The semiconductor laser diode as an excitation source for analytical atomic and molecular fluorescence spectroscopy


Material Information

The semiconductor laser diode as an excitation source for analytical atomic and molecular fluorescence spectroscopy
Physical Description:
vi, 87 leaves : ill. ; 28 cm.
Johnson, Paul Andrew, 1961-
Publication Date:


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


Thesis (Ph. D.)--University of Florida, 1989.
Includes bibliographical references (leaves 84-86)
Statement of Responsibility:
by Paul Andrew Johnson.
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 - 020379490
oclc - 21587606
System ID:

This item is only available as the following downloads:

Full Text








I gratefully acknowledge the unending support of my parents and my brothers and

sisters. They have been a constant source of encouragement.

I express great thanks to Professor James D. Winefordner, my research advisor. It

has been a pleasure and an inspiration to work in his research group. He has always been

available for encouragement and expert advice, and he has always been willing to laugh at

my jokes. My thanks to Dr. Benjamin W. Smith for many good ideas and much good

advice, and to Tye E. Barber, who worked with me on several of the experiments described

in this dissertation.

I would like to acknowledge Rudy Strohschein, Chemistry Department Glassblower,

for his skillful construction of the sheath flow cuvette and glass capillaries used in my

research. My thanks to Chester Eastman, Vernon Cook, Dailey Burch, and Jerry Holland,

Chemistry Department Machinists, for the construction of equipment used in my research.



ACKNOWLEDGEMENTS ....................................... ii

ABSTRACT .................................. ............ v


1 INTRODUCTION ..................................... 1

2 THE LASER DIODE .................................. 3

Description ...................................... 3
Theory of Operation ................................. 3
Beam Characteristics ............................... .. 10
Laser Arrays ................... ................... 11
Construction Methods ................................ 15
Experimental Studies ................................. 16
Discussion ........................................ 22


Need for Frequency Doubling ........................ 26
Experimental ................. .................... 29
Results and Discussion ................................ 31


Background ..................................... 35
Experimental ................. ................... 38
Results and Discussion ................ ................ 48


Introduction ...................................... 57
Atom Reservoirs ................................... 58
Previous W ork ..................................... 62
High-Power LD Array Excitation ................... .... 64
Single-Mode LD Excitation ............................ 70
Frequency-Doubled LD Excitation ....................... 78
Conclusions ....................................... 79

6 CONCLUSIONS ....................................... 80

REFERENCES ............................................... 84

BIOGRAPHICAL SKETCH ...................................... 87

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




August 1989

Chairman: Professor James D. Winefordner
Major Department: Chemistry

The feasibility of laser diode (LD) excitation for analytical fluorescence

techniques was investigated in this work. The spectral characteristics of single-mode and

high-power array laser diodes were examined with a monochromator, a diode array

detector, and a fourier transform Michelson interferometer. The tuning of LD

wavelength with temperature and electrical current was also studied.

Experiments in frequency doubling of the laser diode were carried out and

compared to results predicted by theory. Temperature tuning of the frequency-doubled

LD was demonstrated. The frequency-doubled LD was tuned to 403 nm to excite

gallium atoms in a graphite furnace, but it proved to have insufficient intensity to

produce a detectable amount of gallium atomic fluorescence. The doubled LD was used

more successfully to excite molecular fluorescence. Perylene, a polyaromatic

hydrocarbon, was detected in hexane at a level of 100 ppb.

Both high-power and single-mode LDs were used to excite rubidium atomic

fluorescence. Limits of detection were 2.1 pg in a graphite furnace and 200 ng in a

glow discharge. Amplitude and frequency modulation were used to discriminate against

emission background.

Molecular fluorescence of the organic dye IR-140 was excited with both types of

LDs, in four different instrumental configurations. The first and simplest approach was

a compact, inexpensive filter fluorometer which could be made portable. Luminescence

was measured with a red sensitive photodiode. In the second configuration, the laser

beam was collimated and passed through the cuvette; luminescence was measured with a

monochromator and cooled photomultiplier tube. A concentrational detection limit of

10'14 M was achieved. In the third approach the cuvette was exchanged for a flowing

jet of sample. For the fourth system, a sheath flow cuvette was constructed to achieve

the sub-nanoliter sample volumes required for single molecule detection. For the latter

two experiments the laser diode was focused into the sample stream with a microscope

objective. Limits of detection were 46000 molecules in the liquid jet and 130000

molecules in the sheath flow cuvette.


In fluorescence spectrometry, the source of excitation radiation is of prime

importance, and often provides the ultimate limitation in sensitivity and selectivity for a

technique. Numerical expressions for the intensity of fluorescence reveal a linear

dependence between the power of the excitation source and the fluorescence flux. Thus

a more intense source will achieve greater sensitivity until the saturation of a transition is

achieved. Laser sources have greatly improved some techniques, while others, such as

Raman spectroscopy, are only made practical when a laser source is used. A laser provides

extremely narrow spectral bandwidth radiation, and this gives the added benefit of

selectivity. The spectroscopist may excite an atom or molecule selectively, that is, without

exciting species which absorb at nearby frequencies. With conventional continuum sources,

such as xenon arc lamps, selectivity may be achieved by passing a smaller spectral band of

light through the wavelength selection device (such as a monochromator). However this

is wasteful, since these high-power sources do not have a high spectral radiance --

brightness per nm -- compared to lasers. Lasers in industry are becoming more and more

commonplace for measurement and positioning of manufacturing equipment, and high-

power lasers are commonly used for precision cutting operations. Lasers in spectroscopy

have remained primarily research tools due to their cost and complexity. This may change

since the cost of instruments in general continues to rise. Also, those in the medical field


may become familiar with laser technology through its clinical applications such as tumor

and blood clot removal.

The laser diode is considered a bridging technology because it is making lasers more

desirable for applications like routine spectroscopy, for which lasers were considered too

costly and complex in the past. Lasers diodes are a product of semiconductor technology.

They combine the technology of a microchip similar to those found in computers, and a

light-emitting diode (LED) used as an indicator light for electronic devices such as stereos.

The light in a laser diode is produced by the same mechanism as an LED, with a laser

cavity provided by the cleaved ends of the transparent microchip. This is discussed in detail

in Chapter 2. The wavelength of a laser diode (LD) is temperature tunable. This allows

different atoms or molecules to be excited by the same LD, or the wavelength can be

shifted away from an absorption line for background correction.

The research to be described was carried out to evaluate the spectral properties of

laser diodes, and to perform experiments usually reserved for large expensive lasers.

Success would make possible the use of lasers in routine spectroscopic instruments with the

result of improved performance of these instruments in terms of sensitivity and selectivity.

Laser diodes would be a bridge between laser technology and routine instrumentation.

The spectral and optical properties of commercially available laser diodes are

examined in Chapter 2. The main limitation of laser diodes at this time is wavelength

range, which is currently limited to the red, near-infrared, and the infrared regions of the

spectrum. To access the blue region, experiments were carried out to frequency double

laser diodes; these experiments are described in Chapter 3. Chapters 4 and 5 present the

results of experiments using laser diodes to excite molecular and atomic fluorescence as part

of an analytical technique.



Laser diodes were first constructed in 1962, only four years after the invention of

the laser in 1958. Laser is an acronym for "light amplification by stimulated emission of

radiation." Conventional lasers range in size from the helium neon laser, which can be held

in the hand, to the more powerful lasers such as the neodymium YAG and the argon ion

lasers, which cover a large bench-top. The size of the laser diode is revolutionary compared

to these lasers, and the laser diode can deliver as much optical power as the large lasers,

although with slightly less beam quality. They are very efficient the efficiency in

converting electrical energy to light approaches 50%. The dimensions of a laser diode are

in micrometers rather than meters. Most of the volume of a laser diode is taken up by the

package, which protects the semiconductor chip and provides convenient pin-outs for the

electrical leads. A common package is the HO package, identical to the 8-pin diode and

transistor packages familiar to electronics personnel. Presently, the largest use of laser

diodes is in compact disk players and optical data storage equipment.

Theory of Operation

In conventional lasers, the mechanism of light production involves electronic

transitions of atoms or molecules, or excimers formed from reactant gases. LDs produce


optical emission as a result of transitions between the conduction and valence bands of

semiconductors. Semiconductors, as the name implies, are neither good conductors nor

good insulators, but something in between. The common material used for computer

semiconductor chips is silicon. For semiconductor lasers, gallium arsenide is used. The area

where light is produced is called a p-n junction, where p-type and n-type semiconductor

materials are brought together. In a crystal of GaAs, all of the valence electrons are

involved in bonding in the crystal lattice. Introduction of selected impurities alters the

electrical properties of the material. If some of the As (valence V) atoms are replaced by

Se atoms (valence VI), these atoms have an "extra" electron not involved with bonding.

These donor atoms produce an n-type material, in which the extra electrons are free to

move through the crystal. A p-type semiconductor is produced by introducing acceptors

such as Zn (II) atoms, which are lacking an electron compared to the atoms of the crystal

lattice; "holes" are introduced. These holes are free to move in a similar manner to the

electrons. When p and n type materials are brought together, electrons move from the n

to the p side, where they combine with the holes in the junction region. This process will

take place until an electronic potential between the two materials on the order of 500 mV

is built up. An energy gap or band-gap is produced between electrons in the valence band,

where they are bound to the lattice, and the conduction band, where they are free to move

in the crystal. The p-n junction may be forward biased by applying an external voltage with

electrical contacts. This allows a further flow of electrons and holes to the junction, and

current flows across the junction. In the junction region, electrons and holes combine, with

the release of energy (figure 2-1). The band-gap (Eg) determines the energy released in


this process. The energy may be released in the form of a photon, with wavelength (A, m)

given by

1 = hc/Eg (2-1)

where h = Planck's constant (j s) and c = the speed of light (m/s) (1). The photon is

immediately subject to absorption in the junction by formation of a new pair of electrons

and holes. As the current through the junction increases, a threshold current is reached,

at which point more photons are being produced than consumed. The p-n junction exhibits

optical gain, and is described as having a population inversion of charge carriers between

the bound and valence states.

Optical gain is one of the ingredients for a laser; the other is a resonant cavity to

contain the radiation. Optical confinement in LDs is provided through changes in the

refractive index of the material through which the light travels. The sides of an LD chip

are rough, while the ends are cleaved and polished. At these ends, or facets, the refractive

index changes from 3.6 in the junction to 1, the refractive index of air. This provides high

reflectivity at the ends of the chip; a Fabry-Perot cavity is formed.

This describes homojunction lasers, the first laser diodes produced. The region

where light is produced is very thin as shown in figure 2-2, and the boundaries trail off into

the p and n materials. A very high current density, at least 50 kA/cm2, is required to reach

the lasing threshold. A great improvement in the efficiency of laser diodes was the

introduction in 1969 of double heterostructure semiconductor lasers (2); sandwiched

between the p and n type semiconductor material is a 0.1 /pm to 0.3 /im layer of p-type



Ih f\ hf E,9

; r


Figure 2-1.

Production of photons at a p n junction. The top line represents the
energy of electrons in the conduction band, and the bottom line represents
the energy of electrons in the valence band. The shaded area is the p n
junction, where electrons and holes combine, producing photons.



GaAs .. ..
GaAs A ';

i uncti on

metal lic

of as e:

stri p

t cone
r light

Figure 2-2. Structure of a gain-guided laser diode.



cont act


v e

insul ati on

n p


.c o nt act


1 aye

wavegui de


Figure 2-3.

End view of laser diodes showing two methods of confining output beam.
a) Gain-guiding with a stripe contact; b) index-guiding with a waveguide
active layer.

1 aye



material having a band-gap intermediate between the p and n type materials on either side.

This layer becomes the gain region of the laser. Electrons from the n-type side of the

junction are free to diffuse into this layer, but they cannot cross all the way over to the p-

type side because of the band-gap difference. Similarly, holes migrate to this region, where

recombination and emission of photons occurs. The key feature of this active region is

that it has a higher refractive index than the material above and below it. This layer

exhibits total internal reflectance of light as described by Snell's law:

0 = sin"'(pl/2) (2-2)

If the angle of incidence of a light ray exceeds 0, which is determined by the ratio of the

refractive index of the cladding layers 41, and the refractive index of the active region p2,

it will be reflected without loss. Thus, rays which are nearly parallel to the plane of the

p-n junction are reflected, and the material acts as a wave guide. The efficiency

improvements of heterojunction lasers allow them to be operated in a continuous wave (cw)

mode at room temperature.

The type of laser described so far is an area-emitting laser, because it will emit over

the area where the p-n junction meets the facets of the semiconductor chip. This presents

problems in the quality of the beam emitted, which make it difficult to focus. More

important to the LD developers, area-emitting lasers are difficult to couple to optical fibers.

Either index guiding or gain guiding is used to channel the current in the active region into

a stripe parallel to the emitted beam, to produce a spatially narrow elliptical beam. In

index guiding, the refractive index of the active layer is varied laterally, forming a waveguide

down the length of the chip. Light is confined to the stripe by the same mechanism, total

internal reflection, which confines it to the active region. Index-guided laser diodes have


better beam quality than gain-guided LDs, but they are more limited in output power. In

a gain-guided LD, the broad area electrical contact at the top of the (area-emitting) chip

is replaced by a 10 jpm stripe, with an insulating dielectric layer on either side. Beam

confinement in the vertical and horizontal planes is by different mechanisms, total internal

reflection and charge injection (by a stripe contact), respectively. The beam waist in the

vertical plane appears about 50 pm behind the facet, while the beam waist in the horizontal

plane is at the surface of the facet. This causes the laser output of gain-guided LDs to be

astigmatic, requiring a weak cylindrical lens for correction (3). End-on views of a gain-

guided and an index-guided LD are shown in figure 2-3.

Beam Characteristics

Light of a certain bandwidth is emitted in the active region of the LD. The

bandwidth is determined by the range of recombination energies of electrons and holes.

The spectral and spatial properties of laser diode output are described by the mode theory

common to all lasers (4, p. 89). Within the bandwidth of the lasing transition some

wavelengths are resonant within the Fabry-Perot cavity formed by the cleaved ends of the

laser chip. This cavity can be considered as bounded by two partially reflecting mirrors.

Light leaving the surface of one mirror undergoes a phase change, A0, as it makes a round

trip, reflecting from one mirror and returning to the first. This phase change A0 is given

by A0 = 4rUIA, where A is the wavelength of the light (m) and L is the length of the

cavity (m). For A0 = 2rq, the wave and its reflection are in phase and a standing wave

is set up; q may be any integer. Expressed as frequency, the resonant modes of the laser

cavity are given by

v = q(c/2L) (2-3)

The separation between modes is Av = c/2L For all other types of lasers, L is large (on

the order of several centimeters), so that the spacing between these longitudinal modes is

very small. Thus the output from the laser is made up of several to many discrete lines.

For an LD, however, L = 250 pm, so there are fewer resonant frequencies within the

bandwidth of emitted light. If this bandwidth is limited to homogeneous broadening, only

one longitudinal mode near the emission peak reaches threshold. The output from a single

laser diode is therefore extremely narrow (10 100 MHz).

Laser Arrays

The spectral output of the high-power laser diodes used in this research are more

complicated, because they are actually arrays of lasers. The highest power that has been

obtained from a single laser diode is 60 mW (limited by leakage current across the diode).

To reach higher powers, multiple parallel stripes of active material are fabricated into the

laser chip (figure 2-4). These individual lasers are close enough that their electromagnetic

fields overlap; one laser can "feel" the effect of the others. This results in a definite phase

relationship between the emitters, and a coherent beam emerges. Spectrally, the beam

consists of a number of supermodes. Figure 2-5 shows the spectrum of a 10-stripe gain-

guided laser diode array, operated at 200 mW cw output power.

The spatial pattern of the LD beam is determined by the transverse modes of the

laser cavity, designated as the transverse electronic (TE) and transverse magnetic (TM)

modes. These modes arise because of the three-dimensional nature of the laser cavity.

(Longitudinal modes consider only the case of light travelling along the optical axis of the


me tal 1 ic
stri pe
cont acts


Figure 2-4. Structure of a high-power laser diode array, showing multiple stripes.

796 791
nm nm

Figure 2-5. Multi-mode spectrum of a high-power laser diode array.





Figure 2-6.

Spatial output pattern of laser diodes.
a) low-power, single-mode LD; b) high-power LD array.

-- j' ---


cavity.) One of the properties of the dielectric waveguide (the optical cavity of an LD) is

that the cleaved surfaces are more reflective to TE modes than TM modes. As a result,

the output is a TE mode, 95% polarized. As mentioned previously, the emitters of a laser

diode array are coupled. There is a 180o phase shift between adjacent stripes; this leads

to a dual lobe spatial pattern (figure 2-6) (5). Diffraction of the beam also leads to a

large beam divergence. Considering the laser light to pass through a slit the width of the

active region, a diffraction angle of 9 is given by the half angle diffraction formula (4,

p. 185). Figures 2-5 and 2-6 represent the spatial and spectral character of the high-power,

multi-stripe LD beam. Spectrally, it contains 5 to 7 peaks spanning 3 nm. Spatially it is

a divergent, two-lobed beam.

Construction Methods

Construction of laser diodes requires the deposition of extremely thin layers of

materials on a gallium arsenide substrate; in addition, the layers have to form one

continuous crystal lattice (1, p. 143). This process is called epitaxy. Liquid phase epitaxy

(LPE) was the only method used commercially to produce laser diodes until 1983.

Saturated solutions of the materials making up each layer are brought into contact with the

GaAs substrate. The substrate and/or solution are then cooled, causing semiconductor

material to be deposited on the substrate. This can be repeated for several solutions to

achieve heterostructure epitaxial lasers. Pure gallium is often used as the solvent; the

process is carried out under hydrogen gas at 600-900C. The lasers used in this work were

manufactured by the technique of metallo-organic vapor chemical deposition (MOCVD).

In this technique hydrogen gas is bubbled through a melt of methylated metals or metal


hydrides. The vapors are brought into contact with a hot substrate. Decomposition of the

metal compounds occurs at the surface, leaving epitaxial semiconductor layers. MOCVD

produces interfaces between layers superior to LPE, with excellent uniformity and

reproducibility. The best LD structures are produced by molecular beam epitaxy (MBE),

but this technique is the most difficult and expensive to operate. Atomic or molecular

beams of the layer constituents are produced with effusion ovens. The beams travel into

a high vacuum chamber and strike the substrate, where they react to form epitaxial layers.

This may be the technique of the future once the manufacturing process is optimized.

These methods are used to produce entire wafers of semiconductor material. Next,

the stripe contacts are applied to the wafer, and the wafer is cleaved across the stripes.

The facets exposed by this cleavage are coated for protection, and the slices of wafer are

sawed or cleaved into individual lasers, ready for mounting. A package for inexpensive

laser diodes is shown in figure 2-7. Low-power LDs are mounted on heat sinks, while

high-power LD arrays are mounted on the cooling element of a thermoelectric cooler. A

thermistor is included to monitor the temperature of the LD. A photodiode is placed on

one side of the emitting chip to measure the output power. The photodiode signal may

be used in a feedback circuit to stabilize the power of the laser diode. Controls and

displays of current, power, and temperature are provided in one convenient laser diode


Experimental Studies

Studies were undertaken to determine the performance characteristics of laser diodes

and their suitability to different fluorescence experiments. Spectra of the laser diode output


were recorded on three different spectroscopic systems: a grating spectrometer with

photomultiplier tube (PMT) detection, a Michelson interferometer, and a grating

spectrometer with photodiode array (PDA). This was carried out to avoid artifacts which

can accompany the measurement of laser diode radiation. One of these artifacts is

diffraction of a coherent laser by the entrance slit of a monochromator, resulting in artifact

"peaks" produced by stray light inside the monochromator. To avoid this, the LD beam was

focused on a flat piece of Teflon or frosted glass and scattered into the monochromator.

The scattering process renders the light incoherent. Figure 2-5 shows a spectrum of the

scattered beam from a 200 mW c.w. LD (Spectra Diode Labs Model SDL 2422-H1)

obtained with the monochromator equipped with PMT detection. A slit width of 25 pm

and PMT voltage of 900 V were used.

Since a high-resolution Michelson interferometer was available in the laboratory, it

was used to record high-resolution spectra of both single-mode LDs and LD arrays. The

Fourier Transform Interferometer (Bomem, Quebec, Canada) is capable of recording

spectra throughout the ultraviolet (UV), visible, and infrared regions of the spectrum by

changing beamsplitters and detectors. The near-infrared (NIR) emission of the laser diodes

is easily measured with this instrument, employing a silicon avalanche detector. The

resolution was sufficient to observe the mode structure of a single-mode LD (Mitsubishi

Model ML 4102) evolve from several allowed modes at a power level just above threshold

to a single mode in the operating current range. Figure 2-8 shows the multi-mode

spectrum. The spacing between the modes is 0.275 nm. From equation 2.3, this

corresponds to a cavity length of 125 pm. Upon increasing the power to the normal

operating range to 3 mW, the output becomes single-mode, with a linewidth of 0.02 nm,

as shown in

I aser
di ode

mo ni t or
photodi ode


Figure 2-7. Internal view of a standard laser diode package.

od e....
*l". .

si nk

RES: 0.040 DATE: 14-APR-9 TIM: IS 38t 40
COAD: 10
Laoer DaodS37BA Pour 02 t-20.0 C
X 1.OE+03


7 St
4 Pt
4 HP



127 0. CM-1

Figure 2-8.

Mode structure of a single-mode laser diode operated just above the
threshold current. The full scale of this spectrum is from 782 nm to 792 nm.
The spacing between modes is 0.275 nm.

ES : 0.500 DATE: 1L-AP*R-S TIre 01 : r s
COAU 100 1
Lamar aEse. 41 aA. PEL k 40 --.0 a

X L.0E+05

2.3 i4



L. 4

.P: 7
jLP. 6


- -U -





, 5 _



1285a.0 CM-I

Figure 2-9.

Output spectrum of a single-mode laser diode with 4 mW output. The full
scale of the spectrum is from 778 nm to 782 nm. The width of the line at
half of maximum height is 0.02 nm.

| r Ba 4 a -.0


figure 2-9. Examining this mode structure with the highest resolution obtainable on the

Bomem interferometer, fine structure such as splitting and shoulders were observed on this

single mode. This structure may have been an artifact of the Bomem instrument, as other

researchers report these lasers to be truly single-mode.

The Bomem FT instrument was used to take spectra of the high-power LD arrays,

showing them to have some fine structure on their individual modes. To verify this fine

structure and to observe the changes in the array supermodes on a real time basis, a high-

power LD was scattered into a 1 m monochromator equipped with a photodiode array

detector. The relative height of the supermodes was observed to be quite stable. Of more

interest were the real time views of the spectra as the amount of current and the

temperature of the LD were varied. When the temperature was increased, the wavelength

increased, and the cluster of peaks moved across the screen like a wave, with the higher

wavelength modes getting more intense and the lower wavelengths fading away. The overall

height of the peaks also decreased as the temperature was increased. This is because the

efficiency of the device decreases with increasing temperature. Figure 2-9 shows a series

of spectra as the current to the LD array was increased from threshold at 270 mA to the

maximum current of 608 mA. The total output power measured with the internal

photodiode is listed for each spectrum.

The bandgap energy of a semiconductor changes with temperature; this changes the

wavelength of LD emission, (see equation 2-1). Experimentally, a rate of change of

0.25 nm/oC was observed. With thermoelectric cooling, the LD could be operated at

temperatures from -20C to +40C for a wavelength range of 15 na. The LDs were

supplied at a specified wavelength for operation at 25C, with the tunable range extending


on either side of this value. A finer control of wavelength was obtained by tuning the LD

with the current supplied. A higher current increases the concentration of charge carriers

in the active region. This in turn changes the refractive index of the active region and the

wavelength of light produced. The rate of change was 6 x 103 nm/mA. This fine control

was used in tuning to Rb transitions (see Chapter 5). The temperature and power could

be changed quickly with controls on the high-power LD driver (Spectra Diode Labs Model

SDL 800). The high-power LDs had thermoelectric (TE) coolers mounted inside the LD

package. The single-mode LDs had no internal TE cooler, so they were mounted on a

laser diode mount (ILX Lightwave Model LDM 4412) equipped with a TE cooler and a

temperature controller (ILX Model LDT 5910). The interior of the LD mount was purged

with nitrogen gas to prevent condensation. The exterior (heat sink) was cooled with a

small fan. Single-mode LDs were powered with an ultra-low noise current source (ILX

Model 3620). An important feature of both the SDL and ILX LD drivers was the current

limit option, which was set according to the LD manufacturers specifications. This

prevented the accidental overpowering of an LD, which would damage it.


The high-power LD array is well suited for applications where a broadband

absorption peak ( > 3 nm) is to be excited. Presently, the major application of these

devices is the pumping of yttrium aluminum garnet (YAG) lasers. For this purpose, the

output of the LD may be spectrally as well as spatially broad. The highest power LDs,

available from Spectra Diode Labs, are in the form of 1 cm bars for the side pumping of

YAG. Pumping a Nd-YAG laser, with its excellent beam quality, and then using frequency


conversion techniques to access the visible region of the spectrum, may be a better

approach than direct frequency doubling of semiconductor lasers. High-power LD arrays

are also excellent sources for exciting molecules which absorb in the NIR, since all of the

modes will be absorbed.

For exciting atomic transitions, single-mode lasers are more applicable because of

their narrow linewidth. The high-power arrays are more powerful, but most of the power

lies on either side of the atomic line. Not only is this light wasted, it increases the limit

of detection, which may be limited by noise on the laser scatter. The advantages and

disadvantages of LD for each type of spectroscopy will be discussed more fully in the

appropriate chapters. Table 2 lists examples of commercially available laser diodes. LDs

with short and long wavelengths, high-power, and narrow linewidths are shown.

150 mW

9 0 mW

1 3 0 mW

760 785 760 75 ... 05 810



Figure 2-10. Output spectrum of a high-power LD array as the drive current is increased
from the lasing threshold to the maximum rated current.

2 210 mW

Table 2-1
Examples of Commercially Available Laser Diodes


NEC Corp.




Spectra Diode

Spectra Diode

Active A Power AA
Model # Material (nm) (mW) (nm) Comments











AIGaAs 788

AIGaAs 780

GaAIAs 795

GaAIAs 795









0.02 Used in this research
Single-mode $30

0.01 $84

2.0 Used in this research
Multi mode $1100

5.0 $4000

Toshiba TOLD350S

InGaAsP 1550 4


Need for Frequency Doubling

Atomic or molecular fluorescence is nearly always excited with light sources of

wavelengths less than 600 nm. In the case of atomic fluorescence, there are few atoms

which can be excited from the ground electronic state with the fundamental beam from a

laser diode. With molecules, the lowest energy transitions are the n -. r transitions, which

can be excited with wavelengths as long as 800 nm (6). In order to excite the vast majority

of fluorescent atoms and molecules with a laser diode, a frequency conversion technique

must be used. Second harmonic generation (SHG) is a process where light at one

frequency is converted to twice the frequency. In effect, two low-energy photons enter a

nonlinear crystal and one high-energy photon emerges. This occurs with an efficiency

characteristic of the light source and the crystal. Since the frequency is doubled, the

wavelength of the SHG beam is one-half of the incident, or fundamental beam.

The science of nonlinear optics began in 1961 with the invention of the laser. Its

high spectral brightness makes possible power densities of 109 W/cm2. This generates

electric field strengths of 106 V/cm in a medium through which the laser passes. This field

strength is comparable to atomic field strengths, so nonlinear interactions occur (7, pg. 150).

The electromagnetic field causes the polarization of the electrons in the crystal to oscillate.

This oscillating electronic structure radiates light at the fundamental frequency, as well as


harmonics. Whether the harmonics are 2nd, 3rd, or 4th, etc., depends on the crystal. The

polarization (C m"2) is given by

P = eo X E (3-1)

where P is the polarization, co is the permittivity of free space (C2 N'1 m"2, a constant)

E (N C') is the electric field strength, and x(1) is the linear optical susceptibility

dimensionlesss). This equation describes the polarization of a medium at low spectral

intensity. At high intensity, the complete constitutive relation should be written to take

into account higher order susceptibility.

P = + X + ) ) 2E + )E +...] E (3-2)

The factor x(2) is the second-order nonlinear susceptibility, which gives rise to second

harmonic generation. The induced polarization of the medium and the power of the

second harmonic beam are proportional to the second-order nonlinear coefficient, and to

the square of the fundamental beam power.

The fundamental beam must be tightly focused, as well as powerful. One of the

factors in doubling efficiency is the Boyd and Kleinman focusing factor h(B,C). This factor

is a function of the double refraction parameter and the focusing parameter C = 1/b, where

I is the length of the crystal and b is the confocal distance, or twice the distance over

which the beam doubles. The SHG conversion efficiency is optimum at 1/b u 2.84 (7,

pg 160). In practice, the focusing of a laser beam into a nonlinear crystal is limited by the

damage threshold of the crystal. Above this threshold, heating causes the crystal, which may


cost thousands of dollars, to fracture. The semiconductor lasers used in this work did not

have enough power to damage the crystals used.

For efficient SHG, the fundamental and harmonic waves must stay in the same path

through the entire crystal length. Isotropic crystals cannot be used for SHG because they

are light dispersive. Since the fundamental and harmonic beams are of different

frequencies, they are separated in isotropic crystals. Birefringent crystals are used for SHG.

These anisotropic crystals are doubly refracting, separating an incoming light beam into two

diverging beams, the ordinary and extraordinary. For the proper crystal, at the correct

angle and temperature, birefringence keeps the fundamental and harmonic beams

phasematched and they propagate through the crystal together. Phasematching is used to

offset the dispersion of a crystal and to select a certain harmonic, such as the second

harmonic, to the exclusion of all others (3rd, 4th, etc.). The two kinds of phasematching

are called Type I and Type I (8). Type I requires a plane polarized input beam, while

Type II phasematching may use an unpolarized or linearly polarized input beam. In

addition, Type II phasematching has a larger tolerance of angular phase matching, making

angular alignment less critical.

There are many thousands of crystals known, out of which only a few hundred are

candidates for nonlinear materials. The most stringent requirement is adequate

birefringence for phasematching. A crystal must also have a high effective nonlinear

coefficient P 0 an experimental measure of conversion efficiency. It must be transparent

to both the fundamental and SHG beams, and have a high damage threshold. The most

commonly used SHG crystals are potassium dihydrogen phosphate (KDP), beta-barium

borate (BBO), lithium niobate, and lithium iodate.


The laser diode was operated at its maximum power of 210 mW in the continuous

wave (cw not pulsed) mode. The power was monitored using the built-in photodiode

supplied with the LD. The doubling crystal was lithium iodate cut at 41 for Type II

phasematching. The input beam at 796 nm was collimated and shaped, then focused into

the crystal with a 14 in focal length glass lens. The long focal length was used to maximize

the length of interaction between the focused beam and the crystal (figure 3-1). The power

of the second harmonic beam at 398 nm was monitored with a photodiode after the

fundamental beam was absorbed by a BG-18 colored glass filter. The current from the

photodiode was monitored with a sensitive electrometer, and the power was calculated from

the typical curve of photodiode response (mA/mW) provided by the manufacturer. The.

BG-18 filter passes 50% of the SHG beam with non-detectable passage of the fundamental

beam. For fluorescence work, the removal of the fundamental beam was accomplished with

a copper sulfate solution (10% CuSO4 in deionized water) in a 1 cm pathlength filter

cuvette. This filter passed nearly 100% of the SHG beam, allowing 0.01% of the

fundamental beam to pass. While this small amount of the input beam could be tolerated

during fluorescence experiments, it was unacceptable for doubling efficiency measurements.

For these experiments, the colored glass filter was used. Translational mounts and a

precision rotational mount (Newport Corp. Bikini, CA) were used to position and angle

tune the LilO3 crystal. Stable mounts providing fine adjustments of the positions of optical

components proved essential to this work.

Col I mat or

Di ode

Anamor phi c Crystal BG-I
SFl I ter
Pri sm Pai r

Figure 3-1. Optics used to frequency double a laser diode.

Results and Discussion

The LD was successfully frequency doubled, the blue beam being visible on a

fluorescent index card after passing through the copper sulfate solution filter. The highest

SHG power measured was 92 nW. Allowing for the 50% absorbance of the SHG light by

the colored glass filter, 180 nW was produced. This corresponds to a doubling efficiency

of 180 nW/210 mW = 86 x 10'5%. This is certainly a low efficiency, but comparable to

other workers in the field, as shown in table 3.1. This laser was relatively broadband

(FWHM = 0.5 nm), and also of extremely low power at 100 nW. Since the emitting

aperture was the same as the high-power LD, this corresponds to a spectral radiance of

only 0.1 W/cm2 nm. The power of the second harmonic beam showed a squared

dependence on the power of the fundamental beam as predicted by equation 3-2. The

experimental curve is shown in figure 3-2. Pulsing the laser diode to provide higher peak

powers would increase the doubling efficiency. The short pulse width laser diode driver

required for a high-power pulsed LD would have to be purchased or built. The

polarization of the frequency-doubled beam was orthogonal to the fundamental beam. This

is a characteristic of Type II phase matched second harmonic generation.

Tuning to an atomic line proved difficult since the doubling crystal had to be angle

tuned often as the fundamental wavelength was temperature tuned. This difficulty could

be overcome with a commercial autotracker, which automatically angle tunes a doubling

crystal. The frequency-doubled LD was temperature tunable. The slope of the tuning

curve (figure 3-3) was 0.15 nm/C, half of the value for the fundamental beam.

E 60


0 20


P of Fundamental



Figure 3-2. Relationship of second harmonic power to fundamental LD beam power.






0c m


404 1 p 1ll 111 1111
-30 -20 -10 0 10 20 30 40
temperature (OC)

Figure 3-3. Temperature tuning curve for frequency double laser diode.

Table 3.1
Comparison of LD Frequency Doubling Results

Doubling P(shg) P(fund) Efficiency
Researchers LD type Crystal (nW) (mW) %

Johnson, Smith,
Winefordner (1987)

Bowmurt, Gunther,
Melchior (1983)

Deserno, Kappeler,
Hanke (1987)

Okazaki, Imasaka,
Ishibashi (1988)

Tohmon, Yamamoto,
Taniuchi (1987)






Matsushita Electronics









LiNbO3 106
Proton exchanged
etched waveguide









Ultraviolet-visible fluorescence is a selective and extremely sensitive spectrometric

technique. Various workers have used this sensitivity to pursue the goal of detecting a

single molecule. Hirschfeld detected individual antibody molecules tagged with 80 -100

fluorescein molecules in a solution spread on a microscope slide (9). The minimum number

of fluorescent tags necessary to detect a single molecule may eventually become a popular

way of reporting a limit of detection (LOD). A very low LOD for Rhodamine 6G

adsorbed onto 10 pm silica spheres was achieved by Kirsch et al. (10). When analysis is

performed in liquid solution, solvent and impurity fluorescence and Raman scatter must

be filtered out, as they contribute to the background. Impressive LODs have been achieved

in both static cells (11 13) and in flowing streams, such as an HPLC eluent (14 17).

As well as detecting low concentrations, a technique approaching single molecule

detection must excite fluorescence in small sample volumes. The sheath flow cuvette (SFC)

is a device for obtaining extremely small volumes of liquids. The SFC was developed for

particle counting, and to enable biochemists to study spectroscopically the DNA and protein

content of individual cells dispersed in solution (18). The SFC is at the heart of a flow

cytometer, which enables biochemists to line up cells individually and excite molecular

fluorescence with a laser. It is based on the principle of hydrodynamic focusing, in which


a sample stream is introduced with a glass capillary into the center of a sheath of flowing

solvent. The capillary is centered in a larger diameter glass or quartz tube which contains

both the sample and solvent streams. The tube narrows at the end, to a small opening.

It is in this narrowing region that hydrodynamic focusing takes place. The width of the

sample stream may be varied between two extremes. The stream diameter will be the size

of the tube opening if no sheath solvent is used; it will be zero if no sample is used. Flow

cytometers offer precise control of both sample and sheath flow rates.

The sheath flow cuvette has been used by Dovichi et al. (19) to produce detection

volumes as small as 50 fL A polarized helium-neon laser was focused to a spot size in the

sample of 10 pm with a 25 mm focal length lens. Scattering from particles in the sample

stream was collected with a microscope system for examination of hydrodynamic focusing.

They estimate that the lower limit of the detection volume obtainable with a sheath flow

cuvette is 5 fL Keller et al. excited rhodamine 6G in a sheath fow cuvette with the

514.5 nm line of an argon ion laser (20). The beam and stream diameters were 22 pm and

30 pm respectively. The 1 W argon ion laser was chopped and focused with a 5 cm focal

length lens to give an irradiance in the probe volume of 0.13 MW/cm2. The flow cytometer

optics were used to collect the rhodamine fluorescence with 32X magnification and 6%

collection efficiency. A detection limit of 22,000 molecules was obtained using the formula

mass concentration limit (g) = concentration detection limit (g/L) (4-1)

X volume flow rate (L/s) X time constant (s)


This yields the weight of analyte (g) passing through the probed volume during one time

constant (usually 1 s). The concentration detection limit was based on the signal obtained

from only one standard, which assumes that the calibration curve is linear as the

concentration approaches the detection limit. Still, the mass concentration limit (22000

molecules) is a conservative definition of detection limit. From the probe volume and

concentrational detection limit alone the absolute LOD (concentration X volume) is

0.6 molecules. The key to their success was good spatial filtering with a 0.75 mm aperture

and spectral filtering with three consecutive filters to discriminate against laser scatter. No

monochromator was used.

Low-cost, compact lasers for excitation of molecular fluorescence are needed.

Excitation of molecular fluorescence with high-power multi-mode laser diodes has been just

as successful as excitation with narrow linewidth single-mode lasers, due to the broader

excitation bands involved in molecular fluorescence. Fluorescence from the solvent and

impurities in the solvent often limit the ultimate sensitivity of techniques using UV or

visible lasers for excitation (21). Near infrared (NIR) fluorescence avoids this limitation,

since few compounds fluoresce in this wavelength region. While this also limits the

compounds that can be directly determined, the low background leads to an ultra-low LOD

for those compounds that do fluoresce. Raman scatter background is less than with uv-

vis lasers since the scattering process is less efficient in the NIR and the scatter is shifted

beyond the spectral range of most photomultiplier tubes. In addition, problems associated

with sample photodecomposition are much reduced at longer excitation wavelengths. The

synthesis of NIR fluorescent labelling reagents for fluoroimmunoassay and automated DNA


sequencing would capitalize on this situation and open up new prospects in the biochemistry


Ishibashi and Imasaka et al. (21) have explored the use of the laser diode for

molecular fluorimetry, reporting it to be efficient, compact, rugged, and inexpensive. They

have used an LD as a source for fluorimetric enzymatic assay (22) and as a detector for

HPLC determination of labelled protein (23). A frequency-doubled LD was coupled to a

fiber optic for use as an oxygen sensor based on fluorescence quenching of

benzo(ghi)perylene (24).

In this work, a study was made of various configurations of laser diodes and

detectors for obtaining the lowest detection limits in practical and simple sample

introduction schemes. In the first configuration, a cuvette was used to hold the sample.

A comparison was made between a fluorescence spectrometer and a simple filter

fluorometer. In another configuration, a flowing liquid jet fluorescence spectrometer was

used. This configuration could be used as a flow injection or liquid chromatography

detector. A compact cell glass filter fluorometer was constructed with minimal optics and

cost. Finally, a sheath flow cuvette was constructed to achieve extremely low sample



Laser dye IR140 (Exciton Chemical Co. Inc., Dayton, OH; chemical name 5,5'-

dichloro-1 -diphenylamino-3,3'-diethyl-10,12-ethylenethiatricarbocyanine perchlorate) was

dissolved in methanol for the standards used in this work A 200 mW continuous wave

AIGaAs laser diode (Spectra Diode Labs SDL-2422-H1; San Jose, CA) was used as the


excitation source for most of the experiments. A 3 mW single-mode LD (Mitsubishi

ML 4102) was used with the sheath flow cuvette experiments because of its smaller emitting

aperture. Although the laser diode wavelength is in the NIR, the output could be seen as

a weak red beam due to the high power of the device. This greatly facilitated alignment

of the optics. A comparison was made of four different configurations of equipment used

to focus the laser beam and measure fluorescence. In all cases, the signal current was

measured with a sensitive electrometer (Keithley) with output to a chart recorder (Fisher).

A laboratory constructed low-pass electronic filter with a time constant of 1 s was used

between the electrometer and the chart recorder.

In the cuvette fluorescence spectrometer system, the 200 mW LD beam was

processed with a collimating lens and an anamorphic prism pair (Melles Griot, Irvine, CA)

to provide a collimated, square shaped beam measuring 3 mm by 4 mm. The laser

wavelength was tuned to 794 nm by operating it at a temperature of -7.6 oC. This beam

was passed through a methanol solution of the dye contained in a conventional 1 cm x 1

cm square glass cuvette. Fluorescence at 830 nm was collected at right angles with a 2 in

focal length lens and imaged on the 5 mm entrance slit (64 nm bandpass at 830 nm) of

a monochromator. No optical filters were used. The monochromator was a Spex Minimate

0.25 m (Spex Industries Inc., Edison, NJ) with a grating blazed for 750 nm. The

Photomultiplier (PMT) was red sensitive, with a -2% quantum efficiency and -10 mA/W

photocathodic sensitivity at 830 nm; and a spectral range of 185 900 nm (Hamamatsu

R955, Hamamatsu Corp., Bridgewater, NJ). In preliminary experiments, the limit of

detection (LOD) was determined by the noise on the PMT dark current, so the PMT was


equipped with a thermoelectric cooler (Products for Research, Inc., Danvers, MA). After

this improvement, the LOD was determined by noise on the laser scatter.

A study was undertaken of the signal-to-noise ratio with different slit widths and

different optical filters to reduce laser scatter. The results of this study led to a system

where the monochromator was omitted and only a interference filter (Corion S10-830-F;

10 nm FWHM and 45% transmittance at 830 nm) was used for wavelength selection.

This system is called the cuvette filter fluorometer. The fluorescence was excited in a

cuvette as above, but this time collected with the lens and sent directly to the cooled PMT

through the filter. The interference filter was used primarily to reject laser scatter.

Defining the probe volume is an important step in determining the LOD in a

focused laser experiment Essential to this is an accurate measurement of the diameter

of the beam waist (the diameter at the focus). In the liquid jet and sheath flow cuvette

systems, the LD was focused with a 20X microscope objective. The diameter was measured

by translating a calibrated pinhole along the focal plane. A 10 Jpm pinhole was used for

the 200 mW LD and a 2 pm pinhole for the single-mode 3 mW LD. A Hamamatsu

S2386-45K photodiode was used to monitor light flux passing through the hole. A plot of

intensity vs pinhole position yielded a spatial profile of the focused laser (figures 4-1

and 4-2).

In the liquid jet fluorescence spectrometer system, the collimated, shaped laser diode

beam was focused into a microscope objective using a 14 in. focal length lens. A 20X

microscope objective focused the beam to a diameter of 450 pm. The sample was syringe

pumped through Teflon tubing into a glass capillary with an internal diameter of 400 pm.

This apparatus was constructed by epoxy gluing a 1 cm length of glass capillary into the end










Figure 4-1.

distance (/nm)

Spatial profile of a high-power laser diode array, focused with a 20X
microscope objective.

vert ical









Figure 4-2. Spatial profile of a single-mode laser diode, focused with a 20X microscope
objective. Note that the scale is different from figure 4-1.


of a 3 cm length of small i.d. stainless steel tubing. The steel tubing was then connected

to the Teflon sample tubing with a Swagelock fitting. The laser diode beam was focused

into the liquid jet formed by the solvent immediately after exiting the capillary. Emission

was collected by the lens, monochromator and cooled PMT. A Corion RG850 colored-

glass filter (long-pass, 50% transmittance at 850 nm; Corion Corp., Holliston, NJ) was used

before the Spex Minimate 0.25 m monochromator (2.5 mm) to reject laser scatter. A block

diagram is shown in figure 4-3.

Modern flow cytometers are complex instruments costing at least $100,000. The

work in this lab represents an initial study in the field of exciting molecular fluorescence

in small volumes. A sheath flow cuvette was constructed in the glass shop (see figure 4-4)

to study the feasibility of focusing laser diode radiation into hydrodynamically focused

streams. A pulseless flow of sheath solvent was achieved using a steel ballast tank filled

with methanol. Nitrogen pressure was applied to the tank to force the methanol through

Teflon tubing and through the SFC. The sample was syringe pumped through Teflon

tubing to the glass sample capillary. The experimental setup did not have calibrated

controls for sheath flow rate like a commercial flow cytometer; another difference was the

viewing position of the SFC. In this work, the LD was focused on the sample stream after

it had left the cuvette as a liquid jet. In a commercial SFC the laser is focused through

the optical quality quartz walls of the cuvette. The goal of the SFC experiment was to

provide a sample stream diameter equal to the smallest obtainable LD beam waist (52 Jim).

To achieve this, the liquid jet exiting from the SFC was viewed under 40X magnification

with a microscope. A concentrated solution of the brilliant red dye rhodamine 6G was used

as the sample; this allowed the diameter of the sample stream to be measured. The sheath


flow was adjusted by changing the pressure of nitrogen in the ballast tank, and the sample

flow was adjusted with the syringe pump controls to give a sample stream diameter of

47 pAm. The settings were recorded and duplicated with dilute IR 140 standards for the

fluorescence experiment.

The optics for the sheath flow cuvette fluorescence spectrometer were similar to the

optics for the liquid jet system. A single-mode LD (Mitsubishi ML 4102) had a smaller

emitting aperture and smaller beam divergence. It was collimated (D.O. Industries

collimating lens) and sent into the 20X microscope objective, where it was focused into the

liquid jet flowing out of the sheath flow cuvette. The beam waist measured 52 pm x 9 pm.

The sheath flow cuvette, like the liquid jet, was mounted at a 45 angle with respect to the

optical axis and parallel to the entrance slit. The fluorescence was collected with a 35 mm

focal length lens and imaged 1:1 on the 1.25 mm entrance slit of the 0.25 m

monochromator (Spex Minimate). A cooled PMT (Hamamatsu R928) operated at -900 V

was used for detection. No optical filter was used.

In the compact cell glass filter fluorometer system, the laser diode was placed as

close as possible to the cuvette, and a low noise, red sensitive Hamamatsu S2386-45K

photodiode was also placed in close proximity to the cuvette. This detector photodiode had

a sensitivity of -0.6 A/W at 830 nm and a spectral range of 400 to 1100 nm. The colored

glass filter was the only optical element used in this simple system as shown in figure 4-5.

The frequency-doubled laser diode described in Chapter 3 was used to excite

perylene fluorescence in a cuvette. The experimental setup was the same as the cuvette

fluorescence spectrometer described earlier. The excitation wavelength was 398 nm, and

fluorescence was collected at 464 nm.





Col 1 i mat or


Ca .

Syri r
P m9

LD ---
Anamor phi c \ -
Prism Pair

Sa m

Figure 4-3. Schematic of the liquid jet fluorescence spectrometer system.




El ect ro

me ter

._- s a mp 1 e -..

o---M ol vent .

a b

Figure 4-4. Sheath flow cuvette.
a) complete view; b) closeup of tip. The dashed line represents the sample
stream being hydrodynamically focused.

Figure 4-5. Schematic of the compact cell filter fluorometer.

Results and Discussion

All laser diode fluorescence configurations resulted in high sensitivity although the

systems with the cooled PMT were superior to the photodiode system (25). The best

concentrational detection limit was 5.0 x 10"14 M, achieved with the cuvette fluorescence

spectrometer. The collimated LD beam passed through the cuvette with very little scatter;

the blank noise was only 70 pA. If the cuvette had scattered significant laser light the

blank noise would have been much larger, since a large (64 nm) bandpass was used with

no optical filter.

Replacing the monochromator with an interference bandpass filter increased the

sensitivity of the measurements by a factor of 28, but cost a factor of 44 in laser scatter

noise. The effectiveness of the interference filter was limited by the large solid angle of.

laser scatter striking it. Interference filters must be placed in a collimated beam to achieve

the specified bandpass. With an optimum combination of filters, this system could obtain

the lowest concentration detection limits, because the throughput and collection efficiency

of a filter based detector exceeds a system using a monochromator. A comparison of the

performance of the four fluorescence systems is given in Table 4.1. In all cases, the limit

of detection (LOD) was determined by noise on the scattered laser light reaching the


The liquid jet fluorescence spectrometer system gave a very low absolute LOD, since

this technique involved a very small detection volume. The detection volume is described

as cylinder with diameter defined by the i.d. of the capillary and length given by the width

of the focused laser beam. The logarithmic calibration curve had a slope of 1.05 and was

linear over six orders of magnitude; the sensitivity was 0.029 nA mL pg"'. The


concentrational LOD was calculated by dividing the RMS noise on the blank by the

sensitivity (linear calibration curve slope) and multiplying by 3. The RMS blank noise was

estimated by dividing the peak to peak fluctuation by five. The IUPAC recommended

value of k=3 standard deviations of the RMS noise (26) was used to calculate the LOD.

Based on the concentrational limit of detection and the volume of sample, the

absolute LOD is 59 ag, or 46000 molecules, with a I s time constant of the measurement.

With the liquid jet fluorescence spectrometer, a flow rate of 6.6 mLUmin was used; the mass

detection limit was calculated using equation 4.1:

Mass Detection Limit = cone. LOD x flow rate x time constant (4-1)

=(1.05 ax0"- g/L)(1.11 x 10-4 Ls)(1 s) = 1.2 x 10-13 g

For a molecular weight of 779 dalton, this corresponded to 9 x 106 molecules. The LOD

was determined by the noise on the laser scatter. Background from solvent contaminants

was not observed, since there are few molecules which fluoresce in the red spectral region.

Positioning the liquid jet at a 450 angle with respect to the optical axis and parallel to the

entrance slit greatly reduced the scatter reaching the monochromator entrance slit (27)

although a higher sample flow rate was required to achieve laminar flow.

Figure 4-6 shows the amounts of background corrected analyte signal and

background for monochromator entrance slit widths ranging from 5 mm to 0.5 mm. The

exit slit width was held constant at 5 mm since the modifications for the cooled PMT made

changing the exit slit difficult. The signal to background ratio was optimum for a 2.5 mm

entrance slit width. Also shown are the signal and background with filters placed in front

of the monochromator with 5 mm entrance and exit slits. The two filters tested were: a


10 nm bandpass interference filter at 830 nm (Corion), and an RG-850 colored glass filter

(long-pass, 50% transmittance at 850 nm, Corion). A more important quantity for

optimizing the system is the signal-to-noise ratio (S/N). Figure 4-7 shows the S/N advantage

gained when filters are available to reduce the laser scatter reaching the PMT. The

RG-850 colored glass filter in front of a monochromator with 2.5 mm slits was chosen for

the liquid jet fluorescence spectrometer system; the filter increased the signal-to-noise ratio

by a factor of 33.

The compact cell glass filter fluorescence system gave poorer detection limits than

the other three fluorescence setups discussed previously (see table 4.1). This instrument

had an uncollimated laser beam for excitation and a photodiode for detection, achieving

simplicity but a higher LOD. The uncollimated LD has a divergence angle of 350 by 100

which resulted in inefficient irradiation of the detection volume, as well as greatly increased

scatter. The lack of photodetector gain also limited the LOD of the cell. The photodiode

had a higher sensitivity at wavelengths greater than 900 nm compared to PMTs used in the

other fluorescence spectrometer systems; this could have resulted in an interference from

Raman scatter passed by the glass filter. The advantage of this instrument is its small size

and simplicity. With a battery operated laser diode, it could be completely portable.

The system with the most potential for achieving single molecule detection is the

sheath flow cuvette fluorescence spectrometer. With a sample flow rate of 22 pULmin and

a sheath flow rate of 20 mL/min, a stream diameter of 47 pm was attained. The 3 mW

single-mode laser diode was focused to a beam waist of 52 pm in this stream. A probe

volume of 9.0 x 108 cm3 was calculated for the cylinder with diameter equal to the stream

diameter and length equal to the diameter of the focused LD beam. A calibration curve

5mm 2.5mm 1.25mm 0.5mm INT830 RG850

entrance slit width

E Signal

M Background

Figure 4-6. Signal and Background for IR140 fluorescence excited in a liquid jet. Exit
slit width was constant at 5 mm. INT830 is an interference filter placed in
front of the monochromator; RG850 is a colored glass filter.








6 -

4 F

5mm 2.5mm 1.25mm 0.5mm INT830

Entrance Slit Width

M S/N ratio

Figure 4-7.

Signal-to-noise ratio for a range of slit widths, and with filters before the



(slope 8.84 pA/ng/mL, correlation coefficient 0.9997) was plotted for IR 140 in the SFC

and is shown in figure 4-8. From the blank noise (5.6 pA) and the slope of the calibration

curve, a concentration detection limit of 1.9 ng/mL was calculated. Using a simple

concentration X volume equation, this corresponds to an absolute detection limit of 130,000

molecules. Taking into account the time constant of the system with equation 4.1, the mass

concentration detection limit was 2.1 x 109 molecules. The mass concentration detection

limit allows a direct comparison with the work of Dovichi and Keller. The sheath flow

system used in this work achieved an LOD which is five orders of magnitude more than

the best LOD for an SFC. The work was carried out with optics and filters available in

the laboratory. For future work, optics should be acquired which enable spatial filtering

to be performed so that laser scatter from areas other than the detection volume do not

reach the detector. Also, a search should be made of filter vendors to obtain optical filters

which more effectively discriminate against laser scatter, allowing IR-140 fluorescence to

pass. The detection volume was limited by the LD beam waist of 52 pJm. A high-power

(50X), long focal length microscope objective would enable the beam to be focused smaller,

to about 9 pm.

Three of the five laser diode fluoromeler systems achieved concentrational detection

limits of parts per trillion and below. The fluorescence systems based on a liquid jet and

a sheath flow cuvette achieved an absolute detection limit of 46000 molecules and 130000

molecules, respectively. Dovichi and Keller (20) stress that a sheath flow cuvette system

is designed to detect a small number of molecules not necessarily at a low concentration.

By using a laser diode with higher irradiance and a sharp cut filter to minimize pickup of


laser scatter, the LODs for all five fluorescence systems listed in table 4.1 would be


The frequency-doubled laser diode was used successfully to excite molecular

fluorescence of perylene. The concentration LOD was 0.84 ng/mL, and the absolute LOD

was 110 pmol.


Q -)

E 10'

D 10' 10' 10'
Concentration (ug/L)

Calibration curve for IR-140 fluorescence excited in a sheath flow cuvette.
The calibration curve is linear over at least 3 orders of magnitude.

ix ~ ""'Oiiiiii~ ii i ii" ii"ii i >

iiiii ''iiiiii;iiii iiii0i "" l;i"';;" ""i lO"''i .ii _, r.l; s ,i"..".:"i"
AayiaFiueofMrt for DfentIsrm tal Cofigurations
Sheath Flow Fluorescence 2. x 106 2. x 10-9 1. x 105 9. 6. x 10-.1 2. x 109
"iii~ i ........iii" i

"'iiii ....ii;:8iii~; iR;i i;ii$ .....~s i iii$ i t i i iO
;li: :i,,00 .I; ..8:p: 81,; lls;

Cuet loecec .x1 01 -xle9 o .x 10 5. x 10 I
Cuvette Filter 7. x 10 9. X 10-14 3. x 10 2. x 104 3. x 103 5. x 10

Compact Cell Glass 6. x 104 7. x 10-11 3. x 1011 2x 10 7. x 10' 8. x 10';
F;;ilter Fluoromet
....a:l; ,;ii, s;ii ~ iiililii ii



The use of atomic fluorescence spectroscopy (AFS) as an analytical method was first

suggested by Alkemade in 1962. Winefordner in the U.S. and West in England pioneered

the technique in the mid 1960's (28). In AFS, the sample is atomized and the element of

interest is excited by optical radiation or a combination of optical and thermal radiation.

The selected element then emits light at a characteristic wavelength. AFS provides greatly

increased selectivity compared to atomic absorption and emission techniques, since there

are two methods of selection used. The source is tuned to excite only the element of

interest, and the luminescence provides a second selection when a monochromator or filter

is used to pass only light collected at the fluorescent wavelength. The invention of the

tunable dye laser provided an almost ideal excitation source for AFS. Experiments in laser

excited atomic fluorescence spectroscopy (LEAFS) were reviewed as early as 1973 (29).

Lasers have extremely high spectral irradiance, sufficient to saturate a transition in some

cases. Saturation is defined as promoting the maximum percentage of atoms to the excited

state. In a saturated transition, half of the atoms will be in the excited state, and half in

the ground state.

There is one commercial AFS instrument using hollow cathode lamps for excitation

(30, 31), but LEAFS has never become a commercial technique. The main reason for this


is the cost and complexity of the laser source. Dye lasers require eight different dye cells

and six frequency doubling crystals to cover the wavelength range from 217 nm to 950 nm

(32). Expertise is required to tune the laser to the atom line, and this task must be

performed for each element to be determined. There are many applications where LEAFS

has unmatched qualifications, making cost and complexity less important. In these

applications, an element must be determined at ultratrace levels in the presence of a

complex matrix a matrix which would cause spectral interference (overlapping lines) with

emission techniques. Examples are the elemental analysis of semiconductors and mixtures

of metals in radioactive wastes.

A laser diode source for AFS would be free of the cost and size disadvantages of

tunable dye lasers. Individual LD based modules for determining elements would be smaller

than the element modules on the hollow cathode based instrument. Table 5.1 lists the

atomic transitions which could be excited in a graphite furnace with commercially available

laser diodes. The cost of an LD is about $30, vs hundreds of dollars for a hollow cathode

lamp and tens of thousands of dollars for a tunable dye laser.

Atom Reservoirs

An atom reservoir is a device to produce atoms from the sample and hold them for

a period of time sufficient for spectrochemical analysis. Common atom reservoirs for AFS

experiments are the inductively coupled plasma (ICP), the resistively heated graphite

furnace, the glow discharge, and various flames. In one experiment to be described, a high

concentration of Rb (1000 ppm) was aspirated into an acetylene/air flame for the purpose

of tuning the LD to a Rb spectral line. However, the analytical data was all obtained in

Table 5.1
Strong atomic transitions beginning at energy levels near the ground state,
which are accessible with commercially available laser diodes.

Energy Emission
Element Wavelength levels (cm-1) Intensity gA(10E8/s) gf

Ba 7911.34 0-12637 80 0.0027 0.0026
Ba 767.209 9034-22065 500 0.64 0.57
Ba 778.48 9216-22065 300 0.39 0.36
Ba 705.994 9597-23757 2.7 2.0
Ba 712.033 9034-23074 0.3 0.23
Ba 728.030 9216-22947 1.5 1.2
Ba 748.808 9597-22947 0.27 0.23
Cs 852.124 0-11732 15000 1.3 1.4
Cs 894.359 0-11178 8000 0.48 0.57
Dy 775.732 9211-22099 5
Dy 783.27 7050-19813 7
Er 840.909 0-11887 55
Er 847.242 0-11799 35
Er 876.864 0-11401 22
Eu 686.454 0-14564 360 0.049 0.035
Eu 710.648 0-14068 100 0.016 0.012
Li 670.784 0-14904 480 1.2 0.8
Li 812.652 14904-27206 0.85 0.84
K 766.490 0-13043 18000 1.6 1.4
K 769.896 0-12985 9000 0.78 0.7
Rb 780.023 0-12817 30000 3.0 2.7
Rb 794.760 0-12579 15000 1.3 1.2
Sm 686.093 293-14864 0.11 0.078
Sm 710.454 2273-16345 0.066 0.050
Ti 841.236 6599-18483 0.045 0.048
Ti 842.652 6661-18525 0.063 0.067
Ti 843.494 6843-18695 0.19 0.21
Ti 843.570 6743-18594 0.093 0.099
Ti 867.539 8602-20126 0.055 0.062
U 761.935 7646-20766 0.083 0.073
U 788.194 6249-18933 0.16 0.15
U 875.569 10347-21768 0.15 0.17
V 811.680 8716-21033 0.062 0.061
V 891.985 9824-21033 0.034 0.041
Zr 709.770 5541-19626 0.15 0.11
Zr 695.384 5249-19626 0.039 0.028
Zr 716.909 5889-19834 0.17 0.13
Zr 807.008 5889-18277 0.16 0.16


either a graphite furnace or a glow discharge. Only these two atomizers will be described

in detail.

The graphite furnace, also called the electrothermal atomizer (ETA), is commonly

used as an atom reservoir for atomic fluorescence, with many commercial designs available.

Heating is provided by passing a current of hundreds of amperes at 10 to 12 V through a

graphite tube 2 to 3 cm long and approximately 5 mm in diameter. In most designs, a very

small sample (1-20 ptL) is pipetted through a hole in the top of the graphite tube. The

amount of current is changed to provide three distinct heating stages. In the desolvation

step, the tube is heated to about 110 oC for several seconds while the solvent is

evaporated. Then the current is boosted to provide a temperature of 400-1100 oC for the

ashing step. In this step organic compounds and some inorganic interferents present in the

matrix will be destroyed or volatilized. Finally, for the atomization step, the temperature

is raised very quickly to a temperature of 2000-3000 oC. The sample is vaporized off of

the tube wall and a cloud of atoms is produced within the tube. This cloud of atoms is

then probed with a light beam at a wavelength absorbed by analyte atoms. The graphite

furnace provides excellent absolute detection limits because the amount of sample is small,

and the volume of atomic vapor produced is also small. The atmosphere of the graphite

furnace is free of fuel and oxidant gases reducing chemical interference such as the

formation of refractive oxides.

Neuman and Kriese used a carbon rod as an atom reservoir for atomic fluorescence

as early as 1973 (33). Graphite cups (34) and graphite tubes constructed specifically for

AFS analysis (35, 36) have been employed as atomizers. Graphite cups were used in order

to avoid collecting blackbody radiation from the heated graphite. However, severe matrix


interference can develop due to the large temperature gradient between the atomizer and

the atomic cloud above it. In some cases, condensation will occur in the cooler region

above the cup, causing unacceptable levels of non-specific laser scatter. Tubes are

preferable, and have been used for samples such as metals, seawater, and biological

materials. There is a trend toward using commercially available ETAs from atomic

absorption (AA) instruments in order to take advantage of the temperature programming

and matrix modification experience gained through the extensive routine use of AA on real

samples (37, 38). To use a graphite tube without modifications such as drilling holes in the

side, the fluorescence may be collected with the pierced mirror design of Goforth (39).

Devices invented for the minimization of matrix interference in AA, such as the L'vov

platform, are also finding use in AFS (37).

The other atom source used in laser diode excited atomic fluorescence (LDEAFS)

was the glow discharge. Familiar discharges are sparks, lightning bolts, and arcs like those

in the arc lamps that light our cities. In a glow discharge, a current of a few milliamperes

flows from a cathode in the form of a plate or cup to a cylindrical anode nearby. The

interior of the discharge source is evacuated and the air is replaced by approximately 10

mtorr of argon or helium. The cathode emits electrons under the bombardment of particles

and light quanta from the gas. Under this bombardment, the material of the cathode is

eroded into the discharge by a process called sputtering. Grim made use of this sputtering

in 1968 to produce an analysis technique where the sample was used as the cathode and

the emission spectrum of sample elements was recorded (40). He used the method for

conducting samples like metal plates. Gough et al. studied atomic fluorescence in a cathode

sputtering source (41). Their samples were also metal alloys, in which impurities were


determined at the level of 20 ppm or more. Hollow cathode lamps provided excitation for

atomic fluorescence.

The glow discharge used for these studies, based on the design of Glick et al. (42),

may be used for nonconducting samples as well. A schematic of the glow discharge is

shown in figure 5-1. A graphite cup is used as the cathode, and the sample is placed in

the bottom. Aqueous rubidium standards were determined by pipetting the solution into

the graphite cup and evaporating it with a heat lamp. Then the cup was introduced into

the glow discharge where the rubidium was sputtered off to be excited by the laser diode

source. Gough points out some differences between a glow discharge and other atom

reservoirs for AFS, such as flames. The low pressure of the discharge makes collisions

between atoms less likely, therefore the widths of atomic absorption and fluorescence lines

are reduced to the limit of doppler broadening. Also, there is reduced thermal mixing of

energy levels such as the resonance lines of rubidium. This has important consequences

for AFS schemes that rely on thermal mixing.

Previous Work

Laser Diodes have been used as excitation sources by atomic spectroscopists in

physical studies due to their low cost, narrow bandwidth (single-mode devices) and

tunability. Budkin et al. used LDEAFS for determining the concentration of Cs in a vapor

cell used as a quantum frequency standard (43). Lawrenz et al. used LDs to provide one

step of a two photon LEI scheme for determination of Ba isotope ratios with thermionic

diode detection (44). For frequency sensitive applications such as interferometry, the

output from a laser diode may be stabilized by reference to a rubidium line. Barwood, et


vacuum chamber
a terr eo



IlfJl, lllllllllll llllll lllug -ur*nr* n n -suu-n a- r* -.- i a.. =
"l*Imimlrl ell I ll lllllllll l l ll ll l l ll l l l

------------Ill- l ll- lllll-- 11--I1-l1---I---n-Il-Ai--I-I-.
-1 -, DO I as m o u n ,

p Illlpllll lllllupl
---------I-IIII II-II I IIISI- SI-II-I-I-I-II l--I
I asul a1t l e 1rthldli

i--lIewviag portI aaauI-nu
11- aIe

Mat erl aIs


li br a@ 1


igt ml ea

Figure 5-1. Schematic of the glow discharge atom reservoir.




al. used this technique for both frequency and intensity stabilization (45). Their goal was

to produce a cheap and easy to use alternative to the 633 nm He-Ne laser for

interferometry. Ultralow (10 kHz) linewidths have been achieved for LDs used in

communication applications (46) using external cavity optical feedback. These LDs have

a narrow enough linewidth to allow spectroscopists to study the hyperfine structure of the

Rb D lines as well as isotope shifts.

High-power LD Array Excitation


Fluorescence excitation at 794 nm was provided by a 200 mW c.w. 10-stripe laser

diode array (Spectra Diode Labs SDL-2422-H1) operated at -7.6 oC. The laser beam was

collimated and shaped into a round beam using a collimating lens followed by an

anamorphic prism pair (Melles Griot). The collimated beam was modulated and focused

through a hole in a fat mirror into a graphite furnace (Varian CRA 90) as described by

Goforth and Winefordner (39). This radiation excited the Rb atoms from the ground

state to the 2p2 state. Thermal energy provided additional excitation energy to populate

the 2p, state. Radiational decay occurred from the 2p, state to the ground state. Five

microliter standards of rubidium chloride (Fisher) dissolved in 1% nitric acid were

introduced into the graphite tube where they were dried, ashed, and then atomized at

1800 oC. The resulting rubidium fluorescence at 780 nm was collected with the mirror

and focused on the 1.25 p&m slit of a Spex Minimate monochromator. The signal from the

photomultiplier tube (Hamamatsu 955) was processed by a lock-in amplifier

(EG + G PARC Model 186A) and quantified as the fluorescence peak height on a chart


recorder (Fisher Record-All). A schematic is provided in figure 5-2. Tuning of the laser

wavelength to the atomic absorption line was accomplished by changing the laser diode

temperature while monitoring the fluorescence signal from a flame into which a high

concentration of rubidium was aspirated.

Results and Discussion

Tuning of the laser source to the atomic transition proved to be difficult to achieve,

even with a continuous source of atoms such as a flame. Finding the line required tedious

temperature tuning, and some days tuning was not successful at all. Part of this difficulty

was due to the near-infrared background emission from the flame used for tuning. Perhaps

better use could be made of tuning of the LD by changing the current applied, since this

provides a finer adjustment than temperature tuning. Also, a continuous atomizer with a

lower background could be employed, such as a glow discharge or a heated cell containing

Rb metal. Once tuned, the laser wavelength remained at the line for several days even

after power was turned off, as long as the temperature and power settings for the laser

diode were not changed.

A detection limit of 2.1 pg was obtained for rubidium in water. This limit could be

extended lower by the use of filters to screen the furnace emission in the areas of the

spectrum away from the fluorescence line being monitored. This emission appears as stray

light at the PMT and is the limiting noise in the system, even with lock-in electronics being

used to measure the signal. Initially the sensitivity of the preamplifier and lock-in amplifier

were set at a sensitivity where stray light from laser scatter was being recorded. At this

setting, it was expected that the AFS signal would far surpass the scatter signal. However

when the graphite furnace entered the atomize stage the furnace emission overloaded the


preamplifier and lock-in amplifier, and the sensitivity had to be reduced. A bandpass filter

centered at 780 nm could be used, since cutoff filters are not readily available in the near-

infrared region as they are in the visible and ultraviolet region of the spectrum. An

analytical calibration curve for the technique is shown in figure 5-3. The linear dynamic

range extends from the detection limit up to 100 ppb, where either analyte self-absorption

or saturation of the PMT takes place.

The LD used in this experiment is classified as a "high-power LD array", but the

radiation is not used efficiently in this experiment. The broadband output of an LD array

has a spectral emittance of 5.2 x 104 W/cm2 nm. The 100 ppb Rb standard gave a PMT

signal current of 1.3 x 10' A. Using the cathodic sensitivity of the PMT (15 mA/W at 780

nm), the PMT quantum efficiency (0.025), and the PMT amplification (107), this

corresponds to an atomic fluorescence radiant power of 3.5 x 10'7 W (see equation 5-4).

This assumes a monochromator throughput of 10% and a fluorescence collection efficiency

of 1%.

The laser diode was considered to be a continuum source with (0 m)0

7.0 x 105 W/cm since its linewidth was much larger than the Rb absorption linewidth. This

linewidth was estimated from the Voight profile for Rb absorption in a graphite furnace.

The mechanism of broadening was a combination of Doppler broadening and collisional

broadening. The Doppler broadening in the graphite furnace at T = 2000 K for Rb

(atomic weight M = 85.47 dalton) absorption at Am = 780 nm is given by the expression


AMD = 7.16 x 10-7 (T/2/M1/2) ,m


Monochromat o


Col I mat or

DI ode

Anamorphi c
Pri sm Pai r



Ampl I f I er

Re o d I

\ V._ phi Stop

de? Graphite
Mi rror Furnoce
i __-------- ,J

Figure 5-2. Schematic of the system for high-power LD array excitation of Rb fluorescence in a graphite furnace.



I I I I I 1



Figure 5-3.


I I I I 1 11 |

Calibration curve LD excited fluorescence of Rubidium in a graphite





1 I 11


The Doppler width AID was 0.0027 nm, and the collisional linewidth ML was estimated to

be 0.0040 nm (6, p. 211). The combined linewidth is given by:

Av = AvL/2 + [ (AvL/2)2 + (AvD)2 1/2 (5-2)

and calculated to be 0.0053 nm. The value of the fluorescence radiant power O'FC

expected for this experiment may be calculated using the expression for AFS radiant power

with -continuum source excitation (6, ch. 7):

e2 2 1 ().m)O Y n! fi
*'FC =-e m U (5-3)
4 EO me 2

where e is the elementary charge (C), A is the wavelength (cm), I is the path length (cm),

(OAm)o is the spectral radiant power at the absorption line center (W/cm), Y is the
fluorescence power yield dimensionlesss), ni is the Rb number density (cm"3), fU is the

absorption oscillator strength (dim.), co is the permittivity of free space (C2 N'1 cm'2), me

is the rest mass of an electron (g), and c is the speed of light in a vacuum (cm s'1). The

oscillator strength was taken as 0.68, the path length as 0.4 cm, and the Rb number density

of 2.5 x 1014 cm"3 was calculated from the amount of Rb in the 10 /L standard and the

graphite furnace volume. A fluorescence flux of 0.25 W was calculated. This differs

considerably from the experimental value of 3.7 x 10'7. One possible reason is the very

high background from graphite furnace blackbody radiation, which was the limiting noise

in this experiment, and may have saturated the PMT. This emission, viewed with a 4 nm

monochromator bandpass over the 0.28 cm2 area of the graphite furnace was estimated


using the photon flux of a blackbody at 1800 oC (1020 photon/cm2 s pm). A value of

10'2 W was calculated, which was more than the measured fluorescence flux. Thus the

lock-in amplifier had to discriminate against a very high background.

Single-mode LD Excitation


A single-mode c.w. laser diode (Mitsubishi ML 4102) was operated at 4 mW in a

thermoelectrically cooled mount (ILX Lightwave Model LDM 4412) at -10 oC with a

Spectra Diode Labs driver (M800). The wavelength was temperature tuned to the 780 nm

absorption line of Rb. The LD beam was modulated with a chopper wheel referenced to

a lock-in amplifier. A block diagram of the system is shown in figure 5-4. The high beam

quality of the single-mode LD made the anamorphic prism pair unnecessary, and no

focussing optics were necessary to illuminate the target. The atomizer was a laboratory

constructed (42) glow discharge operated at a voltage of 600 V, a current of 40 mA, and

a helium pressure of 6 torr of He. The vacuum chamber surrounding the cathode was first

evacuated; then He was leaked into the chamber with a needle valve at a sufficient rate

to maintain a pressure of 6 torr. The hollow cathode (depth 4.5 mm, diameter 3.4 mm,

figure 5-5) was constructed of graphite. Fifty /L aqueous rubidium chloride standards were

pipetted into the hollow cathode and evaporated under a heat lamp. The cathode was then

mounted in the vacuum chamber, which was evacuated with a pump (Balzers) and replaced

with a 6 torr pressure of flowing helium. Argon was not used as a fill gas because of

interfering Ar emission lines. The voltage was then applied to initiate the discharge, and

after a period of 2 min (for stabilization of the sputtering), AFS measurements were taken.

Lock- in

Ampl i f i er
Monochromat or


Le a s

Sa mp 1 e

-1- Gl ow
Chopper/\ I i mirror Di s charge

Figure 5-4. Schematic of the system for single-mode LD excitation of Rb fluorescence
in a glow discharge. The LD beam was modulated with a chopper.






Figure 5-5. Diagram of the sample holder of the glow discharge atom reservoir, showing
a closeup of the graphite electrode/sample holder.


The LD illuminated the hollow cathode through a pierced mirror, and the fluorescence was

reflected by the mirror and imaged on the 1 mm entrance slit of a 0.3 m monochromator.

The PMT was a Hamamatsu model R955 operated at -900 V. The LD could be tuned to

the Rb line at 780 nm starting from a range of temperatures. Tuning was accomplished

by very small changes in the driving current of the LD less than 0.1 mA. The current

control knob of the Spectra Diode Labs LD power supply was actually too coarse for this

adjustment, which required some dexterity. The lock-in amplifier was referenced to the

chopper which modulated the LD beam.

In another experiment, frequency modulation was performed with a waveform

generator (Wavetec) which was used to impose a sinusoidal wave in the amplitude of the

current supplied to the LD. The amplitude of the wave was 4 mV peak to peak and the

DC offset was 50 mV. This voltage was supplied to an analog input on the LD driver,

which put out a proportional current (1 A/I V). This alternating current brought the LD

onto the Rb absorption peak (780 nm) and off of it at a frequency of 100 Hz. This is

shown schematically in figure 5-6. The signal from the PMT was measured with a lock-in

amplifier referenced to the modulation waveform supplied to the LD driver no chopper

was used. See figure 5-7 for a block diagram of the frequency modulation system.

Results and Discussion

The single-mode LD required tuning by turning the current control of the driver

by very small increments, yet the tuning was reproducible. With care, the LD could always

be tuned to the line. The absolute detection limit was estimated to be 200 ng. This is a

poorer LOD than that obtained in the graphite furnace. One reason for the low sensitivity

is the low rate of release of analyte from the cathode by the sputtering process. The

Rb absorption

t i me

Figure 5-6.

Schematic representation of frequency modulation of a laser diode The
dashed wave represents modulation of the current supplied to the LD. The
solid wave represents the LD emission wavelength. This wavelength is
brought on and off the Rb absorption line, represented by the solid line.

r Lock- in

A mp 1 i f i er

Functi on

I I e!n

D rro Sampl e

GI Gow
Mi rror Di a charge

Figure 5-7.

Schematic of LD excited Rb fluorescence using frequency modulation for
background correction.


sample in our experiments was consumed over a period of one hour. The LOD based on

the rate of release of Rb from the cathode was approximately 60 pg/s. A Rb number

density of 109 cm'3 was estimated based on a number density obtained by van Dijk et al.

(47) in a glow discharge under similar conditions. In a source like the graphite furnace,

a sample of similar volume would be atomized into a reservoir of similar volume in less

than one second. The single-mode LD used to excite Rb fluorescence in the glow

discharge was much narrower spectrally than the high-power LD, yet it was still considered

a continuum source since the LD linewidth was much larger than the Rb absorption line

width. The absorption profile of atomic lines is narrower in a glow discharge than in a

graphite furnace because of the low pressure maintained in the glow discharge. An

absorption linewidth of 1.4 x 10'3 nm was calculated with equation 5.1, assuming only

Doppler broadening in the glow discharge at 500 K. This is narrower than the LD

linewidth of 2 x 10-2 nm, (but the single-mode laser approaches the width where it could

be considered a line source). The fluorescence radiant power V'FC was calculated from

equation 5.3; for this experiment (0.m)o = 2 x 106 W/cm, I = 0.3 cm, Y = 1, fU = 0.63,

and n1 = 109 cm"3. A value of *'FC = 2 x 106 W was calculated. The experimental value

for the fluorescence radiant power from the 1000 ppm standard (0ep) was calculated from

the instrumental settings and the signal voltage measured from the chart recorder.

Voltage measured x collection efficiency x monochromator throughput
= (54)
amplification x PMT cathodic sensitivity x K(A) x PMT internal gain

The voltage measured 6.04 mV, the factor for collection efficiency was estimated as 100,

the factor for monochromator throughput was estimated to be 10. The amplification was


105 V/amp, the PMT cathodic sensitivity was 1.5 x 10"2 A/W, K(A) the quantum efficiency

of the PMT at 800 nm (0.025), and the PMT internal gain was 107). A experimental

fluorescence radiant power of 2 x 108 W was obtained.

The Stokes fluorescence scheme for rubidium excitation with the single-mode LD

involved excitation to the 2p, level followed by collisional deactivation to the 2p2 level and

relaxation to the ground state with release of a photon. At the low pressure of the glow

discharge, collisions between particles during the lifetime of an excited atom are rare (41).

The scheme just stated relies on these collisions for the fluorescence signal. This may

explain why the experimental fluorescence radiant power was two orders of magnitude lower

than the value predicted by theory.

The anti-Stokes version of this scheme excitation to the 2p2 level with collisional

excitation to the 2p, level would be even more difficult since the atoms would have to

gain energy from collisions rather than lose it. This experiment was attempted with the

high-power LD array. The attempt was unsuccessful due to the anti-Stokes nature of the

transition, and the less efficient excitation with the broad band LD. No single-mode LD

was available for excitation at 794 nm. Single-mode LDs proved difficult to obtain at

arbitrary wavelengths since few retailers would select an individual laser since the quantities

purchased for research were small. Resonance fluorescence of rubidium at 780 nm was also

unsuccessful, this time due to the optics used with the glow discharge. Collection of

fluorescence with the pierced mirror gave a tremendous signal from laser scatter off of the

electrode surface. The techniques of amplitude and frequency modulation were

unsuccessful at discriminating against this scatter.


The main advantage of the He glow discharge is the low spectral interference in

the NIR region. Fluorescence excitation using a frequency modulated LD was successful,

but the LOD was twice as high as single frequency excitation using a chopper for

modulation. The most likely cause for this is the low duty cycle of the LD, which caused

a small amount of time spent on the Rb absorption line compared to the total time the LD

was on. The LD was on the Rb line only at the peak of the wavelength modulation. One

way to improve this would be to decrease the modulation depth, but this would require

electronics for producing a very low amplitude waveform (0.3 mA) which were unavailable

in the lab. This technique should give better correction for spectral and background


Frequency-Doubled LD Excitation

No gallium fluorescence was detected when gallium standards were injected into the

graphite furnace, atomized, and excited with a frequency-doubled laser diode at a

wavelength of 403 nm. This laser was relatively broadband (FWHM = 0.5 nm), and also

of extremely low power at 100 nW. Since the emitting aperture was the same as the high-

power LD, this corresponds to a spectral radiance of only 0.1 W/cm2 nm. Tuning to the

Ga line in a flame was more difficult because the doubling crystal had to be angle tuned

again each time the fundamental LD wavelength was temperature tuned by more than a

few tenths of a nanometer. Concentrations as low as 10 ppb and as high as shavings of

pure gallium ingot were atomized but no fluorescence was measured.


The laser diodes used did not achieve saturation of the Rb fluorescence transitions

that were excited. Es, the irradiance required for saturation, is given by:

Es = 1.52 x 1017 A-5 y-1 [g1 /(gu + gl) (5-5)

where A is the wavelength of the transition (nm), Y is the fluorescence power yield

dimensionlesss), and g. and gl are the statistical weights (2 and 4) of the upper and lower

states, respectively. The LDs fall short of the calculated value, 175 W/cm2 nm, by a factor

of 100.

Two of the atomic fluorescence systems were successful. The first was the excitation

of rubidium in a graphite furnace with a high-power LD array, with a limit of detection of

2.1 pg. Advantages of this system were the efficient production of atoms in the graphite

furnace, and the small amount of laser scatter. Disadvantages were: low spectral brightness

of the LD array, which did not efficiently excite the Rb atoms, and intense thermal

emission from the graphite furnace, which limited the sensitivity of the system. The second

successful fluorescence system was the excitation of rubidium in a glow discharge with a

single-mode LD. This system had an LOD of 200 ng. Advantages were the high spectral

radiance of the LD, and the low background emission from the He glow discharge. The

system suffered from a large amount of laser scatter, which prohibited resonance

fluorescence detection. Also, the sample was sputtered off of the cathode over a long

period, which limited the sensitivity.


Laser diodes are small, inexpensive and rugged. They are easy to mount and align,

and easy to transport between experimental setups. An optical rail carrying the LD mount

and collimating and focusing optics could be mounted on a small optical rail to facilitate

moving the LD between setups. Two types of LDs were used to excite fluorescence the

high-power LD array, which offers high power with a relatively large bandwidth, and the

single-mode LD of lower power and narrower line width. The spectral characteristics of

laser diodes were examined with a monochromator, a diode array detector, and a Fourier

transform Michelson interferometer. The LDs were tuned over a wavelength range of

20 nm with a temperature controlled mount. For fine control of the wavelength, the LDs

were tuned over a 1 nm range by varying the drive current.

Laser diodes are presently available commercially only in wavelengths longer than

670 nm (the red, NIR, and IR regions of the spectrum). Until shorter wavelength LDs are

fabricated, the number of atoms and molecules which can be excited with these devices will

be limited. In order to excite a larger variety of atoms or molecules, experiments in

frequency doubling of the laser diode were carried out. This technique produced a very

low-power blue laser source. Experimental plots of frequency-doubled power vs

fundamental power were compared to results predicted by theory. Temperature tuning of

the frequency- doubled LD was demonstrated.


Both high-power and single-mode LDs were used to excite rubidium atomic

fluorescence. The graphite furnace and the glow discharge were studied as atom reservoirs.

Success was achieved in exciting rubidium fluorescence. A limit of detection of 2.1 pg was

obtained in a graphite furnace with high-power LD excitation. The graphite furnace had

a high atomization efficiency, but also a high blackbody emission background in the near

infrared region. Also, the high-power LD with its large linewidth was not well suited for

exciting atomic transitions. In the glow discharge, the LOD of Rb was 200 ng. Since the

sample was sputtered slowly off the cathode, the sensitivity of this technique was limited.

The glow discharge has a low emission background and is a continuous atom reservoir

which makes it convenient for tuning the LD to an atomic line. Background correction in

the quantitative AFS experiments was performed by chopping the LD beam in phase with

a lock-in amplifier. Frequency modulation (FM) was also successful with single-mode LD

excitation in the glow discharge. This technique provided discrimination against laser

scatter, an advantage over mechanical chopping. FM would be useful in a commercial LD

based AFS system, since there are no moving parts. Improved FM electronics need to be

purchased or built to pursue this research.

Molecular fluorescence of the organic dye IR-140 was excited with both types of

LD. The best limit of detection was achieved with a high-power LD focused into a liquid

jet sample. With an LOD defined as concentration times volume (absolute LOD), the

LOD was 46000 molecules. The LOD based on the sample flow rate and the time constant

of the system (mass concentration LOD) was 107 molecules. A sheath flow cuvette was

constructed which produced a probe volume of 10*10 L. This device hydrodynamically

focused a sample stream with a sheath of flowing solvent. The single-mode LD, with its


smaller (10 pm) beam waist, was focused into this sample stream with a microscope

objective. The absolute LOD was 130,000 molecules and the mass concentration LOD

(which takes into account the time constant of the system) was 109 molecules. The liquid

jet fluorescence spectrometer system had superior detection limits, yet the sheath flow

cuvette offers the most potential for detecting small numbers of molecules. The main area

for improvement is the LD focusing and fluorescence collection optics. Improved spatial

and spectral filtering would decrease the contribution of laser scatter to the noise in the

measurements. Better focusing of the LD beam would enable a smaller sample stream

diameter to be used.

Molecular fluorescence systems with a static sample were also very successful. In

these experiments, a collimated high-power laser diode beam was passed through a cuvette

containing a liquid sample. Excellent concentrational detection limits (as low as 10"14 M)

were achieved with this system. A study of the signal-to-noise ratio of the LD fluorescence

systems suggested that the emission monochromator be dispensed with in favor of a

combination of optical filters. A simple (and potentially portable) fluorometer was

constructed by placing the laser diode directly adjacent to a cuvette. Scatter was rejected

with a colored glass filter, and luminescence was measured with a red sensitive photodiode.

The wavelength availability of LDs limits them to exciting compounds which

fluoresce in the near-infrared. NIR fluorescent tags for the rapidly growing techniques of

fluoroimmunoassay and DNA sequencing could lead to highly sensitive analytical techniques

because of the lack of fluorescent impurities and the diminished Raman scatter in the NIR.

The frequency-doubled LD was tuned to 403 nm to excite gallium atoms in a

graphite furnace, but it had insufficient intensity to produce detectable gallium atomic


fluorescence. The doubled LD was used more successfully to excite molecular fluorescence.

Perylene, a polyaromatic hydrocarbon in hexane, was detected in hexane at a level of

100 ppb.

These studies show the wide applicability of laser diodes to analytical spectroscopy.

As semiconductor researchers strive to provide high quality short wavelength LDs for the

communications and data processing fields, spectroscopists may use their products to



1. G.P. Agrawal, N.K. Dutta, Long-Wavelength Semiconductor Lasers, Van Nostrand
Reinhold Company, New York (1986).

2. G.H.B. Thompson, Physics of Semiconductor Laser Devices, John Wiley and Sons,
New York (1980), p. 156.

3. Optics Guide 4, Melles Griot Corporation, Irvine, California (1988), chapter 18.

4. D.C. O'Shea, W.R. Callen, W.T. Rhodes, Introduction to Lasers and Their
Applications, Addison-Wesley Publishing Company, Reading, Massachusetts (1978).

5. GaAlAs Laser Diode Technical Notes, Spectra Diode Labs, San Jose, California

6. J.D. Ingle, S.R. Crouch, Spectrochemical Analysis, Prentice Hall, Englewood Ciffs,
New Jersey (1988).

7. R.L Byer, Annual Review of Materials Science, 4 (1974).

8. R.S. Adhav, Laser Focus, 4 (1983).

9. T. Hirschfeld, Appl. Opt., 15, 2965 (1976).

10. B. Kirsch, E. Voigtman, J.D. Winefordner, Anal. Chem, 57, 2007 (1985).

11. E. Mroz, C. Lechene, Anal. Biochem., 102, 90 (1980).

12. G.R. Haugen, F.E. Lytle, Anal Chem., 53, 1554 (1981).

13. S. Yamada, F. Miyoshi, K. Kano, T. Ogawa, Anal. Chim. Acta, 127, 195 (1981).

14. GJ. Diebold, R.N. Zare, Science, 196, 1439 (1977).

15. L.W. Hershberger, J.B. Callis, G.D. Christian, Anal. Chem., 51, 1444 (1979).

16. R.A. Mathies, L. Stryer, Applications of Fluorescence in the Biomedical Sciences,
D.L Taylor, Editor; Alan R. Liss, Inc: New York (1986), Chapter 6.

17. Y. Kawabata, T. Imasaka, N. Ishibashi, Talanta, 33, 281 (1986).

18. D. Pinkel, Anal. Chem., 54, 503A (1982).

19. F. Zarrin, NJ. Dovichi, Anal. Chem., 57, 2690 (1985).

20. N. Dovichi, J.C. Martin, J.H. Jett, M. Trkula, R.A. Keller, Anal. Chem., 56, 348

21. T. Imasaka, A. Yoshitake, N. Ishibashi, Anal. Chem., 56, 1077 (1984).

22. T. Imasaka, T. Okazaki, N. Ishibashi, Anal Chim. Acta, 208, 325 (1988).

23. K. Sanda, T. Imasaka, N. Ishibashi, Anal Chem., 58, 2649 (1986).

24. T. Okazaki, T. Imasaka, N. Ishibashi, Anal. Chim. Acta, 209, 327 (1988).

25. P.A. Johnson, T.E. Barber, B.W. Smith, J.D. Winefordner, Anal Chem., 61, 861

26. G. L Long, J.D. Winefordner, Anal Chem., 55, 712A (1983).

27. L Hirschy, B.W. Smith, E. Voigtman, J.D. Winefordner, AnaL Chem., 54, 2387

28. J.D. Winefordner, R.C. Elser, Anal. Chem. 43, 24A (1971).

29. N. Omemetto, LM. Fraser, J.D. Winefordner, Appl. Spec. Rev. 7, 147 (1973).

30. D.R. Demers, DA. Busch, CD. Allemand, Am. Lab. 14(3), 167 (1987).

31. S.F. Zhu, H. Wang, P.N. Keliher, Microchem. Joum. 38, 264 (1988).

32. N. Omenetto, Spectrochim. Acta 44B, 131 (1989).

33. S. Neuman, M. Kriese, Spectrochim. Acta 29B, 127 (1974).

34. M.A. Bolshov, A.V. Zybin, V.G. Koloshnikov, I.I. Smirenkina, Spectrochim. Acta
43B, 519 (1988).

35. F.R. Preli, Jr., J.P. Dougherty, R.G. Michel, Anal Chem. 59, 1784 (1987).

36. K. Dittrich, HJ. Stark, J. Anal At. Spec. 2, 63 (1988).

37. J.P. Dougherty, J.A. Costello, R.G. Michel, AnaL Chem. 60, 336 (1988).


38. N. Omenetto, P. Cavalli, M. Borglia, P. Qi, G. Rossi, J. AnaL At Spec. 3, 231

39. D. Goforth, J.D. Winefordner, Talanta, 34, 290 (1987).

40. W. Grimm, Spectrochim. Acta, 23B, 443 (1968).

41. D.S. Gough, P. Hannaford, A Walsh, Spectrochim. Acta, 28B, 197 (1973).

42. M. Glick, B.W. Smith, J.D. Winefordner, Appl. Spec., Manuscript in preparation.

43. LA Budkin, O.G. Okhotnikov, G.T. Pak, A.I. Pikhtelev, S.L Puzanov, Zhurnal
Prikladnoi Spekrroskopii, 40, 533 (1982).

44. J. Lawrenz, A. Obrebski, K. Niemax, Anal Chem., 59, 1232 (1987).

45. G.P. Barwood, P. Gill, W.R.C. Rowley, J. Phys. E: Sci. Instrm., 21, 966 (1988).

46. R. Wyatt, WJ. Devlin, Electron. Lett., 19, 110 (1983).

47. C. van Dijk, B.W. Smith, J.D. Winefordner, Spectrochim. Acta, 37B, 759 (1982).


Paul Johnson was born on July 21, 1961, in Newport, Rhode Island, the youngest

of eight children. He graduated from Rogers High School in 1979. He pursued the

cooperative plan of education at Northeastern University, alternating classes with

employment in the chemical industry. He graduated with honors from Northeastern

University (located in Boston, Massachusetts) in 1985, with a Bachelor of Science Degree

in chemistry. Since that time he has attended the University of Florida, where he obtained

his Ph.D. in analytical chemistry in August 1989. His hobbies include sailing, SCUBA

diving, fishing, and tropical fish breeding.

I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.

) ames D. Winefordner/Chairman
Graduate Research Professor of

I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.

Vaneica P. Young
Associate Professor of

I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.

Gerhard M. Schmid
Associate Professor of



I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.

Anna Brajter oth
Associate Professor of

I certify that I have read this study and that in my opinion it conforms to acceptable
standards of scholarly presentation and is fully adequate, in scope and quality, as a
dissertation for the degree of Doctor of Philosophy.

Eric R. Allen
Professor of Environmental Engineering

This dissertation was submitted to the Graduate Faculty of the Department of
Chemistry in the College of Liberal Arts and to the Graduate School and was accepted as
partial fulfillment of the requirements for the degree of Doctor of Philosophy.

Dean, Graduate School


HMII BAl All 1111111111lIlIlllDII 11111111 .. l II
3 1262 08554 4459

'^ iti

Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd