MINIMUM”NW!NHINN\leIHLIIHI‘HHIWWM

   

‘z («9 -
.4 .1. ‘
I 1 Jim -'-~ ‘

LIBRARY
I Michigan State ;
University

l\/
C: IV
F9

" ' 1—.—

This is to certify that the
thesis entitled

SEPARATION AND QUANTITATION OF NITRATED
EXPLOSIVES USING THIN-LAYER CHROMATOGRAPHY
AND CHARGE-COUPLED DEVICE CAMERA IMAGING

presented by

Melissa Sue Meaney

has been accepted towards fulfillment
of the requirements for the

M .8. degree in Criminal Justice

 

 

 

Major PEfessTofis Signature

cafe/5&2 ‘30! 1970'?

 

Date

MSU is an affinnative-action, equal-opportunity employer

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:lClRC/DateDue.indd-p.1

 

 

 

SEPARATION AND QUANTITATION OF NITRATED EXPLOSIVES USING
THIN-LAYER CHROMATOGRAPHY AND CHARGE-COUPLED DEVICE
CAMERA IMAGING
By

Melissa Sue Meaney

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF SCIENCE
School Of Criminal Justice

2007

ABSTRACT
SEPARATION AND QUANTITATION OF NITRATED EXPLOSIVES USING
THIN-LAYER CHROMATOGRAPHY AND CHARGE-COUPLED DEVICE
CAMERA IMAGING
By
Melissa Sue Meaney

In recent years, the analysis Of nitrated explosives has become
increasingly important in the forensic community. In cases of national security or
on the front lines of international conflicts, sophisticated technologies for
explosives detection are available and widely used. For less prominent cases,
however, state and local law enforcement and forensic laboratories are not
equipped with such state-of-the-art, field-ready equipment. One of the most
common preliminary screening techniques used in these situations is thin-layer
chromatography (TLC) with spray reagents, followed by identification by visual
inspection. This project has sought to improve the TLC separation of nitrated
explosives using less toxic solvents to decrease safety concerns while
maintaining the inherent simplicity. An analysis procedure was identified to
separate an unknown explosive sample containing up to 12 explosives in two
steps. In addition, fluorescence imaging has been utilized for quantitation of
nitrated explosives and degradation products following TLC separation.
Explosives were quantitated over three orders Of magnitude with practical
detection limits in the low microgram range. This method is appealing because it

can be performed with commonly-used equipment such as a handheld UV light

source and a digital camera.

TO My Friends and Family, and TO Andrew, with Love

ACKNOWLEDGEMENTS

For my start in forensic science, I must thank my father. Without his
fascination with criminal dramas and Court TV in the late 19903 (before they
were popular), I would have never been inspired toward this great career path.
From there, I thank Dr. Jay Siegel for accepting me into the prestigious forensic
science program at Michigan State University, and staying with me through this
process although his own career took him to Indiana University-Purdue
University Indianapolis. He has been a great mentor and my best resource for all
knowledge that is forensic science. It was also Jay who encouraged me to
pursue a Ph.D. in Chemistry, and without his unbelievable insight, I would not be
where I am today.

I must also give many thanks to my other advisors at Michigan State
University, Dr. Victoria McGuffin and Dr. Ruth Waddell Smith, for their continued
support. Vicki helped guide this research project from beginning to end, and
insisted that it continue despite many setbacks. Her knowledge Of separation
science and spectroscopy, In addition to a vast amount of practical expertise, are
directly reflected in the content and structure of this work. Although Ruth arrived
after much of this research was completed, she was willing to handle much Of the
editing aspects of the thesis. Ruth was also invaluable in providing feedback and
suggestions to improve both my research and my writing. I was also blessed
with two excellent women role models, both Of whom are incredibly successful
and more than willing tO share their knowledge and experience. I would also like

to thank Dr. Mahesh Nalla for his willingness to provide feedback as a member of

my thesis committee. In addition, I would like to thank Dr. David Foran and Chris
Bommarito of the Michigan State Police for teaching me so much about forensic
biology and practical forensic chemistry, respectively.

I was fortunate enough to work with Jenny Borowitz, an amazing
undergraduate student, on this work. Jenny worked diligently for nearly a year to
complete many of the separations and acquire much of the data that is presented
in this work. As a senior B.S. Chemistry major, she managed to fit
undergraduate research into her already overloaded schedule. Jenny’s ideas
and creativity helped us overcome a number of hurdles during this research
project, and her enthusiasm and dedication are what made this project so
enjoyable. I am greatly appreciative Of her hard work and truly indebted to her
for the time she spent working with me. I also want to thank my good friend
Amber Hupp for her help with the final parts of this research project by preparing
blind samples for the validation of the methods.

Lastly, I must thank my friends and family who have supported me through
my time in graduate school. Thanks to Elizabeth, Carl, and Audrey for sharing in
the experience of a dual degree program and all of the headaches that entails.
Thanks to Sam, Carl, Xiaoping, Rashad, Amber, Kahsay, Heidi, Lucas, John, and
Melissa for being the best group members a girl could ask for. Thanks to
Elizabeth, Amber, Audrey, Anne, Jason, Alli, Lucas, Dan, and Lisa for being
great friends, and knowing when it’s about work and when it’s not. Thanks to
Lucas, Lisa, and Sarah for being a fun bunch and inviting me to Tuesday lunch.

Thanks to Erin, Sarah, Kristin, and Lisa, my friends outside of chemistry and

forensic science, for keeping me grounded. Thanks to my softball team for
keeping me active and sharing lots Of good memories on and Off the field.
Thanks to Mom, Dad, Melinda, Jerrod, Matt, Stephanie, Gerri, Don, Geoff, Donna,
and Kevin for their willingness to accept my absence at many family events and
for their unquestionable joy when l was in attendance. A special thank you to my
parents for always believing in me and encouraging me to pursue my goals and
dreams. Thanks to my nieces and nephews who just love their Aunt Mel. Lastly,
thanks to Andrew, for being by my side unconditionally through this process that
you don’t understand. We have been on an amazing journey through these past

few years, and I can’t wait for it to continue as I spend the rest of my life with you.

vi

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................... ix
LIST OF FIGURES .............................................................................................. xi
KEY TO ABBREVIATIONS ............................................................................... xiv
CHAPTER 1. INTRODUCTION ........................................................................... 1
1 .1 Explosives .................................................................................................. 2
1.2 Explosives as Evidence ............................................................................. 8
1.3 Established Methodologies ........................................................................ 9
1.4 Thin-Layer Chromatographic Analysis of Explosives ............................... 11
1.5 Detection of Explosives Following Thin-Layer Chromatography .............. 12
1.6 Research Objectives ................................................................................ 15
1 .7 References .............................................................................................. 16
CHAPTER 2. SEPARATION OF NITRATED EXPLOSIVES ............................ 21
2.1 Experimental Methods ............................................................................. 21
2.1.1 Chemicals ................................................................................................ 21
2.1.2 Plate Preparation ..................................................................................... 22
2.1.3 Plate Development .................................................................................. 23
2.2 Theory of Adsorption Chromatography .................................................... 24
2.3 Separation Of TNT and Degradation Products ......................................... 30
2.4 Separation of Nitroaromatic Explosives ................................................... 35
2.5 Separation of Nitroaromatic and Nitrate Ester Explosives ....................... 36
2.6 Separation of Nitramine Explosives ......................................................... 39
2.7 Precision Of Retention Factors ................................................................. 40
2.8 Separation of an Unknown Explosive Sample ......................................... 43
2. 9 Conclusions ............................................................................................. 46
2.10 References .............................................................................................. 47
CHAPTER 3. QUANTITATION OF NITRATED EXPLOSIVES ......................... 48
3.1 Positioning of Imaging System Components ........................................... 48
3.2 Optimization of Image Acquisition and Manipulation ............................... 50
3.3 Correction Using a Background Plate ...................................................... 57
3.4 Theory and Data Analysis ........................................................................ 59
3.5 Quantitation Of Degradation Products ...................................................... 63
3.6 Quantitation of Nitroaromatic Explosives ................................................. 72
3.7 Quantitation of Nitramine Explosives ....................................................... 76
3.8 Quantitation of Nitrate Ester Explosives ................................................... 79
3.9 Quantitation Of an Unknown Explosive Sample ....................................... 82
3.10 Additional Observations ........................................................................... 85

vii

3.11 Conclusions ............................................................................................. 86
3.12 References .............................................................................................. 87

CHAPTER 4. CONCLUSIONS AND FUTURE DIRECTIONS ........................... 88

viii

Table 1.1

Table 1.2

Table 2.1

Table 2.2
Table 2.3

Table 2.4

Table 2.5

Table 2.6

Table 2.7

Table 2.8

Table 2.9

Table 2.10

Table 3.1

Table 3.2

Table 3.3

Table 3.4

Table 3.5

LIST OF TABLES

Explosives and degradation products used to validate TLC
separation and CCD imaging system. ............................................. 3

Common military explosives and formulations. ............................... 7

Group adsorption energies for substituents common to nitrated
explosives and degradation products. ........................................... 26

Net adsorption energies for explosives and degradation products 28
Properties Of common TLC mobile phases ................................... 31

The precision of retention factors for nitroaromatic explosives and

degradation products in 88:12 hexanezTHF. ................................. 34
The precision of retention factors for nitroaromatic explosives in
70:30 cyclohexanezMEK. .............................................................. 37
The precision of retention factors for nitroaromatic and nitrate ester
explosives in 88:12 hexanezTHF. .................................................. 38
The precision of retention factors for nitramine explosives in 50:50
hexanezTHF. ................................................................................. 41
The precision of retention factors for nitramine explosives and picric
acid in 65:35 hexanezMEK ............................................................ 42
Analysis Of unknown explosive sample 1. ..................................... 44
Analysis of unknown explosive sample 2. ..................................... 45

Comparison of image quality for various gain settings Obtained with
an integration time of 500 ms ........................................................ 54

Comparison of image quality for various integration times. ........... 56

Comparison of image quality for summing and averaging pixels in
different size selections of the image ............................................ 58

Detection figures of merit for nitrated explosives and degradation
products using the semi-logarithmic function (Equation 3.4). ........ 69

Detection figures of merit for nitrated explosives and degradation
products using the logarithmic function (Equation 3.5) .................. 70

Table 3.6 Analysis of unknown explosive samples. ...................................... 84

Figure 1.1

Figure 1.2.

Figure 2.1

Figure 2.2

Figure 3.1.

Figure 3.2

Figure 3.3

Figure 3.4.

LIST OF FIGURES

Common nitroaromatic explosives: (1) TNT, (2) picric acid, and (3)
Tetryl. Nitramine explosives: (3) Tetryl, (4) RDX, and (5) HMX.
Nitrate ester explosives: (6) NG and (7) PETN. .............................. 5

Some common degradation products Of TNT used in this work: (8)
NB, (9) 2-NT, (10) 2,4-DNT, (11) 2,6-DNT, and (12) 4-am-DNT. 6

Sample plate containing (I) a reference sample and (II) an unknown
sample. Designated plate areas include (a) the original location
and a completely retained solute (Rf = O), (b) the solvent front, (c,e)
a solute (Rf(C) = 0.21, Rr(e) = 0.66) that is detected in the reference
sample but undetected in the unknown sample, (d) a solute (Rf =
0.56) that is detected in both the reference and unknown samples
at similar concentrations, and (f) a solute (Rf = 0.87) that is

detected at a higher concentration in the reference sample than in
the unknown sample ..................................................................... 25

Effect of percent THF in hexane on the retention factors measured
for TNT (I) and degradation products, including NB (O), 2-NT (O),
2,4-DNT (O), 2,6-DNT (III), and 4-am-2,6—DNT (O). .................... 33

Diagram of TLC plate imaging setup ............................................. 49

Fluorescence of a TLC plate at various times after turning on the
UV lamp: (a) 5 min, (b) 15 min, (c) 25 min, (d) 35 min, (e) 45 min,
(f) 55 min, (9) 65 min. .................................................................... 51

Simulated images and chromatograms for (1) a background plate
and (2) a sample plate containing (I) a reference sample and (II) an
unknown sample. Designated plate areas include (a) the original
location and a completely retained solute, (b) the binning area
around a strongly retained solute that is undetected in the unknown
sample and the corresponding chromatogram, (c) the binning area
around a weakly retained solute detected in both samples but at
different concentrations and the corresponding chromatogram, and
(d) the solvent front. The corrected intensity is calculated by using
Equation 3.1. ................................................................................. 53

F it Of TNT to the symmetric double-sigmoidal function (Equation

3.3). (A) Corrected intensity for 9.8 pg TNT. (B) Symmetric
double-sigmoidal fit of 9.8 pg TNT data (— — —), R2 = 0.937 ..... 61

xi

Figure 3.5

Figure 3.6

Figure 3.7

Figure 3.8

Figure 3.9

Figure 3.10

Figure 3.11

Figure 3.12

Figure 3.13

Figure 3.14

Figure 3.15

Calibration curves for NB based on peak area using (A) the semi-
Iogarithmic fit described in Equation 3.4 and (B) the logarithmic fit
described in Equation 3.5. ............................................................. 64

Calibration curves for 2-NT based on peak area using (A) the semi-
Iogarithmic fit described in Equation 3.4 and (B) the logarithmic fit
described in Equation 3.5 .............................................................. 65

Calibration curves for 2,4-DNT based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the
logarithmic fit described in Equation 3.5 ........................................ 66

Calibration curves for 2,6-DNT based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the
logarithmic fit described in Equation 3.5 ........................................ 67

Calibration curves for 4-am-DNT based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the
logarithmic fit described in Equation 3.5 ........................................ 68

Calibration curves for TNT based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit
described in Equation 3.5. ............................................................. 73

Calibration curves for tetryl based on peak area using (A) the semi—
Iogarithmic fit described in Equation 3.4 and (B) the logarithmic fit
described in Equation 3.5 .............................................................. 74

