A self-cleaning, low-temperature molecular luminescence spectrometer as a detector for gas chromatography

MISSING IMAGE

Material Information

Title:
A self-cleaning, low-temperature molecular luminescence spectrometer as a detector for gas chromatography
Physical Description:
xiii, 136 leaves : ill. ; 28 cm.
Language:
English
Creator:
Clos, John F., 1964-
Publication Date:

Subjects

Genre:
bibliography   ( marcgt )
theses   ( marcgt )
non-fiction   ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 1991.
Bibliography:
Includes bibliographical references (leaves 132-135).
Statement of Responsibility:
by John F. Clos.
General Note:
Typescript.
General Note:
Vita.

Record Information

Source Institution:
University of Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 001693370
notis - AJA5449
oclc - 25222094
System ID:
AA00003717:00001

Full Text












A SELF-CLEANING, LOW-TEMPERATURE MOLECULAR
LUMINESCENCE SPECTROMETER AS A DETECTOR FOR
GAS CHROMATOGRAPHY


















By

JOHN F. CLOS


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


19913















ACKNOWLEDGEMENTS


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

allowing me to be a part of this group. It was a difficult

decision to stay here and work for Jim instead of

transferring to Cincinnati with Dr. Dorsey; however, I feel

very lucky to have had the opportunity to be a part of the

Winefordner research group. I will always be grateful to

Jim for all I have learned while being a member of this

group.

I would also like to thank all the friends I have made

during my graduate school years. Their help and friendship

have been invaluable in my graduate studies. I would

especially like to thank Nancy Mullins for all of her help

while working on this project.

Finally, I would like to thank my parents, Frank and

Betty. I am very lucky to have had parents who guided me in

the right direction as they did.

















TABLE OF CONTENTS


ACKNOWLEDGEMENTS. . . ii

LIST OF TABLES . . .. v

LIST OF FIGURES . . ... vi

ABSTRACT . . xi

CHAPTER 1
INTRODUCTION . 1
Polycyclic Aromatic Hydrocarbons .. 1
Molecular Effects of PAH . 3
Physical Effects of PAH .. 7
Methods for Determining PAHs . 7
Room-Temperature Techniques . 7
Matrix Isolation Spectrometry ... 11
Shpol'skii Spectroscopy . 14

CHAPTER 2
THEORY . 23
Luminescence Spectroscopy. ... 23
Band Broadening . .. 25
Cooling Effect . . 30
Shpol'skii Effect. . . 36

CHAPTER 3
EXPERIMENTAL .. .41
Instrumentation . .. 41
Reagents . . .. .. 51
Procedure . . .. 51

CHAPTER 4
RESULTS AND DISCUSSION .. 54
Phosphorescence Background . .. 54
Chromatography . ... 64
Chromatograms . . 64
Retention Times, Peak Widths, and Limits of
Detection . ... 97

CHAPTER 5
CONCLUSIONS AND FUTURE WORK .. 129


iii










REFERENCE LIST . . .. 132

BIOGRAPHICAL SKETCH . . .. 136
















LIST OF TABLES


Table 2-1 Data showing the dimensions for several PAHs
and their best Shpol'skii solvent. Taken
from reference [50]. . 38

Table 4-1 List of experimental procedures to delineate
and to minimize the phosphorescence
background problem . ... 57

Table 4-2 Retention times and dead volumes measured
using the FID and the LTMLS system ... 108

Table 4-3 Peak widths measured using the FID and the
LTMLS system . ... 110

Table 4-4 Estimated limits of detection. .. .118
















LIST OF FIGURES


Figure 1-1




Figure 1-2



Figure 1-3



Figure 1-4



Figure 1-5





Figure 1-6




Figure 1-7



Figure 1-8

Figure 2-1


Some examples of polynuclear aromatic
hydrocarbons along with their
carcinogenicity. Redrawn from reference
[2] . . .

Possible mechanism for the formation of
benzo(a)pyrene by pyrolysis. Redrawn from
reference [3]. . .

Reaction mechanism for the metabolism of
benzo(a)pyrene. Redrawn from reference
[10] . . .

Structure of benz(a)anthracene showing the
K, L, and bay regions. Redrawn from
reference [11] . .

Diagram of a supersonic expansion showing
the pressure outside the chamber (Po), the
pressure inside the chamber (P,), the
start of free molecular flow (XF), and the
Mach disc (Xm) . .

Diagrams of a) optical system (top view)
and b) cryostat head (side view) used by
Conrad et al. for GC/matrix isolation.
Taken from reference [24]. .

Diagram of experimental system used by
Reedy et al. for GC/matrix isolation/FTIR
Taken from reference [25]. .

Diagram of self-cleaning belt. .

Energy level diagram showing absorption
(A), fluorescence (F), phosphorescence
(P), and radiationless processes
(dashed arrows). Sp is the ground state,
S1 is the excited singlet state, and T, is
the excited triplet state. .










Figure 2-2



Figure 2-3




Figure 2-4



Figure 2-5





Figure 2-6









Figure 2-7





Figure 2-8




Figure 3-1



Figure 3-2



Figure 3-3


Graphs showing a) Lorentzian and b)
Gaussian profiles. Taken from reference
[43] . . .

Diagram showing a) an energy level diagram
and b) well resolved and c) poorly
resolved spectra . .


Spectra of perylene (700 ng/mL) in
n-hexane at three different temperatures.
Taken from reference [46]. .

Schematic showing inhomogeneous
broadening: a) a transition for a single
molecule, b) transitions for many
molecules, and c) how the band actually
appears. Redrawn from reference [47]. .

Fluorescence spectra of various compounds
in different organic matrices at 4.2 K by
ordinary excitation (dotted lines) and
laser excitation (solid lines): a)
perylene in ethanol, b) protonated form of
9-aminoacridine in ethanol, c)
tetraphenylporphin in polystyrene, and d)
protochlorophyll in ether. Taken from
reference [47] . .

PAHs along with their optimum Shpol'skii
solvent showing how they relate: a)
naphthalene and pentane, b) anthracene and
heptane, and c) naphthacene and nonane.
Redrawn from reference [51]. .

Orientation of coronene and heptane
molecules with respect to the
crystallographic axis. Taken from
reference [54] . .

GC/vacuum chamber interface used in these
experiments. Based on the work by Brown
and Wilkins [54] . .

Vacuum chamber used in these experiments:
a) top and b) side views. Belt and spools
are shown as dashed lines. .

Schematic showing how the belt was
tightened: a) the belt loose and b) the
belt tightened. Belt and spool are shown
as dashed lines. . .


vii










Figure 3-4



Figure 4-1



Figure 4-2





Figure 4-3




Figure 4-4





Figure 4-5


Figure 4-6




Figure 4-7


Figure 4-8





Figure 4-9


Figure 4-10




Figure 4-11


Schematic diagram of the spectroscopic
portion of the system. Dashed lines
indicate light rays. . .

Phosphorescence spectrum of the background
when hexane was injected . .


Spectra of 126 Ag/mL phenanthrene in
hexane, hexane, and an empty tube (blank).
Spectra were taken at 77 K using the SPEX
fluorimeter. Excitation wavelength was
250 nm . . .

Spectra of hexane and pump oil diluted in
hexane. Spectra taken at 77 K using the
SPEX fluorimeter. Excitation wavelength
was 317 nm . .

3-D chromatogram of 78 Ag/mL phenanthrene;
impurity peak (I): 8.39 min; phenanthrene
peak (Ph): 8.59 min; phenanthrene
aggregation: Ph(A); time scale is from
7.80 to 8.93 min . .

3-D chromatogram of 10 Ag/mL phenanthrene;
conditions are the same as in Figure 4-4

Spectrum of manual injection of 131 Ag/mL
phenanthrene (Ph); impurity: I; where
aggregation peaks appear in Figure 4-4:
Ph(A) . . .

Spectrum of 78 jg/mL of phenanthrene at
its chromatographic peak maximum .

3-D chromatogram of 94 gg/mL pyrene;
impurity peak (I): 10.76 min; pyrene peak
(Py): 10.95 min; pyrene aggregation:
Py(A); time scale is from 10.42 to 11.22
min. . . .

Spectrum of 94 Ag/mL of pyrene at its
chromatographic peak maximum .

Spectrum of manual injection of 123 /g/mL
pyrene (Py); impurity: I; where
aggregation peak appears in Figure 4-7:
Py(A) . . .

