VITREOUS CARBON TUBE FURNACE FOR ATOMIC
CHARLES J. MOLNAR
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA IN PARTIAL
FULFILLKENT OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
This dissertation is dedicated to all my friends,
and especially to Jude, the best friend I have ever known.
The author is especially grateful to his research
director, Dr. James D. Winefordner, for his advice and
encouragement in the course of this research. The
scholarly assistance of the members of his committee is
also greatly appreciated.
Dr. James D. Winefordner's postdoctoral fellows and
graduate students will always be remembered with gratitude
for their constructive suggestions and invaluable help.
The author also wishes to acknowledge the technical
advice and assistance of Francis D. Ottinger and Arthur P.
Grant, as well as the other staff members of the Machine
and Instrument Shop.
An expression of thanks is due Gail G. Pokrant for
all her help with this and other projects.
TABLE OF CONTENTS
ACKNOWLEDGMENTS . .
LIST OF TABLES . .
LIST OF FIGURES . .
ABSTRACT . . .
I. INTRODUCTION . . . . . . . . .
II. THEORETICAL CONSIDERATIONS . . . . .
Types of Atomic Fluorescence Transitions.
Intensity of Atomic Fluorescence . . .
Fluorescence Quenching Considerations . .
Noise Considerations . . . . . .
Atom Generation and Signal Measurement. .
Concentration of Analyte in the Atom Cell .
Limit of Detection Considerations . . .
III. EXPERIMENTAL . . . . . . . . .
Reagents . . .
General Layout of the
Sources of Excitation
Optical System . .
Electronics . . .
Nebulizers . . .
. . . . . .
Experimental System .
. . . . . .
. . . . . .
. . . . . .
. . . . . .
Desolvation Chamber . . . . . . 45
Vitreous Carbon Tube Furnace. . . . 48
Nebulizer, Desolvation Chamber, and
Vitreous Carbon Tube Furnace Mount .. . 53
Procedure for Evaluation of Experimental
Characteristics of the System. . . 53
Gas flow rates ........... 53
Sample solution flow rates ...... 54
Mean particle diameter . . . .. 54
Desolvation time . . . . ... 56
Scattered radiation. . . . .. 59
Rise velocities. .......... . 59
Efficiency of sample transport ..... 61
Temperature of furnace and gaseous vapor 61
General Procedure for Making Atomic
Fluorescence Measurements. . . . 64
IV. RESULTS AND DISCUSSION . . . . ... 66
Atomic Fluorescence Signal-to-Noise
Optimization . . . . . ... 66
Definitions of Analytical Parameters. . 69
Continuous-Sampling Mode Results. ... . 70
Limits of detection . . . . . 70
Linear dynamic range and sensitivity . 73
Correction for scattering. . . . 80
Precision. . . . . . . 82
Atom population profiles . . . .. 82
Pulsed-Sampling Mode Results. . . .. 86
Limits of detection. . . . . ... 86
Size of sample . . . . . . 87
Linear dynamic range and sensitivity . 92
Precision. . . . . . . . 92
V. SUMMARY AND FUTURE WORK . . . . ... 96
Summary. . . . .... . * * 96
Future Work. ........... .... . 98
Desolvation chamber considerations. . 98
Atomic emission considerations. .... .. 103
Atomic absorption considerations 107
Stop-flow considerations . . 108
Conventional resistively heated
non-flame cell . . . . . .108
I. DEFINITION OF SYMBOLS .......... 111
II. ROTAMETER CALIBRATION . . . ..... . 117
LITERATURE CITED . . . . ..... 122
BIOGRAPHICAL SKETCH. ......... . . . 129
LIST OF TABLES
1. Stability of Aerosol Production with High
Pressure Nebulizer . . . . . .... 55
2. Comparison of Residency Times in Atom Cells. ... 60
5. Experimental Conditions and Analytical Results
for Measurement of Several Elements by Means of
the Continuous-Sample Introduction Tube Furnace
Atomic Fluorescence Spectrometric System . . 71
4. Comparison of Non-Flame Atomization Atomic
Fluorescence Spectrometric Limits of Detection 72
5. Experimental Conditions and Analytical Results
for Measurement of Several Elements by Means of
the Pulsed-Sample Introduction Tube Furnace
Atomic Fluorescence Spectrometric System . . 88
6. Comparison of Flame and Non-Flame Atomization
Atomic Absorption and Fluorescence Limits of
Detection and Plasma Torch Limits of Detection 90
LIST OF FIGURES
1. Physical process leading to atomic absorption,
atomic fluorescence, and atomic emission
spectrometry . . . . . . . .. 3
2. Types of atomic fluorescence . . . ... 10
3. Schematic diagram of the atom cell . . . 13
4. Atomic fluorescence growth curve with a line
source of excitation . . . . . ... 17
5. Atomic fluorescence growth curve with a
continuum source of excitation. . . . ... 18
6. Schematic diagram of continuous- and
pulsed-sample introduction tube furnace atomic
fluorescence spectrometric system . . 33
7. Cutaway view of the EDL thermostating unit. . 36
8. Schematic diagram of nebulizer I. . . ... 43
9. Schematic diagram of nebulizer II . . ... 47
10. Schematic diagram of the stainless steel
desolvation chamber . . . . . . .. 49
11. Schematic diagram of vitreous carbon tube
furnace . . . . . . . . . . 51
12. Gas temperatures in the desolvation chamber at
various heights above the nebulizer . . .. .58
13. Temperatures of non-flame cell at various
points. . . . . . . . . . .. 63
14. Atomic fluorescence analytical curve for Te in
the continuous mode of operation. . . . 74
15. Atomic fluorescence analytical curve for Sn in
the continuous mode of operation. . . . 75
16. Atomic fluorescence analytical curve for Ag in
the continuous mode of operation . . ... 76
17. Atomic fluorescence analytical curve for Bi in
the continuous mode of operation . . ... 77
18. Atomic fluorescence analytical curve for Pb in
the continuous mode of operation . . ... 78
19. Atomic fluorescence analytical curve for Tl in
the continuous mode of operation . . .. 79
20. Decay of atom populations with height above the
vitreous carbon tube atomizer for two elements
as measured by atomic fluorescence spectrometry. 84
21. Fluorescence signal versus sample size for Bi,
Sn, and Co with pulsed mode of operation . . 91
22. Atomic fluorescence analytical curve for Sn in
the pulsed mode of operation . . . . . 93
23. Atomic fluorescence analytical curve for Co in
the pulsed mode of operation . . . ... 94
24. Atomic fluorescence analytical curve for Bi in
the pulsed mode of operation . . . ... 95
25. Fluorescence signal versus time after
introduction of three pulsed samples of 10 ul
of 30 pg ml-1 Sn at 303.4 nm . . . . . 99
26. Schematic diagram of a modified stainless steel
desolvation chamber. . . . . . . ... 101
27. 2.3 Log concentration in the atom cell versus
time after introduction of the pulsed sample
for two volumes of exponential dilution flasks
with identical initial amounts of analyte. .. 102
28. Emission signal versus gas temperature at
grazing incidence to the vitreous carbon tube. 105
29. Emission signal versus external H2 flow rate . 106
30. Ar nebulizer flow rate versus rotameter
reading . . . . . . . . .
31. External Ar flow rate versus rotameter
reading at 60 pounds. . . . . .
32. External H2 flow rate versus rotameter
reading at 12 pounds. . . . . .
33. Internal CH4 flow rate versus rotameter
reading at 8 pounds . . . . . .
. . 118
. . 119
. . 120
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
VITREOUS CARBON TUBE FURNACE FOR ATOMIC
Charles J. Molnar
Chairman: James D. Winefordner
Major Department: Chemistry
Resistively heated non-flame cells previously used
for atomic absorption and atomic fluorescence spectroscopy
are reviewed. A versatile new type of resistively heated
non-flame cell usable in both the continuous introduction
and the pulsed introduction modes is described, and
analytical figures of merit in atomic fluorescence
spectrometry are given.
The theory of the conversion of the analyte to an
atomic fluorescence signal is reviewed, along with some
reasons why non-flame cells in atomic fluorescence
spectrometry should attain better signal-to-noise ratios
than flame atomic fluorescence spectroscopy. The theories
of transient and continuous signals are discussed, and
methods to obtain maximum signal-to-noise ratios for each
The design, construction, and evaluation of the new
versatile resistively heated non-flame cell are discussed.
The non-flame cell system consists of a very efficient
pneumatic nebulizer operating at high pressures with low
sample consumption for introduction of a fine sample aerosol
into a vitreous carbon tube furnace via a desolvation
chamber. Atomic vapor of several metals (Sn, Pb, Te, Ag,
Tl, and Bi) is produced by continuous pneumatic nebulization
of an aqueous sample through a vitreous carbon tube furnace,
and atomic fluorescence is excited with single-element
electrodeless discharge lamps. The high efficiency of the
system is considered along with precision, sensitivity,
linear dynamic range, and limits of detection for the elements
examined. The decay of atom populations above the furnace
outlet using either an Ar-H2 diffusion flame sheath or an Ar
sheath is determined, and the results are discussed.
Pulsed pneumatic nebulization of aqueous samples,
containing various metals (Ag, Bi, Cd, Co, Mg, Pb, Sn, Te,
Tl, and Zn), into a vitreous carbon tube furnace produced
metal vapors which are excited with single-element electrode-
less discharge lamps, and the resulting atomic fluorescence
is measured. This mode of operation improves the absolute
limits of detection; no deleterious effects to the linear
dynamic ranges resulted. The concomitant increase in the
life of the vitreous carbon tube is also stressed. The
atomic fluorescence intensity versus different sample sizes
for a few elements is determined.
