A detailed study of laser enhanced ionization with electrothermal vaporization-flame atomization for trace element analysis


Material Information

A detailed study of laser enhanced ionization with electrothermal vaporization-flame atomization for trace element analysis
Physical Description:
xiii, 216 leaves : ill. ; 29 cm.
Riter, Ken Lynn, 1970-
Publication Date:


Subjects / Keywords:
Chemistry thesis, Ph. D   ( lcsh )
Dissertations, Academic -- Chemistry -- UF   ( lcsh )
bibliography   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1996.
Includes bibliographical references (leaves 205-215).
Statement of Responsibility:
by Ken Lynn Riter.
General Note:
General Note:

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 026437892
oclc - 36703449
System ID:

This item is only available as the following downloads:

Full Text







Dedicated to the loving memories of my mother, Namiko

(Tamanaha) Riter (September 16, 1936 July 9, 1983), and

father, Roger Lynn Riter (July 12, 1945 August 10, 1996).

Without their love, encouragement, and support none of this

would have been written.


First, I would like to thank Dr. Jim Winefordner for the

opportunity to research in his lab. Although I have learned

much about laser spectroscopy in Jim's lab, his example of how

to be a decent person and to treat others with respect is what

will always be with me. I would like to thank Dr. Ben Smith

for all of his advice and help in the lab. Setting up the

instrumentation would have been an enormous task without Ben's

help. I would like to thank Dr. Oleg Matveev for working with

me and lending his expertise in laser enhanced ionization to

the project. I have learned so much about LEI from Oleg in a

very short period of time. Oleg's kindness, patience, and

modesty allowed him to explain very difficult concepts rather


I would like to thank Leah Mordoh and Wendy Clevenger for

assisting with and performing many of the experiments for the

LEIS of Mg. Others I would like to thank include Rob Guenard

for his help with the ultrasonic nebulizer, Chester Eastman in

the machine shop for building the new LEI burner, and all of

the Winefordner and Harrison group members for their help and

friendship. I would like to thank the National Institutes of

Health for funding this research (Grant # 5-R01-GM49638-03).

On a more personal note, I would like to thank my family

for their love and encouragement. Thanks to Jon DeGnore, Dr.

Bill Walden, and Dr. Wei Hang for being such good friends and

making my stay in Gainesville enjoyable. Finally, I would

like to thank my fiancee, Leah. Without her love and support,

I do not know how I would have made it through these final few


All thanks be to God.



LIST OF TABLES . .... .viii


ABSTRACT . . ... xii



Absolute/Standardless Analysis . 1
Analysis of Real Samples ............ 5
Determination of Lead in Whole Blood . 6
Intent of Dissertation . .. 7



The Optogalvanic Effect . 9
General Principles of LEI . .. 10
Atomization of Sample . .. 11
Excitation of Atoms . .. .16
Ionization ................. 20
Charge Collection . .. .20



Introduction . ..... .25
Thermal Ionization . ... .25
Processes Reponsible for Thermal Ionization 27
Thermal Ionization Rate of an Atom in a Flame 29
Modeling of Laser Enhanced Ionization .. .30
Rate-Equation Formalism . .. .30

Degree of Ionization for Two-Step
Excitation . .
Density-Matrix Formalism ....
Detection of the Ionization Signal .. ..
One-Dimensional Approximation .. ..
Point Charge Model .....
Electrothermal Vaporization ...
Absolute Analysis ......



Analytical Performance of Flame-LEI ... .47
Limits of Detection and Sensitivities ... 47
Noise and Interferences . 58
Applications of LEI to Real Samples .. 61
Determinations Without Interferant Removal 62
Determinations With Interferant Removal 66
Hybrid Techniques and Non-Flame Atom Reservoirs 71
Electrothermal Vaporizers . ... 71
Hybrid Combinations of Flame and
Electrothermal Vaporizers . ... 74
LEI in the Inductively Coupled Plasma ... 79
Other Methods and Reservoirs .. .80



LEI ..... . . 83
Burners .. . 83
Graphite Furnace .... ... 88
Procedure and Conditions . .. .97
Flame Gas Flows, Velocity, and Temperature 103
Noise Study . .... 105
Fluorescence Dip and Fluorescence Profile of Flame 108
Transport Efficiency . 113
Transimpedance Amplifier Calibration ...... 118
Atomization Efficiency Measurement .. .119



LEI of Magnesium . . 124
Magnesium As Analyte . .. .124
Mg LEI Signal . ... 127
System Parameter Optimizations for Old Burner

Design .
Flame Profile With Old Burner .. ...
Analytical Curve With Old Burner
New Burner Design .. .....
System Parameter Optimizations With New
Burner ...
Flame Profile With New Burner .. ..
Matrix Modifier/Carrier .. ...
Analytical Curve With New Burner .
Flame Temperature and Flame Gas Velocity
With the New Burner .. ....
Absolute Analysis ......
Vaporization Efficiency .. ...
Transport Efficiency ....
Probing Efficiency ......
Detection Efficiency ....
Atomization Efficiency ...
LEI of Lead .........
Excitation Scheme for Lead ..
Carrier ......................
Calibration Behavior ....



Absolute Analysis .
Pb in Blood .
Future Work .

REFERENCES ..........


. . 202

. . 202
. . 203
. . 203


Table aage

1. LEI limits of detection . ... 48

2. Graphite furnace temperature program for
magnesium .... . .. 99

3. Graphite furnace temperature program for lead in
blood .... .. . 102

4. Values used for flame temperature calculation .175

5. Mg concentration in different cotton samples 177

6. Enhancement of LEI signal for different two-step
excitation schemes for lead . ... .192



Figure page

1. Processes needed for laser enhanced ionization
spectrometry .... . 13

2. Typical experimental setup for flame-LEI ... .15

3. Typical excitation schemes for LEI spectroscopy,
a) one-step excitation using visible light, b)
one-step excitation using ultraviolet light, c)
two-step excitation (direct), d) two-step
excitation (indirect), and e) three-step
excitation . .... . 19

4. Various electrode arrangements used for LEI
spectroscopy, a) split-cathode rod arrangement,
b) split-cathode plate arrangement, c)water-
cooled, immersed cathode arrangement, d) water-
cooled coiled cathode arrangement ... .22

5. Schematic representation of various excitation
and deexcitation processes in a three-level atom:
nj, n2, and n3 are the number densitites of the
three levels, respectively; k21 is the sum of the
collisional deexcitation and spontaneous emission
rates between levels 2 and 1; k12 is the
collisional excitation rate between levels 1 and
2; k3,ion (k2,ion) is the collisional ionization rate
from level 3 (2); Bi2Uv(v12) (B23U,(v23)) and B21U(v21)
(B32U, v32)) are the absorption and stimulated
emission rates and U.(v21) (=U,(v12)) and U,(v23)
(=U,(V32)) are the spectral irradiances of the
laser light . . 32

6. Block diagram of the experimental setup for LEIS 85

7. First burner design used for LEIS ... .87

8. Diagram of new burner design with relative
position of the high voltage electrode and laser
beams .. . 90

9. Detailed drawing of the new burner design .... 92

10. Cut-away view of the graphite furnace showing the
tantalum sample extraction interface ... 96

11. Sketch of laboratory constructed ultrasonic
nebulizer used . ... 107

12. Block diagram of experimental setup for
monitoring of both fluorescence and LEI signals 112

13. Sketch of experimental setup for transport
efficiency measurement . ... 116

14. Block diagram of the experimental setup for the
determination of the atomization efficiency for
Mg by atomic absorption . ... 122

15. Oscilloscope trace of the laser beam timing 126

16. Typical LEI signal for magnesium with older
burner . . 129

17. Argon flow rate optimization for Mg with older
burner .... . 133

18. Burner-to-electrode distance optimization for Mg
with older burner . ... 135

19. Applied voltage optimization for Mg with older
burner . . ... 137

20. Horizontal profile of the flame with the older
burner . . 140

21. Analytical curve for Mg with the older burner .142

22. Log-Log plot of the analytical curve of Figure 21 144

23. Noises for the hydrogen/air and acetylene/air
flames with the older burner design ...... 147

24. Comparison of the noises for the new and old
burner designs . .150

25. Argon flow rate optimization with new burner 152

26. Acetylene flow rate optimization with new burner 154

27. Air flow rate optimization for new burner 156

28. Burner-to-electrode distance optimization for new
burner . ... .. 159

29. Electrode voltage optimization with new burner .161

30. Horizontal profile of Mg atoms in flame with new
burner . . .. 163

31. Fluorescence profile of Mg atoms in the flame
with the new burner . .. 165

32. LEI signal for xylene while scanning dye laser
for X. . . ... .. 168

33. Effect of methanol on LEI signal for Mg ... 171

34. Analytical curve for Mg with new burner and
methanol . . 173

35. Mg LEI signal with increasing laser repetition
rate . . ... 181

36. Dye laser conversion efficiency with increasing
laser repetition rate for . .. .183

37. Dye laser conversion efficiency with increasing
laser repetition rate for X2 ... 185

38. Analytical curve for aqueous lead standards with
and without NaC1 addition . ... 194

39. Log-Log plot of analytical curves for aqueous
lead and blood lead . ... 197

40. Analytical curve for diluted blood lead standards 199

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



Ken Lynn Riter

December 1996

Chairperson: James D. Winefordner
Major Department: Chemistry

A system coupling electrothermal vaporization with flame-

laser enhanced ionization (ETV-FL-LEI) was examined for the

possibility of "absolute" analysis. For a method to be

considered absolute, analytical matrix interference must be

eliminated, the stability of the calibration over time must be

established, and the theoretical equation relating the signal

to the quantity of analyte must be known. For our system, the

LEI signal for magnesium is equal to the product of the

Faraday number (96,485 C/mol), moles of Mg, and overall system

efficiency. In our case, the overall system efficiency is the

product of the vaporization efficiency of the ETV, the

transport efficiency of Mg from the ETV to the flame, the

atomization efficiency of Mg in the flame, the probing


efficiency of the laser beams, and the detection efficiency.

Ideally, these efficiencies should be unity. However, it was

found that all of these except the vaporization efficiency was

less than unity. Also, the LEI signal deviated from linearity

at low Mg concentrations and required the addition of a matrix

modifier to restore the signal. This indicates a dependence

of the LEI signal on the sample matrix. Therefore, ETV-FL-LEI

should not be considered an absolute method.

A second project involved the application of our ETV-FL-

LEI system to the determination of lead concentration in whole

blood. Blood standards from the Centers for Disease Control

(CDC) and the National Institute of Standards and Technology

(NIST) were diluted 21:1 with ultra pure water and analyzed.

Good agreement was found between the CDC and NIST standards.

A linear analytical curve was obtained with a detection limit

(30) of 8.9 X 10-3 pg/dL (890 fg absolute) for lead in whole

blood. This compares favorably with other current methods for

blood lead determinations including isotope dilution

inductively coupled plasma mass spectrometry (ID-ICP-MS) and

graphite furnace atomic absorption spectrometry (GFAAS).



Absolute/Standardless Analysis

There are no analytical procedures which are "absolute"

in the strictest sense of the word, because to analyze

absolutely i.e. without any presuppositions it would be

necessary to identify the atoms and the molecules of the

sample, to sort them out, and to count them individually and

completely [1].

However, if a theoretical expression is known for the

function relating the signal to the absolute quantity of

analyte present that is sufficiently reliable to allow a

direct calculation of the quantity of analyte from a single

measurement, then this method is called "an absolute method of

analysis" [2]. The most complete program for the development

of absolute methods of analysis includes [2]: (1) elimination

of analytical matrix interference, (2) stabilization of the

calibration over time, and (3) theoretical calculation of the

calibration based on fundamental parameters and actual

measurement conditions. This should be distinguished from

"standardless" analysis where the calibration curve is stable

over time. Therefore, once the system has been calibrated for

a particular sample, the calibration should need to be checked

only occasionally (such as once every 8 hours) with a

standard (concentration = 100 to 1000 times the limit of


Many classical methods, such as precipitation reactions,

titrimetry, and coulometry, are considered absolute. When

considering modern instrumental methods for absolute analysis,

usually atomic absorption, where relative measurements are

made, is considered the best candidate rather than emission or

fluorescence procedures where absolute radiometric

measurements are required. According to L'vov [2], this is

not surprising since the atomic absorption method of measuring

the analytical signal is free from many of the uncontrolled or

difficult-to-control factors typical of emission/fluorescence

methods. Also, the stability and consistency of calibration

for modern graphite furnace atomic absorption spectroscopy

(GF-AAS) have brought GF-AAS close to achieving absolute

analysis. However, the sensitivity of GF-AAS, although high,

is still well within the possibility of preparation of

accurate standard reference solutions with minimal

contamination and loss problems. In this respect, absolute

analysis by GF-AAS will most likely never be a necessity.

However, GF-AAS is very amenable to standardless analysis

because of the reproducibility and consistency of the

calibration. Electrothermal vaporization-inductively coupled

plasma-mass spectrometry has achieved, in some cases, low or

even sub-femtogram detection limits. However, it is essential

that standard samples or solutions be used for calibration,

since the transport efficiencies of analyte ions are unknown

and vary significantly with analyte type, matrix type, gas

flow rates, sampling cone and skimmer cone geometries and

electrostatic lens configuration and conditions.

Laser induced fluorescence with graphite furnace

atomization (GF-LIF) is one of the most sensitive methods for

trace element analysis. However, absolute analysis by GF-LIF

involves a difficult and time consuming calibratrion process

relating the signal level to the mass of the analyte and

requires corrections for laser induced ionization, thermal

ionization, and matrix background, and evaluation of or

knowledge of the diffusion coefficient of analyte atoms at the

furnace atomization temperature.

Analytical methods based upon ionization are potentially

the best candidates for absolute analysis. Ions can be

produced with great selectivity and, once an ion is produced,

the probability of detection is generally high. Moreover,

most of the complexities associated with absolute fluorescence

measurements are avoided; the relationship between amount of

analyte and the measured signal is substantially simpler. A

new analytical approach using laser enhanced ionization with

sub-fg (and sub-pg/mL) detection limits has been described by

Smith et al. [3,4] which shows great promise as a standardless

or absolute method. This method involves coupling of a

graphite furnace for sample vaporization with an acetylene/air

flame for laser enhanced ionization (LEI) detection. In

essence, this method can be considered as the analog of

coulometry (i.e. each atom will produce a charge of 1.6 X 10'-

C or 96,487 C/mol). It is hoped that the combination of high

sensitivity and simplicity of detection will make LEI a good

candidate for absolute analysis.

There are several fields in which extreme sensitivity

(sub-fg) combined with absolute analysis would be useful since

the preparation of standards at very low analyte

concentrations is difficult because of sample loss and

contamination problems. This makes the development of an

absolute analytical protocol a pressing need [5]. In medical

research, the determination of the microdistribution of trace

metals in pg amounts of human tissues is essential and

concentration levels at or below pg/g are common [6].

However, the preparation of standards is difficult or

impossible because of the complexity of the sample matrix.

Similar detection capability is required for the determination

of trace elements in small amounts of solid samples of

interest in criminalistic, expert legal, and forensic medical

cases [7]. In situations where the preparation of hazardous

standards needs to be avoided, e.g. the analysis of

radioactive species, a means of quantitation using an absolute

approach would be desirable. These examples are indicative of

the situations where absolute analysis by graphite furnace-

flame-LEI spectroscopy would be desirable: (i) when the

quantities of sample are limited and (ii) when standards are

difficult or impossible to prepare due to the very low

concentrations being used, the difficulty of simulating very

complex sample matrices, or the danger of handling hazardous


Analysis of Real Samples by LEI Spectrometrv

Laser enhanced ionization spectrometry (LEIS) is a

sensitive technique for trace element analysis and has become

well established since its discovery in 1976 [8].

Unfortunately, the application of LEIS has been limited mostly

to simple aqueous systems because of ionization interference

encountered in complex matrices. Thus, the application of

LEIS to complex real samples, such as biological fluids and

environmental samples, has remained largely undeveloped.

The combination of the graphite furnace, used as an

electrothermal vaporizer (ETV), with a flame for LEI detection


has many advantages for the analysis of real samples. By

separation of the graphite furnace vaporization and LEI

detection processes, this system overcomes the problems of

thermionic emission from the graphite tube and poor

collisional ionization encountered with LEIS in the graphite

furnace. This system also allows for the removal of many

interfering matrix species by temperature programming of the

graphite furnace. As a result, ionization interference that

have plagued LEIS of real samples may be reduced or even


Determination of Lead in Whole Blood

Since the Centers for Disease Control (CDC) lowered its

pediatric level of concern for blood lead to 10 jg/dL in 1991,

there has been increased interest in more sensitive methods

for the determination of lead concentrations in blood. There

also exists a need in research laboratories for accurate and

precise measurements of substantially lower blood lead levels

(< 10 g/dL) to establish the levels for chronic lead toxicity

in humans [9]. Some studies indicate that there may be no

threshold for lead toxicity in humans [10,11], and so the need

for more sensitive methods for determining blood lead levels

is clear.

Isotope dilution inductively coupled plasma mass

spectrometry (ID-ICP-MS) and graphite furnace atomic

absorption spectrometry (GFAAS) are both methods currently

used to measure low levels of lead in blood. However, both of

these methods require extensive sample preparation with a

matrix modifier, and have detection limits of only around 1

jg/dL for lead in whole blood [12,13]. With the combination

of graphite furnace with flame-LEIS, little sample preparation

is needed (21:1 dilution with ultra pure water) unlike with

ID-ICP-MS and GFAAS. This not only reduces the complexity of

sample preparation but also reduces sample contamination from

matrix modifiers. This becomes especially important at very

low blood lead levels, so improved detection limits may be

expected. Another advantage of the ETV-LEIS combination is

the high sensitivity afforded by LEIS.

Intent of Dissertation

The aim of the present work was to satisfy the third

condition for absolute analysis (stated earlier) by

characterizing the efficiencies associated with our

experimental setup (consisting of the combination of a

graphite furnace with an acetylene/air flame for LEI

detection) to obtain the theoretical equation relating the

analyte quantity to the LEI signal. This equation and the

efficiencies will be discussed in more detail in Chapter 2.

A second objective was to explore the possibility of

applying this technique to the analysis of real samples with

complex matrices. We chose to analyze lead concentrations in

a blood matrix.


The Optogalvanic Effect

The optogalvanic effect (OGE) is the term for the process

whereby the collisional ionization rate for an element is

enhanced by optical excitation to a higher electronic energy

state [8,14]. This mechanism was first postulated by Foote

and Mohler _n 192E 15]. The first purely optical/collisional

effect, however, was not actually observed until 1928 by

Penning [161. Penning observed the effect as a change in the

voltage drop across a neon discharge when illuminated by a

second neon discharge.

Unfortunately, optical enhancement of collisional

L;nizatis _- too .-eak of an effect to be seen in usual atom

reservoirs with conventionall light sources. Therefore, it

would not be until the advent of tunable lasers that the OGE

could be easily observed and used as a spectroscopic method.

In 1976, researchers at the National Bureau of Standards

'BS, now -he Nati-'al Institute of Standards and Technology,

7IST) decided to investigate the hollow cathode lamp as a

reservoir for laser induced fluorescence (LIF) after

:isappointing results for LIF in a flame [17]. These

researchers fortuitously discovered a change in the voltage


across the hollow cathode lamp when the dye laser was tuned to

an electronic transition of one of the atomic species in the

discharge [14]. Shortly afterwards, the same group decided to

look for a related effect in flames. The effect was observed

and a sub-ppb detection limit for sodium, equivalent to their

results with LIF in a flame, was quickly obtained [8].

The term, laser enhanced ionization (LEI), was first

introduced in 1978 by the group at NBS [18]. Today, LEI is

used to describe the optogalvanic effect in flames and other

reservoirs except discharges. OGE or optogalvanic

spectroscopy (OGS) is now used exclusively for phenomena

taking place in discharges.

General Principles of LEI

Laser-enhanced ionization spectrometry (LEIS) can be

defined as a spectroscopic method where an enhancement in the

normal collisional (thermal) ionization is obtained by optical

excitation of the atoms under study by resonant laser light.

This enhancement is detected as a change in the current

passing through a medium (atom reservoir) between two

electrodes at different potentials.

LEI can be broken down into four distinct processes:

atomization of the sample, optical excitation of the analyte

atoms by resonant laser light, collisional ionization of the

excited atoms, and collection of the charges produced

(Figure 1).

Atomization of Sample

The conversion of the sample into an atomic vapor or

atomization of the sample is dependent upon the particular

characteristics of the atom reservoir used. To date, many

different atom reservoirs have been used for LEI including

atmospheric flames, graphite furnaces, and inductively coupled

plasmas (ICPs). However, the large majority of LEI studies

have been done in atmospheric flames. This is because of the

combination of good qualities that flames exhibit for

atomization, ionization, and the detection processes. A

typical flame-LEI setup is shown in Figure 2 and consists of

a flame, laser system, electrodes, and electronic detection


In conventional flame-LEI, the sample is aspirated by a

pneumatic nebulizer into a spray chamber. The sample leaves

the spray chamber as a spray or mist of fine droplets and

enters the flame. The flame heats this spray and causes the

solvent to vaporize leaving dry aerosol particles. Further

Figure 1. Processes needed for laser enhanced ionization
spectrometry [19]


o at
'o~y^ M^ GS

Figure 2. Typical experimental setup for flame-LEI [19]


Dye Dye
Laser Laser Sample
1 2

:i::i :i:'.r!::: : !: Pum p Laser

:::. . .


heating in the flame volatilizes these dry aerosol particles

producing atomic, molecular, and ionic species.

Many different flames have been used for LEIS including

acetylene-based flames such as CzH,/air and C2H2/N20 and

hydrogen-based flames such as H2/NO, and Hj/Oz/Ar. The cooler

hydrogen-based flames have little flame-ion production so the

background noise from the flame is low. However, the flame

temperature and flame composition have been found to be the

most important factors in obtaining strong LEI signals,

because these factors greatly affect the atomization

efficiency. Therefore, the hotter acetylene-based flames are

most commonly used as atomic reservoirs in LEI. On the other

hand, if the flame temperature is too high, thermal ionization

of the analyte may be considerable resulting in poor detection

limits for atomic LEI. The most common flame used for LEI has

been the air/acetylene flame in which a large number of

elements can be conveniently analyzed. Hotter flames, such as

CH2/N2O, are mostly used for refractory elements.

Excitation of Atoms

Optical excitation of the analyte atoms is usually

performed by a pulsed dye laser. The dye laser may be pumped

by a variety of sources including flashlamps, Nd:YAG, excimer,

and N2 lasers. Continuous-wave (cw) lasers have been used

infrequently for LEI because cw techniques are often too


complicated for analytical applications. This is mostly due

to the ability of pulsed laser sources to produce much higher

intensity light, especially in the ultra-violet region.

Another advantage of pulsed lasers for LEI is that the excess

charge created can be collected during a shorter period of

time, which reduces the influence of background current noise.

Flashlamp-pumped dye lasers usually have pulse durations

in the microsecond range while excimer, Nd:YAG, and N2-pumped

dye lasers usually have pulse durations around 5-20 ns. Pulse

energies vary typically from 0.1 to 100 mJ in the visible and

1 mJ to 10 mJ in the ultra-violet region depending on the

laser system, dyes, and crystals used. The wavelength region

covered by dye laser systems typically ranges from 220 nm to

1000 nm. Repetition rates used are normally around 5-100 Hz.

Many different excitation schemes may be used for LEI

spectroscopy, some of which are shown in Figure 3. One-step

excitation using either visible or ultra-violet light has been

used extensively in flame-LEIS. For elements that are more

easily ionized, a single-step scheme is sufficient to achieve

low limits of detection. However, for many elements, it is

favorable to use a two-step excitation scheme. Most two-step

schemes share an intermediate level although this is not

necessary if the upper level of the first step and the lower

level of the second step are sufficiently coupled by

Figure 3. Typical excitation schemes for LEI spectroscopy, a) one-step excitation
using visible light, b) one-step excitation using ultraviolet light, c) two-step
excitation (direct), d) two-step excitation (indirect), and e) three-step excitation.

Eo I I L
a) b) c) d) e)


collisions. The addition of the second step usually results

in a significant increase in the signal strength (up to three

orders of magnitude) compared to one-step excitation. Three-

step excitation schemes have also been used but are not common

in flame-LEIS.


The common atomic reservoirs (flames, furnaces, and

plasmas) result in thermal ionization. Most atoms ionize

through collisions with thermally excited molecules in the

reservoir. Therefore, the thermal ionization rate depends on

the reservoir constituents, temperature of the medium, and the

ionization potential of the element of interest.

When an element is excited by laser radiation, the atomic

population of that element is greatly altered. Excited atoms

are more easily ionized by collisions than those in the ground

state, so an increased ionization rate results.

Charge Collection

The additional charges produced by optical excitation are

collected by applying an electrical field across the flame (or

other atomic reservoir) using an electrode arrangement and

measuring the current change. Many different electrode

arrangements have been used, some of which are shown in Figure

Figure 4. Various electrode arrangements used for LEI spectroscopy, a) split-cathode
rod arrangement, b) split-cathode plate arrangement, c) water-cooled, immersed
cathode arrangement, d) water-cooled coiled cathode arrangement.

H.V. -H.V. -H.V. H.V.





a) b) c) d)


In all four schemes, the burner head acts as the anode

and is connected to the detection electronics. In the split-

cathode rod arrangement (Figure 4a), two metal rods are

positioned on opposite sides of the flame and biased to the

same negative high voltage with respect to the burner head.

This arrangement was found to be very sensitive to easily

ionized matrix species [20,21]. This led to the development

of the split-cathode plate arrangement (Figure 4b) which was

used for several years [22,23]. This was the case because of

its stable electric field distribution in the flame,

relatively high contamination resistance, and long lifetime.

The stainless steel, water-cooled, immersed cathode

arrangement (Figure 4c) was developed to locate the cathode as

close to the laser excitation zone as possible [24]. The

maximum signal strength and optimal resistance to electrical

matrix interference should occur with the laser excitation as

close to the cathode as possible. This electrode is also easy

to fabricate, robust, and contributes negligible memory

effects. The saturation current is also reduced with respect

to that of the split-cathode plate, which results in a lower

background current. The water-cooled coiled cathode (Figure

4d) has

equivalent sensitivity to the immersed cathode and appears to

have even greater resistance to easily ionized matrix species


Depending on other experimental parameters such as

electrode configuration, the applied voltage varies from about

300 V to ~3000 V. The current drawn through the flame is

usually in the pA range and the amount of charge detected is

in the fC range or larger. A typical LEI signal has a

duration of 300 ns up to 1 4s. The very small LEI current is

directed through a current-to-voltage amplifier which is

usually placed very close to the flame to minimize radio

frequency noise from the pulsed laser. The d.c. current is

usually filtered out using a d.c. blocking capacitor.

The current pulse is detected synchronously using a

boxcar integrator with pulsed laser excitation or a lock-in

amplifier with a chopped continuous wave laser. The signal is

then collected and processed by a computer.




The basic principle of laser enhanced ionization (LEI) is

to enhance the existing thermal ionization rate in the flame

(or alternate atom reservoir) by optical (laser) excitation

and then to detect this increase in the ionization rate as an

increase in the current passing through the flame between two

electrodes. A discussion of the theory for two-step LEI in

flames will be covered here. A more general and in-depth

discussion of LEI theory is given by Axner and Rubinsztein-

Dunlop [26] and Travis and Turk [27].

Thermal Ionization

A combustion flame possesses a small but not always

insignificant amount of thermal ionization [5]. If an atomic



system is retained in a flame, we can define a thermal

ionization rate constant, kh.on for each specific process in

the flame. One such process is the collisional ionization

between a species M and a thermally excited collisional

partner X*:

M + X* -kt M* + e- + X (1)

where M' is the positive ion, e- is an electron, X is the

deexcited collisional partner.

Similarly, there exists a recombination rate constant,

keco,., for the reverse process, given by:

MI + e- + X krecomb M + X* (2)

When the system is in thermal equilibrium, the thermal

ionization and recombination rates balance exactly. This

enables us to write an expression for the relation between the

concentration of the species M, its ions, and the electrons.

This relation is known as the Saha equation [28] and is stated


[M'] [e-] = Kion [M] (3)

where K.o. is the ionization constant given by:

K on = th.ion
K to b (4)


It should be noted that although a third body (collision

partner) is required for the ionization and recombination

processes and the rates for each of these processes is

dependent upon the concentration of the third body, the

ionization constant (and degree of ionization of the atomic

system) is independent of the concentration of this third body

as long as thermal equilibrium conditions prevail in the


As has already been mentioned, in LEI, an electric field

is applied to the volume of interaction within the flame to

separate the created charges and make detection of those

charges possible. Consequently, the electric field will

minimize recombination from occurring at any significant rate.

Therefore, the only remaining process will be thermal

ionization, so the Saha equation is no longer valid.

Processes Responsible for Thermal Ionization

In a flame, a variety of different interactions between

atoms, molecules, and light can take place. The major

processes that lead to ionization can be divided into physical

and chemical ionization processes. The physical processes and

be further divided into collisional and radiative ionization


Collisional ionization processes are most often

considered to be responsible for ionization of atoms in

flames. An example would be the collisional ionization of

sodium atoms in the flame:

Na + X* Na' + e- + X (5)

where X represents any flame molecule [5,29].

Radiative ionization processes are usually

photoionization processes where atoms are irradiated with high

intensity, short wavelength light which results in the

ejection of an electron, such as:

Na + hv Na' + e- (6)

or by the interaction between blackbody radiation and excited


Chemical ionization processes are most easily

characterized as those in which the formation of a new

chemical bond takes place. Most alkaline earth atoms are

believed to ionize in this way. Associative ionization, such

as for the calcium atom:

Ca + OH CaOH' + e- (7)

represents such a process. There are also many charge

distribution processes that may be of importance since they

can constitute one reaction in a chain leading to a net

ionization rate.

Thermal Ionization Rate of an Atom in a Flame

To model thermal ionization, some assumptions must be

made. It is assumed that collisional ionization dominates

other ionization processes in the flame. It is also assumed

that the flame, with all of its atomic and molecular species,

is in a state of thermal equilibrium. This implies that the

concept of detailed balance between all atomic levels is

valid. In other words, the atomic energy levels are populated

according to Boltzmann's distribution:

gi exp(-E,/kT)
n = Z nato (8)

where ni is the population of the ith level (m-3), g, is the

degeneracy of the ith level, Ei is the energy of the i1h level,

n,to,, is the total number density of neutral atoms (m-3), and

Z is the electronic partition function:

Z = Fgi exp(-Ei/kT) (9)

where k is the Boltzmann constant and T is the flame


However, the thermal ionization rate will be

overestimated unless we assume that the condition of detailed

balance is not valid for the highest lying states in the atom.

If we assume that detailed balance is only valid for states up


to a certain level, then the thermal ionization rate, dn,,n/dt

(mW3 s '), can be expressed by:

dnt. 8 kT exp(-Eon/kT) n (10)
dt Z n. 91g..

where is the reduced mass of the system [g =

mato*mx/ (m.to~+mx) where mt.om and m. are the masses of the atom

under consideration and collision species X, respectively],

aion, is the ionization cross section for the species X, Eion is

the energy of the ionization limit, and nx is the

concentration of species X.

Although, with the above simplification, the thermal

ionization rate depends greatly upon what cutoff level is

chosen, the relationship between the enhanced ionization rate

due to laser excitation and the thermal ionization rate is not

very sensitive to the specific cutoff level chosen.

Modeling of Laser Enhanced Ionization

Rate-Equation Formalism

Let us assume a three-level atom illuminated by two laser

beams. For simplicity, it is assumed that the atoms consist

of only three bound levels (see Figure 5) among which laser

Figure 5. Schematic representation of various excitation and deexcitation processes
in a three-level atom: ni, n2, and n3 are the number densities of the three levels,
respectively; k21 is the sum of the collisional deexcitation and spontaneous emission
rates between levels 2 and 1; k12 is the collisional excitation rate between levels 1
and 2; k3,,. (k2,ion) is the collisional ionization rate from level 3 (2); B2U,(v12)
(B23U,(v23)) and B2iU,(vz2) (B32U,(v32)) are the absorption and stimulated emission rates
and U(v21) (=U(v12) ) and Uv(v23) (=Uv(v32)) are the spectral irradiances of the laser

B23U, (v23)

B12 v (12)

B21Uv (v21)

B32U ( v32)


2, lon



excitation takes place in two steps, weakly coupled to an

ionization continuum.

Degree of Ionization for Two-Step Excitation

The degree of ionization here is defined as the fraction

of atoms in the interaction volume that ionize during the

laser pulse. The expression for the degree of ionization is

considerably simplified by assigning an effective collisional

ionization rate from the uppermost laser-coupled level, kfion

With these assumptions, the ionization rate can be written as:

dni _,, eO f f (11)
dt = 3 3,ok io (ntot- ion) )


SBuU,(v12) 23U, (v2)
c =-(C3%) (12)
3 (BU, (v,) +a) (B3U (v23) +b) +d12

g3 (13)
(C, ) = g1+tg+, 13

a = (C3 )tak3i + (Cn) ~k (14)

(C, )
b = -T-- (k3+k32) (15)
(C2 ) sa

d = [ (C) a.2 (k,+k, ) (k21- k f- )t (16)
(C2 ) sat

(C f) ga (17)
2 ) g1+g2

o(C, :g (18)
(C2a g)g1+g2+13

not is the total number density of the analyte (neutral atoms

and ions), nion is the number density of analyte ions, and the

designations of "on" and "off" refer to the second laser beam

(%2) being on or off.