Spectrum of pyrene impurity at its
chromatographic peak maximum .


viii










Figure 4-12




Figure 4-13


Figure 4-14


Figure 4-15


Figure 4-16


Figure 4-17


Figure 4-18







Figure 4-19







Figure 4-20







Figure 4-21


3-D chromatogram of 87 Ag/mL fluorene;
impurity peak (I): 7.09 min; fluorene
peak (Fl): 7.51 min; time scale from 6.97
to 8.09 min . .

Spectrum of 87 ig/mL of fluorene at its
chromatographic peak maximum .

3-D chromatogram of 82 Ag/mL chrysene;
time scale is from 13.72 to 15.01 min. .

Spectrum of 82 Ag/mL of chrysene at its
chromatographic peak maximum .

3-D chromatogram of 82 jg/mL triphenylene;
time scale is from 15.17 to 17.93 min. .

Spectrum of 82 Ag/mL of triphenylene at
its chromatographic peak maximum .

3-D chromatogram of approximately 100
lg/mL mix of pyrene (Py), 1,2- and
2,3-benzofluorene (B), chrysene (Ch),and
triphenylene (Tri); impurity: I; pyrene
aggregation: Py(A); benzofluorene
aggregation: B(A); time scale is from
10.42 to 17.93 min . .

2-D chromatogram of approximately
100 jg/mL mix of fluorene (Fl),
phenanthrene (Ph), pyrene (Py),
benzofluorene isomers (B), chrysene (Ch),
and triphenylene (Tri). Wavelength
monitored: 421 nm. Time scale is from
6.00 min to 17.67 min. . .

2-D chromatogram of approximately
100 Ag/mL mix of fluorene (Fl),
phenanthrene (Ph), pyrene (Py),
benzofluorene isomers (B), chrysene (Ch),
and triphenylene (Tri). Wavelength
monitored: 495 nm. Time scale is from
6.00 min to 17.67 min. . .


84


86


88


90


92


94







96







99







101


2-D chromatogram of approximately
100 /g/mL mix of fluorene (Fl),
phenanthrene (Ph), pyrene (Py),
benzofluorene isomers (B), chrysene (Ch),
and triphenylene (Tri). Wavelength
monitored: 587 nm. Time scale is from
6.00 min to 17.67 min. .. 103










Figure 4-22







Figure 4-23







Figure 4-24

Figure 4-25

Figure 4-26

Figure 4-27


Figure 4-28


Figure 4-29

Figure 4-30


Figure 4-31


2-D chromatogram of mix of approximately
100 jg/mL mix of fluorene (Fl),
phenanthrene (Ph), pyrene (Py),
benzofluorene isomers (B), chrysene (Ch),
and triphenylene (Tri). Wavelength
monitored: 513 nm. Time scale is from
6.00 min to 17.67 min. . .

2-D chromatogram of mix of approximately
100 Ag/mL mix of fluorene (Fl),
phenanthrene (Ph), pyrene (Py),
benzofluorene isomers (B), chrysene (Ch),
and triphenylene (Tri). Wavelength
monitored: 459 nm. Time scale is from
6.00 min to 17.67 min. . .

Calibration curve for phenanthrene .

Calibration curve for chrysene .

Calibration curve for triphenylene .

3-D chromatogram of 1.16 jg/mL of
fluorene . . .

3-D chromatogram of 1.04 Ag/mL of
phenanthrene . .

3-D chromatogram of 12.6 gg/mL of pyrene

3-D chromatogram of 10.9 Ag/mL of
chrysene . . .

3-D chromatogram of 1.09 pg/mL of
triphenylene . .


105







107

113

115

117


120


122

124


126


128
















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

A SELF-CLEANING, LOW-TEMPERATURE MOLECULAR
LUMINESCENCE SPECTROMETER AS A DETECTOR FOR
GAS CHROMATOGRAPHY

By

John F. Clos

August 1991

Chairperson: James D. Winefordner
Major Department: Chemistry

In recent years, there has been much interest in the

presence of polycyclic aromatic hydrocarbons (PAHs). PAHs

are found in many places, including crude oil, foods, and

water. They are formed either by the decay or the pyrolysis

of organic matter. Also, it has been shown that many of

them are highly carcinogenic.

In the early nineteen fifties, it was noticed that many

PAHs show very sharp spectral lines ("quasi-linear") when

frozen in certain n-alkanes. This phenomena was named the

Shpol'skii effect after the person who discovered it. Since

then, much research has been carried out in this area since

it greatly improves the selectivity for the compounds which

show this effect.

Although this technique has many desirable qualities,

it does have some drawbacks. One of the major problems is

xi










the sample throughput, since samples must be cooled to near

absolute zero temperatures before a spectrum can be taken.

Once the spectrum has been obtained, the cell must be warmed

to remove the sample so the next sample can be applied.

This entire cycle can take several hours. To overcome this

problem, a low-temperature molecular luminescence

spectrometer (LTMLS), with a self-cleaning belt rotating

around a cooled spool and an ambient spool was previously

developed in our laboratory. However, this spectrometer had

its problems, including reproducibility and freezing of the

belt to the spool.

In the present research, the LTMLS was interfaced to a

gas chromatograph (GC). Phosphorescence background and

chromatographic broadening severely degraded the limits of

detection. Even though the background degraded the limits

of detection, chromatograms were collected which show the

potential of this technique. The effect of the hexane-to-

PAH ratio was also investigated. The retention times and

peak widths c` ,ained using the GC/LTMLS system were compared

with those found using GC with flame ionization detection.

The peak widths were found to be approximately 5 to 10 times

worse using the LTMLS system. Finally, the estimated limits

of detection were compared to those found previously using

the LTMLS system and other low-temperature work. The limits

of detection were found to be approximately two orders of


xii









magnitude worse than those obtained previously using the

LTMLS system alone.


xiii















CHAPTER 1
INTRODUCTION


Low-temperature luminescence spectroscopy has become

increasingly popular because it provides greater selectivity

than room-temperature luminescence spectroscopy. This is

predominantly due to the reduction in vibrational,

rotational, and environmental broadening. In the early

nineteen fifties, Shpol'skii and coworkers noted that

certain compounds when frozen in n-alkanes would give

"quasi-linear" spectra [1]. Polycyclic aromatic

hydrocarbons (PAHs) are especially suited for low-

temperature Shpol'skii studies because these molecules fit

into the frozen n-alkane matrices and they have strong

fluorescent and phosphorescent properties. They are also of

great concern because of their significant mutagenic and

carcinogenic effects.


Polycyclic Aromatic Hydrocarbons


Some examples of PAHs along with their carcinogenicity

are shown in Figure 1-1 [2]. PAHs are predominantly formed

either by the long-term degradation of biological material

or by the pyrolysis of organic matter. A possible reaction

scheme for the formation of benzo(a)pyrene (BaP) by


















NAPTHALENE
(Inactive)


ANTHRACENE
(Inactive)


CHRYSENE
(Disputed)


BENZO(A)PYRENE
(Highly Active)