Additional advantages of the present non-flame cell
system over some past non-flame cells are that operator
error due to sample placement is eliminated, and that the
conversion of the atomizer from the continuous mode to the
pulsed mode of sampling is achieved by simply altering the
means of sample injection.
Some areas of future research are discussed.
Preliminary results suggest that the present resistively
heated non-flame cell offers atomic emission results
competitive with some flames. A method for the improvement
of the transient signal is discussed. The use of this
system for atomic absorption is discussed, along with a
possible alteration. Some other areas of future research
Atomic Absorption Spectroscopy (AAS), Atomic
Fluorescence Spectroscopy (AFS), and Atomic Emission
Spectroscopy (AES) all depend upon the efficiency of
conversion of the sample into the atomic species of
interest (the analyte) and the ability to differentiate
optically the radiation absorbed or emitted by the analyte
(the analytical signal) from all other optical radiation
(background). In fluorescence, excitation is a result of
absorption of radiation from an external source, whereas
in emission excitation is primarily a result of col-
lisional transfer of thermal energy stored in vibrationally
excited gaseous molecules, for example, N2 in flames.
Flames have been in widespread use as atomization
cells because they are convenient, reliable (barring
flashbacks), and relatively inexpensive. Also, there is a
wide variety of flames and commercially available burners
which result in high signal-to-noise ratios (S/N) for many
elements in the parts per million (ppm) to parts per
billion (ppb) concentration range.
The efficiency of flames in converting analyte in the
sample solution to gaseous atoms is severely limited for
many reasons. To understand the inefficiency (for some
elements) of flames as atomizers, one must first have
examined the processes which lead to flame atomization, as
shown in Figure 1.
With premix burners, the efficiency of transferring
the liquid analyte to the flame as a dry aerosol is about
4-5 per cent . The total consumption burner has a larger
efficiency of transferring the analyte to the flame as dry
aerosol, but has a poorer solute vaporization efficiency .
Processes which lead to low atomic concentrations in the
flame atomization cell include analyte dilution due to the
large flow rates of the fuel and oxidant needed to support
flames and gas expansion upon conbustion. Reactions
occurring in the flame lead to the formation of many flame
gas radicals which may either combine with the analyte to
form stable compounds monoxidess of the analyte species
resulting from 0 radicals) or may emit radiation resulting
in appreciable background in spectral regions of interest
(such as OH, CN, etc.). In some of the higher temperature
flames, there occurs significant ionization of some elements,
such as Na, K, Cs, Ca, Sr, etc.
Ideally, atomization devices would have high
efficiencies of sample transport, low background due to
Fig. l.--Physical process leading to atomic absorption,
atomic fluorescence, and atomic emission
emission, and a gaseous environment which would lead to
efficient conversion of analyte vapor to analyte atoms.
L'vov has been a pioneer in the field of resistively heated
non-flame cell spectroscopy; he has used a graphite tube
furnace contained in an inert atmosphere for extensive AAS
studies [2-6). A complete review of his work is included
in his text . Massmann developed a resistively heated
graphite cuvette atomization cell suitable for AAS and AFS
[8,9]. The AAS system has been developed commercially by
Perkin-Elmer; the Perkin-Elmer non-flame cell AAS system
has been used to analyze many different elements in complex
matrices such as milk , sea water , oils , and
blood plasma . Kahn and Slavin  increased the
sensitivity of this system by stop-flowing the Ar sheath;
this system was then further improved by automation of
sample introduction by Pickford and Rossi , which im-
proved precision from 5-10 per cent to 1-2 per cent.
Samples have also been atomized from carbon fila-
ments into an inert carrier gas by means of resistive heating.
West and Williams constructed the first graphite filament
atomization device  for both AAS and AFS. The filament
was first enclosed within a Pyrex cell fitted with quartz
windows and purged with Ar. Later, the Pyrex cell was re-
moved and a laminar flow of Ar was used to provide an inert
atmosphere about the atomization filament. A number of
elements have been studied with this atomization device [17-
27]. Limited field viewing has led to reduced matrix
effects [21,22]. Applications have been reported for
oil-based solutions [263, Co in soils , Mo in organic
matrices , and V in titanium dioxide . Winefordner
et al. described a similar atomization device which was
evaluated extensively in AAS and AFS and was used for the
analysis of trace metals in both aqueous and oil-based
samples [50-54]. The latter non-flame cell was operated in
both an Ar and an Ar/H2 sheath. Studies were carried out
on the decay of atomic populations with height above the
graphite filament by AAS  and AFS . Temperature
programming of the filament was used to separate temporally
the signals of Ag and Cu in a single cycle ; temporal
resolution was applied to the analysis of Ag and Cu in jet
Amos [35,36] modified the West filament system by
drilling a lateral hole through the filament and creating
the "Mini Massmann" furnace; he compared the results of AFS
to AAS for several elements. Amos found AFS with the
graphite filament to result in better limits of detection
for many elements than did AAS with the .tube non-flame cell;
an Ar/H2 diffusion flame was used to shield the atomized
elements. The "Mini Massmann" furnace has become the com-
mercial Varian non-flame atomization device on which
application studies for elements in oils , Se in natural
waters , and Ag and Au in metallurgical samples 
have been reported. Bratzel and Chakrabarti  measured
the temperature of the atomic vapor produced in the Varian type
of non-flame cell and found that thermal equilibrium did
not exist, which is not surprising in a system containing so few
molecules to aid the attainment of equilibrium.
Brandenburger and Bader were the pioneers in using
resistively heated metal filaments as atomization devices
. Bratzel, Dagnall, and Winefordner employed a
resistively heated Pt filamert for use as an atomization source
for AFS with excellent results for some volatile elements .
Hwang et al.  used a resistively heated tantalum strip in
an enclosed inert atmosphere to analyze Pb in blood serum.
They also reported the determination of thirty-seven elements by
Few resistively heated non-flame cells with continuous-
sample introduction have been reported. Woodriff et al.
[44 46] employed either continuous pneumatic nebulization
of the cold aerosol into a resistively heated graphite tube
furnace or discrete sampling via sample placement with a
microliter syringe into a resistively heated graphite tube.
Applications using the discrete sampling mode were performed
on Ag in snow , Pb in air , and Pb in fish .
M. S. Black et al.  evaluated two types of platinum tube
furnaces with continuous pneumatic nebulization for AFS; no
applications of this system to real samples were described.
M. K. Murphy, S. A. Clyburn, and C. Veillon developed a
unique pyrolytic carbon tube furnace for use in AFS ;
this system resulted in sub-parts per billion (ppb) limits
of detection for some elements. The latter two non-flame
cells utilized a Veillon-Margoshes sample aerosol injection
Non-flame cells other than the resistively heated
non-flame cell types have been employed in various types of
atomic spectroscopy, but they will not be discussed here.
The interested reader may consult the excellent reviews of
non-flame cells by Winefordner , Kirkbright C54], and
Winefordner and Vickers  for a thorough discussion of
these, as well as the previously mentioned resistively
heated non-flame cell studies.
The purpose of this research project was to develop
an efficient, versatile non-flame atomization cell for use
in atomic fluorescence spectroscopy. This entailed the
development of a new high pressure, low sample consumption
pneumatic nebulizer which was coupled to a vitreous carbon
tube non-flame cell via a desolvation chamber into which
the analyte was introduced either in the continuous or
pulsed modes. The resulting system was evaluated as an
atomizer for atomic fluorescence spectrometry. Analytical
figures of merit are given for several elements, and the
possible use of the atomizer for emission spectroscopy is
Types of Atomic Fluorescence Transitions
Resonance fluorescence transitions involve identical
lower and upper levels in the excitation-de-excitation
processes. Direct line fluorescence involves only the same
upper level in the radiational excitation-de-excitation
processes. Stepwise line fluorescence involves different
upper levels in the radiational excitation-de-excitation
processes. If the fluorescence energy exceeds the excita-
tion energy, the fluorescence is termed an Anti-Stokes
process, and if the reverse is true, the fluorescence
transition is then called a Stokes process. A few types
of atomic fluorescence have been shown in Figure 2. For
a more extensive discussion of the types of fluorescence,
and where they have been observed, one should consult the
article by Omenetto and Winefordner .
Intensity of Atomic Fluorescence
The intensity of atomic fluorescence emitted from the
analyte depends principally upon: the intensity of the
exciting radiation; the concentration of the analyte in the
Fig. 2.--Types of atomic fluorescence.
(1) Resonance fluorescence.
(2) Anti-Stokes direct line fluorescence.
(3) Stokes direct line fluorescence.
(4) Stokes stepwise line fluorescence.
atom cell; the efficiency of the conversion of absorbed
radiation into the emitted radiation (quantum efficiency);
and the ratio of the excitation source half-width to the
absorption line half-width. If it is assumed that the
source excitation is imaged on the atomic vapor exiting
from the atomizer (atom cell), that the atom cell completely
collects all source excitation, and that the solid angle of
the monochromator is filled by the fluorescence of the atom
cell, then the most general equation for the integrated
fluorescence radiance (intensity) BF, for an isolated
spectral line is
BF = (-) II-1
-1 -2 -1
where BF is the fluorescence radiance (erg sec cm sr-1),
L is the fluorescence path length (cm), 1' is the atom cell
height (cm), As is the total atom cell surface area (cm ),
EA is the total irradiance per unit area absorbed by the
spectral line which results in fluorescence (erg sec-1cm2),
and Y' is the quantum efficiency (watts emitted by fluores-
cence to the watts of the source power absorbed per unit
The irradiance absorbed by the analyte in an atom
cell is given by
EA = J i BSi i II-2
= B~1 K2i
where J. is the solid angle over which the excitation
occurs, ,i is the summation over all i of (Bi Ki), BS is
the incident spectral radiance of the exciting radiation
-1 -2 -1
(erg sec cm sr ), and Ti is the total absorption factor
(sec-1) for different absorbing lines i. If, as in this
study, only resonance fluorescence is of interest, then
Equation II-2 may be simplified to
EA = LA BS KT
where BS and KT are values evaluated at the resonance
Winefordner et al. [57-59] have rigorously derived
the expressions for the intensity of atomic fluorescence
when using either a line or continuum source at low and
high optical densities in the atom cell. Figure 3 shows
schematically the atom cell, the source intensity direction,
and the direction in which fluorescence is measured.
