Title: Pulsed laser-excited fluorescence
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00098925/00001
 Material Information
Title: Pulsed laser-excited fluorescence
Physical Description: xi, 163 leaves : ill. ; 28 cm.
Language: English
Creator: Weeks, Stephan John, 1950-
Copyright Date: 1977
Subject: Fluorescence   ( lcsh )
Spectrum analysis -- Instruments   ( lcsh )
Lasers   ( lcsh )
Chemistry thesis Ph. D
Dissertations, Academic -- Chemistry -- UF
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
Statement of Responsibility: by Stephan John Weeks.
Thesis: Thesis--University of Florida.
Bibliography: Bibliography: leaves 156-162.
General Note: Typescript.
General Note: Vita.
 Record Information
Bibliographic ID: UF00098925
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000201622
oclc - 03887586
notis - AAW8378


This item has the following downloads:

pulsedlaserexcit00week ( PDF )

Full Text







This work is dedicated to the ones

whose love means so much to me.


I would sincerely like to express my true appreciation to my friends

and colleagues, who through their communication of knowledge and friend-

ship have made this period of my life rewarding and memorable. T partic-

ularly want ti express my thanks to Dr. James D. Winefordner and Dr.

Hiroki Haraguchi for their help and example.

I also wish to thank my family and friends for providing continual

encouragement and understanding.



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

LIST OF FIGURES . . . . . .

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


1. INTRODUCTION . . . . . . .


Types of Fluorescence Transitions. . . . . .

Atomic Fluorescence Radiance Expressions . . . .

II). EXPERIMENTAL . . . . . .

Excitation Source . . . . . .

Detection System . . . . . . ...

Experimental Conditions and Procedure . . . .


Optimization of System . . . . . . .

Analytical Figures of Merit . . . . . . .


Flame Background . . . . . . .

Selective Excitation of Molecular Species in Flames

Laser-Excited Molecular Fluorescence of CaOH and SrOH























REFERENCES . . . . . . . . . ... .. 156

BIOGRAPHICAL SKETCH . . . .... .... . . . 163


Table Page

1 Experimental Components of Laser-Excited Atomic
Fluorescence Flame Spectrometric System 49

2 Molectron Model UV 400 N2 Laser Specifications 51

3 Dye Laser Specifications 56

4 List of Laser Dyes 60

5 Optimization of Optical Arrangement in Laser-
Excited Atomic Fluorescence Spectrometry 73

6A Detection Limits by Laser-Excited Atomic
Fluorescence Flame Spectrometry 83

6B Detection Limits by Laser-Excited Atomic
Fluorescence Flame Spectrometry 84

7 Comparison of Detection Limits in Flame Spectro-
metry and Inductively Coupled Plasma (1CP) 85

8 Investigation of Spectral Interferences Between
Manganese and Gallium 97

9 Spectral Transitions of BaC1 Green System C2 i-X2 111

10 Relative Intonsity of Peak Fluorescence for BaC1
Transitions for the Observed Fluorescence
Excitation Spectra 115
2 2
11 Approximate Energy Transitions of A 2-X2 (i)
and B2E-x2E (ii) 146


Figure Page

1 Types of atomic fluorescence transitions 14

2 Schematic diagrams of cell assumed for luminescence
radiance expressions 18

3 Growth curves in atomic fluorescence 22

4 Two- and three-level atomic systems 30

5 Schematic diagram of laser-excited atomic
fluorescence flame spectrometry system 48

6 Peak, average, and rms power of N2 laser v:
repreition rate 53

7 Optical schematic diagram of dye laser cavity 54

SA Wavelength tuning curves for dyes 58

8B Wavelength tuning curves for dyes 59

9 Optical arrangement around the burner for the
examination of optimization of the optical system 72

10 Excitation and fluorescence profiles of sodium
D lines 76

11 Dependence of atomic fluorescence signal of Na and
T1 on slit width in the ari-acetylene flame 77

12 SNR vs Slit width of Na and T1 in the air-
acetylene flame 79

13 Analytical calibration cirves for elements having
only strong resonance transitions above
355 nm in an air-acetylene flame 88

14 Analytical calibration curves for elements having
both strong resonance and nonresonance transitions
excited above 355 nm in an air-acetylene flame 90


Figure Page

15 Analytical calibration curves for elements having
transitions excited above 355 nm in a nitrous
oxide-acetylene flame 92

16 Analytical calibration curves for elements having
transitions excited below 355 nm in an air-
acetylene flame 94

17 Excitation fluorescence spectrum for: (A) a 1 ppm
Mn solution; and (B) 1 ppm Mn + 4 ppm Ga solution 98

18 Excitation spectrum of CN in a nitrous oxide-
acetylene flame 102

19 Emission spectrum of barium in the air-acetylene
flame 108

20 Energy level diagram of BaC1 showing observed
transitions 110

21 Fluorescence excitation spectra of BaC1 in an
air-acetylene flame 114

22 Fluorescence excitation spectra of BaOH and
BaC1 in an air-acetylene flame 118

23 Fluorescence excitation spectrum of BaOH in
air-acetylene flame 120

24 Fluorescence excitation spectra of BaOH in
an air-acetylene flame 123

25 Fluorescence excitation spectrum of BaO in an
air-acetylene flame 125

26 Fluorescence excitation spectra of BaOH and BaO
in an air-acetylene flame 128

27 Fluorescence excitation spectra of BaOH and BaO
in an air-acetylene flame 130

28 Emission spectrum of CaOH in an air-acetylene flame 133

29 Fluorescence emission spectra of CaOH in an
air-acetylene flame 136

30 Fluorescence excitation spectra of CaOH in an
air-acetylene flame 139


Figure Page

31 Fluorescence excitation spectra of CaOH in an
air-acetylene flame 141

32 Near-resonance fluorescence excitation spectra of
CaOH in an air-acetylene flame 143
7 2
33 Preliminary energy level diagram for A 1-X 2 (i)
and B2X-X2E (ii) systems for CaOH, showing
the sequences 145

34 Fluorescence emission spectrum of SrOH in an
air-acetylene flame 148

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



Stephan John Weeks

December 1977

Chairman: James D. Winefordner
Major Department: Chemistry

Laser-excited atomic fluorescence flame spectrometry with a pulsed

nitrogen laser pumped tunable dye laser has been investigated especially

in terms of improvement of the detection limits. The detection limits

for Ag, Ba, Bi, Ca, Cd, Co, Cr, Cd, Fe, Ga, In, Li, Mg, Mn, Na, Ni, Pb,

Sr, and T1 in an air-acetylene flame and for Al, Ba, Mo, Ti, and V in a

nitrous oxide-acetylene flame are evaluated in the cases of both reso-

nance and nonresonance atomic fluorescence lines. The detection limits

obtained in the present study have been improved by about 10 to 200 times

over those obtained previously by a similar laser-excited atomic fluores-

cence flame spectrometry system. Detection limits for many elements and

lines examined using the frequency doubled dye laser output are reported

for the first time. Now, most elements examined can be detected at the

ng-.nl-1 (ppb) level or less with ranges of analytical curve linearity of

over five orders of magnitude. The improvement of the detection limits

was achieved mainly by expanding the diameter of the laser beam to illu-

minate a larger analyte volume in the flame and by using a low dispersive

spectrometric system. The optical arrangement employed reduced the

scatter signal and improved the signal-to-noise ratio (SNR) even at the

resonance fluorescence wavelengths compared to the previously used sys-

tem. Although a low dispersive spectrometric system was employed, the

spectral resolution of the system was determined by the bandwidth of the

laser (ca 0.03 nm). Conditions for optimizing analytical results are

discussed, and laser-excited atomic fluorescence flame spectrometry is

shown to be superior or equivalent to atomic absorption flame spectro-

metry, atomic emission flame spectrometry, atomic emission induction

coupled plasma spectrometry, and conventional source atomic fluorescence

flame spectrometry.

In addition, the utility of this laser fluorescence system in mo-

lecular flame fluorescence studies was examined. Molecular fluorescence

spectra of BaO, BaOH, and BaCl in an air-acetylene flame have been inves-

tigated and compared to the flame emission spectra of the same species.

Molecular fluorescences of BaO and BaOH in the flame were observed for

the first time. Furthermore, BaO, BaO!I, and BaC1, which have overlapping

emission spectra in the wavelength region from 480 to 535 nm, can be

selectively excited and observed by laser-excited molecular fluorescence

spectroscopy. Also, pulsed laser-excited molecular fluorescence of CaOH

and SrOH in an air-acetylene flame have been investigated. The spectra

were examined in the wavelength region from ca. 500 nm to 700 nm. A pre-

liminary energy level diagram is proposed for CaOH. Results are also

giver, for SrOH.

S Chairman /


Atomic fluorescence spectroscopy has the potential of becoming the

most widely used method for trace elemental analysis. The technique is

based on the production of atoms, via any device which converts the

sample into atoms, such as a flame or furnace; the photon excitation of

the atoms via a suitable primary radiation source, such as a xenon arc

lamp, electrodeless discharge lamp or laser; the absorption of the ex-

citing radiation by atoms existing in the ground state energy level or

a highly populated energy level producing atoms in a higher excited en-

ergy level which undergo radiationalde-excitation over a period of time

on the order of nanoseconds after excitation; and the subsequent detec-

tion and measurement of the induced radiation. The figures o- merit

achievable with the atomic fluorescence spectroscopic method, the sim-

plicity of the experimental arrangement, and the applicability to both

single and multielement trace analysis continue to stimulate develop-

ment of this technique.

tWinefordner (1) has recently written a brief review concerning the

history, principles, instrumentation, methodologies and applications of

atomic fluorescence spectrometry. Despite the fact that the physical

principles upon which atomic fluorescence is based were known in the

1800's (-6), the characteristics of atomic fluorescence were observed

and described in the early 1900's (7-9); and that atomic absorption

underwent rapid development in the 1950's and 1960's (10), it was not

until the early 1960's after Alkemade (11) reviewed the methods by which

atoms undergo excitation in flames and described the use of atomic fluo-

rescence flame spectrometry for measuring quantum efficiencies, that

Winefordner and co-worker (12-14) pointed out and demonstrated the first

analytical use of atomic fluorescence spectroscopy. Since this time,

extensive research has been done, both theoretical and practical, in

order to optimize the experimental conditions and arrangement for chem-

ical analysis by both flame and nonflame methods. This is evidenced by

the large number of references cited in recent reviews and books (1, 15-

23). Because of the linear dependence of the fluorescence signal on the

concentration of the isolated analyte atom and on the spectral irradiance

of the excitation sources for nonlaser sources, most research in analyt-

ical atomic fluorescence spectroscopy has been directed towards the op-

timization and development of high efficiency atomizers (24-31.) and high

radiance excitation sources (32, 33). The impetus behind this research

has been to improve the limits of detection and other figures of merit

(e.g., linear dynamic range) for the atomic fluorescence method.

After the first laser was demonstrated (34) in 1960, spectroscopists

inmnediately realized the usefulness of lasers. However, it was not until

lasers became tunable that they became useful as an analytical tool for

spectroscopy. The greatest advances in tunable lasers for analytical

spectroscopy came after the demonstration of the organic dye laser (35,

36). Several reviews and books describe the developments and properties

of unable lasers (37-42). The lasers considered to be of most interest

to analytical chemists for studies in atomic spectroscopy are the ion-

laser pumped continuous wave (cw) dye laser; the flashlamp pumped dye

laser; and the nitrogen (N 2) laser (or Nd :YAG laser) pumped tunable

dye laser. Of these lasers the N2 laser pumped dye laser possesses the

widest tuning range (from ca 220-950 nm) and the most ease of changing

dyes rapidly during operation. The "tuning" of wavelength in the N2

laser pumped dye laser is accomplished by changing the lasing media

(i.e., the dye) and the angle of the grating, which acts as one of the

laser cavity mirrors. Because of the high spectral irradiances, narrow

spectral bandwidth, wide range of wavelength tunability, coherent pro-

perties of the output beam, and short temporal duration of the pulses,

the N2 laser pumped tunable dye laser seems to be nearly an ideal source

for atomic fluorescence spectroscopy.

In 1971, Fraser and Winefordner (43) reported the first application

of a tunable dye laser used as a primary radiation source in atomic fluo-

rescence spectroscopy. A N2 laser pumped dye laser was used to excite

nine different elements in air-hydrogen and nitrous oxide-acetylene

flames. The detection limits obtained ranged from an improvement by a

factor of ten to within fifty times that of the best reported literature

values for conventional source atomic fluorescence. Besides low detec-

tion limits, they reported long linear dynamic ranges and freedom from

flame background noise and also demonstrated that the spectral resolu-

tion was determined only by the spectral bandwidth of the laser output.

The dye laser source used had a peak power greater than 10 kN, a pulse

repetition rate of 1-25 Hz, a spectral bandwidth of 0.1-1 nm, and a tem-

poral pulse width of about 2-8 ns. A 1P28 photomultiplier tube and a

boxcar integrator (gated amplifier) with a 10 ns gate width were used

for signal detection and processing.

Tn the same year, Denton and Malmstadt (44) also demonstrated the

use of tunable organic dye lasers as excitation sources for atomic flame

fluorescence spectroscopy. They reported atomic flame fluorescence of

barium excited by the output of a dye laser pumped by the second harmonic

of a Q-switched ruby laser. They found conventional total consumption

burners to be totally unusable due to high light scattering levels. An

ultrasonic nebulizer was found to reduce the scatter level and allow

linear calibration curves to be obtained. They also noted that consid-

erable care had to be used to prevent radiation scattered from the en-

vironment (i.e., not the observed analyte volume in the flame) from

striking the detector.

