 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
thesis entitled

Differentiation of Bullet Type Based on Analysis of Gunshot
Residue Using Inductively Coupled Plasma Mass

Master of
Science

 

Spectrometry

presented by

Ruth Norma Udey

has been accepted towards fulﬁllment
of the requirements for the

degree in Forensic Science

 

ﬂake!"

Major Professor’s Signature
77M r‘fi Y in":

&

Date

MSU is an Afﬁrmative Action/Equal Opportunity Employer

o.._.- _ 4

_-._‘_.-._ ..,~-.

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

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

5/08 KIProlecc&Pres/ClRC/DateDueJndd

DIFFERENTIATION OF BULLET TYPE BASED ON ANALYSIS OF GUNSHOT
RESIDUE USING INDUCTIVELY COUPLED PLASMA MASS SPECTROMETRY

By

Ruth Norma Udey

A THESIS

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

MASTER OF SCIENCE
Forensic Science

2010

ABSTRACT

DIFFERENTIATION OF BULLET TYPE BASED ON ANALYSIS OF GUNSHOT
RESIDUE USING IN DUCTIVELY COUPLED PLASMA MASS SPECTROMETRY

By

Ruth Norma Udey

Porcine tissue samples shot with two different types of bullets, jacketed and non-
jacketed, were collected in the fresh state and throughout moderate decomposition.
Wound samples were microwave digested and analyzed using inductively coupled
plasma mass spectrometry (ICP-MS) to detect all elements present in gunshot residue
(GSR). Elements detected included antimony (Sb), barium (Ba), and lead (Pb), which are
considered characteristic of GSR, as well as iron (Fe) and copper (Cu). These ﬁve
elements were used to differentiate shot tissue and unshot tissue, as well as tissue shot by
the two different bullet types, in both the fresh state and throughout moderate
decomposition. The concentrations of Cu, Sb, and Pb were able to distinguish between
the two bullet types in fresh tissue samples at the 95% conﬁdence level. Copper and Pb
were able to differentiate the bullet types throughout moderate decomposition at the 99%
conﬁdence level. The source of these distinguishing elements is likely the different
bullets. A sampling study carried out on a fresh non-jacketed bullet wound determined
that the optimal sampling position to detect all elements considered characteristic of GSR
using ICP-MS is directly adjacent to the wound and as far out as two centimeters from

the entrance wound in any direction.

ACKNOWLEDGEMENTS

I would like to thank the many people who assisted and supported me throughout
this project. Dr. Ruth Smith was my advisor and was always quick with suggestions and
encouragement. Dr. Brian Hunter had great ideas for the project, taught me about
pathology, and sat on my committee. Dr. Steve Dow also took time out of his schedule
and sat on my committee. Al Snedegar and Kevin Turner from the MSU Swine Facility
provided the pigs and helped me transport them. A local certiﬁed ﬁrearms instructor shot
the pigs for me and taught me about ammunition. Matt Parsons assisted with the ICP-MS
analyses, and Ron Crichton provided useful input about results interpretation.

I also need to thank the other students in the Forensic Chemistry program,
including those who started out when I did and have already graduated as well as current
group members. Thanks to Sarah Meisinger, Lucas Marshall, Lisa LaGoo, Jamie
Baemcopf, Patty Giebink, Melissa Bodnar, Christy Hay, John McIlroy, Tiffany Van de
Mark, Seth Hogg, Kari Anderson, Kaitlin Prather, and Monica Bugeja for talking through
problems with me, giving me suggestions, and being such great friends.

Finally, I would like to thank my parents, Dana and Joyce, my sister and her
husband, Amanda and Jonathan, and all of my family members for being constant

sources of encouragement and perspective. I am truly blessed to have you in my life.

iii

TABLE OF CONTENTS

 

 

 

 

 

 

List of Tables vi
List of Figures vii
Chapter 1: Introduction ..... l
1.1 Case Study ..................................................................................................................... 1
1.2 Ammunition .................................................................................................................. 2
1.3 Current Forensic Methods for Gunshot Residue Detection .......................................... 5
1.4 Gunshot Residue Detection Using ICP-MS .................................................................. 7
1.5 Research Objectives ...................................................................................................... 9
1.6 References ................................................................................................................... 11
Chapter 2: Theory 13
2.1 Histology ..................................................................................................................... 13
2.2 Microwave Digestion of Porcine Tissue ..................................................................... 14
2.3 Inductively Coupled Plasma Mass Spectrometry ....................................................... 15
2.4 Statistical Procedures .................................................................................................. 19

2.4.1 Analysis of Variance .......................................................................................... 19

2.4.2 Principal Components Analysis ......................................................................... 24
2.5 References ................................................................................................................... 28
Chapter 3: Materials and Methods - 29
3.1 Experimental Design and Sample Collection ............................................................. 29

3.1.1 Fresh Tissue Studies ........................................................................................... 29

3.1.2 Sampling Study .................................................................................................. 30

3.1.3 Decomposition Study ......................................................................................... 32
3.2 Histology ..................................................................................................................... 32
3.3 Microwave Digestion of Porcine Tissue ..................................................................... 33
3.4 ICP-MS Analysis of Fresh Tissue Samples ................................................................ 34
3.5 Statistical Treatment of Fresh Tissue Data ................................................................. 36
3.6 ICP-MS Quantitation of Fresh and Decomposed Tissue Samples ............................. 37
3.7 Principal Components Analysis of Decomposition Study Data ................................. 39
3.8 References ................................................................................................................... 41
Chapter 4: Results and Discussion 42
4.1 Introduction ................................................................................................................. 42
4.2 Fresh Tissue Sample Observations and Histology Results ......................................... 42
4.3 Selection of Elements to Discriminate Bullet Types .................................................. 44
4.4 Quantitation of Elements in Fresh Tissue Samples .................................................... 48
4.5 Sampling Study Tissue Observations ......................................................................... 52
4.6 Quantitation of Elements in Sampling Study Tissue Samples .................................... 53
4.7 Decomposition Study Observations and Histology Results ....................................... 58

iv

4.8 Quantitation of Elements in Tissue Samples Throughout

 

 

Moderate Decomposition ............................................................................................ 63
4.9 Principal Components Analysis of Decomposition Study Data ................................. 68
4.10 References ................................................................................................................. 73
Chapter 5: Conclusions and Future Directions - - _ -_ 74
5.1 Conclusions ................................................................................................................. 74
5.2 Future Directions ......................................................................................................... 78
Appendix -- 81

LIST OF TABLES

Table 2.1 - Summary of sum of squares and degrees of freedom calculations ................ 21

Table 3.1 — ICP-MS parameters for full mass scan and selected ion monitoring (SIM)
analyses .......................................................................................................... 35

Table 4.1 — Limits of quantitation (LOQs) and SRM recovery results for elements of
interest in the fresh tissue study ..................................................................... 49

Table 4.2 - Limits of quantitation (LOQs) and SRM recovery results for elements of
interest in the sampling study ......................................................................... 54

Table 4.3 — Summary of distances that elements of interest in shot tissue samples were
detected at higher levels than procedural blank and control tissue samples .. 57

Table 4.4 — Limits of quantitation (LOQs) and SRM recovery results for elements of
interest in the decomposition study ................................................................ 63

Table 4. 5— Element concentration ranges throughout moderate decomposition m each
tissue sample type .......................................................................................... 64

vi

LIST OF FIGURES

Figure 1.1 — J acketed (left) and non-jacketed (right) ammunition cartridges ..................... 3
Figure 2.1 — Schematic of an ICP-MS system .................................................................. 16
Figure 2.2 — Example results of PCA visualized as scores and loadings plots ................. 27
Figure 3.1 — Diagram of wound where tissue samples were excised for analysis ............ 31

Figure 4.1 — Gunshot wounds in ﬁesh porcine tissue from (a) jacketed bullets and
(b) non-jacketed bullets, along with corresponding micrographs of H & E
stained histology results (c and d, respectively). Refractive particles of
unburned gunpowder are circled in the jacketed bullet wound histology
slide (c), and dark granules of burned gunpowder are circled on the
non-j acketed bullet wound histology slide (d) .............................................. 43

Figure 4.2 — Mean concentrations of elements of interest in the ﬁrst fresh tissue
study (procedural blank 11 = 3, control n = 3, jacketed n = 9, and
non-jacketed n = 9). The error bars represent one standard deviation.
Signiﬁcant difference between element concentrations at 99% conﬁdence
level (p S 0.01) are indicated by ** .............................................................. 47

Figure 4.3 — Mean concentrations of elements of interest quantiﬁed in fresh tissue
samples (procedural blank 11 = 6, control n = 1, jacketed n = 5, and
non-jacketed n = 5). The error bars represent one standard deviation.
Signiﬁcant difference between element concentrations at 95% conﬁdence
level (p S 0.05) are indicated by * and signiﬁcant difference between
element concentrations at 99% conﬁdence level (p S 0.01) are indicated
by ** .............................................................................................................. 51

Figure 4.4 — Sampling study wound (a) before tissue sections were removed and (b)
after tissue samples had been removed for analysis. The four directions
that the tissue was sampled are indicated with arrows .................................. 53

Figure 4.5 — Concentrations of (3) Fe, (b) Sb, (c) Ba, and (d) Pb quantitated in the
sampling study samples (procedural blank 11 = 3, control n = 3, and
non-jacketed n = 1). Error bars are one standard deviation. In the 3:00
direction data, “6 cm/edge” indicates that the wound edge was at 6 cm ...... 55

Figure 4.6 — Jacketed bullet wounds on days 0, 29, and 49 during the decomposition

study. Corresponding micrographs of H & E stained histology results are
shown below the wounds, with particles consistent with GSR circled ......... 6O

vii

Figure 4.7 — Non-jacketed bullet wounds on days 0, 29, and 49 during the
decomposition study. Corresponding micrographs of H & E stained
histology results are shown below the wounds, with particles consistent
with GSR circled ........................................................................................... 61

Figure 4.8 — Control (stab) wounds on days 0, 29, and 49 during the decomposition
study. Corresponding micrographs of H & E stained histology results
did not show any particles consistent with GSR, as expected ...................... 62

Figure 4.9 — Mean concentrations of elements of interest quantiﬁed in samples
collected throughout moderate decomposition (procedural blank n = 5,
control (fresh) n = 3, control (decomposed) n = 6, jacketed and
non-jacketed n = 6). The error bars represent one standard deviation.
Signiﬁcant difference between element concentrations at 99% conﬁdence
level (p S 0.01) are indicated by ** .............................................................. 66

Figure 4.10 — PCA scores plot displaying clustering of chemically similar samples ....... 69

Figure 4.11 — PCA loadings plot showing the elements responsible for the sample
positions displayed in the corresponding scores plot .................................. 69

viii

CHAPTER 1: INTRODUCTION
1.1 Case Study

Conventionally, gunshot wound determinations are made using gross injury
examination at autopsy by visualization of gunshot residue (GSR). However,
interpretation of gunshot wounds can be difﬁcult even when the deceased individual is
well preserved, since the characteristics of gunshot wounds vary greatly based on the type
of ﬁrearm and ammunition used, range of ﬁre, and the location of the wound on the body.
When post-mortem factors such as decomposition, burial, and insect activity are present,

the identiﬁcation of a gunshot wound becomes even more challenging, as illustrated by

the following case presented to a local forensic pathologist in 2004.I

The decedent was a 25 year old white male whose body was found buried in a remote
area one year after he went missing. His soft tissues had a soapy, mushy quality
consistent with adipocere. Police investigation and the shirt recovered with the victim
indicated that he had been shot in the torso, however, the malleable nature of the tissues
made identiﬁcation of the entrance wounds only possible through palpation. The poor
preservation of the tissue also prevented detection of GSR by gross or histologic
examination.

Clinical scenarios such as this demonstrate the need for ways to increase conﬁdence
in GSR determination by using chemical means of detection that are not affected by
decomposition. In order to design suitable chemical detection methods, the content of

GSR must ﬁrst be understood.

1.2 Ammunition

Small-arms ammunition cartridges consist of a cartridge case, a primer, propellant
(gunpowder), and a bullet [Figure 1.1]. When a weapon is ﬁred, the ﬁring pin strikes the
center of the primer cup, compressing the primer and causing it to explode. Small ﬂash
holes allow the ﬂame to pass into the cartridge case, igniting the propellant. The
combustion of the propellant leads to gas formation and the increased pressure forces the
bullet down the gun barrel. The heated primer also partially vaporizes and is released

from the gun along with the propellant combustion products and unburned propellant.

. . . . . 2 .
This combination of materlals 18 known as GSR. While some research has analyzed the

organic compounds in the propellant,3 the chemical detection of GSR in forensic labs

focuses on the metals in the primer (antimony (Sb), barium (Ba), and lead (Pb)).

However, vaporized metals from the cartridge case, bullet, and gun barrel that were

present in the hot gas cloud emitted during discharge could also be present in GSR.2

J acketed

bullet .. N on-jacketed
‘ bullet
Case
Primer

 

Figure 1.1 — J acketed (left) and non-jacketed (right) ammunition cartridges.

Cartridge cases are usually made of brass, with a composition of 70% copper and
30% zinc, but some are plated with nickel or made of steel or aluminum. The main

functions of the cartridge case are to hold the propellant and to expand and seal the gun

chamber so that all gases escape in the forward direction when the cartridge is ﬁred.2

The primer is located in the primer cup in the base of the cartridge and is used to
initiate ignition of the propellant. The primer is composed of an explosive, an oxidizer,
and a fuel, and it is stable but explosive upon impact. The most common primer
compounds are lead styphnate (explosive), barium nitrate (oxidizer), and antimony
sulﬁde (fuel). The detection of these compounds is the basis for tests to determine

whether an individual has ﬁred a ﬁrearm. Some ammunition manufacturers have

removed the lead styphnate from the primer composition due to environmental concerns,
while other manufacturers use only lead styphnate or only lead styphnate and barium

nitrate. However, these ammunition types are uncommon as they are designed for

. . 2
specrﬁc purposes or specrﬁc ﬁrearms.

The propellant is commonly smokeless gunpowder composed of nitrocellulose
(single-base), nitrocellulose and nitroglycerine (double-base), or nitrocellulose,
nitroglycerine, and nitroguanidine (triple-base). The individual powder grains can be
shaped as disks, ﬂakes, or cylinders, depending on the desired burning rate of the powder.
The propellant itself is relatively stable against ignition, hence the need for the primer to

initiate the burning of the propellant. It is the rapid burning of the propellant and release

of gaseous products that pushes the bullet out of the ﬁrearm.2

The bullet is the part of the cartridge that leaves the barrel of the ﬁrearm when it
is discharged. Bullets fall into two composition categories: lead bullets and metal-
jacketed bullets. Lead bullets are made principally of lead with antimony and/or tin added
to increase hardness. Metal jacketed bullets have a lead or steel core that is covered by an
outside jacket composed of copper with zinc, steel, copper with nickel, or aluminum. The
bullets may be either full-jacketed or partially-jacketed, with the bullet core totally

covered in full-jacketed ammunition and the core partially exposed in partially-jacketed

. . 2
ammunition.

1.3 Current Forensic Methods for Gunshot Residue Detection

Coroners, medical examiners, and forensic pathologists routinely use histology
procedures to determine the presence of GSR in tissue sections from a suspected gunshot
wound. Tissue sections cut from samples taken adjacent to the wound are stained using
hematoxylin and eosin (H & E) and examined using light microscopy. Particles of burned
gunpowder appear as ﬁne, dark granules, and larger pieces of unburned gunpowder
appear lighter in color as they refract the incident light. However, as the stain does not
react with the GSR particles themselves, only the tissue, any particles observed in the
tissue sections cannot be deﬁnitively identiﬁed as GSR. This becomes a problem when
there is debris present on the body, resulting from burial for example, as soil and sand
particles can appear similar to burned and unburned gunpowder during both gross and
histological examination. Tissue also becomes more difﬁcult to section as decomposition
progresses. These difﬁculties would be overcome by a chemical detection method
suitable for use on tissue at any stage of decomposition.

Gunshot residue has been detected in fresh animal hide using staining and visual

examination techniques.4’ 5 Shot human tissue samples have also been analyzed for GSR

using staining and X-ray microﬂuorescence, but these studies focused on fresh, rather

. . . . . 6, 7
than decomposrng, tissue and did not use the instrumental methods described below.

Crime laboratories today use scanning electron microscopy with energy dispersive X-
ray spectroscopy (SEM/EDS) to identify GSR on possible assailants or victims’ skin or
clothing to help conﬁrm a wound as a gunshot wound. Particles suspected to contain
GSR are collected by dabbing an adhesive tab mounted on an aluminum stub across a

surface. Analysis using SEM/EDS provides both particle morphology and elemental

composition which are used to deﬁnitively identify GSR particles.2 Wolten et al.

described four elemental combinations that were observed in GSR particles and were

. . 8 . . .
therefore regarded as characteristic of GSR. The most prominent of these combinations

was Sb, Ba, and Pb, which come from ammunition primers and bullets.9 However,
SEM/EDS is limited by the ability to locate GSR particles on the adhesive tab, which
becomes difﬁcult if the particles are embedded in the adhesive or are obscured by other

debris from the samplem’ H While automated software programs are available, the

. . . . . 2 12
analySis can still be time-consummg, taking up to several hours per cm of sample.

Also, it becomes difﬁcult to effectively collect particles from decomposing tissue using

. . . . 13
the adhesrve stubs, especrally when outdoors, as the tissue becomes Oily.

Differentiation of jacketed and non-jacketed bullets based on elements present in

GSR was demonstrated by Wunnapuk et al. in 2007. ‘4 Fresh wounds in human tissue

were used in the study and were analyzed using inductively coupled plasma atomic
emission spectroscopy (ICP-AES). Wounds were digested using nitric acid and heating at
100 °C for two hours. After ﬁltering and dilution, the tissue digests were analyzed using
ICP—AES in a semi-quantitative manner. Lead and antimony were signiﬁcantly more
concentrated in non-j acketed bullet wounds compared to jacketed bullet wounds, as was
barium although the difference in concentration was not signiﬁcant. Copper and iron
were more concentrated in the jacketed bullet wounds compared to non-jacketed bullet
wounds, though the differences were not signiﬁcant. The concentration of zinc was

similar for both bullet types. However, not all of the elements were useful for

differentiating between control tissue and tissue shot with both bullet types. The study
concluded that a high content of iron and zinc was deposited in wounds shot with
jacketed bullets, and a high concentration of lead was deposited as a result of a non-

jacketed bullet wound. These elements were likely due to the bullet and easing

.. l4 . . . . .
composrtions. While this study demonstrated the analySis of tissue surrounding a

gunshot wound, it is likely that the digestion procedure was not sufﬁcient to completely
digest the tissue and release all of the elements of interest into solution. Also, the study
focused on fresh tissue samples and did not take the effects of decomposition into
account. In addition, the analysis was not rigorously quantitative, so element
concentrations are not expected to be as accurate as they could be. Detection limits for
ICP-AES analysis were not reported in this particular study, however, it is well known
that other techniques, such as inductively coupled plasma mass spectrometry, have

detection limits between one and three orders of magnitude lower than ICP-AES,

depending on which elements are being analyzed. 1 5

1.4 Gunshot Residue Detection Using ICP-MS

Inductively coupled plasma mass spectrometry (ICP-MS) has also been used for

the determination of Sb, Ba, and Pb in GSR, as well as simulated samples of GSR.]6'18

Analysis using ICP-MS was successful in determining these three elements of interest

. . 16
from extracts of swabs spiked wrth the three elements, extracts of swabs from shooter’s

16,17

hands, and from digests of shot cotton tissue. 1 8 Koons described the analysis of

standard solutions containing Sb, Ba, and Pb, as well as extract solutions from cotton

swabs either spiked with Sb, Ba, and Pb or taken from shooters’ hands, using ICP-MS.16

Both standard and swab extract solutions showed linear responses over the entire
dynamic range of the instrument, from the limit of detection to the maximum detectable
signal, indicating that no matrix effects from the cotton swabs were present. The relative
standard deviation of replicate measurements was less than 5% for all three elements,
demonstrating the high precision of the instrument. The detection limits were determined
to be 0.052 ug/L for Sb, 0.020 ug/L for Ba, and 0.014 rig/L for Pb. The limits of detection
for the three elements were at least one order of magnitude less than levels typically
observed in GSR (40-500 ng), and were lower than corresponding detection limits using
graphite furnace atomic absorption spectroscopy and ICP-AES. The ability to detect Sb,

Ba, and Pb in extracts of swabs from shooters’ hands by ICP-MS was also comparable to

the two other analytical techniques.16

Santos et al. used a target of cotton tissue, and the resulting GSR deposition

pattern was analyzed by ICP-MS to estimate ﬁring distance.18 Samples of the cotton

tissue were removed from different distances around the entrance wound and digested
with nitric acid prior to analysis. The detection and quantitation of Sb, Ba, and Pb
deposited on the cotton tissue was possible at ﬁring distances ranging from 20 to 80 cm
in four radial positions ranging from 1.5 to 5.5 cm from the entrance hole. The best radial

position for sample collection was determined to be between 3.5 and 4.5 cm ﬁ‘om the

. . . . 18
entrance hole in order to most accurately determrne the ﬁring distance. However, these

results are only valid for the speciﬁc gun and ammunition used and cannot be applied to

other ﬁrearms.

The use of ICP-MS to analyze digested bullet shavings to differentiate bullets
according to elemental composition has also been described, but the resulting GSR was
not analyzed to determine if the residue could be used to differentiate the bullets as

19-21
well.

The studies outlined thus far have focused on the detection of GSR deposited on the
hands of a shooter or other surfaces using lCP-MS. Until very recently the analysis of
GSR tattooed into tissue surrounding a gunshot wound had not been attempted using

ICP-MS. LaGoo et al. demonstrated the ability of ICP-MS to detect Sb, Ba, and Pb in

. . . . l3 . . .
shot porcme tissue throughout decomposrtion. AnalySis of microwave digested wound

tissue allowed the detection of GSR at all stages during decomposition in both late
summer and winter months. Wounds were also analyzed using SEM/EDS for
comparison, and GSR was only identiﬁed on the ﬁrst day of the study, after which rain
washed away surface particles and prevented collection using adhesive tabs. Detection of
GSR by ICP-MS was shown to be unaffected by environmental conditions, and it was

also shown that concentrations of GSR detected in tissue depended more on the stage of

. . . . l3 . .
decomposrtion rather than the time Since death. While the study showed success in

detecting Sb, Ba, and Pb, there was no attempt to detect any other elements that may be

present in GSR.

1.5 Research Objectives
Gunshot wounds in decomposing tissue are difﬁcult to identify, making chemical

means of detection necessary. While work has been done to detect Sb, Ba, and Pb in

tissue using ICP-MS, the potential to detect additional elements that may provide further
information about the bullet or gun used to make the wound has not been investigated.
Therefore, the goals of this research were:

- To identify additional elements characteristic of GSR in order to increase

conﬁdence in gunshot wound determination using lCP-MS.

- To differentiate two bullet types, jacketed and non-jacketed, using their

characteristic element proﬁles.

- To assess the effects of moderate decomposition on the ability to detect GSR

and differentiate bullet types.

The results of this research will beneﬁt law enforcement agencies as well as
coroners, medical examiners, and forensic pathologists. Law enforcement agencies may
use the method to determine the type of bullet used, which can link a weapon and/or
suspect to a crime scene. Coroners, medical examiners, and forensic pathologists could
also use the method to assist in gunshot wound identiﬁcation and cause of death

determination, even in corpses in a moderate state of decomposition.

10

1.6 References

1. MacAulay LE, Barr DG, Strongman DB. Effects of decomposition on gunshot
wound characteristics: under moderate temperatures with insect activity. J
Forensic Sci 2009;54:443-7.

2. Di Maio VJM. Gunshot wounds: practical aspects of ﬁrearms, ballistics, and forensic
techniques. 2nd ed. Boca Raton: CRC Press, 1999.

3. Scherperel G. Electrospray ionization mass spectrometry for the detection and
characterization of smokeless powders [thesis]. East Lansing (MI): Michigan State
University, 2007.

4. Brown H, Cauch DM, Holden JL, Wrobel H, Cordner S. Image analysis of gunshot
residue on entry wounds - I - the technique and preliminary study. Forensic Sci Int
1999;100:163-77.

5. Brown H, Cauchi DM, Holden JL, Allen F CL, Cordner S, Thatcher P. Image analysis
of gunshot residue on entry wounds - II - a statistical estimation of ﬁring range.
Forensic Sci Int 1999; 1 00: 1 79-86.

6. Glattstein B, Zeichner A, Vinokurov A, Levin N, Kugel C, Hiss J. Improved method
for shooting distance estimation. Part III. Bullet holes in Cadavers. J Forensic Sci
2000;45:1243-9.

7. Brazeau J, Wong RK. Analysis of gunshot residues on human tissues and clothing by
X-ray microﬂuorescence. J Forensic Sci 1997;42:424-8.

8. Wolten GM, Nesbitt RS, Calloway AR, Loper GL, Jones PF. Particle analysis for the
detection of gunshot residue. 1. Scanning electron microscopy energy dispersive X-ray
characterization of hand deposits from ﬁring. J Forensic Sci 1979;24:409-22.

9. Stone IC, Di Maio VJM, Petty CS. Gunshot wounds — visual and analytical procedures.
J Forensic Sci 1978;23:361-7.

10. Matricardi VR, Kilty J W. Detection of gunshot residue particles from hands of a
shooter. J Forensic Sci 1977;22:725-38.

11. Romolo FS, Margot P. Identiﬁcation of gunshot residue: a critical review. Forensic
Sci Int 2001;119:195-211.

12. Zeichner A. Recent developments in methods of chemical analysis in investigations
of ﬁrearm-related events. Anal Bioanal Chem 2003;376:1178-91.

l3. LaGoo L, Schaeffer LS, Szymanski DW, Waddell Smith, R. Detection of gunshot
residue in blowﬂy larvae and decomposing porcine tissue using inductively coupled

11

14.

15.

16.

17.

18.

19.

20.

21.

plasma mass spectrometry (ICP-MS). J Forensic Sci 2010;55:624-32.

Wunnapuk K, Durongkadech P, Minami T, Ruangyuttikam W, Tohno S, Vichairat
K, Azuma C, Sribanditmongkol P, Tohno Y. Differences in the element contents
between gunshot entry wounds with full-j acketed bullet and lead bullet. Biol Trace
Elem Res 2007; 1 20:74-81 .

Skoog DA, Holler FJ, Nieman TA. Principles of instrumental analysis. 5th ed.
Toronto: Thomson Learning, 1998.

Koons RD. Analysis of gunshot primer residue collection swabs by inductively
coupled plasma-mass spectrometry. J Forensic Sci 1998;43:748-54.

Sarkis JES, Neto ON, Viebig S, Durrant SF. Measurements of gunshot residues by
sector ﬁeld inductively coupled plasma mass spectrometry - further studies with
pistols. Forensic Sci Int 2007;172:63-6.

Santos A, Magalhaes T, Vieira DN, Almeida AA, Sousa AV. Firing distance
estimation through the analysis of the gunshot residue deposit pattern around the
bullet entrance hole by inductively coupled plasma-mass spectrometry - an
experimental study. Am J F oren Med Path 2007;28:24-30.

Dufosse T, Touron P. Comparison of bullet alloys by chemical analysis: use of
ICP-MS method. Forensic Sci Int 1998;91:197-206.

Keto RO. Analysis and comparison of bullet leads by inductively-coupled plasma
mass spectrometry. J Forensic Sci 1999;44:1020-6.

Suzuki Y, Marumo Y. Determination of trace impurities in lead shotgun pellets by
ICP-MS. Anal Sci 1996;12:129-32.

12

CHAPTER 2: THEORY
2.1 Histology

Histology is the study of tissues and microscopic cellular details by pathologists

for a variety of medical diagnostic applications. Staining of thin tissue sections allows

visualization of structural details at the cellular level. In order to study pathological
material, sections of tissue are stained in such a way as to impart dark color to the nuclei
of cells and a lighter, contrasting color to the cytoplasm and extracellular structures. The

most common histology staining method used by pathologists is hematoxylin and eosin

(H & E) staining.2 The ﬁrst part of the stain, hematoxylin, is positively-charged and

colors the nuclei of cells blue by binding to negatively-charged nucleic acids in DNA.
The eosin counterstain is negatively-charged and colors both intracellular and

extracellular proteins pink by binding to positively-charged amino acids, resulting in cell

cytoplasm staining. Stained tissue sections are examined usrng light microscopy to

determine any structural irregularities that might give information as to disease or the

2
cause of a wound.

While H & E staining of tissue sections is useful, the stains are not highly
selective, and are only designed to aid visualization. Any debris present on the tissue
section, such as gunshot residue, may be visible, but the H & E stain is not designed to

react with non-biological material. Therefore, histology cannot be used for deﬁnitive

identiﬁcation of GSR and hence is very limited for this application.2

13

2.2 Microwave Digestion of Porcine Tissue

Analytical techniques capable of multi-element quantitation require that solid
matrices, such as skin tissue, be converted into a liquid prior to analysis. As biological
samples contain a mixture of proteins, lipids, and carbohydrates, they are not completely
soluble in either aqueous or organic solvents. It is therefore necessary to decompose the
biological matrix, which can be achieved using microwave digestion. While there are
numerous methods designed to microwave digest biological samples, the majority of
sample preparation protocols employ closed systems. Biological samples to be digested
are sealed inside vessels containing strong oxidizing agents such as acids and peroxides.

Nitric acid is by far the most common digestion agent as it effectively decomposes the

. . . . . . 4
orgamc matrix of the sample and liberates elements into solution as soluble nitrate salts.

The closed microwave digestion vessels are placed in a microwave oven which is very
similar to domestic microwave ovens. Microwave irradiation causes polar molecules
within a sample to rotate, causing friction with neighboring molecules and releasing heat.
As the sample is heated, the pressure inside the sealed vessel increases due to the
evaporation of the digestion acids and gases that are evolved during the breakdown of the
biological matrix. As pressure within the vessel increases, the boiling point of the
digestion reagents also increases, resulting in more efﬁcient breakdown of the sample
matrix. This build-up of pressure is closely controlled by monitoring the temperature
inside the digestion vessels using a temperature probe. Microwave power is only applied
when the temperature reading is below the user speciﬁed level, ensuring that the

temperature and pressure inside the microwave vessel do not exceed safety limits. The

14

resulting digest solution is free of solid material and contains all analytes of interest in a

soluble form that is compatible with the desired analytical detection method.4

Open microwave digestion systems, while devoid of pressure build-up problems,
require an effective fume removal system. The continuous removal of vapors decreases
the concentration of digestion regents in the vessel, and hence the digestion efﬁciency
decreases as well. The variable concentration of digestion reagents remaining in the
vessel also decreases the precision of replicate digestions in open systems. The danger of
sample contamination or loss through the open vessel is undesirable as well. Therefore,
the major advantages of a closed microwave digestion system are that high heating

efﬁciency is obtained with a decreased risk of sample loss. The precise monitoring of

temperature inside the closed digestion vessels also yields very reproducible digestions.4

2.3 Inductively Coupled Plasma Mass Spectrometry

Inductively coupled plasma mass spectrometry (ICP-MS) is one of the most
widely used techniques for elemental analysis because of its low detection limits, high
degree of selectivity, large linear dynamic range, short analysis times, and good precision
and accuracy. The concentrations of numerous elements may be determined
simultaneously in a liquid sample using this technique. Elements are ionized using a
plasma torch and then separated and detected using a mass spectrometer. The mass
spectrometer separates the elements by their mass to charge (m/z) ratio and detects the
abundance of each element present in the sample. Qualitative, semi-quantitative, and

quantitative analyses are all possible with the addition of internal standards and the use of

. . 5
external calibration curves.

15

A schematic of an ICP-MS instrument is shown in Figure 2.1. The liquid sample
solution is introduced into the argon (Ar) plasma torch, and the resulting ions are directed
through two nickel-plated cones into the mass spectrometer, where they are separated
based on their m/z ratio by a quadrupole mass analyzer. Although other mass analyzers,
such as time of ﬂight or ion trap, may also be used for ion separation, in this research a

quadrupole mass analyzer was used. The ions are then detected using a conversion

dynode and electron multiplier.5

Sampling Skimmer

RF generator cone cone Quadrupole
Sample coil

/
Ar NebulizeraniAL—Q‘ ‘ b A (3)) 1/ ﬂ
1 WI \ PlastjTl (L ) \\ Electron

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Per istaltic Plasma j l [on multiplier
pump Ar Plasma Low High path detector
torch vacuum vacuum
pump pump
\NSample solution

 

 

 

Figure 2.1 — Schematic of an ICP-MS system. ,

The ICP torch is composed of three concentric quartz tubes through which
streams of Ar ﬂow. The innermost tube contains sample droplets carried toward the
plasma by an Ar stream. Argon ﬂowing through the middle tube supports the plasma. The
outermost Ar ﬂow cools the induction coil surrounding the torch. The plasma is
generated by exposing the ﬂow of Ar gas to Tesla sparks that initiate ionization of the

ﬂowing gas. The resulting ions and electrons then interact with the high-frequency

l6

oscillating inductive ﬁeld created by the RF current generated in a coil wrapped around
the torch. The ions and electrons are consequently accelerated, collide with more Ar
atoms, and subsequently ionize them. This ionization cascade continues until the
ionization process is balanced with the ion-electron recombination process, resulting in

the plasma. All of the ion and electron collisions cause heating of the plasma to

approximately 10,000 K.6

The sample solution is introduced into the plasma torch using a peristaltic pump
and nebulizer. A ﬂow of Ar gas through the nebulizer produces a ﬁne aerosol spray of
sample that is then carried into the hot plasma. The high temperature of the plasma
rapidly desolvates, vaporizes, atomizes, and ionizes the elements in the sample into
singly charged positive ions. The ions are then extracted from the plasma using a
negative potential and are introduced into the mass spectrometer portion of the instrument

through two cooled nickel cones. The sampling and skimmer cones separate differentially

. . . -5 ' .
pumped regions of an interface which separates the 10 torr vacuum envrronment of the

mass spectrometer from the plasma torch ion source at atmospheric pressure. Once inside

the mass spectrometer, the ion beam is accelerated and focused using an ion lens toward

6
the entrance of the quadrupole mass analyzer.

The quadrupole mass analyzer uses the stability of ion trajectories in oscillating
electric ﬁelds to separate them according to their m/z ratio. The analyzer consists of four
parallel cylindrical rods. Opposite rods are connected electrically, with one pair
connected to the negative terminal of a variable direct current (dc) source and the other

pair connected to the positive side. A variable radio-frequency alternating current (ac)

l7

potential is also applied to the pairs of rods. As ions are accelerated into the quadrupole,
the ac and dc voltages on the rods are increased simultaneously so that their ratio remains
constant. The ions travel in a helical pattern through the quadrupole, and at any given
moment, only those ions with a certain m/z value have stable trajectories and hence reach
the detector. All other ions collide with the rods, are neutralized, and are not detected.
The quadrupole can be used to scan a range of m/z values (full mass scan), or can be used
to allow only selected masses to reach the detector (selected ion monitoring). The latter,

also known as SIM, increases the sensitivity of the instrument as more time is spent

. . . . . 5
monitoring the Signals from the ions of interest.

The ions that pass through the quadrupole are detected using a conversion dynode
and electron multiplier. The positively charged ions are accelerated toward the
conversion dynode using a -15 kV potential. When the ion strikes the conversion dynode,
secondary electrons are ejected from the surface. These electrons are accelerated toward
the next dynode, which is held at a more positive potential, and when they strike the
dynode more secondary electrons are ejected from the surface. This cascade of electron
multiplication results in an ampliﬁed current that is directly proportional to the number of
ions of that speciﬁc m/z ratio that passed through the quadrupole at a given time. It is
important to note that the conversion dynode is slightly offset from the exit of the
quadrupole to prevent light from the plasma torch from striking the conversion dynode,
as that would create secondary electrons as well and cause errors in ion abundance
measurement. The output of the detector is a mass spectrum which is a plot of ion

abundance versus m/z ratio. Ion abundance corresponds to the concentration of each

18

element present in the sample, so that quantiﬁcation of elements is possible using this

. 6
technique.

2.4 Statistical Procedures
2. 4. 1 Analysis of Variance

Analysis of variance (ANOVA) is a powerful statistical procedure that can be
used to separate and estimate the different sources of variation in a sample set. It is used
when there are more than two means to be compared in an analytical experiment, and
thus two possible sources of variation: random error, which is always present, and
variation due to a controlled or ﬁxed-effect factor, i.e. an experimental variable that is
purposely varied. It is used to separate any variation caused by changing the controlled
factor (between-sample variation) from the variation due to random error (within-sarnple
variation), which is useful for testing whether the controlled factor itself leads to a
signiﬁcant difference between the mean values obtained. Analysis of variance can also be
used in situations where there are two random sources of variation, such as the random
nature of sampling aliquots of a whole specimen for testing, in addition to the ever-
present random error in measurement. Since the sampling is done at random, the
variation will be random and is termed a random-effect factor. Analysis of variance is
still useful for separating and estimating two random sources of variation. Both types of

statistical analyses, i.e. where there is one factor, either controlled or random, in addition

. 7
to the random error in measurement, are known as one-way ANOVA.

In the case of two random factors, one-way ANOVA is used to separate and

estimate the different sources of variation: that due to random error, called measurement

19

variance, and that due to real variations in the sample, called sampling variance, which is
a random-effect factor. In this case, the calculated Within-sample variance is used to
estimate the measurement variance. However, the calculated between-sample variance is
no longer an appropriate measure of the sampling variance because the variation between
sample means is caused by both measurement variance and sampling variance
simultaneously. Therefore, methods to adjust the between-sample variance calculation to

more accurately estimate the variation due to sampling variance alone need to be
7, 8
employed.

The magnitudes of the different sources of variation are calculated separately
using sum of squares terms in order to test whether the difference between the sample
means is too large to be explained by the random error alone [Table 2.1]. In general, a
sum of squares term is used to calculate the sum of the differences between each
individual measurement and the mean of all measurements. The difference is squared to
make all the differences positive, instead of some positive and some negative, so that they

can be accurately summed. The within-sample mean square is determined by calculating

. . . . 7
the difference between the total and between-sample mean squares for srmphcrty.

20

Table 2.1 - Summary of sum of squares and degrees of freedom calculations.7

 

Source of Variation Sum of Squares Degrees of Freedom
T2 T 2
Between-samples —l— — — h - 1
,- n N
Within-samples by subtraction by subtraction

 

2 T2
Total 22x1]. —— N-l
i j N

where: n = number of replicates
h number of samples
N = nh = total number of measurements
xij = one measurement
T,- = sum of the measurements in the ith sample
T = sum of all the measurements, grand total

The sum of squares terms have different degrees of freedom associated with them,
and hence cannot be directly compared. Degrees of freedom is a determination of the
number of measurements in a dataset that are free to vary, which is generally all of them
minus one. The minus one restriction is necessary when the calculation of an estimate of
one statistic (i.e. sum of squares) is based on an estimate of another statistic (i.e. mean).
Sum of square terms with different degrees of freedom can be compared based on the

mean square term, which is calculated as follows:

mean square = sum of squares/degrees of freedom.

The mean squares are the variances from different sources that are comparedto

determine if the variance between samples (ﬁxed- or random-effect variance) is

21

signiﬁcantly greater than the variance within samples (measurement variance). To do

this, the mean square terms are used in an F -test as follows:

F = between-sample mean square/within-sample mean square.

The resulting value is then compared to a statistical table of critical F h-l NJ, values at the

desired conﬁdence level. If the critical value of F is larger than the calculated value of F,

the two sources of variance are not signiﬁcantly different and all variance in the dataset is
due to random measurement variance. Conversely, if the calculated F value is larger than
the critical F value, the two mean square values are signiﬁcantly different, and there is

variance between samples due to either a ﬁxed- or random-effect factor in addition to the

random measurement variance.

In the case of a signiﬁcant random-effect factor (i.e. sampling variance),
corrections may be employed to the between-sample mean square calculation to more
accurately estimate the magnitude of the sampling variance alone. There are two

corrections that can be used as estimators of the sampling variance: the method of

. . 8 .
moments and the Klotz-Milton-Zacks estimator. The method of moments IS the most

straightforward, and is calculated as follows:

sampling variance = (between-sample mean square — within-sample mean square)/n.

22

The corrected sampling mean square is the between-sample mean square, which
estimates random variance due to both sampling and measurement, less the within-
sample mean square, which is an accurate estimate of the measurement variance alone.
The Klotz-Milton-Zacks estimator is more sophisticated and accurate for representing the

mean squared terms. The estimator is calculated as follows:

 

sampling variance = -1—
n

SS(A) _ SS(E)
h + 1 h(n — 1)
where SS(A) is the between-sample sum of squares and SS(E) is the within-sample sum of

squares. The only difference between the Klotz-Milton-Zacks estimator and the method

. 8
of moments is the degrees of freedom used to calculate the mean square terms.

The corrected estimate of sampling variance is tested for signiﬁcance using the F
test along with the original within-sample mean square and the same critical value of F,
as described above. If the corrected between-sample mean square is not signiﬁcantly
different from the within-sample mean square, then all variance in the sample set is due
to random sampling and measurement variance. If the corrected between-sample mean

square is signiﬁcantly different from the within-sample mean square, then there is

. . 7
another source of variance in the samples.
One-way ANOVA calculations with random-effect factor corrections are useful
for drawing conclusions about the variability of GSR deposition around a gunshot

wound. For an element to be useful for determining the presence of GSR, it must be

reproducibly detected using ICP-MS (low measurement variance) and somewhat evenly

23

distributed around the entrance wound so that sampling the wound at any location will
yield positive results (low sampling variance). Low levels of measurement and sampling
variances demonstrate the element’s utility for differentiating wounds shot with two

different bullet types.

2. 4. 2 Principal Components Analysis

Principal components analysis (PCA) is a multivariate statistical procedure for
reducing the amount of data in a dataset when there are correlations present within the
data. A data matrix is created containing a list of all of the samples in the dataset and the
amounts of all of the variables present in each sample. A covariance matrix is then
calculated, which examines how the different variables relate to one another. For
example, the covariance matrix will determine if two different variables always increase
and decrease together in the same samples, meaning that they may be correlated. From
the covariance values, linear combinations of variables that ﬂuctuate similarly are
calculated. These linear combinations are called eigenvectors, and are the principal
components (PCs) of the analysis. Each eigenvector has an associated eigenvalue, which
describes the amount of information contained in the eigenvector. Therefore, the higher
the eigenvalue, the more valuable the eigenvector (PC) is for describing the overall
variance in the dataset. The same number of PCs are calculated as the original number of
variables in the dataset. Each PC is calculated to be orthogonal to the preceding PC and
describes the next greatest amount of variance in the dataset. Principal component 1 is
calculated to correlate the highest sources of variation in the dataset, PC2 is orthogonal to

PCI and correlates the next highest sources of variation, and so on. The result of PCA is

24

that the largest and most informative sources of variance in the dataset can now be
examined using a smaller number of variables (i.e. two or three PCs), instead of all of the
original data variables. This also allows simple visualization of data trends as two PCs,
such as PCI and PC2, may be plotted as the x- and y-axes of a single graph, called a
scores plot, which illustrates most of the variance in the samples projected into two
dimensions. Samples with similar chemical composition will cluster together and away
from chemically dissimilar samples on the scores plot, allowing conclusions to be drawn
about how samples relate to one another. A corresponding plot called a loadings plot
displays the variables that are responsible for the variance described by each PC.
Variables that ﬂuctuate similarly in the same samples will be positioned in the same area
of the loadings plot, allowing visualization of variable correlations. Again, PCl is the x-

axis and PC2 is the y-axis so that the position of a variable on the loadings plot

. . . 7
corresponds to sample pOSitioning on the scores plot.

The following example illustrates the PCA procedures with respect to GSR
element concentrations in tissue samples. The dataset would be composed of the
abundances of several elements (variables) in numerous tissue samples of different types,
such as unshot or shot with jacketed or non-jacketed ammunition (samples). Analyzing
for trends in the abundances of elements in the different sample types individually can
become tedious and time consuming. However, if correlations are present between
element abundances and sample types, a new set of variables, the PCs, may be calculated.
For example, if lead and antimony concentrations are always higher in tissue samples
shot with non-jacketed ammunition compared to jacketed ammunition, and copper is

always more concentrated in jacketed wound samples, a single PC is calculated to

25

represent this correlated variation. Linear combinations of elements that vary similarly in
the covariance matrix are calculated as PCs that describe the correlations. The PCs with
the largest eigenvalues describe the largest sources of variance in the different tissue
samples. Then PCl and PC2, the two eigenvectors that describe the greatest sources of
variance, are plotted as scores and loadings plots [Figure 3.2]. Tissue samples with
similar chemical composition will group together in one area of the scores plot and away
from other tissue sample groupings. The elements combined into PCI and PC2 that
describe the highest sources of variance in the dataset are plotted in the loadings plot.
Elements that vary in similar ways will be grouped together in the loadings plot. The
element positions in the loadings plot correspond to sample positions in the scores plot.
For example, if lead and antimony are positioned negatively on PCl in the loadings plot,
then samples shot with non-jacketed ammunition that have high levels of both of those
elements will be positioned negatively on PCI in the scores plot. Likewise, samples shot
with jacketed bullets that have relatively low levels of lead and antimony and high levels
of copper will be positioned away from the samples shot will non-jacketed ammunition
and positively on PCI. The loadings plot shows the position of copper as positive on

PCI, indicating that levels of copper vary inversely to levels of lead and antimony in this

7
dataset.

26

Scores plot

 

 

 

 

Jacketed
samples
N
. Q On
8 . O V
Non-j acketed
samples

 

 

PCI

PC2

Loadings plot

 

 

Lead

Copper

 

Antimony

 

 

PCI

Figure 2.2 — Example results of PCA visualized as scores and loadings plots.

Principal components analysis reveals complex correlations between variables

and samples that may not be discovered with univariate statistical methods. These

correlations will be valuable for drawing conclusions about which elements are most

useful for differentiating shot and unshot tissue as well as tissue samples shot with

different ammunition types.

27

 

2.5 References
1. Geneser F. Textbook of histology. Copenhagen: Munksgaard, 1986.

2. Kieman JA. Histological & histochemical methods: theory and practice. 2nd ed. New
York: Pergamon Press, 1990.

3. Masayoshi H, Yasuo S, Makie K, Hiroyoshi O. Mechanism of hematoxylin and eosin
stains. Journal of Analytical Bio-Science 2003;26:377-83.

4. Lamble KJ, Hill SJ. Microwave digestion procedures for environmental matrices.
Analyst 1998;123:103R-33R.

5. Skoog DA, Holler FJ, Nieman TA. Principles of instrumental analysis. 5th ed. Toronto:
Thomson Learning, 1998.

6. deHoffmann E, Stroobant V. Mass spectrometry: principles and applications. 3rd ed.
West Sussex: John Wiley & Sons, 2007.

7. Miller JN, Miller JC. Statistics and chemometrics for analytical chemistry. 5th ed.
Harlow: Pearson Education Limited, 2005.

8. Miller, RG. Beyond ANOVA, basics of applied statistics. New York: John Wiley &
Sons, 1986.

28

CHAPTER 3: MATERIALS AND METHODS
3.1 Experimental Design and Sample Collection
3.1.] Fresh Tissue Studies

Two studies were conducted using fresh tissue samples. The ﬁrst study was used
to identify elements that were potentially useful in differentiating the two bullet types, as
well as to assess variation in GSR composition within a wound. The second study was
used to quantify the elements of interest to determine statistical differences in element
concentrations between the two bullet types.

For the fresh tissue studies, two euthanized pigs (approximately 150 lbs each)
were obtained from the Michigan State University (MSU) Swine Research Facility in
June 2009. All pigs used in this research were treated in accordance with the MSU
Institutional Animal Care and Use Committee (IACUC) guidelines. A certiﬁed ﬁrearms

instructor from the MSU Campus Police shot both pigs seven times using a Smith &

Wesson® .357 Magnum revolver. One pig was shot using 158 grain copper jacketed

hollow point ammunition with a jacketed bullet base (Remington Arms Co. Inc., Lonoke,
AR, Lot # B 17 HAS 502). The second pig was shot using 158 grain non-jacketed lead
ammunition (Remington Arms Co. llnc., Lot # G 2l YB 6102). All shots were ﬁred at a
muzzle-to-target distance of 5 cm and the wounds were spaced approximately 10 cm
apart to prevent cross contamination between wounds. The gun barrel and chamber were
also cleaned between the different ammunition types. Immediately after all shots were
ﬁred, the wounds were collected by excising the tissue and underlying fat in an

approximately 4 cm radius around each wound. Samples were also removed from these

29

wounds for histology analysis. The tissue samples were wrapped loosely in waxed paper,
sealed in plastic bags, and stored at -80 °C until analysis.

In July 2009, one euthanized pig (approximately 100 lbs) was obtained from the
MSU Swine Facility to serve as a control. Four samples of unwounded skin and
underlying fat (approximately 20 cm x 20 cm each) were removed, packaged, and stored
until analysis as described above.

For the ﬁrst fresh tissue study, two shot tissue samples (one for each bullet type)
and one control sample were used. Three tissue sections were removed from each
gunshot wound and one section was removed from the control tissue sample. The tissue
sections were microwave-digested and analyzed by ICP-MS, following procedures
described below. For the second fresh tissue study, 10 shot tissue samples (ﬁve from each
bullet type) and one control sample were used. One tissue section was taken from each
sample, microwave-digested, and analyzed by ICP-MS, quantifying the elements of

interest.

3 . I . 2 Sampling Study

In June 2009, one euthanized pig (approximately 200 lbs) was obtained from the
MSU Swine Research Facility. The pig was shot ﬁve times with the same ﬁrearm and
non-jacketed ammunition used in the fresh tissue studies described above. All shots were
ﬁred at a muzzle-to-target distance of 5 cm and were spaced approximately 20 cm apart
to prevent cross contamination between wounds. The gun barrel was cleaned after every
shot to make GSR deposition as reproducible as possible. Immediately after all shots

were ﬁred, the wounds were collected by excising the tissue and underlying fat in an

30

approximately 8 cm radius around each wound. The wounds were packaged and stored
until analysis as described above.

One wound was sampled for the study by excising tissue sections in four
directions extending out from the entrance wound: 12 o’clock (12:00), 3 o’clock (3:00), 6
o’clock (6:00), and 9 o’clock (9:00) [Figure 3.1]. The tissue was sampled directly
adjacent to the wound (0 cm) and every 2 cm out to the wound edge in each direction.
The 19 tissue sections (ﬁve in each direction except 3:00, where the wound edge was at 6
cm) as well as three control tissue samples from the same control pig used in the fresh
tissue studies were then microwave digested and analyzed by ICP-MS to quantify the

elements of interest at each location around the wound.

9:00 3:00

 

Figure 3.1 — Diagram of wound where tissue samples were excised for analysis.

31

3. 1. 3 Decomposition Study

In October 2009, three euthanized pigs (approximately 150 lbs each) were
obtained from the MSU Swine Research Facility. Control samples of tissue were
removed from each pig before wounding. Two pigs were shot 12 times each with the
same gun and ammunition as the fresh tissue studies. As before, one pig was shot with
the jacketed ammunition and the second pig was shot with the non-jacketed ammunition.
The gun barrel was cleaned after each shot, and the chamber was cleaned before
reloading. All three pigs were then transported to a research ﬁeld, where the unshot pig
was stabbed 12 times to create open wounds to attract insect activity similar to the shot
pigs. The pigs were placed inside separate wire cages to protect them from predators
while still allowing exposure to the environment. The wounds and histology samples
were collected over a period of 49 days, following procedures for excision, packaging,
and storage described above. One tissue section was removed from each tissue sample
(shot and control) for microwave digestion and ICP-MS analysis to quantify the elements

of interest.

3.2 Histology

Tissue sections were taken from each wound in the area of highest soot density
and placed in formalin ﬁxative (10% buffered for the fresh tissue wounds, Fisher
Diagnostics Co. LLC, Kalamazoo, MI; 37% formaldehyde solution for the decomposed
tissue wounds, Columbus Chemical Industries, Inc., Columbus, WI). The tissue sections
were processed for hematoxylin and eosin (H & E) staining by the Hurley Medical Center

Pathology laboratory according to the established and standard staining procedures of

32

that lab. Stained tissue sections were evaluated using a light microscope (Nikon Eclipse

50i equipped with a 2x-60x lens, Nikon Inc., Melville, NW.

3.3 Microwave Digestion of Porcine Tissue

Tissue samples were removed from the -80 °C freezer and allowed to thaw prior
to sample preparation. Skin directly adjacent to the wounds was removed from the
underlying fat using a disposable scalpel. Between 0.35 and 0.5 g of each tissue sample

were placed into acid and peroxide washed quartz vessels (Milestone, Inc., Shelton, CT).

A 1 mL aliquot of hydrogen peroxide (H202, 30%, Columbus Chemical Industries, Inc.)

and 2 mL Optima grade nitric acid (HNO3, 69%, Fisher Scientiﬁc, Pittsburgh, PA) were

added to each vessel. The quartz vessels were then capped and placed into larger

TeﬂonTM vessels (Milestone, Inc.) that contained 10 mL Milli-Q water (Milli-Q

Academic, Millipore, Billerica, MA) and 2 mL 30% H202. The TeﬂonTM vessels were

then securely closed and placed into a Milestone Ethos EX microwave digester
(Milestone, Inc.). Samples were digested using the following microwave temperature
program: 15 minute temperature ramp from room temperature up to 210 °C, followed by
a 10 minute hold at 210 °C. The microwave wattage was set at 1000 W to provide
enough power to heat all vessels. Nine tissue samples and one procedural blank (prepared
in the same way but without tissue present) were analyzed per microwave digestion run.
The temperature probe was placed inside the procedural blank vessel, and the microwave

system automatically adjusted the applied wattage to obtain and maintain the desired

33

temperature. The quartz vessels were cleaned after each tissue digestion by running
procedural blanks and rinsing with distilled water before reuse.

After digestion, the tissue digests were allowed to cool to room temperature and

diluted to 2% HNO3 by adding 0.5 mL of the digest solutions to 11.168 mL Milli-Q

water in 15 mL polypropylene conical tubes (Corning, Inc., Corning, NY). The dilution
was necessary to prevent degradation of the polypropylene tubes and decrease the acid
concentration for ICP-MS analysis. The remaining concentrated digest solutions were
transferred to separate 15 mL polypropylene conical tubes for storage as well. Both the

concentrated and diluted digest solutions were stored at 4 °C until ICP-MS analysis.

3.4 ICP-MS Analysis of Fresh Tissue Samples

The seven tissue samples from the ﬁrst fresh tissue study (three samples from one
wound from each bullet type and one control sample) were analyzed using full mass scan
mode to identify all elements present at signiﬁcant levels. Prior to analysis, the diluted

tissue digest solutions and procedural blanks were further diluted 1:10 (v/v) with 2%

HN03 and spiked with 20 pg/L In and 20 ug/L Bi (SPEX CertiPrep, Inc., Metuchen, NJ)

as internal standards. The solutions were analyzed by ICP-MS using a Micromass
Platform quadrupole ICP-MS (now Therrno Fisher Scientiﬁc, Inc., Waltham, MA.) The

instrument was equipped with a CETAC ASX-SOO autosampler (CETAC Technologies,

Omaha, NE), and a DynoliteTM detector with a ~15 kV conversion dynode and electron

multiplier. Instrument operating parameters are given in Table 3.1. The instrument was

operated in full mass scan mode using MassLynx software (version 3.4, Waters Corp.,

34

Milford, MA). Digest samples and procedural blanks were analyzed in triplicate in

random order. After sample injection, the injector was rinsed for 90 seconds with 2%

HNO3 to prevent sample carryover.

Table 3.1 — ICP-MS parameters for full mass scan and selected ion monitoring (SIM)

analyses.

 

ICP-MS Operating Parameters
RF power (W)
Argon ﬂow rates (L/min)
Outer
Intermediate
Nebulizer
Sampling cone
Cone Voltage (V)
Skimmer cone
MS resolution
Hexapole gas ﬂow rates (mL/min)
Helium
Hydrogen
Hex Bias (V)
'Data Collection Parameters

Mode

Sample scan time (s)
Dwell time (s)
Interchannel delay time (s)

Autosampler Parameters
Sample read delay (5)
Rinse time (s)

1350

13

08-095

0.67-0.7

Ni with Cu core, 1.14 mm diameter oriﬁce
100-125

Ni, 0.89 mm diameter oriﬁce

Unit mass

45
2
4.0-1.0

Full Mass Scan or Selected Ion Monitoring
(Peak jumping)

75

0.2 (SIM mode only)

0.020 (SIM mode only)

105
90

Instrument responses for elements below atomic mass 155 were normalized to

l

15 . . . . 209 . . .
In, and elements wrth atomic weights above 155 were normalized to B1. Dilution

35

factors were taken into account and concentrations of all elements were expressed as ug

element/ g tissue for comparison.

3.5 Statistical Treatment of Fresh Tissue Data

All statistical analyses were performed using Microsoft Excel 2003 (Microsoft®

Corp., Redmond, WA). The average elemental concentration (pg/ g tissue) was calculated
for each bullet type (three samples analyzed in triplicate from each wound, giving n = 9),
the control sample (triplicate analyses, n = 3), and the procedural blank sample (triplicate
analyses, 11 = 3). The limit of detection (LOD), deﬁned as three times the standard
deviation of the procedural blank concentration, was calculated for each element, as well
as the ratio of the average procedural blank concentration to the lowest concentration in a
shot tissue sample from either bullet type. The relative standard deviation (RSD) was

calculated for each element in each bullet type, and the Grubbs’ test for statistical outliers

was performed on all elements that had RSDs higher than 15%.1 For all elements, any

data points determined to be outliers at the 95% conﬁdence level were removed.
Elements with an RSD greater than 15% in both bullet types after removing statistical
outliers were eliminated from the suite of potentially useﬁil elements.

Analysis of variance (ANOVA) at the 90% conﬁdence level (p S 0.1) was used to
assess signiﬁcant differences in element concentrations between-wounds and within-
wounds, using triplicate analyses of three tissue samples from each bullet type. The null
hypotheses were that there was no difference in the elemental concentrations of replicates
from the same sample and that there was no difference in the elemental concentrations of

samples taken from the same wound. The alternate hypotheses were that the elemental

36

concentrations of replicates and between samples taken from the same wound were not

equal. Random effect corrections were applied to between-sample variances to correct for

. . . . 2
the random nature of sampling the tissue for analysrs, when appropriate.

Student’s t-tests were also performed to compare the average concentration of
each element between the two bullet types. The calculated t—statistic for each element was
compared with statistical tables of critical t-values at the 95% and 99% conﬁdence limits

(two-tail) to assess statistical differences in element concentrations.

3.6 ICP-MS Quantitation of Fresh and Decomposed Tissue Samples

The 11 tissue samples from the second fresh tissue study (one sample from ﬁve
wounds for each bullet type and one control), tissue samples from the decomposition
study (n = 38), and all tissue samples from the sampling study (n = 22) were analyzed
using selected ion monitoring (SIM) mode for the selected elements of interest to

increase sensitivity. The instrument tune conditions were optimized daily using a 10 [lg/L

solution of Be, Co, In, Ce, Bi, and U (SPEX CertiPrep, Inc.) prepared in a 2% HNO3

solution. Ten multi-element external calibration standards containing the elements of
interest were prepared from stock standard solutions of each element (1000 mg/L each; P,

K, Fe, Cu, Sb, Ba, and Pb from SPEX CertiPrep, Inc.; Mg and Zn from Thermo Fisher

Scientiﬁc, Inc.) The stock solutions were diluted with 2% HNO3 to concentrations

ranging from 0.1-500 rig/L, with all elements having equal concentrations in each

standard solution. Each standard solution was spiked with 20 rig/L In and 20 pig/L Bi as

37

internal standards prior to ICP-MS analysis. Calibration standards were analyzed in order
from low to high concentration to minimize carry-over effects.

Standard Reference Material 1643c (SRM, National Institute of Standards and
Technology, Gaithersburg, MD) was analyzed to assess instrument accuracy during

sample analyses as the standard contained known concentrations of all elements of

interest. The SRM was diluted 1:10 (v/v) with 2% HNO3 and spiked with 20 rig/L In and

20 rig/L Bi internal standards prior to ICP-MS analysis.
Due to the large range in concentrations for the different elements in shot tissue
samples, it was necessary to prepare the tissue digest samples at two different dilution

levels. After microwave digestion, the digest solutions (fresh tissue, decomposed tissue,

control tissue, and procedural blank) were diluted to 2% HNO3 as described previously.

One sample set was not further diluted (hereafter referred to as ‘undiluted’) prior to being

spiked with 20 ug/L In and 20 ug/L Bi as internal standards. A second sample set was

prepared by further diluting the digests 1:100 (v/v) with 2% HNO3 (hereafter referred to

as ‘diluted’). The diluted digests were then spiked with 20 ug/L In and 20 ug/L Bi as
internal standards.

The undiluted and the diluted digests were analyzed by ICP-MS using the same
instrument system described previously with additional parameters for SIM mode
analysis, as detailed in Table 3.1. Sets of calibration standards were analyzed at the
beginning of an analysis and after approximately every 30 samples to account for
instrument driﬁ. Samples were analyzed in the following order for all studies: calibration

standard set, SRM, diluted digests and procedural blanks randomized, SRM, calibration

38

standard set, SRM, undiluted digests and procedural blanks randomized, SRM,

calibration standard set. After each sample injection, the injector was rinsed for 90

seconds with 2% HNO3, and two additional 2% HNO3 rinses were performed between

each set of samples.

The resulting instrument responses were quantitated using MassLynx software.

24 31 39 56 63 64 .
Instrument responses for Mg, P, K, Fe, Cu, and Zn were normalized to

l

115 12 138 208 . 209 . . .
In, and Sb, Ba, and. Pb were normalized to B1. Element concentrations in

the tissue samples were determined from the average of the linear calibration curves run
immediately before and after the digest sample set to be quantiﬁed. Element
concentrations in each tissue digest were corrected for dilution factors and normalized to
the original tissue mass to yield ﬁnal element concentrations that were expressed as pg

element/ g tissue.

3.7 Principal Components Analysis of Decomposition Study Data
Principal components analysis (PCA) was performed using SIMCA-P software

(version 1l.5, Umetrics Inc., San Jose, CA). The data matrix included the concentrations

1

(pg element/g tissue) of the ﬁve elements of interest (56Fe, 63Cu, '2 Sb, 138Ba, and

208Pb) in each sample in the deComposition study (procedural blanks, control stab

wounds, wounds shot with jacketed ammunition, and wounds shot with non-jacketed
ammunition) determined using ICP-MS quantitation. Only samples collected on days 0,
5, 14, 24, 34, and 44 were included as the element concentrations did not vary greatly

throughout decomposition. The element concentrations were mean-centered and

39

univariate scaled prior to PCA, which are common data pretreatment procedures. Mean-
centering shifted the concentrations of the elements so that the means were centered at
the origin, which makes results interpretation more straightforward. The element
concentrations were univariate scaled so that the elements present in large abundance
would not dominate the analysis and undervalue the low abundance elements in terms of
their ability to differentiate between the tissue samples as well.

Principal components analysis was performed to examine the systematic variation
of element concentrations in the decomposition study samples and to visualize the
differentiation of samples shot with different bullet types. Principal component 1 (PCI)
was plotted versus PC2 as these components described the elements that varied the most
among the different tissue sample types. The corresponding loadings plot of PC] versus
PC2 was examined to determine which elements varied between the different sample
types and were responsible for associating and discriminating sample groupings in the

scores plot.

40

3.8 References

1. Miller JN, Miller JC. Statistics and chemometrics for analytical chemistry. 5th ed.
Harlow: Pearson Education Limited, 2005.

2. Miller, RG. Beyond ANOVA, basics of applied statistics. New York: John Wiley &
Sons, 1986.

41

CHAPTER 4: RESULTS AND DISCUSSION
4.1 Introduction

The differentiation of two bullet types based on gunshot residue (GSR) was ﬁrst
investigated using fresh porcine tissue. One pig was shot with jacketed ammunition and
two pigs were shot with non—jacketed ammunition. The wounds were collected,
microwave digested, and analyzed using inductively coupled plasma mass spectrometry
(ICP-MS) to determine elemental content. Elements in addition to those already
considered characteristic of GSR (antimony (Sb), barium (Ba), and lead (Pb)), were
identiﬁed in order to increase conﬁdence in gunshot wound determination. The ability of
ICP-MS to differentiate the two bullet types using their element proﬁles was also
investigated. A sampling study was used to investigate the spatial pattern of GSR
deposition around a fresh non-jacketed bullet wound. Finally, to make the results more
applicable in a forensic setting, the effects of moderate decomposition on the ability to

detect GSR and differentiate bullet types were also assessed.

4.2 Fresh Tissue Sample Observations and Histology Results

Dense soot deposition was visible in the wound tract and around the immediate
perimeter of wounds made by both the jacketed and non-j acketed bullets [Figure 4.1].
Wound edges were darkened, exhibiting typical characteristics of a close-range gunshot
wound. No stippling was observed because the pigs were euthanized prior to being shot.
Microscopic examination of the wounds revealed soot staining in the epidermis.
Histology results after hematoxylin and eosin (H & E) staining showed the expected

thermal artifact associated with gunshot wounds (i.e. homogenization of the epidermis

42

with streaming nuclei) in both bullet types, as well as dark granules of burned gunpowder
and refractive particles of unburned gunpowder [Figure 4.1]. There were slight
differences in soot deposition patterns and histology results between the two ammunition
types due to different types and amounts of gunpowder, but this was not a feature of the

study.

(a) I” '3». tr

ll}, 1.9g“ \%‘& A! ' ‘ I'u; 1 ‘
1“" ii ‘1. ...;
\ ‘ EN . I .
~ xxx \
‘ \\

5

0.6 mm
_

 

Figure 4.1 - Gunshot wounds in fresh porcine tissue from (a) jacketed bullets and (b) non-
jacketed bullets, along with corresponding micrographs of H & E stained histology
results (c and (1, respectively). Refractive particles of unburned gunpowder are circled in
the jacketed bullet wound histology slide (c), and dark granules of burned gunpowder are

circled on the non-jacketed bullet wound histology slide (d).

43

4.3 Selection of Elements to Discriminate Bullet Types

The full mass scan ICP-MS analyses detected many potentially useful elements in
the fresh tissue digest samples. Element concentrations from ICP-MS analyses were
analyzed using a series of statistical procedures to select those elements most useful for
differentiating shot from unshot tissue as well as differentiating the two bullet types.
First, elements with average concentrations in tissue samples shot with both bullet types
that were below the instrument detection limits for that element were excluded from
further analyses. Next, if the ratio of the average procedural blank concentration to the
lowest concentration in a shot tissue sample from either bullet type was greater than 30%,
the element was excluded from further consideration since more than one third of the
element signal was due to sample preparation and/or instrument contamination. Finally, if
the relative standard deviation (RSD) for an element was greater than 15% for both bullet
types after elimination of statistical outliers, the element was excluded from further
consideration.

The number of possible elements of interest was slightly reduced by excluding
elements with concentrations in shot tissue samples below the instrument limit of
detection and below levels in procedural blanks. Many elements were also excluded due

to high RSD levels, leaving the following elements as potentially useful for

1

discrimination: 24LMg, 31P, 39K, 56Fe, 63Cu, 66Zn, 12 Sb, 138Ba, and 208Pb. It should be

noted that the RSDs for 121Sb, 138Ba, and 208Pb were higher than 15% in both bullet

types due to the high concentrations of these elements in the samples. However, the three
elements were included in subsequent analyses since they are currently considered to be

characteristic of GSR.

44

One-way analysis of variance (ANOVA) was applied to the three tissue samples
analyzed in triplicate from each bullet type to identify those elements that showed
signiﬁcant differences within and between wounds at the 90% conﬁdence level (p S 0.1).
At this p-value, the probability that the elemental concentrations were the same was 10%
or less. The variation was not signiﬁcant among replicates of the same sample, as
expected due to the high precision of ICP-MS. In addition, and for both bullet types, there

was no signiﬁcant difference in concentrations for samples taken from the same wound

. 24 31 39 56 63 66 . .
for the followrng elements: Mg, P, K, Fe, Cu, and Zn. There was Signiﬁcant
. . . 121 138 208 .
between-sample variance in element concentration for Sb, Ba, and Pb in

samples from jacketed bullet wounds, and in 138Ba and 208Pb for non-jacketed bullet

wounds. However, this variation was due to high concentrations of those elements in the
wound samples. In subsequent analyses, the digests were further diluted prior to
quantitation.

Finally, to test whether the elements of interest could be used to differentiate the
two bullet types, the Student’s t-test was used to compare the mean element
concentrations in each [Figure 4.2]. Elements with signiﬁcant differences in

concentration at the 95% conﬁdence level and above were considered useful for

differentiating the two bullet types. Concentrations of 24Mg, 31P, 39K, 63Cu, and 66Zn

were signiﬁcantly higher in wounds shot with jacketed bullets (99% conﬁdence level),

121 208 . . . .
the elements Sb and Pb were Signiﬁcantly more concentrated in wounds shot wrth

non-jacketed bullets (99% conﬁdence level), and 56Fe and 138Ba concentrations were not

45

signiﬁcantly different between the two bullet types. All elements were also conﬁrmed to
be useful for differentiating shot from unshot tissue by comparison of the mean element
concentrations in a similar manner.

Thus, the suite of elements considered potentially useﬁrl for differentiating shot

. . . . . 24 31 39
from unshot tissue and differentiating the two bullet types consrsted of Mg, P, K,

56 1

Fe, 63Cu, 66Zn, 12 Sb, 138Ba, and 208Pb. These elements could originate from the

ammunition primer, gunpowder, cartridge case, or the bullet itself, as well as from the

gun barrel.1 The three elements currently considered characteristic of GSR, Sb, Ba, and

Pb, come from the primer, and Pb and Sb could also come from the bullet. Copper, zinc
(Zn), and nickel (Ni) are known materials in bullet jackets and cartridge cases. However,
since the sampling cone of the ICP—MS was composed of Ni, data for that element are not
reliable in this study, and the element was not considered in analyses. The gun barrel
used in these studies was made of stainless steel, which is mainly composed of Fe.
Magnesium (Mg) and phosphorus (P) are likely oxidizers in the gunpowder, and
potassium (K) may come from the synthesis of the gunpowder. Further analyses are
required to determine the complete chemical composition of the ammunition as this was

not the focus of the current study.

46

.1. E @88me ohm Cod w 5 $32 859.38 foam 3
2235583 Soﬁa? 5333 monocotmw Hgomawmm dogged 935% one Scampi»: when. Hobo o5. .8 H : US$8.75: 98 .o H c

86on .m H a 75:8 .m u a EB 3258:: beam 253 smoﬂ «mum 05 E “88:: mo £5820 mo mcouﬂaooeoo :32 I new ocswﬁ

Beggar—8 Z I 133.2% I 3.5—SD D «Ea—m— ung—38.5 D

 

 

 

 

 

 

 

 

am so on w: ,E M

1 c O

0

_|_ u

.3. r 66¢ ”

m

.82 m

m.

r 52: m,
an

r. .83 em

47

4.4 Quantitation of Elements in Fresh Tissue Samples

. . 24 31 39 56 63 66 121 138
At this stage, nine elements( Mg, P, K, Fe, Cu, Zn, Sb, Ba,

and 208Pb) were identiﬁed as being potentially useful for differentiating shot from unshot

tissue as well as differentiating bullet type. The levels of these elements in the fresh tissue

samples (one control, ﬁve shot with jacketed bullets, and ﬁve shot with non-jacketed

bullets) were then quantitated by ICP-MS. However, the elements 24Mg, 31P, and 39K

were subsequently eliminated from consideration due to nonlinear calibration curves as a
result of high background levels in the instrument. Limits of quantitation for the
remaining elements were determined as the concentration of the lowest linear point on
the calibration curve or listed as less than the lowest standard concentration analyzed

[Table 4.1]. Koons reported that concentrations of Sb, Ba, and Pb typically observed in

GSR were 40-500 ng.2 The limits of quantitation determined for this study are between

one and three orders of magnitude lower than levels expected in GSR for those three
elements, indicating that the instrument was capable of detecting elements present in
GSR. The mean concentrations for all six elements of interest in the four standard
reference material (SRM) samples were calculated and compared to the reported
concentrations in the material [Table 4.1]. The percent error between the true and
observed Zn concentrations was 25% and hence, Zn was excluded from further
consideration. Errors for the remaining ﬁve elements (Fe, Cu, Sb, Ba, and Pb) were all
less than 10%, indicating that the ICP-MS was accurately quantifying the elements and

hence, one analysis of each digest sample was sufﬁcient.

48

Table 4.1 — Limits of quantitation (LOQs) and SRM recovery results for elements of

interest in the fresh tissue study.

 

Element LOQ (pg/L) SRM Average % Error

 

56Fe 5 7
63Cu 5 1
66Zn 10 25
1213b < 0.1 2
138Ba 0.5

208m) 0.25 3

 

Both the undiluted and diluted tissue digest sample sets were analyzed. The

elements 56Fe and 66Zn were quantiﬁed in the undiluted digests due to the low

concentrations of these elements observed in the ﬁrst fresh tissue study. The remaining

121

four elements (63Cu, Sb, 138Ba, and 206Pb) were one to three orders of magnitude

more concentrated and were quantiﬁed using the diluted digests. This was necessary to
avoid detector saturation and to ensure that the instrument responses were within the
linear range of the calibration curve for those elements.

A comparison of mean elemental concentrations in all tissue samples analyzed for
each bullet type, as well as procedural blank and control tissue samples, is shown in
Figure 4.3. The mean element concentrations in tissue samples shot with either bullet
type are considerably higher than the corresponding procedural blank and control tissue
sample concentrations for all ﬁve elements of interest, indicating that all elements can be
used to differentiate shot from unshot tissue. The Student’s t-test was used to test for

signiﬁcant differences in the mean element concentrations between jacketed and non-

49

jacketed bullet wounds. There were signiﬁcant differences between the two bullet types
in the Sb and Pb concentrations at the 99% conﬁdence level and in the Cu concentrations
at the 95% conﬁdence limit. Antimony and Pb were both more concentrated in tissue shot
with non-jacketed bullets, with mean concentrations of 5444 and 147565 pg/ g,
respectively, compared to 1732 and 6567 ug/g in wounds shot with jacketed ammunition.
Copper was more concentrated in tissue shot with jacketed ammunition, with a mean
concentration of 2524 ug/ g compared to 123 ug/ g in wounds shot with non-jacketed
bullets. These elements most likely originate from the bullets themselves, since the
jacketed bullets have Cu jackets and the non-jacketed bullets are composed of Pb and
possibly Sb. These three elements (Cu, Sb, and Pb) are therefore useful for differentiating

the two bullet types in fresh gunshot wounds.

50

.1. 3 88865 08 God w 5 :52 8:02.38
£30 8 3285588 E0820 28369 oococomtw agomﬁwmm can * 3 @8368 08 God w as .32 ooeowmcoo $3 “a mcouabcoocoo
EoEBo :oozzon cococﬂtw Enumemm 483288 9888 one “command when Hobo 23. .G H c @838 .35: 28 .m I 2 8883mm

4 H : Escoo d H a x53 326883 8383 25m: 58m E womucwaw 888$ MO 35820 .«o 82852850 :32 I md oSwE

138—93.58 Z I USS—own. I 3.5—SD D :55 geououm D

 

 

 
 
 

 

 

 

 

am am :0 am
a _ lilill e -c Q
r r u
. 8— .8.. 888 m
t . 88m 882: w
.88 r i
T 882 -8884 m.
. 8m - 88“ 8888 m
PL 3
. 8.. . 88~ :. -8883 Q

51

Mean Fe and Ba concentrations were not signiﬁcantly different between the two
bullet types, possibly due to the fact that these elements originate from the interior of the
gun barrel or the ammunition primers, which are more similar in elemental composition
between the two ammunition types. Thus, all ﬁve elements of interest (Fe, Cu, Sb, Ba,
and Pb) were successful in differentiating shot from unshot tissue, and three elements
(Cu, Sb, and Pb) were capable of differentiating jacketed and non-jacketed bullet wounds

in fresh tissue samples.

4.5 Sampling Study Tissue Observations

One fresh wound shot with non-jacketed ammunition was selected for the
sampling study to investigate the distribution of GSR around the wound. The gunshot
wound had similar characteristics to all other wounds shot with non-jacketed ammunition
[Figure 4.8]. The gun was ﬁred using a wire hanger as a support, which left the horizontal
void in the GSR pattern that is apparent in Figure 4.8. There was a dense area of soot
deposition around the entrance wound extending approximately 1 cm in all directions
around the wound. A secondary ring of ﬁner GSR deposition was also present, extending
approximately 6 cm from the wound. This secondary ring of GSR, which was likely from
the gap between the gun chamber and barrel, was visually heterogeneous, appearing more
concentrated in the 6 o’clock (6:00) and 9 o’clock (9:00) positions compared to the 12
o’clock (12:00) and 3 o’clock (3:00) positions. There was also some blood staining on the
tissue surface, which was most evident in the 9:00 direction. The edge of the wound was
at slightly different distances from the center of the gunshot wound in each direction:

6.75 cm in the 12:00 direction, 6 cm in the 3:00 direction, 6.5 cm in the 6:00 direction,

52

and 7.5 cm in the 9:00 direction. The outermost edges of the wound did not appear to

have any GSR present.

 

Figure 4.4 - Sampling study wound (a) before tissue sections were removed and (b) after

tissue samples had been removed for analysis. The four directions that the tissue was

sampled are indicated with arrows.

4.6 Quantitation of Elements in Sampling Study Tissue Samples

1

The ﬁve elements of interest (56Fe, 63Cu, 12 Sb, 138Ba, and 208Pb) were

quantitated using ICP-MS, and the limits of quantitation are listed in Table 4.2. The mean
concentrations of the ﬁve elements in the four SRM samples were calculated and
compared to the reported concentrations, as described previously [Table 4.2]. The percent
error between the true and observed Cu concentrations was 20% and hence, Cu was
excluded from consideration during this study. This was reasonable given that Cu levels

were not expected to be useful for GSR determination around a non-j acketed bullet

53

wound. For all other elements, errors were less than 10%, indicating the instrument was

accurately quantitating the elements in all samples.

Table 4.2 — Limits of quantitation (LOQs) and SRM recovery results for elements of

interest in the sampling study.

 

Element LOQ (pg/L) SRM Average % Error

 

56Fe 5 3
63Cu 0.5 20
12‘51) 0.5
1381321 0.25 7
208Pb 0.25

 

Element concentrations were expected to decrease as the sampling distance from

the entrance wound increased, so all element signals were monitored in both the

. . . . 56 . .
undiluted and diluted tissue sample sets. Since Fe was present in relatively low levels

in the fresh tissue study, it was quantitated using the undiluted digest samples so that

. . . . . . . 121
Signals remained Within the linear range of the calibration curve. The elements Sb,

138 208 . . .
Ba, and Pb were more concentrated and were quantitated in the diluted samples to

ensure that sample signals were within the linear calibration curve range and did not
saturate the detector.

The concentration of each element of interest in all shot tissue samples was
compared to the concentration in the procedural blank and control tissue samples [Figure

4.5]. Since there was only one sample taken from each position around the wound, no

54

statistical measures of signiﬁcant differences in concentrations could be made for the shot

tissue samples.

Figure 4.5 - Concentrations of (a) Fe, (b) Sb, (c) Ba, and (d) Pb quantitated in the
sampling study samples (procedural blank 11 = 3, control n = 3, and non-jacketed n = 1).
Error bars are one standard deviation. In the 3:00 direction data, “6 cm/edge” indicates

that the wound edge was at 6 cm.

Fe

A
h?
v

§

Concentration (pg/g)
i-s N
8 8 8

 

 

 

 

 

 

 

as asses 555% 55554 5555a
5% [cm are '8' cot-cg L¢N use?" [a «st-~03]
g 0 12:00 direction 3 6 00direction 9 00direction
‘-
9" 3:00direction
(b) Sb
@4000-
E:
vsooo-
E
.2
E200“
§1000~
E
O
s :3
E .5.
d V
E 12 :00 direction

  

 

3:00 direction

55

Figure 4.5 — Continued

Ba

A

Nntratlon (pg/g) o
, 5 5 i 5""

 

 

 

 

 

 

 

   

 

 
   

 

 

 

=

e

U .
.‘ﬂ
5.
c:
.3. e
9- 3:00 direction

(d) Pb

380000—

3‘;

: —1

.2

g40000«

5200005

5 0 7-‘T-H—ﬁ
55 55555 5555 55555 55555
55% I cave 6:] :N 9% IGNV \o “l IchxoOJ
g 0 i o 6 00direction 9:00 direction
5 12:00direction \O

 

3:00 direction

A summary of the distances from the entrance wound that each element was

detected above the concentration in the procedural blanks and control samples is shown

in Table 4.3.

56

Table 4.3 — Summary of distances that elements of interest in shot tissue samples were

detected at higher levels than procedural blank and control tissue samples.

 

Farthest Distance Element Detected
Element 12:00 3:00 6:00 9:00

 

 

Fe 2 cm edge 2 cm edge
Sb 2 cm 4 cm edge edge
Ba edge 2 cm 2 cm 4 cm
Pb edge edge edge edge

 

The concentrations of Sb, Ba, and Pb were all highest directly adjacent to the
wound, in the area of dense soot deposition, in all directions. The concentrations of these
three elements then decreased in all four directions, with the distance that each element
was last detected above concentrations in procedural blank and control samples varying
for each element. Lead was present in shot tissue samples in concentrations at least one
order of magnitude higher than the procedural blank and control samples all the way out
to the wound edge in all four directions, indicating that Pb spread the farthest from the
entrance wound, possibly because it was the most concentrated element in the GSR.
Antimony was present out to the wound edge in the 6:00 and 9:00 directions, which were
the same directions with the farthest spread of visual GSR, and only a few centimeters in
the 12:00 and 3:00 directions. The Ba and Fe results were more difﬁcult to interpret, as
Ba was deposited farthest out in the 12:00 direction, and Fe was detected furtherrnost in
the 3:00 and 9:00 directions. The Fe results could be due to the presence of blood on the
surface of the tissue, especially in the 9:00 direction, as blood contains Fe. These results
indicate that the optimal sampling position to detect all elements characteristic of GSR by

ICP-MS is directly adjacent to the wound and as far out as 2 cm from the entrance wound

57

in any direction. The high degree of GSR deposition variability from 2 cm outward could
prevent detection of characteristic elements and not allow a wound to be definitively
determined as a gunshot wound.

Overall, these fresh tissue studies allowed the expansion of the list of elements
considered to be characteristic of GSR by adding Fe and Cu to Sb, Ba, and Pb, which are
currently considered characteristic of GSR. The addition of these two elements increases
conﬁdence in GSR determination, as it is unlikely that these ﬁve elements would all be
present on a surface due to other materials. Knowledge of the sample position most likely
to produce conclusive results for the presence of GSR (adjacent to the entrance wound
out to 2 cm) is also essential for coroners, medical examiners, and forensic pathologists.
While these results were promising, the effect of decomposition on element
concentrations was investigated to evaluate the practicality of this method for forensic

applications.

4.7 Decomposition Study Observations and Histology Results

Gunshot wounds from both jacketed and non-jacketed bullets collected on day 0
of the decomposition study were visually very similar in appearance to those from the
fresh tissue studies [Figures 4.6 and 4.7]. This observation was expected as wounds were
inﬂicted using the same weapon and ammunition as in the previous studies. During the
49 day sample collection period, temperatures ranged from a high of 70 °F (day 37) to a

low of 22 °F (day 41). Appreciable rainfall occurred on days 0, 4, 7, 19, 20, 21, 28, 47,

and 48. Full details of the weather conditions are located in the Appendix.3

58

There was minimal insect activity during the ﬁrst week of the study due to cool
temperatures and rain. Bloating was evident and wounds were seeping on all pigs by day
19. By day 24, the heads of all three pigs were beginning to turn black and skin slippage
was apparent. The ﬂesh of all three pigs continued to darken and show more slippage
throughout the rest of the collection period. The carcasses did not reach the
dessication/skeletonization stage of decomposition during this study due to the cold
temperatures experienced. The dense GSR deposition and blackening around the gunshot
wounds persisted throughout the collection period. The stab wounds in the control pig
widened and darkened over time, but stab wounds [Figure 4.8] and gunshot wounds were

visually distinguishable throughout this study.

59

.uﬂoho Mmmv 53> 203628 8398 53> £9303 05 3209 53% 8a 3.32 329$: noﬁﬁm

m we I mo £92?me 36:38:00 4635 :ozmmoaﬁooou o5 waist av 98 .am no 3% so €5.95 6:52 @8832. I 9v onE

 

6O

due—86 #50 53> Eouﬂmcoo mud—0E3 53> £9503 05 323 850% 0.8 3:52 320me 8:83

m a. I we mcmﬁmoaome mcmccommobou Edam 3:68:8ch 2t macaw av 98 .om

 

a

o when so €553 8:3 cowoxomméo Z I 56 23%;

61

.wowoomxo mm .MmO SE “556:8 ave—0E3 b8 25% “on EU 3:58 awe—9mm: 35mg

m as E Co managed: mczocoamotoo Seam :oEmomEoooc 05 macaw 3 new .3 6 want co €553 38$ 3.5.80 I wé Emmi

 

62

Histology samples from all three pigs showed typical patterns of decomposition,
including homogenization of the dermal collagen, separation of the epidermis from the
dermis, loss of cellular detail, and bacterial and fungal overgrowth [Figures 4.6-4.8]. The
sections at the wound edges showed the expected thermal artifact associated with gunshot
wounds, consistent with the fresh tissue studies. The GSR patterns observed in the fresh
tissue persisted but decreased in quantity as moderate decomposition progressed in

wounds inﬂicted using both bullet types.

4.8 Quantitation of Elements in Tissue Samples Throughout Moderate

Decomposition

1

. 56 63 12 '138 208
The ﬁve elements of 1nterest( Fe, Cu, Sb, Ba, and Pb) were

quantiﬁed using ICP-MS, and the limits of quantitation are listed in Table 4.4. The mean
concentrations of the ﬁve elements in the four SRM samples were calculated and
compared to the reported concentrations, as described previously [Table 4.4]. Errors were
less than 10% for all elements, indicating the instrument was accurately quantifying the

elements in all samples.

Table 4.4 — Limits of quantitation (LOQs) and SRM recovery results for elements of

interest in the decomposition study.

 

Element LOQ (pg/L) SRM Average % Error

 

56Fe 5 0.5
63Cu 5 6
me < 0.1 4
138Ba < 0.1 7
208m) 0.25 5

 

63

Since element concentrations were expected to decrease as decomposition

progressed, all element signals were monitored in both the undiluted and diluted tissue

sample sets. Relatively low levels of 56Fe and 63Cu were present throughout the study

and hence, these elements were quantitated using the undiluted digest samples so that

. . . . . . . 121 ,
Signals remained Within the linear range of the calibration curve. The elements Sb,

138 208 . . .
Ba, and Pb were more concentrated and were quantitated 1n the dlluted samples to

ensure that sample signals remained within the linear calibration curve range and did not
saturate the detector.

Concentration ranges for each element of interest throughout moderate
decomposition are listed in Table 4.5. The highest element concentrations did not
necessarily occur on day 0, and the lowest concentrations did not always correspond to

day 49.

Table 4.5 — Element concentration ranges throughout moderate decomposition in each

tissue sample type.

 

 

 

Concentration Raﬂeﬂggg)
Element Control J acketed Non-Jacketed
56Fe 83-235 348-720 242-438
63Cu 12-27 306-835 99-166
me 10-15 197-3677 1722-3923
3:135: ND“ 1721-18159 5057-16599

Pb O-ll 702-11010 42753-131245
* ND indicates “not detected”

 

64

Since element concentrations did not appreciably change as decomposition
progressed, mean element concentrations were calculatedfor six tissue samples from
each bullet type, collected throughout the study at different stages of decomposition on
days 0, 5, 14, 24, 34, and 44 [Figure 4.9]. All elements of interest show higher
concentrations in the shot tissue samples relative to the control tissue and procedural
blank samples, indicating there was no signiﬁcant contamination due to the sample
preparation method. Mean element concentrations for Cu and Ba (99% conﬁdence level)
and for Fe and Pb (95% conﬁdence level) were signiﬁcantly higher in tissue shot with
jacketed ammunition compared to the control tissue. For tissue samples shot with non-
jacketed ammunition, all elements were present at signiﬁcantly higher concentrations in
the shot tissue compared to the control tissue at the 99% conﬁdence level. Thus, all
elements were suitable for differentiating shot from unshot tissue throughout moderate

decomposition.

65

.1. 3 @8868 Ba Cod w 3 56— oocoumcoo £53 8 2235528 “5820 c8253 oocoeomﬁw “5&8me .couagoc 3.2288 25
288%: ES Hobo on... .G n a @8383?— EE @8832 6 u : $888803 35:8 .m u a £85 35:8 .m n a Maw—n E33085

coEmomEooow 38308 323585 380:8 8383 E 63353: 8885 we 35820 no mcowmbaoocoo :32 I oé 8:me

coaoxuan.:ez- venom—2% H Gomez—.838 3.5—SUE Emmy—.5 3.5.5an :5:— Eaeooeum U

 
 
 

 

 

 

 

 

 

 

 

am :0 oh am
.9 a: mu
_|._ as?“ w
-62: r a
é8~ .888 m.
.. _|L rageﬂu For
-825 .625" .5. .683; Fox

66

When mean element concentrations were compared between the two bullet types,
there were signiﬁcant differences in the Cu and Pb concentrations at the 99% conﬁdence
level [Figure 4.9]. Copper was more concentrated in wounds shot with jacketed bullets,
and Pb was more concentrated in tissue shot with non-jacketed bullets, which is likely
due to bullet composition, as previously observed in the fresh tissue study. Mean Sb
concentrations were not signiﬁcantly different between the two bullet types in the
decomposed tissue samples due to the wide concentration range of this element in the
samples shot with jacketed bullets. Some of the variability in element concentrations in
GSR was due to the sample collection procedure. Tissue samples for digestion were cut
directly adjacent to the entrance wound out to approximately 1 cm, where the GSR was
most concentrated. However, as determined in the sampling study, the GSR distribution
was not uniform, which leads to differences in element concentrations in the samples,
larger standard deviation values, and decreased conﬁdence in element concentration
differences between bullet types. Finally, similar to the fresh tissue study, mean Fe and
Ba concentrations were not signiﬁcantly different between the two bullet types and were
therefore not useful for differentiation.

Overall, the element concentration results from the decomposed tissue study were
similar to results obtained from the fresh tissue study, due to minimal decrease in element
concentrations throughout moderate decomposition. All ﬁve elements of interest (Fe, Cu,
Sb, Ba, and Pb) were able to differentiate shot from unshot tissue, and Cu and Pb were

able to discriminate tissue shot with the two different bullet types.

67

4.9 Principal Components Analysis of Decomposition Study Data

Results from ICP-MS quantitation of the ﬁve elements of interest (56Fe, 63Cu,

l218b, 138Ba, and 208Pb) in the decomposed tissue samples were further analyzed to

examine how element concentrations varied with respect to one another. Principal
components analysis (PCA) was used to visualize how different wound samples related to
one another overall by comparing the concentrations of the ﬁve elements simultaneously,
which is more useful than comparing element concentrations individually using Student’s
t-tests as described above. This is also a more objective way to place statistical
conﬁdence on the association and discrimination of samples, which is desirable in all
forensic analyses. The resulting scores plot is shown in Figure 4.10. Principal component
(PC) 1 is the x-axis and accounts for 61% of the variance in the dataset, and PC2 is the y-
axis and accounts for 29% of the variance in the samples. Overall, the scores plot shows
90% of the variance in the dataset, meaning that the sample positions are useful
approximations of the total chemical variance in the samples projected onto only two

dimensions.

68

 

 

 

 

 

 

3 Procedural blank
Non—jacketed
2 Jacketed ‘
Control A ‘
§ 1 A
Ox A
8 0 A
N _ Jacketed
a _1 control A A
Non-jacketed t A
-2 control
-3
A
-4 -3 -2 -l 0 1 2 3 4

PC] (61%)

Figure 4.10 — PCA scores plot displaying clustering of chemically similar samples.

 

 

 

 

 

0.6 - 52081)!)

0.4 - AIZISb
' ‘l
g 0.2 t
a r
N 0
U A
a, 138Ba

-0.2

'094 56

+ FCA A63CII
0 0.1 0.2 0.3 0.4 0.5

PCI (61%)

Figure 4.11 — PCA loadings plot showing the elements responsible for the sample

positions displayed in the corresponding scores plot.

69

 

In the scores plot [Figure 4.10], the procedural blank and control samples group
closely, loading negatively on PCI and near zero on PC2. The ﬁve procedural blank
samples are overlaid, indicating nearly identical chemical composition as expected for
blanks. The control samples taken from the pigs that .were subsequently shot with the
jacketed and non-jacketed ammunition (labeled “jacketed control” and “non-jacketed
control”, respectively) cluster with all of the samples collected from the control (stabbed)
pig throughout decomposition, indicating that differences in the shot tissue samples were
not simply due to the fact that they were from different pigs. A small amount of spread is
observed within the control sample group which is due to natural variability in the tissue
samples as they were from different areas of the pig.

All shot tissue samples are positioned more positively on PC] in the scores plot
compared to the procedural blank and control tissue samples. The corresponding PCA
loadings plot [Figure 4.11] shows all ﬁve elements of interest loading positively on PC 1 ,
which corresponds to the positions of the shot tissue samples in the scores plot. This
indicates that these ﬁve elements are useful for differentiating shot from unshot tissue,
supporting the results from the decomposition quantitation study.

The shot tissue samples are separated according to bullet type by PC2 in the
scores plot [Figure 4.10], with the samples shot with non-jacketed ammunition positioned
positively and the samples shot with jacketed ammunition positioned negatively on PC2.
Again, this positioning can be explained with reference to the loadings plot [Figure 4.11].
Lead and Sb load positively on PC2, indicating these two elements correspond to the
positions of the samples shot with non-jacketed ammunition on the scores plot. The

average Pb concentration in wounds shot with non-jacketed bullets is signiﬁcantly higher

70

compared to jacketed bullet wounds, whereas Sb concentrations, though higher in non-
jacketed bullet wounds, were not statistically different between the two bullet types in the
quantitation study. This explains why Sb is positioned less positively than Pb in the
loadings plot. However, the PCA results show that concentrations of both Pb and Sb are
systematically higher in wounds shot with non-jacketed bullets compared to wounds shot
with jacketed bullets, allowing differentiation of the two bullet types in the scores plot.

All tissue samples shot with jacketed ammunition are positioned negatively on
PC2 in the scores plot. Iron and Cu are positioned negatively on PC2 in the loadings plot,
indicating that these two elements are associated with the jacketed bullet wound positions
on the scores plot. The average Cu concentration was signiﬁcantly higher in tissue shot
with jacketed bullets compared to tissue shot with non-jacketed bullets in the quantitation
study, and, though not statistically signiﬁcant, the average Fe concentration was more
abundant in jacketed bullet wounds as well. Thus, the negative positioning on PC2 of the
jacketed bullet wounds in the scores plot is due to the higher levels of Cu and Fe in these
wounds.

The ﬁnal element of interest, Ba, is positioned close to zero on PC2, indicating
that it is not more closely associated to one group of shot tissue samples than the other.
The average Ba concentrations in the jacketed and non-jacketed bullet wounds were
nearly identical, indicating that Ba is only useful for differentiating shot from unshot
tissue and not for differentiating bullet type.

Therefore, all shot tissue samples are distinct from the procedural blank and
control tissue samples, and the tissue samples shot with different ammunition types are

clearly separated as well. This indicates considerable chemical composition differences

71

between the sample groups, which was expected in view of the results from the
decomposition quantitation study. There is more spread in the shot tissue samples from
both bullet types compared to the tight clustering of the procedural blank and control
samples. This is due to variation in element concentration as a result of the effects of
moderate decomposition and the random tissue sampling around each wound, as
discussed previously.

Overall, PCA results for the ﬁve elements of interest in the decomposition study
tissue samples were in agreement with observations made from the quantitation study
results for Pb, Cu, and Ba. Additionally, Sb and Fe were shown to be useful for
differentiating the two bullet types using PCA, despite the lack of statistically signiﬁcant
concentration differences observed during the quantitation study. This is due to the high
degree of correlation between the Sb and Pb concentrations and between the Fe and Cu
‘ concentrations, which was revealed through PCA. This means that examining all ﬁve
element concentrations simultaneously allows for more useful tissue sample association
and discrimination. Although this is a very small study, it does demonstrate the potential
utility of PCA in casework. A questioned tissue sample could be statistically analyzed
along with all of these known tissue samples, and the position of the questioned sample
on the scores plot could be used to determine whether the questioned wound was a
gunshot wound, and if so, whether the questioned gunshot wound was inﬂicted with
either a jacketed or a non-jacketed bullet. However, this is only a proof-of-concept study
and considerable further research must be done before this methodology could be

implemented in casework.

72

4.10 References

1. Stone IC, Di Maio VJ M, Petty CS. Gunshot wounds — visual and analytical
procedures. J Forensic Sci 1978;23:361-7.

2. Koons RD. Analysis of gunshot primer residue collection swabs by inductively
coupled plasma-mass spectrometry. J Forensic Sci 1998;43:748-54.

3. National Oceanic and Atmospheric Administration's National Weather Service.
http://www.weather. gov.

73

CHAPTER 5: CONCLUSIONS AND FUTURE DIRECTIONS
5.1 Conclusions

Chemical identiﬁcation of gunshot residue (GSR) in tissue samples using
inductively coupled plasma mass spectrometry (ICP-MS) is a promising option to
supplement current gross and histological examination procedures available to coroners,
medical examiners, and forensic pathologists. Three studies using ICP-MS for the
analysis of fresh gunshot wounds shot with two different ammunition types were
completed. The results facilitated selection of an expanded suite of elements capable of
differentiating shot from unshot tissue as well as between tissue samples shot with
jacketed and non-jacketed bullets. The element distribution in GSR around a non-
jacketed bullet wound was also determined in order to develop optimal sampling
procedures for this technique. The results from the fresh tissue studies were then applied
to decomposed wounds to assess the effects of moderate decomposition on the ability to
detect GSR and differentiate bullet types. The ICP-MS results from both fresh and
decomposed wounds were compared to corresponding hematoxylin and eosin (H & E)
stained histology results to determine if ICP—MS was comparable with the current GSR
detection method used by death investigators.

Two fresh wounds, one shot with jacketed ammunition and one shot with non-
jacketed ammunition, were analyzed using ICP-MS to detect all elements present. Using
several selection criteria, the original suite of elements considered potentially useful for

differentiating shot from unshot tissue and differentiating the two bullet types consisted

l

24 31 39 56 63 66 I2 138 208 .
of Mg, P, K, Fe, Cu, Zn, Sb, Ba, and Pb. Due to instrument

I

. . . . . 56 63 12 138
11m1tatlons, the list of elements of Interest was reduced to Fe, Cu, Sb, Ba, and

74

206Pb, which showed acceptably low limits of detection, linearity over the calibration

range, and accurate concentration determination based on standard reference material
recoveries. All ﬁve elements of interest were successful in differentiating shot from
unshot tissue in the fresh state, and three elements (Cu, Sb, and Pb) were capable of
differentiating jacketed and non-jacketed bullet wounds in fresh tissue samples.
Antimony and Pb were both more concentrated in tissue shot with non-jacketed bullets
(signiﬁcant at the 99% conﬁdence level), while Cu was more concentrated in tissue shot
with jacketed ammunition (signiﬁcant at the 95% conﬁdence level). The elements that
successfully differentiated the bullet types most likely originate from the bullets
themselves, since the jacketed bullets have Cu jackets and the non-jacketed bullets are
composed of Pb and possibly Sb. Mean Fe and Ba concentrations were not signiﬁcantly
different between the two bullet types, possibly due to these elements originating from
the interior of the gun barrel or the ammunition primers, which are more similar in
elemental composition between the two ammunition types.

Corresponding results from H & E histology analysis showed particles consistent
with GSR in samples from both bullet types. Due to gunpowder composition differences
in the two different ammunition types used in this study, the wounds could be
distinguished according to bullet type as well. Thus for the fresh tissue studies, the
differentiation of shot from unshot tissue as well as tissue shot with two different bullet
types could be accomplished using gross examination, histology, and ICP-MS analyses.

The ICP-MS quantitation results from the fresh non-jacketed wound sampling
study indicated that the optimal sampling position to detect the ﬁve elements considered

characteristic of GSR is directly adjacent to the wound and as far out as 2 cm from the

75

entrance wound in any direction. The high degree of variability in element composition
from 2 cm outward prevented detection of some of the characteristic elements and thus
did not allow a wound to be deﬁnitively identiﬁed as a gunshot wound.

These fresh tissue studies allowed the expansion of the suite of elements
considered to be characteristic of GSR by adding Fe and Cu to Sb, Ba, and Pb, which are
currently considered characteristic of GSR. The addition of these two elements increases
conﬁdence in GSR determination, as it is unlikely that these ﬁve elements would all be
present on a surface due to other materials. Knowledge of the sample position most likely
to produce conclusive results for the presence of GSR (adjacent to the entrance wound
out to 2 cm) is also vital for coroners, medical examiners, and forensic pathologists.

The optimal sampling procedure was then used to collect tissue samples
throughout the decomposition study. The samples were then analyzed by ICP-MS for the
suite of characteristic elements present in GSR, and comparisons were made between
control tissue samples and samples shot with each bullet type, as well as between the two
different bullet types throughout moderate decomposition. Mean element concentrations
for Cu and Ba (99% conﬁdence level) and for Fe and Pb (95% conﬁdence level) were
signiﬁcantly higher in tissue shot with jacketed ammunition compared to the control
tissue. For tissue samples shot with non-j acketed ammunition, all elements were present
at signiﬁcantly higher concentrations in the shot tissue compared to the control tissue at
the 99% conﬁdence level. Thus, all elements were suitable for differentiating shot from
unshot tissue throughout moderate decomposition. When mean element concentrations
were compared between the two bullet types, there were signiﬁcant differences in the Cu

and Pb concentrations at the 99% conﬁdence level. Copper was more concentrated in

76

wounds shot with jacketed bullets, and Pb was more concentrated in tissue shot with non-
jacketed bullets, which is likely due to bullet composition, as observed in the fresh tissue
study.

Corresponding H & E histology results from the decomposition study samples
showed particles consistent with GSR throughout moderate decomposition. The different
particle patterns from the two different ammunition types persisted but decreased in
quantity over the course of the study, and the tissue samples became harder to section and
analyze using histology as moderate decomposition progressed.

The ICP-MS results from the decomposed tissue samples were further analyzed
using principal components analysis (PCA), which is an objective statistical procedure
that was used to examine how all ﬁve elements of interest varied with respect to one
another in the samples. The results conﬁrmed that all elements were capable of
differentiating shot from unshot tissue and that Ba was not able to differentiate the two
bullet types. The concentrations of Pb and Sb were determined to be correlated and were
higher in non-jacketed bullet wounds. Likewise, the Cu and Fe concentrations were
determined to be correlated and were more concentrated in jacketed bullet wounds. These
results indicated that these four elements were useful for differentiating tissue shot with
the two different bullet types using PCA. Although this was a very small study, it did
demonstrate the potential utility of PCA in casework to determine whether a questioned
wound was a gunshot wound, and if so, whether the questioned gunshot wound was
inﬂicted with either a jacketed or a non-jacketed bullet. However, this was only a proof-
of-concept study, and considerable further research must be done before this

methodology could be implemented in casework.

77

Overall, both histology procedures and ICP-MS were used to identify the
presence of GSR in close-range gunshot wounds from two bullet types in both the fresh
state and throughout moderate decomposition. However, GSR determination using
histology became more difficult as decomposition progressed. Adding the elements Fe
and Cu to the elements already considered characteristic of GSR (Sb, Ba, and Pb)
increases conﬁdence in gunshot wound determination. Using these element proﬁles,
wounds shot with jacketed bullets were differentiated from wounds shot with non-
jacketed bullets throughout moderate decomposition, providing a new level of
information to investigators. While this research demonstrated success in differentiating
bullet types from close-range gunshot wounds, more challenging tissue samples must be
investigated in the future to determine the limitations of these methods.

The results of this research could beneﬁt law enforcement agencies as well as
coroners, medical examiners, and forensic pathologists. Law enforcement agencies may
use the method to determine the type of bullet used, which can link a weapon and/or
suspect to a crime scene. Coroners, medical examiners, and forensic pathologists could
use the method to assist in gunshot wound identiﬁcation and cause of death

determination, even in corpses in a moderate state of decomposition.

5.2 Future Directions

Now that the capability to differentiate close-range jacketed and non-jacketed
bullet wounds using ICP-MS has been established, there are numerous possibilities for
future studies. As all studies using ICP-MS for GSR detection have been conducted using

close-range wounds, it would be useful to examine the capabilities and limitations of

78

ICP-MS for GSR determination in wounds shot from farther distances. Tissue samples
shot from known distances and analyzed using ICP-MS to quantitate the elements of
interest identiﬁed in this study may provide a relationship between the concentration of
certain elements and ﬁring distance. Depending on the results of the initial study with one
bullet type, the differentiation of multiple bullet types at farther distances could also be
investigated. If that were successful, then the potential to determine ﬁring distance and
bullet type throughout the course of decomposition could be tested in a similar manner.
Any information from successful outcomes of these experiments would be considered
very useful to law enforcement ofﬁcers and death investigators.

The current studies investigated only one gun and the two most common bullet
types, but there are many others types of ﬁrearms and ammunition readily available.
Carrying out studies comparing the GSR detected by ICP-MS from different common
ﬁrearms, such as semi-automatics and even riﬂes and shotguns, as well as their
associated ammunition types would add to the knowledge base in this area of forensics.
Also, only one ammunition manufacturer was investigated in the present study. It would
be interesting to examine how elemental compositions vary for the same ammunition
type obtained from different manufacturers. Ideally, elemental proﬁles of the same type
of ammunition from different manufacturers would be similar enough to differentiate
between bullet types from several different manufacturers. In addition, there was a
difference in the type and amount of gunpowder in the two different ammunition types
from the same manufacturer used in these studies. It would be interesting to investigate
differentiation of ammunition with the same gunpowder and different bullet types both

using histology and ICP-MS for GSR detection.

79

All tissue samples that have been analyzed for the presence of GSR using ICP-
MS have been relatively free of environmental debris and have decomposed above
ground. The case that provided the idea for this research involved a body that was buried
for a year before discovery and autopsy. Studies simulating the conditions of this case
need to be conducted in order to address the challenges of gunshot wound determination
routinely seen in casework. Pigs could be shot and then buried throughout the
decomposition process, and the wounds would be analyzed by ICP-MS to determine if
the soil interfered with GSR element detection. Burying the pigs at different depths and
in different soil types in different seasons of the year would also be essential, as the
decomposition process is affected by all of these parameters. The ability to differentiate
ammunition types and possibly also gain information about the range of ﬁre from buried
tissue samples could also be explored if preliminary studies were successful.

Finally, as hematoxylin and eosin (H & E) staining of histology samples is not
designed to enhance visualization of GSR particles, staining tissue sections with different
reagents that are designed to indicate the presence of elements contained in GSR would
be useful. Stains are available that react with the currently accepted GSR elements (Ba
and Pb), and evaluating their ability to determine the presence of GSR in wounds in both
the fresh state and throughout decomposition in comparison to ICP-MS would be useful
for coroners, medical examiners, and forensic pathologists. If successful, the stains could
be used on more challenging tissue samples, such as those shot from farther distances,
shot with different ammunition types, and buried in the ground. Samples could also be
analyzed by ICP-MS as a comparison to determine the most useful technique for GSR

determination in decomposed tissue samples.

80

APPENDIX

Weather data from the decomposition study sample collection period.

 

Temperature (°F)

 

 

Day Date Maximum Minimum Average Rain (in) Samples Taken
0 10/2/2009 57 48 53 0.39 X
1 10/3/2009 57 47 52 0.08
2 10/4/2009 58 48 53 0.01 X
3 10/5/2009 59 ’ 44 52 -

4 10/6/2009 61 42 52 0.20

5 10/7/2009 57 42 50 trace X
6 10/8/2009 60 41 51 0.06

7 10/9/2009 51 44 48 0.56

8 10/10/2009 53 32 43 -

9 10/1 1/2009 47 27 37 - X
10 10/12/2009 49 34 42 0.01

11 10/13/2009 49 32 41 0.01

12 10/ 14/2009 46 3 1 39 trace

13 10/15/2009 45 3 7 41 trace

14 10/16/2009 45 33 39 - X
15 10/17/2009 49 27 3 8 -

16 10/ 18/2009 50 26 38 -

17 10/19/2009 62 3 8 50 -

18 10/20/2009 61 47 54 -

19 10/21/2009 69 52 61 0.17 X
20 10/22/2009 58 41 50 0.32

21 10/23/2009 59 41 50 0.99

22 10/24/2009 50 39 45 0.03

23 10/25/2009 58 37 48 trace

24 10/26/2009 64 48 56 0.01 X
25 10/27/2009 56 47 52 0.08

26 10/28/2009 54 41 48 0.02

27 10/29/2009 59 39 49 trace

28 10/30/2009 69 51 60 1.1 l

29 10/31/2009 54 40 47 trace X
30 11/1/2009 49 35 42 -

31 11/2/2009 52 38 45 0.08

32 1 1/3/2009 46 31 39 -

33 1 1/4/2009 48 29 39 trace

34 1 1/5/2009 49 28 39 trace X

81

35
36
37
38
39
40
41
42
43
44
45
46
47
48
49

1 1/6/2009
11/7/2009
11/8/2009
1 1/9/2009
1 1/10/2009
1 l/ l 1/2009
1 1/12/2009
1 1/13/2009
1 1/14/2009
1 1/15/2009
1 1/16/2009
1 1/17/2009
1 1/18/2009
11/19/2009
1 1/20/2009

49
67
70
67
54
53
52
59
64
59
48
50
56
48
48

25
44
35
49
36
28
22
29
43
4o
38
33
35
41
42

37
56
53
58
45
41
37
44
54
50
43
42
46
45
45

 

82

 

”"t’lll‘llllil [lit [1111111141111'“

637