Calibration curves for picric acid based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the
logarithmic fit described in Equation 3.5 ........................................ 75

Calibration curves for RDX based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit
described in Equation 3.5 .............................................................. 77

Calibration curves for HMX based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit
described in Equation 3.5. ............................................................. 78

Calibration curves for NG based on peak area using (A) the semi-

logarithmic fit described in Equation 3.4 and (B) the logarithmic fit
described in Equation 3.5 .............................................................. 80

xii

Figure 3.16 Calibration curves for PETN based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the
logarithmic fit described in Equation 3.5 ........................................ 81

xiii

2,4-DNT
2,6-DNT
2-NT
4-am-DNT
CCD
DPA
GSR
HMX
MEK

NB

NC

ng
PETN
Picric acid
RDX
RSD
S/N

S/B
Tetryl
THF
TLC
TNT

UV

H9

KEY TO ABBREVIATIONS

2,4-Dinitrotoluene
2,6-Dinitrotoluene

2-Nitrotoluene
4-aminO-2,6-Dinitrotoluene
Charge-coupled device
Diphenylamine

Gunshot residue

1 ,3,5,7-tetranitrO-1 ,3,5,7-tetraazocyclooctane
Methylethylketone

Nitrobenzene

Nitroglycerin

Nanogram (10'9 g)
Pentaerythritol tetranitrate
2,4,6-trinitrophenol

1 ,3,5-trinitro-1 ,3,5-triazocyclohexane
Relative standard deviation
Signal-tO-noise
Signal-to-background
2,4,6,N-tetranitrO-N-methylaniline
Tetrahydrofuran

Thin-Layer Chromatography
2,4,6-Trinitrotoluene

Ultraviolet

Microgram (1043 g)

xiv

CHAPTER 1. INTRODUCTION

In recent years, the determination Of nitrated explosives has become
increasingly important in the forensic community. In cases of national security or
on the front lines Of international conflicts, sophisticated technologies for
explosives detection are available and widely used. For less prominent cases,
however, state and local law enforcement and forensic laboratories are not
equipped with state-of—the-art equipment such as handheld explosives sensors or
dedicated instrumentation for explosives detection. One of the most common
preliminary screening techniques used in these situations is thin-layer
chromatography (TLC) with spray reagents, followed by visual inspection.“3 This
project has sought to improve the TLC separation Of nitrated explosives using
less toxic mobile phases to decrease safety concerns while maintaining the
inherent simplicity. In addition, a fluorescence quenching method has been
utilized for quantitation of nitrated explosives and degradation products in a
single step following TLC separation. This method Offers several advantages
over colorimetric methods for the detection of nitrated explosives following TLC,
and can be performed with commonly-used equipment such as a handheld UV
light source and a digital camera. This method provides qualitative and
quantitative information simultaneously, and the sample can be preserved on the
TLC plate and extracted from the stationary phase for further analysis in its

original form.

1.1 Explosives

Nitrated explosives are divided into groups according to their use, where
high explosives are described as detonating charges and low explosives as
propellants.1 The distinction arises because high explosives detonate at
velocities in km/s, while propellants burn instead of detonating at velocities of
cm/s to m/s.2'4 Low explosives are used to initiate a more powerful explosive
reaction by transferring energy to the main explosive charge. Because the
burning rate of low explosives increases rapidly with pressure, they are also
found as propellants in ammunition cartridges.2 Two common low explosives are
black powder (nitrate salt, charcoal, and Often sulfur) and smokeless powder
(nitrocellulose and possibly nitroglycerin and nitroguanidine).1'4'5 High explosives
or detonating charges are usually shock initiated, and are divided further to
reflect their chemical composition as either organic or inorganic. Inorganic
explosives are common in homemade explosive devices, as ammonia—based
explosives can be easily formulated from fertilizers?4 For example, the devices
used in the 1993 World Trade Center bombing and the 1995 Oklahoma City
bombing were created from ammonia-based fertilizers.6'7 Ammonium nitrate is
one example of a common inorganic explosive.4

Organic nitrated explosives are Of greatest interest in this project because
these materials are very common and comprise a majority of military and
industrial explosives.2'4 Organic explosives are divided into three classes based
on the atom to which the nitro group (N02) is attached. Common nomenclature

and abbreviations for the explosives of interest are given in Table 1.1, and

 

 

Azoémé 6:85 cozmnmfia a: 88.9055u-m.~-oc_E<-¢
R293 6805 cozmuemmo c: ocm:_eos_c_o-o.m
Azoéa 6:85 cofiuemoo a: mcmaeeceaéa
._.z-w 8:85 cozmumcmoo 3v mcoaoaoEzN
mz 6:an cozmnmcooo 8V ocm~cono£z
25a .28 9.32 E 29:55 223595.“.
02 .98 99:2 6V E82852
x2... mc_E§_z 3v 2982038859;Rm. TELEEGEGB. F
xom msEmEz g m:mxm:o.o>om~m§-m.m. 355.35. P
Ea; mEEEzBfiEEmoEz E as:cm_>£mE-z-os_cmsohzdim
new GEE oszemegz E acacaegsfieéd
ER 839822 E mamaepsefiaeN
cow—25.5.: mama ozmoixm uczanoo

 

€296 mEmmé 000 new 8:983 04... 2mg? 2 now: 90:35 cozmumcmon new mozwoaxm ._.._. DEG...

structures are given in Figure 1.1. Nitroaromatic explosives contain nitro groups
attached to aromatic carbon atoms and include explosives such as TNT and
picric acid. Nitramine explosives contain nitro groups attached to nitrogen atoms
and include RDX and HMX. Tetryl is a unique explosive compound in that it has
both nitroaromatic and nitramine character (Figure 1.1). Nitrate ester explosives
contain nitro groups attached to oxygen atoms and include NG and PETN. In
many cases, particularly when investigating aged explosive materials or those
that have been detonated, degradation products will also be identified. Many
degradation products are also synthetic by—products, and may be found in even
recently manufactured explosive materials. Because TNT is the most commonly

encountered aromatic explosive}4

its degradation products may therefore be
important in a forensic investigation. Structures of some common nitroaromatic
degradation products Of TNT used in this work are given in Figure 1.2.

Military explosives Often consist of combinations Of various organic and
inorganic explosives, as shown by the examples listed in Table 1.2.1 The relative
percentage of individual components in each mixture varies slightly between
manufacturers and also between batches from the same manufacturer. In
addition to variation in composition, inert binders such as plasticizers, putties, or
rubbers are added to military explosives to increase mechanical strength,
stability, and lifetime of the materials.4 In their crystalline forms, many explosives
are not easily pressed or are highly shock sensitive. The addition of a binder to

manufactured materials allows the explosive to be formed into any size or shape,

as well as prevent accidental detonation through handling and transport.

02N\ /

OH N
OZN N02 OZN NO2 OZN N02
1 N02 2 N02 3 N02
OZN\
N
O N NO
2 \NAN/ 2 ( \N/NO2
I\ J /N J
T OzN \ N
4 N02 5 N02
N02
NO
0/ O/ 2
/O
O O 7) O/NOZ
6 N02 OZN 7

Figure 1.1. Common nitroaromatic explosives: (1) TNT, (2) picric acid,
and (3) Tetryl. Nitramine explosives: (3) Tetryl, (4) RDX, and (5) HMX.
Nitrate ester explosives: (6) NG and (7) PETN.

NO2

NH2

Figure 1.2. Some common degradation products Of TNT used in this work: (8)
NB, (9) 2-NT, (10) 2,4-DNT, (11) 2,6-DNT, and (12) 4-am-DNT.

 

33 + EC .058
z._.mn_ + R2... + xom N-x._.a
Ex; + EA + xom 3E
._.z.r + 2.55 DEERE
#2... + x2: 680
Louozmma + z._..uE “$593
EA + xom .2226
__o 62.. + £8mnmmc>xm£>£m-mv_v + ocm_>§nom_>.oa + xom v.0 :oEmanoo

822.822 + Em: + E + E? E» + xom
322322.536 + $22.80;: + E + 59. EA + xom

m6 5:68:80
«.0 :oEwanoo

 

xmz, + E2... + xom m 5:68:50
29:... Eacoeem + RE. _2mE<
225.250". 252 coEEoo

 

hchzmSELOL new 828.98 meE coEEoo .N... e32.

Homemade explosive devices are often packaged in metal or glass to
increase detonation power of the explosive or to increase the damage caused by
added shrapnel.3 All explosives, legally manufactured or illicitly assembled, may
be housed in a variety Of casings, for the purpose of confinement, transport, or

concealment.3

1.2 Explosives as Evidence

Explosives are an important class Of forensic evidence, and their detection
is instrumental in many criminal investigations. Undoubtedly, the most common
type Of explosive encountered in the forensic laboratory is smokeless powder in
the form of gunshot residue (GSR). In 2003, the US Bureau Of Justice Statistics
reports over 366,000 victims of firearm-related crimes,8 including over 10,000
homicide victims.9 Comparatively, less than 400 bombings were reported by the
US Bureau of Alcohol, Tobacco, Firearms, and Explosives in that same year.10
Common analyses involve identification Of GSR on the hands or clothing of a
suspect indicating that a gun was recently fired, as well as identification of GSR
on a target to identify bullet holes.2

ln transportation security, passengers and baggage are screened for the
presence Of explosive residues prior to commercial travel. Many federal and
state government buildings also require similar screening of visitors and
employees. These preliminary screening methods are used to identify low levels
of explosive residues present on a person or Object, which may be critical to the
prevention of a dangerous and life-threatening incident. In the case of a

successful bombing or explosion, analysis of post-blast residues can provide

information about the type and source of the explosive used. In both cases,
rapid and accurate determination of trace explosive materials is a critical step in
the investigative process. In the absence of explosive material, the presence of
degradation products may indicate the type or manufacturer Of the explosives
that was used. In many cases, degradation products have higher vapor
pressures than related explosive compounds, increasing their vapor

concentration and thus their ability to be detected.

1.3 Established Methodologies

Forensic detection of explosives and explosives residues is multi-faceted.
In the rare case that an explosive device is discovered intact, the analysis
involves a simple confirmation Of the identity of the explosive. As a preventative
security measure, explosive residues, particles, and vapors must be identified
against a complex environmental background, demanding increased selectivity.
In a third scenario, explosive compounds must be identified in complex post-blast
residues containing ambient air, soot, or ash. Other matrix components may
include degradation products that form during production, storage, or the
explosion itself. In many instances, a separation method is often used to isolate
explosive materials from the complex matrix within which they are contained.
Many methods have been used for such separations, including TLC,"5 gas

1,5,11-15

chromatography (GC), supercritical fluid chromatography (SFC),1'16 liquid

chromatography (LC),1'5'“'17'20 ion chromatography (IC),1'21 capillary

22,23

electrophoresis (CE), capillary electrochromatography (CEC),24 and micellar

electrokinetic chromatography (MEKC).25'26

In addition to separation of explosives and residues from the surrounding
environment, a sensitive and selective detection method is also important. To
this end, detection and characterization of nitrated explosives in the laboratory

have been accomplished by colorimetric reactions,1'5'27 densitometry,"5 UV

1 ,5,17-20,28 5.29.30

absorbence, infrared spectroscopy (IR), nuclear magnetic

1.5.20 1,5,23,31-33

resonance (NMR),1'5 electrochemistry, immunoassay, mass

spectrometry (II/IS).1’3'"36 ion mobility spectrometry (IMS),"3“‘36

13-15,37—4O -

chemiluminescence, indirect laser-induced fluorescence (ILIF),2“'26'41 and

fluorescence quenching.7"8"‘2'50

Unfortunately, many of these methods are
difficult to adapt for field use owing to the size of laboratory-scale instrumentation
and the need for other accessories, such as large magnets for NMR or vacuum
systems for MS, that limit portability.

As mentioned previously, in cases Of national security or on the front lines
Of international conflicts, sophisticated field technologies for explosives detection
are available and widely used. The most common methods for on-site
explosives detection involve trained canines,51 ion mobility spectrometry,34
chemiluminescence,38 or fluorescence quenching Of amplified polymers.52
Detection systems based on these methods are often costly, and for smaller, less
frequent investigations, impractical. Despite the success Of more sophisticated
separation techniques in a laboratory setting, TLC continues to be widely used in

the routine analysis of explosives because it is more rapid, portable, and

inexpensive than more complex methods.

10

1.4 Thin-Layer Chromatographic Analysis of Explosives

Thin-layer chromatography, or TLC, is a separation method that involves a
thin, even layer of sorbent material deposited on a firm support, such as a glass
or plastic plate.53 The sorbent material, or stationary phase, consists of small
silica or alumina particles and is typically deposited in a layer that is 0.10 to 0.25
mm in thickness. The compounds, or solutes, to be separated are deposited
from solution near one edge of the plate. The plate is then placed into a small
volume of mobile phase, which carries the solutes through the stationary phase
by capillary action. Separation of the solutes is based on their affinity for the
stationary phase relative to that Of the mobile phase. Highly retained solutes will
remain near their original location, while less retained solutes will travel with the
mobile phase up the plate. The identity Of a solute is determined from a
comparison of the distance traveled to that of a standard material on the same
plate.

Commonly, separation of explosives by TLC involves the use Of toxic
and/or carcinogenic mobile phases.“5 Chloroform, carbon tetrachloride,
methylene chloride, and toluene are Often used, either individually or in
combination with other mobile phases, because they maximize resolution Of the
solutes in the shortest development time.5 Despite their toxicity, these mobile
phases have been routinely used in a variety of combinations to separate all
classes of explosives.2'5“'59 These mobile phases pose numerous health risks
and Often require use of a fume hood to minimize exposure to harmful vapors.60

The need for a fume hood as well as danger Of accidental mobile phase spills

11

limits the practicality of these mobile phases for field analysis. Identification of a
non-toxic alternative would be an improvement in safety and convenience in the

forensic science community.

1.5 Detection of Explosives Following Thin-Layer Chromatography

Detection following TLC separation is typically accomplished by using a
colorimetric reaction for qualitative analysis and densitometry for quantitative
analysis}5 Colorimetric reactions are conducted on the TLC plate by spraying a
reagent mixture onto the plate and Often accelerating the reaction through
application of a UV light. For nitrated explosives, the Griess and diphenylamine
(DPA) tests are most commonly employed.1’5'61 In the Griess test, nitrite ions
(NOz') are first liberated from nitrate-containing explosives by reaction with base.
Acidic sulfanilamide reacts with nitrites to form a diazonium salt, which in turn
reacts with naphthylamine to form a pink azo dye. Visual detection limits of 5-20
ng for tetryl, RDX, HMX, NG, and PETN are reported using the Griess spray test
following TLC.54 In the DPA test, acidic DPA reacts with nitrites, nitrates, and
other oxidizing agents to form a blue compound. Detection limits of 1 pg are
reported for inorganic nitrites following TLC using the DPA color test.1 The
relative lack Of specificity Of the DPA test, however, makes the Griess test more
favorable and therefore more commonly used.

Although spot tests can indicate the presence or absence of explosives,
visual inspection of the colored product on the TLC plate does not provide
accurate quantitation Of the compounds of interest. For explosives analysis,

quantitation and knowledge of method detection limits are necessary for

12

evaluating utility of a technique. By measurement Of the absorbance of a
derivatized solute zone at a given wavelength, the concentration can be
approximated by using Beer’s Law.53 Because the stationary phase is opaque,
most densitometers operate in reflectance mode, measuring the amount Of light
reflected from the surface Of the plate. In contrast, a typical absorbance
measurement involves quantification of the amount of light transmitted through
the sample. Densitometry detection limits of 0.1 pg TNT/sample in human
urine62 and 2 pg RDX/sample in manufactured explosives63 have been reported.
Linear dynamic ranges for densitometry have been reported as 100 - 1000 pg
TNT/sample in urine,62 1-10 pg RDX/samples?"64 and 1-15 pg PETN/sample in
standard materials,64 respectively.

Although coupling of derivatization and densitometry is sufficient in many
cases, the colorimetric reagent alters the chemical composition of the nitrated
explosive, prohibiting further analysis in its original form. Many colorimetric
reagents require 10-60 minutes of exposure to UV light and all require a second
step using densitometry for accurate quantitation. In addition, inhalation of or
skin exposure to DPA or Griess reagents is known to cause organ irritation and
damage.”67 Consequently, development of a non-destructive, rapid, and simple
quantitation method would greatly improve the current status Of nitrated explosive
analysis by TLC.

An alternative method to derivatization and densitometry for detection Of
nitrated explosives is the use Of trivial fluorescence quenching. Many TLC

stationary phases contain an embedded UV-active fluorophore such as

13

manganese-activated zinc (~2°/o Zn, ~0.5% Mn). A solute that absorbs either
excitation or emission light will appear as a dark spot against the fluorescent
background of the plate. The dark spots are an example of trivial fluorescence
quenching, arising from primary or secondary filtering of the fluorescence by a
quencher molecule.68 Primary filtering results from absorption of the excitation
light from the source by another molecule before reaching the fluorophore.
Secondary filtering results from absorption of the emitted light from the
fluorophore by another molecule before reaching the detector.68 In both cases,
trivial quenching is related to absorptive processes and the Observed decrease in
fluorescence power can be related logarithmically to the concentration or mass of
the quencher through a form of Beer’s Law.

Although densitometers are capable of quantitating fluorescence
quenching as well as absorbance or reflectance, a laboratory must purchase a
densitometer and the relevant light source for such analyses. An alternative
method involves imaging of the TLC plate fluorescence under a handheld UV
lamp by using a digital camera. Photographic software can be used to quantitate
the amount of light reaching the camera at each pixel location. Simon and
coworkers demonstrated this method previously for the quantitative analysis of
several organic pharmaceuticals, including acetaminophen, caffeine, and
acetylsalicylic acid, following TLC.69 The linear range for this method extended
from the detection limit to 450 ng, with detection limits of 4.2 ng, 12.5 ng, and
71.1 ng for acetaminophen, caffeine, and acetylsalicylic acid, respectively. For

on-site detection of explosives, this method is favorable because the sample is

14

retained in its original form and can be extracted from the stationary phase for
further analyses. In addition, handheld UV lamps and digital cameras are
standard equipment in all forensic laboratories, making this method easy tO
implement with existing technology. Furthermore, the use of this method does
not preclude the use of colorimetric reagents, because the presence Of the

fiuorophore on the TLC plate does not alter the results of the Griess or DPA tests.

1.6 Research Objectives

The main Objective Of this research project was tO identify safe, simple,
and cost-effective improvements for the determination Of nitrated explosives by
using TLC. The first goal was to identify and characterize a less-toxic mobile
phase system that would allow resolution Of nitrated explosives in the fewest
possible steps. The second goal was to optimize the CCD camera imaging
system to ensure that the highest quality data could be extracted from TLC plate
images. The final goal was to quantitate various nitrated explosives and
degradation products to determine the analytical figures of merit for this detection
method. Through the work presented here, the forensic science community will
be able to easily adapt current technology to improve their analysis of forensic

explosive evidence.

15

1.7

10.

11.

12.

13.

References

Yinon, J.; Zitrin, 8. Modern Methods and Applications in Analysis Of
Explosives; Wiley: Chichester, West Sussex, 1993.

Midkiff, C. Arson and Explosive Investigation; Prentice-Hall: Englewood
Cliffs, NJ, 1982.

Brodie, T. G. Bombs and Bombings: A Handbook to Detection, Disposal,
and Investigation for Police and Fire Departments, 2nd ed.; Charles C.
Thomas: Springfield, IL, 1996.

Cooper, P. W.; Kurowski, S. R. Introduction to the Technology of
Explosives; VCH Publishers, Inc.: New York, NY, 1996.

Yinon, J.; Zitrin, S. The Analysis of Explosives; Pergamon: Elmsford, NY,
1981.

Guterl, F. The Chemistry of Mass Murder - Homemade Fertilizer Bomb
Used in Oklahoma City Bombing - 1995: The Year in Science - Brief
Article, Discover, January 1996.

Knight, J. In Encyclopedia of Espionage, Intelligence, and Secun'ty,Lemer,
K. L., Lerner, B. W. Eds. World Trade Center. 1993 Terrorist Att_a_clg; Gale
Group: Detroit, 2004.

Bureau Of Justice Statistics. National Crime Victimization Survey, 2006.
http://wwwolp.usti.qov/bIs/qlance/tables/firearmnonfataltab.htm
(accessed 10l02l07).

US Federal Bureau of Investigation. Supplementary Homicide Reports,
2006. http://www.in.usti.gOV/bis/homicide/tables/weaponstab.htm
(accessed 10/02/07).

US Bureau of Alcohol Tobacco Firearms and Explosives. Explosive
Incidents Report for Bombing - January 2003 to December 2003, 2004.
http:l/www.atf.qov/aexi52lbombinqrpts/bombinquOOB.pdf (accessed
10l02/07).

Oxley, J.; Smith, J.; Resende, E.; Pearce, E. J. Forensic Sci. 2003, 48,
742-753.

Brown, H.; Kirkbride, K.; Pigou, P.; Walker, G. J. Forensic Sci. 2004, 49,
215-221.

Crowson, C. A.; Cullum, H.; Hiley, R. W.; Lowe, A. J. Forensic Sci. 1996,
41, 980-989.

16

14.

15.

16.
17.

18.

19.

20.
21.

22.

23.
24.
25.
26.

27.
28.
29.

30.
31.

32.

Bowerbank, C. R.; Smith, P. A.; Fetterolf, D. D.; Lee, M. L. J. Chromatogr.
A 2000, 902, 413-419.

Batlle, R.; Carlsson, H.; TOIIback, P.; Colmsjo, A.; Crescenzi, C. Anal.
Chem. 2003, 75, 3137-3144.

Douse, J. M. F. J. Chromatogr. A 1988, 445, 244-250.

Gholamian, F.; Chaloosi, M.; Husain, S. W. Prop. Expl. Pyrotech. 2002,
27, 31-33.

Gates, P.; Furlong, E.; Dorsey, T.; Burkhardt, M. Trends Anal. Chem.
1996, 15, 319-325.

MonteiI-Rivera, F.; Paquet, L.; Deschamps, S.; Balakrishnan, V. K.;
Beaulieu, C.; Hawari, J. J. Chromatogr. A 2004, 1025, 125-132.

Spiegel, K.; Welsch, T. Fresenius J. Anal. Chem. 1997, 357, 333-337.

Reutter, D.; Buechele, R.; Rudolph, T. Anal. Chem. 1983, 55, 1468A-
1472A.

Casamento, S.; Kwok, B.; Roux, C.; Dawson, M.; Doble, P. J. Forensic
Sci. 2003, 48, 1075-1083.

Bromberg, A.; Mathies, R. A. Anal. Chem. 2003, 75, 1188-1195.
Bailey, C. G.; Wallenborg, S. R. Electrophoresis 2000, 21, 3081-3087.
Wallenborg, S. R.; Bailey, C. G. Anal. Chem. 2000, 72, 1872-1878.

Kennedy, S.; Caddy, B.; Douse, J. M. F. J. Chromatogr. A 1996, 726, 211-
222.

Stevanovic, S.; Mitrovic, M. Int. J. Environ. Anal. Chem. 1990, 40, 69-76.
Goodpaster, J. V.; McGuffin, V. L. Anal. Chem. 2001, 73, 2004-2011.

Pristera, F.; Halik, M.; Castelli, A.; Fredericks, W. Anal. Chem. 1960, 32,
495-508.

Conduit, C. P. J. Chem. Soc. 1959, 3273-3277.

Wilson, R.; Clavering, C.; Hutchinson, A. Anal. Chem. 2003, 75, 4244-
4249.

Narang, U.; Gauger, P. R.; Ligler, F. 8. Anal. Chem. 1997, 69, 1961-1964.

17

33.

34.
35.
36.

37.
38.

39.

40.
41.

42.
43.

44.
45.

46.

47.

48.
49.

50.

Charles, P. T.; Rangasammy, J. G.; Anderson, G. P.; Romanoski, T. C.;
Kusterbeck, A. W. Anal. Chim. Acta 2004, 525, 199-204.

Eiceman, G. A.; Stone, J. A. Anal. Chem. 2004, 76, 390A—397A.
Buryakov, I. A. J. Chromatogr. B 2004, 800, 75-82.

Miller, C. J.; Glenn, D. F.; Hartenstein, S. D. Proc. SPlE-lnt. Soc. Opt.
Eng. 1998, 3575, 335-341.

Jimenez, A. M.; Navas, M. J. J. Hazard. Mater. 2004, 106, 1-8.

Fine, D. H.; Wendel, G. J. Proc. SPIE-lnt. Soc. Opt. Eng. 1993, 2092, 131-
136.

Nguyen, D. H.; Locquiao, S.; Huynh, P.; Zhong, 0.; He, W.; Christensen,
D.; Zhang, L.; Bilkhu, B. In Electronic Noses & Sensors for the Detection

of Explosives; Gardner, J. W., Yinon, J., Eds.; Kluwer Academic
Publishers: Netherlands, 2004, pp 71-80.

Lafleur, A. L.; Mills, K. M. Anal. Chem. 1981, 53, 1202-1205.

Woltman, S. J.; Even, W. R.; Sahlin, E.; Weber, S. G. Anal. Chem. 2000,
72, 4928-4933.

Lu, J.; Zhang, Z. Anal. Chim. Acta 1996, 318, 175-179.

Focsaneanu, K.-S.; Scaiano, J. C. Photochem. Photobiol. Sci. 2005, 4,
817-821.

Albert, K. J.; Walt, D. R. Anal. Chem. 2000, 72, 1947-1955.

Kose, M. E.; Harruff, B. A.; Lin, Y.; Veca, L. M.; Lu, F.; Sun, Y. P. J. Phys.
Chem. B 2006, 110, 14032-14034.

Zhang, 8.; Lu, F.; GaO, L.; Ding, L.; Fang, Y. Langmuir 2007, 23, 1584-
1590.

ToaI, S. J.; Sanchez, J. C.; Dugan, R. E.; Trogler, W. C. J. Forensic Sci.
2007, 52, 79-83.

Yang, J. S.; Swager, T. M. J. Am. Chem. Soc. 1998, 120, 5321-5322.

Rose, A.; Zhu, Z.; Madigan, C. F .; Swager, T. M.; Bulovic, V. Nature 2005,
434, 876-879.

Naddo, T.; Che, Y.; Zhang, W.; Balakrishnan, K.; Yang, X.; Yen, M.; ZhaO,
J.; Moore, J. S.; Zang, L. J. Am. Chem. Soc. 2007, 129, 6978-6979.

18

51.
52.
53.

55.

56.

57.
58.
59.
60.
61.
62.

63.

64.

65.

66.

67.

68.

Furton, K. G.; Myers, L. J. Talanta 2001, 54, 487-500.
Wang, L. Chem. Eng. News 2007, 85, 11.

Wall, P. E. Thin-Layer Chromatography: A Modern Practical Approach;
The Royal Society Of Chemistry: Cambridge, UK, 2005.

Douse, J. M. F. J. Chromatogr. A 1982, 234, 415-425.

Kohlbeck, J. A.; Chandler, C. D.; BOlleter, W. T. J. Chromatogr. A 1970,
46, 173-179.

Chandler, C. D.; Kohlbeck, J. A.; BOlleter, W. T. J. Chromatogr. A 1972,
67, 255-259.

Hoffsommer, J. C.; McCullough, J. F. J. Chromatogr. A 1968, 38, 508-514.
Hoffsommer, J. C.; Glover, D. J. J. Chromatogr. A 1971, 62, 417-421.
Parihar, D.; Sharma, S.; Verma, K. J. Forensic Sci. 1968, 13, 246-252.
Klaassen, 0., 6th ed.; McGraw-Hill: New York, NY, 2001.