In 1972, Fraser and Winefordner (45) obtained limits of detection

ranging from better to within tenfold of the best literature values by

any other atomic flame spectrometric method. They also noted several ad-

ditional advantages pulsed laser excitation has over conventional contin-

uously operated line or continuum sources including the following: only

one source is needed; freedom from analyte emission; and the simplicity

of using nonresonance fluorescence to eliminate noise due to scatter of

exciting radiation. Nonresonance fluorescence measurements yielded about

a ten-fold improvement in the detection limits over resonance fluores-

cence. They observed several types of fluorescence including: resonance

fluorescence; excited resonance fluorescence; Stokes direct line fluo-

rescence; anti-Stokes direct line fluorescence; and excited anti-Stokes

direct line fluorescence. They utilized the same N2 laser pumped tunable

dye laser atomic fluorescence flame spectrometric system as described

above, except that the dynode chain for the photomultiplier tube base

was modified for high current pulsed operation and the output of the

boxcar integrator was fed into an additional analog integrator in order

to minimize pulse-to-pulse variations. Noise sources were briefly dis-

cussed and interference due to CN formed in the nitrous oxide-acetylene

flame was noted.

Since these initial studies, tunable organic dye lasers have been

increasingly used as excitation sources for atomic fluorescence spectros-

copy with both flame atomizers and nonflame atomizers. In these studies,

pulsed (nitrogen, ruby, or Nd+3:YAG laser pumped and flashlamp pumped)

or continuous wave (cw) dye lasers have been employed as excitation

sources for atoms. These studies point out additional advantages and

uses of laser-excited atomic fluorescence spectroscopy, as well as some

of the limitation.

Omenetto et al. (46, 47) examined the atomic fluorescence of some

transition elements and atomic and ionic fluorescence of some rare earth

elements in the nitrous oxide-acetylene flame with a N2 laser pumped

tunable dye laser. Very good results were obtained and the use of non-

resonance fluorescence measurements were stressed.

Because lasers aLe highly coherent sources of radiation, the al-

ready high spectral irradiance output can easily be focused to even

larger power densities. Omenetto et al. (48, 49) and Piepmeier (50, 51)

were the first to visualize the influence of saturation of atomic levels

due to the intense laser excitation, and derive the theories necessary

to investigate saturation conditions. Kuhi et al. (52) discussed the

influence of saturation on the atomic fluorescence measurements obtained

using a flashlamp pumped tunable dye laser. Complete saturation occurs

in a two-level system when the ratio of the populations in the upper

energy to that of the lower energy level becomes equal to the ratio of

the statistical weights for the two levels. Ordinarily, the fluores-

cence radiance, when excited by conventional low intensity light sources,

is linear with source intensity. However, this relationship will hold

only up to the point which the source intensity becomes large enough to

alter the thermal atomic equilibrium distribution which exists in the

atomizer (i.e., flame or other plasma). Saturation or near saturation

conditions offer a number of advantages including: greatly enhanced

signals, less dependence on source intensity and its fluctuations, and

less dependence on quantum efficiency.

In 1973, Kuhl and Spitschan (53) were the first to utilize fre-

quency doubled dye laser output to obtain fluorescence signals from Mg,

Ni, and Pb at wavelengths below ca 350 nm. Because the fundamental

output of dye lasers currently lies above ca 350 nm, nonlinear frequency

doubling crystals must be used in order to investigate the many strong

atomic transitions which lie below this wavelength. A flashlamp pumped

dye laser was used in the above studies to excite atoms formed on a flame

atomizer. To date, no frequency doubled detection limits have been re-

ported using a N2 laser pumped dye laser atomic fluorescence system.

D'e Olivares (54) has investigated several practical and fundamental

aspects of laser-excited atomic fluorescence spectroscopy (AFS) using

a N2 laser pumped dye laser.

Green et al. (55) and Smith et al. (56) have recently used cw dye

lasers to excite atoms in flames. The cw dye laser should be an optimal

excitation source because of the higher average output power, the greater

ease of modulation, and the increased ease of signal processing when com-

pared co the pulsed dye lasers. Good detection limits were reported for

sodium and barium and some nonresonance excited state transitions were

observed. However, the available wavelength range of cw dye lasers is

presently limited to the longer visible wavelength region, primarily

above 520 nm, making cs dye lasers of limited analytical use. Pulsed N2

laser pumped tunable dye lasers, on the other hand, offer the greatest

advantages to date, primarily because a wavelength range from ca 220 to

950 nm can be quickly and easily obtained.

Other recent studies in flames have used flashlamp pumped dye lasers

to excite Na atoms. Again, the advantages of laser-excited nonresonance

fluorescence were pointed out (57) and saturation was investigated (58).

In 1971, Kuhl and Marowsky (59) observed fluorescence in a sodium

vapor (nonflame) cell with a flashlamp pumped dye laser. Later, Fairbank

et al. (60) were able to detect 100 atoms-cm-3 in a sodium vapor cell

with a cw dye laser. In other works (61, 62) where a sodium vapor cell

was employed, flashlamp pumped dye lasers were used to obtain additional

theoretical analysis. Sodium also has been investigated in a quartz

tube and furnace using a cw dye laser (63).

In 1974, Neumann and Kriese (64) reported the detection of subpico-

gram amounts of Pb by nonflame atomic fluorescence spectrometry with a

frequency doubled flashlamp pumped dye laser. More recently, Bolshov

et al (65) reported subpicogram amounts of Pb and Fe by graphite tube

atomic fluorescence analysis using the grequency doubled radiation of a

dye laser pumped by the second harmonic radiation of a Nd3 :YAG laser.

The absolute detection limits obtained in this study are at present the

lowest achieved by any atomic spectroscopic method. In the case of Pb,

saturation was reached. In another study (66) picogram detection of Cs

in a graphite furnace was reported using a N2 laser pumped dye laser to

excite the atoms. Nonresonance fluorescence was utilized.

Several works (67-69, 54) have examined both theoretical and funda-

mental characteristics of saturation of the energy levels of atoms.

Steady state expressions were given for two and three level atomic systems.

Steady state expressions were given for two and three level atomic


The research and theoretical studies performed to date have indi-

cated the many advantages of laser-excited atomic fluorescence spectros-

copy and the great potential of this method for both fundamental research

and practical analysis. So far, laser-excited atomic fluorescence spec-

troscopy with a N2 laser pumped dye laser has produced detection limits

comparable to those obtained by flame emission spectrometry and atomic

absorption flame spectrometry (45-47). Detection limits in laser-

excited atomic fluorescence spectrometry have been comparable with those

obtained in atomic fluorescence flame spectrometry using conventional

narrow line light sources, such as hollow cathode lamps, electrodeless

discharge lamps, and xenon arc continuum sources (70, 71). Unfortunately,

previous results obtained with laser excitation have not been as good as

expected based upon the high intensity output (72). Although many fac-

tors influencing the results have been noted, the experimental arrange-

ment and conditions in laser-excited atomic flurescence spectrometry need

to be more fully and carefully investigated; a pulsed source with high

spectral irradiance combined with gated detection should be a nearly

ideal system for improvement of detection limits and signal-to-noise

ratios (SNR) of atomic fluorescence, as has been predicted (71, 73, 74).

During the investigation of atomic fluorescence detection limits,

strong spectral interference were noted in the 380 nm wavelength region

due to the formation of CN in the nitrous oxide-acetylene flame. Al-

though a N, laser pumped dye laser atomic fluorescence spectrometric

system using gated detection has negligible interference from flame

background or sample emission because of the low duty factor (ratio of

detector on time to the total time of analysis), ca 10 the atomic

fluorescence signal-to-noise ratio can be greatly affected by flame

background fluorescence and sample matrix effects. Knowledge of inter-

ferences from flame background fluorescence and matrix effects is ex-

tremely important for the practicing analytical chemist because often

it is possible to choose the lines used for analysis and the flame con-

ditions and composition to eliminate or minimize the interference with-

out having to use further sample pretreatment methods.

The observation of strong molecular fluorescence in flames indicated

the possibility of using the pulsed dye laser spectrometric system to

conduct a number of studies. These studies included determination of

the fluorescence background due to the flame gases and its dependence on

flame conditions, composition, fuel-to-oxidant ratio, and quenchers, and

a discussion on the possibility of removing spectral interference in

atomic fluorescence spectrometry based upon the results; the possibility

of utilizing molecular fluorescence for the analysis of nonmetallic ele-

ments, such as P, S, 0, C, N, and the halogens, which are difficult to

determine by atomic fluorescence because their atomic resonance lines

are in the far or vacuum UV; fundamental studies such as the identifica-

tion of molecular species via selective excitation, determination of en-

ergy levels and molecular parameters, study of reaction kinetics and

determination of quantum efficiencies; reduction of fluorescence back-

ground in Raman spectrometry; and combustion diagnostics including the

profiling of species concentration and flame or plasma temperature. In

this thesis, the flame background fluorescence of CN in the nitrous

oxide-acetylene flame will be reported, and the molecular fluorescence

of alkaline earth compounds in the perspective of utilizing the pulsed

N2 laser pumped tunable dye laser fluorescence spectrometric system for

the identification of molecular species via selective excitation and for

the determination of energy levels of molecular species formed in flames

or plasmas will be investigated.

The primary purpose of this research is to improve the detection

limits and obtainable signal-to-noise ratio (SNR) through the investiga-

tion of factors influencing the atomic fluorescence signal and the system

noise in a N2 laser pumped tunable dye laser atomic fluorescence spectro-

metric system. Equally important is the need to point out additional

figures of merit for the method such as long linear dynamic range of the

analytical calibration curve and lack of spectral interference which are

of tremendous importance to the practicing analytical chemist. A com-

parison with other trace elemental analysis methods is also necessary for

the analyst and is made here. Another equally important purpose is the

need to consider this as only an initial study and therefore state where

additional improvements could be made in the system. A pulsed laser ra-

diation source with high spectral irradiance, narrow spectral bandwidth,

and wide wavelength tuning range combined with gated detection should

prove to be an ideal system for elemental analysis.


[he usefulness of atomic fluorescence spectroscopy lies in the fact

that it is a method of chemical analysis that is both qualitative and

quantitative and that it has distinct advantages over atomic emission

and atomic absorption spectroscopy and other methods of atomic analysis.

These advantages must be put in perspective by a comparison of the fig-

ures of merit for each technique. Excellent discussions of quantitation

in analysis describing figures of merit can be found in literature (23,

75, 76). Analytical figures of merit refer only to a definite complete

analytical procedure, which is specified in every detail with respect to

the analytical task, the apparatus, the external conditions, the experi-

mental conditions, the experimental procedure, the measurement process,

the means of calibration and the evaluation of data (23). The figures

of merit of an analytical method include: limit of detection; linear

dynamic range of the analytical calibration curve; sensitivity; selec-

tivity; specificity; accuracy; and precision.

The limiting detectable measure, xL, is given by

XL = bl + ksbl

wheru x11 is the average value of the blank measure calculated from a

sufficiently large number of blank analyses (at least 16); k is a pro-

tectioa factor, generally accepted to be 3 (23,77); and sbl is the

standard deviation of the blank measures.

'he linear dynamic range of the analytical calibration curve is

given by the ratio of the upper limit of concentration, at which the sig-

nal deviates from linearity by 5%, to the concentration at the detection

limit, c where c = L bl = bl .

The sensitivity of the measurement procedure is defined as the slope

S of the analytical calibration curve when the concentration, c, is plot-

ted versus the signal intensity, x. The sensitivity is not redundant

with the detection limit.

Many articles and books give detailed discussions of the analytical

figures of merit, atomic fluorescence radiance expressions, the types of

atomic fluorescence, the shape of the analytical curves, nebulization,

atomization, and other theoretical and practical considerations as well

as comparisons of the various methods for analysis (1, 11, 12, 15-23, 31,

48-52, 54, 67-79).

Jypes of Fluorescence Transitions

Since the discovery of atomic fluorescence as an analytical tool,

various types of atomic fluorescence transitions have been utilized for

analytical studies. Due to the rapid development of atomic fluorescence

and with the advent of tunable dye laser sources, which have made it pos-

sible to observe many new fluorescence transitions, Omenetto and Wine-

fordner (78) have proposed a consistent nomenclature to identify atomic

fluorescence transitions, in order to avoid possible confusion in the

literature. Basically, there are five types of atomic fluorescence tran-

sitions, which are reported, together with their possible variation, in

Figure 1.

Figure 1. Types of atomic fluorescence transitions (the spacings be-
tween atomic levels is not indicative of any specific atom), a, reso-
nance fluorescence (either process); b, excited state resonance fluo-
rescence; c, Stokes direct line fluorescence; d, excited state Stokes
direct line fluorescence; e, anti-Stokes direct line fluorescence; f,
excited state anti-Stokes direct line fluorescence; g, Stokes stepwise
line fluorescence; h, excited state Stokes stepwise line fluorescence;
i, anti-Stokes stepwise line fluorescence; j, excited state anti-Stokes
stepwise line fluorescence; k, thermally assisted Stokes or anti-Stokes
stepwise line fluorescence (depending upon whether the absorbed radi-
ation has shorter or longer wavelengths, respectively, than the fluo-
rescent radiation); 1, excited state thermally assisted Stokes or anti-
Stokes stepwise line fluorescence (depending upon whether the absorbed
radiation has shorter or longer wavelengths, respectively, than the
fluorescence radiation); m, sensitized fluorescence (D = donor; D* =
excited donor; A = acceptor; A* = excited acceptor; hv = fluorescence
radiation); n, two photon excitation fluorescence (mulliphoton processes
involving more than two identical photons are even less probably than
the two photon processes). Taken from Reference 78.






i CO

D + hvE -- DT"

D'+ A -- A*+ D

A A + hvE

( I)


--------- 0




i 2







( j)
-- 3



Resonance fluorescence resbults when the same lower and upper energy

levels of the atom are involved in the excitation and de-excitation pro-

cesses. A radiationally excited atom reemits fluorescess) a quantum of

radiation of the same energy as was absorbed. The transition probabili-

ties tor resonance transitions involving the ground electronic state are

typically much greater than for nonresonance transitions. Because reso-

nance fluorescence signal intensities are generally significantly greater

than the intensities observed with other types of transitions, resonance

fluorescence has been shown to be most useful for routine chemical analy-

sis (80).