Figure 1-1. Sone examples of polynuclear aromatic
hydrocarbons along with their carcinogenicity. Redrawn from
reference [2].


~~~














pyrolysis is shown in Figure 1-2 [3]. PAHs are found in

marine sediments [4], coal and shale oil [5], airborne

particulates [6], and vegetable oils [7]. Of particular

importance is the presence of PAHs in tobacco smoke, since

they are inhaled directly into the lungs and have been shown

to be carcinogenic [8].


Molecular Effects of PAHs


Of the many causes of cancer, environmental factors are

by far the largest cause, and of these, PAHs are the most

significant group [9]. It is apparent from most research

that the PAHs themselves are not the cause of cancer, but

rather their metabolites [10].

As shown in Figure 1-3, BaP is eventually converted

into two diol-epoxide isomers by Cytochrome P-450 and

epoxide hydrase [10]. These two isomers are considered to

be the ultimate causes of the adverse biological effects of

BaP. They have been shown to bind to DNA, known as

intercalation, resulting in mutations.

At first, it was believed that the carcinogenic effects

of PAHs might be due to the presence of two structural

regions, K and L regions, shown in Figure 1-4 [11].

However, this theory failed to correlate well with many

carcinogens. Later, it was theorized that the presence of

the so-called bay region might actually be a better test of









NI,


Figure 1-2. Possible mechanism for the formation of
benzo(a)pyrene by pyrolysis. Redrawn from reference [3].


i












CYTOCHROME
P-450


EPOXIDE
HYDRASE


I
OH

CYTOCHROME
P-450


OH


OH


Figure 1-3. Reaction mechanism for the metabolism of
benzo(a)pyrene. Redrawn from reference [10].


OH














Bay


region


I K region

L region





Figure 1-4. Structure of benz(a)anthracene showing the K,
L, and bay regions. Redrawn from reference [11].










7

carcinogenicity of the PAHs [12]. In a recent study, it was

shown that a bay region metabolite of BaP binds 2.7 times

more tightly than a K region metabolite of BaP to DNA [13].

Some researchers have used computer modeling to predict the

carcinogenicity of various PAHs [14]. Much more research is

needed to discover the true pathway by which the PAHs

produce their biological effects.


Physical Effects of PAHs


Regardless of how the PAHs cause their mutations, it

has been well documented that they do result in serious

physical abnormalities. It had previously been known that

exposure to coal dust caused cancer; however, it was not

recognized until the 1930s that exposure to BaP itself

resulted in cancerous growths in humans [15]. It has also

been shown to affect rats and mice [16], algae [17], and

various higher plants [18], to name just a few examples.

Due to the high toxicity of PAHs, an accurate method is

needed to measure the various PAHs in widely differing

complex mixtures.


Methods for Determining PAHs


Room-Temperature Techniques


One of the basic and more popular methods for

determining PAHs is high performance liquid chromatography

(HPLC) coupled with either ultraviolet (UV) or fluorescence











detection. This is because the sensitivity is adequate for

most applications and the cost is low compared to other

techniques. Joe et al. determined various PAHs in smoked

foods by extraction with freon followed by column

chromatography and HPLC analysis [19]. Although quite

appealing for several reasons, there are also several

drawbacks. The technique essentially relies on the

separation for selectivity. Even when using fluorescence

detection, there is little difference between the spectra of

PAHs to provide selectivity. Also, since the spectra are

essentially featureless, one has to rely on the retention

time for identification of the analyte.

Along with HPLC, gas chromatography (GC) with flame

ionization detection (FID) is also very popular for many of

the same reasons stated for HPLC. Liberti et al. determined

various PAHs in several different environmental samples by

extraction with a benzene-methanol mixture followed by

column chromatography, HPLC, and then finally GC separation

[20]. The added sample preparation is needed due to the

fact that the environmental samples analyzed, diesel exhaust

and chimney smoke, are extremely rich in PAHs, both in

concentration and variety. Although GC has much higher

separation capacity than HPLC, it suffers from the non-

selective response of FID to all organic compounds.

Therefore, GC loses some of the advantage it has over HPLC

in resolving power.











One way to overcome the loss in selectivity is by

combining GC with mass spectrometry (MS). Liberti et al.

also did GC/MS analysis of the same samples which he

analyzed by GC/FID [20]. By monitoring certain ions,

compounds which coelute can be differentiated by their mass

spectrum. However, little is learned about the structure of

the PAHs. Therefore, it can be difficult to discern between

the PAHs, especially between isomers, unless GC/MS/MS is

employed. However, this adds greatly to the cost, which is

already considerable in doing GC/MS alone.

Some people have attempted to use spectroscopic methods

other than UV detection to improve the selectivity. One

technique which showed much promise initially as a detector

for GC was supersonic expansion. At room temperatures,

luminescence spectra consist of very broad and featureless

peaks due to rotational and vibrational broadening.

Supersonic expansion involves the flow of a gas from one

region of a moderate pressure to another of a very low

pressure through a small orifice (Figure 5). In Figure 5,

the inner area of the expansion is the supersonic jet and

the outer area is the shock wave. The shock wave prevents

the hot molecules in the chamber from interacting with the

rotationally and vibrationally cooled molecules in the

expansion. Two important points in the supersonic expansion

are the start of free molecular flow (XF) and the Mach disc

(X,) [21]. Free molecular flow means that there are


























P,


t
N


- a- -






- I_,


Figure 1-5. Diagram of a supersonic expansion showing the
pressure outside the chamber (P) the pressure inside the
chamber (P,) the start of free molecular flow (X) and the
Mach disc (X,).


Po











essentially no collisions, which in turn, produces cooled

molecules free from environment effects. The Mach disc

indicates the leading edge of the supersonic expansion. By

allowing the gas to expand rapidly through the orifice into

the area of lower pressure, the rotational and vibrational

energy distributions are decreased. This results in a

lowering of the temperature of the molecule and much sharper

spectral peaks.

At first, one would think GC and supersonic expansions

were well suited for each other. However, several problems

arise when one attempts to interface the two. One problem

is the limit of detection, which is rarely below 1 ppm. The

problem is that the flow from the column is too low to give

a good expansion, therefore the flow is pulsed [22]. This

gives a good expansion, but it also creates dead volume

which reduces the chromatographic performance. Another

problem is that the best chromatographic performance is

obtained with helium or hydrogen, while the most efficient

rotational and vibrational cooling in supersonic expansions

is obtained by using argon.


Matrix Isolation Spectrometry


Another technique which gives better selectivity than

UV spectrometry and room temperature luminescence

spectrometry is matrix isolation (MI) spectrometry. Matrix

isolation involves the dilution of the compound of interest









12

(in the gas phase) with a "matrix" gas, usually nitrogen or

argon, followed by deposition on a cryogenic surface. The

principal characteristics of a matrix are [23]: (1) the

molecules are sufficiently separated in the matrix such that

they are unable to "communicate" with one another, (2) the

solvent and the solutes mix in the gas phase and often poor

solvents for the compounds are good matrix solvents, and (3)

the solutes and the solvents should not react.

A significant amount of research has been done in the

area of matrix isolation spectroscopy for detection in gas

chromatography. Conrad et al. used a 12-sided block

attached to the end of a cryostat for deposition of the GC

eluents [24]. As shown in Figure 6a, the sample is deposited

on one of the surfaces, using the carrier gas (either

nitrogen or argon) as the matrix. The cryostat is then

rotated until the sample comes into view of the excitation

light. A spectrum is then taken using a polychromator to

disperse a section of the emitted fluorescence onto the

faceplate of a silicon intensified target vidicon. They

used the GC/MI fluorescence spectrometer to determine 12

anthracene derivatives in a synthetic mixture. Since the

eluent can not be interrogated continuously, it was

necessary to collect the components in fractions. By doing

this, they relied mainly on the spectral resolution to

determine the original derivatives and were able to

spectrally resolve 11 of the 12 components.











13





It-Visble Filter
a) S


Sphirica Lene ..
Polychomaftor







SIT Vidicon-







b)





CRYOSTAT
""COLD-FINGER"

S.GOLD-PLATED
DISK
-KEL-F SPACER
STEEL
SPRING
-ROTATING
FEEDTHROUGH

DRIVE
SHAFT


-STEPPING
MOTOR






Figure 1-6. Diagrams of a) optical system (top view) and b)
cryostat head (side view) used by Conrad et al. for
GC/matrix isolation. Taken from reference [24].









14

Work has also been done on GC/MI infrared spectrometry.

Reedy et al. built a spectrometer, shown in Figure 7, which

was a disk connected to the cold stage of a cryostat for

continuous sampling, although only for 110 min, unless the

GC capillary was repositioned [25]. Instead of using argon

as the carrier, they used helium with 2% argon since helium

is cheaper and gives much better chromatographic efficiency.

By using infrared spectrometry, they were not only able to

measure pyrene and fluorene, but also several

polychlorinated biphenyls and aliphatic hydrocarbons.

Organic glasses can also be used in matrix isolation.

Brown et al. used hydroorganic mixtures as their matrix and

a laser to produce what is called fluorescence line

narrowing (FLN) [26]. FLN results from the narrow bandwidth

of the laser beam and the weak electron-phonon coupling

which causes broadening of the spectral peaks. They were

able to distinguish between several isomers of anthracene

and found detection limits to be in the parts-per-trillion

range. One advantage of their procedure is that water

samples can be analyzed directly without any extraction

steps since water is a part of the matrix.


Shpol'skii Spectroscopy


As stated previously, a Russian scientist named

Shpol'skii and his coworkers discovered that by freezing

PAHs in n-alkanes, spectral peaks become unusually sharp -




























































Figure 1-7. Diagram of experimental system used by Reedy et
al. for GC/matrix isolation/FTIR. Taken from reference
[25].









16

[1]. Since then, considerable research has been done in the

area of Shpol'skii spectroscopy (SS). SS is essentially a

matrix isolation technique, however, it is almost

exclusively applicable to PAHs in n-alkanes.

Inman et al. studied the fluorescence of 41 compounds

and the phosphorescence of 60 compounds at 77 K in

Shpol'skii matrices [27]. Most of the compounds had

detection limits in the low ppb range with some in the pptr

range. One advantage of the system they used is that the

instrumentation is commercially available.

One of the attractive features of SS is the ability to

resolve compounds of rather similar structure. Colmsjo was

able to show that 13 of the 16 PAHs of molecular weight 328

have distinct excitation and emission wavelengths [28]. He

also did a similar study for PAHs of molecular weight 378

[29].

As previously stated, PAHs are of great concern because

of their mutagenic and carcinogenic effects. Therefore, a

significant amount of research has been done on the

application of SS to environmental samples. Renkes et al.

determined benzo(a)pyrene, benzo(k)fluoranthene, and

benzo(ghi)perylene in particulate samples [30]. The PAHs

were extracted from the samples by heating them in

diphenylmethane or phenyl ether. These extracts were then

diluted 10- to 100-fold with either n-heptane or n-octane.









17

The concentrations determined by this method were comparable

to the certified values for the NBS standard studied.

The analysis of fossil fuels is another area in which

SS has been applied. This is largely due to the high

concentrations and wide variety of PAHs in fossil fuels and

also the complicated matrix. Perry et al. were able to

determine four PAHs in a coal reference standard and

benzo(a)pyrene in a shale oil reference standard [31].

Although SS is a powerful tool for resolving PAHs, it

can still be very difficult to determine them. This is

especially true when analyzing complicated matrices or when

many PAHs are present. Therefore, as with other matrix

isolation techniques, many researchers have tried to combine

SS with liquid or gas chromatography.

Colmsjo and Stenberg looked at six PAHs in automobile

exhaust and air samples by HPLC fractionation followed by

analysis using SS [32]. The samples were collected using a

method which results in a particulate phase (on a glass

fiber filter) and a condensed water phase. They were then

prepared for separation by extraction using

dimethylformamide, water and cyclohexane.

Two HPLC gradient separations were performed. The

first was to separate the polar compounds from the three or

more ring PAHs and the second fractionation was used to

separate the PAHs for E. : alysis at 63 K. By this method,

they were able to idert: :: :ix PAHs in automobile exhaust.









18

Petroleum samples are much more difficult to deal with

due to their complexity if one wishes to analyze them for a

wide variety of PAHs. Garrigues et al. used three

chromatographic separations to analyze shale oil and coal

for alkylphenanthrenes followed by SS [33]. The first

separation used a semi-preparative alumina column to

separate the aromatic compounds by ring class. The

appropriate fraction was then separated on a reversed-phase

column to resolve the PAHs based on their degree of

alkylation followed by a second reversed-phase separation to

further resolve the phenanthrene derivatives. These

fractions were then analyzed by either SS or GC-MS. By

combining these techniques, they were able to identify 17 of

the possible 30 ethyl- and dimethylphenanthrene derivatives.

One problem with the combination of SS with

chromatographic techniques is the necessity to collect

fractions to analyze them. Up to now, fractions were

collected, the eluent was switched to look at a few small

sections of the chromatogram, or, at best, a few

chromatograms could be analyzed continuously. This leads to

errors introduced by the technician, lower effectiveness of

the separation, hinderance of automation, and thus lower

reproducibility. To overcome this, Jones and Winefordner

described a luminescence spectrometric system with a self-

cleaning continuously cooled belt [34].











As shown in Figure 8, the sample in n-hexane is

introduced onto the belt where the belt is in contact with

the cold stage of the closed-cycle helium refrigerator.

Therefore, since the cold stage is at approximately 15 K,

the sample is rapidly frozen to the belt. The sample is

then rotated 90 degrees by means of a feedthrough connected

to the rear spool so that it can be excited by the output

from the xenon arc lamp. After a spectrum has been

obtained, the belt is further rotated so that the part of

the belt covered by the sample is no longer in contact with

the cold stage. The sample then warms and the high vacuum

(105 torr) removes it from the belt; therefore, the system

is self-cleaning. They used this spectrometric system to

analyze discrete extractions from cooked beef for PAHs [35].

Although it allowed continuous sampling without having

to warm the chamber to remove the sample, low

reproducibility of repeated injections of the same standard

was a problem. This was partially due to the samples being

injected manually with a syringe. Because of this, the

spots on the belt were not reproducible in shape, which

changed the amount of sample viewed, and therefore, the

signal.

The current research involves the interfacing and

evaluation of a gas chromatograph with the spectrometric

system described by Jones and Winefordner. This should not

only improve the reproducibility, but also greatly improve







































































*4
C,
















*M
(a






,-4











.r-
P4
^1











!-
__

/__../


o go
o ,Q
( 0<
n l/O(--
(9j









22

the selectivity due to the high resolving power of capillary

gas chromatography. Phosphorescence background and

chromatographic peak broadening were severe problems which

degraded the limits of detection. However, chromatograms

were collected which show the potential analytical

usefulness of this technique. The effect of the hexane-to-

PAH ratio was also investigated. The retention times and

peak widths obtained using the LTMLS system as the detector

for the gas chromatograph were compared with those found

using flame ionization detection. Finally, the estimated

limits of detection were compared to those found previously

using LTMLS system alone and also other low-temperature

work.















CHAPTER 2
THEORY


Luminescence Spectroscopy


Luminescence spectroscopy involves the absorption and

subsequent emission of light. As shown in Figure 2-1, when

an electron is excited to a higher energy level, it can

return back to the ground state in several ways. First, it

can release the energy by radiationless processes, such as

heat or by collision. Second, it can emit light having the

energy equal to the transition from the excited state to the

ground state, which is termed fluorescence. Third, it can

cross over to a triplet state by radiationless processes and

then emit light to return to the ground state, called

phosphorescence, or be quenched by interactions with the

matrix.

There are several differences between fluorescence and

phosphorescence [36]. First, the time scale for

fluorescence is on the order of 108 to 10 s, while

phosphorescence lasts 103 s or longer. This is because of

the delay caused by the intersystem crossing from the

singlet to the triplet state and the lifetime of the triplet

state. Second, in fluorescence, the emission spectrum is

the mirror image of the excitation spectrum, whereas in

23




















t I
./


A












S
0


1'
I


\/


Figure 2-1. Energy level diagram showing absorption (A),
fluorescence (F), phosphorescence (P), and radiationless
processes (dashed arrows). So is the ground state, S1 is