Assuming the following conditions for fluorescence from an
1. The sample cell is 'completely illuminated.
2. The fluorescence radiation is collimated and at
right angles to the source radiation.
3. The atomic concentration is homogeneous throughout
the sample cell.
4. The temperature is homogeneous throughout the sample
5. Only resonance fluorescence contributes to the
Fig. 3.--Schematic diagram of the atom cell.
--4 = source radiation.
+ = fluorescence radiation.
1,1',L = for definitions, see text.
the intensity of fluorescence may be calculated. If a line
excitation source is used, the following equation holds for
the atomic fluorescence radiance, BF (erg sec1 cm-2sr-1) at
low analyte concentrations (no) [57,58] where the total
absorption factor, KT, has been evaluated
= f BY ( LI') II-(
BF = lu nolflu lu s (i- A
Klu = modified atomic absorption coefficient for pure
Doppler broadening (cm 2)
no = concentration of the atoms in the ground state
X1 = fraction of analyte atoms in the lower state
involved in the absorption transition
flu = absorption oscillator strength for transition
1 u, where 1 is the lower state and u is the
upper state dimensionlesss)
lu = factor to account for finite half-width of the
line source compared to the absorption line
S = solid angle of source radiation collected and
impinging upon the flame (sr)
1 = absorption path length (cm).
For high optical densities, the following equation is
derived for excitation with a line source [57,58]
BF = 2 BS 1' Y' (--) a
X KF n XFfF S
aF = damping constant for atomic fluorescence
KF = modified absorption coefficient (same as Klu),
but for reabsorption of fluorescence (cm2 )
XF = fraction of atoms in the lower state involved in
the reabsorption of fluorescence dimensionlesss)
f = absorption oscillator strength for reabsorption
of fluorescence dimensionlesss).
If a continuum source of radiation is used to excite
resonance fluorescence, then the atomic fluorescence
radiance, BF, for low analyte concentrations, no, is equal
BF = C1 lu no X1 lu AD BC (7T
lu 4 A
C1 = X (2 In J,) dimensionlesss)
AND = Doppler half-width of the absorption line (nm)
=lu = absorption line peak for transition 1 -- u (cm)
B k = spectral radiance for a continuum source (erg
lu-1 -2 -1 -1
sec- cm2 sr-l m- ).
At high analyte concentrations, no, the atomic fluorescence
radiance, BF, is equal to [57,58]
B = 2 BcklkD aF 1' Y' (1- )(C ) (l ) II-7
C2 = /T- (ln 2) dimensionlesss).
The experimental analytical growth curves for atomic
fluorescence with a line excitation source and a continuum
excitation source have been measured and the low and high
concentration asymptotic regions have been compared with
the above equations by Winefordner et al. [60,61].
Theoretical growth curves are shown in Figures 4 and 5.
It should be noted from these growth curves that atomic
fluorescence spectroscopy is most useful analytically for
low concentrations of analyte, no, in the atom cell because
the sensitivity (slope of the analytical curve) is greatest
in that region, approaching an asymptote of 1 for both
continuum and line source excitation of atomic fluorescence.
For high analyte concentrations, the fluorescence radiance
is independent of analyte concentration for continuum source
excitation and depends upon for a line source
. ~slope =1/2
slope = I
Fig. 4.--Atomic fluorescence growth curve with a line
source of excitation.
I. 10 102
LOG (Concentration of the Analyte)
// Large no
slope = 0
LOG (Concentration of the Analyte)
Fig. 5.--Atomic fluorescence growth curve with a
continuum source of excitation.
Fluorescence Quenching Considerations
The species in a resistively heated non-flame cell
should have a much lower quenching efficiency for the
excited-state atoms than does a combustion flame, mainly
due to the difference in the gaseous environments (that is,
an inert atomic gas in the non-flame cell (Ar, He, etc.)
versus molecular species in the flame (N2, CO, CO2, etc.)
. This should lead to more intense fluorescence from
a resistively heated non-flame cell than from a flame
(assuming equal atom concentrations and source
characteristics to be the case) by a factor of
Q = -rY 11-8
where the Y'F is the quantum efficiency of fluorescence
of the analyte in the non-flame cell and Y'F is the quantum
efficiency of fluorescence of the analyte in the flame.
This ratio has been expressed  as
KF + fi (KQi n )n- F
Q = i (KQ 11-9
where kF is the first-order rate constant for radiational
deactivation of the resonance level (sec-1), kQ is the
second-order quenching rate constant for deactivation of
the resonance level by collisions of element Z with a
quencher Qi (cm-3sec-l), and hQi is the concentration of the
quencher (cm-). The concentration of the quenching species
is at a minimum in the inert atmospheres of non-flame cells,
kF > i (ki ni)n-F II-10
This leads to 
k + i (ki n F II-
For many common analytical flames [62-64], and for a
typical inert atmosphere non-flame cell, the ratio Q is
ki ki n
which leads to the conclusion that the analyte should emit
fluorescence approximately ten times more intensely in the
inert atmosphere of a non-flame cell, compared to a flame
cell (keeping in mind the assumption of the same atom
concentrations in both cells and equivalent sources of
The major sources of noise in atomic fluorescence
flame spectroscopy are flame flicker noise and phototube
photonoise. With higher temperature, fuel-rich flames,
flame flicker noise may be several orders of magnitude
larger than shot noise, hence increasing the lowest concen-
trations which may be observed by atomic fluorescence flame
spectroscopy. The total noise, Aitot, for various
individual sources of random noise Ail, Ai2...., is
Aitot Ai2 + Ai22 + ... II-13
When using a properly baffled resistively heated
non-flame cell which prevents the black-body radiation
of the hot carbon from reaching the photomultiplier, the
atomic fluorescence measurement is limited by shot noise,
thus providing another advantage of resistively heated
non-flames as atomization devices over flames.
Atom Generation and Signal Measurement
Once the physical processes which constitute atomic
fluorescence and the parameters which govern the intensity
of fluorescence have been understood, one must decide how
atomic population is to be produced and how the analytical
signal should be measured. This was covered in great depth
for transient atom populations by L'vov . If No is the
number of atoms of the analyte consumed, N is the total
number of atoms of the analyte in the atom cell at time, t,
7 is the duration of transfer of atoms into the cell, 2
is the average residency time of analyte in the cell, and
is the time the signal is recorded for, then
S= nl(t) n2(t) II-14
where nl(t) is the number of atoms entering the cell at
time, t, and n2(t) is the number of atoms escaping from
the cell at time, t. For transient sample introduction
nl(t) = No/T1 II-15
n2(t) = (V) N 11-16
n(t) = II-17
where W/V was the ratio of the flow rate of the aspiration
gas (1 sec-1) to the volume of the atomization cell (1).
Thus, the equation relating the change of atomic
populations to change in time is
dN No N 1
It is recognized that for1 7 >> 7 the magnitude of N(t)
approaches a constant value or that 
S= 0 11-19
From Equation 11-19, it followed from Equation II-18 that
Ne = ()N II1-20
It has been widely recognized that electronic
circuits may distort the original pulse shape. From the
theory of electronic circuits [65,66], the relationship
between the distorted peak shape i(t) to the original peak
shape, N(t), and the transient characteristics of the
network A(t) is given by
i(t) = N(O) x A(t) + J A(t-7)N'(rT-7) II-21
where N'(t) is an integration variable, N'(t) is a deriva-
tive function of N(t) where t has been replaced by the
variable, 7", and N(O) is the value of the function N(t)
at time t=0. The time constant, 7', is defined for a
simple RC circuit by
7 = RC II-22
where 7c is the length of time during which the output
signal is within 36.8 (that is, 100 x e-1) per cent of the
input signal and R is the resistance (ohms), and C is the
capacitance (Farads). The transient characteristic of the
circuit passing from unity voltage-to-0 is
A(t) = exp (- 1-) 11-23
and the 0-to-unity voltage transient is
At = 1 exp (- ) II-23a
To obtain the highest value of signal-to-noise,
L'vov  experimentally showed that the time constant of
the measurement circuit, 7-, should be approximately equal
to the value of 72, the average residency time of the
analyte in the cell for a transient signal. He also
stressed that the analyst may increase precision of his
analytical measurements by using integration techniques
for signal measurement rather than just peak heights.
For continuous-sample introduction, the magnitude of
N(t) approaches a constant value, or
) = 0 II-24
Thus, when measuring the equilibrium signal from a con-
tinuous-sampling atomization cell, the time constant, 7 ,
of the recording device may be adjusted to a larger value
than when measuring a transient signal; the larger 7c
reduces the measured noise through time averaging and also
increases the analysis time, and hence the amount of
analyte consumed. At least five time constants should
elapse before performing any analytical signal measure-
ments, to allow the signal to reach greater than 95 per
cent of its maximum value.
Concentration of Analyte in the Atom Cell
For the pulsed mode of sample introduction, it will
be assumed that the analyte is exponentially diluted.
Therefore, the concentration of analyte atoms in the atom
cell, nt, at any time, t, is 
nt = no e-( t/)
t = time (sec)
V = volume of the exponential dilution flask (cm3)
u = flow rate of aspirating gas (cm sec- )
nt= concentration of analyte at time, t, in the atom
cell (cm- )
no= initial concentration at time t = 0 of analyte
in the atom cell (cm-3)
The initial concentration, no, in the atom cell can
be estimated if it is assumed that the entire pulse of
analyte atoms enters the exponential dilution flask and
reaches a homogeneous concentration before any analyte exits
and so [53J
v c NA T -26
n = e- II-26
v = volume of sample solution pulse (cm )
V = volume of exponential dilution flask (cm3)
c = concentration of analyte in sample solution prior
to aspiration (mol cm 3)
NA = Avogadro's number (mol -1)
6 = efficiency of aspiration dimensionlesss)
p = free atom fraction dimensionlesss)
eF = gas expansion factor dimensionlesss)
ZT = ratio of atoms in ground state to total number in
all states dimensionlesss)
For the continuous-sampling flame (or non-flame cell)
system, the concentration, n, of the analyte atoms in the
atom cell (atoms cm -) is related to the concentration, c,
of the analyte (mol 1-1) aspirated into the flame by 
n = 1 x O19 F E )Z II-27
F = sample flow rate (cm min-1)
E = same as defined above
9 = same as defined above
eF = same as defined above
Q = flow rate of the gas entering atom cell (cm3 sec-1
ZT = the ratio of atoms in the ground state to the total
number of atoms in the atom cell dimensionlesss).
The aspiration efficiency, E, is the ratio of the
number of atoms of analyte entering the nebulizer to the
total number of gaseous analyte atoms contained in the
atom cell. This factor corrects for losses of analyte
solution in the nebulizer and associated tubing, and the
incomplete volatilization of the solid particles in the
The free atom fraction,P, is the ratio of the number
of analyte atoms in any form to analyte atoms in all gaseous
forms in the atom cell. This factor corrects for incomplete
dissociation of the analyte compound, combination of analyte
atoms with flame gas radicals, condensation of the analyte
atoms to form the original compounds, and ionization of
the analyte atoms.
The factor ZT accounts for the proportion of free
atoms contained in the atom cell which are not in their
electronic ground state and may be evaluated with the
Boltzmann equation 
n g e s/kT II-28
= e 11-28
n = number of atoms in the ground state dimensionlesss)
ns = number of the atoms in a level with an excitation
energy of Es dimensionlesss)
gs ,g = weighting factors dimensionlesss)
k = Boltzmann's constant (8.614 x 105eV K-)
T = atom cell temperature (OK)
Es = excitation energy of levels (eV)
The total atom concentration will be, nT,
nT = no + n1 + n2 + ... II-29
and so from Equation II-29 and the Boltzmann equation
no [ -E/kT -2/kT
n = io + e + g2e + .'"
= o (B) 11-31
B = g leE/kT + g2e2/kT + ... II-32
and finally ZT is
ZT = 11-33
Limit of Detection Considerations
A very useful analytical figure of merit for use in
trace element analysis is the limit of detection (LOD),
which is defined statistically so that comparisons may be
carried out between different methods of analysis. The
statistical theory used here was developed by Kaiser [69-71].
The limit of detection, C, is defined as
C = mS II-34
where m is the slope of the analytical curve m =
d(log S) (S is the signal resulting from a concentration
of analyte, C) and is assumed to be constant from the LOD
to 1000 x LOD. The smallest analyte signal measurable
S= Ss+b 11-35
where Ss+b is the average limiting signal due to the analyte
plus blank and S, is the average blank signal. Generally,
S is further defined in terms of the fluctuations of the
signal. The standard deviation of these fluctuations is
s =j b2 +b2 II-36
where b is the standard deviation of the blank signal,
s~+b is the standard deviation of the signal plus blank
signal, nb is the number of measurements performed on the
blank signal, and ns is the number of measurements per-
formed on the analyte signal (plus blank). At the limit
of detection, the 6" will approximate i+b. The limiting
detectable signal is then
s = Z -1im II-37
where Z is the z-statistics of the normal distribution,
that is, a protection factor. If the number of observa-
tions are large (about 20), then Z may be replaced by the
Student t and 6 by s which is the estimated standard
deviation of the signal near the limit of detection. If
ns = 1 and nb 15, then Equation II-56 may be reduced to
C = m tsb 11-38
where sb is the estimated standard deviation of the blank
and if sb =ss. However, if ns = nb, then Equation II-36
may be reduced to
C= mt sb V2 11-39
If ns = nb = 1, then Equation II-39 reduces to
C =2 mt sb II-40
In the majority of previous studies , the limit
of detection is defined as that concentration or amount
which will give a signal-to-noise ratio of 2 or 3. If the
only major random errors are electrical and optical noises,
then the rms noise is defined as approximately peak-to-peak
noise divided by 5 ; thus, the limit of detection, C,
may be evaluated from Equation II-39, where nb = 5 and
t = 3, which gives a confidence level of better than 95
Stock solutions were prepared from analytical
reagent grade chemicals of Pb(NO3)2, Na2Te0O42H20, AgNO3,
T12SO4, and K SbOC4HO06.- H20 dissolved in deionized water
to give concentrations of 103 jgml-1. Analytical reagent
grade chemicals of SnO, MgO, and ZnO were dissolved in HC1
and Zn metal, Co metal, and CdO were dissolved in HNO3 and
then diluted to 103 ugm11. With concentrations varying
by about a factor of 3, serial dilutions were performed
with deionized water for concentrations ranging from
3x102 jgml-1 to the concentration resulting in a signal of
2 times peak-to-peak noise.
General Layout of the Exoerimental System
A schematic diagram of the experimental components
in the atomic fluorescence system is shown in Figure 6.
The components of this system are discussed below.
Sources of Excitation
Single-element electrodeless discharge lamps (EDL's)
were prepared from the iodide form of the elements of Co,
Fig. 6.--Schematic diagram of continuous- and pulsed-sample
introduction tube furnace atomic fluorescence
E Electrodeless discharge lamp (EDL).
B Microwave power supply for EDL.
L1, L2, L3 Lenses.
S1, S2 Apertures.
F Vitreous carbon tube furnace.
P Photomultiplier tube.
V High voltage power supply.
L Lock-in amplifier.
R Recorder readout.
I L C L
Sn, Sb, Te, Pb, and Bi. The elements Ag, TI, and Mg were
prepared in their chloride form, while Zn and Cd were pre-
pared in pure form as single-element EDL's. These EDL's
were operated in the thermostated mode described by R. F.
Browner et al. [74,753. The apparatus used for thermo-
stating the EDL's is shown in Figure 7. It consisted of a
fan (Model NTH2, Rotron Mfg. Co., Woodstock, N.Y.) which
blew room air over a heater coil (Model HG 751, Master
Appliance Corp., Racine, Wis.) whose temperature was con-
trolled with a laboratory Variac (Model 2 PF 1010, Staco,
Inc., Dayton, Ohio). The temperature was monitored with a
chromel-alumel thermocouple placed as shown in the metal
conduit with temperature (OC) indicated directly on a volt-
meter calibrated in temperature units (Model K, Barber-Cole-
man Co., Rockford, Ill.). The thermocouple was calibrated
up to 350*C with a common laboratory thermometer, and
because the temperature was in error by only 5 OC at 350C
and the error increased linearly with temperature, the error
was extrapolated up to the maximum temperature used and
then was corrected for. A change in 20 OC had very little
effect (less than 5 per cent) upon the intensity of the
EDL's when the EDL temperature was optimized.
The EDL was situated, as shown in Figure 7, in a
microwave field focused with an "A" antenna. The EDL was
powered by a microwave power generator (Model PGM-10,
Fig. 7.--Cutaway view of the EDL thermostating unit.
All brass construction.
Quartz Jacket k- Metal Conduit
Discharge Lamp ----I Heater
Mount ----- -,-
Raytheon Co,, Manchester, N. H.). The thermostated unit
shown in Figure 7 from the bottom of the "A" antenna up was
enclosed in an aluminum box measuring 4.5 in. wide, 7.0 in.
long, and 6.0 in. high, with a hinged lid to facilitate
changing EDL's. A 0.25 in. diameter hole was placed in a
convenient location for the thermocouple to enter and a
1 in. diameter hole was located in such a position that the
EDL radiation could be easily focused through the lenses
upon the atom cell. When operating this unit at high
temperatures (greater than 300 OC) (Model SP2A2, Rotron
Mfg. Co.), a fan was needed to cool the microwave cable at
the "A" antenna junction. The whole unit was mounted on
two micrometer screw threads which gave both vertical and
lateral adjustment to better than 0.1 mm to facilitate
focusing the desired portion of the EDL upon the atom cell.
The single-element EDL's for these studies were
operated at the temperatures and powers for the continuous
mode studies and pulsed mode studies shown in Tables 3 and
5, respectively. The optimum temperatures and powers were
experimentally determined each time the EDL was started by
i. lighting the EDL at a power of 50 W;
ii. adjusting the monochromator to the approximate
wavelength of interest using light reflected
from the EDL to the photomultiplier tube;
iii. while monitoring the intensity on a strip chart
recorder, slowly raising the temperature of the
EDL until the EDL intensity started to decrease
from the maximum;
iv. decreasing the temperature until the maximum EDL
intensity was attained;
v. fine tuning the wavelength;
vi. adjusting the microwave power to a reasonable
(an increase in EDL intensity of 2-fold per 10 W
power) trade off between high power and EDL
intensity (high microwave power will decrease
the EDL lifetime).
The EDL's for all but two elements were operated
satisfactorily in this manner. However, the Sn and Mg
EDL's needed a microwave power of 90 to 100 W (with
temperature programming) to remain lit, and the optimum
intensities were very sensitive to small changes in micro-
wave power. This latter point was especially true for Mg;
with moderate microwave power (50 W) the Mg EDL discharge
was contained in only a small portion of the EDL volume.
Even at 100 W, the discharge still did not fill the EDL
cavity. In addition, several of the EDL's could be seen to
fluctuate in intensity by the naked eye (through UV ab-
sorbing safety glasses, of course). An increase of micro-
wave power usually eliminated this problem.
Optical grade biconvex quartz lenses with a focal
length of 2.5 in. and a diameter of 2 in. (Esco Products,
Oak Ridge, N. J.) were used throughout for focusing. They
were mounted on a steel plate with the aid of magnetic
mounts (Model M 8-1, Enco Manufacturing Co., Chicago, Ill.)
and were aligned both vertically and horizontally with a
He-Ne lab laser (Model ML 6805, Metrologic, Bellmaur,
N. J.). For all absorption measurements performed, all
three lenses were aligned to 1800 with the entrance optics
of the monochromator in such a manner that a 1:1 image was
created at all focal points as checked with a tungsten
filament lamp, and in such a way that the monochromator's
mirrors and the diffraction grating were totally illuminated.
For AFS measurements, the two lenses farthest from the
monochromator were aligned vertically and horizontally so
that the total solid angle of source intensity which
impinged on the atom cell could be converted to useful
fluorescence, as shown in Figure 6. Aperture S1 of Figure
6 was 0.5 cm by 1 cm, so that all useful fluorescence from
the atom cell would totally fill the slit width and height
of the monochromator, and as mentioned previously, its
mirrors and gratings. Aperture S2 of Figure 6, was
adjusted so that the collected fluorescence would just fill
the solid angle of the monochromator and also fill all of
its related optics. All measurements were performed with
a 0.35 m f/6.8 Czerny-Turner monochromator (Model EU-700/E,
Heath Co., Benton Harbor, Mich.).
The electronic equipment used in these studies is
also shown in Figure 6. The associated electronics included
an RCA 1P28 photomultiplier powered by a high voltage power
supply (Model EU-42A, Heath Co.), a lock-in amplifier tuned
to 667 Hz (Model 391, Ithaco, Inc., Ithaca, N. Y.), and a
preamplifier with variable gain (Model 164, Ithaco, Inc.).
Source modulation was performed with a mechanical chopper
(Model 382, Ithaco, Inc.), and the lock-in output was re-
corded on a potentiometric recorder (Model S. R., Sargent
Welch Scientific Co., Skokie, Ill.).
Different means of creating a wet aerosol mist from
a solution have been used for atomic spectroscopy. Spin-
ning disks, high voltage sparks, ultrasonic radiation, and
pneumatic devices have all been employed to create these
mists; however, only the latter two have received widespread
attention in atomic spectroscopy. Of these, pneumatic
nebulizers have been the commercially accepted choice be-
cause they have been inexpensive, sturdy, and reliable.
However, they have suffered from the problems of creating a
dispersed mist with widely varying particle diameters, and,
when coupled to a spray chamber, poor aspiration efficiency;
furthermore, they needed high gas flow rates to aspirate
properly, and were reported to have a maximum aerosol
density of 0.03 g of solution/liter of aspiration gas [1,76].
Ultrasonic nebulizers have beenplagued with various
problems such as memory effects and sample transport capil-
lary failure, and so were not used in these studies [76,77,
78]. Such nebulizers, however, function on very low flow
rates of gas and hence result in relatively large solution-
to-gas flow ratios. For further information on nebulization
techniques, the reader is referred to the atomic spectroscopy
review by Winefordner and Vickers .
In the present work, it was desired to build a high
pressure, low sample consumption nebulizer using low flow
rates of aspiration gas, although some difficulties were
reported by other workers [1,79]. Such a nebulizer seemed
to be ideally suited to non-flame AAS and AFS work. First,
a miniaturized standard conical design similar to the one
used in pneumatic nebulizers was used with limited success
because the sample transport needle (id 0.005 in.) repeatedly
plugged up. The original design was then slightly altered
to that of nebulizer I, shown in Figure 8. The sample
transport needle (I) used was a 6-in.-long 26 g stainless
steel hypodermic needle (Hamilton Co., Reno, Nev.). This
needle was strengthened by placing a concentric stainless
steel tube around the exterior of the capillary needle, 0.5
in. from the tip until an outside diameter of 0.052 in. was
Fig. 8.--Schematic diagram of nebulizer I.
A Nebulizer cone.
B Rubber 0-ring.
C 0-Ring seating gland.
D Alignment rod.
E Micrometer adjustment.
F Housing ring to hold micrometer.
G Stainless steel needle housing.
H Teflon centering sleeve.
I Hypodermic needle.
J Vertical adjustment plate.
K Nebulizer housing.
L Needle housing.
M Luer lock-luer lock connector.
N Teflon system seal.
0 Retaining screw for N.
.K Stainless Steel
attained. This needle was held firmly in place in the
stainless steel needle housing (F) with the use of a slip-fit
Teflon centering sleeve (H). The transport needle was
epoxied in place to the retaining screw (0) for the Teflon
system seal (N). The nebulizer cone was machined to have a
0.020 in. inside diameter and a convergence angle of the
aspirating gas nozzle to the central solution capillary of
300. This left a clearance of 0.001 in. between the solu-
tion capillary and the aspirating gas nozzle when the former
was centered. The centering of the sample transport needle
was effected with three stainless steel micrometers which
were situated at 1200 about the needle with the aid of a 30
power microscope (Model 70, 266, Edmund Scientific Co.,
Barrington, N. J.). In the continuous-sampling studies,
sample solution was force-fed through the sample transport
needle with the aid of a syringe pump (Model 353, Sage
Instruments, Inc., White Plains, N. Y.).
Nebulizer II was designed for use in both the con-
tinuous- and the pulsed-sampling modes. Nebulizer I con-
tained a dead volume of greater than 0.120 ml between the
syringe and the exit of the sample transport needle, which
caused very severe memory effects when sampling in the
pulsed mode. The dead volume was reduced by shortening the
sample transport needle to 2 in. in length, eliminating the
Luer-lock Luer-lock connection by epoxying and crimping a
male unit to the Teflon tubing, and decreasing the length of
the Teflon tubing from 12 in. to 6 in. The former change
necessitated the complete redesign of the nebulizer housing
and sample transport needle centering apparatus; however,
critical aspiration parameters were not altered, such as
the 300 convergence angle of the aspirating gas nozzle to
the central solution capillary and the clearance between
the solution capillary and aspirating gas nozzle of 0.001
in. Nebulizer II is shown in Figure 9. The overall
dimensions of the nebulizer were 1.375 in. in diameter and
2 in. in length. 10-80 Stainless steel screws (D) were
situated at 1200 around the hypodermic needle to allow
centering of the hypodermic needle (H) in the aspirating
gas nozzle (A).
All other parts, as mentioned previously, functioned
similarly to the parts of nebulizer I, shown in Figure 8,
and will not be discussed further.
A desolvation chamber, heated with heating tape
(Samox Hi Temperature Tape, Arthur H. Thomas Co., Phila-
delphia, Pa.), connected the nebulizer with the vitreous
carbon non-flame cell. Desolvation chambers constructed of
quartz and glass were found unsatisfactory, due to inability
to form a simple, efficient, gas-tight seal between the
Fig. 9.--Schematic diagram of nebulizer II.
A Nebulizer cone.
B Rubber O-ring.
C Rubber 0-ring retaining screw for B.
D Alignment screw.
E Nebulizer housing.
F Stainless steel needle housing.
G Teflon centering sleeve.
H Hypodermic needle.
I Teflon system seal.
J Retaining plate for I.
M Stainless Steel
D Rubber/Teflon Septum
chamber and the nebulizer. Without a good seal, repetitive
pulsed atomic fluorescence signals were obtained while
sampling in the continuous mode. Thus, the desolvation
chamber shown in Figure 10 was machined from type 304
stainless steel. It was bolted to the top of the nebulizer
with sheet asbestos serving as a gasket and, to some extent,
as a thermal insulation for the nebulizer. The desolvation
chamber had an inside diameter of 1.375 in. and was 3.5 in.
long. The top inside of the desolvation chamber was re-
duced to a radius of 0.688 in. The male unit of the de-
solvation chamber was tapered from 0.297 in. outside
diameter at the top to 0.313 in. outside diameter at the
base to facilitate forming a leak-proof seal to the vitreous
carbon tube furnace. Around the inside rim at the base of
the desolvation chamber were entrance ports to add other
gases than were used to effect nebulization.
Vitreous Carbon Tube Furnace
The vitreous carbon tube furnace is shown in Figure
11. Its outside dimensions are 2.5 in. wide by 2.5 in.
long and 2.0 in. in height. All parts ultimately connect
to the water-cooled main brass furnace housing (F). Both
top and bottom water-cooled electrodes (B) contained a
copper slide (A) to facilitate electrical contact. The
top water-cooled copper electrode (B) had three concentric
CHI4 v/C CH4
Fig. 10.--Schematic diagram of the stainless steel
Fig. 1l.--Schematic diagram of vitreous carbon tube
A Copper electrical contacts.
B Water-cooled electrodes.
C Vitreous carbon tube.
D Boron nitride insulation ring.
E Asbestos insulation ring.
F Furnace housing.
G Graphite rings for electrical contact.
Z Boron Nitride
circular arrays of holes (used for inert sheath gases) of
1 mm in diameter at radii from the center of the vitreous
carbon tube of 13 mm, 17 mm, and 19 mm, respectively.
These holes were staggered with each other to minimize 02
and N2 entrainment from the air. The two main electrodes
(B) were electrically insulated from the brass furnace
housing (F) with fiberglass gaskets and nylon screws (the
external surface of the furnace housing was less than 75 C
when the vitreous carbon tube (C) was operating at tempera-
tures in excess of 2000 C). Thermal insulation was pro-
vided with rings of boron nitride (D) and asbestos (E)
between the vitreous carbon tube furnace (C) and the
furnace housing (F).
Perhaps the most critical tolerances in this furnace
are the inside diameters and outside diameters of the upper
and lower graphite rings (G), providing electrical contact,
as well as the thickness of the upper graphite ring (G).
If the rings(G) were not machined to have a snug slip fit
to the vitreous carbon tube (C), the electrical resistance
became so great that the maximum voltage (20 V) from the
power supply (SCR 20-250, Electronics Measurements, Neptune,
N. J.) could not supply enough current to sufficiently heat
the carbon tube. The thickness of the upper graphite ring
(G) determined the resistance of the contact between it and
the vitreous carbon tube (C), and hence the power dissipated
near the atom cell. (It thus was used to influence the
temperature of not only the vitreous carbon furnace, but
also the atom cell.) The lower graphite ring (G) had an
orifice of 0.297 in. in diameter to accept the previously
mentioned male unit of the desolvation chamber.
Nebulizer, Desolvation Chamber, and Vitreous Carbon
Tube Furnace Mount
The nebulizer and vitreous carbon tube furnace were
mounted on a 0.5 in. thick, by 2.0 in. wide by 12.5 in.
long piece of aluminum in such a manner that the former,
with the desolvation chamber in place, could easily be
moved up or down 5 in. and locked in place while the
vitreous carbon tube furnace was kept fixed. This whole
assembly was mounted on two micrometer screw threads which
gave both vertical and horizontal adjustment to better
than 0.1 mm to facilitate imaging the atom cell upon the
monochromator entrance slit.
Procedure for Evaluation of Experimental Characteristics
of the System
Gas flow rates
The rotameters were calibrated for various gas flow
rates at different pressures. These calibrations are given
in Appendix II.
Sample solution flow rates
Flow rates, using the syringe pump and 1 ml disposable
tuberculin syringes (Model 501 S-TB, Sherwood Medical
Industries, Inc., Deland, Fla.), to pump the sample through
the nebulizer, were calibrated versus the coarse flow rate
scale and the percentage flow rate scale of the syringe
pump. The aerosol produced was of a pulsed nature when
samples with very low flow rates were aspirated with an Ar
flow rate of 0.90 1 min1. These results are summarized
in Table 1. A flow rate of 0.12 ml min-1 was chosen as a
nominal sample flow rate for all measurements.
Mean particle diameter
The mean particle diameter of the wet aerosol formed
upon aspiration may be estimated from the empirical equation
which Nukiyama and Tanasawa determined for velocities up to
sonic velocities for a conical pneumatic nebulizer .
Their equation for the mean particle diameter, d, in un
S 585 () + 57 45 1000 V 1.5
OF AEROSOL PRODUCTION WITH HIGH
Sample Flow Rate
Ar Flow Rate
aMeasured stability by monitoring light scattered from
aerosol droplets emerging from nebulizer.
After a time delay of 10 sec.
CNo = stability of aspiration was inadequate for
} = velocity of aspiration gas
V = surface tension
C = density of liquid
V1 = volume of liquid (1)
Va = volume of aspiration gas (1)
7 = viscosity
It should be noted that if Va = 5000 V1, then for water,
the Nukiyama and Tanasawa equation (Equation III-1) reduces
c ^ (G^ III-2
The velocity of the aspiration gas through the nebulization
cone was estimated by dividing the measured flow rate of the
aspiration gas by the area of the ring-shaped orifice. The
value calculated in this manner was 390 m sec- for an
aspiration flow rate of 0.90 1 min-1. Substituting the
speed of sound, 340 m sec- for y in Equation III-2, a
mean particle diameter of 15 .im was calculated. The calcu-
lated velocity of approximately 10 per cent greater than the
speed of sound would further increase turbulence and hence,
tend to decrease the particle size even more.
The aerosol mist was dispersed and desolvated in the
stainless steel desolvation chamber previously described.
The temperature of the gas passing through the center of
the desolvation chamber ranged from 246 C to 278 OC as
shown in Figure 12, while the heat tape wrapped around the
chamber was measured at 393 C. These temperatures were
measured with 0.015 in. diameter chromel-alumel thermo-
couples, and the resulting voltages were monitored on a
digital multimeter (Model 414, Dynasciences, Chatsworth,
Desolvation time, t, of a particle may be approxi-
mated with the following equation .
AC / tn
ts lnL + Cp (T-Tb)/L] II1-3
t = desolvation time (sec)
= density of the liquid (g cm-)
A = initial area of the droplet (cm )
Cp = specific heat of the vapor (cal g- C )
X = thermal conductivity of the gas (cal sec-1
T = desolvation temperature (C)
Tb = boiling point of the solvent (C)
L = specific heat of vaporization of the solvent
This equation neglects the additional transport of heat to
the boiling droplet by convection, which would significantly
reduce the actual desolvation time. Using Equation 1-3,
a desolvation time of 23 msec for a 15 am diameter sphere
Fig. 12.--Gas temperatures in the desolvation chamber
at various heights above the nebulizer.
0 2 4 6 8 10
of water at 250 C was calculated. Also, from the known
nebulization gas flow rate and the volume of the desolva-
tion chamber, a residency time of the sample mist in the
desolvation chamber of 5.3 sec was calculated, neglecting
the additional gaseous volume of the-liquid sample solution.
Thus, desolvation was probably completed in the desolvation
No scattered radiation (in the absorption mode) was
discernible from water or from a 100 pg ml-1 Zr solution
at the Zn 213.8 nm line which was radiated from a Zn hollow
cathode while absorption measurements were being made at
grazing incidence to the desolvation chamber. The scatter
signal was found to be 2 per cent absorption for scatter
measurements performed with 104 ug ml-1 Zr . The
negligible amount of scatter from large concentrations of
solutes, which are difficult to vaporize, qualitatively
supports the above observations concerning the small mean
Comparisons of the rise velocities in the vitreous
carbon tube non-flame to that in various analytical premixed
flames (assuming a primary combustion zone angle of 6) are
given in Table 2 L1]. Ratios of the residency time in the
COMPARISON OF RESIDENCY TIMES IN ATOM CELLS
Rise Velocitya (m sec-1 )
Residency Time Ratiosb
aRise velocities for the flames were calculated from
Vburn = rrise sin 8, where burn is the burning velocity
(taken from Reference 1).
bResidency time was calculated from tR =
r--- where ho for the flames is the observation height
above the burner taken to be 2.5 cm and h for the furnace
was taken to be the tube length of 5 cm.
vitreous carbon tube to the residency time in a flame prior
to measurement (assuming a distance of 2.5 cm) are also
given in Table 2.
Efficiency of sample transport
The efficiency of the nebulizer and desolvation
chamber was experimentally determined by aspirating 2 ml
of 12.5 ug m1-1 Mg. The chamber was then washed with de-
ionized water of the same pH as the original solution, and
the Mg concentration of this solution was measured via
atomic absorption spectrometry with the graphite filament
non-flame cell . The efficiency was found to be 93 per
cent. The chamber was then rewashed in the same manner to
check for any residual Mg, and no detectable Mg was found.
Temperature of furnace and gaseous vapor
The temperatures in and above the vitreous carbon
tube resulting from various applied currents from a constant
current source were measured with 0.010 in. diameter
tungsten-tungsten 26 per cent rhenium thermocouples (Omega
Engineering, Inc., Stamford, Conn.). The voltages were
monitored on a digital multimeter (Model 414, Dynasciences).
The results are given in Figure 13.
Fig. 13.--Temperatures of non-flame cell at various points.
a Temperature in center of carbon tube.
b Temperature at grazing incidence to carbon tube
with Ar sheath.
c Temperature at grazing incidence to carbon tube
with Ar-H2 diffusion flame.
d Temperature at 1 cm above carbon tube with
Ar-H2 diffusion flame.
e Temperature at 2 cm above carbon tube with
Ar-H2 diffusion flame.
f Temperature at 1 cm above carbon tube with
g Temperature at 2 cm above carbon tube with
w 800 d
0 50 100 150 200
General Procedure for Making Atomic Fluorescence Measurements
All electronic equipment (photomultiplier tube,
lock-in amplifier, chopper, etc.) was switched on at least
30 minutes before any experimental data were recorded. The
desolvation chamber required approximately 30 minutes to
reach the steady state in temperature shown in Figure 12,
while the EDL's usually required about 40 minutes to
The nebulizer, desolvation chamber, and vitreous
carbon tube furnace were flushed with Ar for approximately
5 to 10 minutes before any atomic fluorescence measurements
were performed. Increasing amounts of current were applied
to the carbon tube furnace until the desired temperature
was reached after the desired flow rates of Ar, H2, and CH4
were adjusted. The sample (or blank) was then aspirated
into the desolvation chamber in either the pulsed or con-
tinuous modes of sampling, and the resulting aerosol mist
was carried to the vitreous carbon tube where atomization
was effected. The vitreous carbon furnace was operated at
a constant current (temperature) until all atomic fluores-
cence measurements were completed.
The continuous mode of sampling was achieved by
force-feeding the solution to the high pressure nebulizer
with a syringe pump (Model 353, Sage Instruments), while
the pulsed mode was effected by forcing the sample into the
nebulizer operated continuously at 60 pounds pressure of
Ar from a 0.5 ml syringe (Model 1000, Hamilton Co.) con-
tained in a repeating dispenser (Model PB-600, Hamilton
The excitation radiation passed at grazing incidence
to the top of the vitreous carbon tube. The fluorescence
was collected by the monochromator adjusted to the desired
wavelength. The slit width was adjusted to 750 um for all
studies. The Ar, CH4, and H2 flow rates (see Appendix II)
were monitored by rotameters calibrated with a wet test
meter (Model S-39447, Precision Scientific Co., Chicago,
RESULTS AND DISCUSSION
Atomic Fluorescence Signal-to-Noise Optimization
Five 1P28 photomultiplier tubes were examined for
both dark current and relative gain characteristics at
various applied voltages (600 V 900 V); the gain was
determined with a tungsten filament lamp as the source
operated at a constant voltage (intensity), a wavelength
of 500 nm, and a slit width of 200 nm. The optimum photo-
multiplier tube (of the five measured) was used for all
fluorescence measurements; at the voltage (700 V) found to
give the optimum signal-to-noise ratio, a dark current of
1 x 10-10 A at 700 V was measured.
As mentioned in Chapter III, the intersection of the
center of the excitation optics and the center of the
emission optics (L1 and L2, L3 and M, respectively of
Figure 6) was determined; this intersection could be located
to better than 1 mm using a removable marker. The vitreous
carbon tube furnace alignment was coarsely adjusted in this
manner, and then, using the vertical and horizontal screw
adjusts, the fluorescence signal was finely tuned to a
Because ideally a fluorescence signal should be
measured against no background signal, two apertures (S1
and S2 of Figure 6) were used to prevent most source
radiation from being scattered about the room and eventually
finding its way to the photomultiplier tube. Care was
taken to ensure that the aperture did not limit the solid
angle of the monochromator.
No mirrors were used to increase fluorescence
intensity by reflecting the source intensity back upon the
atom cell or in collecting a larger solid angle of the
fluorescence radiation from the atom cell. An improvement
of signal-to-noise (S/N) by a factor of 3-10 by employing
mirrors for this purpose has been reported in atomic
fluorescence flame spectroscopy .
The optimum time constant of the lock-in amplifier
(resulting in optimum fluorescence S/N) used in the pulsed
mode was experimentally determined for several elements and
was found to be 0.4 sec.
Fluorescence signal-to-noise for a number of elements
was determined at various monochromator slit widths, and a
750 um slit appeared to be a suitable compromise choice for
On the average, about five EDL's for each element
determined were characterized as to their optimum tempera-
ture range, operating power, long term lamp stability,
relative intensities at the wavelength of interest, and
short term noise at the optimum temperature (intensity).
These characteristics were determined as described
previously in Chapter III. All results were recorded, and
the respective lamps were labeled. The figures of merit
for each EDL, for example, optimum temperature range, in-
tensity, operating power, long term stability, and short
term stability were checked for the particular lamp in use
before any atomic fluorescence measurements were performed
to determine if the lamp was still near optimum
characteristics. If not, another EDL was used.
The atomic fluorescence signal-to-noise of Pb and Sn
was examined for vitreous carbon tubes with internal
diameters of 0.188 in. and 0.125 in. The S/N for the
0.188 in. tube was larger by approximately a factor of 2
for Pb and Sn than for the 0.125 in. tube. Therefore, it
was used for all further atomic fluorescence measurements.
The upper graphite ring for electrical connection (G of
Figure 11) was changed in thickness to values of 0.050 in.,
0.060 in., 0.080 in., and 0.100 in., and the resulting
atomic fluorescence intensities were measured for Te, Pb,
and Sn. A thickness of 0.080 in. gave the optimum S/N
ratios when the vitreous carbon tube lifetime was taken
into account, so this thickness was used for all further
atomic fluorescence measurements.
The optimum temperatures of the vitreous carbon tube
furnace were determined experimentally for each element by
comparing fluorescence S/N ratios. The vitreous carbon
tube lifetime was also taken into consideration here;
higher temperatures led to shorter lifetimes of the vitreous
The gas flow rates (nebulization Ar flow rate,
external Ar flow rate, external H2 flow rates, and internal
CH4 flow rates) were optimized for each respective element
in the continuous mode study; however, in the pulsed mode
study, average optimum flow rates for all elements were
used. Internal Ar was also added for Ag, Pb, Sn, and Tl
in the continuous mode of operation; however, in all cases,
the fluorescence S/N ratio decreased markedly, and so it
was not further examined. This may be the result of in-
creased atom cloud dilution in the atom cell or decreased
residency times of the analyte in the vitreous carbon tube
Definitions of Analytical Parameters
The atomic fluorescence concentration (,jg m-1)
limit of detection (CLOD) was defined as that concentration
(in ug ml-1) giving a signal of 3 x rms noise. Each LOD
was found by alternately running a signal and blank 3-5
times and then extrapolating back from a signal of about
2 x peak-to-peak noise to a signal of 3 x rms noise and
determining the concentration of the lowest signal level.
Atomic fluorescence measurements of the various elements
were determined for concentrations ranging from 103 pg ml-1
to the concentration resulting in a signal of 2 x peak-to-
peak noise with concentrations varying by factors of three.
The upper concentration limit (called UCL) was determined
by the beginning of serious curvature (greater than 4 per
cent deviation from linearity) of the analytical curve.
The linear dynamic range (LDR) was the ratio of UCL/LOD,
and the sensitivity was the slope of the analytical curve.
The precision of the system (relative standard deviation)
was determined for a concentration of 100-200 x LOD.
Continuous-Sampling Mode Results
Limits of detection
The resulting LOD's for the six elements are given
in Table 3, along with the temperature of atomization, vari-
ous gas flow rates, and the linear dynamic ranges (LDR).
The atomic fluorescence LOD's obtained by the present
non-flame cell were compared to the best limits of detection
obtained with a flame and a resistively heated discrete
sampling non-flame cell in Table 4. Both concentration
(in ig ml-1) limits of detection (CLOD) and absolute (in
ng) limits of detection (ALOD) are listed in Table 4. The
EXPERIMENTAL CONDITIONS AND ANALYTICAL RESULTS FOR MEASUREMENT OF SEVERAL ELEMENTS BY MEANS OF
THE CONTINUOUS-SAMPLE INTRODUCTION TUBE FURNACE ATOMIC FLUORESCENCE SPECTROMETRIC SYSTEM
Ele- Wave- EDL Conditions Tube Furnace Conditions LODe LDRe RSD Sensitivityg
ment length b a Powerb m Ar Flow Ar Flow HFlow
(nm) (C) ( (OC) Rate Rate Rated
Nebu- Ex- Ex-
lizer 1 ternall eternal
(1 min- )(1 min )(1 min-
Te 214.2 430 30 1515 0.78 4.7 1.8 lxlO-2 lxlO4 0.048 0.87
Ag 328.0 530 60 1675 0.78 4.3 1.6 2x10l3 1.5xlO3 0.035 0.98
Pb 283.3 430 30 1675 0.78 4.3 1.6 1x10-2 8x103 0.032 1.0
Sn 303.4 275 90 1800 0.78 5.8 1.8 1xlO2 lxl03 0.045 1.3
T1 377.6 435 45 1675 0.9 4.7 1.6 6xl0-5 3x10 0.036 1.0
Bi 307.7 500 50 1540 0.9 4.7 1.8 2x10l 1x104 0.015 0.96
aTemperature of air flowing by EDL.
bMicrowave power applied to EDL using thermostated "A" antenna.
CTemperature of graphite tube furnace walls.
dFlow rate of Ar and/or H2 around tube furnace exit.
eLOD = limit of detection (see text for definition).
LDR = linear dynamic range (see text for definition).
fRSD = relative standard deviation (taken at a concentration of about lOOxLOD).
gSensitivity slope of Log-Log plots.
COMPARISON OF NON-FLAME ATOMIZATION ATOMIC FLUORESCENCE SPECTROMETRIC LIMITS
This Work Discrete Atomization Flame
CLOD ALOD CLOD ALOD CLOD ALOD
ment (ig ml-1) (ng) (pg ml-1) (ng) Ref. (pg ml-) (ng) Ref.
Sn 1xlO2 1x100 2xlO1 lxlO1 33 xO1-1 5x102 83
Pb x 3 5xl 3 10 l1 5xlO3 3x103 36 lxl02 5x101 84
Te 1x102 lx10 6xl0-3 3x101 85
Ag 2xl0- 2x10-2 8x10-4 4x104 33 lxlO- 5x10 86
T1 6x10- 7x101 4xl0-2 2xlO-2 33 8x10lO 4x10 86
Bi 2x105 2x10-1 5xlO3 2x10 85
flame atomic fluorescence ALOD was approximated by assuming
that measurements were performed for 1 minute with a sample
flow rate of 5 ml min1.
In general, the present system operated in the
continuous mode resulted in concentration limits of detec-
tion which did not vary significantly from those concentra-
tion limits of detection found by flame and non-flame
atomic fluorescence. The absolute limits of detection of
the present system operated in the continuous mode were
generally 1-2 orders of magnitude better than flame atomic
fluorescent limits of detection, but about 1 order of magni-
tude poorer (higher) than filament non-flame atomic
Linear dynamic range and sensitivity
The atomic fluorescence analytical curves for Te,
Sn, Ag, Bi, Pb, and T1 with the present vitreous carbon
tube in the continuous-sampling mode are given in Figures
14, 15, 16, 17, 18 and 19, respectively. The linear
dynamic ranges varied from 1 x 103 for Sn to 1.5 to 104
for Bi, and are listed in Table 4. Bi has a slightly
concave analytical curve; however, it is certainly
analytically useful. The discrete sampling filament
non-flame cell, when measurements were performed by atomic
fluorescence spectroscopy, appears to have significantly
shorter linear dynamic ranges (1 x 102 5 x 10 ) for a
number of elements investigated .
I I 0 I
10 I 10 10 I(
Concentration (p.g mrl)
Fig. 14.--Atomic fluorescence analytical curve for Te in
the continuous mode of operation.
SI I I
0.1 0 10
Concentration (Ipg mTI)
Fig. 15.--Atomic fluorescence analytical curve for Sn in
the continuous mode of operation.
10 1 I I0 102
Concentration (/g mi )
Fig. 16.--Atomic fluorescence analytical curve for Ag
in the continuous mode of operation.
S 10 -
c rI I 2
162 1 1 I0 10
Concentration (/pg ml )
Fig. 17.--Atomic fluorescence analytical curve for Bi in
the continuous mode of operation.
I I I0 10
Concentration (2g m11)
Concentration (ug ml )
Fig. 18.--Atomic fluorescence analytical curve for Pb in
the continuous mode of operation.
I ( I I
1 I 10 10
Concentration (,pg m11)
Fig. 19.--Atomic fluorescence analytical curve for T1
in the continuous mode of operation.
Correction for scattering
All atomic fluorescence signals were corrected for
any spurious scatter signals by one of two methods, and in
all cases the results of the two methods agreed within
experimental error. If a source line was available (that
is, an Iodine line) which was not a fluorescent line of the
analyte, then the ratio of a non-absorbing line intensity,
Ina, of the source to the primary excitation line, IF, of
the source was calculated.
R = IV-1
Since only lines in the approximate spectral region of
fluorescence were used, the variation in photomultiplier
spectral response, in spectral transmission of the lenses
and all other optical components, was not corrected for.
The scatter signal, Ss, and the apparent fluorescence
signal, SF, were then measured, and the corrected scatter
signal, S'na, from the atom cell was calculated from
The corrected fluorescence intensity S'F from the atom cell
S' Sp S'na IV-3
for the first method.
The second procedure used to correct for spurious
i. measure blank deionizedd water) signal, Sb, at
the temperature of atomization;
ii. measure apparent fluorescence signal, SF, for
concentration of interest;
iii. decrease the temperature of the vitreous carbon
tube furnace to 100 OC and measure any scatter
signal, Ss. The maximum scatter signal was then
assumed to be Ss and the corrected fluorescence
signal S' was calculated from
S'F = SF Ss b I
This method assumes that
i. no appreciable atomization has occurred in the
vitreous carbon tube furnace in step iii, and
hence no atomic fluorescence was measurable;
ii. any scatter signal measured at high temperatures
not measured at low temperatures was due to hot
carbon particles and radical formation;
iii. the scatter signal due to the analyte could only
increase by lowering the temperature of the
vitreous carbon tube due to reduced desolvation
and possible coagulation.
Since this second method may over-correct for scatter, it
was not used in cases where scatter was estimated to be
over 4 per cent of the fluorescent signal (see items i and
iii listed above).
The precision for Sn, Pb, Ag, Bi, Tl, and Te is
listed in Table 3. The precision obtained is very competi-
tive with all other atomization devices in use for atomic
fluorescence. For all elements examined, the relative
standard deviation is 0.025 to 0.048, which is comparable
to that observed in flame APS and generally better than
that of graphite rod AFS. The present continuous intro-
duction non-flame cell is not subject to the sample place-
ment errors found in the graphite rod atomizers.
Atom population profiles
The method used to measure the decay of atom
population-height profiles has been previously described
[23,53,34]. Atomic profiles as a function of height above
the vitreous carbon tube furnace for several elements were
examined in a similar manner for the present atomization
cell to determine the effect of only an Ar sheath versus
an Ar-H2 diffusion flame sheath on the decay of atom
populations exiting continuously from the present con-
tinuous-sampling vitreous carbon tube furnace. The same
atomization conditions were used for this study as for the
The factors contributing to the decrease in the
measured peak atomic fluorescence as the vitreous carbon
tube furnace was lowered include: (i) the loss of atoms
by chemical reaction; (ii) transport of atoms outside the
light path by diffusion; (iii) decrease of the solid angle
subtended by the atoms in the sample cavity; (iv) quenching
of the excited-state atom populations by collisions,
particularly with diatomic and polyatomic species. Even
in the absence of condensation, oxide formation, and other
possible chemical losses, a gradual decrease of the
fluorescence signal from factors (ii) and (iii) would be
The decay of atom populations with height above the
vitreous carbon tube furnace for Sn and Pb was measured by
atomic fluorescence spectrometry and the results are shown
in Figure 20. In the case of Pb with the Ar-H2 diffusion
flame, the decay of the atom populations was gradual,
reaching 50 per cent at a height of 20 mm above the furnace
exit. With only an Ar sheath, however, the decay of Pb
proceeded somewhat faster, reaching 30 per cent of the
original population at a height of 20 mm. For Sn, the
Ar-H2 diffusion flame had very great influence on the atom
population as a function of height above the furnace tube
exit; at 12 mm above the carbon tube, the fluorescence
signal actually increased 300 per cent, and then it began
to decrease to slightly less than 150 per cent at 24 mm.
With only Ar, the Sn population dropped to 28 per cent of
the original fluorescence signal at 7 mm.
50 Pb; Ar H2
5 10 15 20 25
Fig. 20.--Decay of atom populations with height above the
vitreous carbon tube atomizer for two elements
as measured by atomic fluorescence spectrometry.
0, Sn, 3034 A; 0 Pb, 2833 A; X, Ar atmos-
phere; *, Ar/H2 atmosphere.
For the two elements (Pb and Sn) studied here, the
atomic population-height profile was less steep for the
present continuous-sample injection non-flame cell compared
to that of the discrete graphite carbon filament atomizer
. For instance, at a height of 5 mm above the atomizer,
the fluorescence signal of Sn with Ar entrainment is about
55 per cent of the initial signal for the present atomizer
as compared to less than 2 per cent of the initial signal
for the graphite filament atomizer. When a H2 diffusion
flame was used with the present system, the fluorescence
signal increased about 3-fold, while with'the carbon
filament atomizers, fluorescence decreased to about 20 per
cent at 15 mm height.
In the case of Pb at 10 mm height above the non-flame
cell, the atomic fluorescence signal was about 60 per cent
of the initial fluorescence signal with either the Ar or
Ar-H2 entrainment sheath, while in the case of the graphite
filament atomizer, the fluorescence signal was less than
30 per cent with a H_ diffusion flame and less than 1 per
cent with only an Ar sheath.
These results tend to indicate that the presence
of the H2 diffusion flame helps to prevent the rapid
attenuation of the atomic populations, and in the case of
Sn, it actually increases the atomic populations at some
heights above the vitreous carbon tube furnace. Previous
studies on a similar type of H2 diffusion flame suggested
that the atomic cell was characterized by a region of an
excess unburned H2, which minimized the entrainment of
In the case of Sn, where the fluorescence intensity
increased to a height of 15 mm over the vitreous carbon
tube, the indication was that H2 was not only protecting
the metal atom cloud from oxide formation, but also was
assisting in Sn-atom production from the molecules formed
previously by reactions such as
MO + H2 --- M + H2
MO + H M + OH
where M was some metal. The explanation for why this
occurs to a far greater extent with the vitreous carbon
tube furnace than with the carbon filament may lie in the
higher temperatures of the furnace, which render the H2
diffusion flame of the former more reactive.
Pulsed-Sampling Mode Results
Limits of detection
The atomic fluorescence limit of detection (LOD)
for this study was defined as that concentration (in ug ml-1)
giving a fluorescence signal of 3 x rms noise of the blank.
The concentration limits of detection, along with the vari-
ous gas flow rates, temperature of atomization, and EDL
conditions for a 10 ml sample size for each element are
given in Table 5. The resulting concentration and abso-
lute LOD's are compared in Table 6 to the best atomic
fluorescence limits of detection obtained by previous
workers with discrete sampling non-flame cells and flames,
as well as to the previous LOD's calculated for the
present system used in the continuous-sampling mode. The
best concentration and absolute limits of detection obtained
by flame atomic absorption, graphite tube atomic absorption,
and the plasmas are also given. Generally the concentra-
tion limits of detection by the present system operated in
the pulsed mode are comparable to most other methods, while
the absolute limits of detection by the present non-flame
cell were generally superior by 1 order of magnitude or
Size of sample
The fluorescence signal did not increase linearly
with increasing sample sizes of 1 pl, 5 gl, and 10 l for
the three elements shown in Figure 21. The figure indi-
cates that the best atomization efficiency is obtained for
1 l and 5 pl samples, while the efficiency drops off
slightly for the 10 pl sample size.