Direct line fluorescence results when the same upper level is in-

volved in the radiational excitation and de-excitatior, processes; however,

the photon energy emitted fluorescedd) is different from that which was

absorbed. Direct line fluorescence is most useful analytically when

scatter of source radiation (e.g., from particles within the analytP

flame volume) is the predominant source of noise in the spectroscopic

system. Direct line fluorescence has also proved to be useful for tem-

perature measurement and species and temperature profiling oL flame

atomizers (81-87).

.tepwise line fluorescence results when different upper level-: are

involved in the radiational excitation and de-excitation processes.

Again, the phocon energy reemitted fluorescedd) is different from that

which was absorbed. Stepwise line fluorescence has analytical usefulness

when !,e excitation source scatter is the predominant noise source and

there is no possibility or the transition probabilities are nor favorable

for tihe direct line fluorescence. Stepwise line fluorescence can be used

to study quenching mechanisms in flames (87).

'he remaining two types of a.omic fluorescence have only theoretical

interest at present in analytical flame spectrometry. Sensitized fluo-

rescence results when one species (called the donor) is excited by an

external light source, collides with an atom of the same or another spe-

cies (called the acceptor), transfers energy to the acceptor, after which

either may de-excite radiationaliy. The final type of atomic fluorescence

is multiphoton fluorescence. This type of fluorescence results when two

or more photons excite an atomic species to an upper energy level after

which radiativu de-excitation occurs. The excitation may involve either

virtual or real energy levels.

it the photon energy of fluorescence is less than the photon energy

of absorption, the process is called Stokes (type) fluorescence. If the

reve:o is true, the process is culled anti-Stokes (type) fluorescence.

If the radiational absorption and fluorescence processes involve only

excited states, then the word exc.,ted is introduced in front of the fluo-

rescence process. If after radiational excitation, radiationless pro-

cesses (e.g., collisions) populate either a higher or lower level from

which fluorescence occurs, then the phrase thermalLy assisted is inserted

prior to the name of thk. fluore-c'ence process (1).

The identification of excitation and de-excitation processes in-

volved in fluorescence measurements is important for several reasons.

First, it helps determine the relative contributions of various radia-

tional excitation energies as well as that of thermal collisions in pop-

ulatriig the upper level from which fluorescence occurs. It also permits

correct use of nomenclature and reporting of fluorescence processes.

Evaluation of quantum efficiency and flame temperatures requires a

clear knowledge of the overall process. Finally and probably most

importantly f: m an analytical point of view, it allows one to take

advantage of tihe difference existing between the excitation and fluores-

cence wavelengths in order to eliminate scattered light, which may cause

the limiting noise in resonance fluorescence measurements. Because

lasers are a single radiation source having an output which has a narrow

spectral bandwidth with high spectral irradiance and are tunable, lasers

represent an ideal source because no filters (or other frequency selec-

tive devices) are necessary to isolate the excitation (absorption) wave-

length. Also, the excited level population is greatly enhanced, thereby

favoring efficient mixing between adjacent levels resulting in strong


Atomic Fluorescence Radiance Expressions

Atomic fluorescence can be considered to be divided into three pro-

cesses: conversion of sample into atoms; absorption of radiation by

atoms; and fluorescence emission of radiation from excited atoms. Figure

2 shows a schematic diagram of the cell assumed for atomic fluorescence

measurements with the common right angle case (i.e., fluorescence is

observed perpendicular to the exciting radiation). The signal expression

is the expression that relates the measured signal to the analyte con-

centrit.ion. The radiance expressions for atomic fluorescence spectroscopy

are given below for various cases of continuum or line excitation

sources; high or low optical atomic densities; and high or low source

spectral irradiances. The expressions hold under the conditions of

homogeneous distributions of atoms in the analyte volume; uniform tem-

peratire, T, throughout the analyte volume; thermodynamic equilibrium

of s;,ecies; negligible effect of source radiation upon the energy dis-

t -^ f'

C) -*i*1 ------- 7-,

b) L


C) t ,L

Figure 2: Schematic diagrams of cell assumed for luninescence radiance
expressions: (a) three-dimensional diagram of cell to show
dimensions of unit cell; (b) top view or cell, right angle
configuration complete illumination of cell, complete
measurement of right angle fluorescence; (c) top view of
cell, right angle configuration incomplete illumination
of cell, incomplete measurement of fluorescence.
Taken from reference 19.

tribution, velocity distribution or temperature of species it the analyte

volume spatial uniformity and constancy of radiation density of the

source throughout analyte volume; and negligibility of polarization and

coherence effects and prefilter and postfilter effects.

Low Source Spectral Irradiance

Continuum source, low optical density case

B -C ( )E K f 6A Y (1)
BF = C () EC iKoij D P (1)
i.. kl

The fluorescence radiance is linearly proportional to concentration,

source spectral irradiance, absorption path length and quantum efficiency.

Self-absorption is negligible in this case. This is the normal condition

under which atomic fluorescence measurements are made.

Continuum source, high optical density case with atomic fluorescence
terminating in ground or nuar ground states
C2 1 L
S C () EC a 6 () Y (2)
1 A P2k
F 4 ij Pkl

Equation (2) assumes reabsorption of some fraction of emitted radiation,

(i.e., self-absorption). The fluorescence radiance is not a .unction of

atomic concentration.

Conti. uum source, high optical density case with atomic fluorescence
terminating in excited states

K nF. Kuf '
B = (-) E ( ) Y (3)
F 4 CA .. kIn2 Pkl

Reabsorption of fluoresced radiation is assumed to be minimal. Fluores-

cenc, radiance is proportional to (ni.)' i.e., the fluorescence signal

is proportional to the square root of the Tuncentr. tion. Therefore, with

a continuum source, many nonresonance fluorescence analysis schemes are

still analytically useful at high analyte concentrations but with reduced

sensit vity.

Line source, low optical density case

Bn = n(4)
B r EL i Ko ifj ij Pk

In the case of a line source, the source half-width is assumed much

smaller than the absorption line width. This case is similar to that

of equation (1) and is again a normal condition for analysis in atomic

fluorescence spectrometry.

Line source, high optical density case with atomic fluorescence termin-
ating in ground or near ground states

S( 4aL \ 1/2 (5
F \q/ EL Kn.f9.2 I1
(A o i kl /

In equation (5) it is assumed that reabsorption of fluorescence radiation

occurs. The fluorescence radiance decreases with increasing analyte


Line source, high optical density case with atomic fluorescence
terminating in excited states

BF E) E (6)
S YPkl

Fluorescence radiance is not dependent on concentration of analyte.

Equations (1) through (6) are shown graphically in Figure 3. These

curves relate the fluorescence radiance, BF, to the concentration of

species in level i, n.. These curves are called growth curves.

'hen absoprtion from excited states takes place, the population of

that level, ni, is found under conditions of thermodynamic equilibrium

by using the Boltzmann equation,

n. = no it ex

where the subscript o refers to the ground state.

Figure 3: Growth curves in atomic fluorescence. (A) Resonance
fluorescence: curve A continuum source; curve B -
line source. (B) Nonresonance fluorescence: curve A
continuum source; curve B line source. Taken from
reference 88.

log n0 -;:-

logj %o-~-;

lTe terms and symbols used in the above equations are as follows:

BF B = fluorescence radiance,
Ai->j Fk-+

where the absorption transition occurs from level i to level j and the

fluorescence transition occurs from level k to level 1, (i,j,k,l designate

-2 -i
energy levels of the Grotrian energy level diagram), W m sr ;

4n = number of steradians in a sphere (fluorescence is
isotropic), sr;

C = J2/2/1n2, no units;

C2 = /T/ln2, no units;

= absorption path length, m:

E = continuum source spectral irradiance of exciting
C.. radiation at the absorption wavelength X.., W m-2m-1
1ij j13

E = line source irradiance of exciting radiation at the
L -2
absorption wavelength A.., W m

n. = concentration of species in level i, m

Y -- Y = fluorescence power (quantum) efficiency,
kl ij W fluoresced/W adsorbed;

K = k /n.f.. = modified absorption coefficient, m ;
o o 1 13

k = absorption coefficient for pure Doppler (Gaussian)
Broadening, m-1;

2/1n2 X XA2 .F
k K nf. f = n.f m-1
o 0 1 13 c 6D 1

e -6 2-1
e = m = 2.65 x 10 m s

F = source factor to account for saturation of
energy levels, no units;

e = electron charge, 1.602 x 10-19C;

c = speed of light, m s ;

m = electron mass, 0.9107 x 10-30 kg;

SAD = Doppler half-width, m:

f., = absorption osci]atorstrength for transition
from energy level i to energy level j, no units;

a = damping constant, Jln2 6 C/6AD, no units;

6AC = Lorentzian (collisional) half-width, m;

L = fluorescence path length, m:

0' = height of fluorescence volume;

6,. = Voight profile factor to account for finite line
width of source compared to absorption line,
no units;

E0 = permittivity of vacuum, 8.854 x 1012 C N m-2

If the sample cell is not completely illuminated, the fluorescence

radiance, BF, emitted towards the detector is reduced by reabsorption of

radiation in the area ALx'xQ. This is the postfilter effect. If the

fluoresiLnce radiation is not completely measured, then the incident

irradiance from the source may be reduced in the area LIxLx&' and therefore

also red'cting the fluorescence radiance, BF, This is known as the

prefilter effect. Both prefilter and postfilter effects are only signifi-

cant for high concentrations of analyte.

iio amount of fluorescence signal reaching the detector is determined

by the uptical conductance, G (m sr), of the spectrometric system.

The radiant flux, I, impinging on the detector is given by:


where T = transmission factor of optical system dimensionlesss).

The concentration of analyte in the lower atomic level, n ,
atoms m must be related to the concentration of the analyte in solu-

tion. ,hen the solution is nebulized, and introduced into the flame as

an aerosol, it can be shown that (19):

n -
o QeZc


23 -1
NA = Avogadro's number, 6 x 10 mol-

3 -1
F = solution transport rate, ms ;

E = nebulization and solute vaporization efficiency,
no units;

S= atomization efficiency, no units;

C = concentration of analyte in solution, mol m ;

go statistical weight of state o. no units:

3 -1
0 = flow rate of unburnt gases into the flame, m s

e = expansion factor for flame gases, no units;

7. (-\}
t p i exp KTi = normalized electronic partition
function, no units:

k = Boltzmann constant, 8.31 x 10- eV K-1

T = flame temperature, K;

E. = energy of level i, eV.

lhe final relationship necessary to relate the solution concentra-

tion to the final signal relates the fluorescence radiant flux to the

detector output. This is given by S = F yRT where S = signal, V;

Y = photodetector sensitivity, A W ; and RT = electronic system

transfer function V A-1

High Source Spectral Irradiance Considerations

ine advent of lasers in spectroscopic analysis made it necessary

to consider the possibility of saturating a particular transition. As

the source spectral irradiance becomes increasingly large, the rate of

induced emission becomes much more important among the deactivation

processes, and the absorption of source radiation becomes nonlinear and

and approaches zero. The use of lasers as excitation sources has the

important consequence that the conventional photon transport equation

(Beer's law) is strictly valid in the limit of zero incident light

flux and is therefore an accurate approximation only for low intensity

sources. The high radiation density of a laser focused into an atomic

vapor is bile to redistribute the populations ot the levels involved

in thi absorption process. When this effect occurs, saturation of the

optical! transition is said to be approached or in the limit to occur,

and the fluorescence signal will no longer be proportional to the

source irradiance and will reach a limiting value depending only upon

the properties of the atomic system (67).

A more general fluorescence radiance expression which holds for

low or high intensity continuum sources and at low optical densities

is given by:

A B = ) Y E /' k dv
Aij Fk-*l 4 / kl ij 0

Pkl E hni
Ek b ii ni (1

+ 2klhvklnk


F = spectral irradiance of continuum source at frequency
ij ., -2 -Hz
Wm Hz

Sok d = integrated absorption coefficient over absorption
line, m-1 Hz;

B.. =i( A.. = e f.. = Einstein coefficient
0 ,"
1J gi) (-1 11 jo mhy

of induced absorption, i.e., absorption transitions per spectral
energyy density per absorbing species, m3 .-1 -1 Hz;

SEinstein coefficient of spontaneous emission, s
Akl = Einstein coefficient of spontaneous emission, s

S= Planck's constant, 6.626 x 10-34 J
- = Planck's constant, 6.626 x 10 J s;

S. = frequency of transition occurring from energy level i
to energy level j, Hz;

hv.. = energy of exciting photon, J.

Using the steady state rate equation approach, where the rate

of excitation equals the de-excitation rate and the assumption made

previously for Equations (1) (6), (except for low source intensity),

fluorescence radiance expressions can be derived for: continuum source

excitation under both high and low spectral irradiances; low optical

densities; and atoms which can be approximated by a two-level system

or a three-level system (see Figure 4). The general expressions are

as follows (69):

Two-level atom, low source spectral irradiance
B ]W EVI2 1T (7)
BF -= )A21hv2 E n (7)

Two-level atom, high source spectral irradiance

BF 2= 1 hv12 2 nT (8)
+ (8)


s CA21
I- = -2 = modified saturation
12 12 1 21 21

spectral irradiance, W m-2 Hz;
spectral irradiance, F m- Hz;

Figure 4: a. Two-level atomic system with radiational, A, and
radiationless, k, rate constants.
b. Three-level case T (e.g.,Na) atomic system with
radiational, A. and radiationless, k, rate constants
k13, k12, A32 0). Left hand diagram is for 1-3
excitation. Right hand diagram is for 1+2 excitation.
c. Three-level case IT (e.g.,Tl) atomic system with
radiational, A, and radiationless, k, rate constants
(kl3, k23, A21 Z 0). Left hand diagram is for 1-3
excitation. Right hand diagram is for 2-3 excitation.
Taken from reference 69.

-----, k

i i
A 2 k2. Z'

A3 I I
.1 I

c c
h .
A ,',A I A. |1 !
I 1/

k A21 ,kfz kal


B" E t 53. B,21E;,Lz
C. -.


