The development of a laser-induced fluorescence method for rapid screening of polyaromatic hydrocarbons


Material Information

The development of a laser-induced fluorescence method for rapid screening of polyaromatic hydrocarbons
Physical Description:
vi, 149 leaves : ill. ; 28 cm.
Kirsch, Barbara A ( Barbara Anne ), 1958-
Publication Date:


Subjects / Keywords:
Fluorescence   ( lcsh )
Hydrocarbons   ( lcsh )
Cancer -- Diagnosis   ( lcsh )
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )


Thesis (Ph. D.)--University of Florida, 1986.
Includes bibliographical references (leaves 145-148).
Statement of Responsibility:
by Barbara Anne Kirsch.
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 - 000951381
notis - AER3624
oclc - 16952126
System ID:

Full Text







Dedicated, again,

to my

father and mother


My deepest gratitude goes to Dr. James Winefordner for his

support, guidance and encouragement throughout these seemingly

impossible, long years. It has been an honor to be a part of his

diverse and always enthusiastic research group. Thanks are also in

order to Dr. John Dorsey for his encouragement along the way, which

sometimes made the difference between coming and going. I would like

to thank Dr. Ben Smith for his guidance and assistance, and for making

our group an enjoyable one. Thanks also to Dr. Ed Voigtman for his

advice and useful discussions.

Special heartfelt thanks go to Dr. Jose Lanauze, a friend to whom

I am deeply indebted and forever grateful.

For their friendship and faithful support, I would like to thank

Dr. Michael Kosinski, Dr. Jodie Johnson and Dr. Jonell Kerkhoff. I

would also like to express my gratitude to the members of Dr.

Winefordner's research group, especially Dr. Shigeki Hanamura and Mike

Rutledge for their time, patience, and friendship, and to Jeanne

Karably for her assistance throughout.

Thanks are also in order to the Chemistry Department Machine Shop

and Glass Shop members, who were always willing to help.

I would like to thank Lisie Torres for being a wonderful friend,

and Mario Tremblay for filling the last few years with interesting


A very, very special thanks for the love and happiness I have

found with Chris Dykstal, and for his help in the final hours.

Lastly, I would like to thank the people at home I love, my

parents and sisters and brother, for giving me the courage to go on.





ABSTRACT . . . . . . .. v


I INTRODUCTION . . . . . 1

Polyaromatic Hydrocarbons . . . 1
Analytical Methods . . . 10
Goals of This Work . . . 17



Introduction . . . . 32
Experimental . . . . 34
Results and Discussion . . . 45


Introduction . . . . .. 55
Experimental . . . . 57
Results and Discussion . . . 61


Introduction . . . . 66
Experimental . . . . 69
Results and Discussion . ... . 88


REFERENCES . . . . . . 145


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




December 1986

Chairman: James D. Winefordner
Major Department: Chemistry

Several methods are evaluated for rapid screening of environmental

samples based on their polyaromatic hydrocarbon content. First, a

laser microfluorometer is constructed and evaluated in terms of

sensitivity and applicability to the in-situ analysis of airborne

particulates. Second, a laser system is investigated for use as a

vaporization and excitation source of polyaromatic molecules adsorbed

to particulate matter. Thirdly, electrothermal vaporization of

polyaromatic molecules contained in particulate matter and other

environmental samples, combined with laser excitation of the resultant

vapor, is evaluated in terms of sensitivity and feasibility as a

routine screening method which could precede more complicated methods

of analysis.

Electrothermal vaporization is combined with multichannel

detection to acquire vapor phase spectra of various polyaromatic hydro-

carbons. A graphite furnace is utilized to provide a thermal

separation of polyaromatic hydrocarbon mixtures, and rapid data

acquisition with a multichannel analyzer provides a three dimensional

map of the eluting vapors.

Finally, suggestions are made as to the future work necessary to

improve the rapid screening capabilities of electrothermal vaporization

combined with multichannel detection.


Polyaromatic Hydrocarbons

The increasing industrialization of modern society and the

increasing demands for new energy sources necessitates the use of

analytical chemistry to closely monitor environmental quality and

assess its effects on human health. Polycyclic aromatic hydrocarbons

(PAHs) are widespread contaminants throughout the environment, produced

mainly by the incomplete combustion of organic material. There are

several hundred compounds included in the chemical class of PAHs, many

of which are carcinogenic or act as cocarcinogens.

PAHs consist of two or more fused benzene rings. The

arrangement and number of rings is related to carcinogenic activity.

PAHs can be arranged in a linear, cluster or angular fashion. A

cluster arrangement contains at least one benzene ring surrounded on

three sides (1). The angular PAHs are the most stable while the

cluster PAHs exhibit the most carcinogenic behavior. Generally, 3 to 6

ring PAHs exhibit carcinogenic activity. The location of an alkyl

group within the PAH can have a crucial effect on carcinogenicity,

since substitution can alter the electron distribution within the

molecule (2,3).

The first step in the link between PAHs and cancer was the

discovery of increased incidence of cancer of the scrotum among chimney

sweeps by a London surgeon in 1775 (4). Eventually, cancer was induced

in animals by repeated application of tar and coal extracts to the

skin. Efforts to isolate the active ingredients of these extracts led

to the identification of various PAHs and subsequent biological testing

of their carcinogenic activities. It has been estimated that the

mortality rate from lung cancer has increased more than 30 times since

1900; PAHs adsorbed to airborne particulates and tobacco products

are believed to be the major contributors to this increased death

rate (4).

Formation of PAHs

The vast majority of PAHs are formed by incomplete combustion of

organic materials. At high temperatures, organic compounds are

partially cracked to smaller fragments pyrolysiss), mostly free

radicals. These free radicals can recombine (pyrosynthesis) to form

larger, more stable aromatic hydrocarbons. Under conditions where

appropriate amounts of oxygen are present, an organic material can be

combusted to CO2 and H20 more completely, and lesser amounts of PAHs

are formed. The amount and distribution of PAHs formed is dependent on

several variables, including temperature, starting material and

combustion conditions. Similar PAH profiles are obtained with

different materials combusted at the same temperature. However, low

temperatures tend to yield higher concentrations of substituted PAHs,

while higher temperatures yield mostly unsubstituted PAHs (2,3).

Pyrolysis of a substance with unsaturation or chain branching produces

increased amounts of PAHs and PAHs are formed more easily from pyroly-

sis of compounds that already contain cyclic structures. In addition,

nitrogen, sulfur and oxygen analogs of PAHs can be expected where

nitrogen, oxygen or sulfur exist in the starting material (2).

Polyaromatic hydrocarbons also occur naturally, through the

carbonization of organic materials. The formation of PAHs from de-

graded biological material at low temperatures (200-300 C) over a

period of millions of years produces complex mixtures which are very

different from PAHs obtained from combustion sources. They are rich in

alkylated and hydroaromatic species, while PAHs formed from high

temperature combustion processes are mainly unsubstituted (2). The

mechanism of formation is not well understood. It has been noted that

PAH fractions from oils of different origin are very similar in

composition (3). Generally, it is believed that PAHs are derived from

preformed structures, such as pigments (5,6,7). Several mechanisms for

the formation of various PAHs in coal, petroleum, shales, crude oil,

and other sediments have been proposed (5-11).

Sources and Distribution of PAHs

The primary technological source of PAHs released into the

environment is the incomplete combustion of fossil fuels for energy

production. Table 1 shows the major emission sources in the United

States for the period from 1971-1973 in terms of Benzo(a)pyrene (B(a)p)

emission and the percentage each emitter contributes to the estimated

total emission (12). In most of the work done in the past, the

prevailing idea has been to assess the carcinogenic potency of an

emitter in terms of a single PAH, namely B(a)p because of its high

relative carcinogenicity and ease of determination. However,

measurement of B(a)p alone has been criticized as underestimating the

total PAH emission and the relative carcinogenicity of the emissions.

Only when emission sources produce the same quantitative profile of

Table I. Major PAH Sources and Estimated B(a)P Emissions (12)

Source B(a)P emission % of Total emissions

Coal refuse fires 310 34.7
Residential furnaces, 300 33.6
coal (hand-stoked)
Coke production 170 19.0
Vehicle disposal 25 2.8
(open burning)
Wood burning 25 2.8
Mobile sources, gasoline 11 1.2
Tire degradation 11 1.2
Forest and agricultural 11 1.2
refuse burning
Open refuse burning 10 1.1
Intermediate coal furnaces 7 0.8
Enclosed incineration 3 0.3
Other (coal steam power, 4 0.5
plants, asphalt air-
blowing, oil and natural
gas combustion, diesel
powered vehicles)

Totals 894 100.0

* (1971-1973)

PAHs can one particular PAH be used to assess the total PAH burden.

Current emphasis is on more detailed analyses of the complete PAH


Polyaromatic hydrocarbons in the air are found mainly adsorbed to

carbonaceous particles (soot) produced in the process of combustion.

They may also be associated with other types of aerosols, such as dust

from wind, sea spray, forest fires, or volcanic dust (2,12). Generally

PAHs are concentrated on submicron size particles, with 70-90% of the

total hydrocarbon content being associated with particles less than

5.0 im in diameter (2-4,12). Transport of particulates occurs

through wind currents, gravitational settling, impaction on solid

surfaces, and removal by precipitation. Smaller, submicron-sized

particles have longer residence times in the atmosphere, on the order

of a few days to several weeks, than larger, 1-10 pm particles, which

have residence times of a few days to less than half a day. In

addition, the residence time of many PAHs depends on photodecomposition

reactions in the atmosphere with ozone, nitrogen oxides, sulfur oxides

and other oxidants (2,12).