the excited singlet state, and T1 is the excited triplet
state.


I
I


.... 9--m-r-











phosphorescence, besides not being the mirror image, the

emission spectrum is also shifted to longer wavelengths.

Third, phosphorescence is rarely observed in gases and

seldom at room temperature in solutions; however

fluorescence is observed in both gases and room temperature

solutions. This is because the time scale for molecular

collisions is similar to that of fluorescence. Therefore,

the absorbed energy is released by radiationless processes

before the molecule has had time to phosphoresce.


Band Broadening



Although molecular luminescence techniques are more

sensitive and selective than absorption spectroscopy, they

still suffer from band broadening. There are several types

of band broadening in molecular spectroscopy. First, an

excited electron does not have a precisely defined energy,

therefore it has an uncertainty in its amount of energy,

depending on its lifetime [37]. Because of this inherent

uncertainty, the spectral lines are broadened into bands;

however, this natural broadening is only 103 to 10-4 cm'

[38].

Molecular bands are also broadened by what is called

the Doppler effect [36]. This is caused by movement of the

molecule either away from or to the detector. If the

molecule moves toward the detector, the apparent r( Luency











is increased, and if it moves away, then the apparent

frequency is decreased.

Another cause of spectral band widening is collisional

or pressure broadening [37]. Collisional broadening is

probably the most significant cause of broadening of

spectral bands. This is caused by the interaction of the

molecule of interest with surrounding molecules. These

interactions cause shifts in the excited state energy

levels, changing the energy level for each individual

molecule slightly.

Band broadening is also caused by what is known as the

Stark effect [39]. In 1913, Stark discovered that an

external electric field caused lines to split in atomic

spectra. This is due to the influence of the electric field

on the dipole moment of the atom. This has also been shown

to apply to molecules as well as atoms.

Another cause of spectral band spreading is known as

resonance broadening [40]. This results from the fact that

like molecules will have the same natural frequency. This

can cause a strong coupling between the two molecules, and

therefore, results in a dispersion of the frequencies.

These interactions also occur between unlike molecules, but

the force between them is not as strong.

Most of these broadening effects fall under what is

known as inhomogeneous broadening [41]. Essentially this is

due to external effects acting on the molecule. These











effects are not consistent throughout the liquid or solid.

This is usually due to the fact that each molecule is not

surrounded by the same environment, so that each molecule is

influenced differently. This, in turn, means that the

excited energy state of each molecule is unique, and so when

it returns to the ground state, it will emit at a slightly

different wavelength. Finally, spectral bands can be

broadened by the monochromator itself.

Broadening effects follow one of two equations, either

the Lorentzian or the Gaussian profiles [42]. The

Lorentzian profile is governed by the equation:

2/(tAvL)
1+[2(VM-V)/AVL]2


whereas, the Gaussian profile follows the equation:

2 (ln2/7) 1/2 e-4(n2) (v-v,)2/Av


where SL is the normalized Lorentzian spectral profile in

Hz 1, AML is the half-intensity width of the Lorentzian

profile,v, is the central frequency, SvG is the normalized

Gaussian spectral profile in Hz 1, and AvG is the half-

intensity width of the Gaussian profile. As can be seen in

Figure 2-2, the Lorentzian profile has more extensive wings

than the Gaussian profile [43].

An energy level diagram and two spectra are shown in

Figure 2-3. The energy diagram shows not only the

electronic levels, but also the rotational and vibrational
















I

a)










i.










b)
I

Io





I,

I














Figure 2-2. Graphs showing a) Lorentzian and b) Gaussian
profiles. Taken from reference [43].









a)


WAVELENGTH -


c)


WAVELENGTH --



Figure 2-3. Diagram showing a) an energy level diagram and
b) well resolved and c) poorly resolved spectra.


I I











levels. One of the spectra has well resolved spectral

bands, while the other has poorly resolved bands. Because

of the broadening effects, electrons are spread into

different vibrational and rotational levels [44]. Also

because of the broadening effects, the fine structure of the

spectrum is no longer resolved, but becomes smeared into

much broader bands. This decreases the utility of

luminescence spectroscopy for identifying the compound

present and the selectivity when two or more luminescent

compounds are present.


Cooling Effect


One way in which to reduce some of the broadening

effects is by cooling the molecules, especially down to

cryogenic temperatures (15 K) [45]. First, since the

thermal motion is greatly reduced, the Doppler broadening is

negligible. Also, because of the reduction in motion,

collisional broadening is removed. Resonance broadening can

also be eliminated by using mixed crystals. Finally,

radiationless transitions to other vibrational and

rotational levels are reduced because of the smaller

available thermal energy.

As shown in Figure 2-4, there is a dramatic effect in

the reduction of the temperature of the solution on the

fluorescence spectra [46]. At room temperature, the

spectrum is basically one broad band, offering no
































450

EMI SSION


470
WAVELENGTH


Figure 2-4. Spectra of perylene (700 ng/mL) in n-hexane at
three different temperatures. Taken from reference [46].


298 K





77 K





15 K


9K Mk


490


( nm)











information on the identity of the analyte. However, when

the temperature is reduced to 77 K, individual bands begin

to appear. At 15 K, there is an even greater improvement in

resolution with some of the peaks becoming completely

resolved. The reduction in band width is partially due to

the Shpol'skii effect, but still demonstrates the usefulness

in cooling the sample to increase the spectral information.

Although many of the broadening effects are removed by

cooling, some of the remaining broadening effects still

cause significant spreading of the spectral bands. This

results from the fact that the matrix in which the compound

is frozen still contains inhomogeneities [41]. As stated

before, these different environments (called sites) cause

the molecule to have different excited energy states, which

in turn results in the slight variations in the emitted

wavelengths. This is shown schematically in Figure 2-5.

When molecules emit light, each produces a sharp peak, known

as a zero phonon line (ZPL), and a broad phonon wing (Fig.

2-5a) [47]. Because of the environmental effects, the ZPLs

occur at different wavelengths (Fig. 2-5b). The combination

of the broad wings and the environmental effects results in

the broad bands (Fig. 2-5c).

One method to reduce the spectral band broadening is by

annealing the sample. When samples are cooled too rapidly,

defects form in the crystalline lattice. These defects

result in the analyte being exposed to different











a)










b)









c)








WAVELENGTH ->

Figure 2-5. Schematic showing inhomogeneous broadening: a)
a transition for a single molecule, b) transitions for many
molecules, and c) how the band actually appears. Reproduced
from reference [47].











environments, and therefore, giving slightly different

spectral characteristics. Annealing involves heating the

crystal to a temperature of about one-half the melting point

of the solvent [45]. This allows the solvent structure to

rearrange, and thus, eliminate the imperfections.

One must be careful, though, not to heat the sample too

much, since the analyte molecules may aggregate together

[45]. Because of this aggregation, coupling of the energy

levels of the analyte may result. This would then lead to

increased spectral broadening. Aggregation can also result

if the solute to solvent ratio is too high or from cooling

the sample too slowly.

Another technique for reducing spectral broadening is

site selection spectroscopy [47]. By using a spectrally

narrow excitation source, usually a dye laser, one is able

to excite only those molecules which are in one or a small

number of environments. This results in spectra with very

sharp peaks, usually regardless of the matrix. Examples of

this effect are shown in Figure 2-6 [47]. The dotted lines

are spectra with ordinary excitation and the solid lines are

with laser excitation. One can clearly see that by using

selective excitation, the spectral peaks become much

sharper. Also, in many cases, one can see the ZPLs along

with their broad phonon wings.



























0|
I








I


- I I [ i f 1 -
o 7100 6700 0 6800 6400
-A .A


Figure 2-6. Fluorescence spectra of various compounds in
different organic matrices at 4.2 K by ordinary excitation
(dotted lines) and laser excitation (solid lines): a)
perylene in ethanol, b) protonated form of 9-aminoacridine
in ethanol, c) tetraphenylporphin in polystyrene, and d)
protochlorophyll in ether. Taken from reference [47].












Shpol'skii Effect


Site selection spectroscopy can be very useful for