S = saturation spectral irradiance (i.e. source spectral
12 irradiance where the fluorescence signal is 50% of the
maximum possible value), W\ m-2 lz:

n = total atom population, m

Three-level system

There are two major types of three-level atomic systems: (I) systems

where the levels 2 and 3 are close together and radiative transitions

between levels 2 and 3 are not allowed e.g., Na; and (II) systems where

1 and 2 are close together and radiative transitions between levels

1 and 2 are not allowed, e.g., T1. Specific expressions are given

by Boutilier et al. (69). More general expressions are as follows:

Case I for 1-3, 3-1 or 1+2, 2+1 [excitation (absorption) transition,

from level i to level j (i-j), fluorescence transition from level

k to level 1 (k-l)]

low source spectral irradiance (E <


B =() kl hlk (9)
13 iJ

S(hi) k k.
S+ + Km
m'l m'l1 m'k

Case I for 13, 3 ---2, 2-1 or 1-2, 2 ---3, 3-1

(dashed arrow, ---, denotes radiationless process)

Bl = (
B = ( ) IA ( E) nT
F(10) 41 kl lk p E "T
(in) I v

kkl +
-_-_- (11)

Ak + kkl+ kkj

S = ( ) Akl hVk nA k
F 01i) k1 Ik Y A k + kj + L kkI

Fi k 4 1
(hi) + (1f) k- kj--3,k1

jkase TI for

case TI for 1-3, 3-1 or 1-+3, 3'2 or 1+2, 2 1 or 2-3, 3-+1


\1 1
A k 1 g Ei nT 1( + )
'( 4ii j 1 + ) exp (-E12/kT)

B,, = ( ) Al Ik
t" ___ CI I( 1k T

where E = i--
1]3 ]1 l

i.. = radiationless rate constant from level i to level j; and

prime (e.g. m') indicates the energy level not involved in the

absorption process.

From the above fluorescence radiance expressions, it is important

to point out and stress the characteristic features of laser-excited

fluorescence which result in great advantages for analytical atomic

spectroscopy (1, 43-69, 88-90).

(1) Attainment of the maximum possible value of the fluorescence

signal for a given nT is reached under saturation conditions,

i.e. E > E f the atomic system closely approximates a two-

level system, then the absolute value nT can be determined if BF is

measured under steady state conditions, i.e., the radiative lifetime

of the excited state must be much less than the temporal pulse width

of the exciting source. For a three-level system, nonradiational

de-excitation rate constants must be known. Tt should be noted that

for high spectral irradiances, E > E. the fluorescence radiance is

determined only by well-known atomic parameters and radiationless

rate constants: therefore, increasing the source spectral irradiance
above E will only increase the fluorescence signal by at most a

factor of two, while the signal-to-noise ratio obtained under these

conditions may deteriorate due to increased noise from larger scatter

or molecular fluorescence signals.

The possibility of simultaneous multiple line detection for

more complicated (mulci-level) atomic systems should not be overlooked

as a means to obtain the largest possible fluorescence signal. In this

case, fluorescence from more than one atomic transition is allowed

to impinge upon the detector.

(2) Increased linear dynanmic range of the analytical calibration

curve is obtained under saturationor near saturation conditions. The

linearity of BF with nT (analytical calibration curve) is greater as

E in,'reases because as E increases nT must increase for k I to

exc,-ed 0.05 (high optical density). If the source irradiance is such

that the atomic system is saturated at any value of the population

density, then the absorption coefficient goes to zero, and no self-

absorption occurs, thus making the calibration curve obtained under

idealized illumination conditions linear over all concentration ranges.

This holds for a continuum as well as for a line source of excitation

(67). However, unavoidable tradeoffs in practical situations do not

allow the observation of such extended linearity. For example,

when the laser beam is focused at the center of the atomizer,

postfilter (self-reversal) effects in the observation path toward

the detector certainly have to be included to account for the observed

shape of the analytical calibration curve. On the other hand, if

the beam is defocussed to illuminate a larger volume of atoms, then

the irradiance can decrease to the point where saturation is no longer

approached, and therefore, the shape of the curve will resemble that

obtained with conventional sources of excitation. Linear dynamic

ranges of 5 orders of magnitude are common for laser excited fluorescence.

Kuhl et al. (89) compared relative fluorescence curves of growth obtained

when lead vapor from a carbon rod atomizer was excited with a hollow

cathode lamp, an electrodeless discharge lamp and a dye laser. The

linear dynamic range, defined as the ratio of upper to lower concen-

5 3 3
tration limits, found here was 1.25 x 10 3.33 x 10 and 1 x 10 for

dye Iacer, electrodeless dischnrge lamp and hollow cathode excitation,

respectively. The lower concentration limit corresponds to the detection

limit while the upper limit was defined as the lead concentration resulting

in a signal whose deviation from linearity did not exceed 5%.

(3) Independence of fluorescence radiance upon the source

irradiance and stability occurs under saturation conditions. Under

the high source spectral irradiance cases described above, E no longer

appears in the equations. As the source irradiance is increased above

E the absorption (excitation) process becomes less and less

efficient; therefore, source fluctuations will produce progressively

smaller variations in the excited state atom population. This means

that upper level of the fluorescence transition will have a rela-

tively stable population, which will in turn produce a relatively

stable fluorescence signal.

(4) Independence of fluorescence radiance on fluorescence

quantum efficiency is achieved under saturation conditions. The

equations describing the high source spectral irradiance cases do not

have a quantum efficiency, Y, dependence. The dependence of fluores-

cence radiance on quantum efficiency is often quoted as a fundamental

and major limitation of the fluorescence process. Therefore, the

importance of this consequence is obvious. Also, it means that atomizers

with high atomization efficiencies can be used even if strongly quenching

species are present in the atomic vapor, (i.e., fluorescence radiance

is now independent of flame chemistry whereas at low spectral

irradiance, the flame gas composition greatly influences Y). When a

high spectral irradiance excitation source is used for fluorescent

measurements, the rates of population and depopulation of the excited

state become predominantly radiationally induced. When such stimulated

processes dominate, any other activation (chemical or thermal) or

deactivation (quenching, spontaneous radiation) pathways do not

strongly affect the equilibrium population of the excited state.

Consequently, the magnitude of the detected (spontaneous) fluorescence

becomes nearly constant (1, 54, 67).

Although not considered in the above fluorescence radiance

equations (7)-(14), two other advantages of using high spectral

irradiance (laser) sources should be mentioned.

(5) Elimination of prefilter and inner-filter (self-absorption)

effects occurs under saturation conditions. The spectral irradiance

of the laser output experiences minimal attenuation as it passes

through the prefilter region. Also because the absorption coefficient

approaches zero as saturation is reached, reabsorption of the fluorescence

radiation becomes negligible. The postfilter effect is unaffected

by source irradiance.

(6) Elimination of scatter noise limitation is possible by using

a nonresonance fluorescence analysis scheme. Generally, resonance

fluorescence transitions have higher transition probabilities than

nonresonance transitions. However, the higher spectral irradiances

of laser sources make nonresonance transitions not only easily detectable,

but sometimes the preferred analysis scheme. Also, because lasers

emit a single narrow spectral bandwidth line, no frequency selective

device is necessary for the exciting radiation.

All of the above advantages (1) (6) are due to the high spectral

irradiances of lasers and are distinct advantages which lasers have

over uLnvcntional sources (e.g. electrodeless discharge namps (EDLIs),

hollow cathode lamps (HCLs), and xenon arc lamps).

(7) Monochromeaticity is a distinct advantage of laser sources.

The fact that their output is a single monochromatic line, and not a

series of lines or bands due to emission of lines from atomic or

molecular vapor and filler gases as is the case with EDLs and HCLs,

or a continuum such as that of xenon arc lamps, means that there is no

off-wavelength background other than source induced emission in a

well baffled (to eliminate stray radiation) detection system.

(8) Narrow spectral bandwidth of the laser output is a tremendous

advantage as can be demonstrated by the lack of spectral interference

occurring in laser-excited atomic fluorescence. Laser sources are

unique in that they can have spectral bandwidths varying from 20 nm to

less than 0.0006 nm. Typically, lasers have bandwidths of ca. 0.01 nm

and, therefore, must be considered to be a pseudocontinuum radiation

source because their bandwidths are greater than the atomic linewidth

in a flame (typically 0.001 nm to 0.005 nm). However, most lasers

have available frequency narrowing etalons to reduce the spectral

bandwidth at the cost of a reduction in average output power. If the

laser output is frequency narrowed so that it becomes a "line" source,

the exact expressions for BF given above in equations (7) (14) are

more complex, but are generally similar. BF is still linearly related

to n but now the source irradiance absorbed is determined by the

width and profile of the exciting line and the velocity distribution

of the absorbers and the broadening mechanism for the absorbers. Line

source excitation has been discussed in literature (62, 67, 91). It

should also be noted that the linewidth of the laser output beam

is a function of wavelength and beam waist. The linewidth decreases

at lower wavelengths. The narrow spectral bandwidth along with a pulsed

output allow for the use of a low dispersive or nondispersive spectro-

metric system in order to gain larger optical collection efficiencies

of the fluorescence detection system.

(9) The co!herernt output_t of the dye laser with its high degree of

beam collimation and directionality, plus the small beam diameter

make the laser beam easy to work with from an experimental point of

view (e.g. ease of focussing, beam expanding, beam direction, etc.).

These factors also make the use of a multipass cell more attractive

and viable even for local plasma diagnostics. The small beam

diameter of the laser output is considered to be essential in those

laser spectroscopic studies where saturation of energy level is

required or where spatial diagnostics (e.g., profiling temperature, T,

or concentration, n ) in flames or plasmas are to be probed. A small

beam diameter, however, is not necessarily required in (analytical)

atomic fluorescence flame spectrometry. In fact, using spectral

irradiances greater than the saturation spectral irradiance, E ,

will at most double the atomic fluorescence signal, while the scatter

signal will continue to increase linearly with source power, E i.e.,

laser power is being wasted and SNR decreased (71, 73). However,

assuming the atomic vapor is excited to near saturation conditions

with the expanded beam as well as with the beam not expanded, the

fluorescence signals will increase in direct proportion to the ratio

of the illuminated volumes. Therefore, laser beam expansion with

sufficiently high powered lasers is analytically useful because the

source irradiance can be optimized (saturation spectral irradiances

for each transition should be calculated or experimentally determined

in order to determine the best compromise source irradiance to use)

(54, 71, 74). Illumination of a larger flame volume excites more

analvte (larger volume) and makes the optical alignment of the system

less critical. Also, near saturation conditions maintain the benefits

achieved with saturation (67), namely: (i) large signal levels;

(ii) large linear dynamic range; (iii) less dependence on source

fluctuations; and (iv) less dependence on quantum yield.

When pulsed sources are used, the temporal behavior of fluorescence

must be considered (54, 67, 73, 74, 88, 92, 93). Practically, a

gated detection system is used, so that the detection system is active

only during the period of time when the fluorescence signal (radiant

flux) is present at the detector. The nitrogen laser pumped tunable

dye laser used in this study has a 2-6 ns temporal pulse width.

Typical fluorescence lifetimes of atoms in flames are on the order of

1 ns. When the duration of the excitation radiation pulse is comparable

to or shorter than the fluorescence lifetime of the upper state of

the transition being observed, the steady state assumption does not

strictly hold. Also, if laser excitation (i.e., high spectral

irradiance) is used and the atomic system approaches saturation under

continuum excitation conditions then an effective lifetime term called
the response time, cL, must replace the spontaneous lifetime, TL Ai

in equations describing the temporal behavior of the fluorescence.

Short pulses mainly affect the growth of fluorescence radiance (i.e.,

the rate of population of the upper level). General radiance expressions

describing the temporal behavior under high and low spectral irradiance

are discussed in literature. The attainment of steady state conditions

is governed by the ratio t/tL where t = any instantaneous time after

initiation of the source and t = E s/(E +F' ), and therefore also
o o
by the spectral irradiance of the source. If the spectral irradiance

is muCh less than the saturation irradiance, then tL is simply the

fluorescence lifetime. However, if E >>E., then steady state conditions

are generally achieved and the previous equation holds. In order to

compare signal intensities with other systems, the signal, SF, present

at the output of the detector must be multiplied by an effective duty

factor, deff. The exact fluorescence intensity expression under
pulsed source-gated detected conditions is complicated because the shapes

of the source intensity growth and decay and the detector response

during gate opening and closing vary greatly with instrument and time,

especially on the very fast time scale required for atomic fluorescence.

The effective duty factor for pulsed (p) laser excitation and a gated (g)

detection system is given by (74):

f{t t [1 exp(-t /t )]}
dPf /g L f
1 exp (-1/ftL)

where f is the repetition rate, and t is the gate width.
Detection systems for pulsed laser-excited fluorescence generally

use a photomultiplier tube to convert the radiant fluorescence flux

into an electrical signal. The signal is then processed with either

a sampling oscilloscope or a boxcar (gated) integrator. It has been

stressed (45, 88, 94-97) that photomultipliers suitable for pulsed

operation must possess certain specific requirements such as short

transit time of the electron cloud, small transit time spread, short

rise time and minimal parasitic capacitance. The photomultiplier

tube must also be able to sustain high peak anode currents. For

such high currents, the dynode chain must be modified. The laser

output can also be monitored with a photodiode. Both the signals

can >e fed into a boxcar integrator and the signals can be ratioed.

The boxcar integrator essentially performs a sample and hold operation.

The sampling time is determined by an appropriate reference pulse which

has a definite time difference with respect to the signal. When

pulse measurements are performed, the timing and width of the sampling

window (gate) are adjusted so as to coincide with the occurrence of

the fluorescence signal pulse (67).

For any measurement system, there are nine major noises to consider.