The potential hazard particulates carrying PAHs present to human

health depends on particle size not only from the standpoint of

residence time in the atmosphere, but also from the point of incorpora-

tion and retention of such particulates into the lung. Figure 1 shows

the retention of particulate matter by the lung in relation to particle

size (4). One can see that the greatest retention occurs for particles

less than about 1.5 pm in diameter, coincidentally the particles which

contain the majority of the PAH content. Particles greater than about

0 0 0
0 k. 0r





o _



0 0 0 0 0
10 10 N -

(%) N011N313y

3.0 pm are largely retained by the mucous membranes of the nose, oral

cavity and pharynx (3,4).

Although the large majority of attention has been focused on PAHs

in the air, PAHs are found throughout our environment; in water bodies,

in soil and sediments, in our fossil fuels, and in our food. The

presence of PAHs in water bodies can be attributed to fallout from air

particulate matter, as well as to such things as spillage of crude oil,

oil drillage, release of effluents from industrial plants, etc.

Generally, the solubility of PAHs in water is low, and a large

proportion of PAHs in water is assumed to be associated with suspended

solids (2). PAHs are incorporated into sediments due to their limited

solubility in water, or from fallout and precipitation from the

atmosphere. The presence of PAHs in foodstuffs can be attributed to

such factors as cooking methods, food processing, food additives, and

curing smokes. Table II lists the PAHs which occur most frequently in

the environment, along with their molecular formulas, molecular

weights, boiling points, and carcinogenic activities (2,3).

The natural degradation of PAHs has failed to keep pace with the

production of PAHs due to the increased industrial development of this

century. Due to the health hazards associated with the increased

levels of PAHs and an increased interest in understanding the mechanism

of their carcinogenic activity, many analytical methods have been

applied to their analysis over the past few decades. Lee, Novotny and

Bartle have reviewed the analytical methods applied to the analysis of

PAHs (2).

Table II. PAHs Occurring Most Frequently in the Environment

Compound Mol. Mol. Bp Carcinogenic
formula weight ( C) activity

Fluorene C13H10 166 293 0

Phenanthrene C14o10 178 338.4 0

Anthracene C14H10 178 340 0

Fluoranthene C16H10 202 383.5 0

Pyrene C6 H10 202 393.5 0

Benzo(ghi)fluoranthene C18H10 226 431.8 0

Cyclopentadieno(cd)pyrene C H10 226 439 +

Benz(a)anthracene C18H12 228 437.5 0/+

Triphenylene C18H12 228 438.5 0

Chrysene C18H12 228 441 +

Benzo(b)fluoranthene C20H12 252 481.2 ++

Benzo(k)fluoranthene C20H12 252 481 ++

Benzo(j)fluoranthene C20H12 252 480 ++

Benzo(e)pyrene C20H12 252 492.9 0/+

Benzo(a)pyrene C20 H12 252 495.9 ++

Perylene C20H12 252 497 0

Indeno(1,2,3-cd)- C22H12 276 531 ?

Indeno(1,2,3-cd)- C22H12 276 534 +

Benzo(ghi)perylene C22H12 276 542 +

Anthanthrene C22H12 276 547 ?

Coronene C24H12 300 590 0/+

* based on the % of test animals which developed tumors:
0 = none, + = up to 33%, weakly carcinogenic, ++ = above 33%,
strongly carcinogenic, ? = not given

Analytical Methods

Analysis of the PAH content of environmental samples involves

extraction, isolation of the PAH fraction, separation and quantitation

of individual PAHs. The choice of the extraction method depends on the

sample type. Problems can be encountered with extraction if the sample

matrix cannot be dissolved in a suitable solvent, i.e. in the case of

samples which contain inorganic constituents or graphite-like

structures. Adsorption of the PAHs by the insoluble matrix consti-

tuents can cause incomplete extraction. Organic solvents such as

acetone, chloroform, etc., are used in cases where the sample is

completely soluble; for the more difficult case of soot and graphite-

like materials, boiling xylene or toluene is more effective. Several

extraction methods suited for different sample types are given by

Grimmer (3).

The PAH fraction usually contains more than 100 PAHs. Thus,

separation techniques of high resolution, giving more than several tens

of thousands of height equivalent theoretical plates (HETP), are

needed. The best results are obtained with glass capillary columns in

gas chromatography, where routine separations give more than 70,000

HETP (3). High pressure liquid chromatography yields approximately

20,000 HETP and packed high capacity columns produce about 25,000 HETP

(3). Other chromatographic methods such as column adsorption, paper,

and thin layer chromatography have been used in the past, but are

generally reserved to simplify complex PAH mixtures, or ascertain

suitable separation conditions for higher resolution methods. High

performance liquid chromatography is widely used for the fraction-

ation of complex mixtures and sample cleanup.

Various detectors have been applied to the gas chromatography of

PAHs; among the most widely used is the flame ionization detector. The

signal is proportional to the amount of carbon combusted and gives a

direct presentation of the quantitative composition of the PAH

mixture. Ultraviolet absorption and fluorescence detectors have found

general use with HPLC systems, as well as some use in gas chromato-

graphy of PAHs. Of course, the combination of gas chromatography and

mass spectrometry (GC/MS) is one of the most powerful, and is commonly

used for the detailed analysis of complicated mixtures. New rapid scan

techniques coupled with high resolution chromatography have greatly

surpassed any other method or combination of methods for PAH analysis

(3). New ionization schemes which can differentiate between isomeric

PAHs are being developed to increase the analytical utility of GC/MS.

Ultraviolet absorption and luminescence techniques are also

widely used for the analysis of PAHs. The absorption of visible or

ultraviolet light (transition of 5 electrons from the I to 5* orbital)

gives characteristic absorption and luminescence spectra (see page 28)-

Luminescence methods have generally replaced ultraviolet absorption

methods due to increased sensitivity (10 to 103 times as great).

However, absorption spectroscopy can be used in conjuction with

luminescence techniques to provide additional information. As

previously mentioned, absorption and luminescence techniques have been

used for detection of PAHs in HPLC, in gas chromatography, and in thin

layer chromatography. Because luminescence and absorption spectra can

be broad and sometimes featureless, efficient separation techniques

must be used to minimize spectral overlap or methods for selective

excitation of components must be used. Such selection methods include

low temperature luminescence in Shpol'skii solvents or in organic

glasses, matrix isolation and supersonic expansion spectroscopies.

These techniques offer a significant reduction in the bandwidth of

absorption, so that selective excitation of fluorescence or phosphor-

escence of an analyte in a complex mixture can be achieved. These

techniques have been applied to extremely complex samples, such as

fossil fuels, with dramatic results (2,13,14).

Enhanced selectivity can be obtained through utilizing more than

one of the four spectra offered through luminescence techniques (emis-

sion and excitation of phosphorescence and fluorescence). Thus, total

luminescence plots of excitation wavelength versus emission wavelength

yield contour maps which can give more spectral information than a

single emission spectrum obtained at a fixed excitation wavelength

(2,15). Various other parameters can be utilized in luminescence

measurements to give multidimensional spectra for enhanced selectiviy.

Multidimensional measurements are finding increased application to

characterization of HPLC and GC effluents (15).

Spectral selectivity can also be improved by synchronous lumin-

escence techniques. In conventional luminescence techniques, either

the excitation or emission wavelength is fixed while the other is

scanned. In synchronous luminescence, both emission and excitation

monochromators are scanned, while maintaining a constant energy or

constant wavelength difference between them. The spectral structure of

the system becomes better resolved because of bandwidth narrowing of

the individual emission lines and decrease of spectral overlap from

various components in the mixture (16). The use of synchronous


luminescence for rapid screening of environmental samples will be

discussed in more detail below.

Rapid Screening

While many of the previously mentioned analytical techniques

provide detailed and compound specific information on PAH content

of complex environmental samples, they may require elaborate, time

consuming analysis procedures and costly, sophisticated instrumenta-

tion. The application of these procedures to routine monitoring

programs which could involve large numbers of samples is not practical.

Simpler, less costly screening techniques can be used to give prelim-

inary information of PAH content and to prioritize samples for further

analysis. It is desirable that such screening methods yield a spectral

profile of the samples and determine whether a sample contains detect-

able PAHs, and if the PAH content of several samples is similar.

Several examples will be given.

Synchronous luminescence has been applied by Vo-Dinh et al. to the

characterization of five synthetic fuel products which were collected

at various locations of a fuel production facility (16). Figure 2

shows the synchronous profiles of the five products (diluted 10-5 fold

in ethanol). The constant wavelength difference between the two

monochromators was chosen to be 3 nm to correspond to the Stokes shift

so that the synchronous fluorescence peaks correspond to the 0,0 band

emissions of most PAHs (refer to Chapter 2, Theoretical Considerations,

for a more detailed explanation on molecular fluorescence).

As a rule, the position of the 0,0 band of high number ring linear

cyclic compounds occurs at longer wavelength than that of lower number

ring compounds. Thus, upon inspection of figure 2, several conclusions

Synchronous luminescence profiles of 5 synthetic fuel
products (16).

Figure 2.





" \

> X3

-i -


300 400 500



can be made right away: 1) Product C contains the least amount of

PAHs. The peak at 285-290 nm is also present in products A, B, D and E

and is known to be a monocyclic aromatic species. 2) Product D

contains more PAHs than product C, but less than products A, B, and E.

