) and 1P, (ls2). This was confirmed by monitoring an increase in the emission

from the 2p, and 2po levels when the exciting light was tuned to populate the 2p,

level. Since the maintenance of the discharge is closely tied to the Ne metastable

density, by depleting the metastable population, one effectively reduces the main

mechanisms for ionization, thereby decreasing the ionization rate. This leads to an

increase in the lamp impedance and a corresponding increase in the lamp voltage.

The positive signals evolve over a much larger time period, probably as a result of

indirect depletion of the metastable density because of the parity forbidden direct de-

excitation from the excited 2p, levels to the 'So ground state of neon. The fast


























O

O








I-
o





f0
Cl




A
-4










5.4
Q



I-


0




0
C








C.
I-











4)





0
0


o
4)






L.
ur
ff














1_






























"-0




CD
l)
Ln




E

F-


A!P/Aw 002
































Figure 10. Partial energy level diagram of Ne.











cm-1


173 932





160 000







150 000









130 000



0


Ne'

//////////////////////////////////^^^/////////


4s


-4p


iSo


Ne









49
negative signal is most probably a result of increased ionization by electron collisions

at the beginning of the laser pulse.


Mathematical Treatment of OG Signals


Only a limited mathematical treatment of the collection of charges formed

under the influence of an electric field will be considered here. Complete

mathematical treatments of the various charge formation and collection phenomena

in the glow discharge as a whole, consistent with the present level of understanding

of glow discharges can be found in references [73-77].

Analogous to the case of double resonance laser enhanced ionization in

flames [78], the simplest rate equation model for the OG process involves

saturation of one transition and a linear dependence of the other transition (and

ionization rate) on laser intensity. The production rate of charge density is given by


an-i K I(xt) (11)
at



where I(xt) is the space-time distribution of the laser intensity, K is a constant which

encompasses the various excitation, ionization and deactivation constants and n. and

n, are the electron and ion number densities, respectively. From Poisson's equation

and the applied potential, the effective field, E(x), between the negative glow and the

cathode is given by [74]












E(x) 2-- 1-I (12)

where V0 is the absolute value of the applied potential, and d is the distance between

the cathode and effective anode (see Figure 11). An electric charge created under

the influence of this field is driven to an electrode according to the equation


dv(x, E(x)-j 0v,(xt) (13)
dt m



where

ve electron velocity (m s')

e = electron charge (C)

m = electron mass (kg)

Be = electron collision frequency (s1)

An analogous equation can be written for ions formed. From the principle

of conservation of energy [79], we can obtain an expression for the induced current

signal:


iVo e f Eo(nv neve) d(A6) (14)
A8

where

E = electric field due to the voltage difference V0 (V/m)

vj, = electron and ion velocities, respectively

(m s-1)








































C.)
U2








U,




I...






-cl





C.)










I-i








52












ea)
o





0

a1
rcl









53
A6 volume between the cathode and effective anode where charges

are moving (m3)

i current signal (A).



E0 is constant and is given by


Eo (15)



where d is the width of the dark space. Neglecting the time-dependent behavior of

the current, and accounting for charges with different velocities being present in a

given position x, at the same time t, the integral in equation (14) can be replaced by

a sum over all charged particles;


i E( nv, n v ) (16)
d Re



Simplistically, we can conclude from equation (16) that the total current is

dependent on several factors, including:

1) population of the originating level of the optical transition and the

rates of ionization processes from all primary energy levels involved,

through ni and ne, and

2) the collision frequencies of the charges, through vi and v,.











Double-resonance Optogalvanic (DROG) Effect


Little work has been done until recently on the double-resonance OG effect.

It was first proposed by Vidal [80] in 1980. The DROG scheme serves to increase

the absolute magnitude of the OG signal by approximately one order of magnitude,

in the case of sputtered atoms. At the time of preparation of this dissertation, only

two papers had appeared in the literature on the double-resonance OG effect

[81,82]. Engleman and Keller [81] reported a 10- and 70- times double-resonance

enhancement of sodium and uranium, respectively, in the HCL, while Behrens et aL

[82] reported on the DROG signals of In and Ga.

For optical double-resonance OG spectroscopy, two lasers are chosen with

wavelengths corresponding to two successive transitions in the atom. In the case of

Na (Figure 12), the first laser, A 12, can be chosen at 588.995 nm and the second, 23,

is set at 568.822 nm. If the intensity of A 3 is chosen such that the transition is

saturated then every atom excited into level 2 will be further excited into level 3 with

unity probability. As a result, level one is coupled directly to level three. In a low-

energy environment and an atom with a high first excited level 2, the effective

population of level 3 is orders of magnitude greater when both 12 and A 3 are tuned

to resonance and applied to the atom reservoir, than if only A 12 (or 23) is applied.

This is a very well known effect and has been used extensively in flame laser

enhanced ionization (LEI) [83]. Double-resonance LEI has proven to be one of

the most sensitive spectroscopic techniques available. As discussed in the previous
































Figure 12. Partial energy level diagram of Na.























.86 eV






Tc'z
aj aj











I aI
C i


cm-1


41 450

34 549













16 973

16 956


2 3/2












2p
1/2


S3/2








2 /2
1/2









57
section, this extremely high sensitivity is due mainly to the high energy levels to which

an atom can be promoted by the successive absorption of two resonant photons.

If one of the more important mechanisms in maintaining the discharge is

energy transfer from the excited species to electrons through inelastic collisions, then

by exciting atoms to much higher energy states, a larger energy increase is imparted

to the electron population. This in turn increases the number of electrons with

sufficient energy to cause ionization of the buffer gas. Since the ionization rate

increases with the exponential of the electron energy, it is a significant effect on the

ionization enhancement in the negative glow region of the discharge.














CHAPTER 5
CONSIDERATIONS ON THE INTERACTION OF ATOMS WITH LIGHT



Absorption of Radiation


Upon passing polychromatic light through an assembly of atoms, it is observed

that distinct wavelength components of the light are absorbed by the atoms. The

wavelengths that are absorbed give a qualitative determination of the atoms present

in the assembly and, in certain cases, the amount of light absorbed is directly

proportional to the number of atoms of that kind present in the assembly. In the

case of atomic absorption analysis, and under an explicit set of assumptions, the most

used relation between the density of atoms and the amount of light absorbed is

Beer's Law (17)


# Goe-4 (17)



where o is the flux of the incident photons (photons s-'), t is the transmitted flux

(photons s'), k(A) is the absorption coefficient (cm-') and I is the thickness (cm) of

the absorber. The magnitude of k(A) is dependent on the wavelength of the incident

radiation, and the nature and concentration of the absorber. It is often more

convenient to express the absorption coefficient, k(A), as the product of the atomic









59
number density of absorbers, na (cm3), and the absorption cross section, a (A) (cm2),

of the particular absorption transition. The measured parameter in an atomic

absorption measurement is the absorption factor, a, defined as


S- (18)
00



or the fraction of the incident light that is absorbed. Integrating over the entire

spectral range of the incident radiation, yields


S- (19)




For a narrow spectral line source, the spectral profile of the source is much narrower

than that of the absorption coefficient, and equation (19) reduces to


aL 1 e-(?') (20)



In the limit where the factor a (A)nal << 1, aL is directly proportional to the

absorber number density. If the factor a(A)nl > 1, then aL = 1, and virtually all

resonant photons incident on the atom assembly are absorbed.













Laser Excitation of Atomic Transitions


The interpretation of the interaction of a laser beam with an assembly of

atoms is adequately and most simply described by a rate equations approach.

Referring to Figure 13, one sees that there are many process which the atom can

undergo upon interaction with radiation whose energy exactly corresponds to an

energy difference between energy states of the atom. The rate equations approach

will be described here only for the simplest case of single-resonance excitation by

absorption of one photon with deactivation from the excited level possible by

collisional ionization, stimulated emission, spontaneous emission and collisional

deactivation back to the originating level of the transition (taken to be the ground

electronic state of the atom). Also, as the quantity measured in this work was the

number of ions created during interaction of the laser beam with the atom assembly,

the rate equations will be solved for the change in number of ions with time.

Single-step excitation

The rate equations for excitation of an atom from level 1 to level 2 are:


dn,
nz2n2 (21)

d n(B12Pnp(X12) + k12)- n2(k,~ + A21 + ik + Bp.On))





















03
(U


,1



















C
(l
I














0
4>
*-*












I-
(d
C
4-









co





4C















I-
.0




























(cVfac


I' l











where

A21 rate of spontaneous emission (s-1)

B12 = Einstein coefficient for stimulated absorption (J-' Hz m3)

k21 = rate of collisional deactivation (s-1)

k2i s rate of collisional ionization (s-1)

PA (112) spectral energy density of the laser

(J Hzi1 m3)at wavelength 1 12



Using standard methods to solve these equations and assuming the transition

is not saturated (B12p (X12) < < A21 + k2) yields [84]


n, nr 1 e B12PA (22)
h21 + k + k .


It is clear from this equation that the ratio between the number of ions formed

during the laser pulse and the total atomic number density is dependent on the rate

of collisional ionization, the duration of the laser pulse the Einstein coefficient for

induced absorption and the rates of relaxation from level 2. Also, the ratio is

dependent on the spectral energy density of the laser.

Optical Saturation of an Atomic Transition

The excited state population of an atom can be greatly enhanced relative to

the thermal population of that state by resonant absorption of photons from a laser

beam. In fact, if the spectral irradiance of the laser beam is sufficiently high, the








64
populations of the two levels will be locked together in the ratio of their respective

degeneracies [85]. This spectral energy density is termed the saturation spectral

energy density (J m"3 Hz-1) and is given by [84]




h [A:'+k21+K I +(g2V +klI" -"
p (12)-/8- 7- t (23)




where

h Planck's constant (J s)

ki = rate of recombination (s"1)

gu degeneracy of upper level (dimens.)

gi degeneracy of lower level (dimens.)


If we achieve optical saturation, B12P(X12) > > (A21 + k21 + k2i), then the

rate of induced absorption is much greater than the sum of the deexcitation rates,

and


n,-nT 1 exp -g+g2 r (24)
I' 82









65
Although ni/nT is now independent of the spectral energy density, the ratio is still

directly proportional to the product of k2i& t,. The value of k2i depends on how close

level 2 is to the ionization continuum. Therefore, we can expect that the closer level

2 is to the ionization continuum, the greater the production of ions. However, for

nanosecond pulsed lasers, even saturation of the transition is not expected to be

significant because of the large energy deficit between level 2 and the ionization

continuum resulting from only single-step excitation. If there is a metastable

electronic state present in the atom which can act as a radiation trap, such as in the

case of Pb, then equation (24) becomes




n- n7k. 1 e -g2 (k.+A+ky) At,l (25)
k2A2+k g[g 1+ 2



In this case, even if the laser does saturate the transition, the ionization rate must

now be fast enough to overcome the loss rate from level 2 to the metastable trap.

Two-step Excitation

When two laser beams, tuned to different absorption transitions, are made

spatially and temporally coincident in the discharge, high lying atomic levels can be

efficiently populated. Under such double-resonance conditions, four distinct cases

are readily apparent:

1) linear dependence of ni on the spectral irradiance of both laser beams,









66
2) saturation of the first step, 1-->2 (p(;A12z)>>p),(,.12), and linear

dependence of ni on p,(A.2),

3) linear dependence of ni on pX (12) and saturation of the second step

{(P 23) >> P.23)}, and

4) Saturation of both steps.

Since we are primarily concerned with detection of low levels of photons

corresponding to i x, cases (2) and (4) are not applicable in our system. Either case

(1) or case (3) can be operative, case (3) being preferred.

Assuming that the rate equations approach is valid in our experiments,

Omenetto et aL [84] derived the expressions for the fraction of atoms ionized for

these two cases. Assuming collisional ionization to proceed instantaneously from

level 3, in case (1), where saturation conditions for neither step have been met,


n-ny 1-exp (- Bx()B,3(X3) At (26)
"21 k* l I
the number of ions created during the laser pulse, ni, is dependent on the spectral

energy densities of both laser beams and on all de-excitation processes which deplete

the first excited level, 2.

In the case of resonance ionization detection, the second step will always be

saturated, while the first step will be linear (since the photons being measured are

at A 12). For this case, assuming no collisional or radiative losses from level 2, the

number of ions produced is given by











n,-n [ 1 exp ( -B12PA(X12)At,)] (27)

In equation (27) it is also assumed that the atoms reach a level whose energy defect

with the ionization potential is so low that collisional ionization proceeds

instantaneously. This is a valid assumption in most double resonance excitation cases

of atoms in an energetic environment (e.g. flame, plasma, electrical discharge).

Therefore, in such a case, the number of ions created is directly proportional to the

laser pulse length and the spectral energy density (B12p, ( 12)) of the photon flux of

A12 Also, we see no dependence of ni on the spectral energy density of Az, or on any

de-excitation pathways from level 2, since Bzp, ( 3) > > A21 + k21.

Two conclusions that can be drawn from the above discussion are that, for

significant ionization from the uppermost laser excited level, a two-step excitation

scheme must be used and that, for unity ionization of all atoms excited to level 2 by

absorption of a photon (, 12), the spectral energy density of the second laser, p, (.23),

must be sufficient to saturate the transition.

Assuming some typical values (Table 1) for the variables in equations (26) and

(27) we find that in case of a linear interaction of both steps,


i- 10-2 (28)
nT

while for saturation of the second step


1 (29)
nT






















Values used for calculating ni/nt in eauqtions (26) and (27).


Variable Value Units

B12 1018 J' m3 s-'Hz

B3 1018 -1 m3 s-'Hz

A21 109 s-1

k21 109 s-1

At, 10 s


PA ( 12) 10-5 J m-2 Hz


Table 1.














CHAPTER 6
EXPERIMENTAL



General Experimental Configuration


The general experimental system used for all experiments is shown in Figure

14. A detailed listing of experimental components is given in Table 2. A frequency

doubled Nd:YAG laser (532 nm) operated at 30 Hz was used as the pumping source

for the dual dye laser system. The pump beam was split equally to pump each dye

laser. Typical output energy from the frequency doubled Nd:YAG laser was 240 mJ

per pulse with a pulse duration of 12 ns. The dye laser output was either used

directly for transitions in the visible region of the spectrum or frequency doubled for

UV transitions. In the case of UV transitions, the visible laser light was passed

through a KDP frequency doubling crystal; an autotracking system with angle

matching of the frequency doubled light ensured maximum output intensity while

scanning. The fundamental and second harmonic waves were separated by a

dispersive Pellin-Broca prism. If the laser wavelength was to be scanned, then the

Pellin-Broca prism was replaced with a right angle prism to prevent "walking" of the

beam during scanning. The two harmonics were then separated with a low bandpass

colored filter, which absorbed the visible wavelengths.

























































0
4-







1-
co
E




'-




a




F4
























































.4














Listing of experimental components.


Component Model No. Manufacturer
Nd:YAG laser YG 581-30 Quantel International, Santa
Clara, CA'
Dual Dye Laser TDL 50 Quantel International, Santa
Clara, CA'
Frequency Doublers HD 50 Quantel International, Santa
Clara, CA1
900 Quartz Prisms and Esco Products, Inc., Oak
Quartz Lenses Ridge, NJ
Neutral Density Filters -- Corion Corp., Hollistong, MA
High Power Laser Neutral FN-10,30,10,80 Optics for Research, Caldwell,
Density Filters NJ
Amplifier 113 EG&G PARC, Princeton, NJ
Boxcar Averager Gated SR250 Stanfor Research Systems, Palo
Integrator Alto, CA
Computer Interface SR245 Stanford Research Systems,
Palo Alto, CA
Computer PC-AT Northgate Computer Systems
Digital Oscilloscope 2430A Tektronix, Inc., Beaverton, OR
Chart Recorder D-5000 Houston Instruments, Austin,
TX


1 Now Continuum, Santa Clara, CA.


Table 2.
























Table 2.


--- continued


Component Model Manufacturer
Hollow Cathode Lamp L233 series Hamamatsu Corp.,
(Lamp 1) (Na, Pb and U) Bridgewater, NJ
Galvatron (Mg) L2783-12NE- Hamamatsu Corp.,
Mg Bridgewater, NJ
Hollow Cathode Lamp PMT-20A/N Bertan Associates, Hicksville,
Power Supply NY
Fast Photodiode ET 2000 Electro-Optics Technology,
Fremont, CA
Photodiode for Absorption PIN 10DP-SB United Detector Technologies,
Measurements Hawthorne, CA








74

Two commercially available hollow cathode lamps were used with different

hollow cathode designs. One (lamp 1) was a common HCL used for atomic

absorption analysis (Figure 15). In this lamp only one end of the hollow cathode was

open. The other lamp (lamp 2) (Figure 16) was a "Galvatron" used in wavelength

stabilization of dye lasers by the optogalvanic effect. This lamp is a "T" design in

which both ends of the hollow cathode are open.


Pick-off Circuitry for OG Signal


The circuit used to power the hollow cathode lamps and measure the AC

voltage across the discharge is shown in Figure 17a. It was entirely housed in a

shielded Pomona box (Figure 17b) to reduce pick-up of radio frequency noise. The

high voltage was supplied from a dc-dc high voltage power supply through a current

limiting resistor, RB. This resistor could be made external with the use of an isolated

BNC connector. A variable resistor within a decade resistor box was used for initial

studies. For final experiments, the optimum RB (20 kn) was hard-wired inside the

shielded box. The voltage across the discharge was AC coupled to the detection

electronics by a high voltage coupling capacitor.


One-step Excitation Experiments


For one-step excitation, laser 1 was steered via right angle prisms into the

discharge. The laser beam could be focused, depending on the experiment to be

performed. When the laser was not focused, it was apertured to just fill the hollow


















































E





0
1-1





ol
a














wi
QL
















ji-T-P ^
Eu U



o "7
LI' b

/ 3
I a
































Figure 16. Diagran of a Galvatron (lamp 2).









Cathode Anodes


Optical /
axis




Insulated ^y Insulated
support I support














Negative Dark
glow space


LasezL



Detail of
hollow cathode






























































o
05
So.



fo

S 0



Q-0


Q)*'


t~-











*8


p
on 6








81
cathode volume. Optogalvanic spectra were obtained by scanning the dye laser

wavelength while monitoring the AC voltage across the discharge.


Two-step Excitation Experiments


Lamp 1

The alignment of the lasers when using the single-open ended hollow

cathode is shown schematically in Figure 18. The two laser beams were made to

enter the hollow cathode at a slight angle to the cathode axis. Different focussing

configurations of the two beams were used. Placement of a lens at position 1

allowed only laser 1 to be focused, while placement of the lens in position 2

focused both beams into the hollow cathode. In either case, the position of the lens

was such that the beam(s) was (were) focused at the back surface of the cathode.

Lamp 2

The alignment of the lasers when using the double-open ended hollow

cathode is shown in Figure 19. In this case alignment is much easier since the beams

can be made counter-propagating, entering the cathode from opposite ends. Again,

different focussing configurations were used.


Timing of Laser Beams in Two-step Excitation Experiments


For either lamp, temporal coincidence of the two laser beams in the hollow

cathode was ensured with a fast photodiode (risetime <200 ps). The output of the

photodiode was connected to a fast digitizing oscilloscope. For timing optimization,






























































0

C(



c


0




1-1
o
















od













0.
.O
4a 0

2


<0-
u.


f^
-2 I


0
0 s
0 3
44
cis
pw


MIMI,-


0


u o


Ile






























































0

'-i





0
E













o,
Y.-
I-








c;
Q\























ta)
z


a)
T3
0


C-
u



























cu
ro
0
-4J














oj

cd
u
\d














()
C.)
04
Q,

o
5-4
'a








86

the fast photodiode was placed at the position to be occupied by the HCL Then

laser 1 and laser 2 were alternately made incident on the photodiode. The arrival

time of the two lasers at the photodiode was then noted. The laser pulse arriving

earlier at the photodiode was delayed by way of a prism delay line until temporal

coincidence with the other laser pulse at the photodiode was observed. To avoid

skewing of the measurements toward longer times, care was be taken to ensure

operation of the photodiode was within its linear operating range. Figure 20 is an

oscilloscope trace of the photodiode output. The two traces were obtained

consecutively for laser 1 and laser 2. Temporal coincidence of the two laser beams

was within 1 ns.


Absorption experiments


The experimental setup used to make absorption measurements (with lamp

2 only) is shown in Figure 21. The transmitted laser light was incident on a

photodiode while the laser was scanned through the transition of interest. Linearity

of the photodiode response was insured with a 0.3 neutral density filter. The output

of the photodiode was fed into the input of a boxcar. Output from the boxcar was

then sent to a personal computer and strip chart recorder for subsequent data

analysis.