UvA-DARE (Digital Academic Repository) Chemical … · Chemical profiling of explosives PhD thesis, University of Amsterdam, The Netherlands ... General introduction 3 Chapter 1 1.1

UvA-DARE is a service provided by the library of the University of Amsterdam (http://dare.uva.nl)

UvA-DARE (Digital Academic Repository)

Chemical profiling of explosives

Brust, G.M.H.

Link to publication

Citation for published version (APA):Brust, G. M. H. (2014). Chemical profiling of explosives

General rightsIt is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s),other than for strictly personal, individual use, unless the work is under an open content license (like Creative Commons).

Disclaimer/Complaints regulationsIf you believe that digital publication of certain material infringes any of your rights or (privacy) interests, please let the Library know, statingyour reasons. In case of a legitimate complaint, the Library will make the material inaccessible and/or remove it from the website. Please Askthe Library: http://uba.uva.nl/en/contact, or a letter to: Library of the University of Amsterdam, Secretariat, Singel 425, 1012 WP Amsterdam,The Netherlands. You will be contacted as soon as possible.

Download date: 30 Jun 2018

http://dare.uva.nl/personal/pure/en/publications/chemical-profiling-of-explosives(4a4280f7-b284-4c09-b573-78f14d6bb556).html

Hanneke Brust


Hanneke Brust

NO

O

O

O

O N

O

210

89

63 180

134

30149164

103120

194


PhD thesis, University of Amsterdam, The Netherlands

The research described in this thesis was financially supported by the CBRN-program of the Netherlands Forensic Institute.

Author: Hanneke BrustISBN: 978-94-6203-599-7

Printing: CPI Whrmann


ACADEMISCH PROEFSCHRIFT

ter verkrijging van de graad van doctor

aan de Universiteit van Amsterdam

op gezag van de Rector Magnificus

prof. dr. D.C. van den Boom

ten overstaan van een door het college voor promoties ingestelde

commissie, in het openbaar te verdedigen in de Agnietenkapel

op woensdag 25 juni 2014, te 14.00 uur

door

Gerritje Maria Hendrika Brust

geboren te Utrecht

Promotiecommissie:

Promotores: Prof. dr. ir. P.J. Schoenmakers Prof. dr. A.C. van Asten

Overige leden: Prof. dr. ir. J.G.M. Janssen Prof. dr. M.C.G. Aalders Prof. dr. M.W.F. Nielen Dr. M.S. Beardah Dr. A.E.D.M. van der Heijden Dr. M. Koeberg

Faculteit der Natuurwetenschappen, Wiskunde en Informatica

Table of contents

Chapter 1 1

General introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1

1.1 .Explosives in forensic casework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3

1.2 .Forensic explosives analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

1.3 .Chemical profiling in forensic science . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

1.4 .Chemical profiling of explosives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

1.5 .Scope of this thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12

Chapter 2 17

Impurity profiling of trinitrotoluene using vacuum-outlet GCMS . . . . . . . . . . . . 17

2.1 .Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.2 .Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

2.3 .Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24

2.4 .Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Chapter 3 37

Pentaerythritol tetranitrate (PETN) profiling in post-explosion residues to

constitute evidence of crime-scene presence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.1 .Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.2 .Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44

3.3 . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48

3.4 .Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56

3.5 .Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62

3.6 .Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

Chapter 4 67

Accurate quantitation of pentaerythritol tetranitrate and its degradation

products using LCAPCIMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67

4.1 .Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69

4.2 .Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

4.3 .Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74

4.4 .Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83

Chapter 5 85

Investigation of isotopic linkages between precursor materials and the

improvised high-explosive product HMTD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85

5.1 .Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87

5.2 .Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88

5.3 .Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96

5.4 .Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98

5.5 .Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106

Chapter 6 109

Isotopic and elemental profiling of ammonium nitrate in forensic explosives

investigations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109

6.1 .Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111

6.2 .Experimental . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115

6.3 .Results and discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120

6.4 .Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141

Summary 145

Samenvatting 149

Dankwoord 153

Chapter 1General introduction

General introduction 3

Chapter 1

1.1 Explosives in forensic casework

Explosives analysis is an important discipline present in many forensic-science

laboratories. Although in most countries the case load in this field is relatively low

compared to other disciplines, such as DNA or illicit drugs, the impact of cases involving

explosives or explosions is often high. The criminal use of explosives is a global issue.

Attacks are committed worldwide by terrorist organizations or individuals, mainly by

the use of improvised explosives. Recent attacks include the bombing during the Boston

Marathon in 2013, utilizing pressure-cooker bombs, and the attack on the government

quarter in Oslo in 2011 with a fertilizer-based car bomb. Attacks on aircrafts were

carried out using improvised explosive devices, hidden in either shoes (in 2002) or

underwear (in 2009), but both attempts failed. These examples illustrate the need for

improved detection and analysis of explosives.

ba c

d e

Fig. 1.1. Typical forensic casework involving explosives: intact explosive material (ac) with (a) plastic explosive Semtex, consisting of RDX and PETN; (b) a pipe bomb, often containing pyrotechnic material; (c) an IED based on PETN and the home-made explosive TATP as used by the shoe bomber; (d) a post-explosion crime scene (Oslo bombing); and (e) raw materials for the production of explosive devices.

4 Chapter 1

Cha

pter

1

Forensic cases that involve explosives may typically include post-explosion investigations,

intact explosive devices, or materials possibly used to construct explosive materials or

associated devices (Fig. 1.1). A wide variety of explosives is encountered in forensic

casework, as a result of the alleged criminal use of military and commercial explosives

that are obtained through illegal channels, or by the use of home-made explosives

(Table 1.1). In general, military and commercial explosives are organic compounds with

nitro groups, belonging to the nitramines, nitroesters or nitroaromatics (Fig. 1.2). One

exemption is explosive-grade ammonium nitrate (AN), with applications in mining and

civil construction [1,2]. The organic military and commercial explosives are generally

characterized by high stability required for safe handling and storage and their high

detonation velocities, depending on their specific application.

N N

N

NN

NN

N

N

N

OO

O

O

OO

O

O

N

N

NNN

N

O

O

OO

O

O

N NN N

NO2NO2O2N

O2N

NN O

O

O

O

CH3N

N

NO

O

OO

O

O CH3N

N

O

O

OO

N

N

OO

O

O

CH3N

O

O

NO

O

N

N

NO

O

OO

O

O

OHN

N

NO

O

OO

O

O

NN

N

NO

O

OO

O

OH3C N O

O

O

O

O

O

OO

N

O

OON

O O

NOO

N

ON

O

O

O

O

N

N

N

N

OO

O

O

OO

O

O

OO

NO O

ON

O

O

O

N

H2C

H2C

O

O O CH2

N

CH2OH2C O O

H2C

OO

OOO

O

CH3H3C

CH3CH3H3C

H3C

HOO O

O OOH

HMX

RDX

CL-20

2,4,6-TNT

1,3,5-TNB

1,3-DNB

2,4-DNT

2-NT

Picric acid

TetrylNB

PETN

NG

EGDN

DEGDN

HMTD

TATP

MEKP

Nitroamines Nitroaromatics Nitrate esters Peroxides

Fig. 1.2. Chemical structures of different types of organic explosives, categorized by their functional groups. Full names of the compounds are given in Table 1.1.


Chapter 1

Table 1.1. Overview of the different types of explosives mentioned in this thesis, including their classification according to their use.

Explosive Full name(s) CategoryAN Ammonium nitrate Civil, home madeCL-20 Hexanitrohexaazaisowurtzitane MilitaryDEGDN Diethylene glycol dinitrate Military, civil1

EGDN Ethylene glycol dinitrate Military, civil1

ETN Erythritol tetranitrate Home madeHMTD Hexamethylene triperoxide diamine Home madeHMX Cyclotetramethylene tetranitramine; octogen MilitaryMEKP Methyl ethyl ketone peroxide Home madeNG Nitroglycerine Military, civilPA Picric acid MilitaryPETN2 Pentaerythritol tetranitrate MilitaryRDX Cyclotrimethylene trinitramine; cyclonite; hexogen;

research department explosiveMilitary

TATP Triacetone triperoxide Home madeTNT3 Trinitrotoluene Military, civil1

UN Urea nitrate Civil, Home made

In forensic casework, military explosives are often encountered as grenades or plastic

explosives. Another class of explosives is the home-made explosives. Because access

to military and civil explosives is restricted, improvised explosive devices (IEDs) are

often constructed using more readily available materials. These include fertilizers,

pyrotechnics, or peroxide explosives. Based on the chemical knowledge of the suspect,

compounds such as PETN are occasionally encountered as a home-made explosive.

Fertilizer-based explosives are ammonium nitrate (AN), often mixed with a fuel, and urea

nitrate (UN), obtained by nitration of the fertilizer urea. Because of their availability,

1 In mixtures with other types of explosives2 Common degradation products include:Pentaerythritol trinitrate (PETriN)Pentaerythritol dinitrate (PEDiN)Pentaeryhtritol mononitrate (PEMN)3 Common impurities include:Dinitrotoluene (DNT) isomersDinitrobenzene (DNB) isomersTrinitrobenzene (TNB) isomers

6 Chapter 1

Cha

pter

1

low sensitivity to shock and friction, and simple application, fertilizer-based explosives

are typically used in large explosive devices. Examples of attacks involving fertilizer-

based explosives are the Oklahoma City bombing in 1995 and the Oslo bombing in

2011. IEDs based on pyrotechnic material can contain fireworks ingredients or mixtures

of raw materials, such as potassium or sodium chlorate, potassium perchlorate or

potassium permanganate (oxidizers), sulfur and metal or metalloid powders (fuels) and

various additives. An example of an attack involving a pyrotechnic mixture is the Bali

bombings in 2002. Peroxide explosives can be synthesized using household chemicals,

such as acetone, hydrogen peroxide and acid. The most frequently encountered peroxide

explosives are hexamethylene triperoxide diamine (HMTD) and triacetone triperoxide

(TATP). These explosives are characterized by their sensitivity to shock and friction

and therefore production in large quantities is dangerous. Casework involving home-

made explosives does not only comprise intact explosives or post-explosion residues,

but frequently also raw materials.

1.1.1 Casework in the NetherlandsThe Explosions and Explosives group at the Netherlands Forensic Institute (NFI)

deals with about 200 cases per year. Approximately 60% of these involve intact

explosive devices, such as IEDs, grenades and illegal fireworks (Fig. 1.3). Other types

of casework include the examination of parts of intact explosive devices (~10%), such

as high-explosive (HE) substances and blasting caps, post-explosion casework (~15%)

and identification of possible precursors for explosives (~10%). The other 5% of the

cases include, for example, assistance on crime scenes and sampling of suspects for the

presence of traces of explosive material.

In the period of 20092013, a total of 58 cases were submitted to the NFI involving intact

high-explosive substances, compared to 31 post-explosion cases. Four cases included

both intact explosives and post-explosion material. In total, 28 intact hand grenades were

encountered, as well as more than 70 blasting caps and close to 50 plastic explosives.

An overview of the different types of high explosives encountered in pre- and post-blast

casework is given in Fig. 1.4. This figure illustrates that the most frequently encountered


Chapter 1

high explosive in the Netherlands is PETN, followed by TNT, RDX and HMX. Note that

cases involving pyrotechnics are not included in the figure.

HE

Other

Precursors

Miscellaneous

HE

Illegalfireworks

Other

Intact explosives

Post-explosion

Fig. 1.3. Average caseload in the Explosions and Explosives group at the NFI in the period 2010-2013. The types of explosives are specified for intact explosives and post-explosion casework. HE stands for high explosives and other largely concerns pyrotechnics4.

4 IEDs based on pyrotechnic material, either home-made or originating from fireworks ingredients.

0 20 40 60 80 100

Inta

ct e

xplo

sive

sPo

st-e

xplo

sion

Occurence in NFI casework 20092013

PETN TNT

HMX RDX

AN NG

ETN HMTD

TATP UN

PA DEGDN

mixed unknown

Fig. 1.4. Frequency of encountering specific types of high-explosives in the period 20092013 in cases submitted to the NFI for intact explosives and post-explosion samples.

8 Chapter 1

Cha

pter

1

1.2 Forensic explosives analysis

The choice of analytical techniques to study an explosive material depends on its nature

(bulk material or residues) and on its type (organic or inorganic). Beveridge [3] and

Yinon [4] provide overviews of the analytical techniques used. Various techniques can

be used for the analysis of bulk explosive material. Infrared and Raman spectroscopy

can be employed for the identification of both inorganic and organic materials [5]

either in the laboratory or in the form of portable instruments for on-site detection of

explosives [6]. X-ray diffraction (XRD) can also be used for the analysis of inorganic and

organic materials, provided that the material is crystalline. Information on the elemental

composition of inorganic intact explosives can be obtained by X-ray fluorescence (XRF).

In case of material that is suspected to be of pyrotechnic nature, small-scale burn tests

can be performed to test whether the ingredients are mixed in the appropriate ratio to

yield an effective pyrotechnic composition. In addition, the burning behavior of the

material gives an indication of its chemical composition.

Analysis of residues from organic explosives is usually performed using high-

performance liquid chromatography (HPLC) or gas chromatography (GC). Various

detection methods have been employed, such as photo-diode array (PDA) detection

for HPLC and the thermal-energy analyzer (TEA) [7], a detector exhibiting a high

selectivity and high sensitivity for nitro and nitroso groups, and the electron-capture

detector (ECD) for GC analysis. However, current research mostly relies on mass

spectrometric (MS) detection, as it provides high sensitivity, high selectivity and

structural information. GC instrumentation is more widely available than LC in many

forensic laboratories due to lower costs, faster analyses, robustness, and easier coupling

with MS, but it operates at higher temperatures, rendering the analysis of thermolabile

explosives (e.g. RDX, HMX and PETN) difficult [4,8-10]. Therefore, LCMS is

generally the method of choice for identification of organic explosives [11,12]. Residues

of inorganic explosives are best analyzed using ion chromatography (IC) [13,14],

offering high sensitivity and repeatability, or capillary electrophoresis (CE), providing

short analysis times and application in portable devices [15]. Because these techniques


Chapter 1

are based on complementary separation mechanisms, they are often used in parallel

for confirmation of the identification of inorganic explosives [13,14]. It should be noted

that interpretation of analytical results involving inorganic residues is complicated by

the natural occurrence of these materials. Scanning electron microscopy with energy

dispersive X-ray spectroscopy (SEMEDS) has been employed for the elemental

analysis of pyrotechnic residues [16,17]. Although the aforementioned techniques are

routinely used in forensic laboratories, recent research explored the use of techniques

such as laser-induced breakdown spectroscopy (LIBS) [18], direct analysis in real

time MS (DARTMS) [19] and secondary-ion MS (SIMS) [20] for the detection and

identification of explosives.

All of the aforementioned approaches focus on identification of the explosive material.

A major trend emerging in forensic science in general is individualization, which is

necessary when investigating relationships between material recovered from a crime

scene (e.g. an explosive device) and material associated with a suspect (e.g. explosive

material or raw materials). Basic identification does not suffice to study such relationships.

Chemical profiling can be an important tool for individualization of the material.

1.3 Chemical profiling in forensic science

According to Inman and Rudin [21], individualization is the ultimate goal of a forensic

examination. To achieve individualization, properties need to be determined that are

characteristic for a common source (e.g. a chemical profile). In casework, the evidential

value of matching characteristics can be assessed according to Bayes theorem using

likelihood ratios (LR). Bayes theorem is described in detail by Aitken and Taironi [22],

Berger [23] and Inman and Rudin [21]. This theorem indicates that true individualization

does not exist in forensic science as there is never absolute certainty on the existence of

a relationship between materials. However, the term illustrates the quest for chemical

profiling methods that establish strong support for such a relationship. The probability

of the evidence (e.g. a matching profile) needs to be viewed in light of two hypotheses:

(1) the questioned sample (i.e. crime-scene sample) originates from the same source as

10 Chapter 1

Cha

pter

1

the reference sample (i.e. suspects sample) and (2) the questioned sample originates

from another source. To determine the likelihood of a matching profile, background

information is needed on the occurrence of typical profiles for both hypotheses. This

means that samples originating from the same source should be studied to determine the

within-source variation, and samples from different sources to establish the between-

source variation. To accurately model the within- and between-source variations,

sufficiently large sample sets are required. Considering the relatively low numbers

of cases and amounts of explosives in circulation, chemical profiling of explosives is

challenging. Chemical profiling is more advanced in other forensic disciplines, such as

drug analysis, where the caseload and sample throughput are higher. In addition, most

types of drugs are organic substances that are volatile, or can be made volatile through

derivatization, and can therefore be analyzed using a single analytical technique (i.e. gas

chromatography), in contrast to explosives as described in section 1.2. Individualization

using GC-based impurity-profiling methods has been achieved for different types of

drugs, mainly for MDMA (3,4-methylenedioxy-N-methylamphetamine, known as

ecstasy or XTC) [24] and other amphetamines [25], but also, for example, for heroin

[26] and cocaine [27]. In addition, impurity profiles proved to be useful for deducing the

synthetic route and type of precursors used, implying that impurity profiling cannot only

be used as evidence in court, but also for intelligence purposes [28]. This illustrates the

importance of using knowledge on production processes and distribution of the material

to understand and interpret chemical profiles.

As described above, individualization deals with establishing the source or origin of

questioned samples. The hypotheses are therefore constructed at source level. However,

according to the hierarchy of propositions [21], hypotheses can be proposed at two other

levels, i.e. activity level and offense level. Besides individualization, moving towards

evaluation of the evidence on the activity level is another major trend in contemporary

forensic science. An interesting example is shown in the area of DNA analysis. Matching

DNA profiles are routinely used to define the source of a biological trace. The recent

development of RNA typing enables determination of the cell type, thus providing


Chapter 1

valuable information regarding the activity that led to the suspects DNA on the crime

scene or evidence items [29].

1.4 Chemical profiling of explosives

Several studies have been aimed at chemical profiling of explosives. The majority of these

studies focused on the use of isotope-ratio mass spectrometry (IRMS). Small variations

occur in the isotopic composition of materials based on their origin and history. Based

on the isotopic composition it was possible to discriminate between different batches

of both organic and inorganic explosives, such as Semtex [30], triacetone triperoxide

(TATP) [31], pentaerythritol tetranitrate (PETN) [31,32], black powder, flash powder,

nitromethane, plastic explosive no. 4 (PE4) [33], trinitrotoluene (TNT) [32-34] and

ammonium nitrate [32,35]. In addition, it was shown that the isotopic signatures of the

explosives cyclotrimethylenetrinitramine (RDX) [36] and urea nitrate [37] depend on

the isotopic composition of the precursors. This information can be used to establish a

chemical link between an explosive from a crime scene and precursors seized from a

possible suspect.

Besides the isotopic signatures that may vary based on the precursors used and the

synthesis mechanisms, other characteristics of explosive samples are impurities that

can be introduced into the sample during synthesis, are formed due to degradation over

time, or result from contamination in the presence of other materials. TNT samples

from different origins showed varying by-product profiles based on the presence of

dinitrobenzene, dinitrotoluene, trinitrobenzene and trinitrotoluene isomers detected

using LCMS [38]. For the home-made explosives HMTD and TATP it was observed

that impurities present in improvised sources of the precursor hydrogen peroxide (e.g.

hair-bleaching products) could be detected using GCMS after the synthesis of the home-

made explosives HMTD and TATP [39]. These impurities could even be detected after

initiation of the HMTD and TATP, giving an indication of the type of precursors used for

production of the explosive device. Benson et al. [35] studied the possibility to link post-

blast ammonium nitrate (AN) to the intact AN used for the explosion using IRMS. Post-

12 Chapter 1

Cha

pter

1

blast AN was found to be more enriched in 15N compared to pre-blast AN, but there was

no clear relationship between the post and pre-blast samples. Relationships between the

isotopic composition of post-explosion samples and the corresponding intact explosive

could also not be established for different types of organic explosives [40]. McGuire et al.

[40] suggested that post-explosion samples originating from aromatic explosives could

be distinguished from non-aromatic explosives based on their carbon isotope ratios,

but this can also easily be accomplished using, for example, LCMS. Post-explosion

profiling is difficult because of possible interference by environmental contaminants.

In addition, as far as isotope analysis of post-explosion material is concerned, isotope

exchange with the atmosphere may occur. Isotope fractionation during an explosion is

difficult to predict and is likely to be irreproducible, because of the uncontrolled nature

of an explosion.

As discussed above, most of the studies concerning the chemical profiling of explosives

focus on comparison of different batches of the explosive. Knowledge on the chemical

processes used in the manufacturing of explosives is important to understand and

interpret the chemical profiles. Most of the studies used sample sets consisting of only

a few samples. This limits these studies in determining the evidential value of eventual

inferences.

1.5 Scope of this thesis

This thesis focuses on novel applications of chemical profiling of explosives for

forensic purposes. This includes development of analytical methods for profiling as

well as evaluation of the results from a forensic point of view. Various approaches are

considered with respect to utilizing sufficiently large sample sets to establish or indicate

evidential values of matching profiles, evidence evaluation at activity level and studying

relationships between home-made explosives and their associated precursors.

In Chapter 2 we describe the use of vacuum-outlet GCMS for impurity profiling of

TNT to discriminate between different sources of the explosives. This approach is


Chapter 1

similar to the approach used for illicit drugs as described in section 1.3. Depending

on the synthesis and clean-up procedures and possible degradation, impurities can be

introduced in TNT, resulting in characteristic profiles for samples with different origins

and histories. Vacuum-outlet GCMS provides short analysis times and is promising for

the analysis of thermolabile explosives because of low elution temperatures. Studying

the variation in impurity profiles between samples from the same origin and between

samples from different origins allows assessing the evidential value of matching profiles

in terms of likelihood ratios.

Chapter 3 focuses on profiling of PETN and its common degradation products

pentaerythritol trinitrate (PETriN), pentaerythritol dinitrate (PEDiN) and pentaerythritol

mononitrate (PEMN). In NFI casework, LCMS had been used to identify PETN and

its degradation products on the clothing of a suspect allegedly involved in a series of

safe crackings. As the suspect denied his involvement it was questioned whether the

identified degradation products were a result of a PETN explosion. In this perspective,

differences between post-explosion PETN, intact PETN and naturally-degraded PETN

are studied using LCMS. It is demonstrated that this information can be used to

investigate possible crime-scene presence of a suspect and therefore to evaluate the

evidence on activity level. However, the unavailability of commercial standards of the

degradation products of PETN rules out accurate quantitative analysis. This latter issue

is addressed in Chapter 4, where we describe accurate quantitation based on the use of

custom-made standards. This renders the discrimination between post-explosion PETN

and naturally degraded PETN more robust.

The relationship between the isotopic composition of the explosive HMTD and its

precursors are studied using isotope-ratio MS in Chapter 5. The general availability of

precursors makes HMTD popular as home-made explosive. Several batches of HMTD are

synthesized with precursors from different origins. In addition, the influence of varying

synthesis conditions (relevant to more or less realistic crime-scene circumstances) on

the isotopic composition of HMTD is investigated.

14 Chapter 1

Cha

pter

1

In Chapter 6 isotope and elemental analysis were combined for the chemical profiling

of ammonium nitrate (AN). AN is popular for application in explosive devices, because

of its explosive properties and it is widely available as fertilizer. The geographical

origin of AN may be reflected in its isotopic composition, whereas raw materials used

in the manufacturing of AN may result in characteristic elemental profiles. Therefore,

IRMS is employed to determine the isotopic composition of AN and the trace-elemental

composition of AN is studied using inductively-coupled plasma with MS detection

(ICPMS). In addition, the potential of laser-ablation ICPMS for elemental profiling of

single AN granules is investigated.

The chapters of this thesis have been prepared for publication in international scientific

journals and can be read independently. Therefore, some overlap may occur.


Chapter 1

References[1] E.G. Mahadevan, Ammonium Nitrate Explosives, in: Ammonium Nitrate Explosives for Civil

Applications, Wiley-VCH, Weinheim, 2013.

[2] J. Akhavan, Royal Society of Chemistry, The Chemistry of Explosives, Royal Society of Chemistry 2004.

[3] A. Beveridge, Forensic Investigation of Explosions, 2nd ed., CRC Press, Boca Raton, FL, 2012.

[4] J. Yinon, S. Zitrin, Modern Methods and Applications in Analysis of Explosives, Wiley, Chichester, 1996.

[5] M. Lpez-Lpez, C. Garca-Ruiz, Trends Anal. Chem. 54 (2014) 36-44.

[6] D. Moore, R.J. Scharff, Anal. Bioanal. Chem. 393 (2009) 1571-1578.

[7] D.H. Fine, F. Rufeh, D. Lieb, D.P. Rounbehler, Anal. Chem. 47 (1975) 1188-1191.

[8] A.J. Bednar, A.L. Russell, C.A. Hayes, W.T. Jones, P. Tackett, D.E. Splichal, T. Georgian, L.V. Parker, R.A. Kirgan, D.K. MacMillan, Chemosphere. 87 (2012) 894-901.

[9] M.E. Walsh, Talanta. 54 (2001) 427-438.

[10] J.M. Perr, K.G. Furton, J.R. Almirall, Talanta. 67 (2005) 430-436.

[11] J. Yinon, Analysis of Explosives by LC/MS, in: Yinon J. (Ed.), Advances in Forensic Applications of Mass Spectrometry, CRC Press, Boca Raton, FL, 2003.

[12] X. Xu, M. Koeberg, C. Kuijpers, E. Kok, Sci. Justice. 54 (2014) 3-21.

[13] C. Johns, R.A. Shellie, O.G. Potter, J.W. OReilly, J.P. Hutchinson, R.M. Guijt, M.C. Breadmore, E.F. Hilder, G.W. Dicinoski, P.R. Haddad, J. Chromatogr. A. 1182 (2008) 205-214.

[14] B. Stuart, Separation Techniques, in: Forensic Analytical Techniques, Wiley, Chichester, 2012.

[15] J.P. Hutchinson, C.J. Evenhuis, C. Johns, A.A. Kazarian, M.C. Breadmore, M. Macka, E.F. Hilder, R.M. Guijt, G.W. Dicinoski, P.R. Haddad, Anal. Chem. 79 (2007) 7005-7013.

[16] S.A. Phillips, Sci. Justice. 41 (2001) 73-80.

[17] K.L. Kosanke, R.C. Dujay, B.J. Kosanke, J. Forensic Sci. 51 (2006) 296-302.

[18] J.L. Gottfried, F.C. Lucia Jr, C.A. Munson, A.W. Miziolek, Anal. Bioanal. Chem. 395 (2009) 283-300.

[19] J.M. Nilles, T.R. Connell, S.T. Stokes, H. Dupont Durst, Propellants Explos. Pyrotech. 35 (2010) 446-451.

[20] H. Tllez, J.M. Vadillo, J.J. Laserna, Rapid Comm. Mass Spectrom. 26 (2012) 1203-1207.

[21] K. Inman, N. Rudin, Classification, Identification, and Individualization - Inference of Source, in: Inman K., Rudin N. (Eds.), Principles and Practice of Criminalistics: The Profession of Forensic Science, CRC Press, Boca Raton, FL, 2002.

16 Chapter 1

Cha

pter

1

[22] C. Aitken, F. Taroni, Statistics and the Evaluation of Evidence for Forensic Scientists, 2nd ed., Wiley, Chichester, 2004.

[23] C.E.H. Berger, B. Robertson, G.A. Vignaux, Interpreting Scientific Evidence, in: Freckleton I., Selby H. (Eds.), Expert Evidence, LBC, Sydney, 2012.

[24] R.J. Waddell-Smith, J. Forensic Sci. 52 (2007) 1297-1304.

[25] K. Andersson, E. Lock, K. Jalava, H. Huizer, S. Jonson, E. Kaa, A. Lopes, A. Poortman-van der Meer, E. Sippola, L. Dujourdy, J. Dahln, Forensic Sci. Int. 169 (2007) 86-99.

[26] D.R. Morello, S.D. Cooper, S. Panicker, J.F. Casale, J. Forensic Sci. 55 (2010) 42-49.

[27] J. Casale, R. Waggoner, J. Forensic Sci. 36 (1991) 1312-1330.

[28] N. Stojanovska, S. Fu, M. Tahtouh, T. Kelly, A. Beavis, K.P. Kirkbride, Forensic Sci. Int. 224 (2013) 8-26.

[29] A. Lindenbergh, P. Maaskant, T. Sijen, Forensic Sci. Int. Genet. 7 (2013) 159-166.

[30] G. Pierrini, S. Doyle, C. Champod, F. Taroni, D. Wakelin, C. Lock, Forensic Sci. Int. 167 (2007) 43-48.

[31] S.J. Benson, C.J. Lennard, P. Maynard, D.M. Hill, A.S. Andrew, C. Roux, Sci. Justice. 49 (2009) 81-86.

[32] D. Widory, J. Minet, M. Barbe-Leborgne, Sci. Justice. 49 (2009) 62-72.

[33] D. Wakelin, Isotope Ratio Analysis of Explosives - A New Type of Evidence, in: Proceedings of the 7th International Symposium on Analysis and Detection of Explosives, 2001.

[34] A. Nissenbaum, J. Forensic Sci. 20 (1975) 455-459.

[35] S.J. Benson, C.J. Lennard, P. Maynard, D.M. Hill, A.S. Andrew, C. Roux, Sci. Justice. 49 (2009) 73-80.

[36] C.M. Lock, W. Meier-Augenstein, Forensic Sci. Int. 179 (2008) 157-162.

[37] R. Aranda IV, L.A. Stern, M.E. Dietz, M.C. McCormick, J.A. Barrow, R.F. Mothershead II, Forensic Sci. Int. 206 (2011) 143-149.

[38] X. Zhao, J. Yinon, J. Chromatogr. A. 946 (2002) 125-132.

[39] A. Partridge, S. Walker, D. Armitt, Aust. J. Chem. 63 (2010) 30-37.

[40] R.R. McGuire, C.A. Velsko, C.G. Lee,E. Raber, The use of post detonation analysis of stable isotope ratios to determine the type and production process of the explosive involved, UCRL-ID-113275, 1993.

Chapter 2

Impurity profiling of trinitrotoluene using vacuum-outlet GCMS

H. Brust, S. Willemse, T. Zeng, A. van Asten, M. Koeberg, A. van der Heijden,

A. Bolck, P. Schoenmakers

Manuscript in preparation

18 Chapter 2

Cha

pter

2

Abstract

In this chapter, a reliable and robust vacuum-outlet gas-chromatographymass-

spectrometry (GCMS) method is introduced for the identification and quantification

of impurities in trinitrotoluene (TNT). Vacuum-outlet GCMS allows for short analysis

times; the analysis of impurities in TNT was performed in four minutes. Impurities may

be introduced during or after synthesis of TNT and through degradation. This results

in a characteristic profile for samples with different origins and histories. This study

shows that impurity profiling of TNT can be used to investigate relations between TNT

samples encountered in forensic casework. A wide variety of TNT samples was analyzed

with the developed method. Dinitrobenzene, dinitrotoluene, trinitrotoluene and amino

dinitrotoluene isomers were detected at very low levels (< 1 wt%) by applying the MS

in selected-ion monitoring (SIM) mode. Impurity profiles based on seven compounds

showed to be useful for discrimination between TNT samples from different sources.

Statistical analysis of these impurity profiles using likelihood ratios demonstrated the

potential to investigating whether two questioned TNT samples encountered in forensic

casework are from the same source.

Impurity profiling of TNT using vacuum-outlet GC-MS 19

Chapter 2

2.1 Introduction

Currently, forensic explosives analysis mainly focuses on the identification of the

explosive that is encountered in casework. However, more information can be retrieved

from explosives samples by chemical profiling. Profiling explosives may result in more

knowledge on the origin of the explosive and the characteristic profiles obtained may be

used for investigating possible relationships between crime-scene samples and samples

obtained from suspects. In this chapter, we present the use of vacuum-outlet GCMS

for the profiling of impurities (including chemically-related additives) in trinitrotoluene

(TNT) and we assess its potential to discriminate between different sources of TNT.

In forensic explosives analysis, liquid chromatographymass spectrometry (LCMS)

is the method of choice for identification of thermolabile explosives [1]. LCMS is

operated at moderate (often ambient) temperatures. GC requires elevated temperatures

(typically 100300C). However, GCMS is more readily available in forensic

laboratories, due to lower costs, faster analyses, and greater ease of use. In addition,

the development of portable GCMS instruments is promising for on-site detection and

identification of explosive compounds. This has been demonstrated by Bednar et al.

[2] for the identification of TNT and related compounds in groundwater. Several other

studies also employed GCMS for the analysis of TNT and related compounds [2-5]. In

contrast to TNT, thermolabile explosives such as RDX, HMX and PETN often showed

signs of degradation during GC analysis [2,4,6,7]. These compounds exhibit thermal

degradation at temperatures lower than those required for elution from a conventional

GC system.

2.1.1 Vacuum-outlet GCMSVacuum-outlet GCMS, also referred to as low-pressure GCMS [8-10], is characterized

by the direct coupling of a short wide-bore analytical column (5-10 m in length and an

internal diameter of 0.32-0.53 mm) to a mass spectrometer. In this way, the analytical

column is operated at reduced pressure along its entire length, hence the name vacuum-

outlet GCMS [11-13]. From theory it is estimated that under the conditions in the present

20 Chapter 2

Cha

pter

2

work the pressure at the inlet of the analytical column is below 30 kPa. The principles

of this concept have been described in detail by Cramers and coworkers [11,14,15].

Low pressures in the analytical column result in higher optimum linear velocities

compared with atmospheric-outlet GC (i.e. conventional GC and GCMS conditions).

Therefore, the analysis times are significantly shortened [11,14,16]. In several studies a

reduction in analysis time by a factor of 3 to 10 is reported [8,9,11-13]. The increased

linear velocities also result in narrower peaks and improved signal-to-noise ratios and

thus to lower detection limits [9,11,13,17]. Reduced pressure across the entire length of

the analytical column would require injections at sub-atmospheric pressures in a fast

stream of carrier gas. To allow the use of conventional sample-injection techniques, a

narrow-bore deactivated column is inserted between the injector and the GC column

[12,13]. In addition to faster separation, vacuum-outlet GCMS offers more advantages

over atmospheric-outlet GC analysis. Lower elution temperatures have been reported

for vacuum-outlet GCMS [13], which make it possible to elute thermally-labile and

high-boiling compounds at moderate temperatures. Lower elution temperatures and

minimized thermal degradation were reported, for example, for the analysis of pesticides

[9,17] and polybrominated diphenyl ethers [10]. This suggests potential benefits of using

vacuum-outlet GCMS for analysis of thermolabile explosives. The use of a wide-

bore column also results in an increased sample capacity, which is beneficial for trace

analysis. On the other hand, wide-bore columns exhibit increased plate heights and

thus reduced separation efficiency. In addition, the large relative pressure drop (i.e. Pin/

Pout) along the column length results in a strong decompression of the carrier gas. A

resulting loss in efficiency of 12.5% was calculated if identical columns were to be

operated under vacuum conditions compared to atmospheric conditions [11]. However,

no such loss in efficiency was observed experimentally. Finally, the use of short columns

limits the attainable number of theoretical plates. This renders vacuum-outlet GCMS

less attractive for the separation of complex mixtures. However, it appears a powerful

approach for the fast analysis of samples containing a limited number of compounds.


Chapter 2

2.1.2 Impurities in TNT2,4,6-TNT is manufactured by stepwise nitration of toluene using a mixture of nitric

acid and sulfuric acid [18]. Intermediate products are mononitrotoluene (NT) and

dinitrotoluene (DNT) isomers. In addition to TNT, DNT and NT isomers, other

compounds, such as tetranitromethane, trinitrobenzoic acid, trinitrobenzene (TNB),

and di- and trinitrocresols can be formed [7]. The majority of the synthesis by-products

are removed by washing with a sodium sulphite solution [7,18]. Remaining impurities

that were detected in military grade TNT include 1,3-DNB (dinitrobenzene), 1,3,5-TNB

and several DNT and TNT isomers [19-21]. Amino dinitrotoluenes (A-DNT) can also

be detected in TNT, although they are usually a result of microbial degradation in soil

[6,19,22]. The potential of chemical profiling of TNT has been the subject of several

studies, either based on impurities and chemically-related additives using GC analysis

[19,20] or LCMS analysis [21], or based on isotope ratios [23,24]. However, the sample

sets used for these studies were small, making it difficult to assess whether origin

determination and individualization were actually possible.

2.2 Experimental

2.2.1 Analytical standards and chemicalsAnalytical standards of TNT and TNT-related compounds (listed in Table 2.1) were

obtained from AccuStandard (New Haven, CT, USA), except for 2,3,4-TNT (Dr.

Ehrenstorfer, Augsburg, Germany). Quantitation of impurities in TNT samples was

performed using calibration solutions ranging from 0.05 to 1.5 g/mL. All dilutions

were made using acetonitrile (Biosolve, Valkenswaard, The Netherlands or Rathburn,

Walkerburn, UK). Methyl decanoate (Sigma-Aldrich, Steinheim, Germany) was added

as an internal standard (IS) to all samples and standards at a concentration of 0.5 g/mL.

Calibration of 2,3,4-TNT was separately performed in the range of 0.3 to 7.5 g/mL

without the use of an internal standard and dilutions were made with cyclohexane

(Rathburn, Walkerburn, UK) as the commercial standard was a cyclohexane solution.

Although chemically not similar to TNT, methyl decanoate proved to be stable under

the analysis conditions and it did not coelute with any of the compounds of interest.

22 Chapter 2

Cha

pter

2

Therefore, it served the purpose as an internal standard to correct for variations in

sample preparation and injection volume.

2.2.2 Sample setA total of 10 TNT samples were obtained from TNO (Rijswijk, The Netherlands).

In addition, 41 TNT samples were obtained from the Netherlands Forensic Institute

(NFI, The Hague, The Netherlands). Most of these samples were retrieved from cases

previously dealt with. For these samples limited background information was available.

Three NFI samples were from the same origin as three samples obtained from TNO.

Samples were prepared for GCMS analysis by dissolving them in acetonitrile at a

concentration of 4 mg/mL. High TNT concentrations were necessary for detection of the

low-level impurities. Sample solutions were prepared at 2 mg/mL and methyl decanoate

was added as IS at a concentration of 0.5 g/mL, similar to the levels of impurities in

the TNT samples.

Table 2.1. Retention times and selected ions for all compounds. Target ions were used for quantification. For methyl decanoate and both A-DNT isomers the sum of three ions was used to enhance signal-to-noise ratios.

Compounds Retention time (min) Selected ions (m/z) Target ion(s) (m/z)2-NT 2.203 0.006

65, 91, 120, 137120

3-NT 2.290 0.010 1374-NT 2.329 0.011 137Methyl decanoate 2.550 0.002 87, 143, 155 87, 143,1551,3-DNB 2.767 0.006

63, 122, 165, 168168

2,6-DNT 2.786 0.003 1651,2-DNB 2.802 0.005 1682,5-DNT 2.857 0.004

89, 119, 165, 182

1652,4-DNT 2.924 0.004 1653,5-DNT 2.958 0.003 1823,4-DNT 3.021 0.003 1822,4,6-TNT 3.232 0.006 167, 213 -*

2,3,4-TNT 3.405 0.006 134, 180, 210 1344-A-2,6-DNT 3.580 0.009

104, 180, 197 104, 180, 1972-A-4,6-DNT 3.674 0.010

* m/z 167 and 213 were used for qualitative monitoring of the signal.


Chapter 2

2.2.3 GCMS analysisAll samples were analyzed using a 6890N GC system coupled to a 5973N mass

spectrometer (both Agilent Technologies, Palo Alto, CA, USA). Samples were injected

using an Agilent 7683 automatic liquid sampler with an injection volume of 1 L,

and split injection at a temperature of 250C and a split ratio of 1:10. Vacuum-outlet

conditions were achieved by coupling a wide-bore Rtx-TNT analytical column (6 m

0.53 mm 1.5 m) (Restek, Bellefonte, PA, USA) to the MS. A 1 m 0.1 mm deactivated

restriction column (Restek) was inserted at the injector side of the system and coupled to

the analytical column using a Meltfit connector (NLISIS Chromatography, Veldhoven,

The Netherlands). Helium was used as the carrier gas and the system was operated in

constant-pressure mode at an inlet pressure of 276 kPa. The GC oven was held at 35C

for 1 min, then ramped to 250C at 80C/min and finally held at 250C for 1 min. The

MS was operated in selected-ion monitoring (SIM) mode using electron ionization (EI),

with a dwell time of 10 ms. Retention times and monitored ions for all compounds are

shown in Table 2.1.

2.2.4 Calculations of likelihood ratiosTo evaluate whether the impurity profiles can be used to discriminate between TNT

samples from different batches or to link TNT samples from the same batch we have

used the likelihood ratio (LR) approach. The LR approach is often used in forensic

science as a measure for the strength of evidence whether items come from the same

source or not. In general the LR is:

LR P E HP E Hp

d= ( )( ) (2.1)

Were E represents the evidence, or more specifically the impurity profiles. Finding this

specific evidence is evaluated under two competing hypotheses often associated with

the prosecution (Hp) and the defence (Hd):

Hp: The two samples are from the same batch

Hd: The two samples are from different batches.

24 Chapter 2

Cha

pter

2

LR values of around 1 represent neutral evidence. High LR values support the hypothesis

that samples are from the same source and low LR values (less than 1) support Hd.

LR values were calculated based on two different types of models, i.e. a score-based

model and a feature-based model. Score-based models use a score for the (dis)similarity

between the profiles of two samples, such as a similarity index or distance measure as

the evidence, while feature-based models use profiles, or at least the concentrations of

all compounds, of the two samples to be compared as the evidence. Here, scores were

calculated using Pearson-correlation distances between the levels of seven compounds

in two TNT samples. Distributions of the Pearson-correlation distances for samples

from the same source and for samples from different sources were both smoothed using

kernel-density estimation (KDE) (as in [25,26]). The drawback of score-based methods

is the reduction of the, in this case, seven variable values to a single score, resulting in

a loss of information concerning the variation in the concentrations of the individual

compounds, their correlations and the rarity of all this. This is likely to result in LR

values representing low evidential strength. Multivariate comparisons can be performed

using feature-based models. Here, the LR is based on the multivariate information of

two profiles and their rarity resulting in often large LR values. However, feature-based

models require many data for robust covariance estimations and, with that, robust LR

calculations. The feature-based model used here is a so-called two-level random-effect

model [27] and assumes normality within a single TNT batch. The distribution of

samples from different batches were again estimated using KDE. LR calculations were

performed using the software R1.

2.3 Results and discussion

2.3.1 Method developmentThe optimal inlet pressure was determined by constructing a curve relating the observed

plate height (H) to the linear gas velocity (u) (Fig. 2.1). At a given carrier gas pressure

u was measured by injection of air and H was determined for a 2,4,6-TNT standard at

a column temperature of 180C. The TNT standard was injected in triplicate and the

1 Available on http://cran.r-project.org.

http://cran.r-project.org


Chapter 2

average width at half height of the detected peak was determined. The lowest plate

heights were reached at linear velocities around 0.8 m/s. Higher linear velocities could

not be obtained due to the limited pumping capacity of the MS system. To perform robust

analysis well within the specifications of the instrument we chose a linear velocity of

0.78 m/s for further experiments. This corresponded to an inlet pressure of 276 kPa. The

observed minimum plate height of 1.1 mm corresponds to N = 5500 for TNT at 180C

(retention factor k = 5) on the given analytical column.

0

1

2

3

0.4 0.6 0.8 1

H(m

m)

(m/s)

Fig. 2.1. H-u curve obtained for the vacuum-outlet GCMS system. Analyte 2,4,6-TNT; oven temperature 180C.

The effect of different temperature ramps (20, 30, 40, 60, and 80C/min) on the elution

temperature of 2,4,6-TNT was investigated next. The observed elution temperatures

were 164, 174, 178, 189 and 199C, respectively. Elution temperatures below 164C may

be achieved with temperature ramps lower than 20C/min. However, this significantly

increased the analysis time. It was also observed that elution temperatures decreased

with increasing flow rates, but, again, the maximum flow rate was limited by the MS

system. These results show that to exploit the lower elution temperatures offered by

vacuum-outlet GCMS moderate to slow temperature-programming rates have to be

used. However, since 2,4,6-TNT and related compounds are not thermolabile, the elution

temperatures do not need to be minimized in the present application. Fast temperature

programming resulted in narrower peaks, improved signal-to-noise ratios and higher

sample throughput, while maintaining sufficient chromatographic resolution. Therefore,

26 Chapter 2

Cha

pter

2

a temperature ramp of 80C/min was selected for the analysis of impurities in TNT,

resulting in typical peak widths at half height of 0.80.9 s.

Using the method developed as described above a mixture of TNT-related compounds

was separated in 4 min (Fig. 2.2). This demonstrates the power of vacuum-outlet GC

MS for fast separations at moderate resolution. As can be observed from Fig. 2.2, 2,6-

DNT was not completely separated from 1,2-DNB and 1,3-DNB. However, separation

of these peaks was still possible using spectral deconvolution. Based on appropriate

ion selection in SIM mode all three compounds could be accurately quantified without

mutual interference (see inset in Fig. 2.2).

0

10

20

30

2 3 4

Abun

danc

e (

105 )

Retention time (min)

12 3

5,6

7

8

910

11

1213

14

46

4 5

Fig. 2.2. Separation of a standard mixture of TNT-related compounds (10 g/mL each, full-scan mode), containing 2-NT (1), 3-NT (2), 4-NT (3), 1,2-DNB (4), 2,6-DNT (5), 1,3-DNB (6), 2,5-DNT (7), 2,4-DNT (8), 3,5-DNT (9), 3,4-DNT (10), 1,3,5-TNB (11), 2,4,6-TNT (12), 4-A-2,6-DNT (13), 2-A-4,6-DNT (14). The inset shows the extracted-ion chromatograms of m/z 168 (gray line; 1,2-DNB and 1,3-DNB) and m/z 165 (dashed line; 2,6-DNT). Conditions as described in section 2.2.3.

For chemical impurity profiling in actual TNT samples low levels of TNT-related

impurities have to be quantitatively analyzed in an excess of TNT. High concentrations

of TNT need to be injected to enable detection of the low-level impurities. However, this

is limited by overloading phenomena that result in deterioration of peak shapes and

resolution. Acceptable chromatography was obtained for a maximum TNT concentration


Chapter 2

of 2 mg/mL. At this concentration and for typical levels of the impurities insufficient

sensitivity was obtained when operating the MS in full-scan mode. To improve the

detection of the impurities a SIM method was developed. For all compounds, except for

2,4,6-TNT, two or three of the most abundant ions were selected to obtain the lowest

possible detection limits. Because of the very high relative concentration of 2,4,6-TNT,

low-abundant ions, m/z 167 and m/z 213, were chosen in order to suppress the 2,4,6-

TNT signal as much as possible. Improved selectivity resulted in the detection of several

impurities in the actual TNT samples (Fig. 2.3). The levels of impurities were found to

0

2

4

6

2 2.5 3 3.5 4

Abun

danc

e (

104 )

1 2 3 4

5

6

7 8

9 10

0

0.5

1

1.5

2

2.5

2 2.5 3 3.5 4

Abun

danc

e (

104 )

Time (min)

1

7

8

9 10

2.7 2.9 3.1

m/z 168m/z 165m/z 182

2

34

5 6

a

b

Fig. 2.3. GCMS (SIM) chromatograms of 10 TNT samples (a) showing the presence of different impurities: methyl decanoate IS (1), 1,3-DNB (2), 2,6-DNT (3), 2,5-DNT (4), 2,4-DNT (5), 3,5-DNT (6), 2,4,6-TNT (7), 2,3,4-TNT (8), 4-A-2,6-DNT (9) and 2-A-4,6-DNT (10). Fig. 2.3b shows the detailed chromatogram for one of the samples with the extracted ion chromatograms of the target ions of 1,3-DNB and the DNT isomers in the insert. Conditions as described in section 2.2.3.

28 Chapter 2

Cha

pter

2

vary between the different samples. This indicates that TNT samples can be discriminated

based on their impurity profiles. 1,3,5-TNB was found to elute too close to 2,4,6-TNT to

be included in the profiling method.

2.3.2 Quantitation of impuritiesThe levels of impurities were determined using methyl decanoate as an internal

standard. Calibration was performed in the range of 0.05 to 1.5 g/mL, except for 2,3,4-

TNT. Based on the levels of 2,3,4-TNT in the samples, calibration was extended to

higher concentrations for this compound (0.3 to 7.5 g/mL). Good linearity across the

calibration range was observed for all compounds, with coefficients of regression R2

ranging from 0.99 for 3,4-DNT to 0.999 for 4-NT. Limits of detection (LOD) and limits

of quantitation (LOQ) for all compounds are shown in Table 2.2. LODs and LOQs were

determined as three and ten times the signal-to-noise ratio, respectively.

Table 2.2. LODs and LOQs for all compounds obtained using vacuum-outlet GCMS in SIM mode and the target ions listed in Table 2.1. LODs and LOQs are listed in units of ng/mL; LOQ values are also converted to ppm in TNT, based on a TNT concentration of 2 mg/mL.

Compound LOD (ng/mL) LOQ (ng/mL) LOQ (ppm in TNT)2-NT 9 30 153-NT 10 33 174-NT 15 50 251,3-DNB 10 32 162,6-DNT 6 21 111,2-DNB 11 38 192,5-DNT 8 26 132,4-DNT 8 27 143,5-DNT 18 60 303,4-DNT 17 58 292,3,4-TNT 24 81 414-A-2,6-DNT 43 142 712-A-4,6-DNT 41 137 69


Chapter 2

Table 2.3. Recovery of analytes spiked in a TNT sample at 500 ng/mL of each compound.

Compound Concentration (ng/mL) Recovery (%)2-NT 510 1023-NT 501 1004-NT 520 1041,3-DNB 577 1152,6-DNT 575 1151,2-DNB 535 1072,5-DNT 527 1052,4-DNT 484 973,5-DNT 559 1123,4-DNT 516 1034-A-2,6-DNT 699 1402-A-4,6-DNT 675 135

To evaluate the accuracy of determining low levels of impurities in real TNT samples,

one of the TNT samples containing relatively low levels of impurities was spiked with

a standard mixture resulting in a concentration of 500 ng/mL of each compound. This

experiment resulted in the recoveries listed in Table 2.3. The recoveries for all compounds

were between 97% and 115%, except for both amino dinitrotoluene (A-DNT) isomers,

which show a recovery of 135140%. A reason for this could be that the A-DNTs elute

after 2,4,6-TNT and that the target-ion sensitivity is affected by the presence of 2,4,6-

TNT in the mass spectrometer.

2.3.3 Analysis of TNT samplesAll 51 TNT samples were analyzed for the presence of impurities. The vacuum-outlet

GCMS method was applied in full-scan mode for possible unknown impurities

and with the developed SIM method for accurate quantification of the TNT-related

impurities. The levels of impurities were expressed as ppm (mg/kg) in the original TNT

sample. In full-scan mode, 2,4-DNT and 2,3,4-TNT were the only impurities that were

detected in several samples. In SIM mode, 2,4-DNT and both A-DNT isomers were

detected in all samples, whereas the NT isomers and 1,2-DNB were not detected. 1,3-

DNB was detected in 40 of the 51 samples, 2,6-DNT in 46 samples, 2,5-DNT in 25,

30 Chapter 2

Cha

pter

2

3,5-DNT in 36, 3,4-DNT in 21 and 2,3,4-TNT in 39 samples. All of the impurities were

present at levels well below 1 wt% indicating the high purity of the TNT samples. To

estimate the within-sample variation, one of the samples was prepared and analyzed ten

times. The results shown in Table 2.4 indicate that the coefficient of variation (CV) of

the vacuum-outlet GCMS method in SIM mode ranged from 2% to 13% for the TNT-

related impurities. This within-sample variation included sample homogeneity, sample

preparation and GCMS analysis.

Table 2.4. Detected levels of impurities and associated repeatability for one of the TNT samples (n = 10).

Compound ppm in TNT %CV1,3-DNB 65 8.02,6-DNT 65 6.02,5-DNT nd2,4-DNT 263 9.93,5-DNT 49 2.03,4-DNT nd2,3,4-TNT 259 12.54-A-2,6-DNT 151 10.52-A-4,6-DNT 134 8.9

0

500

1000

1500

2000

2500

3000

1 2 3 4 5 6 7 8 9 10 11

Con

cent

ratio

n (p

pm in

TN

T)

Sample number

1,3-DNB

2,6-DNT

2,5-DNT

2,4-DNT

3,5-DNT

2,3,4-TNT

4-A-2,6-DNT

2-A-4,6-DNT

0

150

300

Fig. 2.4. Impurity profiles for different TNT samples. The inset shows an enlargement of sample 1, which was used to study the within-sample variation (the error bars represent 1).


Chapter 2

Substantial variation was observed between impurity profiles of different TNT samples.

An example is given in Fig. 2.4. In the majority of the samples, 2,4-DNT and 2,3,4-TNT

were detected as the major impurities. Both A-DNT isomers were detected at roughly

the same levels in all samples (157 26 ppm in TNT for 4-A-2,6-DNT and 145 20

ppm for 2-A-4,6-DNT). For these analytes the variation between different samples

was comparable to the within-sample variation (see Table 2.4) and the recoveries were

unrealistically high, possibly due to interference of TNT (see Table 2.3). Therefore, the

A-DNT isomers were excluded from the impurity profiles.

2.3.4 Sample comparisons using likelihood ratiosLikelihood ratios were calculated to evaluate whether the impurity profiles can be used

to discriminate between TNT samples from different batches or to link TNT samples

from the same source. To examine the effect of using a different number of compounds,

LR values were calculated using the feature-based model for one compound (2,4-DNT),

for two compounds (3,5-DNT and 2,3,4-TNT), for three compounds (2,4-DNT, 3,5-

DNT and 2,3,4-TNT), and for all seven compounds (as listed in Table 2.4, without the

A-DNTs).

All 60 samples were compared with each other, without comparing more than one

sample from a single source with another source and comparing of samples with itself

to avoid double comparisons, using the four feature-based models and the score-based

model. LR values were calculated for each comparison. This resulted in a total of 1176

comparisons, of which 48 comparisons concerned samples from the same source and

1128 concerned samples from different sources. The distribution of LR values for these

comparisons using the feature-based methods and the score based method is shown in

Fig. 2.5.

Apart from the sample that was analyzed ten times, three pairs of samples (obtained

from the NFI and TNO) shared the same origin. Calculated LR values for these three

duplicates are shown in Table 5, as well as measures for misleading evidence, that are the

false-positive and false-negative rates and the log-likelihood-ratio cost (Cllr) [28]. False

32 Chapter 2

Cha

pter

2

positives represent here LR values higher than the natural value for neutral evidence of

1 for two samples that are from different sources and false negative are comparisons of

samples from the same source that returned LR values less than . The Cllr is a measure

for the validity of the LR method that gives a higher penalty to false positives with a

high LR value and false negatives with a low LR value compared to false negatives

and positives closer to one. The Cllr should be below 1 and when comparing similar LR

models a lower Cllr indicates better accuracy of the model.2

D EG DEG all 7 Pearson

-20

24

68

10

same-batch

method

10lo

g(LR

)

D EG DEG all 7 Pearson

-300

-250

-200

-150

-100

-50

0

different-batch

method

10lo

g(LR

)

ba

Fig. 2.5. LR values obtained for comparisons of samples from the same source (a) and from different sources (b) using four feature-based methods and a score-based method (Pearson-correlation distance). Feature-based LRs were calculated using one, two, three or seven compounds. Compound identification D = 2,4-DNT, E = 3,5-DNT, and G = 2,3,4-TNT. The score-based model was calculated based on the same seven compounds as the last feature-based model. Feature-based comparisons of samples from different sources often resulted in LR values rounded by the R software to 0. These have been replaced by 110-300 to allow meaningful graphical representation on a logarithmic scale.

From Fig. 2.5 and Table 2.5 it can be observed that higher absolute LR values are obtained

when more compounds are included in the feature-based model. In addition, the false-

negative and false-positive percentages and Cllr values indicate that the feature-based

method using seven compounds is the best discriminating LR method. The feature-

based methods using less compounds, especially just one compound, show similar 2 The Cllr is only given for prior odds of 1 (thus where P(Hp)=P(Hd)). So-called ECE (empirical cross-entropy) plots provide similar values for other prior odds.


Chapter 2

performance compared to the score-based method (based on the seven compounds).

However, as mentioned before, the more features are included in the feature-based model

the more data are needed to obtain robust LR calculations. The exact LR values are,

therefore, not very informative, but the orders of magnitude provide useful information

on the evidential value. The present results indicate the potential of LR calculations for

impurity profiles of TNT obtained using GCMS. All LR methods show the ability to

discriminate between samples from different sources and to link samples from the same

source.

Table 2.5. Likelihood ratios for three pairs of samples from the same source calculated with five different methods. False-positive and false-negative rates for the different methods are also given.

Feature (# compounds) Score

1 2 3 7Pearson-correlation distance

LR pair 1 3 60 1473 4.54109 13LR pair 2 15 455 10653 6.56109 41LR pair 3 10 455 9679 1.26109 10False positives (%) 0 4.9 1.2 0 11.4False negatives (%) 16.6 6.3 14.6 0 0Cllr 0.35 0.26 0.27 0.00026 0.27

2.4 Conclusions

Vacuum-outlet GCMS operates with a wide-bore column under vacuum conditions. It

is characterized by high linear velocities resulting in short analysis times. For chemical

profiling of the explosive TNT this resulted in an analysis time of 4 minutes. 51 actual

TNT samples were analyzed with the developed vacuum-outlet GCMS method.

Several impurities have been detected in these samples, such as DNT isomers, 1,3-

DNB and 2,3,4-TNT. Impurity profiles based on seven compounds showed to be useful

for discrimination between TNT samples from different sources. The impurity profiles

allowed for the calculation of likelihood ratios to assess the evidential value. This

approach can be used to study relationships between TNT from a crime scene and TNT

obtained from a suspect.

34 Chapter 2

Cha

pter

2

References[1] J. Yinon, Analysis of Explosives by LC/MS, in: Yinon J. (Ed.), Advances in Forensic Applications

of Mass Spectrometry, CRC Press, Boca Raton, FL, 2003.

[2] A.J. Bednar, A.L. Russell, C.A. Hayes, W.T. Jones, P. Tackett, D.E. Splichal, T. Georgian, L.V. Parker, R.A. Kirgan, D.K. MacMillan, Chemosphere. 87 (2012) 894-901.

[3] M.E. Sigman, C.Y. Ma, J. Forensic Sci. 46 (2001) 6-11.

[4] J.M. Perr, K.G. Furton, J.R. Almirall, Talanta. 67 (2005) 430-436.

[5] J. Yinon, J. Chromatogr. A. 742 (1996) 205-209.

[6] M.E. Walsh, Talanta. 54 (2001) 427-438.

[7] J. Yinon, S. Zitrin, Modern Methods and Applications in Analysis of Explosives, Wiley, Chichester, 1996.

[8] K. Ravindra, A.C. Dirtu, A. Covaci, Trends Anal. Chem. 27 (2008) 291-303.

[9] K. Matovsk, S.J. Lehotay, J. Hajlov, J. Chromatogr. A. 926 (2001) 291-308.

[10] A.C. Dirtu, K. Ravindra, L. Roosens, R. van Grieken, H. Neels, R. Blust, A. Covaci, J. Chromatogr. A. 1186 (2008) 295-301.

[11] C.A. Cramers, G.J. Scherpenzeel, P.A. Leclerq, J. Chromatogr. A. 203 (1981) 207-216.

[12] M. Van Deursen, H. Janssen, J. Beens, P. Lipman, R. Reinierkens, G. Rutten, C. Cramers, J. Microcolumn Sep. 12 (2000) 613-622.

[13] J. de Zeeuw, J. Peene, H. Jansen, X. Lou, J. High Resolut. Chromatogr. 23 (2000) 677-680.

[14] P.A. Leclercq, G.J. Scherpenzeel, E.A.A. Vermeer, C.A. Cramers, J. Chromatogr. A. 241 (1982) 61-71.

[15] P.A. Leclercq, C.A. Cramers, J. High Resolut. Chromatogr. 8 (1985) 764-771.

[16] J.C. Giddings, Anal. Chem. 34 (1962) 314-319.

[17] M.J. Gonzlez-Rodrguez, A. Garrido-Frenich, F.J. Arrebola, J.L. Martnez-Vidal, Rapid Comm. Mass Spectrom. 16 (2002) 1216-1224.

[18] T. Urbanski, Chemistry and Technology of Explosives. Vol. 1., Pergamon Press, Oxford, 1964.

[19] T.F. Jenkins, D.C. Leggett, P.H. Miyares, M.E. Walsh, T.A. Ranney, J.H. Cragin, V. George, Talanta. 54 (2001) 501-513.

[20] D.G. Gehring, J.E. Shirk, Anal. Chem. 39 (1967) 1315-1318.

[21] X. Zhao, J. Yinon, J. Chromatogr. A. 946 (2002) 125-132.

[22] J. Yinon, Forensic and environmental detection of explosives, Wiley, Chichester, 1999.

[23] A. Nissenbaum, J. Forensic Sci. 20 (1975) 455-459.


Chapter 2

[24] D. Wakelin, Isotope Ratio Analysis of Explosives - A New Type of Evidence, in: Proceedings of the 7th International Symposium on Analysis and Detection of Explosives, 2001.

[25] A. Bolck, C. Weyermann, L. Dujourdy, P. Esseiva, J. van den Berg, Forensic Sci. Int. 191 (2009) 42-51.

[26] M.R. Rijnders, A. Stamouli, A. Bolck, J. Forensic Sci. 55 (2010) 616-623.

[27] C.G.G. Aitken, D. Lucy, J. R. Stat. Soc. Ser. C Appl. Stat. 53 (2004) 109-122.

[28] G.S. Morrison, Science & Justice. 51 (2011) 91-98.

Chapter 3Pentaerythritol tetranitrate (PETN) profiling in post-explosion residues

to constitute evidence of crime-scene presence

This chapter has been published as:

H. Brust, A. van Asten, M. Koeberg, A. van der Heijden, C.J. Kuijpers,

P. Schoenmakers

Forensic Sci. Int. 230 (2013) 37-45

38 Chapter 3

Cha

pter

3

Abstract

Pentaerythritol tetranitrate (PETN) and its degradation products are analyzed to

discriminate between residues originating from PETN explosions and residues

obtained under other circumstances, such as natural degradation on textile, or after

handling intact PETN. The degradation products observed in post-explosion samples

were identified using liquid chromatographymass spectrometry as the less-nitrated

analogues of PETN: pentaerythritol trinitrate (PETriN), pentaerythritol dinitrate

(PEDiN) and pentaerythritol mononitrate (PEMN). Significant levels of these

degradation products were observed in post-explosion samples, whereas only very low

levels were detected in a variety of intact PETN samples and naturally degraded PETN.

Based on the peak areas of PETriN, PEDiN and PEMN relative to PETN, it was possible

to fully distinguish the post-explosion profiles from the profiles obtained from intact

PETN or after (accelerated) natural degradation. Although more data are required to

accurately assess the strength of the evidence, this work illustrates that PETN profiling

may yield valuable evidence when investigating a possible link between a suspect and

post-explosion PETN found on a crime scene.

PETN profiling in post-explosion residues 39

Chapter 3

3.1 Introduction

Recently, a series of safe crackings occurred in the Netherlands in which the explosive

PETN (pentaerythritol tetranitrate) was used. A suspect was apprehended and his

clothing was sampled and analyzed by the NFI (Netherlands Forensic Institute).

Analysis by liquid chromatographymass spectrometry (LCMS) led to the detection

of PETN, but also relatively high amounts of its degradation products pentaerythritol

trinitrate (PETriN), the dinitrate (PEDiN) and the mononitrate (PEMN). This triggered

the question whether the identification of these degradation products could be explained

by the suspects presence at the crime scene. The suspect stated that the residues

found on his clothing had originated from handling intact PETN, which was also

found at his home. The degradation products could then be explained by the presence

of impurities in the material or by natural degradation of the intact PETN as such or

on the clothing. In this context, differences between PETN chemical profiles obtained

by natural degradation and explosion were studied. This chapter reports on the study

that was undertaken to investigate whether post-explosion PETN residue profiles can

be differentiated from PETN profiles arising from other processes such as natural

degradation, or during synthesis.

PETN is a powerful high explosive prepared by nitration of pentaerythritol (PE). It

was brought into use during World War II, after formaldehyde and acetaldehyde

(precursors of PE) had become industrially available [1]. PETN is relatively stable

both chemically and physically in comparison with other nitrate-ester explosives,

such as ethylene glycol dinitrate (EGDN) and nitroglycerin (NG) [2,3]. This has been

attributed to the symmetrical molecular structure of PETN [1]. Major applications of

PETN as an explosive are as main charge in detonation cords and blasting caps [1,2] and

in formulations [4-6] (e.g. Semtex, pentolite, PEP 500).

In addition to its use as an explosive, PETN also acts as a coronary vasodilator. It is used

as the active ingredient in heart medicines for the treatment of angina pectoris [2,7]. The

40 Chapter 3

Cha

pter

3

amounts of PETN used for medical purposes are very small compared to the amounts of

material required to cause explosions.

3.1.1 Decomposition of PETNTraditional explosive-residues analysis only considers identification of the explosive used

and, therefore, limited information is available on the presence of degradation products

of PETN in post-explosion samples. Moreover, levels of explosives and degradation

products in post-explosion residues tend to be low, making their detection challenging.

Although detonation of PETN mainly results in the formation of gaseous products [1],

solid decomposition products are usually also formed, because of incomplete detonation.

Thin-layer chromatographic (TLC) analysis of post-explosion extracts showed

additional spots apart from PETN, which were later identified by chemical-ionization

mass spectrometry (CIMS) and nuclear-magnetic-resonance (NMR) spectroscopy as

pentaerythritol trinitrate (PETriN) and pentaerythritol dinitrate (PEDiN). Another spot

could not be identified, although it was suggested that this spot could be attributed to

pentaerythritol mononitrate (PEMN) [8]. Fig. 3.1 shows the chemical structure of PETN

and its less-nitrated analogues.

O

O

O O

NOO

NO

O

N

O

O

NOO

O2NO

ONO2

O2NO OH

HO

ONO2

O2NO OH

HO

ONO2

HO OH

PETN PETriN

PEDiN PEMN

Fig. 3.1. Chemical structure of PETN and its degradation products PETriN, PEDiN and PEMN.

More literature is available on chemical and environmental degradation of PETN. The

degradation of PETN is influenced by a variety of parameters, such as temperature, the


Chapter 3

presence of microorganisms, and humidity. There is consensus in literature that the first

and rate-determining step in the decomposition of PETN is the scission of the O-NO2 bond, resulting in the release of nitrogen dioxide (NO2) [2-4,9,10]. This was observed for

nitrate esters in general [9]. Several mechanisms for the following decomposition steps

have been postulated, depending on the physical and chemical environment.

PETN is stable compared to other organic explosives [4] and therefore the majority of

research into thermal decomposition of PETN has involved elevated temperatures (i.e.

above 100C). However, decomposition mechanisms are different at higher temperatures

than under ambient conditions [2,4,11,12]. Thus, the results of accelerated-degradation

studies may not accurately reflect the natural degradation occurring on, for instance, the

clothing of a suspect. However, previous research on high-temperature decomposition

of PETN yields useful information on the identity of degradation products formed.

As no condensed-phase decomposition products were detected after analysis of

naturally aged PETN at ambient temperatures, it was suggested that at low temperatures

only gaseous decomposition products are formed [2]. Decomposition of PETN at 53C

was studied by monitoring the released NOx (mainly NO2) using a chemiluminescence

analyzer [12]. PETN was found to have an NOx evolution rate that was roughly 1000

times lower than the evolution rate for nitrocellulose. By extrapolating the NOx-emission

data, the half-life time of PETN was estimated to be 12 million years, confirming

its stability [12]. Chambers et al. [4] reported on the possible formation of peroxide

[(O2NOCH2)3C-CH2OO], nitrate [(O2NOCH2)3C-NO2] and aldehyde [(O2NOCH2)3C-

CHO] products at ambient temperatures, but the presence of these products has not

been experimentally confirmed. It was also suggested that the alkoxy radical formed by

scission of the O-NO2 bond could attack PETN, resulting in polymer-like side products,

such as dipentaerythritol hexanitrate (DiPEHN) and tripentaerythritol octanitrate

(TriPEON) [2,4].

Thermal ageing studies of PETN at 80C [1] and 100C [2] did not show significant

degradation, but continued heating to temperatures above the melting point of PETN

42 Chapter 3

Cha

pter

3

(141.3C) resulted in gradual decomposition [1]. Decomposition of PETN at higher

temperatures yielded a greater variety of primarily gaseous decomposition products

resulting from further breakdown of PETN [2-4,10]. It was also reported that the

second step in the decomposition of PETN (after cleavage of the O-NO2 bond) is the

loss of a formaldehyde molecule [3,9-11]. Infrared analysis of the residual material

after degradation of PETN in benzene at 185C resulted in the identification of a

polyketo oxetane [9]. This led to a proposed decomposition mechanism involving cyclic

intermediates. Shepodd et al. [11] identified several decomposition products (including

PETriN, PEDiN and DiPEHN) using LCMS and capillary electrochromatography

mass spectrometry (CECMS) after heating PETN under vacuum at temperatures up

to 135C. PETriN formation during high-temperature decomposition of PETN was also

suggested by Makashir and Kurian [3].

3.1.2 Other factors influencing PETN decompositionSeveral environmental factors have been found to accelerate PETN decomposition or to

result in different decomposition pathways. These include the presence of water, soil, or

microorganisms.

The presence of water has a detrimental effect on the stability of PETN. Moisture results

in sequential hydrolysis of the O-NO2 bonds, resulting in hydroxyl end groups [2,4,13].

Several studies showed the formation of PETriN, PEDiN and PEMN [1,4,13]. Hydrolysis

proceeded more rapidly under acidic or basic conditions [1,11].

Microbial degradation of PETN also resulted in the formation of PETriN, PEDiN

and PEMN [14,15]. This behavior was also observed for other nitrate esters, such as

nitroglycerin, EGDN and nitrocellulose [15,16] and it was suggested that biodegradation

of nitrate esters generally follows a hydrolytic pathway [17]. Binks et al. [14] isolated

a microbial culture (Enterobacter cloacae PB2) from explosive-contaminated soil.

Several metabolites of PETN were detected, including PEDiN. The enzyme PETN

reductase was also isolated from the culture, showing conversion of PETN to PETriN

and PEDiN. In another study, PETN was buried in soil and after 20 years, 90% of the


Chapter 3

PETN was found to be remaining. From these results, the half-life time of PETN in soil

was estimated to be 92 years [18].

Although PETN is relatively resistant to chemical reagents [1], several compounds can

accelerate its decomposition, such as carbamite (1,3-diethyl-1,3-diphenylurea), calcium

carbonate, magnesium oxide [3], ferrous chloride [1] and granular iron [5]. When

analyzing degraded PETN, it should be considered that some of the proposed degradation

products may also have originated as side products during synthesis. Yasuda [19] used

TLC to identify PETriN, DiPEHN and TriPEON in PETN samples. Other commonly

encountered impurities are pentaerythritol (PE), PEMN and PEDiN [4].

3.1.3 Case assessmentIn the present study, the possibility to discriminate between PETN degradation during

explosion and other scenarios is investigated. This is important in assessing the evidential

value of an observed PETN chemical profile in cases as the example described above.

The probability of the evidence should then be considered under different hypotheses

that may be postulated by the prosecution (Hp) or the defense (Hd), in line with the

Bayesian framework for evidence interpretation [20,21]. To discriminate between post-

explosion samples and other scenarios, the following hypotheses were formulated:

Hp: The observed PETN degradation products on the suspects clothing originate from

a PETN explosion.

Hd,1: The observed PETN degradation products on the suspects clothing were present as

impurities in the intact PETN handled by the suspect.

Hd,2: The observed PETN degradation products on the suspects clothing were formed

by chemical and environmental degradation of PETN.

To determine the specificity of post-explosion PETN profiles it should be investigated

whether similar profiles can be generated by other processes than PETN detonation.

In this study, PETN-detonation experiments were conducted. Samples were taken and

44 Chapter 3

Cha

pter

3

analyzed using LCMS to establish the PETN chemical profile and to observe the

variation therein. In addition, a selection of PETN samples of different origins were

analyzed to determine the impurity profile that might have been expected if intact

PETN material were present on the suspects clothing. Finally, numerous experiments

were conducted to effectuate PETN degradation through chemical and environmental

processes. These experiments included various textile matrices and variation in

parameters such as temperature and humidity. The LCMS profiles of all experiments

were compared to establish to what extent observed PETN profiles can provide support

for the hypothesis that PETN residues originate from an explosion.

3.2 Experimental

3.2.1 Chemicals and materialsHigh-purity PETN (containing a low level of PETriN as a minor impurity) was

provided by TNO Technical Sciences, department of Energetic Materials (Rijswijk, The

Netherlands). Rathburn (Walkerburn, UK) HPLC grade methanol was used for both

sample preparation and LCMS analysis. Ultra-pure water, prepared using a Milli-Q

(Millipore, Bedford, MA, USA) or a PureLab Ultra (Elga, High Wycombe, UK)

system, was used both

Documents

UvA-DARE (Digital Academic Repository) Chemical … · Chemical profiling of explosives PhD thesis, University of Amsterdam, The Netherlands ... General introduction 3 Chapter 1 1.1