Jenkins, T.; Walsh, M. Talanta 1992, 39, 419-428.

Liu, Y.; Wei, W.; Wang, M.; Pu, Y.; Lin, L.; Zhang, J. J. Planar
Chromatogr. 1991, 4, 146-149.

Gholamian, F.; Chaloosi, M.; Waquif-Husain, S.; Dehghani, H. J. Planar
Chromatogr. 2001, 14, 296-299.

Bladek, J.; Miszczak, M.; Popiel, S. Chem. Anal. (Warsaw) 1995, 40, 723-
729.

Diphenylamine, MSDS NO D7728. Mallinckrodt Baker: Phillipsburg, NJ,
February 26, 2007. http://www.itbaker.comlmsdslenq|ishhtmlld7728.htm
(accessed 9/12/07)

Sulfanilamide, MSDS NO S7658. Mallinckrodt Baker: Phillipsburg, NJ, May
25, 2007. http://www.itbaker.com/msdslenqlishhtmlls7658.htm (accessed
9/12/07)

1,8-Diaminonaphthalene, MSDS NO SLD2968. Sciencelab.com: Houston,
TX, December 10, 2005. http://wwwsciencelab.com/xMSDS-
1 8 Diaminonaphthalene-9927150 (accessed 9/12/07)

McGuffin, V. L.; Goodpaster, J. V. In Encyclopedia of environmental

analysis and remediation; Meyers, R., Ed.; Wiley: New York, NY, 1998,
pp 3815-3831.

19

69. Simon, R. E.; Walton, L. K.; Liang, Y.; Denton, M. B. Analyst 2001, 126,
446-450.

20

CHAPTER 2. SEPARATION OF NITRATED EXPLOSIVES

The TLC separation Of nitrated explosives and degradation products was
accomplished using mobile phase systems that are less toxic than those
currently in use. An analysis procedure was developed to separate an unknown
explosive sample containing up to 12 explosives in two steps. In the first
analysis step, nitroaromatic and nitrate ester explosives can be separated and
identified, while picric acid and nitramine explosives are analyzed in the second

step. This method was successfully applied in the analysis of two blind samples.

2.1 Experimental Methods

2.1.1 Chemicals

Conventional manganese-activated zinc silicate TLC plates (Silica gel 60,
5x10 cm and 5x20 cm, 150-A pores, 250-pm layer thickness, Whatman, Florham
Park, NJ) were used for explosives separation. Reagent and spectral grade
solvents used in TLC plate preparation and explosives separation included
acetone (Sigma-Aldrich, St. Louis, MO), acetonitrile (Sigma-Aldrich),
benzaldehyde (Fisher Scientific, Waltham, MA), benzonitrile (Eastman Organic,
Rochester, NY), cyclohexane (Sigma-Aldrich), hexane (Honeywell Burdick &
Jackson, Morristown, NJ), methanol (Sigma-Aldrich), methylene chloride
(Spectrum, Gardena, CA), methylethylketone (Columbus Chemical Industries,
Columbus, WI), pentane (Jade Scientific, Canton, MI), tetrahydrofuran
(Spectrum), and triethylamine (Spectrum). Analytical standards of 2-NT, 4-NT,

2,4-DNT, 2,6-DNT, 4-am-DNT, TNT, tetryl, picric acid, HMX (Sigma-Aldrich),

21

RDX, NG, PETN (Radian, Austin, TX), and NB (Mallinckrodt, Hazelwood, M0)
were used as received in methanol or acetonitrile solution. Standard solutions
(10'1 to 104 M) were prepared for each explosive in spectroscopic grade
methanol or acetonitrile (Honeywell Burdick & Jackson). Working concentrations
and application volume of the nitrated explosives solutions were chosen for TLC

analysis to yield solute zones visible to the eye under UV irradiation.

2.1.2 Plate Preparation

Prior to chromatographic analysis, plates were pretreated to remove loose
particles and contamination that may lead to irreproducible retention or
fluorescence backgrounds. The TLC development chamber (Kodak, Rochester,
NY) used for plate pretreatment consisted Of a glass mobile phase reservoir and
two textured glass plates between which the TLC plates were placed during the
washing step. Plates were incubated in pure solvent in the development
chamber for 1 hour and heated at 90 °C for at least 24 hours to remove residual
solvent or adsorbed compounds. Plates were stored in the 90 °C oven until use
to prevent adsorption Of water or other contaminants from the laboratory
environment that may deactivate the silica.

Several common solvents Of varying solvent strength were investigated for
use in plate pretreatment, including acetone, acetonitrile, hexane, methanol, and
methylene chloride. Test plates were partially developed in each solvent and the
fluorescence was observed for the washed and unwashed sections of each plate.
Upon visual comparison, plates developed in acetonitrile, hexane, and methanol

had equivalent fluorescence after washing, while the fluorescence of plates

22

cleaned using acetone and methylene chloride had significantly less
fluorescence. In addition, a build-up of contamination at the solvent front was
Observed on plates cleaned in acetonitrile and methanol that was not observed
on plates washed in hexane. The visible build-up indicated greater removal of
non-fluorescent compounds from plates washed in methanol and acetonitrile. In
consideration of application to field use by forensic scientists, methanol is a
better choice of solvent because the cost of acetonitrile is approximately 2.5
times greater than that Of methanol. For these reasons, methanol was used for
plate pretreatment for all subsequent measurements.

Standard solutions were applied in 5 pL increments ~1 cm from the
bottom Of cleaned plates using a capillary tube (PCR Micropipettes, to deliver 1-5
pL, Drummond, Broomall, PA). Application was repeated in 5-pL increments
until the desired mass of solute was reached. An alternative method in which
solutes were applied using a 10-pL syringe (Hamilton, Reno, NV) was found to
be more time intensive than using a capillary tube. The increase in time required
for application with the syringe allowed evaporation Of volatile solutes (e.g.,
nitrobenzene) and resulted in decreased signal for the same solute concentration

compared to the capillary tube application.

2.1.3 Plate Development

An enclosed developing chamber (Desaga, Heidelburg, Germany) was
used for development Of plates after solute application. The glass lid was sealed
to prevent loss of mobile phase vapors and to allow plate development under

equilibrium conditions. Plates were placed inside the apparatus such that the

23

mobile phase level was less than 1 cm from the bottom Of the plate. Plates were
developed until the mobile phase front reached 1 cm from the top (Figure 2.1, a
and b). The plate was removed from the development chamber and the mobile
phase was allowed to evaporate. A handheld UV lamp (254 nm, Ultraviolet
Products, Upland, CA) was then used to visualize the solutes on the plate, seen
as dark zones against a fluorescent background (Figure 2.1, c-f). The distance
traveled by the solute was measured from the center of the zone at both the
initial and final locations on the plate. Relative solute retention was determined
by comparison Of calculated retention factors, or Rf values, as described by
Equation 2.1.

_ solute distance
solvent front distance

 

(2.1)

Solutes that are highly retained and remain near the original location will have
small Rf values, near zero. Solutes that are non-retained or travel to the solvent

front will have large Rf values, near unity.

2.2 Theory of Adsorption Chromatography

The retention of any compound in TLC depends primarily on the relative
adsorption energies of the functional groups within the solutes. The Q?
parameter is a dimensionless representation of the free energy of adsorption for
a functional group i.1 Values of Qi" depend on the nature Of the functional group
and are different for substituents on aliphatic, aromatic, and combined aliphatic
and aromatic backbones (Table 2.1). Assuming that each substituent is

independent, the energy of adsorption for each nitrated explosive is estimated by

24

 

a ""0 TTTTTTTTTT 1

Figure 2.1. Sample plate containing (I) a reference sample and (II) an unknown
sample. Designated plate areas include (a) the original location and a
completely retained solute (Rf = 0), (b) the solvent front, (c,e) a solute (Rf(c) =
0.21, Rf(e) = 0.66) that is detected in the reference sample but undetected in the
unknown sample, (d) a solute (Rf = 0.56) that is detected in both the reference
and unknown samples at similar concentrations, and (f) a solute (Rf = 0.87) that
is detected at a higher concentration in the reference sample than in the
unknown sample.

 

 

 

25

Table 2.1. Group adsorption energies for substituents common to nitrated
explosives and degradation products.1

 

 

 

Qlo

Substituent, i Aliphatic Aromatic
Carbon (-C-) 0.25 0.25
Methylene (-CH2-) -0.05 0.07
Methyl (-CH3) 0.07 0.11
Ether (-O-) 3.61 0.87
Hydroxyl (-OH) 5.60 4.20
Tertiary amine (-N<) 5.80 2.523
Primary amine (-NH2) 8.00 5.10
Nitro (-NOg) 5.71 2.77

 

a Tertiary amine value is for mixed aliphatic-aromatic substituent.

the sum of the group adsorption energies, 050 (Table 2.2).1'2 When comparing
two solutes, the molecule with the greater net adsorption energy will be predicted
to be adsorbed most strongly and therefore most retained.

Of the substituents common to nitrated explosives, carbon-based groups
contribute the least to the adsorption, while amine, hydroxyl, and nitro groups
contribute most. For carbon-based functional groups, aromatic substituents have
higher adsorption energies than aliphatic groups. Conversely, for stronger
adsorbing groups such as those containing nitrogen or oxygen, aliphatic forms
are more strongly adsorbed than aromatic groups. In addition, increasing the
number of functional groups increases the adsorption of the molecule. From 2-
NT to 2,6-DNT to TNT, for example, the adsorption energy increases with
addition Of aromatic nitro substituents (Oi0 = 2.77). For 4-am-DNT, the addition
of an aromatic amine substituent (Qio = 5.10) results in increased adsorption
energies relative to 2,6-DNT. Similarly for picric acid, the addition of an aromatic
hydroxyl substituent (Q.0 = 4.20) results in increased adsorption energies relative
to TNT. Tetryl contains an additional aliphatic nitro substituent (Q:0 = 5.71) and
an aliphatic/aromatic tertiary amino substituent (0.0 = 2.52), also increasing the
adsorption energy relative to TNT. In addition, nitramine and nitrate ester
explosives are expected to be more strongly adsorbed than nitroaromatic
explosives, as the group adsorption energy for aliphatic nitro substituents (Q;0 =
5.71) is twice that for the same group on an aromatic molecule (an = 2.77). The
adsorption energies of HMX and PETN are greater than those for RDX and NG,

respectively, as expected from the addition of the fourth nitro substituent.

27

Table 2.2. Net adsorption energies for explosives and degradation products.1

 

 

Compound (QI°)net"
TNT 9.92
Picric acid 14.01
Tetryl 17.99
RDX 34.38
HMX 45.84
NO 28.21
PETN 37.53
NB 4.27
2-NT 4.38
2,4-DNT 7.15
2,6-DNT 7.15
4-am-DNT 12.25

 

a Net values calculated according to Reference 2.

28

In addition to primary adsorption effects represented by Q? values,
secondary effects must also be considered.3 Although adsorption energies
calculated in this way can provide some information about relative adsorption
strength, secondary structural and electronic effects are also important in
molecules with substituent groups that are not independent. For example,
concerted adsorption may be possible for molecules in which two non-interacting
functional groups in close proximity can simultaneously interact with the surface.
Concerted adsorption can enhance adsorption relative to the same groups when
separated.3 Conversely, interaction between groups in close proximity can
reduce or prevent adsorption of either group, resulting in an antagonistic effect
on adsorption relative to the same groups when separated. Electron-donating
and -withdrawing substituents can also enhance or diminish adsorption by
causing intramolecular inductive effects in aromatic systems. In addition,
adsorption can be diminished sterically by bulky, non-interacting groups that can
limit access Of other substituents to surface sites. Similarly, increased linearity
and planarity of molecules can lead to increased adsorption through
simultaneous interaction of multiple sites with the surface. Understanding the
Impact of secondary effects can allow predictions about the adsorption Of a group
of solutes, such as nitrated explosives.

The interaction between the mobile phase and the surface must also be
considered when predicting retention behavior. In adsorption chromatography,
the mobile phase acts as a displacing agent during the separation. The retention

therefore depends on the solvent strength parameter, 2°, Of the mobile phase.

29

The solvent strength parameter describes the relative adsorption of a pure
mobile phase on a given adsorbent. Solvent strengths are measured relative to

pentane, the simplest hydrocarbon mobile phase (defined as 3° = 0), where a

0 value implies increased displacing power.1 Measured strengths for

large a
mobile phases used in this study are given in Table 2.3. When a single mobile
phase cannot provide adequate separation Of solutes of interest, two or more
solvents can be combined to produce a mobile phase of intermediate strength.

For a binary mixture of a bulk mobile phase component (a) and modifier (b), the

combined solvent strength can be estimated by using Equation 2.2.1

n 0_ 0
log[Xb10a ”I8” EaI+1-—x,,)

8 =8 +
a. . anb (2.2)

 

From Equation 2.2, the combined solvent strength parameter (soab) is related to
the solvent strength parameters Of the bulk mobile phase constituent (303) and
modifier (3%), the adsorbent activity function ((1, taken to be unity). the mole

fraction Of the modifier (Xb), and the molecular area of the modifier (nb).1

2.3 Separation of TNT and Degradation Products

In the literature, common mobile phase mixtures used to separate
nitroaromatic explosives have ranged from 7:3 cyclohexanezchloroform (503., =
0.17) to 3:1 hexanezacetone (soab = 0.33).4 The numerical average of the 80
values for eleven mobile phases used to separate these compounds is 0.24.4 In
our studies, this solvent strength was first approximated by using a mixture Of

80:20 cyclohexane: MEK(e°ab = 0.25). This mixture, however, did not move the

30

Table 2.3. Properties of common TLC mobile phases.5

 

 

Solvent Strength Molecular

Mobile phase Parameter, 3° (SiOg) Area, n (A)
Pentane 0.00 5.9
Hexane 0.03 6.0
Cyclohexane 0.03 6.0
Carbon tetrachloride 0.1 1 5.0
Benzene 0.25 6.0
Chloroform 0.26 5.0
Methylene chloride 0.32 4.1
THF 0.35a 5.0
MEK 0.393 4.9
Acetone 0.47 4.2
Acetonitrile 0.50 10.0
Methanol 0.73 8.0

 

a Values estimated based on Reference 2.

31

solutes from their original location. A combination Of 85:15 hexane:THF (soab =
0.20) is also of appropriate strength to provide sufficient separation of
nitroaromatic explosives and degradation products. However, this combination
did not provide sufficient resolution of TNT from NB (Figure 2.2). To determine
the mobile phase composition necessary to fully resolve TNT from all
degradation products, the percent THF in the mobile phase was decreased
systematically from 15% to 8%, and the resulting retention factors are compared
in Figure 2.2. Because THF is a stronger displacing agent than hexane,
decreasing the fraction Of THF decreased the Rf values for NB, 2-NT, 2,4-DNT,
2,6-DNT, and 4-am-DNT at a similar rate. The Rf value for TNT, however,
decreased more dramatically in this range, overlapping NB at high percentages
of THF and nearing 2,6-DNT at low percentages Of THF. The optimum mobile
phase for separation of nitroaromatic explosives and degradation products was
88:12 hexane:THF (soab = 0.18), which allowed separation of TNT and NB while
maximizing the Rf value for 4-am-DNT, which was highly retained. Using this
mobile phase composition, the observed retention order was 4-am-DNT, 2,4-
DNT, 2,6-DNT, TNT, NB, and 2-NT, and the precision of Rf values are
summarized in Table 2.4.

Of TNT and the degradation products, 4-am-DNT was the most retained,
as expected from the calculated adsorption energies (Table 2.2). Both 2,4-DNT
and 2,6-DNT were more retained than TNT, although the group adsorption
energy Of TNT is greater. The presence Of the electron withdrawing p-nitro

substituent of TNT (Figure 1.2) may have an intramolecular inductive effect on

32

mr

.on 59.8... 85 .EV 50.3 .5 Egan .Aev EN .AOV mz 8855 .9885
538908 can A! ._.Z._. .6”. omSwmmE 9903 53:99. 05 co 8me: E ".1... E855 .6 «omtw .N.N e59“.

Z<mDuOmD>I<mHMF o\o

 

 

 

 

 

 

 

 

 

 

 

 

- > U...
0 < <
fl-I - - D .----H_ -
- I --®I: IIC
,,,,,, Qi: IOI- - II-
.......... o d-e- m

 

 

 

 

od

“4
o

‘3‘.
o
HOlOVd NOIlNElElEI

°°.
o

33

Table 2.4. The precision of retention factors for nitroaromatic explosives and
degradation products in 88:12 hexane:THF.

 

 

Solute Retention Factor, Rf
NB 0.698 1 0.0073
2-NT 0.754 1 0.007
2,4-DNT 0.383 1 0.006
2,6-DNT 0.442 1 0.006
4-am-DNT 0.080 1 0.005
TNT 0.618 1 0.005

 

3 Precision calculated as the standard deviation of 8 replicate measurements.

34

both O-nitro substituents, decreasing adsorption of these groups relative to the
same substituents in 2,6-DNT. In addition, simultaneous adsorption of the two 0-
nitro substituents may be sterically hindered by the presence of the methyl group,
decreasing adsorption for TNT and 2,6-DNT relative to 2,4-DNT. Taken together,
these secondary adsorption effects result in TNT being less retained than was
predicted by the adsorption energy. This trend was also reported by Zou and
coworkers in the separation of TNT, DNT, and am-DNTs in normal phase liquid
chromatography using a silica stationary phase.6 Lastly, although NB is
predicted to be the least retained compound, 2-NT is less retained than NB. For
2-NT, the steric effect of the methyl substituent may decrease the adsorption

energy Of the nitro group, resulting in less than predicted retention.

2.4 Separation of Nitroaromatic Explosives

The effect of mobile phase composition was also investigated for the
nitroaromatic explosives. Tetryl has an adsorption energy more similar to that of
the nitroaromatic than nitramine explosives, and was therefore separated with
the nitroaromatic compounds. A mixture of 88:12 hexane:THF (coat, = 0.18)
resulted in strong retention of picric acid (Rf = 0). Picric acid was also completely
retained in a mixture Of 80:20 hexane:THF (803-, = 0.22), while TNT traveled in the
solvent front (R- = 1). Based on these two separations, no mixture of hexane and
THF would retain TNT and cause migration Of picric acid simultaneously. A
combination of 70:30 cyclohexanezMEK (803-, = 0.28), however, yielded less
extreme retention of both picric acid and TNT. Using this mobile phase

composition, the observed retention order was picric acid, tetryl, and TNT, and

35

the reproducibility of the retention factors is summarized in Table 2.5.

Although picric acid was most retained, tetryl was predicted to be the most
retained compound based on calculated adsorption energies (Table 2.2). One
explanation is that intramolecular inductive effects are possible for picric acid in
which the presence of the electron-donating hydroxyl group may increase
adsorption of nitro substituents. In addition, concerted adsorption of the hydroxyl
and both O-nitro substituents may also be possible for picric acid, resulting in
increased retention. Combined, these two effects may have resulted in the

retention of picric acid being stronger than predicted by net adsorption energies.

2.5 Separation of Nitroaromatic and Nitrate Ester Explosives

In the literature, several mobile phase combinations have been used for
the separation of nitrate ester explosives, ranging from 4:1 petroleum
etherzdichloroethane (803., = 0.17) to 4:1 carbon tetrachloridezacetone (soab =
0.32).4 The numerical average of the so values for seven mobile phases used to
separate these compounds is 0.24.4 Because this so value is similar to that used
to separate TNT and its degradation products, a similar mobile phase Of 88:12
hexane:THF (303-, = 0.18) was used to separate the nitrate ester explosives from
TNT. In addition, because TNT and tetryl have similar adsorption energies,
separation Of tetryl from TNT, NG, and PETN was also possible using this mobile
phase. The Observed retention order was tetryl, NG, TNT, and PETN, and the Rf
values measured for these solutes are summarized in Table 2.6.

Despite the Observed retention order, calculated values of adsorption

energy indicate that the nitrate ester explosives should be most highly retained,

36

Table 2.5. The precision of retention factors for nitroaromatic explosives in 70:30
cyclohexanezMEK.

 

 

Solute Retention Factor, Rf
TNT 0.53 1 0.013
Tetryl 0.40 1 0.02
Picric Acid 0.07 1 0.01

 

3 Precision is calculated as the standard deviation 018 replicate measurements.

37

Table 2.6. The precision of retention factors for nitroaromatic and nitrate ester
explosives in 88:12 hexane:THF.

 

Solute Retention Factor, Rf

 

TNT 0.59 1 0.023
Tetryl 0.096 1 0.005
NG 0.28 1 0.04

PETN 0.78 1 0.02

 

3 Precision is calculated as the standard deviation of 8 replicate measurements.

38

followed by tetryl and TNT. As mentioned previously, the calculated adsorption
energies assume that substituents are independent because they are bonded to
aliphatic or aromatic carbon atoms, which is not the case for NG and PETN.
Thus, the adsorption energies for -O-NOz groups may be greater than those for
aromatic nitro groups alone. Interestingly, PETN is less retained than NG,
although PETN contains an additional —O-NOz functionality that increases its
calculated adsorption energy and predicted retention. From the construction of
molecular models, it appears that concerted adsorption is more sterically

favorable for NG than for PETN, therefore increasing the retention.

2.6 Separation of Nitramine Explosives

Compared to those used for separation of nitroaromatic and nitrate ester
explosives, stronger mobile phase combinations have been reported in the
literature for separation of nitramine explosives.4 These mobile phase mixtures
range from 10:1 chloroform:nitromethane (303-, = 0.31) to 9:1
dichloromethane:acetonitrile (goat, = 0.41).4 The numerical average of 3° values
for six mobile phases used to separate these compounds is 0.37.4 Because
RDX and HMX have been separated in stronger mobile phases than most
nitroaromatic and nitrate ester explosives, a combination of 88:12 hexane:THF
(goat, = 0.18) was not sufficient for their separation, resulting in Rf values of zero
for both solutes. Increasing the proportion to 75:25 hexane:THF (coat, = 0.23) still
resulted in unsatisfactory separation, with both solutes having similar retention
factors. Increasing the proportion further to 50:50 hexane:THF (203-, = 0.29)

improved the separation, and the reproducibility of retention factors obtained with

39

this mobile phase is shown in Table 2.7. As predicted by adsorption energies,
HMX was more retained than RDX.

Because picric acid was highly retained in 88:12 hexane:THF, analysis Of
picric acid with the nitramine explosives may be more effective in method
development than analysis with nitroaromatic explosives. Using 50:50
hexane:THF (303-, = 0.29), picric acid and HMX were not resolved. TO increase
solvent strength, a combination 0165:35 hexanezMEK (goat, = 0.30) was used with
a retention order Of picric acid, HMX, and RDX. Measured Rf values for these
solutes are summarized in Table 2.8. As predicted by the calculated adsorption
energies (Table 2.2), RDX was less retained than HMX. Picric acid, however,
was significantly more retained relative to RDX and HMX than predicted by
calculated adsorption energies. As discussed previously, concerted adsorption
Of the hydroxyl and both O-nitro substituents may be possible for picric acid,

resulting in greater than predicted retention.

2.7 Precision of Retention Factors

Reported in Tables 2.3-2.7 is the precision associated with replicate
measurements of Rf values for each solute. For all solutes except NG, the
relative standard deviation (RSD) in the Rf value was less than 8%. For weakly
retained solutes (Rf > 0.38), the RSD was less than 4%, with an average of 1.6%.
For more strongly retained solutes (Rf < 0.38), the RSD was significantly greater
at 7.9%. The greatest error in measurement of Rf values in TLC resulted from
accurately determining the center Of the solute zone, which was increasingly

difficult for weekly retained solutes. Although the magnitude of the deviation is

40

Table 2.7. The precision of retention factors for nitramine explosives in 50:50
hexane:THF.

 

 

Solute Retention Factor, Rf
RDX 0.21 1 0.028
HMX 0.138 1 0.009

 

a Precision is calculated as the standard deviation of 8 replicate measurements.

41

Table 2.8. The precision Of retention factors for nitramine explosives and picric
acid in 65:35 hexanezMEK.

 

 

Solute Retention Factor, Rf
Picric Acid 0.13 1 0.01a
RDX 0.509 1 0.007
HMX 0.32 1 0.02

 

8 Precision is calculated as the standard deviation of 8 replicate measurements.

42

consistent for all solutes, the RSD was greater for weakly retained solutes. The
large RSD observed for NG (15%) was likely a result Of both low signal and
nonlinearity of the adsorption isotherm, as a high concentration was needed to
accurately determine the Rf value. This combination made determination of the

center Of the solute zone difficult, leading to more variability in the Rf value.

2.8 Separation of an Unknown Explosive Sample

Based on the mobile phase Optimization studies discussed above, an
analysis scheme was developed and applied for the separation of two unknown
explosive mixtures. Each blind sample was applied by a volunteer onto three
5x10 cm TLC plates. The use of multiple small plates allowed rapid development
and therefore reduced analysis time. Three of the nine explosive standards were
also applied to the plate for direct comparison. The three plates were developed
simultaneously in 88:12 hexane:THF for determination Of nitrate ester and
nitroaromatic explosives and degradation products in less than 10 min. By
comparison of the Rf values for the unknowns with each of the standards (Table
2.9), unknown 1 was determined to contain TNT and 4-am-DNT. Unknown 2
was determined to contain 2,6-DNT (Table 2.10), as well as highly retained
compounds (picric acid, HMX, or RDX) that must be separated in the second
step using 65:35 hexanezMEK for the separation of nitramine explosives and
picric acid. Standards Of picric acid, HMX, and RDX, as well as unknown 2, were
applied to a fourth plate and developed in 65:35 hexanezMEK in less than 10 min.
By comparison of the Rf values for the unknown and standard samples (Table

2.10), unknown 2 was determined to contain picric acid and RDX, in addition

43

Table 2.9. Analysis of unknown explosive sample 1.

 

 

Rf Value
Solute Standard Unknown

2-NT 0.74
NB 0.68

TNT 0.60 0.60
2,6-DNT 0.47
2,4-DNT 0.44
Tetryl 0.13

4-am-DNT 0.11 0.11

 

44

Table 2.10. Analysis of unknown explosive sample 2.

 

 

 

Rf Value
Solute Standard Unknown
2-NT 0.71
NB 0.75
TNT 0.59
2,6-DNT 0.44 0.45
2,4-DNT 0.48
Tetryl 0.10
4-am-DNT 0.095
RDX 0.62 0.61
HMX 0.44
Picric acid 0.25 0.25

 

45

to 2,6-DNT determined in the first step. Using this method, all explosives
contained in two blind samples were correctly identified with a maximum total

analysis time of 30 min.

2.9 Conclusions

Using the two mobile phase systems described and identified here, a two-
step separation can be used to determine all Of the explosives considered in this
study. After application of an unknown mixture, plate development in 88:12
hexane:THF will separate nitroaromatic and nitrate ester explosives. Nitramine
explosives and picric acid will remain at the original location, as the hexane:THF
mobile phase is not strong enough to cause their migration. After identification
and quantitation Of the nitroaromatic and nitrate ester explosives, development in
65:35 hexanezMEK will provide separation of the nitramine explosives and picric
acid. Using this method, up to 12 explosives and degradation products can be

separated in two steps.

46

2.10 References

1.

Snyder, L. Principles OfAdsorption Chromatography, Marcel Dekker: New
York, NY, 1968.

Barton, A. CRC Handbook of Solubility Parameters and Other Cohesion
Parameters; CRC Press: Boca Raton, FL, 1991.

Lakowicz, J. Principles of Fluorescence Spectroscopy, Plenum Press:
New York, NY, 1983.

Yinon, J.; Zitrin, S. The Analysis Of Explosives; Pergamon: Elmsford, NY,
1981 .

Snyder, L. J. Chromatogr. 1966, 25, 274-293.

Zou, H.; Zhou, S.; Hong, M.; Zhang, Y.; Lu, P. Anal. Chim. Acta 1994,
291, 205-210.

47

CHAPTER 3. QUANTITATION OF NITRATED EXPLOSIVES

Quantitation of nitrated explosives and degradation products following
TLC was accomplished by using images acquired with a computer-controlled
charge-coupled device (CCD) camera. The positions Of system components
were adjusted for imaging of fluorescent TLC plates under UV irradiation, where
areas containing solute zones were dark against a fluorescent background. The
settings associated with image acquisition and manipulation were also
investigated to ensure collection of the highest quality data. Using the software
associated with the camera, images were corrected for illumination gradients.
The peak height and area of each dark zone were correlated to the mass of
solute using both a semi-logarithmic and logarithmic relationship. Explosives
were quantitated over three orders of magnitude with practical detection limits in
the low microgram (pg) range. This method was successfully applied for the

quantitation of two blind samples.

3.1 Positioning of Imaging System Components

An uncooled, computer-controlled CCD camera (model DXM1200F, Nikon,
Melville, NY) was used to image TLC plates under continuous illumination by two
mercury lamps (254 nm, Ultraviolet Products, Upland, CA) (Figure 3.1).
Illumination by UV lamps resulted in a fluorescent green background of the plate
with darkened zones that correspond to the solute locations. The relative
positions of the system components were adjusted to maximize fluorescence
from the plate as well as provide consistency between consecutive

measurements. The fluorescence power of the TLC plate was dependent on the

48

 

CCD

 

 

 

 

CAMERA I

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

75 cm
7.5 cm \ A’
TLC Plate
I —
l‘ t
I 75m1 I

Figure 3.1. Diagram of TLC plate imaging setup.

49

distance and angle between the lamp and the plate. The greatest fluorescence
power was Observed when the lamp was placed directly above and parallel to the
plate. Because this orientation was not possible for imaging, two UV lamps were
used to provide illumination from both sides Of the TLC plate.

For positional reproducibility, the TLC plate was placed on a grid that was
fixed to the bench top. The two lamps were secured 7.5 cm above the plate at
~30° to the surface with a 7.5 cm separation between the two lamps. In addition,
the UV lamps required one hour of stabilization to Obtain a constant fluorescence
power of the TLC plate, as shown in Figure 3.2. The stabilization time was a
result of fluctuation in the lamp output, not in the degradation of the fluorophore
or stationary phase. The lens Of the CCD camera was positioned approximately
75 cm from the plate and the position was fixed by using a locking clamp. This
height was sufficient to image the entire plate with appropriate resolution. To
reduce extraneous light that may result in increased noise, all images were

acquired in the absence of natural or ambient light.

3.2 Optimization of Image Acquisition and Manipulation

Grayscale images were collected by using V for Windows software
(Version 3.5, Photometrics, Tuscon, AZ), which controlled exposure time of the
camera as well as manipulation of the image for extraction Of chemical data. To
correct for the dark current of the CCD camera, a bias image, acquired with the
shutter closed, was subtracted from every image. As described by Simon and
coworkers,1 the intensities of the sample plates were also corrected for

illumination gradients using intensities from a background image that does not

50

.sE 8 :9 .:_E 8 9 .5E 1.13 .55 8 av .51.
mm A3 .EE 2. 3V .EE m A8 dEm. >3 9: :O @553 5cm 8E: 80:9 um. 98 O._._. m .6 8:88.025 N.” 239“.

29.—.50.“ 4<._.ZON_M_OI

 

con com com oov com com 09

- I I .- - _, _ I -_ - - . - IL m
E

, v

i 2 W

. s

1 3

W 3 l

W N

1 I.

1 3

m . N

- t m.
m?

 

51

contain the solutes Of interest, as described by Equation 3.1.

Raw Intensity
Background Intensity

 

Corrected Intensity = x Average Background Value (3.1)

Corrected intensity values were plotted versus horizontal pixel position to result
in a peak at the location Of the solute zone (Figure 3.3). This area has an
increased signal, or ratio of intensities of the sample and background plates,
above the background level, and the center of the zone corresponds to the peak
maximum. The peak height represents the mass at the center Of the zone, while
the peak area represents the mass of the entire zone.

Prior to image acquisition, values for gain and integration time are
selected by the Operator. Gain is commonly used to describe the electronic
amplification of a signal, where a greater value for gain implies a greater
amplification.2 TO examine the effect Of gain in this experimental system, images
Of TLC plates were collected before and after application of 3.6 pg Of NB at gain
levels 1, 2, and 3 using an integration time of 500 ms. The signal-tO-noise and
signal-tO—background ratios were compared for the corrected images to
determine the optimum gain setting (Table 3.1). The signal-tO-noise (S/N) ratio
was constant and the signal-to-background (SIB) ratio decreased when the gain
was increased. The decrease in S/B ratio resulted in decreased peak height and
area at higher gain levels. All subsequent images were Obtained using gain level
1 to ensure maximum peak heights and areas were calculated.

An integration time is also specified by the Operator prior to image

acquisition. The integration time describes the time that the shutter is exposed to

52

.50 80035 9.8 .3 00.0528

0_ 380.5 00.00500 05.. ..:0.. E0200 05 g 0:0 .E0.mo.0:.0.:0 050808500 05 0:0 8000:8800 E0.0t_0

.0 30 83800 500 :_ 00.00.00 0.200 0050.0. >200; 0 0:090 00.0 05:50 05 A0. .E0.00.0E0.:o 050808500

05 0:0 03800 :26ch 05 :_ 00.00805 0. .05 0.200 00:08. 20:08 0 0:80 00.0 05:50 0:18 0.200 0050.0.
20.03:.00 0 0:0 :00002 059.0 05 E 00205 000.0 0.03 00.8980 0.8.8 5105:: :0 :0 0:0 03:00 00:08.0.
0 3 05:00:00 0.0.0 03500 0 3 0:0 0.03 0:30.308 0 E .0. 8.0.8.0805... 0:0 800:: 00.03850, .0.” 0.59".

:05000 _0.:ON_.OI

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

0 ..... i m
M nIIMIIIIHPIIJ. 0
:0280 _0.:0N_.OI n w
__ _ .w
M - I- I
W .1I IIUIL o
O 0. IIIIIII 0 N
O .M
.0.
0 m. 0
W
W :0580 0.8000:
mm 8
.0 w
M.- /II\ m IIIIIII L 0
O - IIIIIII J
O D m. . IIIIIIII . n—
M
e
6
8
M _..--...-.--..-.-_ o
/\ m iiiiiii L U P
a 0.

53

Table 3.1. Comparison of image quality for various gain settings obtained with

an integration time of 500 ms.

 

 

Gain Level 1 2 3
Signal (S) 0.0799 0.0501 0.0346
Noise (N) 0.0173 0.0119 0.00736

Background (B) 1.01 0.999 0.987
SIN Ratio 4.62 4.21 4.70
SIB Ratio 0.0793 0.0501 0.0351

 

54

the illuminated plate, where long integration times generally result in higher
signals. At significantly large integration times, however, overexposure may
occur, resulting in decreased pixel resolution. To determine the optimum
integration time, images of TLC plates were collected before and after application
of 3.6 pg of NB at integration times from 100 to 4000 ms. The SIM and SIB ratios
were compared for the corrected images (Table 3.2). Because the signal
corresponds to a ratio of the sample to the background intensity (Equation 3.1),
the signal and background did not increase with increasing integration time. An
increase in both the S/N and SIB ratios was observed with increasing integration
time. Although the SIB decreased slightly at integration times greater than 1000
ms, this decrease was considered acceptable relative to the greater gain in SIN
ratio. Based on this study, an integration time of 4000 ms was utilized for all
subsequent measurements.

The V for Windows software also allows selection of a portion of the image
for further numerical analysis. Once the selection was made, pixel values could
be summed or averaged in the horizontal or vertical direction. For TLC plate
images, the horizontal direction was perpendicular to the solvent flow, while the
vertical direction was parallel to the solvent flow (Figure 3.3). Because the solute
zones traveled with the solvent in the horizontal direction, the mathematical
operation was performed in that direction to be meaningful. For both summing
and averaging of the pixel values, the amount of fluorescent background included
in the selection was also important. The program assigned the greatest values to

illuminated pixels and lesser values for darker areas. Because the solute zones

55

 

 

0.000 00000 0.00.0 0.000 «000.0 0000.0 000: 0.0

0.00 0.: E. :0. 00.0 0.00 0000. 2.0

30.0 00.. .000 .0000 0.. 0: .0. 050.0000
000000 20000 000000 00000 030.0 00.0 .2. 0002
0.000 0000.0 0000.0 300.0 0000.0 0000.0 .0. .0090

0000 0000 0000 000. 000 00. .05. as: 00.00.02...

 

.00:._. 50060.5 0:0..0> .0. 3:000 0.000: .0 50:09:00 .N.« 030...

56

appeared dark, the assigned pixel values were low in these areas. Incorporation
of background fluorescence into the sum or average around the zone therefore
resulted in higher backgrounds without an increase in signal. To compare the
effect of summing and averaging of pixels, images of TLC plates were collected
before and after application of 3.6 pg of NB. The SIM and S/B ratios were
compared for selections of various heights of the corrected images (Table 3.3).
Overall, signal summing and averaging resulted in comparable SIM and S/B
ratios. Increasing the size of the selection, however, decreased both the SIN and
S/B ratios, because more fluorescence background was incorporated into the
larger selection. Based on this observation, all selections in subsequent
measurements were chosen around the solute zone to minimize fluorescence
background inclusion. Although summing and averaging resulted in similar
figures of merit, summation was chosen for subsequent measurements.
Because mass diffuses radially from the center of each solute zone throughout
the development process, the areas near the edge will contain less mass and
therefore appear lighter in the image. Summation of the pixels will not discount

the lighter areas of less mass, and will therefore be more physically meaningful.

3.3 Correction Using a Background Plate

As described by Equation 3.1, raw images were corrected using the
fluorescence of a background TLC plate that did not contain the solutes of
interest. The best method for background correction would be to acquire an
image of the same plate before and after chromatographic development. In the

field, however, this approach would be time consuming and impractical. To

57

 

0N5... oomod ommod R50 00.0.0 0.00.0 5.0m m.w

 

00.0 00.0 0... 00.0 .00 0... 0000 2.0
00.. 00.. 0... .000 000.0 .0.0 .0. 008.0000
00.000 .0000 0.0000 00.000 00.000 00000.0 .2. 0002
0.00 200.0 00000 50.0 50.0 0000.0 .0. 000.0
000 .00 0... 000 .00 00. .0000. .000... 000800
0_0x_n_ .0 000.0>< 0_0x_n_ .0 E00

 

000:5 0:. .0 0:0_.00_00 050 .5050 5 0.0000 0500.0>0 0:0 05.5.50 .0. 3:000 0005 .0 50000800 .0..” 0.00...

58

determine if different plates could be used for the background and sample
images, several situations were considered. First, repeat measurements of the
same plate were conducted in which the plate was removed from the imaging
location between acquisitions. The RSD of the background fluorescence for
repeat measurements of the same plate was less than 5% when the dimensions
of the summation area were consistent between images. The fluorescence of
several TLC plates from the same lot was similarly compared, with an RSD of
less than 5%. The fluorescence of several TLC plates from the same lot was
also compared for orientations relative to the direction that the solvent traveled
during the washing step. The orientation of the plate did not affect the measured
fluorescence, with an RSD less than 5%. Because the fluorescence of
background plates was consistent, a single background plate can be used for all
image corrections within that lot. The background image must be acquired at the
same time as the sample image, however, to provide a consistent illumination
profile and appropriate background correction. Similarly, the selection from the
background plate must be from the same location as the selection from the
sample plate. In addition, the background and sample plates must be pretreated

in methanol at a similar time to ensure consistent fluorescence emission.

3.4 Theory and Data Analysis

Peak heights and areas were determined by visual inspection in Microsoft
Excel as well as through the use of computer software (PeakFit, version 3.18,
SYSTAT Software, San Jose, CA). Because mass diffuses radially from the

center of the zone during the chromatographic process, peak area represents the

59

entire mass of the sample. Conversely, the peak height represents the mass
only at the center of the zone, and although it can be used for quantitation, peak
height is less representative of the entire sample. For peak height, manual
determination in Microsoft Excel was accomplished by subtracting the average
baseline level from the maximum intensity value. For peak area, a right Riemann

sum was used to approximate the integration, as described by Equation 3.2

i Yr (xi T xi—1) (3'2)
i=1

x and y are the coordinates of each point being integrated. A Riemann sum
calculated from the average baseline level was subtracted from the Riemann
sum of the solute zone to normalize all baselines to zero.

Peak heights and areas were also determined by fitting to a symmetric

double-sigmoidal function using PeakFit software, as shown in Equation 3.3

l: a 1_ 1

I0 x_b+E 7 X-b-E ’ (3.3)
1+exp - d 2 1+exp - d 2 .

where x is horizontal position. The symmetric double-sigmoidal function was

 

 

 

 

 

 

determined to best represent the shape of the solute zones through iterative
determination of constants a, b, c, and d. Figure 3.4 demonstrates the quality of
fit generated by the PeakFit program. The PeakFit program also incorporated
the shape of the baseline in the generated function, and the best fit equation was
used to determine peak height and area. In general, a more accurate

representation of the peak height and area was obtained by using PeakFit and

60

29.0 -- —— ~— - - -. - -f -. -0- v f
28.0 —
27.5 -

27.0 ~

 

 

 

CORRECTED INTENSITY

26.5 J

700 800 900 1000 1 1 00
X POSITION

 

29.0

 

28.5 - . . w. ,l h. l. W50.

28.0 ‘ i.

27.5

CORRECTED INTENSITY

27.0 _ I!“ W ‘

 

26.5 '

700 800 900 1000 1 100
X POSITION

Figure 3.4. Fit of TNT to the symmetric double-sigmoidal function (Equation 3.3).
(A) Corrected intensity for 9.8 pg TNT. (B) Symmetric double-sigmoidal fit of
9.8 pg TNT data (— — —), R2 = 0.987.

61

therefore all reported values were determined with this method.
Using both the peak height and area, calibration curves were constructed
for each of the solutes. At low concentrations, a semi-logarithmic relationship is

predicted by Beer’s Law, as shown in Equation 3.4
A = —log[%) = sbc (3.4)
As traditionally defined, A is the absorbance, a is the molar absorptivity, b is the
pathlength, and c is the concentration. I is the intensity after sample interaction,
and lo is the intensity of the source light. For this application, the corrected
image is a ratio of the intensity of the sample plate (I) to the intensity of the
background plate (lo) according to Equation 3.1, and is therefore analogous to
the intensity ratio (mo) in Equation 3.4. In addition, the concentration (c) is
related to the mass of the sample, as the solvent was evaporated prior to the
measurement. As a result, Beer’s Law predicts that the mass of the sample is
linear with the logarithm of the peak height or area. This relationship was
observed by Simon and coworkers for a variety of organic pharmaceuticals?"4
The underlying assumptions of Beer’s Law require that the solute be low
in concentration and be distributed randomly throughout the medium. On a TLC
plate, however, as the concentration of molecules increases, solute molecules
may arrange non-randomly to maximize the interaction between them as the
solvent evaporates. The process of solute application onto the plate may also

result in layering of the molecules, which can in turn affect uniformity. In order to

extend the linear range to higher masses, an empirical equation (Equation 3.5)

62

was identified to describe the nonlinear relationship between mass (m) and the

logarithm of peak height or area (l/lo).
log [It] = w + 2 log m (3.5)
0

Using this equation, the logarithm of the peak height or area was linearly related
to the logarithm of the concentration. Detection limits were calculated based on
peak height data using both relationships (Equations 3.4 and 3.5) as the
theoretical concentration that would yield a S/N ratio of 4.353 at the 90%
confidence level.1'5'7 Linear ranges for each of the explosives and degradation
products were taken as the mass range extending from the detection limit to the
maximum concentration used in this study. The detection limits and linear
ranges determined for each class of explosives is discussed in the following

sections.

3.5 Quantitation of Degradation Products

At low concentrations, calibration curves were calculated using Equation
3.4 and are shown in Figures 3.5A-3.9A for each of the degradation products.
For the entire concentration range, calibration curves were calculated using
Equation 3.5 and are shown in Figures 3.5B-3.9B. Similar trends were observed
when curves were calculated based on peak height and area, and only the
results for peak area are shown here. Table 3.4 summarizes the correlation
coefficients for the semi-logarithmic fits (Equation 3.4) of both the peak height
and area data for each of the solutes. Table 3.5 summarizes the correlation

coefficients for the logarithmic fits (Equation 3.5) of both the peak height and

63

 

1O6

105

PEAK AREA

 

 

1O4

 

MASS (#9)

 

107

A
O
0
r

PEAK AREA
3.5.

 

 

 

 

104 . . .
10-1 10° 101 1o2 103

MASS (r49)

Figure 3.5. Calibration curves for NB based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described in
Equation 3.5.

64

106

 

105‘

PEAK AREA

 

 

 

104 I

o 5 1o 15
MASS (us)

 

107

106 _

_'L
O
(7|

 

PEAK AREA

 

 

104 _ , .
100 101 102 103
MASS (pg)
Figure 3.6. Calibration curves for 2-NT based on peak area using (A) the semi-

logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described in
Equation 3.5.

 

65

 

106

105

104.

PEAK AREA

 

 

 

103 . , .
o 0.1 0.2 0.3
MASS (#9)

0.4

0.5

 

1O7

.3.

O
O)
1

PEAK AREA
5.5,.

_\.
O
.0.

 

103

 

 

 

10-2 10-1 100 1o1
MASS (rug)

I

102

103

Figure 3.7. Calibration curves for 2,4-DNT based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described

in Equation 3.5.

66

1O6

 

 

 

 

 

 

 

 

 

A.

< O
35" O
< 1051 %'//
x O
'<
I.”
n.

104 I I ,

O 1 2 3 4 5
MASSWQ)
106
B

'<
32 5
<10
X
.<
m
a.

104 , T E

10'1 100 1O1 102 1O3

MASS (119)

Figure 3.8. Calibration curves for 2,6-DNT based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described
in Equation 3.5.

67

 

 

 

 

 

 

 

 

 

106
A
<
I.”
25 5 °
x 10 a .
< .
3.“
C
104 r
0 1 2 3 4
MASS (H9)
107
B

106-
<
I.”
I:
<
§
"1105-
O.

C
104 1
10'1 100 101 102 103

MASS (#9)

Figure 3.9. Calibration curves for 4-am-DNT based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described
in Equation 3.5.

68

60:39.60 Emma: 88 E0: guano: on 9 858.9% m:_m> .o_nm__m>m 6: 0. <2 a
hand u 2% mafia 85,5900 .0

 

 

 

com I 8 <2 E35 EBB 250
o8 I mm <2 58.0 386 02
we I no <2 $86 :93 x2...
3 I 00 v H m ~83 $5.0 xom
3 I 3.0 <2 am :3 83¢ .200 0:20
m .. to F a m 255 owned E5»
0 I No F H m Bond :25 eze
m I no no H to <53 mmuad 5953
we I no <2 593 856 #293
8.0 I mod 8.0 0 8.0 586 flood 50:}
3 I .1 “<2 mmmmd 886 #2-.“
ad I no to q. me $86 38.0 m2
33 encam— Eocfi ~33 cozooaoo No.3 soon. Ego... goon. 27.0.an
*0 :5... my: 2205on 5:22.00 8.222

 

is $ch 8:82

o_E£:moo_-_Emm o5 mEm: 30:85 cozmcEmmu new $28.98 8.9:: .8 E9: Co 353 cozomamo .v.» 2%...

69

.83. 0 2a 0%: 85.538 .0

 

 

 

 

0mm I mm or H or 326.0 Nhomd Z._.Mn_
own I mm v H _. vaod monod 02
com I o N H 0 59.0.0 mowmd XS...—
omm I m N H m mmmwd mmhmd XDN—
:~ I no.0 5006 H ovod :~de 3506 20m urea
09. I m _. H m owomd mwvmd 3:8.
ONV LVN 10 H YN g6 Nnmmd ._.Z._.
com I md no.0 H omd mwwmd hmmmd HZDIENIV
CNN I md 00.0 H 3.6 wand vawd FZDIQN
our I v.0 mod H and mommd mowmd FZOIVN
own I V4. 50.0 H 220 NNNQO mmuwd ._.Z..N
on? I N._. Nd H N._‘ NNde m—‘mmd m2
3.: macaw. .35.. .33 cozooaoa no.2 god 232.. xmon. ozmoixm
“.0 :5: r5 220580 5:20:00 322:2
.65, 8:83:

5:82 255230. 9: 05m: $2603 cozmnfimou new mozwoaxo 828:: .8 #5.: Co menu: cozofioo .md min...

70

area data for each of the solutes. The correlation coefficients, or R2 values,
represent the quality of fit between the data and the mathematical relationship of
interest. Correlation coefficients can vary between zero and unity, with a value of
zero indicating no correlation and a value of unity indicating a perfect correlation.

For most of the degradation products, the logarithmic expression over the
entire concentration range (Equation 3.5) was a better fit to the data than the
semi-logarithmic expression at low concentration (Equation 3.4), as seen from
the correlation coefficients (Tables 3.4 and 3.5). The semi-logarithmic correlation
coefficients for NB, 2-NT, and 4-am-DNT were greater than 0.9, indicating a
reasonable fit for these solutes in the low concentration range. The semi-
logarithmic correlation coefficients for 2,4-DNT and 2.6-DNT, however, were
significantly less, and the lower quality of fit is evident in Figures 3.7A and 3.8A.
The quality of fit was improved significantly by using the logarithmic expression
for all of the degradation products except 2-NT, for which the fit was slightly
poorer for the logarithmic expression. Despite the decrease in fit quality, the
correlation coefficient for 2-NT was greater than 0.9. Using the logarithmic fit, the
linear ranges were. extended to three orders of magnitude, compared with one
order of magnitude for the semi-logarithmic fit. By considering only the linear
range of the two mathematical fits, the logarithmic fit will be much more useful in
the analysis of a sample of unknown concentration.

Detection limits were also calculated for each of the degradation products
using both the semi-logarithmic (Table 3.4) and logarithmic (Table 3.5) fits of

peak height data. For many of the degradation products, the semi-logarithmic fit

71

was so poor that the calculated detection limits were negative and were therefore
not meaningful. Detection limits calculated for NB, 2,4-DNT, and 4-am-DNT and
were in the low pg range. Using the logarithmic fit, however, detection limits for
all degradation products were in the low pg range. The detection limits for NB,
2,4-DNT, 4-am-DNT calculated by using the logarithmic function were
comparable to those calculated by using the semi-logarithmic function, and were

comparable to those reported for nitroaromatic explosives using densitometry.8'9

3.6 Quantitation of Nitroaromatic Explosives

For quantitation, tetryl was classified with nitroaromatic explosives despite
its nitramine character, as the aromatic character was most important in the
absorption-based quantitation process. Calibration curves were calculated using
Equations 3.4 and 3.5 and are shown in Figures 3.10-3.12 for TNT, tetryl, and
picric acid, respectively. Again, because curves for peak height and area
followed the same trend, only peak area curves are shown. Tables 3.4 and 3.5
summarize the correlation coefficients for the semi-logarithmic and logarithmic
fits of the data, respectively.

For TNT, tetryl, and picric acid, the logarithmic expression over the entire
concentration range (Equation 3.5) was a better fit to the data than the semi-
logarithmic expression at low concentration (Equation 3.4), as seen from the
correlation coefficients (Tables 3.4 and 3.5). The quality of fit for the semi-
logarithmic expression was acceptable for picric acid, with a correlation
coefficient greater than 0.9. The regressions for TNT and tetryl were less

satisfactory, as seen from Figures 3.10A and 3.11A. As observed with the

72

 

1O6

 

 

 

 

 

 

 

 

/\
105.
g .
i . ‘
fig 104— I.
C
103 l . f T I
0 1 3 4 7
MASS (P9)
107
BI
106.
‘1
33 105
‘<
K
<(
III-I
O. 104 -
103 . .
10'1 100 101 1O2 103

MASS (#9)

Figure 3.10. Calibration curves for TNT based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described in
Equation 3.5.

73

 

105

104

PEAK AREA

 

 

 

103 I I I I I

3
MASS (I19)
106

 

105 -

104“

 

PEAK AREA

 

1O3

 

 

10-1 100 101 102 103
MASS (I19)

Figure 3.11. Calibration curves for tetryl based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described in
Equation 3.5.

74

1O6

 

 

 

 

 

 

 

 

 

<
U]
a:
< 1052
E
a.
104
0 1.5
MASS (I19)
107
B
106-
<
l.l.|
0:
<
if 5
3.‘ 10 .
. O
104 I .
10-2 10-1 10° 101 102
MASS (#9)

Figure 3.12. Calibration curves for picric acid based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described
in Equation 3.5.

75

degradation products, the logarithmic expression increased the quality of fit and
linear range significantly, making it much more useful in the analysis of a sample
of unknown concentration. The linear range for the semi-logarithmic fit was one
order of magnitude, which was extended to 2 — 3 orders of magnitude by using
the logarithmic fit. The regressions for all three explosives were greatly improved,
although the quality of fit for tetryl was lower than those for TNT and picric acid.
In fact, the logarithmic fit for picric acid had the greatest correlation coefficient of
all of the nitrated compounds.

Detection limits were also calculated for each of the nitroaromatic
explosives using both the semi-logarithmic (Table 3.4) and logarithmic fits (Table
3.5) of the peak height data. The semi-logarithmic fit was so poor that the
calculated detection limit was negative and was therefore not meaningful.
Detection limits for TNT and tetryl were in the low pg range, as observed for the
degradation products and were comparable to densitometry“9 Using the

logarithmic fit, detection limits for all solutes were also in the low pg range.

3.7 Quantitation of Nitramine Explosives

Calibration curves were calculated using Equations 3.4 and 3.5 and are
shown in Figures 3.13 and 3.14 for RDX and HMX, respectively. Tables 3.4 and
3.5 summarize the correlation coefficients for the semi-logarithmic and
logarithmic fits of the data, respectively. Again, because curves for peak height
and area followed the same trend, only peak area curves are shown.

For RDX and HMX, the logarithmic expression over the entire

concentration range (Equation 3.5) was a better fit to the data than the semi-

76

 

 

 

 

 

 

 

 

 

 

105 $—
A
<
I..|.|
a:
<
X
<
LU
O.
104 5
0 1 2 3 4 5
MASS (I19)
106
B
<
35‘ 5
< 10
x
<
|.l.l
D.
104 . .
10-1 100 1o1 102 103

MASS (I19)

Figure 3.13. Calibration curves for RDX based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described in
Equation 3.5.

77

 

 

 

 

 

 

 

 

 

 

 

 

104
A O
f 4 . 4

, <

LI]

0:

<

x

<

I.l..|

O.

106 . 5 5

0 0.5 1 1.5 2
MASS (119)
106
B

105 a
<
|.I.l
0:
<
x
5 104‘
n.

103 , _ ,

10-1 10° 101 102 106

MASS (#9)

Figure 3.14. Calibration curves for HMX based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described in
Equation 3.5.

78

logarithmic expression at low concentration (Equation 3.4), as seen from the
correlation coefficients (Tables 3.4 and 3.5). For the semi-logarithmic expression,
correlation coefficients varied from 0.03 to 0.5, and the poor qualities of fit were
obvious from Figures 3.13A and 3.14A. Correlation coefficients for the
logarithmic expression were improved to 0.87-0.92, although the fits were still
quite poor (Figures 3133 and 3148). Although the correlations were poor, use
of the logarithmic expression increased the linear range to two orders of
magnitude, making it more useful in the analysis of a sample of unknown
concentration. The correlation coefficients for the nitramine explosives were
more comparable to those for nitroaromatic explosives, and were less than those
for many of the degradation products.

Detection limits were also calculated for RDX and HMX using both the
semi-logarithmic (Table 3.4) and logarithmic (Table 3.5) fits of peak height data.
The semi-logarithmic fit was so poor for HMX that the detection limit was
negative and was therefore not meaningful. The detection limit for RDX using
the semi-logarithmic fit was determined to be 5 pg, which was the highest
concentration used in the semi-logarithmic fit. Detection limits calculated using
the logarithmic expression were also in the low pg range, which were comparable

to those for nitroaromatic compounds and literature values for densitometry.1

3.8 Quantitation of Nitrate Ester Explosives
Calibration curves were calculated using Equations 3.4 and 3.5 and are
shown in Figures 3.15 and 3.16 for NG and PETN, respectively. Again, because

curves for peak height and area followed the same trend, only peak area curves

79

 

 

 

 

 

 

 

 

 

106
A
<
I“
a:
< 105 5 o
x
E
J.
“- o
g C
104 5 . . I
50 100 150 200 250 300 350
MASS (119)
106
B
<
35‘ 5
< 10 0
a:
fi /1_/‘
o
n. . .
104
1 O1 102 1 03
MASS (lug)

Figure 3.15. Calibration curves for NG based on peak area using (A) the semi-
logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described in

Equation 3.5.

80

 

 

 

 

 

 

 

 

 

1 O5
A
o
' ’1
< .2
I.”
‘5‘:
x o
<
I.”
O.
1 04 I 5
50 100 150 200 250 300
MASS (I19)
1 O5
B
o
://o
<
|.I.|
a:
< o
x
<
m
a.
1 04 5
1 01 1 02 103

MASS (119)

Figure 3.16. Calibration curves for PETN based on peak area using (A) the
semi-logarithmic fit described in Equation 3.4 and (B) the logarithmic fit described
in Equation 3.5.

81

_‘ ‘_

are shown. Tables 3.4 and 3.5 summarize the correlation coefficients for the
semi-logarithmic and logarithmic fits of the data, respectively. For both
expressions, the quality of fit was very poor, with correlation coefficients less than
0.05. The low correlation coefficients led to difficulties in estimation of linear
ranges and detection limits. Because NG and PETN have limited absorbance at
254 nm, higher concentrations were necessary to obtain a measurable
quenching signal. For the nitrate ester explosives, masses below 50 pg were
undetectable, and masses above 250 pg had asymmetric zones, making
determination of R5 values difficult and narrowing the workable concentration
range. The need for higher concentrations also led to nonlinear behavior of the
adsorption isotherm (vide infra), resulting in a decreased upper limit of the
concentration range relative to the other nitrated explosives. In addition, the
estimated linear ranges and detection limits for NG and PETN were at
significantly larger masses than those obtained for the other nitrated explosives

(Tables 3.4 and 3.5).

3.9 Quantitation of an Unknown Explosive Sample

Using the imaging system described here, quantitation of two unknown
explosive samples was achieved. For each blind sample, a known quantity of
explosive compounds was applied by a volunteer as described in Chapter 2
(Section 2.8). After development in the solvent systems discussed previously,
the TLC plates were imaged using the CCD camera. A background plate was
also imaged at the same time. The images were manipulated as discussed in

Section 3.2 and the peak areas and peak heights were used to calculate the

82

mass of sample by using previously acquired calibration curves and Equations
3.4 and 3.5 (Table 3.6). Using the semi-logarithmic relationship (Equation 3.4),
the calculated masses were less than the actual masses for all solutes except
picric acid. For 2.6-DNT, TNT, and picric acid, the mass in the unknown was
outside the linear range for this relationship (Tables 3.4 and 3.5), so the
discrepancies were not surprising. The calculated mass of picric acid was within
error of the actual mass. The calculated masses for 4-am-DNT and RDX were
within a factor of two of the actual masses. For 2,6-DNT and TNT, however,
logarithmic fit (Equation 3.5) was much more accurate, again with calculated
masses within a factor of two of the actual mass.

The error in each determined value was determined by the method of
propagation of error, not by replicate measurements. The error on each
calculated mass was large, greater than 10% for most of the compounds. For
some solutes, the error was nearly 50%. Although replicate measurements may
provide a more accurate mass for an unknown sample, the time required would
be too great for field analysis. Despite the large error, the values were within a
factor of two of the true value using previously acquired calibration curves. With
frequent calibration or internal calibration using standards of known concentration,
however, the accuracy of the quantitation would be greatly improved. Therefore,
the most practical procedure would be to use this imaging method to estimate the
concentration of an unknown explosive, but to use a more sophisticated
laboratory method or simultaneous calibration standards to determine the

accurate quantity.

83

.m.m cozmzum 05m: 85.3030 5
in cozmzcm oEm: cmumfiofio 5

 

 

 

3an mam F; In 4.: eze
mod H 8N No a 3 to a 3 No a 3 mom xom
m H mm N a e m H K m a e 8.5 .28 0:20
3.0 5 :3 No a 3 mod 5 Ed no a 3 mod ezoémé
om a ow m 0 : om 5“ E m H 2 0.8 e293
5:“. 2555?; 5:". 55. 25525.9... 5:“. as «was. 228
o_:.5tmmo_._Eom o_E5_.mmo_._Eom 3304.
$2 3.85. .570... 0.35.

 

3m: moms. boa—2:030

 

.moEEmm 02.06.30 5505...: 5 20.222 dd 2%...

3.10 Additional Observations

As mentioned for NG and PETN, many of the solutes exhibited streaking
on the TLC plate at high concentrations, analogous to a tailing peak shape often
observed in liquid chromatography under similar conditions. At high mass, the
linear region of the adsorption isotherm was likely exceeded, causing overloading
of the adsorption sites of the stationary phase. Although the shape of the zones
was irregular in these regions, the values measured for peak height and area
remained in the linear region of the calibration curve. Hence, one advantage of
this method was that even with large sample amounts, the linearity of the
calibration curve was maintained and quantitation was possible.

The linearity and detection limits calculated in this study were at much
greater concentrations than those cited in the literature. A high degree of
linearity was observed by Simon1 for concentrations 1 to 2 orders of magnitude
lower than those used in this study, based on images obtained using a cooled,
spectroscopic-grade CCD camera with reduced background noise. In this study,
higher concentrations were necessary to visualize the solutes on the TLC plates
using a standard digital camera. Although this method was less sensitive, use of
an uncooled camera was more relevant to proposed field applications. For field
applications, linearity and detection limits cannot be directly compared to those
using more sophisticated laboratory equipment, but the detection limits

determined in this study should be sufficient for preliminary screening.

85

 

3.11 Conclusions

Quantitation of nitrated explosives and degradation products following
TLC was accomplished by using images acquired with a computer-controlled
CCD camera. The imaging system was optimized to provide the most reliable
quantitative data. The peak height and area of the solute zone were correlated
to the mass of solute deposited on the plate using both a semi-logarithmic and
logarithmic relationship. The semi-logarithmic relationship was only valid at low
solute concentrations, while the logarithmic relationship could describe the entire
concentration range. In general, the logarithmic relationship provided the best fit
to the data and therefore the most reliable quantitation. Explosives and
degradation products were quantitated over three orders of magnitude with
practical detection limits in the low microgram (pg) range. For routine use in a
forensic laboratory, this method is advantageous because no reagents are
required and quantitation can be performed with a handheld UV lamp and a
digital camera. Using this standard equipment, the calculated detection limits
were comparable to more sophisticated quantitation techniques. The practical
utility of this method was explored, and quantitation of the explosive content in

two blind samples was achieved.

86

1.

3.12 References

Simon, R. E.; Walton, L. K.; Liang, Y.; Denton, M. B. Analyst 2001, 126,
446-450.

Skoog, D. A.; Holler, F. J.; Nieman, T. A. Principles of Instrumental
Analysis, 5th ed.; Saunders College Publishing: Philadelphia, 1998.

St. John, P. A.; McCarthy, W. J.; Winefordner, J. D. Anal. Chem. 1967, 39,
1495-1497.

Pearson, E. 8.; Hartley, H. 0., Eds. The Biometrica Tables for
Statisticians, 3rd ed.; Cambridge University Press: Cambridge, 1966.

Bladek, J.; Miszczak, M.; Popiel, 8. Chem. Anal. (Warsaw) 1995, 40, 723-
729.

Goodpaster, J. V.; McGuffin, V. L. Anal. Chem. 2000, 72, 1072-1077.
McGuffin, V. L.; Goodpaster, J. V. In Encyclopedia of environmental
analysis and remediation; Meyers, R., Ed.; Wiley: New York, NY, 1998,
pp 3815-3831.

Liu, Y.; Wei, W.; Wang, M.; Pu, Y.; Lin, L.; Zhang, J. J. Planar
Chromatogr. 1991, 4, 146-149.

Gholamian, F.; Chaloosi, M.; Waquif-Husain, S.; Dehghani, H. J. Planar
Chromatogr. 2001, 14, 296-299.

87

CHAPTER 4. CONCLUSIONS AND FUTURE DIRECTIONS

This project has sought to improve the TLC separation and detection of
nitrated explosives for infrequent cases involving local law enforcement and
forensic scientists. Unlike cases of national security or on the front lines of
international conflicts where sophisticated technologies are available and widely
used, local cases are less frequent and must be solved with the same accuracy
but with fewer resources. One of the most common preliminary screening
techniques used in these situations is TLC with spray reagents, followed by
determination by visual inspection.

The first improvement made to this technique was through the
incorporation of less toxic solvents to reduce safety concerns while maintaining
the inherent simplicity of TLC. Using the two mobile-phase systems described in
this work, a two-step separation can be used to determine up to 12 nitrated
explosives. After application of an unknown mixture, plate development in 88:12
hexane:THF will separate nitroaromatic and nitrate ester explosives. Nitramine
explosives and picric acid will remain at the original location, as the hexane:THF
mobile phase is not sufficiently strong to cause their migration. After
identification of the nitroaromatic and nitrate ester explosives, a second
development in 65:35 hexanezMEK will provide separation of the nitramine
explosives and picric acid. The identification of these compounds is
accomplished by visual inspection under a UV lamp, which allows comparison of
retention factors in an unknown and standard sample. This method was

successfully utilized in the identification of two unknown explosive samples. One

88

blind sample was found to contain 4-am-DNT and TNT and a second was found
to contain 2.6-DNT, RDX, and picric acid. Future work could extend this method
for use with other compounds not considered in this work, such as inorganic or
peroxide explosives.

Improvements have also been made in the quantitation of nitrated
explosives following TLC separation. Quantitation of nitrated explosives and
degradation products following TLC was accomplished by using images acquired
with a computer-controlled CCD camera. The imaging system was optimized to
provide the most reliable quantitative data. The peak height and area of the
solute zone were correlated to the mass of solute deposited on the plate using
both a semi-logarithmic and a logarithmic relationship. The semi-logarithmic
relationship was only valid at low solute concentrations, while the logarithmic
relationship could describe the entire concentration range. In general,
logarithmic relationship for peak area provided the best fit to the data and
therefore the most reliable quantitation. Explosives and degradation products
were quantitated over three orders of magnitude with practical detection limits in
the low microgram range. For routine use in a forensic laboratory, this method is
advantageous because no reagents are required and the quantitation can be
performed with a handheld UV lamp and a digital camera. Using this standard
equipment, detection limits were comparable to more sophisticated quantitation
techniques such as densitometry. In addition, this method provides qualitative
and quantitative information simultaneously, and the sample can be preserved on

the TLC plate and extracted from the stationary phase for further analysis in its

89

original form. This quantitation method was utilized in the analysis of two
unknown explosive samples. Quantitative determination of picric acid was
successful, with the concentration determined accurately within experimental
error. For four other compounds, this quantitation method estimated the
concentrations to be within a factor of two of the actual concentrations. The
utility of the method for estimating unknown sample concentrations has been
demonstrated, and improved quantitation can be achieved through further
laboratory analysis.

Future work on this project would involve design and implementation of a
field analysis system to incorporate the separation and quantitation components.
This system should be relatively inexpensive and contain all components
required for complete analysis. Mobile phase mixtures could be prepared in
sealed pouches that would be opened immediately prior to use. Pre-washed
TLC plates could also be sealed and opened immediately prior to use. Additional
investigations would be necessary to determine the optimal extraction conditions
to be used for unknown samples. Two small development chambers (~20x15x5
cm) would allow complete, simultaneous analysis of 2 unknowns in less than 10
minutes. Following separation, the plates could be placed in a static holder
under UV illumination for imaging by a digital camera. A computer program
could be developed that would identify solute zones and measure Rf values. The
program could also be used to store background images and calibration curves,
select areas for pixel value summation, and calculate corrected intensities from

images. Ideally, the program would also calculate peak heights and areas and

90

estimate concentrations from stored calibration curves for each solute.
Incorporation of these tasks into a computer program removes some subjectivity
of the technician or forensic scientist, and increases the objectivity of the analysis.
The program could then provide a printout or sound an alarm to notify the
technician of the presence of an explosive. This work, and the work that will
follow, can provide a simple, non-hazardous, and cost-effective method for
separation and quantitation of nitrated explosives for local law enforcement

agencies and forensic laboratories.

91

AAAAAAAAAAAAAAAAAA

Illll’lllllfllljflljllllllll‘ljlltj’ll