3) Products A and D are similar, but A contains more PAHs with 0,0

bands at 346, 402, and 442 nm. 4) Products B and E contain even more

PAHs at these 0,0 bands. 5) The profiles of B and E are similar. The

resulting preliminary ranking of these five products for PAH content

was given as B > E > A > D > C.

Kerkhoff et al. applied constant energy synchronous fluorescence

(CESL) to the spectral fingerprinting of ambient air samples (17).

Samples included gasoline engine exhaust, diesel engine exhaust, air

near a heavily traveled interstate, and air from a moderately polluted

urban area. Spectral fingerprints of these samples were obtained, and

5 major PAHs in the gas phase extracts were characterized and

estimated. CESL was applied by L. A. Files and workers to the

fingerprinting of gasoline exhaust samples (18).

Room temperature phosphorescence (RTP) has also been applied to

the screening of cigarette smoke condensate (16). In RTP, a sample is

spotted onto solid substrates such as filter paper, silica gel or

alumina and the phosphorescence is measured by a spectrofluorometer

equipped with a rotating phosphorescope. RTP has been shown to be

enhanced by the addition of an external heavy atom (19). Both the

selectivity and sensitivity are improved. The sensitivity can often be

improved by orders of magnitude. RTP was used to assess the total

particulate matter in cigarette condensate. Cigarettes were smoked by

a smoking machine and the total particulate matter was collected on a

glass fiber filter. After the filter was extracted in ethanol, the RTP

signals (enhanced by lead acetate as the heavy atom) were found to be

proportional to the number of cigarettes smoked per extract. Ranking

according to the RTP signals of three different cigarette brands was in

keeping with their tar contents.

Laser-induced fluorescence has been applied to the spectral

fingerprinting of oils. Oils can be classified into groups based on

their fluorescence spectra as seen in figure 3 (20). Fantasia, Hard

and Ingrao examined the fluorescence characteristics of 29 crude oils

and concluded that each sample could be characterized by a measurement

of peak emission wavelength, fluorescence lifetime, and fluorescence

efficiency (21). As an example, figure 4 includes the fluorescence

spectra of several crude oils, along with their "fluorescence decay

time spectra." Although the fluorescence spectra are broad and

featureless, the combination of the fluorescence spectrum and time

decay curve provides a definitive fingerprint of that oil. The same is

illustrated in figure 5 for several refined petroleum products.

Again, these simple screening methods are not meant to compete

with the more complicated methods of analysis previously mentioned.

They may only provide a rapid assessment of PAH content, and yield a

spectral "fingerprint", indicating something about the type of PAHs

present, or simply allow one to distinguish one sample from another, as

in the fingerprinting of oils. Advantages are simplicity, speed of

operation, and relative low cost.

Goals of This Work

Laser-induced fluorescence is a highly sensitive technique for the

analysis of PAHs in solutions. Sub-parts-per-trillion limits of detec-

Figure 3. The classification of oil groups based on their fluorescence





















Figure 4. Fluorescence spectra and fluorescence decay time spectra for
several crude oils (21).

#11(Leduc crude, 39.80 API)


4C *13(Cabinda crude 30.2 )

I 4Aed)umcude, 2g.*fA)

..- .-."'#6(Bunker,16.7 API)

400 450 500 550 600

Emission wavelength (nm)

11 (Leduc crude, 39.

So10 (NSO,36.0 APR)

S/ 13 (Cablnda crude
S/30.20 API)

S#1 (Medium crude,
4 y26.4 API)
*6(Bunker, 16.70AP

0 2

Fluorescence wavelength (nm)

3 API)

400 500 600 700


Figure 5. Fluorescence spectra and fluorescence decay time spectra for
several refined petroleum products (21).

~4 Furnace fuel
S~ Diesel fuel
S30- Esso extra
-Esso gasoline
Stove oil

o 10-

0 400 500 600
Emission wavelength (nm)


Furnace fuel


28 -


Diesel fuel
20 -

S 8-

30 1 912-
350 390 430 470

Emission wavelength (nm)

tion have been reported (22,23). It was the primary goal of this work

to develop a sensitive laser fluorescence detection technique for the

rapid screening of the PAH content of airborne particulates and other

environmental samples. Several approaches were tried.

In the first approach, a laser microfluorometric system was

utilized to detect PAHs directly on particulate matter and discern if

the fluorescence emission was characteristic of sample types. The

sensitivity of this system was tested using Rhodamine 6G, a very

efficient fluorophore, as a model. The fluorescence spectra of

individual particles in particulate matter were collected and evaluated

in terms of feasibility as a screening tool.

Secondly, laser vaporization of particulate matter was attempted

to determine if PAHs adsorbed to soot could be vaporized nondestruc-

tively and simultaneously excited to produce fluorescence. Detection

with a multichannel analyzer to produce a full wavelength spectrum was


Thirdly, electrothermal vaporization of particulate matter was

utilized to thermally remove PAHs from particulate matter. Laser-

induced fluorescence of the vapor produced was detected in several

ways. First, to test the sensitivity of the combination of electro-

thermal vaporization and laser-induced fluorescence, limits of detec-

tion for several PAHs were determined using an N2 laser/monochromator/

PM tube system. Alternately, multichannel detection was employed to

give the entire wavelength spectrum of the vapor. Limits of detection

were determined with a PAR model 1205D Optical Multichannel Analyzer

(OMA) for several PAHs.

Due to limitations of the speed of data transfer of the 1205D OMA

to a peripheral storage device and limited memory space inside the OMA

itself, only one spectrum could be acquired during the entire vaporiza-

tion event (generally 20 seconds). It was desired to obtain several

spectra during the vaporization so that a third dimension of time could

be added to the intensity versus wavelength information to give a three

dimensional map of the eluting vapors. A manual means of doing so was

devised and the separation of PAHs based on their boiling point differ-

ences was demonstrated.

An OMA III system capable of extremely rapid spectral data aquisi-

tion was donated by PAR. This system consisted of a 1024 element sili-

con diode array detector with a thermoelectric cooler and a model 1460

system processor. This system was used to rapidly obtain three

dimensional "fingerprints" of a variety of environmental samples, and

limits of detection for several PAHs were determined.


The absorption of ultraviolet or visible light by a molecule

raises the molecule to a higher electronic state. In order to return

to the ground state, the molecule must lose its excess energy. Both

radiative and nonradiative decay processes can occur. These processes

are illustrated schematically in figure 6.

The electronic states of a molecule consist of associated vibra-

tional and rotational levels. The energy spacing between vibronic

levels and between electronic levels is on the order of 0.1 eV, and

several eV, respectively. An energy of 0.1 eV corresponds to approxi-

mately 1160 K, so that at room temperature there is insufficient

energy to populate any of the excited vibrational levels of the ground

state. Rotational levels have energies of about 0.01 eV, transitions

too small to be observed at room temperatures in liquids. Electronic

states containing electron pairs of opposite spin are termed singlet

states, while those containing electrons of the same spin are known as

triplet states.

Absorption (A) of radiation raises an electron from the ground

state (SO) to one of many vibrational levels in one of the excited

electronic levels (usually the first excited singlet state, S ). If

the energy absorbed exceeds the bond energy of the molecule, the excit-

ation energy is lost by dissociation. If not, the excess vibrational

energy of the electron is then rapidly dissipated in collisions with



0 tW


0 <-
D (D


0 I-D

* CL

v 00



,, v'~c~j-O

10 CM- 0
t o


neighboring molecules, termed vibrational relaxation (VR), to bring the

molecule to the lowest vibrational level of the excited state. From

there, several pathways exist for the return of the molecule to the

ground state. The direct return to one of the vibrational levels of

the ground state with subsequent emission of a photon is known as

fluorescence (F). Again, excess vibrational energy is then dissipated

by vibrational relaxation.

Processes competing with fluorescence include internal conversion

(IR) and intersystem crossing (IC). Internal conversion is the

nonradiative return of the molecule to the ground state with energy

being given off as heat. Internal conversion is favored in molecules

where there is a large degree of vibrational overlap between the

excited state and the ground state. Rigid ring structure molecules,

such as PAHs, do not possess high vibrational degrees of freedom and so

frequently exhibit fluorescence. In certain instances, a molecule can

undergo a change in spin multiplicity and undergo intersystem crossing

from the lowest excited singlet state to the lowest excited triplet

state. Even though this is a forbidden transition, some probability

exists for its occurrence because the lowest vibrational level of the

triplet state is lower than the singlet state. At this point,

vibrational relaxation brings the molecule to the lowest vibrational

level of Lhe triplet state; return from this level to a vibrational

level of the ground state by emission of light is termed phosphor-

scence (P). This process also requires spin reversal and is character-

ized by a long lifetime due to the low probability of this transition.

Deactivation of the triplet state may also occur by spin reversal and

internal conversion.

While absorption occurs within 10 s, the lifetime of fluor-
escence is generally on the order of 10 s, and that of phosphor-

escence is 10 -101 s. Vibrational relaxation and internal conversion

occur on a time scale of 10-13-10-12 s.

The intensity of a luminescence process can be related to concen-

tration through Beer's law. The amount of light absorbed is given by

I = I I = I (1-e-ebC) (1)
a o o

where I is incident light intensity, I is the transmitted light
intensity, is the molar absorptivity, b is the path length, and C is

the concentration of the absorber. The fluorescence intensity If will

be proportional to the intensity of light absorbed I and to the

fluorescence efficiency D :

I I (l-e- ) (2)

The fluorescence efficiency q is related to the rate of fluorescence kf

and to the rate of competitive deactivation processes (klR and klcC

k f (3)

k + klR + klC

For solutions in which not over 2% of the total excitation energy is

absorbed, equation 2 reduces to

If =- I ebC (4)
f o

Organic molecules are capable of absorbing electromagnetic

radiation because they contain valence electrons which can be excited

to higher energy levels, either electrons involved in bonding or those

that are unshared, or nonbonding. The valence electrons can form two

types of bonding orbitals: sigma (a); and pi (5). Sigma orbitals are

associated with single bonds and require large amounts of energy for

excitation, restricting absorption to the vacuum ultraviolet. Double

bonds involve the formation of one a and one bond. Pi bonds are de-

localized and the electrons move freely throughout the atoms in the


The electronic transitions in the ultraviolet and visible region

involve 5 electrons and nonbonding electrons. Nonbonding and

electrons can be excited to the 4* level. For PAHs, the 4- *

transition is responsible for the characteristic absorption and

fluorescence spectra. Generally, the absorption spectrum contains more

than one electronic band, indicating absorption to more than one

excited state, and the fluorescence spectrum contains only one band,

indicating transitions from the first excited state to the ground

level. Absorption spectra contain information on the vibrational

spacings of excited electronic states while the fluorescence spectra

contain information concerning the vibrational spacings of the ground

electronic state.



The initial goal of this study was to investigate the usefulness

of a laser microfluorometric system for the direct, in-situ analysis of

particulate matter. It was thought that the fluorescence emitted by

the particulates might produce a useful fingerprint of the PAHs

present, based on such factors as fluorescence maxima, minima,

intensity ratios, and lifetimes.

Laser-induced fluorescence microscopy is a very sensitive tech-

nique for ultratrace analysis. The use of a laser as a light source in

fluorescence microscopy was first suggested by Masberg and Kusnetz

(24). They reasoned that 1000 times greater irradiance could be

obtained with continuous wave lasers as compared to a conventional

mercury arc lamp. Since then, various combinations of laser sources

and detection systems have been used to achieve increased sensitivity

(25-28). Hirschfeld reported on the detection of one molecule of

polyethyleneimine bound to y globulin, tagged with 80-100 fluorescein

isothiocyanate molecules (27). This was accomplished by illuminating

the sample at light intensities high enough to produce photochemical

bleaching during the observation period, thus producing a short

fluorescence pulse which contained the maximum signal the fluorescent

tags could produce. More recently, Dovichi et al. achieved a detection

limit of 8.9 x 10-11 M for Rhodamine 6G (R6G) in the liquid phase using

a flow cytometer system with a probe volume of 11 pL (28). This is the


equivalent of 22000 molecules of R6G through the probe volume during a

1 second integration time.

Additional advantages of using a laser as an excitation source

include the ability to study the temporal decay of fluorescent events,

and the ability to focus the laser to near-diffraction limits, permit-

ting high spatial resolution.

The polyaromatic hydrocarbons are characterized by high quantum

yields and thus particulates carrying PAHs would appear to be ideal

candidates for analysis by laser microfluorescence. Advantages would

include no solvent fluorescence or Raman scatter, no need for solvent

extraction, and the particles themselves constitute a source of pre-

concentrated PAHs.

In this work, a laser microfluorometric system was evaluated in

terms of sensitivity for the analysis of small (10 pm) fluorescent

particles. Limits of detection were determined for R6G. R6G was

chosen as a model fluorophore due to its high quantum efficiency. The

R6G was adsorbed onto the surface of small silica spheres which were

viewed individually with a fluorescence microscope.

This system was modified to include a scanning monochromator in

place of the spectral filters, and a pulsed N2 laser and boxcar

average. Various PAHs were adsorbed to silica spheres and graphite

powder to simulate particulate matter, and the fluorescence spectra

collected and compared. This system was used to collect the

fluorescence spectra of individual particles in urban dust. A

discussion of the usefulness of laser microfluorometry for

fingerprinting of airborne particulates is given.



A block diagram of the laser microfluorometer as used for the

determination of limit of detection of R6G is shown in figure 7. An

Ar-ion laser was used as the excitation source with all-line visible

output in the light regulated mode. A laboratory constructed fiber

optic coupler (70% efficient) was used to direct the output to an

optical fiber. The coupler included a solenoid shutter to block the

beam from the microscope when the sample was not being observed. This

minimized fading of the fluorescence.

The output end of the fiber was positioned to illuminate the area

under the objective of a fluorescence microscope. The fluorescence was

collected by a 40X objective and directed upward to a phototube attach-

ment onto a 1 mm spatial filter. Spectral filtering of the fluores-

cence was accomplished with a 550 nm (40 nm FWHM) bandpass filter. An

IP28 photomultiplier tube operated at 700 V was used as the detector.

The signals were amplified by an electrometer and then filtered by a

laboratory constructed low pass filter with a cutoff frequency set at 1

Hz. The output of the filter was fed to a strip chart recorder.

Figure 8 shows the modifications made to the system following the

determination of limit of detection for R6G. The spectral filters were

replaced by a scanning monochromator which was mounted directly to the

microscope phototube. The output of a Nitromite pulsed N2 laser (337

nm) was coupled by a Nikon X20 objective to a optical fiber. The

output was measured as 35 PJ/pulse. The fiber was again positioned to

illuminate the area under the objective of the microscope. An LG 350

cutoff filter was placed inside the microscope to reduce scatter of the



Cl -O

0 L.
o i.










laser. A portion of the N2 laser was directed to a photodiode to

trigger the boxcar. The output of the boxcar was sent to a strip chart

recorder. This system was used to examine the fluorescence of PAHs

adsorbed to silica spheres and graphite powder.

This same system was used to detect fluorescence of individual

particles of urban dust, with the exception that a Molectron UV-24

pulsed N2 laser was used in place of the Nitromite due to its increased

output power (8 mJ/pulse).

Table III contains a complete listing of the above described

equipment and manufacturers.


Limits of detection of R6G (Eastman Kodak, Rochester, NY) adsorbed

to silica were determined first. Silica spheres were prepared as

standards in the following way. Weighed amounts of 10 pm silica

spheres (Spherisorb, Phase Separations, Hauuppauge, NY) were placed in

test tubes. To each tube was added a 0.1 mL aliquot of a different R6G
solution in ethanol, the solutions ranging in concentration from 10 M
to 10 M. The solvent was allowed to evaporate, leaving the R6G

adsorbed to the silica spheres. The test tubes and glassware used in

these steps were silanized with dichlorodimethysilane to minimize loss

of R6G by adsorption onto the glass surfaces. Rinsing the test tube

walls with small portions of ethanol combined with treatment in an

ultrasonic bath was repeated several times to minimize loss of R6G and

to ensure uniform coating of the spheres. To estimate the concentra-

tion of these standards in terms of numbers of molecules of R6G per

silica bead, a 2 mL portion of the silica beads was packed into a test

tube with constant tapping to aid settling of the spheres. Two milli-



Ar-ion laser


X40 objective

PM tube

HV power supply



N2 laser



N2 laser




CF Fluor



61 OB



SRS 250


UV 24

162 with
164 plug

Mountain View, CA

Nikon, Garden City, NY

Nikon, Garden City, NY

Hamamatsu Corp.,
Middlesex, NJ

Heath Corp., Benton
Harbor, MI

Keithley Instruments
Inc., Cleveland, OH

Fisher Scientific,
Lexington, MA

Photochemical Research
Associates, Inc.
London, Ontario

Stanford Research
Systems, Inc.
Stanford, CA

Jobin Yvon American,
ISA, Inc., Metuchen, NJ

Molectron Corporation
Sunnyvale, CA

Princeton Applied
Research, Princeton, NJ

liters of the spheres weighed 1.38 g. The number of spheres in

1.00 g can be estimated by assuming a certain packing arrangement,

for instance, cubic packing. The percent free space in a cubic

packing arrangement is 47.6% (29), hence 52.4 % is volume occupied

by the packing material. From the above and the volume of a 10 Pm

sphere, one can estimate there are 1.5 x 10 silica spheres/g in a

cubic packing arrangement. From the known weights of the silica in

each tube, and the known concentration and volumes of R6G added, con-

centrations were calculated in terms of molecules of R6G/sphere. It

should be noted that the assumption of cubic packing is a "worst case"

possibility, i.e., this is a very inefficient packing arrangement. It

might be more reasonable to assume an arrangement like hexagonal

closest packing, in which one sphere would be surrounded by six

spheres, all of which lie in the same plane and touch each other. By

assuming cubic packing, the number of spheres that are present in a

gram of the material is probably an underestimate, and hence the amount

of R6G/sphere is overestimated. In theory, this would mean the detec-

tion limit for R6G is even lower.

To prepare a calibration curve, the spheres from a given standard

were spread on a quartz slide. The fluorescence emission was measured

for each of ten individual spheres to obtain a mean value and a percent

relative standard deviation. The microscope focus was adjusted so that

the image of a silica sphere would fill the 1 mm aperture, giving

maximum response at the photodetector. A halogen lamp in the base of

the Nikon microscope was used to illuminate the field while positioning

a sphere under the aperture to minimize fading of the fluorescence by

laser radiation.

Figure 9 shows an scanning electron micrograph taken of a mono-

layer of the spheres showing a large variation in sphere size. Manu-

facturer's literature states the mean diameter as 10 pm + 20 %.

Generally spheres which deviated from the mean diameter by more than

approximately 15% were not selected for measurement. The mean

fluorescence intensities were background corrected (the background was

taken as the mean signal from 16 blank spheres). The background

consisted of contributions from the solvent-coated blank silica spheres

and the quartz slide.

Synthetic particulates were prepared in a similar fashion using

the 10 lm silica spheres and spectroscopic graphite powder (Grade SP-2,

Union Carbide Corporation, Carbon Products Division, New York, NY) as

solid supports. Graphite powder resembles soot in that it is a

carbonaceous material. Silica is one of the major inorganic consti-

tuents of fly ash (30). Approximately 10 mg portions of silica or

graphite were coated with 0.1 mL portions of various 100 ppm PAH

standards in hexane (American Burdick and Jackson, Muskegon, MI) to

give concentrations of 1 pg PAH/mg solid support. This is more than

1000 times the concentration of most PAHs found in particulate matter

such as urban dust. (PAHs were obtained from Aldrich Chemical Company,

Milwakee, WI, Eastman Kodak, Rochester, NY, and Analabs, North Haven,

CT). These synthetic particulates were spread on a quartz slide.

Fluorescence was excited by the Nitromite N2 laser operating at 20 Hz.

Individual particles of National Bureau of Standards Reference

Material 1649 Urban Dust were examined using a Molectron UV 24 nitrogen

laser as the excitation source. A small portion of the urban dust

was spread on a quartz slide. The entrance slit of the monochromator

Figure 9. Scanning electron micrograph of a monolayer of
Spherisorb 10 Pm spheres.

was replaced by a 1 mm circular aperture to spatially reject fluores-

cence from particles other than the one of interest.

Results and Discussion

Sensitivity of Laser Microfluorometer

For the five R6G standard groups measured (ranging in

concentration from 10 to 107 molecules R6G/sphere), the percent

relative standard deviations varied from 18 to 48%. The fluorescence

intensity was proportional to concentration for standards used, with a

linear correlation coefficient (r) of 0.9994. The LOD (30) was 8 x 103

molecules of R6G/sphere.

There are several reasons why the LOD should be lower than what

was found in this study. As previously mentioned, a worst case esti-

mate was made of the number of spheres/g, giving a high estimate for

the number of R6G molecules/sphere. Also, it was assumed that during

evaporation of the solvent from the test tubes containing the silica

spheres, all of the R6G was adsorbed to the spheres. Loss of R6G by

adsorption to the walls of the glassware would mean there is actually

less rhodamine per sphere than is believed, and hence a lower LOD.

Despite the difficulties in making a more accurate calculation of the

LOD, it has been shown through conservative estimates that the detect-

ion power of this technique is quite good.

The sensitivity of this technique for the detection of PAHs on

small particles can be estimated by a relatively simple comparison to

R6G. If one calculates the ratio of fluorescence intensities from

equation 4 for a typical PAH and for R6G, then:


IR6G DR6Go1 (C)R6G

Assuming equal light intensities at 514 nm for R6G excitation and at

355 nm for PAH excitation by an Ar ion laser, and assuming equal path

lengths and concentrations, equation 5 reduces to



where is the molar absorptivity of a PAH excited at 355 nm and
ER6G is the molar absorptivity of R6G at 514 nm. From literature

values (31-36), R6G in ethanol ~ 0.9 and ER6G ~ 5xl04 For fluorene

as an example, ~ .68 and E ~ 4x103 at 355 nm so that IPAH/IR6G =

6x10-2. For anthracene, ~ .30, E ~ 8xl03 and IPAH/IR6G = 5x102.

Thus this method is about 100 times less sensitive for the detection of

PAHs. Concentrations in excess of 800,000 molecules of PAH per 10 Pm

particle would be necessary to be detectable by this method.

Fluorescence of PAHs on Solid Supports

Because the majority of PAHs are found adsorbed to carbonaceous

particles and because silica is one of the major constituents of fly

ash, the fluorescence characteristics of PAHs adsorbed to graphite

powder and silica spheres was investigated. Figure 10 shows the

results for B(a)P.

Uncoated graphite displays a 10 times lower background spectrum

than plain silica. However, there is no evidence of B(a)P fluorescence

on graphite. Significant fluorescence of B(a)P adsorbed to silica was

observed and the spectrum compares favorably with fluorescence of 100

ppm B(a)P in hexane. The fluorescence spectrum of a single crystal of

B(a)P is also shown and shows a large red shift as well as significant

broadening. At high concentrations and in crystalline hydrocarbons

Figure 10. Fluorescence spectra of B(a)P adsorbed to silica and
graphite substrates.


Solid B(a)P /

100 ppm B(a)P in hexne

Silica with B(a)P



with B(a)P

Graphite 0
I I I gl l




Wavelength ( nm)







(x 10)





where high molecular overlap can occur, the formation of dimers or

excimers is known to occur and causes the appearance of broad,

structureless fluorescence at longer wavelengths (34).

For all PAHs used in this study pyrenee, fluoranthene, B(a)P, and

perylene), fluorescence was not observed on graphite powder but was

evident on silica spheres. The absorption characteristics of graphite

in the ultraviolet and visible regions have been measured by Ergun and

McCartney (37). Figure 11 shows the results they obtained plotted as

extinction coefficient (k) versus wave energy, where k was defined as

SIln(I /I)
k = (7)


and b = cell path length and X = wavelength. Equation 7 can be

arranged to give:

log I /I = A = 4kb (8)

Selecting the maximum extinction coefficient as k = 2.0 and substitut-

ing A= 259 nm and A = 1.0 in equation 8, the path length which would

result in absorbance of 90% of the incident light is only 0.024 Pm.

Even at a minimum in figure 11, at k = 1.0 and X= 450 nm, 90% of the

light would be absorbed at t = 0.082 Pm. From this, it can be con-

cluded that graphite is an almost perfect absorber of ultraviolet and

visible radiation and can not only absorb essentially all of the

ultraviolet radiation used for excitation of PAHs, but can also absorb

light emitted as fluorescence.

Figure 11. Extinction coefficient versus wave energy for
graphite (40).






S1.4 -



1.0 -

09 I

2 3 4 5 6 7

6000 4000 3000 2000

In addition, oxygen is known to form surface complexes by chemi-

sorption with almost all types of carbons (38). In chemisorption, the

oxygen molecules are linked to the surface by valency bonds and

constitute a monolayer thickness. Oxygen is a very efficient quencher

of fluorescence of organic molecules and it is likely that oxygen plays

a role in the quenching of fluorescence of PAHs adsorbed to graphite.

The exact mechanism of fluorescence quenching of PAHs on carbona-

ceous materials cannot be determined from the results of this work.

However, it can be concluded that since the bulk of PAHs in particulate

matter are adsorbed to carbonaceous particles which prevent or quench

fluorescence, laser microfluorescence cannot be a useful tool for

screening airborne particulates.

Fluorescence of Particulate Matter

The fluorescence spectra of several individual particles found in

urban dust are shown in figure 12. These particles are not carbon-

aceous in nature, but in general are odd-shaped and translucent.

Particles which appeared to be fluorescent under ultraviolet radiation

were randomly chosen.

The fluorescence spectra were broad and nonspecific and a wide

variation was found in emission maximum among only these few particles

of only one type of particulate sample. For these reasons, it was

again concluded that no useful fingerprint could be obtained using

laser microfluorescence of particulate matter.

Figure 12. Fluorescence spectra of individual particles of Standard
Reference Material 1949 Urban Dust/Organics.

0 0 0
V gt0 to

Wavelength (nm)

0 0 0

o 0
0 .0



Since its introduction in the 1960's, the laser microprobe has

been widely used in the study of solid surfaces. The laser microprobe

consists of a lens or microscope which focuses a pulsed laser beam to a

microarea of a sample. The high irradiances produced by focused laser

beams can be used to advantage to vaporize and ionize a wide variety of

samples. The interaction of a focused laser beam with the sample

surface can range from thermal desorption to plasma formation.

A large amount of attention has been focused on the use of high

irradiances to produce a plasma which can be observed by emission,

absorption and fluorescence techniques, or to produce a source of ions

for mass spectrometric techniques, as in the Laser Microprobe Mass

Analyzer (LAMMA). The interaction of high power radiation with solid

surfaces is complex and the mechanisms of vaporization and ionization

are not completely understood. Several reviews discuss the mechanisms

of ion formation and vaporization (39-41). The temperature rise of a

surface is governed by such factors as the absorption characteristics

and thermal properties of the sample, as well as the laser pulse

characteristics. Laser radiation absorbed at the surface of a material

will produce heat in a thin layer under the surface and heat conduction

into the interior of the material will occur. The velocity of heat

transport due to heat conduction decreases with time so that the

temperature at the surface rises; evaporation at the surface occurs

when the energy deposited exceeds the latent heat of vaporization L ,

of the sample. The minimum absorbed irradiance E in, above which
evaporation will occur can be give by (40)

min = Lv a1/2t -1/2 (9)

where p is the mass density of the target, a is the thermal diffusi-

vity, and t1 is the laser pulse duration. For substances like metals
2 3
with a ~ 0.1 cm /s, p ~ 10 g/cm L ~ 400 J/g (assuming a laser pulse

duration of 10-9 s), the minimum irradiance is approximately 3 x 108
2 -3 2
W/cm (40). For organic materials with a ~ 10 cm /s (42), L ~ 400
J/g, and p 1.5 g/cm3 (43), the minimum irradiance would be

approximately 6 x 105 W/cm2.

Considerable interest is currently being focused on the laser

microprobe for thermal desorption of intact molecular species. A wide

variety of compounds have been shown to be amenable to molecular

analysis by laser desorption mass spectrometry, including high

molecular weight, nonvolatile or thermally labile compounds such as

peptides, oligosaccharides, and nucleotides (44-46). Intact molecular

ions have also been observed for PAHs (47). In fact, Ramaley reported

on the use of a focused laser to thermally desorb PAHs from thin layer

chromatography plates. The thin layer chromatography scanner was

interfaced to a quadrapole mass spectrometer and an auxiliary reagent

gas was used to sweep the desorbed PAH into a chemical ionization

source (48). Vanderborgh and Jones reported on the use of the LAMMA

technique to characterize coal and shale samples (49). A recent report

by Burgess et al. discusses the calculations of surface temperature

rise and desorption temperature in laser-induced thermal desorption


The laser microprobe is finding increased application to single

particle analysis, primarily for quick, routine qualitative and semi-

quantitative analysis of individual particles of micrometer size for

minor and in some cases, trace level inorganic constituents (51).

Waele et al. have recently used the LAMMA (Leybold-Heraeus) for the

analysis of organic pollutants, including B(a)P, adsorbed to asbestos

fibers (51).

The purpose of this work was to investigate the feasibility of a

laser microprobe system for the thermal desorption and excitation of

PAHs adsorbed to particulate matter. Detection of the excited vapor

with a Silicon Intensified Target (SIT) camera was attempted in order

to produce a full wavelength spectrum of either single particles or of

the integrated signal from many particles.


A block diagram of the laser microprobe is shown in figure 13.

The laser microprobe consisted of a Nikkon fluorescence microscope

equipped with an epi-illuminator. The epi-illuminator allowed the

introduction of a pulsed N2 laser through the rear of the microscope.

The laser was deflected downward through a microscope objective by a

337 nm dichroic mirror which was 97% relective at 45. A 40X objective

was used to focus the laser on the sample. The laser spot size was

approximately 90 x 40 pm2 and could be reduced by the use of an

aperture in the epi-illuminator. The laser emitted ~ 300 ps pulses

with an energy of ~ 30 pJ. A portion (~ 10%) of this light reached the

sample; losses were caused by reflection at and absorption by the

internal optical components. The energy density at the sample was thus

~ 108 W/cm 2, and could be reduced by neutral density filters.

Figure 13. Block diagram of the laser microprobe.


The fluorescence and reflected light emerging from the sample was

collected by the 40X objective and directed upward through a phototube

attachment to a monochromator. A WG 360 long wavelength passing filter

(Corion Corp.) was used to reject scattered laser radiation. The

fluorescence was detected by an SIT camera.

The SIT camera used in this work is part of a Hamamatsu streak

camera system that includes a streak unit termed a temporal disperser,

a microprocessor termed a temporal analyzer, and the SIT camera and

camera control unit. In its normal operation with the streak camera

system, the temporal analyzer converts the signal stored on the SIT to

an intensity versus time display, and allows the user to control the

data acquisition conditions. The 256 horizontal scanning lines of the

vidicon are arranged in a direction which is perpendicular to the time

direction. During the scanning of each frame, the target current of

a horizontal line is analog integrated, A/D converted and stored in

digital memory. At the end of each frame scan, 256 8-bit intensities

are in memory. Variables of importance which can be controlled from

the temporal analyzer include signal integration, persistence

integration, dark current correction, and data output mode. The number

of integration can be selected from 1 to 256. The number of persist-

ance integration can be set from 1 to 16, and instructs the read beam

of the camera to read the current frame the selected number of times.

Vidicons suffer from "lag", or incomplete readout of signal, and in

some cases it may improve the signal-to-noise ratio by reading the same

frame more than one time. The signal acquired by the temporal analyzer

can be sent to a computer or to a recorder.

For this work, the SIT was removed from the streak unit. The

light was dispersed across the SIT by the monochromator so that the

wavelength axis corresponded to what was previously the time axis. The

temporal analyzer then produced an intensity versus wavelength display.

The temporal analyzer was triggered by a photodiode.

The signal from the camera was either processed by the temporal

analyzer and displayed on the video monitor, or it could be sent

directly from the camera control unit to the video monitor. From the

camera control unit, the read beam could be turned off (frame blanking)

for controlled periods of time to allow many light events to accumulate

on the SIT. Since the SIT acts like a photographic emulsion, essen-

tially recording all incoming energy and retaining it until the read

beam impinges upon it, the signal is allowed to accumulate. This

amounts to increasing the exposure time. The camera control unit

allowed either the use of the temporal analyzer, or the use of frame

blanking directly from the camera control unit, but not both.

Particulate matter (SRM 1649) was placed on a quartz slide and

focused under the laser spot. Problems were encountered with

scattering of the particulate dust by the laser spot, and several means

were attempted to immobilize the particles. The most effective was the

use of a non-fluorescing mounting medium (Aquamount, American Scienti-

fic Products). The particles were dispersed in a small drop of the

fluid medium and the mixture was allowed to harden.

Results and Discussion

The action of the laser on the particulate matter could be

observed through the eyepiece mount of the Nikkon microscope. Faint,

visible flashes of light could be observed with each laser pulse as

long as there was particulate matter directly under the beam. After

several laser pulses, particles were usually loosened from the mounting

medium and scattered out of the field of view.

Because of the low light level, the temporal analyzer was used in

the integration mode. Up to 256 integration with 16 persistence

integration were acquired in an effort to produce an observable

signal. No signals above the noise level were obtained. In this case,

the light level reaching the detector within each frame scan was

insufficient to be recorded.

In an attempt to improve the signal-to-noise ratio, the temporal

analyzer was abandoned for the frame blanking option of the camera

control unit. However, there are several disadvantages to the use of

the camera control unit without the temporal analyzer. The temporal

analyzer stores the integrated information (intensity) from each

scanning line into memory. The camera control unit can only give

intensity information along one selected vertical line of the 256

horizontal scanning lines. The advantage of frame blanking is only

available at the cost of losing a large portion of the available sig-

nal. Also, after the read beam is instructed to read the accumulated

signal on the SIT, the intensity versus wavelength information along

the vertical line is displayed instantly on the video monitor, but is

erased in 33 ms by the next sweep of the read beam. There was no

method available to transfer the intensity versus wavelength

information to a permanent storage device in that time period. At the

very least, this method only allowed the user to evaluate if there was

any observable signal produced by frame blanking, not to record it.

Turning off the read beam for periods of more than a few seconds

resulted in overload of the detector, partly due to dark current

integration. However, by turning off the read beam for 1 s intervals

(triggered manually by a front panel switch), this user was able to

observe a broad emission signal in the region from ~ 420-500 nm.

Because the intensity versus wavelength display was only available for

observation instantaneously, no real characterization of this emission

could be performed. The limitations of the available instrumentation

in handling the low light levels resulted in the termination of further

work. However, several conclusions and suggestions for future

improvements can be made.

The most obvious improvement to this system would be a more

flexible multichannel detection system, i.e. one that would allow the

acquisition of long exposure time signals with a means to store that

information. Various commercial multichannel image detectors are

available today which are considerably more flexible and appropriate to

the nature of this work (the OMA III photodiode array system described

in Chapter V is one such system, but was not available at the time of

this work).

In addition to complicating factors with the detection system, the

nature of the laser microprobe is a "photon-starved" one; the light

levels available to the detector are low for several reasons. One

factor is the limited availability of the sample. Particulate matter

is not compact in nature, and problems were encountered in maintaining

the particulates under the focused laser spot. It is suggested that

single particle resolution be discarded in favor of obtaining an

integrated representation of the fluorescence emission of particulate

matter, either by collecting enough particulate matter to pack into a

sample cup or press pellets which can be observed for long periods of

time. Optionally, the laser can be focused directly on filters used to

collect particulate matter, and the fluorescence of thermally desorbed

PAHs can be integrated as the filter is moved under the laser. Cremers

and Radziemski have described such a method for the direct detection of

beryllium on filter paper, using the laser spark (52). A cylindrical

lens was used to focus a pulsed Nd:YAG laser onto a rotating stage

which held the filter paper. The emission was detected as the filter


The inefficiency of the present optical setup must also be

considered. Losses of light can occur throughout the internal optical

components of the microscope system, through focusing and refocusing

onto the monochromator and the SIT. The microscope could be eliminated

as the means of signal collection at the expense of single particle

resolution. The laser can be focused on the sample without the use of

the microscope. Also, because the photocathode of the SIT was recessed

from the front of the camera, it was necessary to use a lens to focus

the image from the exit focal plane of the monochromator onto the SIT.

A flat field spectrograph with an external focal plane would allow the

image to be focused directly from the spectrograph to the SIT.

Photon starvation can also be caused by the inefficiency of the

source of emission. In laser microprobe atomic emission, a secondary

excitation source is often necessary to increase the emission intensity

of the atoms and ions in the plasma. In laser-induced thermal

desorption, the desorption of neutral species has been shown to

continue for a considerable amount of time (500 ps to 1 ms) after the

laser pulse (47). The laser pulsed duration in this work is very


short, ~ 300 ps. Thus, there is a substantial loss of available signal

between laser pulses, which could be accessed by cross excitation with

an appropriate second laser.



Electrothermal atomizers have found widespread use in atomic

absorption and atomic fluorescence spectroscopies for the vaporization

of elements in the temperature range from 2000 to 2700 0C. Heated

graphite furnace atomizers have been used almost exclusively for the

production of atomic vapors and little attention has been paid to the

performance of these atomizers in the temperature regions between 25

and 1000 C.

Recently, however, several groups have reported on the use of

electrothermal atomizers for the study of the ultraviolet absorption of

molecular vapors (53-58). Thompson and Wagstaff used ultraviolet

absorption spectrometry to monitor organic pollutants in water (53).

They found that by monitoring the absorption at one or more wavelengths

as a small amount of sample was vaporized from 35 to 900 0C, they could

rapidly characterize many types of substances based on peak shapes,

intensities, and appearance times. The higher temperatures obtained

with the graphite furnace allowed the rapid characterization of samples

often too difficult to analyze by gas chromatographic methods, such as

heavy oils, tar residues and resinous materials. This technique has

also been used by Tittarelli et al. for the identification of crude

oils spilled in seawater (55), for the identification of organic and

inorganic pigments (56), and for recording the vapor spectra of several

aromatic hydrocarbons (57). The latter work involved using the

graphite furnace to supply a steady vapor supply while scanning over

the absorption spectrum. Several sample aliquots were generally neces-

sary to acquire the entire spectrum. The use of photodiode arrays for

fast spectral acquisition was suggested for the programmed heating

studies of molecular compounds. Subsequently, the graphite furnace was

combined with a diode array detector for the ultraviolet investigation

of crude oils, pigments and polymers (58). Thermal cycles from 150 to

as high as 2750 C were run while monitoring the ultraviolet spectra.

Both atomic and molecular emission and absorption can take place at

high temperatures exceeding 1000 C and emission was evidenced in the

occurrence of negative absorbance peaks. The collected spectra were

plotted as a three dimensional representation of the vaporization

cycles to produce spectral "fingerprints" characteristic of the vapor-

ized oils or pigments. The simultaneous measurement of more than one

parameter in this way provides enhanced selectivity over conventional

spectroscopic techniques, especially where spectra are too broad-banded

to be useful for analytical measurement of similar compounds.

While the graphite furnace has been used as a vaporization source

for laser excited atomic fluorescence, it has not been applied to the

laser-induced fluorescence of molecular vapors. In this work, laser-

induced fluorescence of PAH vapors produced by electrothermal vapori-

zation was investigated. The capabilities and behavior of

laser-induced fluorescence in a graphite furnace were evaluated in

terms of sensitivity and potential use as a fingerprinting technique

for particulate matter and other environmental samples.

Limits of detection for several PAHs commonly found in the envir-

onment were determined using a monochromator/PM tube combination and a

N2 laser as the excitation source. These LODs were compared to those

obtained using a Princeton Applied Research Model 1205D Optical Multi-

channel Analyzer (OMA) and an excimer laser.

The vapor phase spectra of several PAHs excited by the N21aser

were obtained using the multichannel analyzer. Generally, little

information is available on the gas phase spectra of PAHs. Laser-

induced fluorescence is being developed as a technique for the real

time in-situ measurement of PAHs in combustion sources, and the

emission spectra of individual PAH must be known for such work (59-61).

Electrothermal vaporization combined with laser-induced fluorescence

and multichannel detection was shown to be a convenient method for the

measurement of vapor phase PAH fluorescence spectra.

Due to limitations of the speed of data transfer of the 1205D OMA

to a peripheral storage device and limited memory space inside the OMA

itself, only one spectrum could be acquired during the entire vapori-

zation cycle (generally 20 s). Because the vapor phase fluorescence of

PAHs is generally broad and featureless, it was desired to obtain

several spectra during the vaporization so that a third dimension of.

temperature could be added to the intensity versus wavelength

information to give a three dimensional map of the eluting vapors. A

manual means of doing so was devised and the separation of a PAH

mixture was demonstrated.

An OMA III system capable of extremely rapid spectral data aquisi-

tion was donated by Princeton Applied Research Corporation. This

system consisted of a model 1421 1024 element silicon diode array

detector with a thermoelectric cooler and a model 1460 system

processor. Three dimensional spectra were obtained for several PAHs

demonstrating the improved resolving power achieved by combining the

fluorescence spectral information with the evolution behavior.

Finally, laser-induced fluorescence with electrothermal vaporiza-

tion and diode array detection was applied to the rapid screening of

various environmental samples, including particulate matter, cigarette

smoke condensate, and crude oils. A new sampling technique was

developed which allows the direct collection of samples within the

graphite rods of the furnace system. The advantages and limitations of

this technique as a routine screening technique will be discussed.


A block diagram of the various detection systems used in this work

is shown in figure 14. Fluorescence could be excited from opposite

directions by either a Lumonics excimer laser or a Molectron N2 laser

passing directly above the graphite furnace. Either laser was brought

to a focus a few millimeters beyond the vapor produced by the furnace.

Lenses on either side of the furnace were used to form a 1:1 image of

the fluorescence onto the entrance slit of a 0.35 m Heath monochromator

and a 0.2 m ISA flat field spectrograph. Apertures were used to

reduce stray light and background emission from the furnace. For the

same reason, the entire optical setup was surrounded with a tent of

black felt. Limits of detection for several PAHs were determined using

an R1414 photomultiplier tube with the N2 laser as the excitation

source. A photodiode was used to trigger the boxcar. Fluorescence was

detected at the wavelength of maximum emission for the particular PAH.

The ISA flat field spectrograph had a dispersion of 24 nm/mm and a

spectral range of 200-800 nm. Once mounted to an OMA detector (either

the model 1205D or model 1421), the grating could be accessed and

I 0

0 0
4-4 co

0 0

4- CO

o w
u U

W O 4-




I ~O
I *

rotated back and forth to shift the spectral region of interest onto

the detector.

Graphite Furnace System

The graphite furnace is illustrated in figure 15. Two water-cooled

brass blocks were mounted on opposite sides of a round phenolic block

and were connected beneath the phenolic block to an SCR power supply.

A groove in the center of these brass electrodes supported the graphite

rod which contained the sample. Electrodes to fit both 1/4" diameter

and 3/8" diameter graphite rods were constructed. The 1/4" rods were

solid throughout, except that a small 2 mm circular depression approx-

imately 1/8" deep was drilled out to hold the sample. For this type

rod, argon carrier gas was connected underneath the phenolic block and

fed through a burner head to flow upward past the graphite rod. Room

air currents were found to grossly affect the reproducibility of signal

measurements. A glass cover with quartz windows was used to minimize

this problem.

The 3/8" diameter rods were hollow, with a small hole drilled

through the middle of the top for sample introduction. The carrier gas

was made to flow in through both ends of the graphite tube via brass

end caps (see figure 16) so that a vaporized sample would flow out

through the hole in the top of the rod. The vapor was carried into the

path of the laser beam directly above the rod. Argon flow rate was

optimized at 0.3 L/min.

The heating rate and duration of the heating cycles for the

graphite furnace were controlled in a manner similar to that described

by Goforth (62). A variable timing circuit controlled the duration of

the heating cycle, and the output of the power supply could be

Figure 15. The graphite furnace.

Glass Cover

Figure 16. Design of 3/8" graphite tubes, illustrating carrier gas
flow and position in the brass electrodes of the furnace.

Graphite sleeve

end caps

Ar in

Ar in

Brass electrodes


ite tube


controlled by a front panel voltage setting. The timing circuit, which

is shown in figure 17, consisted of 555 timer which controlled a relay.

A toggle switch was used to activate the 555 timer. The timer then

activated the relay for a length of time determined by the capacitor CT

(10 )iF) and a variable resistor RT (0-5Mn). With the relay activated,

the normally closed switch was opened so that a 100 Q resistance

existed across the controlling wires of the power supply, corresponding

to full power. Generally, the furnace was activated for a 25 s time

period at a voltage sufficient to produce an approximately linear

temperature rise from 20 to 450 C. Figure 18 shows the reproduci-

bility of the temperature rise versus time for two different graphite

tubes (3/8" diameter). The temperature was measured with an Iron-

Constantan (type J) thermocouple in contact with the inside of the

graphite tube. The solid lines indicate successive heating cycles with

the same tube. The temperature rise is reproducible to within about 10

C. The dashed lines indicate successive heating cycles with a

different graphite tube. The voltage setting on the power supply was

adjusted so that the graphite tubes would heat up to approximately the

same degree of visual brightness. The deviation in temperature rise

between graphite rods adjusted in this manner was found to be

approximately 15%.

OMA 1205D

The model 1205D OMA consisted of a 500 channel silicon intensified

target (SIT) detector and an OMA console. The detector contained an

electron image intensifier section ahead of the target section. The

detector target was 5 mm x 12.5 mm. The top 2.5 mm x 12.5 mm portion

of the detector was used to compensate for vidicon dark current and for





4 .)

