RAPID SCREENING OF POLYCYCLIC AROMATIC COMPOUNDS
AND FINGERPRINTING OF COMPLEX SAMPLES BY
LASER-EXCITED FLUORESCENCE SPECTROMETRY
A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF
THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
I gratefully acknowledge the sage advice and saintly patience
shown me over the last five years by Jim Winefordner, without whom this
work could not have been accomplished. I cannot begin to express my
appreciation of all the members of Jim's group, who have not only helped
to make my time in graduate school more productive and enjoyable, but
have, each in his or her own way and without exception, enriched my
life. I would, however, like to thank those who helped to make my
endeavors more productive. Benny Smith and Ed Voigtman were always
willing to help in any way, both materially and spiritually. Brad Jones
and Jorge Vera were also helpful in experimental designs. Mike Mignardi
and Mark Click helped often in taming the computers. I will always have
fond memories of the time shared with Ben Womack, Paul Johnson and Tom
I owe much to those who have inspired me over the years to seek
graduate education. My parents have always stressed the value of
education. My mother never lost faith in me. My father never tired of
encouraging me forward. Dr. Popham, an instructor at SEMO, and Dr.
Matthews, my boss at Washington University, were both inspirations to
me. I am deeply indebted to Ingrid, my wife. She has shown me
patience, kindness and understanding, not to mention 70 words per
TABLE OF CONTENTS
1 INTRODUCTION ..............................................1
Polycyclic Aromatic Compounds .......................1
Analysis of Polycyclic Aromatic Compounds............3
2 LASER-EXCITED FLUORESCENCE OF AROMATIC MOLECULES
IN A GRAPHITE FURNACE .................................... 14
Experimental ........................... ...........15
Results and Discussion .................. ............18
3 SURVEY OF COMPLEX SAMPLES BY GRAPHITE FURNACE,
LASER-EXCITED FLUORESCENCE ................................36
Results and Discussion..............................40
4 LASER-EXCITED FLUORESCENCE FINGERPRINTS OF
ENVIRONMENTAL MATERIALS WITH A PULSED LASER,
GATED PHOTODIODE ARRAY SYSTEM ............. ..............77
Introduction...................... ................. 77
Results and Discussion..............................81
5 SUMMARY AND FUTURE WORK.................................. 89
BIOGRAPHICAL SKETCH ................................................. 97
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
RAPID SCREENING OF POLYCYCLIC AROMATIC COMPOUNDS
AND FINGERPRINTING OF COMPLEX SAMPLES BY
LASER-EXCITED FLUORESCENCE SPECTROMETRY
Chairman: James D. Winefordner
Major Department: Chemistry
Polycyclic aromatic compounds (PACs) are widespread contaminants
in our environment. They are encountered as complex mixtures in real
samples due to the nonspecific processes by which they form. Routine
methods for PAC analysis are costly and time-consuming. The aim of this
work was to develop and evaluate laser-excited fluorescence of aromatic
compound vapors produced in a graphite furnace (GFLEF) by direct
atomization of untreated complex samples as a screening method for PACs
and as a fingerprinting technique for environmental and other complex
Untreated samples were placed into the graphite furnace. The
furnace temperature was slowly raised. As components vaporize according
to volatility, sequential fluorescence spectra were collected by a
photodiode array. These spectra were plotted as fluorescence intensity
versus wavelength versus scan number, producing a three-dimensional plot
representative of the sample. Scan number was correlated to furnace
temperature. A continuous wave (cw) laser was used to excite
fluorescence. The system was used for PAC screening and fingerprinting
of crude oils, particulate matter and petroleum products. Limited
thermal resolution and broad fluorescence features precluded the
identification of individual components in complex mixtures.
A second experimental system with improved imaging of the graphite
furnace enabled use of the system over a wider temperature range.
Fingerprinting and PAC screening capabilities of GFLEF were evaluated by
analysis of a wide array of samples types including used and unused
cooking and motor oils, dairy products, fruits, meat, paper, tobacco,
vitamin formulations and over-the-counter medications.
An experimental system employing a pulsed dye laser and gated
photodiode array enabled even greater extension of the useable
temperature range by discrimination against furnace wall emission. By
tuning the laser to an atomic absorption line, atomic fluorescence
information could be included into the fingerprints for complex samples.
The various GFLEF systems provided rapid and simple screening for
PACs and fingerprints for complex samples. The high temperatures
attainable with the graphite furnace enabled the direct analysis of
solids and resinous materials often too difficult to analyze by
Polycyclic Aromatic Compounds
Polycyclic aromatic compounds (PACs) are widespread contaminants
in the environment that are produced mainly by the burning of fossil
fuels and refuse. The industrialization of modern society resulted in a
greater dependence on fossil fuels. Concurrently, refuse production
increased from both commercial and domestic sources. PACs are,
therefore, closely linked to our industrialized society and desire for
consumer products and a high standard of living. Table 1-1 is an
estimation of worldwide PAC production and sources (1).
Polycyclic aromatic compounds consist of two or more fused
aromatic rings, and may contain substituted alkyl groups. Heteroatoms
may be incorporated into the rings or substituted groups. Most PACs are
produced by the incomplete burning of organic material. At temperatures
of 500-800 OC, organic compounds are cracked to smaller free-radical
fragments. This process is pyrolysis. These fragments then recombine
to form larger, more stable aromatic compounds in a process known as
pyrosynthesis. The amount and types of PACs produced vary greatly.
Under ideal conditions of temperature and oxygen content, organic
material is combusted to CO2 and water to a greater degree (2).
Table 1-1. Worldwide PAC Productiona
Heating and power generation
Incineration and open burning
Total PAC emissions
a. From reference (1).
Generally, lower temperatures yield higher concentrations of
substituted PACs. Higher temperatures yield higher concentrations of
unsubstituted PACs. The organic starting material and other combustion
conditions will effect the pyrolysis products. Nitrogen, sulphur and
oxygen, when present, can be incorporated into PACs (2,3). Materials
containing PACs usually have extremely complex compositions because
pyrolysis is an unspecific process.
Polycyclic aromatic compounds are formed to a much lesser degree
by natural processes. They have been identified in coal, crude oil,
sediments and airborne particulates. Sources of these naturally
occurring PACs include volcanic activity, forest fires and the
degradation of organic material in the processes forming fossil fuels.
Many members of the PAC class of compounds are known or suspected
carcinogens. The Environmental Protection Agency (EPA) has recognized
many PACs as carcinogens (4). The carcinogenic effect of these PACs in
microgram quantities and their prevalence in the environment dictate
close monitoring of their presence and investigations of their sources
and health effects in order to regulate their detrimental effects.
Polycyclic Aromatic Compound Analysis
Because of the carcinogenic nature of many PACs, there is intense
analytical interest in the development of methods to analyze specific
components of the complex samples in which PACs normally occur.
Interest in PAC analysis has focused on the analysis of polycyclic
aromatic hydrocarbons (PAHs), a subset of the PAC family of compounds.
PAHs contain only carbon and hydrogen, and are usually present in much
higher concentrations than PACs that may contain heteroatoms.
By far, the most popular method of analyzing PACs is by gas
chromatography (GC). Advantages of GC include high resolution,
availability of universal and element-specific detectors, availability
of cross-linked polymeric and chemically bonded stationary phases that
are stable at high temperatures and compatibility with mass
Gas chromatographic analysis requires that the analyte be volatile
and thermally stable at the temperature used. These conditions are
generally met for PACs of seven or fewer rings (5). High resolution GC
methods make them the techniques of choice for separation and
quantitation of PACs in complex samples (6).
High-performance liquid chromatography (HPLC) is also routinely
used for PAC analysis and is suitable for analysis of PACs soluble in
the mobile phase. This includes most PACs of less than 13 rings (5).
Advantages of HPLC include the separation of thermally labile compounds,
compatibility with absorbance, fluorescence and electrochemical
detectors and the availability of many suitable solvents and mixed
phases (7). These mobile and stationary phases make possible many
different separation mechanisms that can be used to separate analytes
based on physical and chemical differences. The spectroscopic and
electrochemical detectors used with HPLC are selective and sensitive.
The limited resolution of HPLC usually makes more than one separation
necessary. The initial separations can be extractions, other forms of
chromatography or preliminary HPLC fractionation (8).
The areas of GC and HPLC applicability overlap, but generally the
methods are complementary. These two methods, with mass spectrometric
confirmation of peak identifications, are often used for the most
complete analysis of complex samples such as air and diesel particulate
matter, petroleum products and coal tar (8-10).
Supercritical fluid chromatography (SFC) has experienced
tremendous growth as a method for PAC analysis. This growth in the last
decade is due to the introduction of capillary SFC columns and the
availability of commercial SFC instrumentation (11). A 1981 reference
(3) treats SFC in about two pages while a 1989 reference (11) devotes an
entire chapter to PAC analysis by SFC.
Supercritical fluid chromatography offers easier detector coupling
and better separations than HPLC plus a wider mass range than GC and has
been used to identify components in automobile exhaust (12), petroleum
products (13) and coal tar (14). Presently, most applications of SFC to
PAC analysis have been aimed at method and instrument development.
Thus, the full potential of SFC is yet to be realized.
Paper and thin layer chromatography have also been used for PAC
analysis, but suffer from poor reproducibility and sensitivity (15).
Absorption and luminescence spectra of PACs, although specific,
are broad and featureless. If a sample contains more than two or three
individual PACs, identification may be difficult. To detect PACs in
complex samples, a separation is required prior to spectroscopic
detection. Spectroscopic techniques, thus, are widely used as
chromatographic, particularly HPLC, detectors (3). Fluorescence has at
least two advantages over absorption for PAC analysis. First, lower
detection limits are achieved because of a lower background. Second,
selectivity is higher because the excitation wavelength can be selected
based on the requirements of a particular analysis. Lasers are used
extensively because they are a well-collimated, intense excitation
At low temperatures, absorption and fluorescence peaks of PACs are
narrowed, which greatly increases the usefulness of spectroscopic
techniques for PAC identification. UV/visible absorption in solid
matrices has been used to identify PACs (16). Matrix isolation Fourier
transform infrared spectroscopy (17) and low temperature fluorescence
(18) also have been used to selectively detect PACs in simple mixtures.
Allowing molecular vapors to expand through a small orifice into a
vacuum is another approach to low temperature spectroscopy. In a free
jet expansion, molecules are quickly cooled. As a result, spectra
exhibit the characteristically narrow absorption and fluorescence peaks,
allowing selective detection of PACs in mixtures (19). If a dye laser
is used as an excitation source, analytes may be selectively ionized by
the absorption of two or more photons. The ions may be detected as a
current across electrodes. Alternatively, the ions can be detected by a
mass spectrometer, greatly increasing the amount of analytical
information available (20). This form of low temperature spectroscopy
has the advantages of fast cooling, high resolution and compatibility
with gas chromatography and mass spectrometers.
Improved selectivity is accomplished by synchronous luminescence
spectroscopy by taking advantage of absorption and luminescence
information simultaneously. This is done by scanning the excitation
source and emission detector together with a constant energy or
wavelength difference (21). Signals are obtained only for compounds
that have the selected energy or wavelength difference between
excitation and emissions transitions. Lifetime-selective and phase-
resolved fluorescence detectors have been used in HPLC determination of
Other specialized luminescence techniques for PAC analysis take
advantage of selective phosphorescence and fluorescence enhancement or
quenching (23). While fluorescence detection is the most widely used
spectroscopic technique for PAC analysis, other techniques used in
special cases are gaining in popularity.
Raman spectroscopy detects inelastically scattered photons. Like
infrared spectroscopy, Raman spectroscopy has excellent selectivity for
chemical groups. Surface-enhanced Raman was first observed in 1974 for
molecules adsorbed on metal surfaces, particularly silver (24). The
advantages of surface-enhanced Raman spectroscopy are ease and
versatility of sample handling, applicability to water systems and the
ability to investigate adsorption dynamics. Raman spectroscopy has been
used to selectively detect components in environmental samples (25).
Photothermal spectroscopy is an absorption technique which is applicable
to large, irregular solid samples (26) as well as to HPLC detection
(27). The numerous variations of this up-and-coming technique as well
as theoretical aspects have been described by T. Vo-Dinh (28). Nuclear
magnetic resonance (NMR) is most helpful in the elucidation of unknown
structures and has also been used to selectively detect PACs in mixtures
extracted from environmental samples and tobacco smoke condensates (29).
Mass spectrometry (MS) has been an indispensable tool for PAC
identification and quantitation since 1951 (30). High resolution
chromatographic retention data combined with mass spectral information
can often provide an unambiguous identification of components of complex
mixtures. Structure elucidation is often possible based on the
fragments present in the spectrum. "Soft" ionization techniques such as
chemical ionization produce mainly molecular ions, yielding molecular
weight information. Selective ionization of isomers can be achieved in
a free jet expansion by the use of lasers (20). Mass Spectrometry is
compatible with GC, capillary supercritical fluid chromatography and
capillary electrophoresis. Secondary ion, fast atom bombardment, laser
desorption and field ionization desorption are vaporization methods for
thermally labile compounds. The most used method for unambiguous PAC
analysis in complex mixtures is GC separation followed by electron
impact ionization MS. Although MS is usually less sensitive than
fluorescence for PAC analysis, its high specificity and compatibility
with GC make it one of the most useful methods for PAC analysis.
The complex samples in which PACs are present often contain
thousands of different PACs in widely varying concentrations. For a
given sample type and target PAC, appropriate methods exist for
quantitative and qualitative analysis. Common to all techniques are
time-consuming sample pretreatment steps. These steps may, beyond the
initial extraction, involve further extractions, adsorption
chromatography, sample preconcentrations, derivatization and HPLC
fractionation. These procedures aim to remove interference and reduce
the number and increase the concentrations of components to manageable
and detectable levels. For every pretreatment step, confidence in
conclusions about the original sample is reduced. Precise results are
not necessarily accurate. Problems include evaporative loss of volatile
components, reactions with oxidants and surface-catalyzed photochemical
reactions that sometimes result in products and mixtures exhibiting a
greater carcinogenic nature. Extraction efficiencies have been shown to
vary widely for different PACs and for certain compounds in different
sample types (31). These degradation and loss complications are
affected by the particle size and the microscopic nature of the surface
to which PACs are adsorbed (32).
Since biological tests are performed using the same extracted,
fractionated samples, it follows that misconceptions may exist about the
composition and biological effects of PAC-containing samples.
Unacceptable variability in PAC quantitation, in some cases order of
magnitude differences, have been reported for identical aliquots
analyzed by different methods (33). In the same study, compounds
detected by HPLC were not confirmed by GC-MS. The problem of
identifying particular compounds is due, at least in part, to the above
complications as well as variability among different lots of the same
stationary phases which can, on occasion, change the elution order of
PACs, differences in mass spectra dependent upon instrumental parameters
and a lack of standards available for confirmation of peak assignments.
In general, PAC analysis is tedious, time consuming, uses very
expensive, state-of-the-art equipment and requires much technical
expertise. The most selective and sensitive techniques are also the
Due to the complex nature of PAC-containing samples, the
variability of results and expense in both time and equipment required
for analysis of individual components of these samples, some
investigators have questioned the necessity and wisdom of current
methods for routine analysis (1,34). They suggest that such
sophisticated methods are not cost- or information-effective.
Two solutions to this problem are possible. Samples can be
characterized by analysis of one PAC and inferences made concerning the
concentrations of other PACs. Alternatively, a method measuring total
PAC content irrespective of isomers could be used. Both of these
solutions are screening methods.
Screening methods are techniques that provide incomplete
information about a sample in a rapid and simple manner. Screening
methods indicate the absence or presence of a few compounds or class of
compounds. They are helpful in characterizing samples and for
prioritizing samples for further analysis by alternate techniques if
more specific information is desired.
If the characterization provided by the screening method is unique
to a particular sample, it can be used as fingerprint for that sample.
Fingerprints can be used to differentiate samples and can be helpful,
for example, in tracing the origin of pollutants such as an oil spill.
Numerous problems are associated with screening methods. Methods
that are sensitive to PACs irrespective of isomers yield nonspecific
information from which no inference concerning health effects can be
made. Methods that rely on measurement of target compounds that are in
some manner representative of the very large and complex group of PACs
that comprise nearly all real samples suffer from lack of reliability in
the choice and analysis of the target compound and in the inferences
made concerning the concentrations of other components. Target
compounds are referred to as proxy, indicator, surrogate or signature
The most commonly used target compound for proxy PAC analysis is
benzo(a)pyrene (BaP). This, apparently, is due to the carcinogenic
nature of BaP, its presence in most PAC-containing samples and the
wealth of information available in the literature concerning the health
effects and analytical considerations of BaP.
Mixing of PACs from various local sources takes place to the
extent that there is a high degree of correlation among PAC species in
the atmosphere for a given locale (35). Any one compound is therefore a
good index of the concentrations of the others. In another study, PAC
concentrations relative to BaP were found to vary depending on the
location of the sampling station, the time of day and the weather, with
location showing the most significant variations (36). In addition, the
concentrations of PACs adsorbed to particles vary with particle size and
emission source (37). Since size is a determining factor in retention
of particles in the respiratory tract, this consideration cannot be
overlooked in determining PAC impact on health.
Another problem with proxy methods is that any inaccuracy in
determining the proxy compound would result in inaccurate inferences
about the concentrations of other components. For example, the C20H12
isomers BaP, benzo(k)fluoranthene (BkF), benzo(b)fluoranthene,
benzo(e)pyrene and perylene are among the most commonly encountered
PACs. They are also difficult to separate chromatographically. It is
likely that BkF, a noncarcinogenic PAC that exhibits a nearly identical
fluorescence spectrum as BaP, has in some cases been erroneously
measured as the carcinogen BaP (38).
As stated earlier, extraction efficiencies and degradation losses
of PACs vary greatly among the various PACs and for PACs in different
sample matrices. Field tests have shown that tracers introduced at the
time of sampling could not be used to account for these losses due to
the variability of recovery of the tracers, ranging from near 0 to 500%
(39). P.W. Jones estimated that 80% of all PAC studies have failed to
take these variances into account (1). This indicates that preliminary
studies that would be required to choose a proxy PAC and to ascertain
the relative concentrations of other PACs in order to infer their
concentrations from subsequent proxy studies may not yield reliable
After careful characterization of a particular emission source,
reliable proxy PAC analysis should be possible; however, a universal
proxy PAC would not be possible. Since considerable synergism for
biological activity exists among PACs, mixtures often exhibit greater
carcinogenic activity than would be expected from their individual
carcinogens (40). It is, therefore, important to correlate PAC
concentrations as well as health effects with proxy PACs, data which are
totally lacking for most PACs.
The aim of this work was to develop laser-excited fluorescence of
aromatic compound vapors produced by direct electrothermal atomization
of untreated complex samples in a graphite furnace (GFLEF) as a
nonspecific screening method for PACs and as a fingerprinting technique
for environmental and other complex samples.
LASER-EXCITED FLUORESCENCE OF AROMATIC MOLECULES
IN A GRAPHITE FURNACE
Many PACs are classified as EPA priority pollutants because they
are known or suspected carcinogens. The two main sources of PACs in the
environment arise from incomplete combustion of fossil fuels for energy
production and from incomplete combustion of refuse. Interest in
carcinogenic compounds led to the development of many gas and liquid
chromatographic techniques for analysis of specific compounds in such
samples as crude oils, petroleum products, diesel particulates and urban
dust (3). These methods generally involve elaborate extraction and
fractionation schemes prior to analysis.
Recently, several groups have reported the use of electrothermal
atomizers for the study of the ultraviolet absorption of molecular
vapors. Tittarelli et al. found that by monitoring ultraviolet
absorption at 190 nm as a crude oil sample was vaporized from 140 to 900
OC, spectral fingerprints characteristic of the samples could be rapidly
obtained (41). In later work (42), a photodiode array was used to
monitor absorption in the wavelength range of 190 to 355 nm. Plotting
the sequential spectra produced a three-dimensional plot characteristic
of the sample vaporized. These representations were also used as
fingerprints for the samples, and had more distinguishing
characteristics than earlier two-dimensional plots. Absorption of
ultraviolet light by vapors produced in a graphite furnace has also been
used to characterize pollutants in water and sediments and as a
fingerprinting technique for various oils, petroleum products and soaps
(43). The high temperatures obtained with a graphite furnace
facilitated the direct analysis of heavy oils, resinous materials and
other complex samples often too difficult for analysis by gas
In this work, a novel system involving laser-induced fluorescence
of PACs in a graphite furnace was investigated as a screening method for
PACs and as a fingerprinting technique for environmental samples and
other complex mixtures. Direct analysis of crude oil, petroleum
products and solid materials was accomplished without sample
A block diagram of the experimental system is shown in Figure 2-1.
The laser was a He-Cd continuous wave (cw) laser (Model 4120N, Liconix,
1390 Borregas Ave., Sunnyvale, CA, 94089) with a maximum output of six
mW at 325 nm. The laser beam passed through a 1.5 mm hole drilled
through a polished aluminum off-axis parabolic mirror before passing
lengthwise through the inside of the graphite tube.
The graphite furnace was an atomic absorption spectrometric system
(Model CTF555, Instrumentation Laboratories, Thermo-Jarrell Ash, 590
Lincoln St., Waltham, MA, 02254). The controller had adjustments for
six temperature settings and allowed selection of the ramp rate between
Block diagram of the experimental system.
the six stages. The furnace was set to slowly ramp from room
temperature to 50 C over one minute to dry the sample, then ramped
linearly from 50 to 500 OC over approximately two minutes as
fluorescence spectra were collected. The furnace was contained in a
small enclosed housing, which allowed volatization of the sample to
occur in an inert environment of argon. To monitor the temperature, the
housing also contained a thermocouple in contact with the graphite tube.
The fluorescence was collected at 1800 relative to the laser
excitation. The cone of fluorescent light exiting the graphite tube was
collimated by an off-axis parabolic mirror. Loss of signal by light
escaping back through the 1.5 mm hole was approximately 0.3%. The
collimated light was focused on the entrance slit of the spectrograph by
a second off-axis parabolic mirror.
The flat field spectrograph (Instruments SA, Inc., 173 Essex Ave.,
Metuchen, NJ, 08840) has a dispersion of 24 nm/mm, a focal length of 0.2
m and a spectral range of 200 to 800 nm. A 1024 pixel intensified diode
array (Model TN 6100, Tracor Northern, 2551 West Beltline Highway,
Middleton, WI., 53562) was mounted to the spectrograph with a
laboratory-constructed flange, which held the array in the focal plane.
The diode array head contained a Peltier effect thermoelectric cooler
capable of cooling the head to -20 C, which reduced the dark current.
A dedicated computer controlled data acquisition, storage, processing
and output. Generally, the array was exposed for 2 s before being read
and stored. Spectra were displayed on the screen and stored after
automatic subtraction of a restored background spectrum. Additional
software, Quadra (Tracor Northern), allowed for the generation, scaling
and plotting of three-dimensional representations of fluorescence
intensity versus wavelength versus scan number of sequentially
accumulated spectra. Scan number was correlated to time and, therefore,
Results and Discussion
To determine the sensitivity of the system, analytical calibration
curves were prepared by placing 100 p1 of standard concentrations
ranging from 100 to 0.1 pg/ml in methanol into the graphite furnace. A
drying cycle was employed to evaporate the solvent. The appearance time
of the sample has been found to depend linearly on the boiling point of
the compound (44). Figures 2-2 through 2-5 show typical vaporization
cycles for several standards. The temperature ramp rates were chosen as
a compromise between the increased sensitivity of fast ramp rates and
the increased thermal resolution of slower ramp rates. Small variations
in ramp rates have little effect on the appearance of the three-
dimensional plot. Since all plots have the same temperature scale, ramp
rate affects the number of spectra taken during the vaporization cycle,
assuming the scan rate remains unchanged. For instance, the plot of a
standard vaporized at a 30% greater ramp rate would appear the same if
the number of scans displayed was reduced from 55 to 39.
For preparation of calibration curves, spectral information was
collected and stored in a slightly different form. The software allowed
the user to define a spectral region of interest. This region was
chosen to correspond to the fluorescence peak of a particular compound.
As the sample vaporized, only the area above baseline of the region of
650 600 550 500 450 400 350
Figure 2-2. Vaporization plot for anthracene.
Scr- ___ _______
650 600 550 500 450 400 350
Figure 2-3. Vaporization plot for benzo(b)fluoranthene.
> 10000 -- ------------ 500
-- ------ -___ 00
-0 0 -0-_
50000 50 40 40
Figure 2-4. Vaporization plot for pyrene.200
650 600 550 500 450 400 350
Figure 2-4. Vaporization plot for pyrene.
S- --- -2
a- ma--""" -"" I-----"'""""""" '""" "
650 600 550 500 450 400 350
Figure 2-5. Vaporization plot for coronene.
Figure 2-5. Vaporization plot for coronene.
interest was stored. A representation of the area of the fluorescence
peak versus time is called a histogram. The area under the histogram
peak was proportional to the volume under the peak in the three-
dimensional representation but required much less memory. For the six
compounds for which standard curves were prepared, the log-log plots
were linear over three or more orders of magnitude. The limits of
detection (LODs) were calculated as three times the standard deviation
of the blank measurement divided by the slope of the analytical
calibration curve (counts/ng). The slopes of the analytical calibration
curves, the linear dynamic ranges and the LODs for six PACs are shown in
Table 2-1. The major limitation to lower LODs is the high standard
deviation of the blank measurement. Uncertainty in the background
subtraction causes the low concentration points to deviate from the
linear calibration curve. At higher concentrations, the curves bend
over, especially for strongly absorbing compounds due to prefilter,
postfilter and self-absorption effects. Also, furnace emission
interfered with compounds fluorescing at longer wavelengths and/or
vaporizing at higher temperatures. These detection limits were about 10
times higher than those reported in previous work in this laboratory by
Kirsch (45). However, Kirsch (45) used more powerful lasers and the
fluorescence collection angle was not limited by baffles in the graphite
tube housing. In addition, averaging pixels produced a high signal-to-
noise ratio at the expense of spectral resolution. Detection limits
obtained in this work were comparable to the approximate determination
limits reported by Shekiro and Skogerboe (44) for absorbance
measurements of PAC vapors. If detection limits were calculated using
Table 2-1 Limits of Detection (LOD) Obtained by Laser Fluorescence
Spectrometry of Vaporized Polycyclic Aromatic Hydrocarbons (PAH)
Linear Dynamic Ranpea
Orders of magnitude
Slope of log-log calibration curve.
the same criteria as Shekiro and Skogerboe (44), the detection limits
were three to five times lower than those shown in Table 2-1.
To demonstrate the thermal resolution of the system, standard
mixtures were vaporized. Figure 2-6 shows the thermal resolution of two
PACs. Standard compounds evolved over about a 25 s, or 145 oC interval.
In the thermal dimension, the FWHM for most of the compounds tested is
about 9 s, or about 50 C. Since the PACs fail to undergo sharp
vaporization transitions into the vapor phase, the thermal resolution
capabilities are limited. The highest number of compounds thermally
resolved was three. Vapor phase PAC spectra exhibit peak widths on the
average of 60 to 90 nm FWHM. This lack of sharp features and limited
thermal resolution precluded the identification of individual components
of a complex mixture. It also seems likely that components in a sample
matrix vaporize at slightly higher temperature than the pure materials,
further complicating the extraction of qualitative information.
Finally, to demonstrate the fingerprinting capability of the system,
three samples of crude oil, three samples of particulate matter and
several types of petroleum products were analyzed. The resulting three-
dimensional maps all contained distinguishing features. Samples
producing similar plots could be differentiated by the relative
intensities or peak widths of the various features. Figure 2-7 through
2-14 show plots obtained from the vaporization of some of these samples.
The simplicity and speed of laser-excited fluorescence of aromatic
molecular vapors produced in a graphite furnace, the characteristic
fingerprints obtained and the ability to analyze solids and resinous
materials indicate that it will be a useful method for PAC screening and
600 500 400
Figure 2-6. Vaporization plot for a standard mixture of anthracene and
perylene, showing the thermal resolution of two PACs.
600 500 400
Figure 2-7. Vaporization plot for Kirkuk crude oil.
600 500 400
Figure 2-8. Vaporization plot for SRM 1582 crude oil.
600 500 400
Figure 2-9. Vaporization plot for E.S. Sider crude oil. Note the
broader peaks in the 200-300 C range compared to the previous figure.
, ,0 ..-,-,.I ,,, ....i ,,,i .
0A 111.l ii ii ii r --- I
600 500 400
Figure 2-10. Vaporization plot for SRM 1580 shale oil.
-~~~ ~~ ,.^^ l^ J- .-?-- -
600 500 400
Figure 2-11. Vaporization plot for regular gasoline.
600 500 400
Figure 2-12. Vaporization plot for SRM 1649 urban dust.
600 500 400
Figure 2-13. Vaporization plot for SRM 1648 urban dust.
intense fluorescence in the 200-300 C range compared to
Note the more
600 500 400
Figure 2-14. Vaporization plot for SRM 1650 diesel particulate matter.
Note the narrower peaks in the 200-300 OC range compared to the previous
SURVEY OF COMPLEX SAMPLES BY GRAPHITE FURNACE,
Preliminary investigations into the use of laser-excited
fluorescence of aromatic molecules in a graphite furnace indicated that
it was a sensitive method for screening of PACs in complex samples such
as crude oils, particulate matter and petroleum products. The three-
dimensional plots obtained by direct vaporization of untreated samples
were useful as distinguishing fingerprints for the samples analyzed.
Most of the PACs in our environment result from the use of fossil
fuels for energy production and from the burning of refuse. These PACs,
therefore, are associated with particulate matter in the atmosphere,
water and in sediments.
Despite the fact that most PACs are produced by combustion of
fuels and refuse, significant human exposure to PACs occurs as a result
of the use of consumer products such as tobacco and the ingestion of
PAC-containing foods as well as from exposure from air and water.
For many people, the major risk from PAC exposure may be from PACs
in food. Of particular concern are smoked products such as smoked meats
and cheeses. Regardless of PAC exposure from other sources, people
ingest with their food amounts of PACs exceeding an acceptable, safe
Because of the variability in diets and food preparation,
benzo(a)pyrene (BaP) exposure from food may be the most significant
exposure, even for tobacco users. In the same study, average daily
intake of BaP in water and air was found to be within the allowable
daily intake (46). For example, one charcoal-broiled steak may contain
an amount of BaP equivalent to that from the smoke of 22.5 packs of
cigarettes (47). Evidence concerning PAC health effects comes largely
from occupational and community air pollution studies, while the effects
of ingested PACs have been largely ignored (46).
Another PAC contamination that is often overlooked is used oil,
often discarded haphazardly by the general public. Of the PACs produced
by an engine, about 85 percent is retained in the crankcase oil and
about 15 percent is emitted into the atmosphere as part of the exhaust
(48). Total PAC concentration of used motor oils has been found to be
as high as 15 g/l (49).
To a certain extent, lack of information concerning the PACs in
foods and used oils is due to the tedious and time-consuming procedures
involved with PAC analysis. In this work, the applicability of GFLEF
for PAC screening in a wide array of consumer products and in used oil
was demonstrated. Distinguishing fingerprints for foodstuff, oils, used
oil and over-the-counter medications were also obtained by GFLEF.
The experimental system used for the survey of consumer products
was the same as that described in Chapter 2, with two modifications
(Figure 3-1). First, a graphite furnace with more accurate and flexible
temperature control was employed (Shimadzu, model GFA-4A, Kyoto, Japan).
Temperature was controlled by a microprocessor. Temperatures and ramp
rates for up to six stages could be chosen. Temperature control was
achieved by a current sensor for temperatures below 1100 OC. A silicon
photocell monitored furnace wall emission for temperature control
feedback for temperatures exceeding 1100 OC. This allowed temperature
cycles to be tailored to different sample types with greater ease than
with the previous furnace. In addition, the three-dimensional plots
obtained were more reproducible due to better temperature control.
Second, the off-axis parabolic mirrors were abandoned in favor of
a flat mirror with a four mm hole drilled at a 450 angle and a lens.
The excitation laser passed through the hole in the mirror and into the
graphite furnace. The cone of fluorescent light exiting the graphite
tube housing at 1800 relative to the laser excitation was folded by the
mirror onto a lens that focused the light onto the entrance slit of the
spectrograph. The advantages of this arrangement are lower cost, easier
alignment of optical components and better imaging of the graphite
furnace. Since a better image of the furnace emission is obtained, it
can more effectively be masked, reducing the background emission from
the furnace wall collected by the photodiode array. This was important
because of the higher temperatures used and lower PAC concentrations
encountered in some of the samples in this work.
For solid samples, approximately two to three mg were placed
inside the furnace. The tobacco samples were blended by the retailer
and were composed of a mixture of tobacco from various sources. These
Block diagram of the experimental system.
samples were powdered in a mortar and pestle in order to obtain a small,
representative sample for analysis. Liquid samples were pipetted
directly into the furnace. For oils, milk and juices, the volume
vaporized was 5.0 pi. No sample pretreatment was involved in the
analysis with the exception of grinding the tobacco, which is not a
pretreatment in the usual sense of extraction or fractionation. After
the vaporization cycle, the furnace temperature was raised to an
empirically determined temperature high enough to prevent memory
effects, usually 1000 OC.
Results and Discussion
Fourteen tobacco samples were analyzed by GFLEF. The samples were
numbered one through fourteen. Sample numbers one through ten were
blended pipe tobacco samples. These samples were blended by the
retailer with tobacco from various sources. Many of the mixtures have
components in common. Sample numbers eleven through fourteen were four
different brands of cigarette tobacco. All four cigarette tobacco
samples and one of the pipe tobacco samples produced unique fingerprints
that could be used to identify the samples. The remaining nine pipe
tobacco samples were divided into three groups of two samples and one
group of three samples. The plots obtained from the vaporization of the
nine samples for which unique plots were not obtained could be used to
differentiate the samples from other samples outside the group. This
was the basis of the grouping. Examples of tobacco fingerprints are
shown in Figures 3-2 and 3-3.
Various investigators have reported the presence of PACs in food
products such as vegetable oils (50), dairy products (51), fruit and
meat (52). Figure 3-4 shows the plot obtained from the vaporization of
beef fat. Figure 3-5 shows the vaporization cycle of a sample of beef
hot dog, showing fluorescence from components vaporizing at higher
temperatures than the beef fat. If, however, this fingerprint was
plotted with a different intensity scale, fluorescence similar to that
seen in the beef fat vaporization was visible. Figure 3-6 is the
vaporization cycle for a beef hot dog plotted on a different intensity
scale to show the low intensity contributions from the beef fat.
Vaporization plots were also obtained for four fruit samples.
GFLEF screening revealed PAC content in all four samples. In addition,
plots obtained from various forms of the same fruit exhibited similar
spectral characteristics. Figure 3-7 is the vaporization plot of canned
orange juice concentrate. This plot is very similar to the plot for
fresh orange juice (Figure 3-8) except that the fluorescence intensity
is greater. Figure 3-9 shows the vaporization cycle for an exterior
orange peel sample. Although components of this sample vaporize at a
higher temperature, the spectral characteristics are similar to the
spectral features of the other two orange samples. For comparison, a
sample of canned grape juice concentrate was analyzed (Figure 3-10). A
comparison of Figures 3-7 and 3-10 reveals that the fluorescence
intensity and the intensity versus furnace temperature profiles of the
two samples are virtually identical. However, spectral features differ.
Figure 3-2. Vaporization plot for tobacco #1.
Figure 3-3. Vaporization plot for tobacco #8. Note the difference in
the intensity vs temperature profiles between this plot and the previous
Figure 3-4. Vaporization plot for beef fat.
Figure 3-5. Vaporization plot for beef hot dog.
Figure 3-6. Vaporization plot for beef hot dog plotted on a different
intensity scale to show the lower intensity contributions from the beef
fat visible at lower temperatures.
- ,1 0 000
Figure 3-7. Vaporization plot for orange juice concentrate.
0 j200 -
o lrr loo
Figure 3-8. Vaporization plot for fresh orange juice. Note the
spectral and intensity vs temperature similarities with the previous
Figure 3-9. Vaporization plot for orange peel. Note that the spectral
characteristics are similar to the previous two figures, however, these
features appear at higher temperatures.
Figure 3-10. Vaporization plot for grape juice concentrate. Note the
different spectral characteristics compared to the three previous
This demonstrates the advantage of the three-dimensional technique GFLEF
with photodiode array detection in providing distinguishing
fingerprints. One might wonder how GFLEF would be helpful in
distinguishing grape juice from orange juice. However, the technique is
sensitive enough to provide fingerprints for dilute solutions or small
amounts of residue.
Marked differences in the fingerprints for whole milk, American
cheese, butter and margarine are apparent (Figures 3-11 through 3-14).
The characteristic features of these plots indicate that fingerprinting
of different brands would be possible, particularly for cheeses.
Five different refined vegetable oils were obtained from a local
retail grocer. Corn, safflower, soybean, olive and canola oils all
exhibited PAC fluorescence. The low intensity of the fluorescence
indicated low PAC concentration. Because the fluorescence intensity was
low, photodiode array background noise contributed significantly to the
plots. The noisy, low-intensity plots could not be used as fingerprints
to differentiate the five vegetable oils. Figure 3-15 is a vaporization
plot for a corn oil sample.
While GFLEF could not be used to fingerprint refined vegetable
oils, the method was sensitive to changes that occur in the composition
of oils as the oils are used in food preparation. Two samples of
vegetable oil were obtained from a local restaurant. One sample was
unused fry oil (Figure 3-16). The other sample was taken from a deep
fryer (Figure 3-17). The used oil exhibits additional PAC fluorescence
not present in the unused oil. The origin of the additional PACs
Vaporization plot for whole milk.
0 IIIII-r I 100
Figure 3-12. Vaporization plot for American cheese.
Figure 3-13. Vaporization plot for butter.
Figure 3-14. Vaporization plot for margarine.
peaks at 250 OC compared to the previous plot.
Note the lower intensity
Lj j 10 0
Figure 3-15. Vaporization plot for corn oil.
300 400 500 600
Figure 3-16. Vaporization plot for unused fry oil.
300 400 500 600
Figure 3-17. Vaporization plot for used fry oil. Note the additional
features near 300 C compared to the previous figure.
present in the used oil may come from the pyrolysis of components of the
oil or food, or from incomplete removal of heavily used oil when the oil
In order to simulate the changes that occur in heavily used oil,
corn oil was heated in a cast iron frying pan on high heat for one hour.
Figures 3-18 and 3-19 are plots obtained from the vaporization of the
unused and used corn oil, respectively, plotted with the same intensity
scale. Again, no conclusions can be drawn about the origin of PACs in
the used oil. They may occur as a result of pyrolysis of components of
the oil or residue from the considerable previous use of the pan in food
preparation. However, the significantly higher PAC content of the
synthetically aged corn oil indicates that the PACs are pyrolysis
products of corn oil components.
Crude oils contain PACs, as do the many of the petroleum products
refined from crude oil. One of the potentially important uses for GFLEF
is as a fingerprinting technique for petroleum products. GFLEF has
provided a distinguishing fingerprint for every crude oil and petroleum
product analyzed to date. In order to further investigate the
fingerprinting capabilities of GFLEF, several brands of motor oil were
obtained from a local retail merchandiser. Plots obtained by GFLEF
could be used as fingerprints to selectively identify different brands
of motor oil. Fingerprints of Castrol 30W, Quaker State 30W and Penzoil
10W-40 motor oils are shown in Figures 3-20 through 3-22, respectively.
As noted previously, used motor oils contain additional PACs as a
result of use. Figure 3-22 is a fingerprint of unused Penzoil 10-40
motor oil. Figure 3-23 is a fingerprint of the crankcase oil of an
Figure 3-18. Vaporization plot for corn oil, plotted with the same
intensity scale as the next figure.
Figure 3-19. Vaporization plot for used corn oil.
300 400 500 600
Figure 3-20. Vaporization plot for Castrol 30W motor oil.
400 500 600
Figure 3-21. Vaporization plot for Quaker State 30W motor oil. Note
the difference in spectral characteristics between this plot and the
other motor oil fingerprints.
300 400 500 600
Figure 3-22. Vaporization plot for Penzoil 10W-40 motor oil. The
spectral characteristics and intensity vs temperature profile of this
plot differs from the other motor oil samples.
300 400 500 600
Figure 3-23. Vaporization plot for crankcase oil immediately after a
fresh change of Penzoil 10W-40 motor oil.
300 400 500 600
Figure 3-24. Vaporization plot of Penzoil 10W-40 after 1000 miles of
automobile immediately after a fresh change of Penzoil 10W-40 motor oil.
The differences between Figures 3-22 and 3-23 are a result of incomplete
removal of the old oil. Figure 3-24 is a fingerprinting of the same
crankcase oil after 1000 miles of driving. The compounds responsible
for the additional features of Figure 3-23 have increased in
concentration during the 1000 miles of use. While PAC appearance in
used oil may not be a direct result or cause of oil wear, PAC content
detected by GFLEF should be a useful method to monitor oil wear. The
changes that occur in PAC content in oil during use are cumulative,
detectable by GFLEF and could be correlated to the useful life of motor
Several paper samples were analyzed by GFLEF. All samples
exhibited PAC fluorescence. The vaporization plots were not useful as
fingerprints. Based on the plots, paper samples could be grouped into
only two categories. White paper such as typing, computer and notebook
paper all exhibited low intensity fluorescence. As with the refined
vegetable oil vaporization plots, photodiode array noise contributed
significantly to the plots and masked any spectral characteristics that
might help distinguish samples. Figure 3-25 is a plot obtained from the
vaporization of a sample of computer printer paper. The second group of
paper products included unbleached paper such as brown paper used in
paper bags, institutional brown paper towel and cardboard. The plots
obtained from the vaporization of this group of paper products failed to
exhibit differences needed to use the plots as fingerprints. Figure 3-
26 is a vaporization plot for a sample of brown paper towel.
L 50 4 O 00
Figure 3-25. Vaporization plot for computer paper.
,> 2000 3 0 0 o C
A0 VE IIE H 1 |
Figure 3-26. Vaporization plot for brown paper towel.
Reproducibility problems were also encountered in the analysis of
paper samples. In order to obtain reproducible plots, the sample had to
be finely divided, care had to be taken to assure that the entire sample
was in contact with the furnace wall and the furnace temperature had to
be raised slowly. If the entire sample was not in contact with the
furnace wall, components that should have vaporized at lower
temperatures were vaporized instead at higher temperatures, causing
irreproducible fluorescence signals. This problem was more severe with
samples that were not finely divided. During an investigation of
appropriate temperature ranges and ramp rates, an interesting
irreproducibility was observed with paper samples. Figure 3-27 is the
vaporization plot obtained using a fast ramp rate from 50 to 800 C.
Under these conditions, plots sometimes contained features in the 500 to
600 OC temperature range. Although the intensity of these additional
features varied greatly, when present, the temperature of the appearance
of the features and their spectral characteristics were reproducible.
In addition, the fluorescence spectral characteristics were different
from the fluorescence characteristics of components detected at lower
temperatures or with slower ramp rates. Using faster ramp rates, the
composition of the vapor at furnace temperatures of 400 to 500 C was
different from that observed at lower temperatures and different from
that observed in the same temperature range when slower ramp rates were
used. This indicated a more complicated process than components
vaporizing at higher temperatures than they would have under conditions
normally used. Possibly, components of the paper sample that were not
vaporized at lower temperatures due to the fast temperature ramp rate
0 lI l I l l I
Figure 3-27. Vaporization plot for brown paper towel using a faster
ramp rate. Note the additional features in the 500-600 OC range.
were pyrolyzed at higher furnace temperatures instead. Since components
of complex samples analyzed generally vaporize at temperatures below 500
C, pyrolysis should not complicate the screening of those samples for
PAC content. In addition, if pyrolysis does occur but can be made to be
reproducible, the fingerprinting capabilities of GFLEF will not be
affected and may even extend the applicability of GFLEF fingerprinting
to materials that contain organic material but no PACs.
The PAC class of compounds includes many compounds with beneficial
uses. Many hormones, amino acids, pesticides and drugs are PACs.
Vaporization plots were obtained from two vitamin formulations and two
over-the-counter pain medications. Figure 3-28 through 3-31 are plots
obtained from the vaporization of a multivitamin, vitamin C, aspirin and
ibuprofen, respectively. This method is potentially useful for
fingerprinting and monitoring the uniformity of such products.
The versatility of GFLEF with photodiode array detection in PAC
screening and fingerprinting of a wide array of samples types has been
demonstrated. Liquid and solid samples were analyzed with no sample
pretreatment in less than five minutes per sample. Appropriate
temperature ranges and ramp rates are easily ascertained and ensure
reproducible plots and credible PAC screening.
Figure 3-28. Vaporization plot for a multivitamin.
w iVL 300 C
0 Il l ] 1 1 1
Figure 3-29. Vaporization plot for a vitamin C formulation.
Figure 3-30. Vaporization plot for aspirin.
S2002 30 o c
S-1 1 100
Figure 3-31. Vaporization plot for ibuprofen.
LASER-EXCITED FLUORESCENCE FINGERPRINTS OF ENVIRONMENTAL
MATERIALS WITH A PULSED LASER,
GATED PHOTODIODE ARRAY SYSTEM
Analysis of crude oil, petroleum products, particulate matter,
cooking oil, food and tobacco by laser-excited fluorescence of molecular
vapors in a graphite furnace with photodiode array detection was shown
to be a simple and reliable method for PAC screening in these complex
samples. In addition, GFLEF provides reproducible, characteristic plots
useful as fingerprints for individual samples.
Using the previously described continuous wave (cw) experimental
systems, compounds that vaporized at higher temperatures or fluoresced
at longer wavelengths gave higher limits of detection due to
interference from furnace wall emission. This problem is most severe at
higher temperatures and at the higher wavelength, or red, end of the
detected wavelength range. Furnace wall emission also made interfering
contributions to fingerprints obtained from the vaporization of complex
samples. This interference was greater with samples exhibiting low
intensity fluorescence and with resinous samples such as crude oils,
which scattered furnace wall emission toward the detector to a greater
extent than samples with less complex matrices. As furnace temperature
rose, furnace wall emission increased, eventually saturating the
photodiode array. The temperature at which furnace wall emission
saturated the photodiode array varied depending on the alignment of the
optics, how well the emission was masked and how much scattering the
sample caused. Generally, photodiode array saturation occurred at
approximately 700 to 1000 C, thus limiting the usable temperature range
of the method.
Another limitation of the cw laser system was that the lasing
wavelength was not tuneable. Although 325 nm is an effective wavelength
to excite PACs with three or more rings, this lack of tunability
inherently limits the flexibility of the method.
A GFLEF experimental system utilizing a tuneable, pulsed laser and
gated photodiode array would have several advantages over the cw system.
A pulsed excitation source and gated photodiode array would allow
discrimination against background emission. This would provide
fingerprints with less contribution from background emission such as
furnace wall emission and stray light. The useable temperature range
would also be extended significantly. Tunability would enable the
choice of an appropriate excitation wavelength based on the analytes of
interest. By tuning the laser to an atomic absorption line, atomic
fluorescence information could also be used in fingerprinting complex
samples. For a given set of samples, a wavelength that provides the
most distinguishing fingerprints could be employed.
The focus of this work was to investigate these and other
prospective advantages of a pulsed laser, gated photodiode array GFLEF
experimental system. The system was used to obtain improved
fingerprints and to detect molecular as well as atomic fluorescence.
A block diagram of the experimental system is shown in Figure 4-1.
The laser was a flashlamp dye laser (model CMX-4, Chromatix, 560 Oak
Mead Parkway, Sunnyvale, CA, 94086) with about one mW average power (one
kW peak power) in the wavelength range used. The dye used was rhodamine
6G (Eastman Kodak Co., Rochester, NY, 14650). The laser output passed
through a four mm hole drilled in a flat mirror at a 450 angle before
passing lengthwise through the inside of the graphite tube.
The graphite furnace assembly was manufactured by Instrumentation
Laboratories (Model CTF555, Thermo-Jarrell Ash, 590 Lincoln St.,
Waltham, MA, 02254). The operational characteristics of this furnace
were described in Chapter 2.
The fluorescence was collected at 1800 relative to the laser
excitation. The cone of fluorescent light exiting the graphite tube
housing was folded by the mirror onto a lens that focused a 1:1 image of
the graphite furnace onto the entrance slit of the spectrograph. The
spectrograph was described in Chapter 2.
An intensified photodiode array manufactured by Princeton
Instruments, Inc., (model IR4-1024, P.O. Box 2318, Princeton, NJ,
06580) was mounted to the spectrograph with a laboratory-constructed
flange. Data collected by the photodiode array was stored and processed
by a PC's Limited 286 computer supplied by Princeton Instruments.
An air-acetylene burner and an additional lens aided in tuning the
laser to atomic absorption lines. A standard solution was aspirated
into the flame as the laser was tuned, while fluorescence was monitored
Figure 4-1. Block diagram of the experimental system. PDA, photodiode
using the same detection system employed with subsequent graphite
furnace atomization. This configuration simplified the change from
laser tuning to sample analysis.
Results and Discussion
The most obvious advantage of the pulsed laser, gated photodiode
array system was a reduction of furnace wall emission compared to the cw
system. This background emission occasionally interfered with
fingerprinting of oil samples, where furnace wall emission was scattered
to a greater extent than with standards. With the pulsed laser, gated
photodiode array system, furnace wall emission was not apparent. Figure
4-2 shows the three-dimensional plot obtained from the vaporization of a
crude oil sample using the pulsed laser system. The photodiode array
controller was triggered by a photodiode positioned to detect a small
fraction of the laser pulse scattered from the back side of the mirror.
The gate for data collection was set to correspond temporally with the
excitation pulse. Thus, fluorescence was collected only during the one
us excitation pulse. The furnace wall emission and stray light
contributions to the plot were reduced by the same factor as the duty
factor of the laser. Using a laser pulse repetition rate of 15 Hz,
these background contributions should be reduced by a factor of about
One possible advantage of the pulsed system was selectivity based
on the fluorescence lifetime of PACs. However, the laser pulse width,
the narrowest useable photodiode array gate and the smallest gate delay
all were too large to allow selective PAC detection based on
80 240 400 550 710 870 t
Figure 4-2. Vaporization plot for crude oil. Note the absence of
furnace wall emission.
fluorescence lifetime. In addition, the photodiode array, when
operating in the delayed gate mode, could collect spectra at a maximum
rate of only three Hz. These properties precluded taking advantage of
this potential capability of a pulsed, gated system. The laser and
photodiode array would require precision on the order of, at most,
several nanoseconds and a high pulse rate to achieve sensitive
fluorescence lifetime selectivity.
Advantages of the wavelength tunability of the dye laser were also
realized. The flashlamp dye laser could be tuned to an excitation
wavelength that produced the most distinguishing fluorescence
fingerprints for a given group of samples. A novel additional use for
the pulsed laser, gated photodiode array system was as a diagnostic tool
for laser-excited fluorescence of atoms produced in a graphite furnace.
Photomultiplier tubes generally used for atomic fluorescence have
greater sensitivity and linear response than photodiode arrays, so
sensitivity degraded compared to previous laser-excited atomic
fluorescence studies (53). However, the multi-wavelength nature of the
photodiode array provided a unique look at some of the processes
occurring in the graphite furnace.
As previously noted, oil samples caused increased furnace emission
to be scattered to the detector, a phenomenon which could not accurately
be accounted for by standards. The photodiode array allowed assessment
of the extent of this problem and whether it significantly interfered
with detection and quantitation of atomic species. In order to simulate
a situation in which an atomic fluorescence determination was hindered
by furnace wall emission, an oil sample containing iron was vaporized
with the laser tuned to the fringe of an iron absorption line. Figure
4-3 is the photodiode array scan exhibiting maximum fluorescence
intensity of the iron 375.5 nm line with the laser tuned to the
absorption maximum at 296.7 nm. Figure 4-4 is the photodiode array scan
exhibiting maximum iron fluorescence from a similar plot of the same
sample for which the laser was detuned to 296.5 nm. Severe interference
from the furnace wall emission is apparent. The photodiode array
allowed the assessment of the nature and extent of any spectral
interference that might be encountered. This information would be
useful in obtaining a reliable background for baseline subtraction. In
spectra and plots obtained with the Princeton Instruments photodiode
array, furnace wall emission apparently falls off at about 710 nm.
This, however, is an artifact of the software, which was written to
control a 1024 diode array instead of the 700 diode array used.
The thermal evolution, laser-excited atomic fluorescence system
was also useful as a fingerprinting technique for complex samples such
as oil. Figure 4-5 shows a three-dimensional plot obtained from the
vaporization of a sample of used jet engine oil. Fluorescence of
molecular and iron vapors contributed to the fingerprint for this
sample. In addition, this plot could be used to confirm that the
organic matrix had been vaporized prior to atomization of the iron.
The pulsed laser, gated photodiode array system provides
fingerprints with greatly reduced background contributions. Furnace
wall emission did not saturate the photodiode array, even at furnace
temperatures of approximately 2600 C required to vaporize iron.
240 400 550 710
Figure 4-3. Photodiode array scan exhibiting maximum iron fluorescence
from the vaporization plot of jet engine oil. Excitation was at 296.7
Ou e4u 4UU 550 710 870
Figure 4-4. Photodiode array scan exhibiting maximum iron fluorescence
from the vaporization plot of jet engine oil with off-line (296.5 nm)
80 240 400 550 710 870
Figure 4-5. Vaporization plot of jet engine oil.
Wavelength tunability imparts flexibility to the technique not present
with the previous cw system. This flexibility includes the ability to
add atomic fluorescence information to the fingerprinting of complex
samples based on PAC content, enhancing the capability to differentiate
SUMMARY AND FUTURE WORK
Polycyclic aromatic compounds are widespread contaminants in the
environment. They are produced mainly by the burning of fossil fuels
for energy production. Many PACs are known or suspected carcinogens and
mutagens. They are usually found as components of complex samples and,
therefore, present a difficult challenge to the important role of the
analytical chemist. Due to interest in PACs and the presence of these
compounds in complex samples, many chromatographic and spectroscopic
techniques have been and continue to be developed that provide compound-
specific information on PAC content. Unfortunately, these methods often
involve elaborate, time-consuming sample pretreatment, sequential
determination of target compounds and the use of standards for
confirmation of identification. With each step in the analysis,
confidence in the analytical result decreases. In general, PAC analysis
is tedious, time-consuming, uses very expensive state-of-the-art
equipment and requires highly trained technical experts. The most
selective techniques are also the most sophisticated.
Recently, the efficiency of current methods of analysis has been
questioned. Methods that can provide information on the presence of
PACs in a rapid and simple manner would be useful for routine PAC
screening. In addition, if the characterization of the sample is unique
to a particular sample, it can be used as a fingerprint for the sample.
Fingerprints can be used to differentiate samples and, thus, would be
helpful in tracing the origin of pollutants such as an oil spill.
The aim of this work was to show the usefulness of laser-excited
fluorescence of aromatic compound vapors produced by direct atomization
of untreated complex samples in a graphite furnace as a screening method
for PACs and as a fingerprinting technique for environmental and other
complex samples. The untreated samples were placed in a graphite
furnace and were vaporized by a gradual temperature gradient. As the
components were vaporized according to volatility, sequential
fluorescence spectra were collected by an intensified photodiode array.
These sequential spectra were plotted in the three-dimensional format of
wavelength versus fluorescence intensity versus scan number. Scan
number could be correlated to furnace temperature.
The GFLEF system was used to provide fingerprints and screen for
PACs in complex samples. Crude oil, petroleum products, urban dust
particulates, diesel particulates, standards and standard mixtures were
analyzed by GFLEF. Polycyclic aromatic compounds exhibit broad,
featureless fluorescence spectra. They also have a significant vapor
pressure over a broad temperature range and fail to undergo a sharp
transition into the vapor phase. Thus, GFLEF was able to provide
specific identification of components for only simple standard mixtures
of two or three components. However, the speed, simplicity and
sensitivity of the method to PACs allowed rapid screening for PACs in
complex samples and fingerprinting of the samples based on PAC content.
Analysis time was less than five minutes per sample.
A diverse array of sample types were analyzed to evaluate the
performance of the technique as a routine screening and fingerprinting
method. Petroleum products, tobaccos and many types of foodstuffs and
over-the-counter medications could be selectively identified based on
the fingerprints for the samples. The useful lifetime of cooking and
engine oils could be monitored due to the buildup of PACs during use.
Samples that were difficult to fingerprint generally had low PAC levels.
With these samples, such as unused refined vegetable oils, photodiode
array noise contributed significantly to the vaporization plots.
Vaporization of paper indicated that at higher temperatures, pyrolysis
may contribute to the vaporization plots.
A similar experimental system utilizing a flashlamp-pumped dye
laser and a gated intensified photodiode array was developed to
investigate the advantages of such a system. Fingerprints devoid of
background emission were obtained due to the discrimination by gateable
detection against continuous background emission. Reduction of furnace
wall emission, the most intense background contribution, greatly
extended the useable temperature range. This, in conjunction with
wavelength tunability, enabled the inclusion of atomic fluorescence
information into the fingerprints of complex samples.
Another use for the pulsed, gated system was as a diagnostic tool
for methods normally using discrete detectors. The system was used to
evaluate spectral interference encountered in atomic fluorescence
normally detected using a spectrometer and photomultiplier tube.
Laser-excited fluorescence of molecular vapors produced in a
graphite furnace was a useful PAC screening technique and fingerprinting
method for environmental and other complex samples. Optimum
experimental parameters such as sample size, temperature ranges and ramp
rates were easily and quickly ascertained without prior knowledge of
sample composition. Analysis time for samples was less than five
minutes and required no sample extractions, fractionation or cleanup of
any kind. In order to obtain reliable, reproducible fingerprints, care
was taken to obtain homogeneous, representative samples for analysis.
The complexity of samples in which PACs are found and the variability of
component concentrations facilitated fingerprinting, in contrast to the
hindrance that sample complexity usually imparts to PAC analysis. The
high temperatures attainable in a graphite furnace enabled direct
analysis of solids and resinous materials often too difficult to analyze
by alternate methods.
The informing power of this method would be increased greatly by
coupling it with a method that increases selectivity for specific
compounds. Low temperature and lifetime-selective techniques hold the
most promise in this respect. If the vapors produced by the furnace
underwent a supersonic expansion, narrowed absorption and fluorescence
peaks would allow identification of specific components. Alternatively,
the vapors from the furnace could be condensed onto the cold finger of a
helium refrigerator for subsequent low temperature fluorescence
spectroscopic analysis. Both of these methods are capable of cooling
analytes to a temperature facilitating selective identification of
components in mixtures by low temperature fluorescence.
Selective detection of components with long fluorescence lifetimes
is also possible. Requirements for lifetime selectivity include narrow
laser pulses, detector gate widths and gate delays in the nanosecond or
sub-nanosecond range. Increased spectral selectivity would enable the
investigation of the conditions and products of pyrolysis by which PACs
are produced. Studies of the matrix, adsorption and pyrolysis effects
that contribute to the appearance time of the components responsible for
the various features of the vaporization plots would also be possible.
Increased selectivity would require additional experimental system
complexity and detract from the usefulness of GFLEF as a non-selective
PAC screening technique, but would also engender other avenues of
1. P. W. Jones, Polynuclear Aromatic Hydrocarbons: Chemistry and
Biological Effects, A. Bjorseth and A. J. Dennis, eds., Battelle,
Columbus, OH, 1979
2. G. Grimmer, Environmental Carcinogens: Polycyclic Aromatic
Hydrocarbons, CRC Press, Inc., Boca Raton, FL, 1983
3. M. L. Lee, M. V. Novotny and K. D. Bartle, Analytical Chemistry of
Polycyclic Aromatic Compounds. Academic Press, New York, NY, 1981
4. U.S. Environmental Protection Agency, Scientific and Technical
Assessment Report on Particulate Polycyclic Organic Matter,
Author, Washington, D.C., 1975
5. J. C. Fetzer, Chemical Analysis of Polycyclic Aromatic Compounds,
T. Vo-Dinh, ed., John Wiley, New York, NY, 1989
6. R. C. Lao, R. S. Thomas, T. C. Mankman, J. Chromatogr. 112, 681,
7. K. Ogan, E. Katz, W. Slavin, Anal. Chem. 51, 1315, 1979
8. W. A. MacCrehan, W. E. May, S. D. Yang, Anal. Chem. 60, 194, 1988
9. I. L. Davies, K. D. Bartle, P. T. Williams, G. E. Andrews, Anal.
Chem. 60, 204, 1988
10. S. A. Wise, B. A. Benner, G. D. Byrd, S. N. Chesler R. E. Rebbert,
M. M. Schantz, Anal. Chem. 60, 887, 1988.
11. B. W. White, R. D. Smith, Chemical Analysis of Polycyclic Aromatic
Compounds, T. Vo-Dinh, ed., John Wiley, New York, NY, 1989
12. R. E. Jentoft, T. H. Gouw, Anal. Chem. 48, 2195, 1976
13. R. M. Campbell, N. M. Djordjevic, K. E. Markides, M. L. Lee, Anal.
Chem. 60, 356, 1988
14. J. C. Fjeldsted, R. C. Kong, M. L. Lee, J. Chromatogr. 279, 449,
15. E. Sawicki, T. W. Stanley, W. C. Elbert, J. D. Pfaff, Anal. Chem.
36, 497, 1964
16. E. Clar, Spectrochim. Acta 4, 116, 1950
17. G. Mamantou, E. L. Wehry, R. R. Kemmerer, E. R. Hinton, Anal.
Chem. 49, 86, 1977
18. E. V. Shpol'skii, T. N. Bolotnikova, Pure Appl. Chem. 37, 183,
19. D. M. Lubman, M. N. Kronick, Anal. Chem. 54, 660, 1982
20. D. M. Lubman, D. M. Trembreull, C. H. Sin, Anal. Chem. 57, 1084,
21. T. Vo-Dinh, Anal. Chem. 50, 396, 1978
22. C. B. McGown, Prog. Anal. Spectrosc. 11, 383, 1988
23. M. Zander, Chemical Analysis of Polycyclic Aromatic Compounds,
T. Vo-Dinh, ed., John Wiley, New York, NY 1989
24. M. Fleischmann, P. J. Hendra, A. J. McQuillan, Chem. Phys. Lett.
26, 163, 1974
25. R. Rumelfanger, S. A. Asher, M. B. Perry, Appl. Spectrosc. 42,
26. M. J. D. Low, C. Morterra, Appl. Spectrosc. 41, 280, 1987
27. M. D. Morris, Detectors for Liquid Chromatography, John Wiley, New
York, NY, 1986
28. T. Vo-Dinh, Chemical Analysis of Polycyclic Aromatic Compounds,
T. Vo-Dinh, ed., John Wiley, New York, NY, 1989
29. K. D. Bartle, M. L. Lee, M. Novotny, Analyst 102, 731, 1977
30. M. J. O'Neal, Jr., T. P. Wier, Jr., Anal. Chem. 23, 830, 1951
31. R. J. Hurtubuse, J. D. Phillip, G. T. Skar, Anal. Chim. Acta. 101,
32. P. J. Arpino, I. Ignatiadis, G. De Rycke, J. Chromatogr. 340, 329,
33. S. G. Zelenski, G. T. Hunt, N. Pangaro, Polynuclear Aromatic
Hydrocarbons: Chemistry and Biological Effects, A. Bjorseth and
A. J. Dennis, eds., Battelle, Columbus, OH, 1979
34. J .M. Shekiro, Jr., R. K. Skogerboe, H. E. Taylor, Environ. Sci.
Technol. 22, 338, 1988
35. G. Lunde, A. Bjorseth, Nature, 268, 518, 1977