Time dependent solutions to these equations for an

interaction time, -, of laser light can then be readily

expressed as:

njn( ) = aion ( ) nt (19)

where a," represents the degree of ionization of the atomic

system and can be expressed as:

al(T)= l-exp(-C)n k~ i,,0) (20)

Density-Matrix Formalism

The rate-equation formalism, however, is primarily valid

only for one-step excitation so an effective model for two-

step excitation is needed. Effects not taken into account are

those primarily caused by intense light fields, such as two-

photon excitation and dynamic Stark broadening, splitting, and

shifting. In order to account for these effects while keeping

the model as simple as possible, a theory for LEI based on the

density-matrix formalism was proposed by Axner et al.


The density-matrix formalism of light and matter

interactions and its theoretical assumptions have been

discussed in detail in many sources previously

[30,34,35,36,37,38,39,40,41,42,43]. Therefore, only a brief

overview will be given here.

Assuming that the laser frequency profile is Lorentzian,

the system of density matrix equations can be solved under

steady-state conditions when the time dependencies of the

level populations are neglected. The steady-state

simplification is valid if the pulse duration is substantially

longer than the inverse of the deexcitation collision rate.

The following expressions result for the fractions of atoms

excited, C2" and C3DM (where D denotes the Density Matrix

formalism), for the density-matrix formalism:

CD_ n2 1 R12(R23+k31+k32) + (R13 (R12+R23+k32) ( )
C, an (21) D

Ca = nt- -- (22
C3 3 (22)
ntom 3

where the denominator, 0, is given by:

S= R12R23+R13 (Ri2+R,) +2/3R12 (k31+k32)
+V/R23 (k21+k31) +1/R13 (2k21+k32) +1/k21 (k3l+k32) (23)

and where R12, R23, and Ri, are the excitation rates for the

atoms given by:

1 2 2 1 1 2 ) + 3( ( 2(
O ) 4 A23A 13 (24) 23)2

2 23 )2 (0 23 2 + 1 2 2)
R 23 2 m D R (2)

0) 1A 2 1 2 ( ) 2 3)2 )

R 2( R ) (25)
R 3= 2 Im (26)

D = 4A,2A23Aa3 (R2)2A -( 232A23 (27)

where Im() denotes the imaginary component, aR12 and OR23 are

the Rabi flop frequencies and A12, A23, and A13 are the complex

detunings given by:

R'= [ (As,12 312) / (2ihc) ]" (28)

)R"= [ (A32X233123) / (2ihc) ]" (29)

A12=)1,2-ni2- i722 (30)

A23=23 -- 23- 732 (31)

Al3= (0)2+023) ( 12+f23) IY31 (32)

where 012 and 0)3 are the energy between the levels in angular

frequency units, i12 and 023 are the angular frequencies

(rad/s) of the laser light related to the wavelength by

f=2tc/A, and y12, 713, and Y23 are the "off-diagonal" decay rates

between levels 1 and 2, 1 and 3,and 2 and 3, respectively,

given by:

Y21=k21+Yc+YL (33)

Y31=B (k31+k32) +Yc+2yL (34)

Y32=S ( k21+k31+k32) +Yc+YL (35)

where k21, k31, and k32 are the inelastic collision rates

between the levels, Yc is the elastic collision rate, and yL is

the half-width half-maximum laser bandwidths in angular

frequency units.

These equations are analyzed in more detail by Axner and

Ljungberg [36]. Some of the main properties of these

equations are briefly discussed below.

Computer simulations based on this theory reproduce the

main features of experimental curves fairly well, although

exact lineshapes and peak heights are not always

satisfactorily reproduced [5]. In other words, there is a

qualitative agreement between experiment and theory in

predicting the LEI signal lineshapes. However, the

quantitative agreement is still unsatisfactory and a

refinement of the theory is required. Several studies have

been published in the areas where the theory is not

satisfactory [31,32,33,44,45,46], such as the description of


the properties of the laser light and the collisional

broadening and ionization mechanisms. Consequently, these

will not be discussed here.

It should be noted, however, that many assumptions were

made in solving the density-matrix equations. Some of these

approximations included: neglect of the mode structure of the

laser light, assuming Lorentzian-shaped wings for the laser

light, reduction of the system to a three-level non-

degenerative system, and neglect of Doppler broadening,

depletion, and the influence of other non-laser-connected

levels. Consequently, the above theoretical model will not

adequately describe experimental results in certain situations

where these assumptions do not hold. However, in general, the

theoretical model is more than adequate in most analytical


Detection of the Ionization Signal

As has been noted by other authors [27,47], a good theory

for LEI signal detection in the air/acetylene flame has

remained largely underdeveloped. Therefore, a theory of LEI

signal detection and optimization will not be presented in

detail. A brief overview of the concepts and simple

approaches to modeling the detection of natural and laser-

produced ions and electrons will be given.

One-Dimensional Approximation

A one-dimensional model of the distributions and dynamics

of ions, electrons, and fields in flames has been described by

Travis and Turk [27]. In this model, the only axis

accommodated mathematically is perpendicular to the plane of

the electrodes assuming that plate electrodes are used. Other

assumptions include: the flame uniformly fills the space

between the plates and that both the flame and laser light

extend infinitely in the directions parallel to the plates

(this results in a laser beam shape of a plane rather than a

more realistic line).

Point Charae Model

The point charge model can be used to describe the LEI

signal when pulsed laser excitation is used. This theory is

described in detail elsewhere [48,49,50]. This model assumes

that pulsed laser excitation instantaneously deposits charges

(ions and electrons) in the flame. Travis and Turk [27] have

written a computer program to numerically model the LEI signal

by convolving the point charge model with charge distribution


The simulated current vs voltage curves and experimental

results for the air/acetylene flame (with the immersed

electrode configuration) are not in good agreement. The


reasons for this have been discussed by Travis and Turk [27]

and will not be presented here.

Therefore, there still exists a need for development of

an adequate model for LEI signal detection in the

air/acetylene flame and with the immersed electrode

configuration. Many opportunities exist for the extension of

the existing theory to higher dimensionalities and to

different geometries.

Electrothermal Vaporization

Electrothermal vaporizers have been studied extensively

for sample introduction into the inductively-coupled plasma

(ICP) and other plasmas. A detailed theory for electrothermal

vaporization and vapor transport has been described by others

[51,52,53,54]. A brief discussion will be presented here.

The production of dry aerosols by high temperature

processes, such as in the graphite furnace, is known as

thermal dispersion. Once the vapor is produced, condensation

of the vapor is needed for effective transport to the

observation well. This can be attained by physical

condensation of the vapor, vapor condensation on foreign

particles, or chemical condensation of the vapor.

Physical condensation is the attainment of

supersaturation by cooling the vapor or vapor-gas mixture

[52]. The simplest method of effective cooling is the mixing

of vapor with a turbulent stream of cold gas. When

condensation nuclei are generated by the vapor itself, the

process is known as homogeneous nucleation or self-nucleation

[55]. When a high concentration of stable nuclei is attained,

condensation of vapor takes place on existing particles which

is called heterogeneous condensation [55]. When a high

concentration of fine particles is reached, their growth is

governed by Brownian coagulation. Coagulation occurs when two

particles collide and adhere to form a single particle.

The condensation of the vapor of a volatile analyte on

the stable nuclei formed from another substance in the furnace

is likely to take place under ETV conditions. If adequate

mixing is achieved, a component of low volatility can increase

the transport efficiency of a volatile analyte, even if the

evaporated masses of the two components are equally small.

Condensation of the analyte vapor can also take place on

stable nuclei produced from organic substances as solid sample

matrices or additives. In general, the condensation of

analyte vapor on foreign particles of different origin is

expected to be one of the most important processes in the ETV

methods. It is also expected that this process enhances the

transport efficiency of the analyte. In other words, a

carrier effect will result.

Chemical condensation takes place when the vapor

undergoes a reaction by which a compound is formed which is

less volatile than the vaporized substance [40]. A good

example of this is the formation of a metal oxide. Therefore,

it should be kept in mind that the chemical form of the

analyte vapor as released initially is not necessarily

relevant from the point of view of aerosol formation.

Absolute Analysis

As stated in the introduction, for a method to be

considered absolute, the following conditions should be met:

(i) elimination of analytical matrix interference, (ii)

stabilization of calibration over time, and (iii) theoretical

calculation of the calibration function based on fundamental

parameters and actual measurement conditions [2]. The third

condition, a theoretical expression for the LEI signal, will

be discussed below.

The integrated LEI signal, S, in coulombs for our flame-

LEI system with ETV (graphite furnace) sample introduction is

given by:

96, 4 85 W (3
S =h m EyEpE (36)

where W, is the mass (g) of analyte in the sample, A, is the

atomic weight of the analyte (g mol-1), e is the vaporization

efficiency of the graphite furnace, s, is the transport

efficiency between the furnace and the flame, es is the

atomization efficiency (or the free atom fraction, 0) of the

analyte in the flame, e, is the laser beam probing efficiency,

and ed is the detection efficiency (efficiencies are all

dimensionless). All of these efficiencies must be measured

before considering the possibility of using this technique as

an absolute method. Ideally these efficiencies should all be

unity for absolute analysis, because, if they are not, there

is a greater chance that they will vary from sample to sample.

Also, the sensitivity of the method will be reduced if the

efficiencies are not unity.

The transport efficiency, ET, describes how effectively

the analyte vaporized in the furnace is transported to the

flame. This value must be experimentally determined, as it is

a characteristic of the experimental setup. The probing

efficiency of the laser beams, pE, is a product of the spatial

probing efficiency, E,, and the temporal probing efficiency,

Et. The spatial probing efficiency describes what portion of

the flame the laser beams encompass. If the lasers encompass

the entire flame, then E, = 1. The temporal probing

efficiency is given by:

E, =(37)

where Db is the diameter of the laser beam (cm), f is the

frequency of the pulsed laser (Hz), and v is the flame

velocity (cm s-1). The vaporization efficiency of the furnace,

,, is assumed to be unity, provided the proper temperature

and ramp times are chosen to control the graphite furnace.

The atomization efficiency, a, describes the portion of the

sample that is actually present in the flame as free atoms

capable of being ionized. This is a function of the analyte

and the flame conditions and can be found in tables or

experimentally determined [5,56,57].

The detection efficiency, Ed, describes the charge

collection by the electrode and the ionization efficiency

induced by the laser in the flame. It is given by:

Ed = YEo (38)

where Yi is the ionization yield and !, describes the

efficiency of charge collection. The actual signal detected

in our case is the induced charge, Qi, which is related to the

total charge, QTot, by the equation:

QI= Qo V (39)

where AV is the actual potential in the flame at the point of

ionization and V is the potential applied to the electrode


[58]. The fraction AV/V describes the efficiency of charge

collection, F., in our case, is given by:

AE = (40)

The ionization yield describes the fraction of atoms

which become ionized in the flame due to the laser induced

process. According to Omenetto et al. [59], when the second

excitation transition reaches a Rydberg level from which

collisional ionization occurs very rapidly, the ionization

yield can be determined by measuring what is known as the

fluorescence dip. This parameter describes the decrease in

resonance fluorescence from the first excited level occurring

when adding the second pumping process. This second

excitation process depletes the atomic population of the first

excited level such that they do not return to the ground

state. The resonance fluorescence signal is always

proportional to the number density of excited atoms in the

first excited state [60]. As it has been shown from simple

theoretical considerations [20,48,61], the ionization yield

will approach unity when an optically saturating laser pulse

has a duration that significantly exceeds the reciprocal of

the effective ionization rate of the laser populated excited

state. However, before using the fluorescence measurement to

evaluate directly the ion yield, one must be sure that there


is no loss due to quenching collisions into a metastable

level. In the absence of collisional quenching, the

ionization yield can be calculated as:

I,, (?only) -1, addede)
Y = (41)
I2, (%only)

where I21 is the signal intensity for the resonance



Analytical Performance of Flame-LEI

Limits of Detection and Sensitivities

The primary advantage LEI offers over other analytical

methods is its very high sensitivity. With LEI, detection

limits in the sub-ppb range have been achieved for many

elements. There are several reasons for the high sensitivity

of LEI: (i) electrical detection implies an almost 100%

signal collection efficiency in a substantial part of the

flame, (ii) high ionization efficiency in the interaction

region of the flame, (iii) no background from scattered light,

and (iv) molecular formation of the analyte atoms may be of

less importance than in other methods since the sensitivity is

high enough to compensate for this.

Table 1 lists over 200 experimentally measured limits of

detection for flame-LEI of 34 elements. For the best cases,

the measured limits of detection are within a factor of three

of the theoretical detection limit. There are a number of

elements which have detection limits in the sub-ppb range.


Table 1. LEI limits of detection [62]

El. X, (nm) X2 (nm) Laser Flame Time LOD (ppb) Ref

Ag 328.068 421.094 E AA 0.075 63
Ag 328.068 546.549 Y AA 0.3 64
Ag 328.068 547.155 Y AA 0.4 65
Ag 328.068 Y AA 2 65
Ag 328.068 F AA 1.1 1 65
Al 265.248 E AA 1 3 66
Al 266.039 E AA 1 2 67
Al 308.215 F AN 1 0.2 67
Al 309.271 F AN 1 0.2 68
As 278.022 E AA 1 3000 67
As 286.044 E AA 1 50000 67
Au 242.795 479.266 Y AA 1.4 1 68
Au 242.795 Y AA 1.4 1000 69
Au 264.148 E AA 1 4 67
Au 267.595 E AA 1 1.2 67
Au 267.595 294 E AA rod 0.02 69

Au 274.825 E AA 1 200 67
Ba 270.263 E AA 1 0.6 67
Ba 307.158 F AA 1.1 0.2 66
Bi 302.464 E AA 3 45 70
Bi 306.772 F AA 1.1 2 66
Ca 272.165 E AA 1 0.4 67
Ca 300.686 F AA 1.1 0.1 66
Ca 422.673 468.527 N AA 1 0.05 71
Ca 422.673 468.527 N MA 1 0.5 72
Ca 422.673 585.745 Y AA 1.5 0.03 72

Table 1 Continued

Ca 422.673 518.89 N AA 1 0.02 72

Ca 422.673 518.89 N MA 1 0.1 72

Ca 422.673 518.89 N AA 1 0.02 73

Ca 422.673 Y AA 1.5 15 73

Ca 422.673 518.885 N HA 3 100 74
Ca 422.673 Kr cw AA 0.3 1 75
Ca 422.673 N HA 3 30000 75

Ca 422.673 N AA 1 1 72

Ca 422.673 N MA 1 10 72
Cd 228.802 466.235 Y AA 1.4 0.1 69

Cd 228.802 Y AA 1.4 100 69

Co 252.136 591.680 Y AA 1.4 0.08 69
Co 252.136 Y AA 1.4 10 69

Co 273.112 E AA 1 50 67

Co 274.046 E AA 1 25 67
Co 276.419 E AA 1 6 67

Co 281.556 E AA 1 7 67

Co 304.400 E AA 3 6 71
Co 315.878 531.678 Y AA 0.2 65

Co 315.878 534.339 Y AA 0.2 65

Co 315.878 Y AA 2 65

Co 321.915 515.405 Y AA 0.3 65
Co 321.915 Y AA 4 65
Cr 272.651 E AA 1 0.9 67

Cr 278.070 E AA 1 1.5 67
Cr 298.600 F AA 15 2 76
Cr 301.492 E AA 3 36 71

Cr 301.757 F AA 15 2 77

Table 1 Continued

Cr 427.480 529.74 E AA 0.5 77

Cs 455.531 N PBA 0.004 78

Cs 455.531 N PBA 1 0.004 79
Cs 455.531 N PBA 0.1 0.1 80

Cs 455.500 N AA 0.002 81

Cs 852.124 Diode HA 1 0.25 82

Cu 276.637 E AA 1 50 67

Cu 282.437 F AA 15 100 77

Cu 282.437 E AA 1 40 67

Cu 296.116 E AA 1 600 67

Cu 324.754 453.078 E AA rod 0.02 70

Cu 324.754 453.078 Y AA 1.4 0.07 69

Cu 324.754 Y AA 1.4 3 69

Cu 324.754 F AA 15 100 18

Cu 324.754 Y AA 1 2 83

Cu 324.754 Y AA tc 1 2 84

Cu 510.600 453.078 N HA 3 500 75

Eu 459.404 564.02 N AA 0.1 4000 84

Fe 271.902 E AA 1 0.1 67

Fe 273.358 E AA 1 2 67
Fe 273.548 E AA 1 3 67

Fe 274.698 E AA 1 30 67

Fe 278.810 E AA 1 1.5 67

Fe 281.329 E AA 1 5 67

Fe 298.357 F AA 15 4 77

Fe 302.064 E AA 3 0.12 71
Fe 302.064 F AA 15 2 77

Fe 318.490 Y AA 100 65

Table 1 Continued

Fe 319.166 Y AA 4 65

Fe 319.323 Y AA 3 65

Fe 321.440 Y AA 200 65
Fe 364.784 538.337 N HA 3 100 75

Fe 364.784 N HA 3 2000 75

Ga 265.987 E AA 1 0.1 67

Ga 271.965 E AA 1 0.04 71

Ga 287.424 F AA 15 0.07 77

Ga 287.424 E AA 1 0.06 67
Ga 294.364 E AA 3 0.06 71

Ga 294.364 F AA 15 0.1 77
Ga 417.200 Kr cw AA 0.3 60 76

In 271.026 E AA 1 0.001 67

In 271.394 E AA 1 0.008 67

In 275.388 E AA 1 0.005 67

In 293.263 E AA 1 0.03 67

In 303.936 532 Y AA 75 0.0004 85

In 303.936 786.4 Y AA 0.03 4

In 303.936 F AA 1.1 0.006 66

In 303.936 F AA tc 1 0.1 84

In 303.936 F AA 1 0.02 84

In 303.936 Y AA tc 1 0.1 86

In 303.936 Y AA 1 0.02 87
In 303.936 F AA 15 0.008 77

In 303.936 Y AA 0.007 82

In 303.936 E AA 3 0.03 71

In 410.176 Kr cw AA 0.3 20 76

In 451.131 571.0 E AA rod 0.0004 87

Table 1 Continued

In 451.131 501.8 N HA 3 0.6 75

In 451.131 501.8 N AA 0.007 82

In 451.131 502.3 N AA 0.03 82

In 451.131 525.4 N AA 0.003 82

In 451.131 526.3 Y AA 0.01 82

In 451.131 571.0 N AA 0.001 82

In 451.131 572.8 N AA 0.03 88

In 451.131 572.8 N AA 0.003 82

In 451.131 Kr cw AA 0.3 0.1 76

In 451.131 N HA 3 100 75

Ir 266.479 562.004 E AA 1 0.3 89
+ 642.0

K 294.268 F AA 15 1 77

K 296.321 E AA 1 1.5 67
K 404.414 N PBA 0.1 79

K 580.200 E AA 0.1 90
K 766.490 Kr cw HA 0.3 0.1 76

Li 274.119 E AA 1 0.005 67

Li 460.286 Kr cw AA 0.3 20 76

Li 610.362 F AA 1.1 0.01 66

Li 639.146 639.146 F AA 1.1 0.4 66

Li 670.784 460.286 E AA 0.0003 64
Li 670.784 610.362 N HA 3 0.04 75

Li 670.784 610.36 Y AA 0.1 0.03 81

Li 670.784 F AA 1.1 0.001 66

Li 670.784 N HA 3 4 75

Mg 285.213 435.2 Y AA 0.002 4
Mg 285.213 470.3 N AA 0.1 0.4 91

Mg 285.213_ F AA 15 0.1 18

Table 1 Continued

Mg 285.213 E AA 3 0.005 71

Mn 279.482 E AA 1 0.04 67

Mn 279.482 521.482 Y AA 1 0.02 92

Mn 279.482 F AA 15 0.3 18

Mn 279.827 E AA 1 0.05 67
Mn 279.984 F AA 15 5 18

Mn 280.106 E AA 1 0.08 67

Mn 292.557 E AA 1 3 67

Mn 292.557 E AA 3 3 71

Mn 403.076 602.180 N HA 3 5 75

Mn 403.076 N HA 3 30 75

Mo 267.985 F AN 1 30 68
Mo 306.428 F AN 1 400 68

Mo 307.437 F AN 1 500 68
Mo 308.562 F AN 1 500 68

Mo 311.212 F AN 1 900 68

Mo 313.259 F AN 1 70 68

Mo 315.816 F AN 1 70 68

Mo 317.035 F AN 1 20 68

Mo 319.397 F AN 1 10 68

Mo 320.883 F AN 1 50 68

Na 268.037 E AA 1 0.1 67

Na 268.046 E AA 1 0.1 67
Na 285.281 E AA 3 0.0015 71

Na 285.301 F AA 15 0.05 18

Na 540 540 Y PBA 70 79

Na 550 550 Y PBA 3 79

Na 578.732 578.732 E AA 0.001 93

Table 1 Continued

Na 578.732 578.732 Y PBA 0.9 79

Na 588.995 568.266 E AA 0.003 64

Na 588.995 568.821 N HA 3 0.04 75

Na 588.995 568.821 Y AA 0.012 94

Na 588.995 568.821 Y PBA 0.002 79

Na 588.995 568.821 N AA 0.0006 82

Na 588.995 616.075 Y PBA 0.01 79

Na 588.995 Kr cw AA 0.3 0.03 76

Na 588.995 N HA 3 6 75

Na 588.995 E AL HA 10 0.3 95

Na 588.995 E AA 0.02 94

Na 588.995 F AA 0.01 96

Na 588.995 F AA -20 97
Na 588.995 Y HA 1 0.8 98

Na 588.995 Y AA 1 0.6 99

Na 589.592 568.263 N AA 0.001 82

Na 589.592 568.26 N AA 0.1 0.005 92

Na 589.000 449 E AA rod 0.0002 70

Na 588.995 F AA 15 0.1 18

Ni 269.649 E AA 1 24 67

Ni 279.865 E AA 1 0.4 67

Ni 282.129 E AA 1 0.3 67

Ni 300.249 F AA 1.1 7 66

Ni 300.249 F AA 15 8 77

Ni 300.249 576.755 Y AA 1.4 0.08 69

Ni 300.249 Y AA 1.4 8 69

Ni 301.200 E AA 3 1.5 71

Ni 324.846 Y AA 2 65

Table 1 Continued

Pb 280.199 E AA 1 0.4 67

Pb 280.199 F AA 15 0.6 18
Pb 282.320 E AA 1 0.5 67

Pb 282.320 600.193 Y AA 75 0.0007 86
+ 1064

Pb 282.320 F AA 15 3 18

Pb 283.306 600.193 Y AA 1.4 0.09 69

Pb 283.306 600.193 E AA 0.3 64
Pb 283.306 Y AA 1.4 3 69
Pb 283.306 E AA 1 0.2 67

Pb 287.331 E AA 3 3 71

Pb 287.331 E AA 1 0.6 67

Rb 420.185 Kr cw HA 0.3 0.7 76

Rb 420.185 N PBA 0.1 79

Rb 420.185 N AA 1 0.0006 74
Rb 780.023 K HA 0.3 0.09 76

Rb 780.023 Diode HA 1 0.3 83

Sb 276.995 E AA 1 90 67

Sb 287.792 E AA 1 50 67

Si 288.158 F AN 1 40 68
Sn 266.124 E AA 1 30 67

Sn 270.651 E AA 1 8 67

Sn 270.651 F AN 1 2 68

Sn 283.999 597.028 Y HA 1.4 0.3 69

Sn 283.999 Y HA 1.4 8 69

Sn 283.999 E AA 1 2 67
Sn 283.999 F AN 1 0.4 68

Sn 283.999 F AA 15 6 77

Sn 286.333 F AN 1 2 68

Table 1 Continued

Sn 286.333 F AA 15 10 77
Sn 286.333 E AA 3 20 71

Sn 286.333 E AA 1 3 67

Sn 300.914 F AN 1 10 68
Sn 303.412 F AN 1 6 68
Sn 317.505 F AN 1 3 68
Sn 326.234 F AN 1 2 68
Sr 293.183 E AA 1 0.01 67
Sr 459.513 E AA 15 99

Sr 460.733 K AA 0.3 0.4 76
Sr 460.733 Y HA 1 3 99

Sr 460.733 Y AA 1 1 99
Sr 460.733 554.336 E AA 0.3 100
Ti 294.200 F AN 1 10 68
Ti 294.826 F AN 1 8 68
Ti 295.613 F AN 1 6 68

Ti 300.087 F AN 1 20 68

Ti 318.645 F AN 1 1 68
Ti 319.199 F AN 1 1 68
Ti 319.992 F AN 1 1 68

Ti 331.442 F AN 1 3 68

Ti 334.188 F AN 1 2 68
Ti 335.469 F AN 1 3 68
Ti 337.145 F AN 1 4 68
T1 276.787 377.572 E AA 1 0.008 100
T1 276.787 E AA 10 0.02 101
T1 276.787 E AA 1 0.006 67
T1 291.832 E AA 3 0.02 71

Table 1 Continued

T1 291.832 E AA 1 0.008 67
T1 291.832 F AA 15 0.09 77
T1 377.572 E AL HA 10 3 96

T1 377.572 655.6 Y AA 0.01 4
V 292.362 F AN 1 20 68
V 305.633 F AN 1 6 68

V 306.046 F AN 1 4 68
V 306.638 F AN 1 3 68
V 318.398 F AN 1 0.9 68
V 318.540 F AN 1 0.9 68
W 283.138 E AA 1 300 67

Yb 267.198 E AA 1 1.7 67
Yb 555.647 581.2 Y AA 0.1 82
Zn 213.856 396.545 Y AA 1 1 101
Zn 213.856 Y AA 1 3 102
Zn 307.590 472.216 Y AA 1 15 102

E=excimer pumped dye laser, E AL=excimer pumped atomic line
laser, F=flashlamp pumped dye laser, Kr cw=krypton ion
pumped cw dye laser, N=nitrogen pumped dye laser, Y=Nd:YAG
pumped dye laser, AA=acetylene/air, AN=acetylene/nitrous
oxide, HA=hydrogen/air, MA=methane/air, PBA=propane/butane/
air, rod=graphite rod in flame, tc=total consumption burner.

The lowest detection limits are obtained for elements with

good atomization in the flame and low ionization potentials

such as Li, Na, In, and Tl. This demonstrates that

collisional ionization is very efficient from excited states

close to the ionization limit. The low limits of detection

obtained for one-step LEI for some of the other elements with

comparatively higher ionization limits, such as Mg and Cd,

suggest that alternative ionizing routes may exist.

Unfortunately, in many cases, the measured detection

limits are much worse than the theoretical values [63]. This

is due to a variety of reasons: poor atomization fractions,

high contamination levels in blanks, radio frequency

interference, low repetition rate lasers, high thermal

ionization fractions, and the use of non-optimum excitation


Noise and Interferences

LEI detection limits are usually limited by noise or

spectral interference during the measurement. Sources of

noise can be separated into two categories: multiplicative

and additive. Multiplicative noises in LEI arise from the

fluctuations in atomic population, fluctuations in the

ionization yield, and fluctuations in the detection

efficiency. Sources of additive noise include fluctuations in


the thermal background ionization, fluctuations in the laser-

induced background ionization, and electronic noise.

Fluctuations of the atomic population within the

irradiated volume of the flame result from fluctuations in the

nebulization rate and in the flame gas flows. Fluctuations in

the ionization yield are a result of changes in the laser

output properties, such as the pulse-to-pulse power variation

in the dye laser output. A 4% RSD is typical but may be worse

in some circumstances. This is further complicated by the

variation in laser power across the beam profile, where the

power is higher at the center of the beam than at the edges.

However, this problem may be minimized by saturation of the

atomic transitions. Fluctuations in the temperature of the

atom reservoir also contribute to fluctuations in the

ionization yield.

Fluctuations in the detection efficiency can result from

fluctuations in the high voltage power source for the

electrode, variations in the flame composition, and spatial

fluctuations of the laser beam.

As the concentration of the analyte decreases,

multiplicative noise decreases; however, additive noise

remains even in the absence of analyte. Therefore, it is

usually the additive noises that ultimately limit the

detection capability of the system.

Fluctuations in thermal ionization additive noise are a


result of fluctuations in the number of natural flame ions and

sample matrix ions. These fluctuations may be a result of the

flame flow fluctuations and nebulizer-induced noise.

Electronic additive noise results from the noise of the

various electronic components used to measure the LEI current.

Of the detection electronics used, the current preamplifier is

the noisiest. Another source of electronic noise that may be

significant is radio frequency (rf) noise. The LEI electrodes

and preamplifier seem to act as an excellent antenna and

detector for rf noise, so care should be taken to shield and

ground the LEI instrumentation.

Random fluctuations in the laser-induced background may

result from laser-induced ionization of spectral

interference. Spectral interference can be caused by any

matrix element but are most often a result of easily ionized

elements. Also, spectral interference may result from the

overlapping of atomic lines but are usually a result of an

overlap between the analyte line and some broadband spectral

feature of a matrix component.

Line overlaps are rare and are easy to eliminate when

two-step LEI is used. However, one disadvantage of using two-

step excitation is that interference could occur at one or

both wavelengths.

Overlaps between analyte lines and broadband spectral

features of matrix constituents are, again, much more likely


to be encountered than direct spectral line overlaps. Such

interference include line wings, molecular bands, and

thermionic ionization of particles. The wings of atomic lines

are easily observed in laser spectroscopy, including LEI. An

example of this is the line wing interference from Na on the

determination of Ni by two-step (300.249 nm and 561.479 nm)

LEI [63]. The most common molecular band interference

encountered in LEI is due to LEI of CaOH [102]. The

prevalence of Ca in many sample matrices, the incomplete

dissociation of CaOH in the air/acetylene flame, the rather

low ionization potential of CaOH (5.7 eV), the broad spectrum

from green to red wavelengths, and the location of many

second-step LEI stepwise excitation lines in this wavelength

range combine to make this a common problem. Laser-induced

particle ionization, may occur when a fuel-rich flame is being

used or when certain organic solvents are aspirated. The

mechanism for this is thought to be thermionic in nature

[103]. It may be possible to correct for these by scanning

the laser wavelength across the analyte line and performing

the appropriate background correction [104]. Wavelength

modulation has also been used to deal with these interference


Applications of LEI to Real Samples


LEI is one of the most sensitive analytical methods for

trace element analysis. Unfortunately, LEI has found limited

applications to real samples because of its susceptibility to

easily ionized matrix elements. The inherent ease of

collecting and sensing ions that contribute to the simplicity

of the LEI detection scheme also makes it vulnerable to these

easily ionizable elements (EIEs) [106]. So, although the

laser affords a good amount of selectivity, it cannot

compensate for an undiscriminating detector.

Sample dilution was the first solution to matrix

interference. It was often possible to dilute sufficiently

the sample matrix and still detect the analyte because of the

high sensitivity afforded by LEI. The use of an immersed

electrode also helped to reduce the loss of LEI signal due to

ion collection interference [107], but did not reduce the dc

background current from EIEs.

Today, approaches to analyze real samples by LEI can be

categorized as involving interferant removal or those without

interferant removal.

Determinations Without Interferant Removal

LEI is particularly amenable to samples of high purity

with small amounts of EIEs. In these cases, little

accommodation for interference is necessary. Alloy samples

are particularly well suited to LEI because they typically

contain low levels of sodium and potassium. The determination

of indium in nickel-based, high-temperature alloys [22] is an

early example of application of LEI spectrometry to a

difficult analytical problem. An acetylene/air flame on a

slot burner and plate electrodes produced satisfactory results

because of the low levels of EIEs. Similar samples usually

require a time-consuming extraction before conventional

furnace atomic absorption analysis, in contrast to LEI where

the alloy samples were successfully analyzed without sample

preparation. The results were also in close agreement with

values obtained with furnace atomic absorption.

Lowering the temperature of the atom reservoir is also a

potential solution for analytes with low atomization

temperatures such as cesium. Using a solid stainless-steel

rod immersed in a low temperature propane/butane/air flame,

researchers were able to determine accurately low

concentrations (ng/mL) of cesium in tap water samples by LEI

even with tens of mg/mL of sodium, potassium, and calcium

present [80].

Natural water samples are also ideal for LEI. The

concentrations of several elements at pg/mL levels were

validated in a simulated rainwater Standard Reference Material

(SRM 2694) by researchers at NBS (now NIST) using LEI

spectroscopy [73]. LEI was one of the unrelated methods used

to certify the concentration of the standards at NBS. Some


spectral interference due to excitation in the wings of

nearby peaks were corrected by standardization using matrix

matched standards.

Two-step excitation has been used to determine zinc in

SRM 1643a, trace elements in water, in the presence of a

background interference [102]. The experimental value for

zinc was slightly high, but no attempt was made to remove

potential interference beyond using an immersed electrode and

sample dilution. It was felt that matrix matching of the

standards would have improved the accuracy of the measurement.

As part of an environmental monitoring program, lead was

determined in unpolluted waters from mountainous regions and

compared with results for natural waters impacted by

industrial development [108]. Many spectroscopic techniques

do not have adequate sensitivity to determine species which

are naturally present at very low background levels. These

pristine waters presented little difficulty because of the

very low levels of impurities. In the case of water impacted

by industry, the concentrations of calcium, potassium, sodium,

and magnesium impurities were 4-5 orders of magnitude larger

than the lead concentrations and produced broad background

signals. It was found that CaOH molecules were responsible

for the interference at both excitation wavelengths but, by

tuning off the resonance lines, it was possible to use

background subtraction successfully.


Several elements have been determined in rock samples by

LEI spectrometry [109]. Most of the other analytical methods

require the use of complicated procedures prior to analysis

unless the sample is preconcentrated or interference are

removed. However, for LEI the dissolved samples were

aspirated into a propane/butane/air flame with an immersed

electrode used for detection. Although a broad ionization

background was found (due to CaOH), by reduction of the laser

powers used, good agreement with certified values was obtained

using aqueous standards. Detection limits of 0.002, 0.001,

and 0.5 ig/g were obtained for cesium, lithium and rubidium,


LEI spectrometry has also been demonstrated as a viable

approach for detecting dopants and impurities in acid-

dissolved bulk gallium arsenide [110]. By using a two-step

excitation scheme, background subtraction was possible. Trace

amounts of chromium, iron, nickel, indium, manganese, and

cobalt were detected. Two-step LEI has also been used to

determine sodium in semiconductor silicon [111].

LEI has been used for determination of trace amounts of

nickel in petroleum products because nickel poisons the

catalysts used in petroleum processing [112]. Samples of both

heavy-oil flash distillate and an oil-based SRM were diluted

with a xylene/n-butanol solvent mixture and aspirated into an


air/acetylene flame. Nickel determination in the SRM was in

good agreement with the NIST certified value. Because of the

high sensitivity of LEI, it was possible to dilute the samples

considerably which nearly eliminated the need for matrix

matching of the standards.

Determination of indium in a CdHgTe alloy was

accomplished in both liquid solutions and solid samples

without sample preparation [70,113]. Electrothermal

atomization was coupled with LEI spectrometry by inserting a

resistively heated graphite rod in a premixed flame of a slot

burner. Propane/butane/air and acetylene/air flames were

used. No matrix interference was found for the samples and

aqueous standards were used for calibration. A good

correlation between results for liquid and solid samples

indicated analytical accuracy and an absence of analyte losses

for solid sampling.

Determinations With Interferant Removal

Preionization has been used for removal of spectral

interference and is described in more detail elsewhere [114].

Magnesium was chosen as the analyte since it is very

susceptible to interference from sodium (atomic wing

absorption). Several preionization schemes were investigated

using up to three photons of different energies. Up to an 83%

sodium depletion in the flame was achieved. A probe laser

then interrogated the preionized "hole" with 285 nm photons to

enhance thermal ionization of the analyte. Although

satisfactory results were achieved, the technique will

probably not be widely utilized because of the cost involved

for the two separate laser systems, the complexity in timing

the arrival of the ionization laser beam(s) and the probe

laser beam, and because signal collection interference are

related to the bulk flame environment and are not relieved by

laser preionization.

Chekalin and others determined copper and sodium in

concentrated orthophosphoric acid using their rod-flame system

[70,114]. The sodium interferant was removed by selective

volatilization from the dried sample at 1000"C. When the

temperature was raised to 2000'C, the copper signal could be

detected in the absence of noise. Detection limits were

determined by the purity of the rod material.

The determination of lead in a blood matrix has also been

reported [1151. A graphite furnace, used for sample

vaporization, was coupled with an acetylene/air miniature

flame for the analysis. With only a 21:1 dilution using ultra

pure water and temperature programming of the graphite

furnace, a detection limit of 0.089 ng/mL (890 fg absolute)

for lead in whole blood was obtained.

Solvent extraction has been shown to be effective for the

determination of trace amounts of manganese using a single-


step excitation scheme [116]. Manganese was completed in

water with sodium diethyldithiocarbamate and extracted into

diisobuytl ketone. The extraction resulted in a 10-fold

increase in the concentration of the manganese as well as

interferant removal. This method was successfully applied to

the analysis of ng/mL of manganese in groundwater, river and

lake waters, seawater, tap water, and wastewater.

An extraction also made the determination of 0.001%

calcium in aluminum alloys possible [72]. The separation was

based on the different solubilities of calcium and aluminum

chlorides in methanol. The detection limit was determined by

the purity of the methanol used.

Chromatography using a chelating resin to separate the

interfering elements from the analyte was used to determine

trace amounts of copper in a sulfate plating solution and

seawater [65]. The removal of EIEs was accomplished using

Chelex 100 [117]. In this process, transition and heavy

metals were chelated in the 5.2-5.6 pH range while ammonium

acetate was used to selectively elute the alkali and alkaline

earth metals by ion exchange. In the final step, the trace

metals were eluted with nitric acid and introduced into a

burner for LEI spectrometry. A microsampling cup coupled to

a premix burner (acetylene/air) permitted absolute

determinations of copper as low as 50 pg. Silver, cobalt,

iron, and nickel were also detected.


Turk and Kingston have combined automated chelation

chromatography with computer-controlled LEI spectrometry to

determine a large number of elements in a wide range of NIST

SRMs [105]. Chelex 100 resin was used for the separation

which was automated with a laboratory robot after preliminary

work. The elements determined and the reference materials

analyzed included the following: Cd, Co, Cu, Mn, Ni, and Pb

in Trace Elements in Water (SRM 1643b), Mn and Ni in Inorganic

Constituents in Bovine Serum (SRM 1598), Ni and Pb in Buffalo

River Sediment (SRM 2704), Cu, Mn, and Ni in Total Diet (SRM

1548), and Mn and Ni in Apple Leaves (SRM 1512' and in Peach

Leaves (SRM 1547). Concentrations determined ranged from the

mg/g to the ng/g range while precisions were in the range from

0.8% to 36% RSD. It was hoped that this impressive display of

technology would go a long way towards establishing LEI

spectrometry as a practical analytical method.

Alkyltins in sediment were determined by ion-exchange

chromatography coupled with LEI detection [106,118].

Tributyltin was extracted into 1-butanol and two-step (284.0

nm and 603.8 nm) LEI was performed in an acetylene/air flame.

The detection limit determined with the LEI detector was 3

ng/mL tin as tributyltin or 60 pg of tin.

The coupling of liquid chromatography with LEI has also

been reported for the measurement of organolead species [119].

Two Nd:YAG pumped dye lasers were used to optically excite


lead at 283.31 nm and 600.19 nm. A reversed phase LC column

was attached to the LEI acetylene/air flame. A detection

limit of 0.9 ng/mL (20 pg Pb for 20 pL injection) for

tetraethyllead was calculated. Oyster tissue samples (SRM

1566a) were analyzed. Trace levels of trimethyllead were

observed in the Oyster tissue, but concentrations varied among

the samples tested. This seemed to indicate that the

extraction and digestion procedures used were inadequate.

The use of flow injection analysis to reduce the

electrical interference from a sodium matrix for LEI has been

reported [120]. A frequency doubled Nd:YAG laser provided

optical excitation at 325.62 nm for LEI of indium. A typical

slot burner and nebulizer were used. A flow injection

apparatus was used to handle the solution prior to the

nebulizer. The standard addition method was also used with

this system to recover the original In concentration. This

combination of flow injection analysis and LEI was capable of

detecting In in a Na matrix of over 40 ppm, which is about 20-

fold more than the conventional LEI apparatus could tolerate

alone. This system also exhibited a larger linear dynamic

range for In, which was extended to 30 ppm with a mixture of

8 ppm Na matrix, up from 5 ppm for the conventional LEI


Although the future for applications of LEI spectrometry

looks good, for LEI to become more widespread will require

continued evolution of hardware and software to accommodate

routine analysis by LEI. Coupling of LEI to other techniques

has also enhanced its ability to handle real samples. It has

been suggested that perhaps a multi-capability "laser

spectrometer" may be considered more promising for commercial

LEI instrumentation [121].

Hybrid Techniques and Non-Flame Atom Reservoirs

Although the flame is the most commonly used atom

reservoir for trace element analysis by LEI, the flame

atomizer alone has some serious drawbacks for trace element

analysis: (i) dilution of sample vapors by flame gas

combustion products, (ii) limited range of temperatures used,

(iii) a small (0.1-0.15) sample utilization factor (fraction

of the sample that reaches the flame), (iv) the impossibility

of separating the processes of sample evaporation and

atomization, (v) problems associated with handling

microvolumes of liquids and solid samples, and (vi) combustion

products of flames may hinder some of the spectral regions for

successful implementation of LEI [122]. These limitations

have prompted consideration for hybrid techniques and

alternate atom reservoirs for LEI spectrometry.

Electrothermal Vaoorizers


The first attempts to use a graphite furnace for LEI

determinations of elements failed [108]. Later, Gonchakov et

al. reported the successful application of electrothermal

atomization for determination of small amounts of sodium using

a three-step ionization scheme [123]. A graphite cup in an

argon atmosphere was used as an atomizer and a tungsten loop,

positioned 2 cm above the graphite cup, was used as an

electrode. A detection limit of 1 fg was calculated for Na.

Torres used a Varian-Techtron CRA-90 electrothermal tube

atomizer for LEI spectrometry [124]. The electrode and

graphite tube were positioned end-to-end on the same axis with

the laser beams passing through the graphite tube. A

detection limit of 5 pg was achieved for Cs. However, other

metals could not be determined because of thermionic emission

at higher temperatures and also because of arcing between the

electrode and graphite tube.

Graphite furnace-LEI was also used to determine the

presence of sodium and indium at the fg level [82]. A

tungsten wire was place axially inside the graphite tube. The

LEI signal was found to be 10-100 times larger than in a

flame, but the reproducibility was poor. Arcing between the

electrode and graphite tube also occurred when the temperature

was increased.

The most detailed investigations of LEI in the graphite


furnace were made by Magnusson and others [125,126,127].

However, they were faced with the same problems encountered in

the previously mentioned studies. Thermionic emission from

the graphite tube and electrode made it impossible to detect

elements which atomized at high temperatures. Although the

sampling efficiency (fraction of sample introduced that

reaches the probing area) was 2-3 orders of magnitude higher

than that in a flame, the limits of detection were on the same

order of magnitude. This was partially due to the lower rate

of collisional ionization in the argon atmosphere of the

graphite furnace.

A T-shaped furnace was suggested by Magnusson and

demonstrated by Sjdstrom to solve the thermionic emission

problem by spatially separating the regions of atomization and

detection [128]. A flow of argon was used to transport the

atoms to an external cavity adjacent to the tube, in which

laser excitation and LEI detection were performed. Detection

limits in the pg range were obtained for manganese and

strontium. However, this system had some drawbacks. As a

result of the temperature gradient between the center of the

graphite tube and the detection region, relatively few of the

atoms in the sample actually reached the detection region.

The use of modern furnace technology with probe

atomization for LEI has been described by Butcher et al.


[129]. In this system, a graphite probe was used for both

sample introduction and as the high-voltage electrode (-50 V).

The sample was vaporized off of the probe into an already

isothermal, furnace environment, which reduced the matrix

interference mentioned above and prevented atom condensation

onto the probe. For elements such as Tl, In and Li, detection

limits were between 0.7 and 2 pg. For Pb, Mg, and Fe,

detection limits were between 10 and 60 pg. The linear

dynamic range was between 3 and 4 orders of magnitude with a

precision between 12 and 16% for aqueous solutions. Sodium

matrix effects were also investigated and found to suppress

the LEI signal in the same manner as in the flame, so similar

difficulties would be encountered in the graphite furnace with

complex matrices as in the flame.

A novel design for a graphite furnace atomizer-ionizer

was examined by Chekalin and Vlasov [70]. Their outer

electrode design along with careful selection of the applied

electrode voltage suppressed interference from thermionic

emission of the heated graphite tube. Detection limits of

0.08 pg/mL for In and 100 pg/mL for Yb were obtained. This

system seems very promising for analysis of high purity


Hybrid Combinations of Flame and Electrothermal Vaporizers

Hybrid combinations of the flame and electrothermal


vaporizer were developed in order to alleviate the problems

associated with LEI in the graphite furnace alone, which were

mentioned above. The combination of flame and electrothermal

vaporizer/atomizer was first proposed by Chaplygin et al.

[130]. Their system consisted of a specially designed

cylindrical burner which contained an electrically heated wire

loop or filament in the central channel. An argon flow up the

central channel carried the sample vaporized off the wire loop

up into the flame. This design helped to minimize the sample

matrix by using small injection volumes and by temperature

programming of the sample vaporization loop. Detection in the

flame eliminated interference from thermionic emission from

heating of the sample loop and also maximized the LEI signal

because the flame region irradiated by the laser beams was

located much higher than the combustion zone. A detection

limit of 0.5 pg for Cs with reproducibility of 5-6% was


Miyazaki and Tao have reported the use of a commercial

electrothermal vaporizer as a sample introduction system for

LEI spectrometry [131]. The commercial electrothermal

vaporizer (ETV) used (Seiko Instruments Inc., Tokyo) consisted

of a tungsten boat and a glass chamber. A Nd:YAG pumped dye

laser provided optical excitation at 276.79 nm and 291.83 nm

for one-step LEI of thallium. EIEs were found to interfere at


a 10-fold excess or more. Tl was extracted into 2,6-dimethyl-

4-heptanone (DIBK) with ammonium tetramethylene

dithiocarbamate (APDC) and hexamethyleneammonium

hexamethylenedithiocarbamate (HMAHMDC) at pH 6 to remove EIE

interference. Detection limits of 0.043 ng/mL (276.79 nm)

for Tl with the extraction and 0.11 ng/mL (291.83 nm) for Tl

without the extraction were obtained. This system was then

applied to and found to be useful for the analysis of

different types of natural water samples.

In order to acquire the advantages of both graphite

furnace vaporization and flame ionization detection while

maintaining independent control of each process, Smith et al.

have reported the coupling of a graphite furnace with a

miniature acetylene/air flame for LEI detection [132]. The

graphite furnace was used to vaporize samples which were

transported to the flame by a flow of argon gas. Two Nd:YAG

pumped dye lasers provided optical excitation at 285.2 and

435.2 nm for Mg, 377.6 and 655.6 nm for Tl, and 303.9 and

786.4 nm for In. Limits of detection of 0.0017 ng/mL (17 fg

absolute) for Mg, 0.012 ng/mL (118 fg absolute) for Tl, and

0.026 ng/mL (260 fg absolute) for In were obtained.

Riter et al. have reported the use of a modification of

the above system for trace element analysis [133]. A

commercially available graphite furnace (Finnigan MAT/SOLA,


Bremen, Germany) was used to vaporize samples which were

subsequently transported to a redesigned miniature burner,

which supported an acetylene/air flame, by a flow of argon

gas. Two XeCl excimer pumped dye lasers were used to provide

optical excitation at 285.213 and 435.191 nm for the analysis

of Mg. A complete determination of the instrumental

characteristics of the above system for the analysis of Mg was

performed. A blank-limited detection limit of 2 ng/mL (20 pg

absolute) was obtained for Mg. A detection limit of 590 fg/mL

(5.9 fg absolute) was calculated in the absence of a blank

signal and a reduction of the radio frequency noise.

Riter et al. also evaluated the above system for the

determination of trace Pb concentrations in whole blood [116].

Optical excitation at 283.3 and 509.0 nm was provided by two

XeCl excimer pumped dye lasers. Bovine blood samples from the

Centers for Disease Control (CDC) and NIST (SRM 955a) were

analyzed. With temperature programming of the graphite

furnace, there appeared to be no interference from the blood

matrix, and, instead, the matrix appeared to produce a carrier

effect, increasing the transfer efficiency between the furnace

and the flame over that for aqueous standards. The authors

concluded that there was sufficient matrix removal from

temperature programming of the graphite furnace and that the

matrix remaining acted as a carrier. A detection limit of


0.089 ng/mL (890 fg absolute) was calculated for Pb in whole

blood. This technique appears to have great promise for the

analysis of samples with complex matrices.

Marunkov reported the first LEI experiments with sample

vapors being introduced into a flame by an electrically heated

graphite rod and by diffusion through the wall of an

electrically heated closed graphite tube [134]. Detailed

experiments were not carried out, but results appeared to be


A hybrid "rod-flame" arrangement was proposed by Chekalin

et al. to combine the advantages of flame and electrothermal

atomizers [70,88,114,135]. In their system, the sample is

evaporated by an electrically heated graphite rod into the

flame where the analyte is atomized, laser excited, ionized,

and then detected. Advantages of this system included an

increase of the sample utilization factor (fraction of the

sample introduced that reaches the probing region), the

analysis of microsamples, and the direct analysis of solid

samples. However, the researchers encountered problems with

nonselective background ionization from the compounds

evaporating from the heated graphite rod and impurities in the

graphite were contributing to a blank signal. Good detection

limits were obtained for many elements including Au (2 pg/mL),

Co (100 pg/mL), Cr (20 pg/mL), Cu (2 pg/mL), In (0.04 pg/mL),


Mn (30 pg/mL), Na (0.02 pg/mL), and Ni (8 pg/mL). The

analysis of high-purity substances, such as Na and Cu in

orthophosphoric acid, Cu in germanium, In in Cd-Hg-Te alloy,

and Cr, Co, Mn, and Ni in fluorine-containing materials for

optical fibers, was also demonstrated. Detection limits

ranged from 0.1 to 7,000 ng/g for the different elements.

LEI in the Inductively Coupled Plasma

The first measurements of LEI in the ICP were made by

Turk and Watters [136]. Resonant LEI was detected for Fe, Mn,

Na, and Cu, however, the population of free atoms was too

small so the sensitivity was very poor. Turk et al. used a

power modulated ICP to reduce the rf interference from the

plasma [137]. With their approach, detection limits of 80

ng/mL for Fe and 20 ng/mL for Ga were achieved which are still

poorer than for the flame.

Ng et al. reported significant improvements in the

detection limits by using an extended-torch ICP, modifying the

torch and electrode designs, and using a continuous wave laser

[138]. The smaller electrode dimensions and smaller

separation between electrodes used by Ng et al. probably

accounted for the 2 orders of improvement in the detection

limit over Turk and Watters. Limits of detection ranged from

30 ng/mL for Ca to 810 ng/mL for Sr. Although these are an

improvement, they are still worse than the detection limits

for flame and furnace LEI.

The coupling of ICP and flame LEI with mass spectrometric

detection has been reported by Turk and others [139,140,141].

With a modified commercial ICP-mass spectrometer (ICP-MS), Sr

was examined employing laser-induced ionization (460.733 nm

and 308 nm). An enhancement of only 11% in the Sr' ion signal

was observed with the addition of laser excitation. The flame

was found to offer a much better environment for laser-induced

ionization or LEI than the ICP. A hydrogen/air flame was used

for Na, K, and Fe while an acetylene/air flame was used for

Ca. A 350 times increase in the signal was observed for Na'

with the addition of two-step optical excitation (589.0 nm and

498.3 nm). A detection limit of 0.05 ng/mL for Na was

calculated. Detection limits of 0.14 ng/mL for K (766.5 nm

and 580.2 nm), 35 ng/mL for Fe (302.1 nm), and 61 ng/mL for Ca

(422.7 nm and 585.7 nm) were obtained. The poor detection

limit for iron was believed to be due to one-step excitation,

and the detection limit for Ca was thought to be due to the

poor atomization efficiency of Ca in the acetylene/air flame.

Other Methods and Reservoirs

Gorbatenko et al. have reported the use of a Nd:YAG laser

as a laser microprobe solid sampling device for flame LEI

[142]. The sample was positioned at the edge of the burner

head to allow for the direct transport of the sample vapors to

the combustion zone. Lithium was chosen as the analyte and

was ablated from an aluminum alloy sample. Optical excitation

was provided by two Nd:YAG pumped dye lasers at 670.8 nm and

610.4 nm. It was found that the atomization efficiency was

determined by flame temperature and composition regardless of

how the sample was introduced into the flame. It was also

found that this technique allowed for the study of the

distribution of Li impurities over the surface of a solid

sample with a spatial resolution of about 100 pm. A detection

limit of around 30 ug/g for Li in solid samples was achieved

with a very short (s 20 s) analysis time.

Churchwell et al. have investigated the atmospheric-

pressure microarc atomizer as an atom reservoir for LEI [143].

Microvolumes of analyte solution were deposited on the tip of

the tungsten cathode loop and dried with a heat gun. LEI

measurements were performed directly in the plasma above the

microarc discharge. A detection limit of 3 ng for Na was

estimated. Preliminary results indicate that the helium-

microarc-induced plasma may be feasible for LEI spectrometry

but further studies are needed.

From the above examples, it can be seen that alternate

atom reservoirs and hybrid techniques have been used with

varying degrees of success for LEI spectrometry. The flame

still appears to be the optimal atom reservoir for LEI.


However, the hybrid techniques appear to be the most promising

of those discussed, especially for dealing with the matrix

interference from real samples. Further research on their

application to samples with complex matrices is warranted.



A block diagram of the experimental setup used for

graphite furnace-LEI is shown in Figure 6. An excimer laser

(Model LPX-240i, Lambda Physik, Acton, MA), operated with

XeCl, was used to pump two dye lasers (Model Scanmate 1,

Lambda Physik, Acton, MA). The output of the dye lasers was

directed into an air/acetylene flame for LEI spectrometry. A

repetition rate of 30 Hz was used for all LEI experiments.


Two different burners (both designed in this laboratory)

were examined in this work. The first burner design (Figure

7) consisted of a Teflon base and a 2 7/8 in. diameter brass

top. A flow of argon carried the vaporized sample up through

a central stainless-steel capillary (o.d.=1/8 in, i.d.=1/16

in) where the sample was injected up into the flame. Premixed

air and acetylene flowed up in stainless-steel capillaries

(o.d.=3/32 in, i.d.=1/32 in) surrounding the central capillary

and were ignited to form the flame.


Figure 6. Block diagram of the experimental setup for LEIS


Transfer Line






Figure 7. First burner design used for LEIS



< Ar Sheath


S! Flame Gas o

: __ Ar Sheath
Ar + Sample

Ar +
Top View
Side View