reducing the amount of spectral broadening; however, there

are two significant problems with the technique. First, it

is only useful for those molecules in certain sites which

can be excited at the specific wavelength of the laser.

Secondly, phosphorescence, and in some cases fluorescence,

spectral bands are not narrowed in site selection

spectroscopy [48]. This results from the molecules being in

similar sites, but not identical sites [49]. Therefore, the

molecules may have the same S1-S0 (fluorescence) transition

energies, but different T1-So (phosphorescence) transition

energies. To overcome this, researchers have tried exciting

the molecules directly to the triplet state [50]. However,

since it is a forbidden transition, this excitation process

has a low probability.

Shpol'skii spectroscopy, although not as universal as

site selection spectroscopy, is applicable to nearly all

PAHs. Therefore, an entire class of compounds can be

determined, qualitatively and quantitatively, by the use of

one technique. It is also useful for some compounds other

than PAHs.

Even though Shpol'skii spectroscopy has been studied

extensively, it is still not fully understood. However, it

has been established that the Shpol'skii effect is related









37

to the shape and dimensions of the PAH in relation to the n-

alkane used. This relationship is shown in Table 2-1 [51].

These data show that the optimum Shpol'skii solvent for each

compound matches the linear dimension of that compound. As

shown in Figure 2-7 [51], the alkanes have a zig-zag

conformation which allows the PAHs to fit nicely into the

frozen matrices. This was referred to as the "key and hole"

rule by Pfister [52].

Although the principle works well for the linear PAHs,

it becomes more difficult to see for other PAHs, such as

coronene and 1,2-benzo(a)pyrene, which are not linear. The

proper Shpol'skii solvents have to be determined

experimentally for each individual PAH. Lai et al.

determined the best solvent for 23 different PAHs [53].

Work has also been performed using X-ray diffraction

and ESR to study the Shpol'skii effect [54]. From the X-ray

experiments, Merle et al. were able to determine the

crystalline form of heptane and from the ESR experiments,

the orientation of the coronene molecule with respect to the

heptane lattice. The results of their studies are shown in

Figure 2-8. The coronene molecule replaces three of the

heptane molecules and lies in the plane of the other heptane

molecules. Of course, the orientation and number of n-

alkane molecules replaced will depend on the PAH and the n-

alkane used for Shpol'skii spectroscopy.



































Table 2-1. Data showing dimensions for several PAHs and their
best Shpol'skii solvent. Taken from ref. [50].


PAH
naphthalene
anthracene
naphthacene

*in angstroms


length*
7.2
10.0
12.8


n-alkane
pentane
heptane
nonane


length*
7.4
10.0
12.8










a)


b)


c)


Figure 2-7. PAHs along with their optimum Shpol'skii
solvent showing how they relate: a) naphthalene and
pentane, b) anthracene and heptane, and c)naphthacene and
nonane. Redrawn from reference [51].




























































Figure 2-8. Orientation of coronene and heptane molecules
with respect to the crystallographic axis. Taken from
reference [54].















CHAPTER 3
EXPERIMENTAL


Instrumentation


A Perkin-Elmer (Norwalk, CT) Model 8500 gas

chromatograph with a J & W Scientific (Folsom, CA) 30 m x

0.32 mm DB-5 capillary column was used for separating the

PAHs. For the transfer line, part of the column itself was

used. The heated transfer line was made from 0.030" x 1/16"

316 stainless steel tubing. A 1/16" stainless steel male

union with an Alltech Associates, Inc. (Deerfield, IL) 1/16"

to 0.4 mm graphite reducing ferrule was used to seal between

the capillary column and the stainless steel tubing at the

GC. A three way stainless steel tee was inserted into the

transfer line for the introduction of hexane as a mixture of

781 ppm hexane in helium. The hexane/helium mixture was

introduced with a Newport Corporation (Fountain Valley, CA)

BV100 Molecular Beam Valve. The transfer line was heated

using an SPC Technology universal step-down transformer

distributed by Newark Electronics (Jacksonville, FL) and a

Variac. The current was applied to the transfer line by

silver soldering the wire to the union at the GC and silver

soldering the other wire to the stainless steel tubing

approximately 1.5 cm from the end of the tubing at the

41











vacuum chamber. Supelco, Inc. (Bellefonte, PA) Supeltex M-

2A 1/16" ferrules were used for the three-way tee and the

reducing union inside the chamber (see Figure 3-1) and at

the GC.

The GC/vacuum chamber interface used in these

experiments is shown in Figure 3-1. This design was based

on the work by Brown and Wilkins [55]; the interface was

built in our departmental machine shop. The 0.25" ID x

0.75" OD Teflon V-rings were obtained from Hyarco (Skokie,

IL). The 1/8" ID x 1/4" OD ceramic tube was obtained from

Omega Engineering, Inc. (Stamford, CT). High vacuum grease

from Dow Corning Corp. (Midland, MI) was used to obtain a

more efficient vacuum seal between the ceramic tube and the

Teflon V-rings. Supeltex M-2A 1/16" and 1/4" ferrules were

used for the reducing union. The interface formed a vacuum

seal by tightening the lower nut which compressed the Teflon

V-rings. The upper nut had retainer rings on each side of

it, which stayed in place since the ceramic tube had ien

notched. When the upper nut was turned, it pushed on the

retainer rings causing the ceramic tube, and therefore the

tip of the capillary, to move. Thus, the tip of the

capillary could be positioned precisely while still

maintaining a high vacuum. When doing manual injections,

the interface was replaced with a septum and a Cajon fitting

welded to a face plate.









CAP. >
COLUMN


CERAMIC
TUBE -


REDUCING
UNION


< SS


TUBING




WER NUT


TEFLON
V-RINGS


Figure 3-1. GC/vacuum chamber interface used in these
experiments. Based on the work by Brown and Wilkins [55].


--F









44

The vacuum chamber consisted of a Vac-U-Flat five-way

cross and a Vac-U-Flat three-way cross, each with 6"

flanges, a Vac-U-Flat three-way cross with 2" flanges, a 6"

quartz viewport, and two direct-drive rotary motion

feedthroughs which were all obtained from Huntington

Mechanical Laboratories (Mountain View, CA). A Turbo-V80A

turbomolecular pump, an SD-90 direct-drive mechanical pump,

a Model 860A cold cathode ionization gauge, and a Model 801

thermocouple gauge were obtained from Varian Vacuum Products

(Lexington, MA). The vacuum pressure was in the 105 torr

range when the beam valve and the flow from the GC were off.

When the beam valve was on, helium was flowing from the GC,

and the system was cooled, the pressure was approximately 1

mtorr and the belt temperature was between 50 and 60 K. A

laboratory built liquid nitrogen trap and a molecular sieve

trap were place between the turbo and mechanical pumps to

reduce backstreaming. An APD Cryogenics, Inc. (Allentown,

PA) displex Model CS-202 closed-cycle helium refrigeration

system was used to maintain the cryogenic temperatures.

The vacuum chamber, which is similar to the one used by

Jones et al. [56], is shown in Figure 3-2. The Cajon

fitting, which had been welded directly onto the five-way

cross, was removed and replaced with a plate which was

welded on. In this way, either the GC interface or the

Cajon fitting could be used for introducing the samples.








) TURBO
PUMP

E

QUARTZ
WINDOW


CRYOSTAT


HAND
CRANK













STEPPING


MOTOR


VACUUM


GAUGES


Figure 3-2. Vacuum chamber used in these experiments: a)
top and b) side views. Belt and spools are shown as dashed
lines.









46

Also, the Vac-U-Flat flexible coupling, which had been

used to tighten the belt, was removed, since it was

difficult to work with it and it was thought that the

coupling was causing contaminants to remain in the chamber.

The method which was then used for tightening the belt is

shown in Figure 3-3. The two feedthroughs which had been

used for rotating the rear spool to turn the belt were

mounted on plates. These plates were held onto flanges with

bolts and O-rings were placed between the plates and flanges

to form the vacuum seal. The plates which held the

feedthroughs onto the vacuum chamber had oval holes for the

mounting bolts allowing the plates to be moved in the

direction of the belt to tighten the belt.

To turn the belt continuously at a constant rate, a

Hurst (Princeton, IN) Model ABS3008-003 stepping motor with

1800 steps per revolution and a Hurst Model EPC-015 stepping

motor controller were employed. The stepping motor was

mounted to a U-bracket which had one of the feedthrough

plates as a part of the bracket. A knob was mounted on the

other feedthrough for manually turning the belt, but the

stepping motor had to be disconnected before using it.

As previously done by Jones et al., the front spool was

bolted directly onto the end of the cold stage with an

indium gasket between them and did not rotate. The spool

was made of oxygen-free copper and was 3/4" in diameter and

7/8" wide. It had a groove cut in it 1/64" deep and 11/16"






































VACUUM
-----CHAM-BER--/


Figure 3-3. Schematic showing how the belt was tightened:
a) the belt loose and b) the belt tightened. Belt and spool
are shown as dashed lines.


a=3 ~

(O












wide so that the belt would remain on the spool. The rear

spool was made of aluminum and also had a 11/16" wide groove

so that the belt would remain in place. The portion of the

spool which the belt passed over was 1/2" in diameter and

the entire diameter was 3/4". The belt itself was made of a

0.003" thick brass sheet cut 5/8" wide and 23 1/2" long.

This strip of brass was then spot welded with a 1/4" overlap

to form the belt.

The spectroscopic portion of the system is shown in

Figure 3-4. The entire output of an Oriel Corp. (Stamford,

CT) Model 8500 75-W xenon arc lamp was used for excitation.

The output from the arc lamp was directed onto the belt by

passing it through an adjustable aperture, reflecting it off

a mirror, and then passing it through another adjustable

aperture and a 6" focal length quartz lens. Another 6"

focal length quartz lens was used for collecting the

emission and f causing it into an Instruments SA, Inc.

(Metuci .i, NJ) Model UFS-200 flat field spectrograph. This

spectrcgraph allowed the Princeton Instruments, Inc.

(Princeton, NJ) Model IRY-700 photodiode array to see the

200-800 nm range. Along with the photodiode array, a

Princeton Instruments, Inc. Model 120 OSMA Detector

Controller and a PCs Limited (Austin, TX) Model AT113

personal computer were used for collecting data. An Ithaco,

Inc. (Ithaca, NY) Model 382A optical chopper with a three-

hole chopping wheel operated at 700 Hz was used for





























4-1

0>
0
0
(0










0
*4










4J.
0
i,













0
M




















4-
0

5
4




0 t







4.)
*0
a


a
































*0
*4

10
'O
(0



















3 Q)
CPC ~










w

I I
LO 0
2(


LLIl
CIII

K-
Cc


LLI

0<

0<
I
1


z-J
z _



W0
<


-i
LUI
C13


5 :

>5









51

modulating the light. For the measurements at 77 K, a SPEX

Fluorolog double monochromator fluorimeter was used.


Reagents


Pyrene, 1,2-benzofluorene, 2,3-benzofluorene, and

triphenylene were obtained from Aldrich (Milwaukee, WI) and

phenanthrene, fluorene, and chrysene were obtained from

Eastman Kodak (Rochester, NY). All were labeled as 98% pure

or better and were used as received. Hexane (UV Grade) was

obtained from Burdick & Jackson (Muskegon, MI). It was

labeled as 99.9% pure and also was used as received. Stock

solutions of approximately 100 Ag/mL for each PAH was made

and dilutions were made from the stock solutions as needed.


Procedure


When doing transfers from the gas chromatograph, the

capillary was fed through the transfer line and an

approximately 1 cm section of capillary was left protruding

from the end of the stainless steel tubing. The liquid

nitrogen trap was then filled and the roughing pump started.

After approximately two minutes, the turbo pump was started.

The transfer line was then heated and the helium

refrigerator turned on.

It was necessary to wait until the system was under

vacuum and the transfer line was heated to adjust the height

of the capillary tip above the surface of the belt, since









52

the capillary would draw up into the stainless steel tubing

upon heating the transfer line. The capillary was adjusted

to a height of approximately 0.5 mm above the belt. It has

been recommended to have the tip of the capillary

approximately 100-200 Am above the deposition surface [55].

The tip of the capillary was farther from the surface than

recommended because the column would become plugged by

hexane freezing in the tip of the capillary unless it was

atleast that high above the surface.

One microliter of the standards were then injected into

the gas chromatograph. The injector was operated in the

splitless mode for 2 min, then the split was opened for the

rest of the run. The oven temperature was maintained at

1300 C for 2 minutes, the temperature was then ramped at 300

C/min to 3000 C and was held there for 10 min. The entire

run was approximately 18 min. The injector and detector

were maintained at 3200 C.

Each spectrum was averaged for 2.5 s. Since large

files were produced and they required a significant amount

of memory, it was necessary to wait until just before the

peaks eluted before beginning to collect data. Since the

amount of hexane being introduced through the molecular beam

valve was too low (see Chapter 4), only the lowest

concentration injected and the origin were used to estimate

the limits of detection. The average noise was determined

by zooming in on a 30 diode section of the background









53

spectra where the emission peaks used for the calibration

curves appeared. The photodiode array program automatically

calculates the mean and standard deviation of those 30

diodes. The standard deviation was then multiplied by

three, which was taken as the intensity at the limit of

detection.















CHAPTER 4
RESULTS AND DISCUSSION


Phosphorescence Background


One of the major problems encountered in this project

has been that of phosphorescence background, a spectrum of

which is shown in Figure 4-1. This background is seen

whenever a liquid is injected onto the belt, whether it is a

pure solvent or a standard. Hexane, heptane, methanol, and

acetonitrile all have been injected and each give the same

background; however, it is not seen when nothing is

injected. It is not scatter, since the phosphorescence can

actually be seen on the belt with the naked eye as a blue

glow. Also, it appears whether the solvent is injected

manually with a syringe or by first injecting it onto the

gas chromatograph. The background is also seen continuously

when the molecular beam alve is used to introduce the

hexane. Along with the background, the vacuum pressure is

now in the 10'5 torr range for the chamber alone, which is

approximately 2 orders of magnitude higher than previously

obtained.

In order to remove or reduce the phosphorescence

background, a number of experimental procedures have been

performed (see Table 4-1). First, spectra were taken with

54






















(0

















U
t0










-,
(0
a)
x
40











0)







4-)
-i
0

















134
0)














0
U
U)










s-I




c-
(U

Ul



X!
0<
in












-1-1
Fr





















O
0














0
'0
Q















o @ *
o






'A



-QS
0
oC



0 6










0







- 0






-Io
0


o 0 0 0
o o o o
t c< -D *


kHmuoluI












Table 4-1. List of experimental procedures to delineate and
to minimize the phosphorescence background problem.


Source
1. belt (no
solvent)

2. belt (with
solvent)

3. hexane


4. pump oil
in hexane

5. leaks


6. belt (with
hexane)


7. belt (with
hexane)


8. belt (with
hexane)

9. belt (with
hexane)


10. ion gauge


11. ion gauge


Experiment
spectrum at 15 K


same


spectrum at 77 K
on SPEX

spectrum at 77 K
on SPEX


leaks were found
and corrected

cleaned chamber
with acetone
(see text)

same as #6, but also
cleaned cryostat and
turbo with Freon

added molecular sieve
and nitrogen traps

wrapped chamber with
heating tape while
under vacuum

put on cold cathode
and hot filament ion
gauges simultaneously

put on hot filament
ion gauge alone


Replicates Results
many NSB*


many


once


once


background
present


NSB


PSB


no change
from #2

several no change
from #2


once


several


once


no change
from #2


no change
from #2

no change
from #2


no change
from #2


pressure
dropped
tx 2 x
10 torr


CONCLUSION: Since the pressure has dropped, most likely
background is gone; therefore, phosphorescence background
probably due to contamination in the ion gauge, most likely
pump oil.

*NSB: not source of background; PSB: possible source of
background.









58

the belt cooled, but no solvent or sample injected onto the

belt. As stated previously, the phosphorescence background

is not observed when there is no solvent or sample present.

Since the phosphorescence background is only seen when

a solvent is injected, the first thought was that the

solvents were contaminated. Figure 4-2 shows spectra of

phenanthrene in hexane, hexane, and an empty tube (blank)

taken at 77 K using the SPEX fluorimeter. As one can see,

the signal is larger for the empty tube alone than when

hexane is present, indicating that the signal must be due to

scatter.

Another possible source of phosphorescence is pump oil

from the roughing pump used to back-up the turbomolecular

pump. Therefore, a spectrum of pump oil was taken using the

SPEX fluorimeter, which is shown in Figure 4-3. One drop of

pump oil was diluted in 20 mL of hexane. As one can see,

the hexane gives essentially no signal, as is expected;

however, the pump oil phosphoresces very strongly.

A large vacuum leak was discovered in the vacuum

chamber near where the cold spool is located. It was hoped

that when the leak was corrected, the background would go

away. However, the background still remained after the leak

was sealed. All of the leaks which were located using a

helium leak detector were corrected. Therefore, the

background was not a result of leaks in the system.

















a)

4=)






o
40 *


0)
a

X




M
Q)
4*

0 *
CX -

0
X-.d



W M
0 00











v-4 C(
S-)


0 4)
CX









0W
a-)





NM






0)



00








CM
(054
CJ
a)












<. .
Q) S

3~ (0
O'r 1













































































o 0 0 0 0
o 0 0 0 0
(' '0 ff~orr


AXlsuoluI


A

O
*-<


00
o











*n
0









o
o '





















4)






4
(0

*4
40



















*HT








4J
0)

0 r:



4 0
X






















X*
Q)r





cO












4-I
'lC
X- r



























-H (0
ui X
ML
&
*r w
nr~





(spuasoqL)
Ai!suotuI


C





C,










C





p









o
0O





0:
ff





0)
0


I











The system has also been completely dismantled and

thoroughly cleaned. All parts of the system which are

bakable were placed in an oven at 2000 C for 24 hr. The

vacuum gauges were cleaned as thoroughly as possible with

acetone. After reassembling the system, the vacuum pressure

and the background remained the same. The cleaning process

was then repeated with the exception that this time the

turbomolecular pump and the cryostat were cleaned with

Freon. The background still remained unchanged.

A molecular sieve trap and a liquid nitrogen trap were

placed in line between the turbomolecular pump and the

roughing pump to prevent backstreaming of pump oil when the

turbomolecular pump was off. Also, the chamber was heated

using heating tape while under vacuum. Neither of these

things had any effect on the vacuum pressure or on the

phosphorescence background. Since no other possibilities to

correct the problem remained, it was decided to go ahead and

collect the initial data in order to prove the efficacy of

the approach.

After the data had been collected, the turbomolecular

pump was connected directly to the ion gauge in an attempt

to isolate the problem. The vacuum pressure was still

approximately 105 torr. Another high vacuum gauge had been

previously used on the system, but with the ion gauge also

connected to the vacuum chamber. With a different ion

gauge, the pressure now reached 2 x 10"7 torr. Therefore,











the problem may have been contamination in the ion gauge,

since it read the same as the old gauge when both were on

the system and no leaks were detected around the ion gauge

with the helium leak detector. Since then, the chamber has

been completely reassembled and a pressure of 2 x 10"7 torr

has been achieved.

Since there were no leaks present and the solvent had

been checked for phosphorescence background on another

system at 77 K, the background was most likely a result of

contamination in the system, probably from pump oil. One

way to determine whether or not pump oil caused the

background is to first correct the problem and then to

introduce pump oil manually with a syringe and to see if the

background returns. This experiment might be unwise since

the system could simply become contaminated again with pump

oil.


Chromatoqraphy


Chromatograms


Although the phosphorescence background degraded the

effectiveness of the system, data were still collected which

showed the potential of the GC/LTMLS system. All

chromatograms shown have been background subtracted. Figure

4-4 shows a 3-D chromatogram of 78 jg/mL phenanthrene. One

can see that an impurity comes off at 8.39 min with a

spectral maximum at 423 nm (I) just ahead of the

















.0






0 r-q
As
c ^

**01 (








4

e>
m













.C
4-1



0W















O
-4













0














0o





-4J 4



0r.
-1 0
*H M
CO *









Sco











4 4-I
OQ)


6 C



coa

M ,C









O
i j








F--
(5
E
C
1..


-.J










0
,/,_ISN _INI
I







~~O
~O











phenanthrene peak. One can also see several peaks in the

spectra when phenanthrene elutes which are not seen in

either the 10 gg/mL phenanthrene chromatogram (Figure 4-5)

or in the manual injection of 131 Ag/mL of phenanthrene

(Figure 4-6). These peaks can be seen more easily in Figure

4-7, which is a spectrum of 78 Ag/mL of phenanthrene at its

chromatographic peak maximum.

In Figure 4-6, one can see a peak corresponding to the

impurity (I) and the peaks for phenanthrene which is seen at

both concentrations of phenanthrene in the chromatograms

(Ph). However, the peak due to aggregation (Ph(A)) in the

chromatogram of 78 Ag/mL is not seen in the manual

injection. One explanation of this anomaly is that there

may not have been a sufficient amount of hexane introduced

through the beam valve for the higher concentration of

phenanthrene. If the ratio of hexane to phenanthrene were

too low, the phenanthrene molecules would not be isolated

from one another enough to prevent interaction of them which

could lead to spectral peaks arising from those

interactions.

A similar effect to the one which was observed for

phenanthrene was also observed for pyrene. A 3-D

chromatogram of 94 gg/mL pyrene is shown in Figure 4-8, a

spectrum of pyrene at its chromatographic peak maximum is

shown in Figure 4-9, and a spectrum of 123 Ag/mL pyrene is

shown in Figure 4-10. Once again, there are peaks which are

















c4






U)








0)




a,
c


















0
M


























tu
.O









t(




















0
r.
u








F4
c





0



cII
4-


c
a

0







(U
o



(0


aa)







- .
*4-4
6F
0r
Er1








69


C(












IL

z
-t-


LU
LU













JISNM INI




















>4



a)





4*


0


r.



A-












0)4
-4
*r







0)






41








0 C






ow
c)





0)0
O 0-











w .
4rl I















040)
*i- *
4- a
ua
E)








C (
(0
(5
\or
4- (
0 0
a)Q
!~C
h 0
*p -
U 4-
(D n






























































































0!m s s
C CC C
C C C


C








Cc







C
'0















o ao
0E
0)





C,
C





4.







C







-
C o





















x









4
r:4










0




o
a,









4-)


o
0
ct


4











0


o

r--
43







0
a)




*r-
















.-4
f43
c0



a)
Ci,




a




u-'




















fa




































































o O G


73




o
0











S- I





--o





























I I I I I I o
'0 V.IM 01
(SU nIL










Afl o o
-
ss ~ -





^^-^ ^

~^ ^


tD i^^lr -l

(r^t ^ -^l





IL; ___51--

















0
4J
(N






0
c o






.. aw
e-,

















r0
*4



SON





03
*

































0U
M <





















V.
,d i
Z)C
CO0






\ tP
"dS


Q)

oa












e -
cne







SC N

h Q-l












O
O







AI


C



F--
C


7
Li
O















O
o


A_LISNM N I o






















0
0

x

M
4
0








0
















41
0
0


0












*p
























I
4,
0
in










0;
0p









.r4

r4
iy

ki

14-
o,

vl



U,
Q)

I-
*r












77





o
o












o
'r




E0



0


y sO
Q

























I I I I I oo
S C 0 0 0 o 0




AJLISNILNNI
^^OO Q Q Q O Q O
--QO O O O Q Q
O O Q Q Q Q Q Q





















H

*>4








>14



>4
0,
I-I
















r4
r-














0
0
-4
4^








0
C





















o
Ore























a)
1 0)

0
C







4 3;
C M
7(fl(

&a























































































C C 0 C 0 0
0 0 C C 0 0
oo~ f f r o r


AlisuaiuI


O
C














o *
o O


0
'-
- 0
C
















-

o(3











apparently due to impurities (I) (a spectrum of its peak

maximum is shown in Figure 4-11), as well as ones that are

a result of pyrene (Py) and ones that are a result of

interactions among the pyrene molecules (Py(A)).

Apparently, the interactions between like molecules are

stronger for pyrene than they were for phenanthrene since

the aggregation peak (Py(A)) is much larger than the

Shpol'skii peaks for pyrene (Py).

Other PAHs investigated included fluorene (Figure 4-12

and 4-13), chrysene (Figure 4-14 and 4-15), and triphenylene

(Figure 4-16 and 4-17). Although an impurity is seen for

fluorene (I), no spectral peaks are present for these PAHs

which are not present in their spectra of manual injections.

Evidently, these PAHs are not as sensitive to the hexane/PAH

ratio as pyrene and phenanthrene are. This is especially

true for triphenylene, since its major peak is extremely

narrow, even at 82 gg/mL (approximately 6.2 nm wide at half-

height). A 3-D chromatogram of a mixture of approximately

100 Ag/mL of pyrene, 1,2- and 2,3-benzofluorene, chrysene,

and triphenylene is shown in Figure 4-18. Fluorene and

phenanthrene are not shown since phenanthrene obscures

pyrene and the benzofluorene isomers. The benzofluorene

isomers were barely baseline resolved using the flame

ionization detector, and with the added chromatographic

broadening, are not resolved using the LTMLS system.




























x

0





0
x







(a

(0
tp
0


0

u

4-J
.r4




41
(0











0
>4
.!






0
0
U





















,-4
0






04
-n
91


>1

cec

















0)
ii
tT
CP
*r-l
































O
-0



0



E









0 0
-o

-Q -

















I I I I I I I o


0 0 0 0 0 0 0 0
o o o on o o

AJ-ISNaJ-NI






















*-l


0
o




H



0.
-v-I

4-
-v-I O


co

4-HO
04co


*S


0
0














4
-H
C






Q)













0

ON











I



r-1 0
*0 0
00 (U
4-1 *r-1
0 -P

E












I
n *'










b 4-1












0
O









E

r-



Li
t-

II






0







_LISN IiNIN
























x
r4



a0





(o
*-i


,
tp

41

0

(0












-4
0












C4
0










4-4
cz






r-








0)

0
E










41





r4

rz
*3










86




o






-Q








o







-/ Q






















IISHHNINI
_a= a
QQr
QQ@'
















,-I


0
41



rl




0
4J
C1





r.-





r-l


uV



4r-




















0
a)




















oC



(0
*p












0





a,
c
(0


cu







c,
Ua




c^