These are detector shot noise (e.g., photomultiplier dark current);

background shot and flicker noise (e.g., Flame background from OH

emission): scatter shot and flicker noise (e.g., Rayleigh and Mie

particle scatter from flame gases); source induced fluorescence

background shot and flicker noise (e.g., CN fluorescence in a nitrous

oxide-acetylene flame); and analyte luminescence shot and flicker

noise. All other noises such as 60 Hz pickup or radio frequency

interference (RFI) noise or noises that arise in the electronic signal

processing are system noises not inherent in the signal flux detection

[i.e., they occur after the radiant flux strikes the photoc:.-hode and,

therefore, do not result from signal generated in the sample cell

or the photocathode (transducer) surface]. Shot noise is due to the

fundamentally discontinuous (quantized) nature of radiation (photons)

and electrical current (electrons). Photons arrive at the photocathode

randomly even though the overall radiant flux is constant. The same

is true for thermionic emission from the photocathode. Signal shot

noise has a square root dependence to the signal level. Flicker noises

describe the relative fluctuations in the signal level and are, therefore,

proportional to the signal level. Boutilier et al. (74) have compared

signal-to-noise ratios for puls.,a and cw sources in atomic ,ind molecular

luminuscence spectrometry. Bowse and Ingle (98) give a procedure for

deter:iining the noise which limits the precision with which a measurement

can be made in flame atomic absorption spectrometry. Obviously, the

source, of the limiting noise is There refinements of the system and/or

research should be directed.

10) The gain for pulsed operation compared to cw operation as

proposed by Omenetto et al.(67, 71-73) is an advan-.ge for the N2 laser

pumped dye laser fluorescence system. Assuming that the total noise

in a fluorescence setup is due to atomizer-source-detector shot noise,

the :.in, G, in signal-to-noise acio due to source pulsing (po) and

detecor gating as compared to continuous wave operation (cw) of both

sour. and detector can be expr.:;sed as follows:

P cI } 1
,(po/cw) = {(E ); /(E ) avft

= {E ) /(E ) {ft }1/ (15)
A peak A av p

in !ite case of background shot noise limited systems; and

G(po/cw) = {(E P) /(EC)c )'/2

( ) pea /(Ec) 1/2{ft l/2 (16)

in the case oi source-related or source-inauced shot noise limited

syscem.. Here, (E C) stands lur the aveiage spectral irr -;i.nce
sys~~ r3Vl

-, -l
(W nm ) -f an excitation s *.rce operated in the cw modt and

(E )peak and (E ) represent the peak and average spectral irradiance

of the pulsed source, which is characterized by a pulse duration tp (s)

and a repetition rate f, Hz. Tie product ft is the duty cycle. If

the a.urce stability is poor ca:sLinp flicker noise which is source-

related (scatter) or source-indited (molecular fluorescence of flame

gas species) signal, then there will be no gain or a reduction of

the gain with the pulsed excitation source. For narrow line sources,

Ei is replaced by EL, the line i-radiance, i.e., EL = /oEdA .

As can be seen in Equation (15), the advantages of gated detection

appear to be exploited by the high peak power and low duty factor of

a N2 laser pumped tunable dye lser fluorescence instrumental system.

The highest average power, (EP )v, consistent with the lowest duty

fact.Jr, ft appears to be the r..:ot dcsir..cle operating conditions.

For resonance Lluorescence, the Rayleigh matterr shot noise limit

shout be approached. For nonresonance fluorescenc, cases where higher

spettral irradiances are tolerable, the lowest detectable signal will

be limited by dark current noise, amplifier noise, molecular fluorescence

background noise or simply the ability to collect efficient, the

fluorescence radiation. However, because not all elements have strong

nonreonance fluorescence lines, system should be designed to optimize

as bisc possible, both resonance .luorescence and ,onresonnnce

fluorescence signals in order thai the the system be generaiiy

anal- ically useful.

.% can be seen in Equation (16), when the system is souire-

carri.j shot noise limited, the characteristic advantage of pulsed

excitation-gated detect on is 1-., ; chis case will be encountered in

manr resonance fluorescence measurements, as pointed out by come

workers (43, 45). Therefore, efforts to reduce the scatter must

be male. By loser beam expansion and proper baffling, aperturing and

trapping of the optical light path, sourLe-related scatter can be

greatly reduced.

(11) Mlinimization of the emission background noise results from

the extremely low duty cycle of the N2 laser pumped dye laser source

(10 ). A narrow gate width (t 10 ns) of the boxcar integrator

can t.hn be utilized so that th. detection system only "views" the

flame emission background for a. extremely small period of time. The

ratio of detector on time to th- rotal analysis time is the duty

factor, ft .
(12) High optical conductance (i.e., high optical throughput or

luminosity, L GT) results from the combination of Low duty cycle

with 5ated detection (i.e., low I .ty factor) and the narrow spectral

bandwidth of the laser. The former minimizes the emission background,

while the latter helps to minimize scatter over the spectral bandpass

of the. detection system (i.e., negligible scatter exists at wavelengths

different from the excitation wavelength). Therefore, low dispersive

spectrometric detection systems (e.g., interfere~ filters) can be

used. The larger collection efficiency of fluorescence gives high

fluorescence signal levels. The spectral iesolutio:, of these systems

is still excellent because it i., determined solely by the ; trial l

band.dti~th of the dye laser output.

(13) The wide wavelenglt tjiingj. range of the N, laser pumr.ped Jve

laser (ca. 217 950 nm), particularly its extension into the UV

where most of the strong atomic fluorescence lines exist, i of obvious

benefit for atomic excitation. Moreover, this is accomplished

through the use of onl; one source.

(14) Relative ease of wavelength selection and dye changing of

the N, laser pumped dye laser compared to most other dye laser systems

aids greatly in the operation of the system. Most dye laser systems

use a flowing dye system which contains ca 4 1 of expensive dye

solution. However, the N2 laser pumped dye laser system can operate

with only 2 ml of dye solution in a cuvette. Cuvettes on a carousel

can be rotated into the laser cavity. Selection of wavelength, then,

involves rotating the proper dye solution into place and changing the

angle of the grating, which acts as the wavelength selective device

and the back laser cavity mirror.

(15) Fluorescence itself must be considered as an advantage to

the spectroscopic system. The reasons are inhere;,: to the fluorescence

process. First of all, at low concentrations, a low signal level

must be detected which is easier than detecting a small change in a

large signal level. Second, fluorescence is emitted in 4T steradians.

This niit only .nakes optical alignment simpler, but also allows the

use nf two detectors so that optical conditions for the detection of

resonance and nonresonaice fluorescence could be optimized separately.

Multi-element analysis is also possible using only one source and a

slew-scan type system similar to that used by Johnson et al. (99).

Irom the theoretical and practical advantages; described above,

the N2 laser pumped dye laser fluorescence spectroscopic system with

gate~i detection should provide greatly improved analytical figures of



Any fluorescence instrumental system consists of an excitation

radiation source, a sample cell, associated optics and a means of

detecting the fluoresced radiation. The basic components of the atomic

fluorescence flame spectrometric system used in this study are a

nitro-en laser pumped tunable dye laser excitation source; a beam

expander for increasing the diameter of the dye laser output beam;

a pneumatic nehulization sampling system; a flame atom reservoir;

a monochromator for wavelength selective radiant flux isolation; a

photo.nultiplier detector and tra.,sducer which converts the radiant

flux impinging upon the photoca-. de into a signal current; a boxcar

(gateu) integrator for -gnal pr,, essing of transient repetitive

signals; and a strip clhart record lr. A schematic diagram of the

present laser-excited atomic fluorescence flame spectrometric system

is shown in Figure 5. The experimental components used in thL system ,

are described in Table 1, along with their model numbers and manufacturers.

Excitation Source

'lie nitrogen laser pumped unable dye laser he many characteristics

which ,iakc it an apparently ideal excitation source for atomic

fluor.escence spectroscopy. Many of these characteristics have been

indicated in the previous two chapters. Much research continues

to be done in the field of dye laser development alone, such as: dye

research to extend the wavelength running range and lifetime of the

dyes and to improve the output -i-,v.r of the dye laser; N2 la, r res-.rch

* 1-

C *H 0 I | d

- U O C o

O rd *i *- 'U
0 4 i O **

S *. .H
0] D. *- C U H


'C a C C c

&|U uc U
U) U- -- 0
. 0 ** ; a' .C
. U] .-ICjI
4. 4 On

cC C 4-J 4W

mE 7, E M
-- d- )-< ^. 0 0
EC X 4-4

-0 O C C -
S0C 4J

** C O- ** 4- **H

O ^ C- C C )
CN CO Ci 0 *-i- rC C >.

^ iio'ia o c


j -


0 Hn ----

oo to
0 t0 to

G C 0 (- C

I cf to o --

CoI o C o :
0 a) a > 1 -

to t o c -' H 4< c-!

j C D o
0 0 0 U cc r 0 C tn 1o

tu -n to- rO -i r- 10

to oa a o o 0 c
O Or 4 O O O O 00 0
< >-j t 00 t 0 -H '. -3 t

S0 t O 0 C 0 7 -C
-O O H O 0 o 0 t a -
0 4O to a -to 0
O I O t i CC < 0 4 C
F I 4- 0 H on: 0 o Co t
to o 0N c o -0 o o 4 -
I 0 44 C n 4- 4 (2 tE- -C H
G mEO m C; (i >
SC 0C w 0
S0 2 N + C I C 0


S z c c 0 2

to tc 1< 0 C o
o e 4 0
to O 0 H 0 N

Co 0 o I I O I S
-H3 0 -- C 0 0j

Oi0 i

0) 0 CO

V i -> 0 O

I S B -

E S' t & j 0j W i' u) to C lJ
0I '- e 0 rO u

to 3 > 33 t O t H Or 0

i 0Cr-,'' to HO HO > U 0 0 to

to :[o' imize its peak power, pulse width, pulse energy, triggering and

switcLing circuit with respect to being used as a pump source for

a dye laser and to minimize its radio frequency output so that its

operation does not interfere with the operation of sensitive detection

instrumentation and also to improve its reliability and stability; and

dye laser cavity optics and amplifier configuration in order to obtain

the most efficient use of the N2 laser pump source and to provide

continually improved characteristics demanded by spectroscopists.

However, because the N2 laser and the dye laser are commercially

available such details will not be discussed. The ease of operation

and the dye laser output beam characteristics are of primary interest

to the analytical spectroscopist, and these will be described.

The pulsed N2 laser unit consists of a laser head, power supply,

triggering unit, and vacuum pump. The laser head houses the laser

disccirge channel and high voltage circuit, thyratron (fast switch)

pulser circuit, and pressure and N, gas control valves. The laser

channel is typically operated at 50 torr with the flow of N2 gas

through the channel determined by the pulse repetition rate used.

The specifications for the four basic components are given in Table 2.

Figure 6 shows the typical peak, r':s, and the average power of the

Model IV-400 N2 laser as a function of repetition rate.

lThe dye laser is based on i Hansch design (100). Figure 7

shows a schematic of the optical design as viewed from above. The dye

laser js transversely pumped by the output of the \ 2 laser, which is

focusr,-I onto the dye cell with a cylindrinal quar z lens. Some of

the ;Iuorescent radiation produced in the dye cell travels along the

laser cavity, which is formed by tne grating (acting as the back

i'ble 2. Molectron Model UV 400 N2 Laser Specifications


Peak Power Output

RMS Power:

Average Power:


Pulse Width:

Repetition Rate:
Beam Dimensions:

Beam Divergence:


Time Delay From Trigger

Time Delay Jitter:

System Leakage:

Gas Supply:

400 kW @ 20 Hz; 100 kW @ 100 Hz

190 W @ 50 Hz

130 mW @ 50 Hz

337.1 nm

10 nsec

1 100 pps

6 mm vert.; 25 mm horiz.

<1 mrad vert.
<7 mrad horiz.

15% power variation @ 100 Hz
500 1000 nsec

<4 ns @ 60 pps

<1 Torr/s

Molectron recommends the use of stand-
ard high purity gas (99.9%) or the
use of a LN dewar with a built-in
gas converter.


Input Requirements
Charging Voltage:
Max. Avg. Output Current:

117 V(ac) nominal, 20 A, 60 Hz

0 to + 30 kV(dc)

40 mA(dc) @ 100 pps


Electrical Service


Output Voltage:

Output Energy:
Internal Rate Generator

Pulse Rise Time:

Repetition Rate:
Manual Operation:

208/440 V(ac), 3 phase, 60 Hz,
5.2/2.6 Amp

-250 V peak
0.3 mJ

Variable from 1 100 pps

0.5 psec nax.
Single pulse to 100 pps

Single pulse with local or remote

Table 2. (Continued)

External Signal Operation:


External Sync Output:

External pulses +5 to +30 V peak,
with maximum 1 psec risetime into
a 50 0 load

MODE switch (including power OFF
positions), COARSE and FINE pulse
rate controls, MANUAL operation
pushbutton switch

Output pulse of 4 V peak @ 400 ilsec
is available at the SYNC OUT con-
nector. Sync pulse occurs at same
time as output pulse and may be
used to trigger an oscilloscope,

500 _
400 o PE K POIER
iU 3U0 .t -
400 -

0 200 l-
< 100 -
,D 0 ---,--- I l
a- o 0 I _
10 20 30 40 50 GO 7 0 0 90 100



o v-I I
50c 10 I
< 0
> 10 20 30 40 50 60 70 80 E0 100

1 --i ... I___

5200 I- - i

, / 'i i i "
100 L '

10 20 30 40 50 00 70 0o 90 100

Figure 6: Peak, average, and rms power of N2 laser vs repetition rate

EU w

'- C-
, P
ccnO :
;- : I U











mirror) and the front output mirror. The grating acts as the frequency

selective (tuning) device and returns a particular wavelength within

the fluorescent emission band of the dye to the dye cell for amplifi-

cation in the laser cavity. Because the single pass gains are so

high with this configuration, only a 4% reflectivity of the front

output mirror is required to sustain lasing action. Table 3 lists the

specifications for the three available cavity configurations. The

DL100 is the basic dye laser consisting of the grating, dye cell,

and output mirror. The DL200 adds the beam expander. The beam

expanding telescope collects a larger solid angle of radiation from

the dye cell; expands the beam so that a larger surface area of the

grating is illuminated; and reduces the beam divergence on the grating.

This has the effect of increasing the output power, narrowing the

bandwidth of the output beam and reducing cavity losses. The

DL300/DL400 inserts a Fabry-Perot etalon between the beam expander

and grating. This provides an even narrower laser output bandwidth

and also improves the frequency stability of the laser cavity; however,

it also reduces the output power. Such narrow bandwidths are extremely

useful for high resolution spectroscopy because source broadening

effects (54) can be essentially eliminated. The mode structure of the

laser output beam for this Hansch-type dye laser has recently been

reported (101).

Of particular importance to analytical spectroscopy is the

capability to frequency double the fundamental output of the dye

laser. The most sensitive lines of many elements lie in the

ultraviolet region of the spectrum. However, because the efficiency

of frequency doubling is typically 5 10%, the dye laser output power

C, ci E

C 0

o on

cl 0 N N 0 N C C0

- I I

C i i N

CJ 7 7 '- C* 0

N m C C

3 -+

N r


-r- U

1000 D N
0 0

3 00 0

4 a


a) C

44 I 0 0

uj 0 H- -c a (n

4 C 0 N 44 0 0 N N U

c to 44 4 C ci o u

O C 0 4 4 4 4 C N 4
CC C 0 0 j 0 C
0 09|0 0 0 Oo

( n 0 ICr ) 4 d 4 u z3
r c- c c) C) w a a N1 > C c .
E2 C) O 4 40 0 q2
O 3 t0 i C C) i4 2 C) C C) C)
0. a a) C 0 E 1 s 0
000> 11 c ^ ^^ 0

0 E

r-N f


fn 0 0 C0
0 -



r -.


C 0

c0 c

()3 V

0H 0

Ol C
- a

is considerably reduced. Frequency doubling is accomplished by

changing the output mirror to one that can focus the fundamental dye

laser beam into one of a series of KDP or KPB nonlinear crystals.

The crystal used is determined by output wavelength desired. The

crystals are mounted on a rotation device, so that they can be

angle phase matched (necessary for frequency doubling to occur). Filters

or a prism can be used to eliminate the fundamental beam from the

frequency doubled output. Because the nonlinear crystals only double

the horizontally-polarized component of the fundamental dye laser

beam, a linear polarizer can be inserted into the dye laser cavity

in order to maximize the horizontal component and therefore, increase

the doubling efficiency. Because the extracavity power is almost

equal to the intracaviLy power, efficient frequency doubling is

accomplished through the use o0 a nonlinear crystal external to the

laser cavity, which is far more convenient and less critical.

A scan control unit controls the sine-bar grating drive and a

synchronized etalon-tilting meciianism. It also houses the dye cell

interchange control. Six dyes can be in covettes on a carousel.

The dye chosen for the desired wavelength is rotated into the dye

laser cavity. Each cuvette contains a magnetic stirrer which is set

in motion when the dye cell is being irradiated by the N2 laser output,

so th.it the dye does not degrade or plotodecompose quickly.

The N2 laser pumped dye laser has the widest continuous tuning

range of any laser, in Figure 8, typical dye tuning curves are given.

The organic dyes used are listed in Table 4 where the concentrations

of the dyes, solvents, and their peak lasing wavele.?iith and wavelength

range are also given.

FuLndaler Ii

IC000 2 4 O 11.

S/ ,, / \

Siu0-i I I/; / I I

0 i I -- ,

3 0 37 30 30 0 410 410i 420 430 4 0 4L1D 4661 40 Ii0 490

o Fundamen'al
S 11
Sl00- 10 14

400 "\ - \

F i /r -. t n g c v fo dyes. S0 T 4 for
u 00 : \ e p /
..i I ,'

I (MO -I VEclT .
I it"
I 118

', / ,, / "\ -
F 1 -I 1

. . [ -o-! ... i.. .. ___ _.

F.;'i I b i M,3 C '0 f ;l j i.' ,'l ':'1 ,'J' 740 7";0

Figure 8A: Wivelen gth tuning curves for dyes. See Table 4 for
dye description.

6 3

210 220 23 240 0 2


.... .. 13

\ /


Fr ..j"i.oy Do b.bled




\j v\

2 70 2,0 20C 2?' 310 20C z3.1 3;n 35 360


2 23 ;'4
2 i ; '.

S -- r -- -
'*{* ? ,o7 /"'" i .* .'

\','A 'V' Lf I (N f (1 -i

Figure SB: Wavelength tuning curves for dyes.
dye description

See Table 4 for



25 21

0r 1
,GO ginlr


Table 4. List of Laser Dyesa

No. Dye Concentration Solvent

1 PtD 5x10-3

2 BRQ 2.5x10-3

3 PBO 5x10-4

4 DF5 Snturated
<1.2 10

5 BIs-MSB 1.2x0-3

6 C120 5x10-3

7 C2 10-2

8 7D4MC 10-2

9 C102 10-2

10 7D4THC 10-

11 C500 10-2

12 C485 10-2

13 C495 10-2

14 O6G 5xlO-3

15 RB 5x10-3

16 R3 + CVP 2.5x10-3


I R66G + CVP 2.5x10-3

3 0
19 RB + OX1P 5x10-3

20 OXIP 2x0-2

21 DOTC 1.2.10-3

22 21 + 24 10:1

23 21 4 24 1:1

24 ITC 1.2I0- 3

25 DTTC 1.2xl0-3

26 IR144 2.5x10-3

27 IRI25


t luene/dchanotl



p-diox oce



















Vavclnagth R.nng (nn)
Peak (nl,) (1%p t a )

366, 378
































660 641-487

691 683-710

725 705-750









''toIo "'oect1Cra Dye List" by Molectron Corp., Sunnyvnle, CA 94096.

The characteristics of the N, pumped tunable dye laser excitation

source are summarized below:

(i) high peak power;

(ii) narrow spectral bandwidth;

(iii) wide wavelength tuning range;

(iv) low duty cycle:

(v) small diameter, nearly collimated, coherent output beam;

(vi) single monochromatic source; and

(vii) ease of wavelength selection.

The advantages derived from these characteristics were discussed in

Chapter II.

Detection System

The wavelength selective device used in these studies was a

0.35 m Czerny-Turner grating monochromator with a f-number (i.e.,

relative aperture) of 6.8. Fluorescence systems, in general, allow

the use of low cost, low dispersion monochromators as compared to

emission or Raman systems which generally require higher dispersion.

Nondispersive fluorescence systems, which allow a greater solid angle

for collection of fluorescence radiation, are also attractive possibilities.

The photomultiplier tube (PMT) used for most studies was a

R212U!, (which is a high sensitivity variant of the 1P28, more

commonly used in UV-Visible spectrometers). As mentioned earlier,

photomultiplier tubes used for the detection of fast transient signals

must meet specific requirements. They must have a rapid rise time,

short transit time, small transit time spread, minimal parasitic

capacitance, and be able to sustain peak anodic currents on the order

of amperes (80). To provide the best transient response and largest

signal levels, the highest possible high voltage not causing PMT

electrical breakdown or excessive dark current should be applied.

At high signal levels, care should be taken to ensure that the PMT

is still responding linearly to the signal, that PMT fatigue does

not occur and that space charge buildup around the last few dynodes

is minimal. Modification of the conventional PMT detector base

circuit is necessary for high gain, clean transient response, linearity,

and prevention of space charge buildup at high currents (88, 96).

Impedance matching to avoid pulse distortion, reflections, and reduce

ringing is necessary. The use of neutral density filters prior to

the monochromator entrance slit uas used to extend linearity at

high signal levels. At high signal levels the PMT will saturate and

respond nonlinearly. Particular care in the above considerations

would have to be taken if a nondispersive optical system were used

with a high background atomizer.

Coupling the signal from the anode of the photonultiplier to the

signal processing device (e.g., gated integrator) is crucial to

maintain pulse shape and amplitude and avoid reflections and additional

noise pickup (e.g., RFI pickup). Cables should be kept as short as

possible, a good ground plane should be maintained, and impedances

should be matched.

To take full advantage of the increased gain provided by a high

intensity, low duty cycle pulsed excitation source, it is necessary

to use a gated detector synchronized with the excitation source.

The gated detector used in these studies was a boxcar integrator.

The gated integrator is a device that is used to measure and improve

the signal-to-noise ratio of repetitive pulsed signals. Any electrical

signal pulse whose intensity is to be measured, but which is too

fast to be recorded directly on a strip chart recorder, can be effectively

measured and averaged with a gated integrator. The gated integrator

is basically an RC integrator with a switch (gate) at the input. When

the switch is closed, an input pulse can charge the capacitor. If the

switch is opened immediately following the pulse and if the leakage

current is small, the capacitor will remain charged. If the RC

time constant is much longer than the signal pulse width, t a true

time integration of the pulse intensity will occur. However, the

capacitor voltage will be reduced from the signal voltage level by the

factor t /RC. When processing repetitive pulses, the capacitor
charge will increase with each pulse as long as the output voltage is

less than the signal voltage. Therefore, the integral of many pulses

appears on the capacitor, and an averaged value is obtained and displayed

at the output of the gated integrator. Because some leakage does

occur, the average is weighted towards more recent pulses. Not

only does the averaging process improve the SNR by approximately the

square root of the number of pulses averaged, but it also ignores all

noise signals which occur outside of the gate. Therefore, the main

advantage of using gated detection is that the detection system is

only active when the pulsed fluorescence signal (i.e., the signal of

interest) is present at the detector and that all other noise signals

are not processed. Only the noise signals present during the short

period the gate is open will be processed. Gated detection also

helps eliminate large transients (e.g., RFI) that often plague

pulsed systems. A trigger signal is necessary to synchronize the boxcar

integrator to the laser output pulse (and therefore the fluorescence

signal). This can be accomplished by either using the "sync pulse"

from the laser trigger unit or by beam splitting off a small fraction

of the laser output onto a high speed photodiode and using the photodiode

signal pulse to trigger the boxcar integrator.

The PAR Model 160 and Model 162/164 boxcar integrators use these

sampling and averaging techniques to extract synchronous waveforms

from noise. They synchronously sample the input signal with a

variable gate width, and a variable a;ate delay, which can be fixed

at any point on or slowly scanned across the input signal. That

signal which is passed by the gate is averaged by variable tine

constant integrators, the output of which is the average of some

number of repetitions of the input signal over the gate width interval.

Because the average of noise over a large number of repetitions is zero,

an improvement in signal to noise occurs. If the gate is fixed, the

output rises asymptotically tow.. I the amplitude of the input signal.

The boxcar integrators used here are sophisticated instruments whose

operation is complex; many inter-related and often subtle factors

must be considered for achieving optimum performance for the detection

of fast transient signals. In particular, the time constants used must

be suitable for analytical work, and the settings should yield the

highest possible signal-to-noise improvement ratio and precision with

respect to the chosen time constants.

Pulsed lasers used as excitation sources present many problems

which can be solved by a gated integrator. The troublesome characteristics

of pulsed lasers are: pulses are very short (2 8 ns); pulse-to-pulse

reproducibility is poor fluctuationss as large as 10% are common);

long term drift can be severe; electrical pickup can be comparable to

or larger than the desired signal voltage; jitter in the time between

triggering and laser pulse can be significant; and repetition rates

are low (1-100 Hz). Some of these problems are solved directly by

using a gated integrator. Others may be solved by the use of com-

binations of integrators or signal processing followed by integration.

As mentioned previously a single gated integrator rejects any signals

or electrical noise occurring outside the gate. Also, an integrator

with a small leakage current is not bothered by low repetition rates.

Further SNR improvements can result from using two integrators. A

small fraction of the incident laser excitation beam may be split

off using a small quartz flat. This fraction of the laser beam is

allowed to impinge upon a photodiode, and the signal from the photodiode

is fed into one integrator (reference channel). The fluorescence signal

processed by the other integrator may then be ratioed to the reference

channel. This provides a correction for both pulse-to-pulse laser

intensity fluctuations and long term drift as long as the fluorescence

signal remains on the linear portion of the saturation irradiance

curve (i.e., the log-log plot of fluorescence intensity vs source

spectral irradiance). Few integrators can respond accurately to the

3 10 ns fluorescence signal pulse produced from the N2 laser pumped

dye laser excitation pulse. The minimum gate width of the boxcar

integrators used in this study was 10 ns. Although the ideal gate

width would exactly match the fluorescence signal pulse, the wider

10 15 ns gate width was found to be necessary to compensate for

jitter and drift in the time between the trigger pulse and the

signal pulse. Another way to avoid jitter problems and improve the

SNR can be realized by signal processing followed by gated integration.

The signal processing consists of preintegration, amplification, and

pulse stretching. Stretching the pulse out to 1 5 Us will allow

the integrator to perform more effectively and will also reduce the

effective noise bandwidth and increase the charging duty factor. Pulse

stretching is generally followed by pulse shaping, during which high

frequencies are attenuated.

Experimental Conditions and Procedure

The nitrogen laser (Model UV-400, Molectron Corp.) was operated

under the following conditions: applied high voltage 26 28 kV;

pressure of nitrogen in laser channel 50 torr; and repetition rate

20 lHz. The N2 laser had a 10 ns pulse width and ca 400 kW peak

output power at 337.1 nm. The dye laser, which was pumped by the N2

laser, had 6 dye cells mounted in a carousel which rotates the desired

dye cell into the optical path of the laser cavity. Each cell contained

ca 2 ml of dye solution, which is magnetically stirred during operation.

The output of the dye laser had ca 3 8 ns pulse width and ca 1 80

kW peak power depending upon the dye and wavelength. The beam

diameter of the dye laser output was ca 0.3 mm. A beam expander/

spatial filter was placed in the optical path of the dye laser beam.

The beam expander was mounted on a support with fine x, y, z adjustments.

The pinhole (spatial filter) could be used to minimize the fluorescence

background of the dye laser output and also produced a more spatially

homogeneous excitation beam of radiation. The beam diameter at the

flame position was expanded to about 5 mm. A panel fitted with an

adjustable diaphragm was placed between the beam expanded and burner.

As will be seen and discussed in Table 5, this panel with diaphragm

played an important role in minimizing the fluorescence background

and the stray or scattered light of the laser beam.

A standard capillary burner head 10 mm in diameter (about 50

capillaries of 1 mm i.d. for the air/C H flame and of 0.84 mm i.d.

for the N20/C2H5 flame, arranged into an area 10 mm in diameter) was

used with an inert gas sheath which provides flame "separation" (84).