4 H

4 J4
. .-





Figure 18. Temperature rise versus time for two different graphite
rods, measured with an Iron-Constantan (type J) thermo-




- Rod I
--- Rod 2

0 6 12

18 24

Time (s)




stray light on the detector. Any light falling on this portion was

automatically subtracted from the signal on the corresponding area of

the bottom 2.5 mm x 12.5 mm portion. Therefore alignment of the signal

of interest only on the bottom portion of the detector target was

critical. The 12.5 mm area across the detector target was divided into

500 vertical channels. Vertical alignment of the slit image on these

channels could be accomplished by rotating the detector head. The real

time display was used during optical alignment procedures. The flat

field spectrograph dispersion of 24 nm/mm combined with a detector

target length of 12.5 mm resulted in an approximately 300 nm spectral


The OMA console digitized the signal from the detector. Once in

digital form, the data could be stored in one of two 500 word, 21 bit

memories. A single exposure of 32.8 ms could be stored in memory, or

the OMA could be programmed to signal average many exposures and then

store the signal averaged data. Generally, memory B was used to

collect a background spectrum and was subtracted channel by channel

from memory A (the signal plus background) by the A-B memory mode. The

contents of memory A, memory B, or memory A-B could be viewed on a

video monitor or output to a chart recorder. The excimer laser was

triggered by a pulse from the OMA that was synchronous with each 32.8

ms scan, corresponding approximately to a 30 Hz repetition rate.


The model 1421 silicon diode array detector was mounted to the

spectrograph in the same manner as the model 1205D. Each diode element

was 2.5 mm x 25 Pm, to give a total 25.6 mm length. Thus the spectral

region dispersed across the diode array was approximately 600 nm wide.

The entire area of the detector target was used to acquire signal; the

diode array was not divided into "dark" and "signal" portions as was

the SIT. The model 1421 OMA contained a Peltier effect thermoelectric

cooler. With the cooler connected to a source of chilled H20, the

array temperature could be brought down to -20 0C. Generally, for each

7 decrease in array temperature, the dark current was reduced by a

factor of two. This was an important consideration since some of the

work done here required long periods of integration and the dark

current could represent a large percentage of the saturation current.

Detector setup and data processing was accomplished with the model

1460 system processor and Version 19 OMA III operating software.

Because the microprocessor contained over 450 Kbytes of random access

memory, many more "memories" could be acquired with this system than the

two available with the earlier model 1205D. Each memory contained a

number of scans, which could be programmed by the user, and each scan

was of a length also determined by the user. For this work, generally

each scan was 33.333 ms long. Ten scans were usually acquired and

stored in one memory, and 40-50 memories were acquired. Data was thus

acquired over a 13 to 16 s interval which was initiated with the start

of the graphite furnace. The data could then be stored on a floppy

disc and/or could be plotted on the system screen as a three dimen-

sional representation of intensity versus wavelength versus memory

number. In this case, memory number corresponds to time, and hence to

temperature of the graphite furnace.

The variety of data acquisition and processing options which were

available are too numerous to mention here. Generally, spectra were

collected and displayed so that a restored background was automati-

cally subtracted, and a restored calibration was used to present the X

axis as a wavelength display. After the data had been satisfactorily

selected and scaled, it was sent to a plotter.

When the OMA III was used, no convenient method was available to

synchronize the triggering of the laser with each scan event of the

OMA. The laser was operated in a free running mode at 30 Hz and the

OMA was synchronized to line frequency.

Direct Sampling System

A new sampling technique which allowed the direct collection of

exhaust samples in the graphite rods was developed. Figure 19 illus-

trates the sampling system. A brass H20-cooled jacket was constructed

to hold four 3/8" diameter graphite tubes within its length. Connec-

tion was made at one end to a source of exhaust, such as a lawn mower.

The brass was cooled by the continuous flow of H20 through it and

through copper coils resting in a dewar of ice and rock salt. The

exhaust source was turned on for a short period of time, generally 5 to

10 min. The graphite tubes were cooled by the action of the brass

jacket and exhaust condensed inside the tubes. The tubes were removed

from the cooler and each rod was analyzed in the graphite furnace.

Alternately, a pump could be attached to the sampling system to draw

air through the graphite tubes. Cigarette smoke condensate was

collected by inserting the cigarettes in the opposite end of the brass

jacket, allowing the action of the pump to "smoke" the cigarette.

The equipment used in this work and manufacturers are

given in Table V.

Figure 19. The sampling system for direct collection of exhaust
samples in 3/8" graphite rods.



I 2

8 \c\\

/ / >
5~/^ / /$
^ ~ ~~ / // [-
*^ ~// /"--
CD rr ,
.Q N ,,,


/o,', ft

Table V. List of Equipment and Manufacturers

Equipment Model Manufacturer

UV 24

Molectron Corp.
Palo Alto, CA

Furnace power

0.35 m mono-

PM tube


HV supply for
PM tube



Video monitor

Flat field

Excimer laser

Photodiode array

system processor


SCR 20-250



162 with
164 plug in






TE 861S




Electronic Meausurements
Neptune, NJ

Acton, MA

Middlesex, NJ

Princeton Applied Research
Princeton, NJ

Princeton Applied Research
Princeton, NJ

Houston Instruments
Austin, TX

EG & G
Princeton Applied Research
Princeton, NJ

Tektronix Inc.
Beaverton, Oregon

Instruments SA, Inc.
Metuchen, NJ

Lumonics Research Limited
Ontario, Canada

EG & G
Princeton Applied Research
Princeton, NJ

EG & G
Princeton Applied Research
Princeton, NJ

Martinez, CA

N2 laser

Results and Discussion

Comparison of Limits of Detection

Standard PAH solutions in hexane ranging in concentration from

100 ppm to 0.1 ppm were used for determining LODs in the graphite

furnace. Five iL portions of the standards were used with the 1/4"

diameter rods and 10 pL portions were used with the 3/8" diameter

tubes. A drying cycle was used in all cases to evaporate the solvent.

For the LODs determined using the PM tube and N2 laser, the voltage on

the furnace power supply and the duration of the furnace cycle were set

to vaporize the sample within 6 s. With the 3/8" tubes, the voltage

was decreased on the furnace and the cycle length increased to ~15

s to allow the carrier gas to completely remove the vapor from inside

the graphite tube. Table VI lists the LODs obtained for several PAHs

for both the PM tube and the 1205D OMA.

The LODs obtained with the PM tube are only slightly better than

those obtained with the SIT, even though higher sensitivity is expected

with the PM tube. This is attributed to the increase in laser power

and in some cases more efficient excitation at 308 nm. Percent

relative standard deviations obtained with the PM tube varied from 5 to

34%. Those obtained with the SIT varied from 2 to 47%. The large

relative standard deviations reflect irreproducibility in the sample

vaporization by the graphite furnace. Fluctuations in carrier gas

flow, aging of the graphite rods, sample pipetting and room air

currents all affected the way the sample was vaporized and contributed

to the relative standard deviation.

Detection limits with the SIT are all at the nanogram level. The

limiting noise is equivalent to the noise on the baseline in the A-B

Table VI. Limits of Detection of PAHs by Laser-Induced
Fluorescence with Electrothermal Vaporization

PAH PM tube* 1205D SIT-

Xem (nm) LOD (ng) LOD (ng)

Phenanthrene 365 18 -

Fluoranthene 460 1.1 2.1

Pyrene 383 1.1 1.8

Benz(a)anthracene 385 1.2 3.2

Chrysene 6.1

Benzo(b)fluoranthene 450 1.4 7.9

Benzo(a)pyrene 410 0.3 -

Perylene 438 0.9 3.3

Benzo(ghi)perylene 416 0.4 1.6

Anthanthrene 3.7

* x = 337 nm,
pulse energy = 8 mJ/pulse
2 nm bandpass

+ = 308 nm
pufse energy = 20 mJ/pulse
12 nm bandpass

mode, when A and B are both background accumulated for the same period

of time. According to previous work done in this lab with this SIT,

this noise is predominantly shot noise and improvements in LODs could

be accomplished by increasing the light throughput (63). Possible

methods to accomplish this include increasing the residence time of the

vaporized sample in the laser beam and turning the monochromator so

the slit is horizontal and parallel to the direction of the exciting

beam. Although improvements in LODS are possible, the sensitivity of

this technique is already good.

Vapor Phase Fluorescence of PAH

The emission spectra of PAH in the vapor phase are broadened over

that of the solution phase. In order to more fully characterize the

emission spectra of PAH vapors produced by the graphite furnace, and to

determine to what extent thermal broadening occurs, the SIT was used to

collect the vapor phase spectra of PAHs evolved from the graphite

furnace. A small amount of the PAH of interest was placed in the

graphite tube and fluorescence was integrated over a period of 15 s

while the furnace was on. The spectra collected in this manner are

shown in figure 20. Although the spectra are broadened over what is

typical of solution phase, there is some structure apparent in a few

cases (B(a)P and benz(ghi)perylene), and there is still an obvious

increase in maximum emission wavelength with an increase in boiling

point, or molecular weight. Spectra are uncorrected for detector


The broadening of the fluorescence spectra in the vapor phase in

comparison to solution phase is illustrated in figure 21, for chrysene

and benz(ghi)perylene. In addition, the gas phase spectra are shown at

Figure 20. Vapor phase fluorescence spectra of several PAHs collected
with the 1205D OMA. Bandpass = 12 nm. X = 337 nm.

Benzo(ghi)perylene (542)

Perylene (497)

Benzo(a)pyrene (496)

Chrysene (441)

Benzo(a)anthracene (438)

Pyrenp (394)

*K. Fluoranthene (243)











0 IT OD N w
"1 0 to 0
10 ^- << 10 10



4 4

0 0
u 0



0 0

0 0


zoo II
II am
II ,-<



r e

00 0
-1 \

N I0

04 0 0
0 0 :3

OO rn

A1ISN31NI 30N30S3IOfnli 3AAIIV13I

AIISN31LNI 30N30S3Ona-U 3AllV13