The burner head was mounted on either a Perkin-Elmer chamber/

nebulizer assembly or an Instrumentation Laboratory chamber/nebulizer

assembly. The entire burner system was placed on an adjustable mount

which allows x, y, z, 0 positioning of the burner system. Flame

gases were passed through calibrated rotameters after two-stage

regulation at the cylinders. Before the flow meter, the acetylene

gas was passed through a trap of charcoal (for elimination of acetone

vapor) and a phosphine (PH3) trap (for elimination of phosphine

included in commercial acetylene gas as an impurity)(102-104).

Air was passed through a trap of Drierite (for removal of water vapor,

The :;.A. Hammon Drierite Co., Xenia, OH 45385). The flow rates of the

flame gases passed through rotameters were calibrated with a linear

mass flowmeter (Model ALK 50k, Hastings-Teledyne, Hampton, VA).

The fluorescence was imaged on the monochromator entrance slit

through a lens (76 mm focal length, 50 mm diameter), which was mounted

inside a light baffle. The light baffle (60 mm i.d.) had an adjustable

diaphragm (30 mm i.d. maximum) mounted on the front and also allowed

for positioning of the lens. The diaphragm was set to optimize the

fluorescence signal with respect to the scatter. The monochromator

was a 0.35 m focal length Czerny-Turner grating spectrometer (2.0 nm/mm

reciprocal linear dispersion, 1180 lines/mm grating blazed at 250 nm).

The detection system consisted of a photomultiplier tube, a boxcar

integrator, and a chart recorder. The electronic circuit for the

photomultiplier base was the same as the one previously used by

Fraser and Winefordner (45). The photomultipliers were operated at

1000 V for the R212 UH and 1200 V for the R818. The boxcar integrators

(PAR Model 160 or Model 162/164) were operated with 502 input, an

aperture time ca 15 ns, and for the Model 160 an observed time constant

between 0.5 s and 5 s with 50 mV to 10 V imput sensitivity, and for

Model 162/164 an observed time constant for the gated integrator

(Model 164) of 3 s and the main frame (Model 162) time constant

0.1 ms to 1 s, with an input sensitivity of 100 mV.

The gas flow rates were ca 2.3 1-min-1 of acetylene and 12.3 1-min-

air for the air-acetylene flame, and ca 6 1-min-I of acetylene and
10.6 1 min- of nitrous oxide for the nitrous oxide-acetylene flame.

Most measurements were performed under slightly fuel-rich flame

conditions. The atomic fluorescence signals were maximized by changing

the flow rate of acetylene in both of the flames. Argon gas with
ca 10 15 1-min- flow rate was used for flame separation in both

of the flames. An 800 or 1000 pm slit width of the monochromator

was used in most resonance fluorescence cases, unless scatter and

background noises were found to be small, in which case 2000 pm slit

widths were used. In most nonresonance fluorescence cases and most

cases in the UV (<350 rm), 2000 pJm slit widths were used.

After optimization of the experimental arrangement and conditions,

analytical curves were measured for each at their atomic fluorescence

lines in suitable flames. The rms noise levels were evaluated by

measuring 1/5 of the observed peak-to-peak noise on the baseline (98)

while aspirating deionized water (blank). The detection limits were

then evaluated by extrapolating the analytical curves to a signal-to-noise

ratio of 3 (77).


All the chemicals used were reagent grade. Stock solutions of

each element (1000 Ug-ml-) were prepared in accordance with the

procedure given by Smith and Parsons (105). Deionized water was used

as a blank solution in all the experiments.


Based upon the preceding considerations, a pulsed N2 laser pumped

tunable dye laser should provide nearly an ideal source for atomic

fluorescence spectroscopy. The flame has proven to be the most common

and convenient atomization source in atomic spectroscopy. Combining

the above with a gated detection system to greatly improve the signal-

to-noise ratio (SNR) should produce results which surpass any

other analytically useful method for trace elemental analysis. The

results previously obtained were only comparable to those obtained

by other atomic methods. In this study, the experimental arrangement

and conditions were investigated in order to optimize the analytical

results. Analytical calibration curves were obtained for 23 elements

and 37 transitions, many of which lie in the UV region of the spectra

and, therefore, require frequency doubling of the dye laser output.

Several analytical figures of merit are discussed for the method

of pulsed laser-excited atomic fluorescence flame spectrometry.

Optimization of System

Reduction of Scatter Signal

As mentioned earlier, the characteristic advantage of the pulsed

excitation-gated detection system is lost when the fluorescence

signals are limited by source-carried (scatter fluorescence background)

shot noise. This circumstance can be avoided by using nonresonance

atomic fluorescence lines. However, because many elements such as

Ba, Ca, Cd, Li, Mg, Na, and Sr (mostly alkali and alkaline earth

elements) have only one strong resonance fluorescence line, source-

carried shot noise can deteriorate their detection limits. Tn order

to overcome this situation for resonance fluorescence lines, an effort

to reduce the scatter noise and optimize the optical system was made

in the present experiment.

The experimental system used for the optimization study is shown

in Figure 9, and the results are summarized in Table 5. In these studies,

strontium was chosen as a typical element which only has available a

resonance atomic fluorescence line. As can be seen from Table 5, the

use of a focussing lens between the dye laser and the flame does not

improve the SNR because of the low signal levels and the large scatter

signal levels. The low signal level in case (i) is due to the

excitation and observation of smaller flame volume. The use of a

panel with a diaphragm mounted on it and a light trap (cases (ii)-(iv)

in Table 5) reduced the scatter signal level to almost half of that in

case (i); this indicated that the scatter signal observed in these

cases was due to the scattered light reflected by the surroundings, not

due to the unvaporized particles in the flame. It should be stressed

here that the addition of only the beam expander (case (v)) without

a spatial filter enhanced the atomic fluorescence signal, but also

increased similarly the scatter signal. The large scatter signal,

which varied with wavelength, was clearly a result of radiation scattered

by the surroundings. However, the panel and the light trap which were

set at the positions shown in Figure 9 reduced the scattered signal

level as well as the noise level. Therefore, the best SNR was obtained

in case (xii), where the beam expander, the panel, and the light trap





Figure 9: Optical arrangement around the burner for tle examination of
optimization of the optical system (see Figure 1 and Table 30)
FL: focusing lens for laser beam (focal length 17.8 cm),
BE: beam expander, P: panel with diaphragm, LT: light trap.

Table 5. Optimization of Optical Arrangement in Laser-Excited
Atomic Fluorescence Spectrometrys

Code Arrangement

(i) FL

(ii) FL, P

(iii) FL, LT

Signalc Noisec Scatterc
S N Sc S/N S/Sc

12 1

12 1

12 1

(iv) FL, LT, P 12 0.8

76 12 0.16

45 12 0.27

35 12 0.34

38 15 0.32

48 3 145 16 0.33

(vii) LT

(viii) P

(ix) P, LT

(x) BE, LT

(xi) BE, P

21 0.8

22 0.8

23 0.8

25 0.8

48 26 0.44

32 27 0.7

45 29 0.6

42 31 0.6

41 0.7 27 59 1.5

41.5 0.4 6.5 104 6.4

(xii) BE, P, LT 42 0.3

3.5 140 12

aA solution of 0.1 ppm Sr was used. Experimental conditions and
instruments were described in the previous chapter.

See Figure 9. In the optical arrangement, each device or a
combination of the devices was set around the burner (flame),
as can be seen in Figure 9.

Relative units.

(v) BE

were simultaneously used. According to this optical arrangement, the

SNR was improved by more than a factor 10, and the fluorescence signal-

to scatter signal ratio by about a factor 75, compared to those obtained

by the use of only a lens (case (i)), as can be seen in Table 5. As

a result of these studies, the optical arrangement of case (xii) was

employed in all of the following experiments. It should be noted that

a rigorous comparison was not attempted, because arrangement (xii)

proved to be significantly superior. Considerations such as reflection

losses at lens surfaces, the laser beam shape, variation in laser

intensity with time, collection efficiency, etc., needed to compare

the arrangements on an absolute basis were not taken into account.

The scatter signals for many analyses using arrangement (xii) were

found to be very small and due mainly to scatter from the environment

(not the flame). Due mainly to the lower output and the lower

reflectivity of surrounding materials in the frequency double range,

scatter was found to be negligible for lines in the 220-330 nm range.

Slit Width vs SNR

For most spectral measurements using conventional dispersive

spectrometers, relatively narrow spectrometer slit widths must be

used to distinguish the analyte signal from the signals of other

elements and background components; this leads to collecting less

signal with the monochromator-detection system (i.e., lower optical

conductance). The spectral narrowness of the source allowed larger

slit widths to be used with only a solid angle consideration for

collection of scattered light at resonance wavelengths (i.e., additional

radiation from a spectral continuum or inert gas lines does not exist

and the fluorescence background of the laser beam is orders of

magnitude smaller and can be minimized and spatially filtered if

necessary). The fact that the laser is a single tunable source

with only one narrow spectral line emitted is a primary advantage for

atomic analysis. Recently, interferometric and multiplex instruments

which permitted the use of a wide slit width (Jacquinot advantage)

have been investigated in order to overcome the disadvantage of

conventional dispersive spectrometers mentioned above. However,

these situations are not valid when the analyte signal can be spectrally

differentiated from any background by other means (106, 107).

In laser-excited atomic fluorescence spectrometry, signal selectivity

depends upon the spectral bandwidth of the laser output, not upon the

spectral bandwidth of the monochromator. An example of signal

selectivity is shown in Figure 10 (a) in terms of the excitation of

sodium D lines doublett separated by 0.6 nm). The spectral bandwidth

of each line is ca 0.03 nm despite the use of an 800 pm monochromator

slit width, i.e., spectral resolution is better than 0.03 nm, which is

determined only by the spectral bandwidth of the laser output (a

spectral profile of sodium atomic fluorescence observed by wavelength

scanning the monochromator spectromether is shown in Figure 10 (b),

where an 800 pm slit width was used). Because the emission signal

of sodium is essentially negligible due to the low duty factor (3 x 10-7

here) of the pulsed source-gated detection system, wide monochromator

slit widths can be used in the present instrumental system. The

spectral purity of the source is seen in Figure 11 more clearly, where

the fluorescence signal vs slit width is shown in terms of sodium

(resonance atomic fluorescence; excited at 589.0 nm and observed at

589.0 nm) and thallium (both of resonance and nonresonance atomic

II G irm
0.03 nrnI

I _____ L L L

509.0 58 9. 592 588

\avelength (n!m)

Figure 10: Excitation and fluorescence profiles of sodium D lines
(at 589.0 and 589.6 nm) in the air-acetylene flame

(A) Profile observed by scanning tlhe laser
(B) Profile observed by scanning monochromator
Slit width was 800 im in both cases


S500 o -
( )

.o; 0
*- ;0 / /
2 / /

S/ /
L 1001 /< -

-2} / 0/-
-5 tool- /o/

0 500 1000 !500 2000

Slit Width,/ irn

Figure 11: Dependence of atomic fluorescence signal of Na and T1 on
slit width observed in the air-acetylene flame

-0- : resonance fluorescence of sodium. Excited at
589.0 nm, observed at 589.0 nm (---: Scatter level)

0 : resonance fluorescence of thallium
Excited at 377.6 nm, observed at 377.6 nm

I : nonresonance fluorescence of thallium
Excited at 377.6 nm, observed at 535.0 nm

fluorescence; excited at 377.6 nm and observed at 377-6 nm and 535.0 nm;

respectively). As can be seen in Figure 11, all fluorescence signals

increase almost linearly with increase of slit width. In the case

of sodium, the curve has three different slopes in the ranges of

0-300 um, 300-1000 pm, and >1000 pm. This can be interpreted as follows.

Spectral overlap between the lines at 589.0 nm and 589.6 nm takes place

with the slit width wider than 400 pm, i.e., both of the sodium D lines

are observed by the spectrometer with the wide slit width (simultaneous

multiple line observation). This gives the larger signals (i.e., a

larger slope in Figure 11) observed by spectrometer in the slit width

range 300-1000 pm. The simultaneous multiple line observation suggests

another advantage of laser-excited atomic fluorescence spectrometry,

when multiple lines of each element are within the slit width used

(e.g., Mn), in that greater signal levels are easily obtainable.

In Figure 11, the slope of the curve for sodium becomes smaller for

slit widths exceeding 1000 pm slit width because of saturation of the

photomultiplier anodicc current); saturation was evaluated by using the

neutral density filters to reduce the signal to the photomultiplier

detector. For analyses at high concentration, the monochromator

slit width could also be reduced to minimize saturation of the photo-


In Figure 12, the curves of SNR vs slit width are shown in the

cases of sodium and thallium, which were observed under the same

experimental conditions as those in Figure 11. For both the Na and Tl

resonance cases, the SNR's reach plateaus at slit widths between 800 Um

and 1000 pm. Therefore, the system becomes source-carried noise

limited above a slit width of 1000 pm in the present experimental

Figure 12: SNR vs Slit width of Na and T, in the air-acetylene flame

og ( si .idhA )

Figure 12: SNR vs Slit widtli of Na and T1 in the air-acetylene flame

0 : resonance fluorescence of sodium
Excited at 589.0 nm, observed at 589.0 nm
o : resonance f]iiorescence of thallium
Excited at 377.6 nm, observed at 377.6 nin
S: nonrcsonance fluorescence of thl llii m
Excited at 377.6 nm, observed at 535.0 nm

arrangement, and Equation (16) applies. Scatter signal levels are

shown in Figure 11 for Na. Considerations which determine the optimum

monochromator slit width in atomic fluorescence are (i) type of

transition; (ii) source irradiance; (iii) optical configuration (i.e.,

efficiency of signal collection and degree to which system is baffled

and apertured to eliminate environmental scatter); (iv) saturation of

photomultiplier anode current; (v) flame background fluorescence; and

(vi) spectral bandwidth of source radiation and spectrometer. In

the present experimental arrangement, an 800 or 1000 Um slit width

provides a generally useful setting, because the scatter noise limit

is approached here as determined experimentally for Na and T1 by

the SNR vs slit width plot (Figure 12).

However, in the nonresonance case, no leveling off of the slope in

SNR vs slit width curve was observed. This indicates that the signal

was not source-carried noise limited (Equation (15) approximately

applies here). Therefore, for nonresonance fluorescence, a nondispersive

system, using, for example, only interference filters and neutral density

filters before the photomultiplier, should yield the optimal system

(i.e., lowest limit of detection and largest linear dynamic range). It

should also be noted that because fluorescence is emitted in 4r steradians,

a system could be designed using two detectors and two optical arrange-

ments to optimize both resonance and noiresonance signals with little

added expense or inconvenience.


Flame conditions were adjusted to give the maximal signal. For

most elements in either air/C H or N20/C 2 a slightly fuel-rich

flame with its good reducing properties yielded the best results in

atomic fluorescence spectrometry. A more fuel-rich flame was used in

some cases (e.g., Al, Bi, Cr). A fuel-lean flame yielded slightly

better results for Ni, Ag, Cu. The flame height at which the atomic

fluorescence measurements were taken was ca 20-30 mm above the burner

head. The position of the exciting beam in the flame was not extremely

critical due to the larger beam diameter. No significant difference

in signal intensity or SNR was observed between the "separated" or

unseparatedd" flames (i.e., use of ai inert sheath gas around the

flame does not appear to be necessary).

Air-acetylene and nitrous oxide-acetylene flames were used here

because of their relative freedom from matrix interference, good

atomization characteristics, and simple procedure for safe utilization.

Analytical Figures of Merit

Limits (Powers) of Detection

The usefulness of any analytical method is given by its figures

of merit. The limits of detection are the most often quoted figure

of merit because these values indicate the detection power of the

technique. For proper choice of a method for any analysis, it is

necessary to compare the limits of detection of all possible methods.

Therefore, limits of detection will be mainly discussed in the

following discussion.

The detection limits and analytical curves were obtained for

23 elements, including 37 transitions, and 7 different fluorescence

processes. Ten elements were observed using the frequency doubled dye

laser output, most of which represent the first time these transitions

have been observed by laser-excited atomic fluorescence. The work also

demonstrates the wide tuning range of the N2 laser pumped dye laser

as a single line source for atomic spectrometry as elements from Cd

at 228 nm to Li at 671 nm were observed.

The limits of detection have been improved by a factor of ca 10-200

times over those previously obtained with a N2 laser pumped tunable

dye laser system for all elements examined, except cobalt. The

linearity of the analytical curves has also been extended to about

five orders of magnitude for almost all elements examined. These

results are presented in Table 6 together with all pertinent data and

conditions necessary for discussing detection limits except for the

system time constant which was 1 s for most cases. In all cases,

at least five time constant intervals were allowed before signals

were measured. The noise level was evaluated by taking 1/5 the

peak-to-peak noise over at least a 50 s time period when only the

blank was being aspirated (98). This approach was equivalent to

calculating the standard deviation of 16 separate measurements for

evaluating the noise, N. The limit of detection (LOD) was calculated

on the basis of a signal-to-noise (SNR) equal to three [as recommended

by IUPAC (77)]. However, it should be noted that, it is generally

accepted that a SNR 2 10 should be used for practical determination.

In Tables 6 and 7, the results are compared with previous laser-

excited fluorescence studies in flames, and with the best values for

the limit of detection obtained by conventional source flame atomic

fluorescence (AF), flame atomic emission (AE), flame atomic absorption

(AA), and by atomic emission induction coupled plasma (ICP). It

should be noted that all limits of detection except those determined

in the percent work were bosed on a SNR = 2 value. In Table 7, it

is clearly shown that laser-excited atomic fluorescence flame spectrometry

Table 6A. Detection Limits by Laser-Excited Atomic Fluorescence
Flame Spectrometry (Experimental Conditions and Spectral Parameters)

Elicrt (e. )7

Ag 324.1
A1 331.4

Ba 553.7
BS 306.8
Ca 422.7
Cd 224.E
Coe 357.5/347.4
Cr 359.3
Cu 324.7

Fe 296.69/373.49
Ga 4C3.3
In 410.4
LI 670.8

Mg 265.2
Mn 403.1
No 379.8
Na 593.0
Ni 232.0

Pb 405.8
Sr 4'0.7
Ti 3,9.9
11 377.6
V 411.2

0 IS300

0 23652


0 16973

2425-23.1/79425-26 61

Type of
- AFt'' _p v lu.

5-DI F














0 .00




Eye' _______Typed

(E1:4;CVP) A/A
Pt'l N/A
rl0io N/A

C.95 A/A
C435 N/A

(FF) A/A
Eis- -B A/A
(7'V:!C) A/A
(7M."C) A/A


(RF CV") or A/A
(C495) A/A
(C495) A/A

877 N/A

R(G- A/A
P.tE A/A
(7'.4HC) A/A
rcD A/A

U;3 A/A
(C;95) A/A
(0495) A/A
7U4i;C A/A
Bhq A/A

'EX/Fi. eclr,-il i iavelenCrti/fluoIe nc5cr ,..l\rugch (if dAfE.r.enc. chine, ecltlon s vlelngth ), ndi
EL energy I-vl.
IValuc tan fro-I C.H. Corlli, .F.. soa5o ,n "E>perimenta1 Transition ?lobab'llIles for Spectral Lines
of Seventy El-i. .nc x..F ,nraph 51 (1162).
RF rcoi:,,,,ce [Iloresn ce; E-RF ejcit:. resii ience f luorrscrcr; S-I.F Stok-s direct lnre fluorcs-
Ci-:" A%-DL ant- Sto-l, dict l1ine le. f ',, nrce TA-S F t!eri l n tscel stp'ice linc
fltr-c,,' e-r; F-AS- DLF exc-ltd ,,-l..tke, Jitot l1IL f I i',.., n,-e; alld F-S-SL .cltcJ Stokes
steP.Ise line [u1.1ruenct. (re Relerence 10 for dec.a1cd disunlsien).

cSe Table 4. Farantheses ecan thit the dye la.rt output is frtquency doubled.
dA/A alr-a.iceyl e a itee and f/4 nitrous oxide-acerylere flr.e.

E 2xperenital diff culty 4,i encountered In obttintrg god lae.r ouIp... at th Co line. attenmped.

Slit idtch





Table bB. Detection Limits by Laser-Excited Atomic Fluorescence

Flame Spectrometry* (Analytical Figures of Merit)

Ele: at ~_(onl)

ip. 4PO

Al -
60 1

Bi IP0

85 3PO0

Ca 8P-2

Cd 8FO

Co 1IP3

Cr 1PO

Cu 1PO

Fe 3P1

Ca 7PO

In 8P-1

LI 5P-1

K6 2F-1

Y.n ] PO

Ko IP1

Na <1P-1

NI 100

Fb 2P2


Sr 3P-1


T1 A0ro

i a LOnrk
UL aoLO.)

no,/nl) LPlR

6P4 ?P4

61'5 3P5
3P5 5P5

1P7 P15
>106 >1P5

5'5 215

6P3 8P4

3P3 3i'3

>2P6 >IF3
>2P6 >1P3

3?5 3F5

1P5 105

>1P6 >3P4

305 404
2P5 2P5

4P5 53'5
3P5 2P6

>1P4 >2P4

2P4 1F5

8r4 8!4
1P5 3P5

>1P6 1F5
>1P6 1P5

5P4 5P5
6Pi 6F4

1F5 1'4
1F6 5P5

>1P6 >5P3
>1P6 >3P4
1P6 5P4
>1P6 >1P5

3P4 1P5

1P6 2P5
4P5 2P5

3P0 704
6P5 9P4

-__ (' ,)__ (ny/a!)

1F2 2i'5
3P2 1F5
5P(1 2?5

PE0 >ir6
4PI --

5P0 5P0
5'0 5r4

L.hFla Re ferr eb

1P3 1
IF4 1
1F5 1

>1?5 2
-- 3

2P4 7P3

1P5 102
1P5 3P2

2P5 4F3

V 5P1 >IF6 >1P5 --- -- --
3PI >1P6 >3P4_ 5P2 --- 5
*APB tL.an Ax 106.

LOll nit Ir of Detection. UL Uperr .lrrit of linearity, and IDR linear dyna"lc r.ar. -
UI /1.,'.

Vallu taken fro-,: (I) Referenre 45; (2) Refercnce 55; (3) Refcrence 56: (4) Roferenoe 53;
(5) iferefr nc 47; nai (6) Relfre nce 54.

Ctxperlrcntal difficulty er n rncountered in obtainlnp o pd laser outpu at th, Co Ilno-
mL Iro'' 'l.

Pr~lour vo~t


Table 7. Comparison of Detection Limits in Flame Spectrometry

and Inductively Coupled Plasma (ICP)

Detection limit. n cM-1

laser convent onaL
Thi YOrkn source
4 0.18e
0.63 100
B --
33 58
0.08 20
8 0.0016
1000 56

1 '0.5
30 8-
0.98 10
0.23 100
0.5 ---
0.2s 0.16
0.40 1
128 500
<0.1B --
28 38
138 100
0.2 30
28 4000

43 88
30 70

Ab pnec- nitc ultrasonic
rebulieationo Anebul i tion

1 4t --



30* 2
20 O.1 t
50 50-4
1' 0.07!8
1 1.0
23 20
25 0.93
1* 0.23
4 0.28
50 141
30 30t
1 --
0.18 0.7t
0.8 0.18
30 43
0.8 0.2f
58 48
106 103
5 0.02f5
90* 38*
20 203"
20 18






alrlfts of detection represent concentration o required to produc-. a line signal three rimes
ao reat as .te standard deviat-on of Ith b-pcVround noise. All othri values licid in the
able Irereesent concenratlons req.. red to produce a Ilne Esignal tEice as great as the
standard devialion of the background noise, except heree noted.

bValues Ctaen from J.D. WVinefordner, J.J. Fitzgerald. and N. C2-netc'o, eApl. fcctrose., 32,
369 (1975) except for crhsc Jeslgnuce *, lhich .ere taken from V.A. Fassol and R.N.
Fnlseley, Anal. C-tn., 46, 111OA (1974).

CALl values tanci fro'i K.'. Olson, S.J. ins, Jr., and V.A. Fasel, An1l. Ch m., 1. 632 (1977)
enc.pt those with V. whlch are from V.A. Fassel and R.N. Kliseley, An:.il. LChlI., 46. 1110.1 (1974).

dAll values taken froci P.V.J.M. BOLIL, ,2 an F.J. der ocr, rSectrochin. Acca., InB, 3 19 (1975)
exc r t those de..)t-njted ihlcih Lcre ''-.'n front .W Olso0100, V.J. F .i. Jr., and V.A. Fassel,
Fl Cheo .,. 49, 632 (1 777) ..d. theCo e d ,s i.n.:rid fhich w re akl, f~rom '.H. AbC il.'h,
R. Dl.3ieli zuni k, J. Ja-r z, Jt.M. crcmi t, Rnhin, anl C. Trm-.y, An I. Chei Acti, 8', 271
(1976). Linlt of dteLittio ds1 .nactRr- ** tler event cuoicent.ratin~t rnl-rcid to produce line
sigonl six tcl-,s es great ay thet stad.idi dviatlion of the background no-ise.

e* dcs gnnteI b.cs vut-.- vwith pneua:tic i b hu lizacon. All values within a factor of 3 are
considered to be equal.

--- indicates co v :c reported.

L'rPt Int'tical dif iclulties wrie etiounttetd in hob.iinng good liser output at the t. lines.









is equivalent or superior to all the methods listed where pneumatic

nebulizers are used. The use of an ultrasonic nebulizer should also

provide ca 10 fold improvement in the detection limits (108). As

noted earlier, optimization of the fluorescence emission optics should

also improve nonresonance fluorescence detection limits.

The analytical calibration curves plotted in Figures 13-16 are

divided according to whether the element (i) has transitions which

are excited by the fundamental (>355 nm) or frequency doubled (>355 nm)

dye laser output; (ii) has only strong resonance transitions or has

both strong resonance and nonresonance transitions; and (iii) is

best excited in an air-acetylene or nitrous oxide-acetylene flame.

The limits of detection for the nonresonance cases are typically

one order of magnitude lower than for the resonance cases, as seen

in Figure 14.

Linear Dynamic Range

A very important figure of merit to the practical analyst, which

often is not reported and compared, is the linear dynamic range (109),

i.e., the linearity range of the analytical calibration curve. A

technique which offers a low limit of detection, but whose analytical

curve is only linear over one or two orders of magnitude is analytically

practical but is difficult to use, because often it means time consuming

dilutions will have to be made until one is ensured to be within the

linear range for that technique or one is destined to work within

the nonlinear portion of the analytical curve requiring extensive


The linear dynamic range is determined by the lower and upper

limits of concentration. The upper limit of concentration is determined

Figure 13: Analytical Calibration Curves for Elements having only
strong resonance transitions above 355 nm in air-acetylene
flame ,I _

resonance fluorescence of Ba
resonance fluorescence of Ca
resonance fluorescence of Cr
resonance fluorescence of Li
resonance fluorescence of Mn
resonance fluorescence of Na
resonance fluorescence of Sr
indicates the limit of detection.)

=fl knm)



o /

S1 0 / 0
I- /

0Io /

10o 10 io 102 Loi fo5 106

Analyle Concen:trai.cn (ng-:n,')

Figure 14: Analytical Calibration Curves for elements having both
strong resonance and nonresonance transitions excited
above 355 nm in an air-acetylene flame.

eA (nm) A f(nm)
ex fl
A : resonance fluorescence of Ga 403 403
: nonresonance fluorescence of Ga 403 417
O : resonance fluorescence of In 410 410
S: nonresonance fluorescence of In 410 450
S: nonresonance fluorescence of Ni 361 352
O : resonance fluorescence of Pb 405 405
X : nonresonance fluorescence of Pb 405 283
-0- resonance fluorescence of T1 377 377
: nonresonance fluorescence of T1 377 535
(@ : indicates the limit of detection)

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs