WNWWWWlWHIHHIHWWWWWI

      

THS_

,1, LIBRARIES
‘-. MICHIGAN STATE UNIVERSITY
EAST LANSING, MICH 48824-1048

as: WW3

This is to certify that the
thesis entitled

THE USE OF LASER ABLATION INDUCTIVELY
COUPLED PLASMA MASS SPECTROMETRY
(LA-ICP-MS) FOR THE DISCRIMINATION OF GLASS
FRAGMENTS IN FORENSIC CASEWORK

presented by
David William Szymanski

has been accepted towards fulfillment
of the requirements for the

MS. degree in Forensic Chemistry

 

 

 

 

 

Major Professor’s Signature

/I////04f

Date

MSU is an Affirmative Action/Equal Opportunity Institution

-.—.-.-.---.—.-.-‘— - - -.-.-.-.--.—.-—.-.-.-.-.

o .0...--;- --—--.-.

'1")

 

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

 

rig/till; 632007.

 

li"R 1 8 2007

 

 

 

 

 

 

 

 

 

 

 

 

6/01 c:/C|RC/DateDue.p65-p415

THE USE OF LASER ABLATION INDUCTIVELY COUPLED
PLASMA MASS SPECTROMETRY (LA-ICP-MS) FOR THE DISCRIMINATION OF
GLASS FRAGMENTS IN FORENSIC CASEWORK
By

David William Szymanski

A THESIS
Submitted to
Michigan State University
In partial fulfillment of the requirements
For the degree of
Master Of Science in Forensic Chemistry

School Of Criminal Justice

2004

ABSTRACT
THE USE OF LASER ABLATION INDUCTIVELY COUPLED
PLASMA MASS SPECTROMETRY (LA-ICP-MS) FOR THE DISCRIMINATION OF
GLASS FRAGMENTS IN FORENSIC CASEWORK
By

David William Szymanski

Inductively coupled plasma mass Spectrometry (lCP-MS) is becoming a widely
used method for elemental analyses of glass fragments, with laser ablation (LA) of solid
samples being the most efficient introduction system for forensic applications (e.g. Trejos
et al., 2003). Trace element abundances in glasses are widely variable, even in glasses of
the same refractive index. This variability allows for a high degree of discrimination
when comparing ratios of trace element abundances. Laser ablation requires little or no
sample preparation, which reduces Opportunity for contamination, and is virtually non-
destructive. In order to use the technique for routine forensic analyses of glass fragments
in casework, it was necessary to test the precision of the technique and the homogeneity
of samples to avoid erroneous interpretations of the results. A protocol for the analysis of
glass fragments by LA-ICP-MS was developed using a set of ten “unknown” automobile
float glass fragments and a National Institute of Standards and Technology glass. A
graphical technique was employed for comparing element ratios in samples, based on the
work of Watling et al. (1997). After confirming the homogeneity of glass samples and
ensuring reproducibility of individual analyses, the technique was applied to casework for
the Michigan State Police. When samples cannot be distinguished by physical
parameters such as refractive index, they are analyzed by LA-ICP-MS. Three cases are

presented as examples of the discriminatory and associative power of the method.

For Jack Brownstein.

iii

ACKNOWLEDGEMENTS

This research was accomplished with the support of many colleagues, friends
and members of my family. Jay Siegel, Lina Patino, Christopher Bommarito, and
Thomas Vogel served as advisors throughout the project, offering numerous discussions,
critiques, and ideas to help me transfer the research into practical application.
Christopher Smith is thanked for his role as a reader from the School of Criminal Justice.
Special thanks are Offered to Lina Patino for her patient, comprehensive teaching of ICP-
MS techniques and data interpretation. Likewise, the friendship and exceptional
instruction of Christopher Bommarito fueled my interest in glass and encouraged my
pursuit of excellence in criminalistics. I am grateful that the Department Of Geological
Sciences and School of Criminal Justice at Michigan State University allowed me to
pursue dual degrees and supported my interdisciplinary approach to learning and
teaching. The result here is a cooperative program that involves both departments and
the Michigan State Police, working together to solve real-world problems through
science. Accordingly, I must thank the Michigan State Police and detectives throughout
Michigan for their help in completing this research.

The loves of my life, Anne and Hannah, serve as a constant reminder of why all
of this matters. Likewise, my parents, brother, Sisters, and extended family are a source
of endless support and joy in my life. I would also like to thank numerous friends from

Michigan and Minnesota for encouraging me.

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. vi
LIST OF FIGURES ................................................................................. vii
INTRODUCTION .................................................................................... 1
EXPERIMENTAL SETUP ......................................................................... 5
Instrumentation and Materials ...................................................... 5
Experimental Design ................................................................. 5
RESULTS AND DISCUSSION ................................................................... 8
Precision and Discrimination ...................................................... 18
Precision and Sample Homogeneity .............................................. 29
APPLICATION OF THE METHOD: CASEWORK EXAMPLES ......................... 31
Case #1: Lansing, MI Homicide ................................................... 34
Case #2: Saginaw, MI Homicide .................................................. 34
Case #3: Dundee, MI Breaking and Entering .................................... 36
CONCLUSIONS AND FUTURE WORK ...................................................... 36
REFERENCES ....................................................................................... 41

LIST OF TABLES

Table 1. List Of ten samples treated as unknowns in this study and sample codes used in

text ...................................................................................................... 6
Table 2. Integrated peak heights for all analyses. (n/a = not acquired) ........................ 9
Table 3. Sr—normalized peak heights for analyses .............................................. 14

Table 4. Precision RSD values for non-float and float side analyses for all samples in
Phase I and II. Each RSD value includes five replicate analyses on the non-float side
and five analyses on the float Side of each fragment. RSD values are given for peak
heights normalized to Sr ........................................................................... 24

Table 5. Precision RSD values for analyses in two cross-sectional traverses in Phases I
and II. RSD values are given for peak heights normalized to Sr ............................. 33

vi

LIST OF FIGURES

Figure 1. Plot of all analyses for the set of six unknown fragments and NIST 612 from
Phase 1. Symbols represent five replicate analyses for each sample on the non-float side
of fragments. Note that only two samples, 89GrandAm and 93Cav, are not clearly
differentiated using these three elements (Zr-La~Ba) .......................................... 19

Figure 2. Plot of all analyses for the set of six unknown fragments and NIST 612 from
Phase 1, replacing La in Figure 1 with Sr at the top apex. Symbols are the same as Figure
1. Notice that all samples, including 89GrandAm and 93Cav from Figure 1 are clearly
separated ............................................................................................. 21

Figure 3. Plot of all analyses for the set of Six unknown fragments and NIST 612 from
Phase II. Symbols represent five replicate analyses for each sample on the float side of
fragments ............................................................................................. 22

Figure 4. Plot of all analyses for the set of six unknown fragments and NIST 612 from
both Phase I and Phase 11. Closed symbols represent five replicate analyses for each
sample on the non-float Side of fragments (Phase I) while open symbols represent five
replicate analyses on the float side of fragments (Phase II) .................................... 23

Figure 5. Plot of all analyses for the set of additional four unknown fragments and NIST
612 from Phase III. Symbols represent ten replicate analyses for each sample, five
replicates on the non-float side and five replicates on the float side of fragments. Note
that the fragments that are not clearly discriminated in the plot (Sr-Y-Zr) can be
differentiated using an additional three elements (Ce-Ba-La, inset) ......................... 26

Figure 6. Plot Of all analyses for the set of additional four unknown fragments and NIST
612 from Phase 111 Showing the effect of multiplying one component (Y) by a factor of
ten. Symbols are the same as Figure 5 ........................................................... 27

Figure 7. Plot of all analyses for the entire set of ten unknown fragments and NIST 612
from Phase I, II, and III. Symbols represent ten replicate analyses for each sample, five
replicates on the non-float side and five replicates on the the float side of fragments. See
text for discussion of samples 90Accord and 9OSunbird, which cannot be graphically
discriminated using any combination of elements ............................................. 28

Figure 8. Schematic diagrams of cross-sectional traverses of sample 86LabGTS from
Phase II. Analyses were spaced 300 um apart for a total of 13 ablation points. A. Plot of
distance versus peak height (counts) for Ce and Y along the cross-section. B. Plot of
distance versus peaks heights for Ce and Y normalized to Sr. Note the relatively flat
profiles using normalized values .................................................................. 3O

vii

LIST OF FIGURES, CONTINUED

Figure 9. Schematic diagrams of cross-sectional traverses of sample 94Metro from Phase
III. Analyses were spaced 300 um apart for a total of seven ablation points. A. Plot of
distance versus peak height (counts) for La and Ba along the cross-section. B. Plot of
distance versus peaks heights for La and Ba normalized to Sr. Again, note the relative
flat profiles using normalized values ............................................................. 32

Figure 10. Plot showing one of the diagrams used to conclude that the questioned and
known fragments in the homicide case could have originated from the same source. Each
sample was analyzed three times. Combinations of other elements produced the same
results; none of the fragments could be discriminated .......................................... 35

Figure 11. Plot Showing one of the diagrams used to conclude that the questioned
fragment from the front porch and the known fragment from the south window could
have originated from the same source. Note that the questioned fragment from the
hammer handle is easily excluded as coming from the same source as the other two
fragments. Again, each sample was analyzed three times .................................... 37

Figure 12. Plot showing one of the diagrams used to conclude that the questioned
fragment from the suspect’s shoe and the known fragment from the convenience store
window did not originate from the same source. As in previous cases, each sample was
analyzed three times ................................................................................. 38

viii

INTRODUCTION

Glass is used universally in packaging, architecture, and motor vehicles. Float I
glass, named for the process by which molten glass is cooled and thinned as it is floated
atop a layer of liquid tin, defines an important and frequently encountered type of glass
evidence. Commonly used in architectural and automotive windows, float glass is easily
dispersed and transferred during the commission of crimes. With the increasing physical
and chemical homogeneity of float glass, forensic scientists must use innovative
techniques to strengthen associations and add discriminating power to analytical schemes
that rely upon traditional optical techniques, such as the determination of refractive index.

In the absence of a physical match, forensic scientists typically classify glass
fragments according to their physical properties of color, thickness (when parallel sides
are present), and refractive index. When exclusions cannot be made based on these
properties, elemental analyses can provide crucial additional information for
classification and discrimination of fragments (Hickman, 1987). Major element variation
in glass is often associated with changes in refractive index (Parouchais et al., 1996).
Trace elements, however, are more variable, even among glasses with analytically
indistinguishable refractive indexes (e. g. Duckworth et al., 2002). For this reason, it has
been clearly demonstrated in pairwise comparison studies that while refractive index
measurements alone can discriminate roughly 40-50% of different glasses in large sample
sets (e.g. 40-70 sample samples, or roughly 1000-2500 pairs), the addition of trace
element data decreases the number of indistinguishable pairs to O-5%, depending on the

rigor of the statistical comparison (Trejos et al., 2003; Duckworth et al., 2002).

Inductively coupled plasma mass spectrometry (ICP-MS), with multielement
capabilities, low detection limits, and a dynamic range that normally spans eight orders of
magnitude, is well suited for trace element analyses of glass fragments. The use of ICP-
MS for elemental analysis of glass is well documented by solution nebulization of acid-
digested fragments (e.g. Parouchais et al., 1996; Suzuki et al., 2000; Duckworth et al.,
2000; Duckworth et al., 2002; Trejos et al., 2003; Montero et al., 2003) and by laser
ablation (e.g. Watling et al., 1997; Watling, 1999; Trejos et al., 2003; Trejos and
Almirall, 2004). In the case of solution nebulization, fragments are first dissolved in an
acid cocktail consisting of two or more of the following acids in varying proportions:
hydrofluoric (HF), nitric (HNO3), hydrochloric (HCl) and/or perchloric (HClO4). The
solutions are then aspirated through a small annulus and carried to the plasma in liquid
aerosol form.

During laser ablation of glass, radiation from the laser couples with the sample
and produces intense heat at its surface. The sample is partially melted and vaporized,
with some of the material being quenched into microscopic glass beads. The solid
particles are subsequently swept away by an inert carrier gas to be atomized, then ionized
by the plasma and sent to a mass spectrometer. Elemental fractionation, whereby the
composition of the material analyzed is not representative of the material being ablated,
can occur due to a combination of factors, including but not limited to: depth and
morphology of the crater produced by ablation, laser beam properties, transport of
various particle sizes, and sample matrix effects (Russo et al., 2002). However, in a

recent study using glass standards and float glass samples, Trejos and Almirall (2004)

found that most elements show low levels of fractionation during ablation and
fractionation does not appreciably affect elemental comparisons in forensic casework.

As reviewed by Durrant (1999) and Russo et al. (2002), laser ablation has been
used in a wide variety of geological, biological, materials science, and forensic
applications. Depending on the sample and ablation parameters, as little as microgram
(1045 g), nanogram (10'9 g), or even femtogram (10'15 g) quantities may suffice for an
analysis (Russo et al., 2002). Glass fiagments submitted to the forensic scientist are often
microscopic; the small volume of material required for LA-ICP-MS makes the method
almost non-destructive even for the smallest fragments. The lack of sample preparation
minimizes opportunity for contamination of samples and significantly reduces analysis
time. The efficiency of LA-ICP-MS makes it ideally suited for forensic elemental
analysis of glass fragments.

Although true quantification of elemental abundances in glass samples is not
possible without the use of other techniques to determine the concentration of at least one
element to be used as an internal standard, the use of elemental ratios has been
demonstrated to be a precise and effective discriminating tool for comparisons, even
when using solution nebulization. Parouchais et al. (1996) note that in cases where acid-
digested fragments are too small to be accurately massed for quantification, elemental
ratios are well suited for sample comparison. Watling et al. (1997) suggested the use of
ternary discrimination diagrams for comparing trace element ratios in samples. Such
diagrams (see below, Figure 1) allow for fast graphical discrimination of samples based

on a number of different combinations of elements.

This research took a two-fold approach to overcome several roadblocks to LA-
ICP-MS analysis of forensic glass samples. The first goal was to develop a protocol for
the ablation of glass fragments in the ICP-MS laboratory at Michigan State University
and determine the precision of the technique for the standard reference glass from the
National Institute of Standards and Technology (NIST 612) and a set of automobile float
glass samples collected by the Michigan State Police. In order to use the technique for
casework, it was necessary to demonstrate that the analyses are precise, regardless of
which part of a fragment is sampled. As float glass is produced, tin (Sn) is imparted to a
very thin (um-order) portion of the “float” side of the glass, the side in contact with the
liquid tin. A clear concern, therefore, was whether trace elements were heterogeneously
distributed throughout a fragment, which would lead to imprecise analyses and possibly
erroneous exclusions in comparisons.

The second goal was to develop a method for efficiently evaluating the data,
discriminating between samples using trace element ratios in actual casework. As
described above, much work is being done to develop the method in terms of optimizing
discrimination power of the technique (e.g. Trejos et al., 2003) and in the statistical
comparison of sample sets (e.g. Aeschliman et al, 2004; Koons and Buscaglia, 2002;
Koons and Buscaglia, 1999; Curran et al., 1997a, 1997b). The purpose of this study was
to employ the laser ablation technique for discrimination in actual casework. The method
was designed to compliment comparison techniques currently employed by the Michigan
State Police, including refractive index determination by the Emmons double-variation
method and scanning electron microscopy energy-dispersive spectroscopy (SEM-EDS)

for major element composition.

EXPERIMENTAL SETUP
Instrumentation and Materials

A CETAC LSX-200® Plus Q switched Nd:YAG laser (266 nm) attached to a
Micromass Platform® ICP-MS, a quadrupole mass spectrometer, was used in all
experiments. The Platform features and in-line hexapole collision cell, which reduces
elemental interferences with the argon carrier gas (e.g. 40Arlf’O+ on 56Fe). The LSX laser
focus ranges from a 10-250 um beam diameter and has precision translation stage
movement (x, y and 2) within 0.25 um. NIST 612, a glass standard with ~50 ppm
concentrations 'of the trace elements that is used for tuning the instrument, was
incorporated as a known standard for precision in all experiments. Ten sets of
automobile float glass fragments were provided by the Michigan State Police Forensic
Science Division. Fragments were collected by the Michigan State Police and represent
ten models from seven manufacturers (Table 1). All unknown samples had parallel sides;
the side of each fragment in contact with the liquid tin during production, the “float side,”

was identified by UV fluorescence and labeled with permanent maker.

Experimental Design

In the first phase of the project, a protocol for ablation of NIST 612 and one
“unknown” glass fragment was developed. Based on previous glass discrimination
studies by Parouchais et al. (1996), Watling et a1 (1997), Becker (2000), and Duckworth
et al. (2002), a menu of eight isotopes was selected for data acquisition by LA-ICP-MS.

A scan of these eight isotopes (85Rb, 88Sr, 89Y, 90Zr, 98Mo, 138Ba, 139La, and 14°Ce) was

.38“ E vow: £68 638% 28 35% £5 E £5653: me 3:35 338% :8 mo 35 A 93:.

 

 

9323 can: 80 32
853% 85:3 588»... $2
assuage» Samara 23336 32
3238 39:5 838.. 82
genomes «seem $6526 82
>88 5:98 33.520 82
$338 2.0 8E3 .230 £2
:80? 3880 Eon 32
€553 5. Bee 8:8,. $2
2828 282 38: 82
230 9.955 .232 Shauna—.52 30>

conducted for 60 s in each trial of protocol development, during which the transient
signal was acquired by the detector. Ablation parameters were optimized in depth-profile
ablation mode, using both NIST 612 and 1993 Chevrolet Cavalier (93Cav, see Table 1).
Variables optimized included: spot size, rate of penetration into sample by raising the
sample stage (z—rate), and duration of ablation. In each trial, a 1 s pre-ablation burst of
the laser at the sample surface was conducted (100% energy output) to eliminate any
possible contamination. In each trial, the pre-ablation spot size was 100 um larger in
diameter than the ablation spot size. The 60 s scan was initiated before ablation began
and continued after ablation was complete, making it possible to easily identify any well-
defined transient peaks above background, with a 0.1 s dwell time for each isotope. The
technique was optimized for the smallest spot size (best for forensic applications) that
produced adequate signal above background. For both NIST 612 and the unknown
sample, this condition was met using a z-rate of l urn/s for 30 s and an ablation spot size
of 100 pm. (The pre-ablation laser burst for contamination prevention was consequently
set with a Spot size of 200 um.)

Using this protocol, NIST 612 and a set of the samples from the Michigan State
Police (n=10) were ablated to test 1.) the precision of the technique and 2.) sample
homogeneity. In this first phase, a subset of samples (n=6) was analyzed, including:
93Cav, 89GrandAm, 86LabGTS, 95Cont, 90Beretta, and 90Accord. The non-float side
of each sample was ablated five times, with each ablation spaced 500 um away from the
preceding point. In the second phase of experiments, eight months later, the float side of
each unknown sample was ablated in the same manner to ascertain the degree of

homogeneity of the samples between the float and non-float sides. In the third phase,

another subset of samples (n=4) was ablated using the same parameters seven months
later to expand the data set for comparison. From Table 1, these samples included: '
94Metro, 9OSunbird, 89Sundance, and 89Brougham. Samples were similarly ablated five
times on both the float and non-float sides of the samples. In addition, a cross-sectional
traverse of two samples was performed during Phase II (86LabGTS) and Phase III
(94Metro) of the experiments. In addition to testing reproducibility of measurements, the
purpose of these trials was to ensure homogeneity of the samples when no parallel sides

are present, as is often the case when glass samples are submitted to analysts in caseWork.

RESULTS AND DISCUSSION

All data were reduced using the MassLynx software provided by the ICP-MS
manufacturer (Micromass, Ltd., Manchester, England), which performs a standard
guassian integration of peaks, yielding a maximum peak height for each isotope (Table
2). As precision in this study is determined by measurements of element ratios, relative
standard deviation (RSD) is considered for peak heights normalized to 88Sr (Table 3).
Data are presented in chronological order by groups of analyses (Phases 1, II, and III).

Overall, precision for sets of replicate analyses is good, with RSD values of
<10% for most elements, many <5% (Table 3). In the first two phases of research, 85Rb
and 98Mo, which had the lowest peak heights of all isotopes, were the only two elements
to demonstrate precision consistently >10% RSD. In all subsequent analyses, 85 Rb and
98M0 were removed from the element menu. In rare cases, for the other elements in the
menu (“Sn 89Y, 902r, 138Ba, 139La, and 140Ce), precision RSD values exceed >10%. For

example, 139La for 9OSunbird has a precision RSD of 29.7% (Table 3). However, the

mmv>m
meNc
oha_o
Vmwmo
mow—o
a~o_>
cvwmc
_-vo
combo
w_mN~_
mofimm_
wwm¢m~
wmwmm~
oommv_
~mw~¢
VVNNV
oov_v
afimnv
aoehv
amvmw
NN~cw
Na_m>
NQOmw
cocoa
amm_om
vomhhm
acNovm
omwwhN
comm—m

oO

wowow
acwoa
¢¢¢~N
~oom~
vamm
NNNoN
o_¢N~
~_amm
~vm¢~
_mamv
cmvav
N~¢¢¢
mama?
oomaw
a__n~
Nwm>_
Nfimh_
wvmw_
cacw~
mmmvfi
h_cm~
~m¢m~
h¢¢m~
thnu
ao¢¢a~
thooa
fiomfin—
wwhvmfi
ao_m~N

«A

vomvw
hmhmw
awa~o
m~m0¢~
mmmmN_
thmV—
-cm-
~mm_m~
o~__v_
vmmwon
h©~©m>
mm—mmb
fiovnmh
hmamoh
hccvmfi
fimmom—
ohmmmfi
ommcvg
owwmmfi
v¢wca
Nnmva
Nahwh
anma
hbovo_
m_m~a—
wmowom
mcheofi
ommma_
Nowmmm

«m

thN
ww—~
n—Nm
hmam
hhmm
mmvv
awam
ovvv
b_av
om_N
mvom
n~©~
omen
mcwu
Nwmm
_OmN
hmwm
voom
fivwm
w~m~
~mm~
mm¢~
vwom
wQMN
mmcme
onmwo
abocm
mcvoc
mmmfiw

o2

memo
mmamc
~mhww
mmomh
mwmmub—
mwcmvom
wwmmowfi
Nochmw~
mmn~va~
Nvmnwm
wemvom
hmaocm
oa_NmN
ammofim
ahhw_m
gcea_m
oe_mmm
MVVth
mNmNen
mommhm
vovwnm
~0mmmm
mom—am
aov__v
moons
oamwn
avnvo
“meow
hnvom

HN

388 5 am sea

ovam
mmhn
wmmm
came
mmamv
amwfim
~ownv
thcv
ommwv
oaham
ammom
mmmNm
m_N~m
mohmm
wah~m
h-~m
NchNN
~m¢¢~
amwNN
mmvofi
wmhmfi
vme¢~
m_ao_
mvac~
mo~mv~
mmm~m~
vamom~
mw~¢m~
mnemo—

>

thvmm
coma—m
v~o>vm
cNaNmm
hm—bov
wavmon
mcmhmv
ommefim
vmm_mm
mmmch
nm-¢m
oa~m_m
mammfim
mNOVNm
cofiwoo
NhMOho
owmwcc
mbNm_h
ommamh
mvam_m
_Nmm_m
cmwwhN
wobmmm
~Ncomm
cmmNNm
mwwmvm
hmfimom
vmfiaNm
wfihowm

Hm

c_wm_
enun—
mus—N
mmoc_
voom_
bhvmm
~mm>~
wmomm
mmNhN
waovom
ammmmm
ammomm
evwcmm
mc_wvm
«mne—
«Nom—
vv—c_
anoo~
NVN¢~
mmmw
Nv—~v
tha
ww¢-
wvnofi
mnbmw_
Nowhm_
owwhefi
moocwm
w—wNNN

am

88 3328 Z
0va Hocéoz
02m «no—.282
38 32252
36 Decade Z
3% Eocéoz
02m 23:-qu
03m :8?qu
02m 83:82
02m Socéoz
02m ”8:-qu
oEm “no—mac Z
35 32.28 Z
02m “no—.282
02m US$52
02m “aoaéoz
02m “mo—.282
03m Sowéoz
02m “moaéoz
62m «no—.282
02m «so—.252
02m “co—EH24
02m :8?qu
3.6 «3%-qu

ehflzuam

abaznam

vbaxuam

cbaznam

c:a=:am

an»;

vigobmoo
m-§o~omoa
Néteomoa

_ $99695
98003
«15003
miaoOma
@2503
7:503
mrmhhvnqch
Vagaqwm
mwmhhunqucw
NfihOfixGw

~ -mHOnaAww
nrfiaflvqnmuaw
fiicuwqugvmw
TEA—30am
mmndfluagxuaw
~-::<c=nmuaw
950mm
750mm
M4603
N456?

_ 460mm

m-N _ @ Hmmz
TN _ c Hm HZ
m-~ ~ c 892
N-N ~ o Hmmz

_ -N _ 0 H92
~nmm4qwm

88: :2 8mm

@9368 you n 35 H8398 =e .5.“ 3:30: xeoa US$35 .N 03.;

-aoo~
hoe—N—
mwmva_
arema~
ommnw_
awwmom
chm~N
~w>vc
wVOme
cflmmh
mthw
eccoa
vwaovfi
eve—mg
wmchmfi
boome—
omem_
acmm_m
oofimvm
mmmewu
w-v~m
one—fin

chVn
c—ooe
Nemhm
hh~mn
mmw_h
o_vwm
Ru

N_cwm
Newmw
vmahh
thmh
vwwmh
~vn~w
mow~w
ooowm
wcchm
ebaaw
ambom
cfiwmm
mmgmm
wwwom
avaom
cmoam
~m~mm
acmth
~h~va~
bvwa_m
Nwwnmm
mthVN

amhmfi
wvoflN
camcm
wmmom
vmvvm
n_ocm
«A

vawmm
wnmoam
whomo~—
wooom~_
w~mwao_
omvfio_~
o-o-~
mvnomm
fime—N
mnmamm
m~momm
Omhmmm
envenfi
~5v¢>~
hence“
wcaahg
o—vmvm
NNmmmm
om~ww~
~wma—N
cmhmvm
hh—MMN

m0>m¢~
mmbm>~
mvwvv_
mmmcvg
~___om
mmew
em

mw_o_
flux——
come
vmmc
swab
ovum
a_h~—
Nwfih
Nwhm
owva
mmo¢~
mvnvfi
came
hfimc
mach
owns—
ooan~
on—mm
wmmoc
Nwmma
whhwa
mmvvw

mNNN
ov¢~
~mo~
mmom
cmhv
hmw~
ova

mafia—Co a

aaammwu
N—Vha_m
oavawv
cvmcmv
«mahmv
nwnhov
~o~wwv
othmw
vmmmwh
aw_wmw
~mo~ww
wvmvna
Nmmwvo
mmowhc
moenoc
mnoeao
vmvwch
ohmma
bmvmn
VNocw
meam
cmh~a

wowaom
NNNVMN
mev~N
mwmmmm
m—mficm
@m_Nc
HN

an 88

an—ah
women
macaw
Nvawv
omeow
v—cfim
ommmn
v_mom
mmwwm
mmwcm
wwmwm
mmmme
momma
Nmoam
mmoom
ombmm
moomm
~wc¢a~
ow¢av~
mm~m>~
ocwmm_
~nomw~

ecom~
hNMV—
macmfi
memmfi
wvwm_
awhm

vwonhw
amvmaa
mow—on
one—av
amohww
omwawm
wooown
HmVON__
ch->o~
wm~oa-
obcvm~_
CVthmu
cobmwn
cmw-e
onwawm
vwhcme
cgthh
hwmmvv
cmcmVM
owac_v
whammv
b_Nwmv

fiwammn
ma_wow
mooaen
a—nman
~v¢v~h
n—mmNm
um

~ommv
mhMVW
mmmem
oovmmm
vomwmm
menwm
cawmfic
cmwcm
cmbmm
NvaCn
Neowm
o~cmv
v~mmm
amomm
ww~¢¢
ochM
vwvmb
owvmam
swamafi
coo—om
meCmN
mahwvm

NNQ~N
~¢vo~
mON_N
Nmfivm
aovmm
whocfi
am

as... so;
8:. .8;
8:. 8E
02... 8E
02... 8E
88 .8;
88. .8;
88 8E
88 .8:
88 8E
88 so;
02... 8oz
88 as...
88 sea
88 .8;
88 83
88 8E

e883

8888

885.0.

‘8qu

385m

36 «8:-qu
62m «no—.252
02.... 32.282
38 «mo—«-82
02m «nocéoz
oEm Socéo Z

3.484%

Nix—00mm
7230mm
miHOfiAow
oimhhvneqow
mtmhgai—ow
megax—cw
~-munvnqucw
mvnaflcnmxuaw
"innfluanumw
mwndflqumuow
anuflvquumw
_L::<u:uhuaw
930mm
«160mm
WED?
N->~Oma
7303
9N3 HmHZ
vim—c HmHZ
WNS mez
N-N$ HmHZ
TNS Hmmz
.finmmaawm

@8828
1:828
3828
$8838
72828
n-828m8

0302mm<£m§m

Acne—Econ So: u 35 down—new =a co.“ 3:30: anon 3225:: dos—5:8 .N 033—.

10

mgmmh
omoNeh

3mmm3m
vwoamm
vaomm
moamoN
mowcaN
-~na~
enhmwm
moowmm
coma—m
cooaOm
mm—num
emovam
hvooov
meah
ohNno
~3wwb
envvw
wwmam
wvooh
newgw
h—NVw
VmBmw
Vmabm
gmfiha
353ma
mmmmo
R0

mnV—mo
wwwhmc

mmw~3~
-N3m_
~3m>-
30mo3_
owhoo_
choc——
vwm3m_
m3¢-~
omnomd
_o_3-
hm—w~—
3wa33~
wmacmfi
hmmom
cmvmm
w3m_m
mwmmm
momcm
ovwmm
emcmm
33mmm
wmamm
mo_3m
osmmm
wcocm
mwmmm
«A

amuamm
mv—hcn

w3~m3-
enema—N
ovwmhww
uncann—
3mm3m>~
mmhmmhfi
amhoaHN
mwocwh~
3NN-¢~
35mmcw~
ows¢mw~
amowhmm
w—nmcvn
oooa—N
omemvm
~vommm
ammmvu
mwaaN
wmwhvfi
wwV~3~
hNVhM~
oo-v~
momwn_
unwoow
_oaw~N
aNaNNN
3m

e:
e:

-aw
name
~wbh
thb
Nwoo
~3wm
~www
3n_3
comb
coca
NN3¢~
3mmm_
eoha_
N033
when
mmnw
330m
_wm__
vomv
wnov
momv
when
3333
meow
mmnn
0005
034

3 _ 33.3.
3% _ 3N

mmween
mbwhmh
Nw_vch
NhNOhm
chwmve
oma_N3
whamow
~amovo
awmhco
nmaoco
hvamwo
ovw_vw
own—mm
hNNoVM
mcvwmm
wmmmmm
nwmomm
whwwov
333no_
a~woc~
~w33-
3oOm__
wm_a__

oOBVBwN
ww~335~
mom—mom

3N

38. 90 =3 ”I: 03. ole

a—mhhv
m~m~w¢

wofiow
~m—_w
3mm~w
¢mom3
vwomc
gwamn
aowfia
whoob
o~3mn
monmn
3m~hh
33350
wdmmo
v~no~
avo-
owwom
o~N_N
mnmmm
voNa
wflha
ehmo
mm~o~
ohoo
ca__h
333mb
omens

38mg—
3835—

oa_mww
Nmmavw
wwg—Nw
oooame
Noauhh
vwcwoh
mvmvoo_
wwoweh
w—wwmw
Noomow
hNhamw
fioaamo_
wavno—
ommovo
Novmom
mmowoa
mmvgoa
m0~h¢¢~
Newman
fiwommm
0333mm
mmaocm
~m->m
haemah
hmwhmw
vethm
3m

8
e:

mma¢0-
wmowma
oNtho_
thmbh
mow—ca
ch—haw
mo_hvm_
o_mNNm
mumowa
thvca
“Vanna
vanaom—
wwa_3~_
_wmwv
ovmmm
~o0m3
ocean
_womn
cnhmm
unvam
NwNmN
cvoam
Queen
Rwanm
mhwom
a_nmm
QM

3.835%
33355

30308-880
30308-880
30308-880
00308-880
30308-880
30308-880
30308-880
00308-880
30308-880
30308-880
30308-880
00308-880
30308880
038 300E
038 80$
038 80E
038 80E
038 3005
038 300E
038 3005
038 300E
038 300E
038 300$
038 3003
038 300E
038 300E

8. 81830 0.0.. gm.

N-NS H92
723 #92
.fiumamdammn

m _ -8033-X-333w
N ~ 0030-ch
g Tun-3043
o _ 083043

o-X-nnqow
Tun-333»
Run-333w
3-X-333w
m-X-nach
«iannuom
Tum-333w
TVA-333w
~ flea-X333
9.3300350
333000.050
93300330
”3.800130
~ -3~000<om
9038380
3038095
T038095
T038095
m 038850
933003
«13003
m-EoUma

 

0300:900\0_§m

8830.000 300 n 35 883030 :0 00.3 33303 0800 308.803: 30:33:00 .N 030,—.

11

300bo~
3~33__
mmw30
00mw0
0m3No~
~_hmo—
h_ncm_
~0w03~
3303m—
3m0mm—
NNN~3~
won~3~
-o3m~
mnw¢m~
mm—3m—
~3m—3—
mom0m~
m—w3w
hw33w
m33—w
~w-w
m0om0
~0oow
~mm3>
~m3~w
03330
mw3_o~
3m>~w3
oan33
m~3mh3
£0

3mo3m
mnN3m
m0oom
mmmfim
33NNm
muomm
03~0m
3mwmh
0m333
00hm0
m33¢0
333mm
wm3m3
33033
H0033
0N3—0
003mm
O3wwm
3wwo3
mono3
omn3m
ww333
~m~3m
mm03m
whowm
03333
bum—3
0w3w3m
NmN3mm
_Mo~mm
«A

wmm_m_
00mm3_
0Nm0N~
30—m-
w3~m~—
000w-
3~Nmm_
w03¢0~
whfimmm
mom30N
00N03N
O3wm0N
Nmm~mm
mnoamm
3m~N3N
mmmm3~
hooowm
33w~om
003m~m
0nh-m
b3~mhm
~mbw~m
osmomm
mmw33~
N_m—wN
30wm0N
N3w33m
m~m>_m
3_m003
m03003
0m

«a
«a
«E
«a
«a
«a
«a
a:
“a
«E
«a
«a
“a
«a
“a
«a
a:
a:
«a
«a
«a
s:
S
a:
«a
“a
«a
a:
as
.0:
oz

cm03_m
m3wn~w
w33ww3
mmwo03
3—3003
owwomm
3mw_~3
33m300
m0¢3m0
Nw0o_w
33mwhh
3333—w
33nh¢0
~0mnoh
33¢mm0
m—m3ch
wm0030
3w33m3~
30mc33_
N00303—
30¢033~
m033mb—
m3m30m—
Nwmmmm_
3w—~mm_
Nw0wwm~
hO3h~M~
m0m0_N
03mm¢~
33MNNN
HN

8:8 3 am 080

033nm
>330N
m00m~
3303m
3333w
~mh3m
hmo~m
ooOw3
~00m3
m3mw3
00003
33330
w_m_3
33333
30Nm3
cam—3
N3—33
Nomm3
0NNm3
30033
N—Nmm
hwwm3
w3hbm
w~m0m
m_owm
0N—0m
3~mhm
_w_~m3
0m3>~3
300333
>,

m303hh
Nm3mow
~mhnw3
_om~03
~>m_Nn
Nw3m3h
~N03ww
w3~moo_
30—330
3303mo_
N33~00
c—0wo—_
3~0w~0
wm~m—0
0m3—m0
m33_30
monocog
_n3003—
w3m3mm~
O330_m~
m~03¢m~
on3mnm_
30mm0m~
mem—m—
w3033m—
w3nn—3—
00wmmm_
Nomw30
33~3~0
0_Nmm0
8m

«a
an
«a
«a
«a
«E
«a
an
an
«a
“a
“a
a:
«a
«a
«a
S
“a
«a
an
an
S
a:
a:
«a
«a
«a
«a
a:
S
8

038 300.0
038 3005

038 3003-002
038 3003-002
038 3003-002
038 3003-002
038 3003-002
038 300E

038 300E

038 3005

038 3003

038 300E

038 3003-002
038 3003-002
038 3003-002
038 3003-002
038 3003-002
038 300—0

038 3003
038 000E
038 300—0

038 300—0
038 3003-002
038 3003-002
038 3003-002
038 3003-002
038 3003-002
383008
383008
383008

0%

Nésza
3823
8.2523
0.2328
8.0823
«-3028
3823

8.85588
#850588
8-880.388
08509.88
78809.88
8.830588
3.850588
8.85388
3850288
785588

8.828208

0.820858

8.828888

wag—0:208

78203208

8.8308008

3.8280208

$820858

2580808

72500208

8-20 82
3.20 82
8-20 82

03003 30. Sam

383.000 00: n 85 880380 :0 80.3 $380: 0800 30308035 30:33:00 .N 030B

12

3mmow_
~N33—N
3ww33~
~3000~
00033N
0NN03—
33~m-
m330m
N0330
33333—
303N3~
33033—
N~N30
w0m30
330333
30~_3~
33w-_
mm~3-
0_3~3_
m3—m3~
£0

m3~h3
33~w3
~3033
__303
33333
~3333
30w03
30N0m
33333
03333
03033
w0333
333~3
~3NN3
30033
N3M33
-mmw
33333
N3333
omNmm
34

3w33-
0b~mwm
3hwm_N
ofimomm
M3~w3m
w-3mm
30mwhN
Nwommm
~3333N
mmw33N
03303N
3w030~
hm333~
333M3N
00—33N
_N003N
03b3w~
3w303~
33w3m_
h0_0-
3m

“a
3:
«a
«E
e:
«a
«a
a3
«a
«a
«a
«a
«a
“a
“a
“a
s:
«a
«a
“a
o:

~03Nww

30_3mo_

330000
33ww3w
0NON0w
30Nh>w

0~0~33~

333333
3N00m3
3Nme3
w30333
w30__3
3m3333
303333
30N333
Nhnmh3
~330w3
n_>033
3_3N03
m33w33
3N

1333-00.. 3. 380m 08. -_ 0.0-

~3~w3
«N333
33—33
~0333
~h-3
3~333
30Nh3
3N30N
w~33m
033wN
00NNm
0-3m
3w0w~
33ww~
>300N
>333m
w—333
N0nwm
3030N
3~m3~
>.

3333—3—
3~3333~
3m~30-
w3~3~m—
N_wM33_
hw303m_
333330—
3MN033~
33303——
30033——
3N333N~
m0m33N~
ww3~03~
333303—
>_3~3~_
n3~3>-
_0_M~N~

3w3—3w
w3~>m>
30__30

3m

3:
«a
«a
3:
«a
“a
“a
3:
“a
“a
3:
S
a:
«E
«a
s:
s:
s:
.0:
“a
03

30308-880
30308-880
30308-880
30308-880
30308-880
303008-880
30308-880
038 300.0

038 300$

038 300$

038 300$

038 300E

038 3003-302
038 3003-30 Z
038 3003-302
038 3003-302
038 3003-302
038 300.0
028820
028820

8.38.83

3888033023
0-03.3028
8-032328
33828
8.03523
N-x-§oza

Eek-9823
8-30588
#000588
8.030588
300588
7030388
8.03588
3.30588
8.30588
330880
730588

8.2528
«.2328
8-2328

 

0330330323333

€833.00 003 n «\5 880—033 :0 30.3 83803 0803 30308035 30333300 .N 033,—.

13

Table 3. Sr—normalized Peak Heights for analyses.

PHASE I

Rb
Zr

Ba
La
Ce
Rb
Zr

Ba
La
Ce
Rb
Zr

Ba
La
Ce
Rb
Zr

Ba
La
Ce
Rb
Ba
La

Zr
Ce

Sample and peak/Sr for each replicate (n=5)

NIST 612
0.57618
0.43616
0.22362
0.21 166
0.60199
0.58210
0.82620

93Cav
0.05081
0.05141
1 .24813
0.00728
0.31575
0.04695
0.27453

89GrandAm
0.02534
0.03008
0.74092
0.00506
0.20263
0.02462
0.06271

86LabGTS
1.07430
0.10420
0.97639
0.00884
2.46869
0.16141
0.43934

95Cont
0.05137
0.00925
0.26557
0.04581
3.65430
0.12779
0.09082

0.54719
0.42289
0.21460
0.20176
0.59315
0.59184
0.84719

0.03413
0.05038
1 .16742
0.00800
0.27928
0.04607
0.25347

0.02332
0.03360
0.76536
0.00420
0.20442
0.02551
0.06336

1.04293
0.09961
0.93234
0.01 174
2.41697
0.15796
0.42703

0.04388
0.00860
0.25479
0.04631
3.53862
0.12419
0.09054

0.57397
0.43214
0.22137
0.19481
0.57016
0.58567
0.82147

0.03330
0.05248
1 .19620
0.0071 1
0.28256
0.04673
0.26965

0.02416
0.03397
0.78584
0.00428
0.19959
0.02620
0.06196

1.04813
0.10258
0.95475
0.00830
2.38635
0.15677
0.42602

0.03485
0.00802
0.25944
0.04507
3.62661
0.12638
0.09229

14

0.57520
0.44358
0.22766
0.19882
0.58458
0.60158
0.80610

0.13131
0.05020
1.20791
0.00744
0.30184
0.04793
0.27487

0.02241
0.03299
0.77518
0.00373
0.20369
0.02593
0.06302

1.09090
0.10141
0.97813
0.00678
2.44481
0.16415
0.44891

0.03989
0.00787
0.25510
0.04653
3.62926
0.12603
0.09198

0.56714
0.44437
0.22657
0.20189
0.59744
0.60350
0.81 108

0.02637
0.05259
1 .1 9945
0.00805
0.29029
0.04548
0.26669

0.02504
0.03262
0.77642
0.00386
0.20155
0.02562
0.06216

1.06134
0.10397
1.00286
0.00750
2.47283
0.16030
0.42760

0.02736
0.00780
0.25256
0.0471 1
3.57162
0.12432
0.09244

RSD ”/0

2.1%
2.0%
2.3%
3.1%
2.1%
1.6%
1.9%

78.8%
2.2%
2.4%
5.6%
5.1%
2.0%
3.3%

5.0%
4. 7%
2.2%
12.3%
0.9%
2.4%
0.9%

1.8%
1.9%
2.7%
22.1%
1.5%
1.8%
2.3%

23.0%
7.4%
2.0%
1.7%
1.3%
1.2%
1.0%

Table 3, continued. Sr-normalized Peak Heights for analyses.

Rb

Zr

Ba
La
Ce

Rb
Y
Zr
Mo
Ba
La
Ce

PHASE II

Rb

Zr

Ba
La
Ce

Rb

Zr
Mo
Ba
La
Ce

Rb

Zr

Ba

La
Ce

90Bcretta

90Accord

NIST612

93Cav-float

0.04549 0.06274 0.05456
0.01812 0.01613 0.01815
0.20427 0.19812 0.20626
0.01 1 16 0.00926 0.00685
0.28565 0.26509 0.28100
0.06517 0.06180 0.06531
0.18671 0.17860 0.19510
0.04676 0.04075 0.03722
0.02218 0.02279 0.02228
0.36604 0.37722 0.37624
0.00665 0.00510 0.00360
0.28150 0.24689 0.25426
0.03419 0.03431 0.03580
0.10055 0.09815 0.10104

Sample and peak/Sr for each
0.56774 0.55009 0.63531
0.41685 0.42428 0.42146
0.20948 0.21789 0.20934
0.19273 0.20786 0.22774
0.53210 0.53897 0.53534
0.56455 0.56556 0.53501
0.71119 0.71083 0.69686
0.09823 0.05371 0.06815
0.04659 0.05149 0.05025
1.05594 1.09315 1.12866
0.02193 0.01663 0.01306
0.33311 0.28264 0.28285
0.04555 0.04562 0.04570
0.26856 0.26646 0.26734

89GrandAm-float

0.03655 0.04952 0.04259
0.03194 0.03240 0.03078
0.71778 0.74371 0.71745
0.01086 0.00902 0.00793
0.21786 0.21 146 0.20003
0.02638 0.02594 0.02506
0.06679 0.05684 0.06037

15

0.04725
0.01775
0.19585
0.00674
0.25273
0.06234
0.17155

0.04356
0.02356
0.38512
0.00401
0.28388
0.03461
0.09867

replicate (n=5)

0.57130
0.43456
0.21404
0.20275
0.53967
0.56600
0.71464

0.04091
0.04842
1 .10967
0.01065
0.27862
0.04395
0.24786

0.0371 1
0.03254
0.72959
0.00540
0.19883
0.02583
0.06073

0.04973 13.3%

0.01778 4. 7%
0.19224 2.9%
0.00587 27.3%
0.25503 5.6%

0.06376 2.5%
0.18066 4. 9%

0.03867 9.3%
0.02359 2.9%
0.37884 1.8%
0.00401 26.4%
0.25951 6.3%
0.03381 2.2%
0.09795 1.4%
RSD %
0.66161 8.1%
0.42986 1.6%
0.21614 1.8%

0.22356 6.9%
0.53050 0.8%
0.53556 3.0%
0.71 127 1. 0%

0.04035 39.9%

0.05012 3.9%
1.1 1251 2.5 %
0.00772 39.3%
0.29248 7. 6%
0.04313 2. 6%

0.25222 3. 7%

0.02751 21.0%

0.03241 2.3 %
0.74367 1.8%
0.0064] 27.1%
0.20594 3.9%
0.02499 2.3 %
0.05781 6.4%

Table 3, continued. Sr-norrnalized Peak Heights for analyses.

86LabGTS-float

Rb 1.05561 1.10841 1.10322 1.12363 1.09350 2.3%
Y 0.09565 0.09627 0.10173 0.09955 0.09953 2. 6%
Zr 0.84067 0.88279 0.89762 0.88877 0.93635 3.9%
Mo 0.02191 0.01 179 0.01420 0.01268 0.00989 32.9%
Ba 2.1 1 178 2.24842 2.25134 2.29036 2.32543 3.6%
La 0.14090 0.15426 0.14814 0.15353 0.15541 4.0%
Ce 0.37037 0.38477 0.38390 0.39192 0.38768 2.1%
95Cont-float
Rb 0.05462 0.04955 0.04074 0.03208 0.06919 28.6%
Y 0.08696 0.09048 0.08824 0.09028 0.08948 1.7%
Zr 3.21206 3.22940 3.20355 3.30166 3.23587 1.2%
Mo 0.01 188 0.01849 0.00966 0.00923 0.00636 41.0%
Ba 0.29792 0.26795 0.26933 0.26128 0.25994 5. 7%
La 0.04409 0.04412 0.04299 0.04305 0.04433 1.5%
Ce 0.12216 0.11532 0.11266 0.11157 0.12207 4.4%
9OBeretta-float
Rb 0.06387 0.06960 0.04558 0.05516 0.04502 19.5%
Y 0.01747 0.01806 0.01689 0.01821 0.01819 3.2%
Zr 0.20864 0.20503 0.20676 0.20578 0.20844 0.8%
Mo 0.01 121 0.00905 0.00776 0.00872 0.00987 13.9%
Ba 0.27710 0.25154 0.24776 0.26512 0.29144 6.8%
La 0.05971 0.06048 0.06015 0.06200 0.06235 1.9%
Ce 0.15400 0.15288 0.15183 0.15339 0.15033 0.9%
90Accord-float
Rb 0.06842 0.05471 0.04956 0.05532 0.05144 [3.2%
Y 0.02148 0.02206 0.02266 0.02249 0.02202 2.1%
Zr 0.36716 0.37068 0.38913 0.37247 0.37130 2.3%
Mo 0.01036 0.00618 0.00964 0.00590 0.00499 32. 6%
Ba 0.26823 0.25225 0.25665 0.25212 0.23285 5.1%
La 0.03309 0.03368 0.03455 0.03476 0.03228 3.1%
Ce 0.09057 0.08784 0.08685 0.09900 0.08458 6.2%
Phase 111
Sample and peak/Sr for each replicate (n=5) RSD %
NIST 612
Y 0.46124 0.45984 0.46893 0.45687 0.45574 1.1%
Zr 0.23138 0.23508 0.23328 0.22704 0.23154 1.3%
Ba 0.52378 0.53865 0.52422 0.52213 0.54552 2.0%
La 0.60107 0.59872 0.57807 0.58658 0.59632 1.6%
Ce 0.73008 0.72518 0.70878 0.70323 0.71892 1. 6%
85Brougham
Y 0.02814 0.02760 0.02822 0.02844 0.02752 1.4%
Zr 0.99588 0.97983 1.00389 1.01014 1.00165 1.2%
Ba 0.26023 0.2071 1 0.20885 0.20186 0.20446 11.4%
La 0.03097 0.02822 0.02827 0.02662 0.02633 6.6%
Ce 0.07629 0.06678 0.06047 0.05818 0.05836 12.0%

16

Table 3, continued. Sr—normalized Peak Heights for analyses.

Zr
Ba
La
Ce

Zr
Ba
La
Ce

Zr
Ba
La
Ce

Zr
Ba
La
Ce

Zr
Ba
La
Ce

Zr
Ba
La
Ce

Zr
Ba
La
Ce

SSBrougham-float

0.02916
1 . 10373
0.20893
0.02827
0.05917

89Sundance
0.06414
0.76770
0.31991
0.07346
0.15916

89Sundance-float
0.06452
0.73644
0.26680
0.06634
0.14583

94Metro
0.03502
0.68983
0.17839
0.04415
0.14716

94Metro-float
0.03408
0.65288
0.17803
0.04502
0. 141 55

9OSunbird
0.02507
0.40343
0.23556
0.06785
0.09302

9OSunbird-float
0.02405
0.40728
0.23638
0.03588
0.08417

0.02924
1.10724
0.28705
0.02661
0.06212

0.06501
0.74801
0.28210
0.07598
0.15013

0.06850
0.78412
0.26928
0.07097
0.14227

0.03594
0.70041
0.17343
0.04447
0.14215

0.03534
0.66554
0.16908
0.04389
0.13904

0.02589
0.40193
0.21910
0.03857
0.08603

0.02677
0.41518
0.22132
0.03876
0.08461

0.02952
1.1 1818
0.2061 1
0.02682
0.05378

0.06791
0.78686
0.28137
0.07169
0.14397

0.0661 1
0.78224
0.26651
0.0731 1
0.14850

0.03692
0.68880
0.17083
0.04469
0.14227

0.03599
0.69552
0.17669
0.04545
0.13965

0.02626
0.40777
0.22804
0.03856
0.08829

0.02467
0.41545
0.22122
0.03925
0.08625

17

0.02813
1 .06757
0.20403
0.02661
0.05629

0.06615
0.77094
0.2743 1
0.07096
0.14299

0.06884
0.80346
0.27526
0.07276
0.14271

0.03575
0.70881
0.17790
0.04524
0.14209

0.03674
0.66802
0.17068
0.04132
0.13770

0.02632
0.41634
0.231 15
0.03848
0.08699

0.02660
0.42691
0.22310
0.03926
0.08406

0.02892
1 .10663
0.20163
0.02594
0.05665

0.06601
0.76171
0.27080
0.07065
0.14642

0.06785
0.76869
0.26991
0.07270
0.14755

0.03496
0.71257
0.17547
0.043 89
0.14124

0.03381
0.65499
0.17579
0.04302
0.141 15

0.02638
0.41796
0.22605
0.03799
0.08635

0.02570
0.41712
0.21371
0.03680
0.08392

1.8%
1.8%
16.6%
3.2%
5.5%

2.1%
1.8%
6.9%
3. 0%
4.4%

2. 7%
3.2%
1.3%
4.0%
1.9%

2.2%
1.5%
1.8%
1.2%
1.7%

3.5%
2.6%
2.3%
3.8%
1.1%

2.1%
1.8%
2. 7%
29. 7%
3.2%

4.6%
1.7%
3.7%
4.1%
1.1%

importance of replicate values is apparent, as when the first of the five replicates is
removed, the precision RSD decreases to 0.7%. Graphical plots of the data allow for
quick recognition of such uncommon imprecise analyses by comparing relative
proportions of each element in a sample.

Element proportions were compared using ternary diagrams plotted with IgPet
for Windows®, a geological plotting software package (Terra Sofia Inc., New Jersey,
USA). In each diagram, integrated peak heights are plotted for three elements. The
location of a sample in the triangle represents the relative proportion of each of the three
elements with respect to the other two. Therefore a sample plotting in the center of the
triangle, equidistant from each of the three apexes (X, Y, and Z), would have a relative

composition of approximately 33.3% X, 33.3% Y and 33.3% Z.

Precision and Discrimination

The results from the first set of analyses (Phase I) on six unknown fragments
and NIST 612 are presented in Figure 1, showing relative peak heights of Zr, La and Ba
for each sample. Each symbol represents one analysis for the non-float side of a given
fragment. Each sample was ablated five times to determine precision. Precision for this
set of analyses is good (except for Rb and M0 in several samples, as described above).
Aside from low RSD values (Table 3), the precision can be evaluated graphically by the
relatively tight grouping of replicate analyses in Figure 1. Notice that the only two
samples that are not clearly differentiated by ratios of these three elements are
89GrandAm and 93Cav. Both samples have relatively low counts of La, plotting near the

Zr—Ba axis with similar proportions of those two elements. In this diagram, the samples

18

Amanda—NV $5820 025 805 main “82362on

base? 8: 2a .503 98 Ego—35am .moafiwm 23 3:0 35 802 .mEoEwfim .«o

02m waging 65 so 2958 some 8m mug—.28 832%: 95 832%“ flonfixm ._ 89E
88m m8 HmHZ Ea fiaofiwfim 2305?: cam mo How 05 8% mug—«aw =a mo SE 4 «ENE

mm hN

 

Q «teem 83320 ca
53260 “20325 3

3.0 883 3.an 3

.8950 Bow 3

E... 380 838m 3

Ecoo< some: ca

m5 HmHZ

déoxom+

m.—

19

overlap within the range of the precision of the samples and cannot be clearly
differentiated. However, when La is replaced with a more abundant element, Sr, the two
samples separate (Figure 2). Using only these three elements (Zr-Sr-Ba), all six unknown
samples can be clearly differentiated.

Although the precision decreased slightly for several elements in select
samples, the results from analyses of the float side of the same six unknown fragments in
Phase II were very similar to those from Phase I. The RSD values for replicate
measurements remained less than 10% for all elements (Table 3). Using the same Zr-Sr-
Ba plot as used for the non-float analyses (Figure 2), all six samples are still clearly
distinguished from one another, even with slightly lower precision (Figure 3). In this
diagram, for example, one of the five 93Cav (open squares) analyses plots slightly closer
toward the Sr-Ba axis than the others, which is reflected in the slightly higher RSD value
(7.6% vs.5.1% in Phase I, Table 3). However, the significant overlap of this data point
with other replicate measurements for the sample forms a tight cluster that is plainly
dissimilar from other samples.

Analyses from the non-float (Phase I) and float (Phase II) sides, which were
acquired eight months apart, are shown together in Figure 4. Note that although there is
slight drift for some samples (e.g. 95Cont), the range for all samples remains quite
narrow. Analyses for individual samples still form distinct clusters. Table 4 gives
precision RSD values for elements in each sample using analyses from both Phase I and
Phase II. RSD values remain below 10%, with the exception of a value of 10.1% RSD for

Ce in one sample (90Beretta). Remarkably, the molten tin in the float process does not

20

080.508 3320

05 ~ £0me 80.0 >008 0:0 8305.50» menus—05 £2083 =0 005 00002 0 0.59m
3 050m 05 Ba flonfimm .0600 03 05 00 um 55 _ 2:me E 3 mamas—08 A 89E
80¢ NS .52 0:0 $00893 0305?: 00m .00 00m 05 00.“ 6.09298 :0 .00 SE .N 9...»:—

mm ._N

 

a Sachem 00—00550 00

§ 5:98 03:80 Q

Q .. 2.0 8380 500:0 3
‘ 800000 30m m0

:2 280 80:8 3

00003.. 00:03 o0

N 5 52

.w

<09XOD+

21

300830
.«0 02m 000m 05 00 0383. some 00.“ 89328 88:02 o>w Sewage £008? d 895
800 NS H92 05 8530.0 0305?: xmm .00 How 05 00.“ 89:28 :0 .«0 SE .m charm

mm .N

 

ma 0320mm 00—05020 ca
0 3:960 00—03000 8

G & who 833 500:0 3
s 53:00 Rom mo
:2. 0:80 80:00 on

2003.. 300m 00

NS HEZ

em

400%0D0

22

.9 08:00 800.508 00 020 88c 05 00 89:08 88:08
0>: 8000808 £005.00. 0000 0:53 9 08:00 80080.: 00 020. 880-000 05 :0 0:088
:08 000 89:08 88:08 0.5 E08008 2008.00 008—0 .: 08:0 000 0 08:0 50:

8000 m8 .502 00.0 800888 850578 020 .0 80m 05 .80 8.0.3000 =0 00 80:0 .80 008E

mm .N

 

800m 880980 00
0338 .2258 8
m0.0 00080.: 00:00.30 on
58000 0000 m0

:2 0:80 8080 3
8000/0. 800$ 00

N fl 0 Hmmz

 

<0oxom+

.0

400%on0

23

.0m 8 0000:8000 8:80: 880

000 003w 0.8 000:8 0mm 008088 :08 00 00% 08: 05 00 803000 02.8 000 020
080-000 05 00 083000 88:08 020 800—00: 00:0.» 0mm 000.0 d 000 0 00000 0:
8:00am :0 000 0000300 020 08m 000 8000-000 00.: 82.0.» 0mm 003080 .8. 030,—.

 

 

 

 

 

 

 

 

 

 

 

 

 

£30 ..\o_ .o— 080.0 o\ow.o $0.0 000$ foo.» 00
..\&.N o\.._ .m o\oN.m $0.0 o\oN.N o\oc.m o\om.v «A
$00 0006 $0.0 o\ao.m o\&.N o\o_.o o\om.m 0m

o\om.wm o\oN.—N £06m ..\&.wm o\oN.¢v $0.3 o\ow.m 02
o\oc.N {and {and o\om.m a\om.m o\om.n o\oc.m 0N
o\oc.m 000d $0.— ..\&.N 000.0.” food o\oN.N >
£00.0— o\o¢0_ fobfiN o\oo.N o\a_.cm $0.5m 0000 £—
SM0 2H0 2H0 Eu: EH0 SH: SH:

0.503050 038095 8000mm mspg 05600000” >003 NS 0.92

 

 

 

 

 

 

24

appear to affect the measured trace element ratios of the glass by imparting additional
elements to the float side, as is observed with Sn.

Results for the analyses of the second subset of samples in Phase III of the
experiment are given in Figure 5. Data points include five analyses conducted on the
float side and five analyses of the non-float side of each sample for a total of 10 analyses.
Note that a different set of elements is used in this comparison (Sr-Y-Zr) than in Figures
2-4 (Zr-Sr-Ba), as they provided the better discrimination between all four samples. The
four samples plot in distinct groups, with slight overlap between two samples
(89Sundance and 94Metro). Again, using another group of elements (e.g. Ce-Ba-La,
inset Figure 5), these samples can be clearly separated. However, samples with roughly
similar peak heights of one element can also be differentiated by modifying the relative
value of the component, as suggested by Watling et a1. (1997). Figure 6 demonstrates
this effect by multiplying Y by a factor of ten. All peak values are multiplied by the
same value, resulting in separation of different groups while maintaining the precision of
replicate measurements.

Figure 7 shows the entire set of ten samples. Note that only two samples,
90Accord and 9OSunbird, cannot be excluded as having come from the same source.
Using the six elements in the abbreviated menu (88Sr, 89Y, 9°Zr, 138Ba, 139La, and 140Ce)
and four additional isotopes (“TL 5 5 Mn, me, and 208Pb), these two samples could not be
distinguished from one another graphically. However, the samples were known to be
different based on thickness measurements (90Accord, 2.04 mm; and 9OSunbird, 2.24
mm). The samples were also discriminated using simple t-tests comparing p0pulation

means of measured elemental ratios in each sample at the 95% confidence level (P<0.05).

25

.3000: 0:85-000 80080—0 00.0.0 00002000 00 wfimz 0800088000

0: 08 CN->.0mV 80:0 05 5 0800:8880 b.0080 0.00 80 00:0 80808.: 05 00:0 802
8000080 00 006 08c 05 00 088:08 0.»: 000 006 0000-000 05 00 088:08 00,:
.8088 :08 000 80.2000 88:08 08 0000808 20:00.3 .0: 08:0 0000.: NS 0.92
000 808580 0300000 000.: 00002000 .«0 80m 05 0.: 89:00.0 :0 00 80:0 .m 0.0—um...—

.N 0w

 

 

 

 

 

 

202 80 a
850.3 0:850 00
52880 £0 a
2350 2.08.0 8
00 52

 

 

 

 

<1+OI+

26

.m 809m m0 808 05 80 838% .08 .«o 0808
a 3 9V 80000908 000 $502000 mo 8on 05 336% a 83m 08¢ NE HmHZ
000 8000038 03920: .58 300225 «c 8m 05 08 monk—000 :0 mo SE .e 95»:—

.N 0m

 

 

83: so .3
855m 5885 a
60385 35 3
293 825 8
ms 52

<1+OI+

or;

27

8080000 .00 0000005800 .000 w0000 0800000000000 000003000w

80 000000 00003 00000500 000 0.500.050 00000000 00 0000000000 000 0008 00m
8003000 .00 0000 0000.0 05 80 00 08000008 02.0 000 0000 00000000 000 00 08000008
0>m0 6000000 .0000 00.0 00900000 80000000 08 00000008 0.000000% .000 000 .00 .0 00000 0000.0
N00 0M0 Z 000 8080300 03000000 08 .00 000. 000000 80 00.0 00900000 :0 .00 0000 .0. 0.530

.N 0w

 

202 80 a
85050 58:0: 00

Q 8208.0 .20 8
+ § 00505 808.0 8
000.030 00003800 00

0000050 000030000 8

who 80.3 57.0.00 8

3280 08.0 3

E 00000 0000000 0w

00000.0~ 00000.0 oo

20 0.22

 

O
Q
Q

<0000U0+0u+

0 Fan.

28

There are two feasible explanations for the similarity of these two samples.
First, the samples may have come from completely different sources and by chance have
a trace element composition that cannot readily be differentiated using the graphical
method. Alternatively, the glasses may have come from the same manufacturer, having
been produced at different times on the same float line (explaining the different
thicknesses but similar trace element compositions). However, the vehicles from which
these samples originated were manufactured in Sayama, Japan (90Accord) and
Lordstown, OH, USA (9OSunbird). Unless the glass from the 1990 Accord was installed
after-market, it appears unlikely that the second option is the case. The Similarity of
these two samples merits further investigation, as the first Option is the preferred

explanation.

Precision and Sample Homogeneity

In each of the second and third phases of the experiment, a cross-sectional
traverse of one fragment was conducted to ensure reproducibility of measurements
through the thickness of the sample. In Phase 11, sample 86LabGTS was ablated 13 times
in a line perpendicular to its parallel sides, with ablation spots 300 um apart. Figure 8
shows a schematic diagram of the fragment and distance between ablation points plotted
against raw peak heights (in counts) for Ce and Y, as well as ratios for Ce/Sr and Y/Sr.
There is variability from analysis to analysis in the raw peak heights for each element
(Figure 8a), but normalization to one element (Sr) has a smoothing effect on the data
(Figure 8b). Although many factors influence elemental signals in an analysis (laser

power, sample coupling, etc.), the ratio of counts between two elements should remain

29

00000> 0080000000 w0000 000.0000

0000 0300000 0000 802 .0m 00 0080000000 > 000 00 00.0 00030000 000000 00000> 00000000

.00 0000 .m 000000000000 05 w0000 > 000 00 00.0 00000000 0008000 00000 00000> 00000000 .00
0000 .< 000000 000000000 2 .00 00000 0 00.0 00000 001 com 000000 0003 000.0000».~ .00 00000
00000 3.00003 0000000 .00 0000000000 00000000000000 00 00000w000 000000000000 .0 0.0—$0.0

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

00. 10 0000005 00.10 00.00005
89. 800 82.. 89 o . 89 800 88 080 o .
d d 4 CO . q 0 OO
. . 0.0
L0 . r
.aaaaaaaaaaaa. 3 50> .«aaaaaaaaaaca..iw000
. . 0.0
. . 0.0 . W
p 0 - NO 0 . 0 m w
0
00:10 00000000 00010 0000005 m
89. 080 080 89 o 89 88 88 080 o mu
- u 0 . O u a # O w
fl 0
T I 1 gr I"
W .0. . . 808....
n .0. a .0. a. a .0 00005000 .4 000.580
. .q. a, a. > f a. a. a. . 8800 $0
n a a. .0. #889.
T L . . 8800 . . . 0808 .

 

 

(um O)19°|:l

30

approximately constant, assuming minimal fractionation effects (Trejos and Almirall,
2004). This effect is observed in the relatively flat profiles in Figure 8b.

Similar profiles are seen for a cross section of sample 94Metro (Figure 9),
which was analyzed in Phase III, seven months after the cross-section for sample
86LabGTS. Again, ablation spots were spaced 300 um apart. The inconsistency of
analyses in terms of raw peak counts for La and Ba (Figure 9a) is greatly diminished
when analyses are normalized to Sr (Figure 9b). Given the relatively flat profiles in
Figures 8b and 9b, replicate measurements of trace element ratios through sample
transects are considered both precise and homogeneous. The precision RSD values (”Sr-
norrnalized) for all elements within an entire transect are ~5% or less (Table 5). In
addition to the compositional homogeneity demonstrated between the float and non-float
sides of fragments in the previous section, these results show that the technique is precise
even in cases where no parallel sides are present, as is often true of casework samples,

where only irregularly shaped questioned fragments are found.

APPLICATION OF THE METHOD: CASEWORK EXAMPLES

Since completion of the experiments and evaluation of the results, the technique
has been applied to casework for the Michigan State Police Forensic Laboratory in
Lansing, MI, USA. In cases where fragments cannot be discriminated on the basis of
physical properties, such as refractive index, or major element composition as determined
by SEM-EDS, trace element ratios are determined by LA-ICP-MS. The following case
examples demonstrate the discriminatory and associative power of the method. In cases

where an exclusion is made, the analytical scheme is improved by eliminating Type II

31

00000.» 0000000000 m0000 000.0000 000.0 0300000

0000 800 .0003. .5 00 0080000000 000 000 0.0 00.0 000E080 000000 00000> 00000000 .00 0000
.0 000000000000 05 w0000 0m 000 00 00.0 00000000 0002000 00000 00000> 00000000 .00 0000
.< 000000 00000000 0960 .00 00000 0 00.0 00000 8: 00m 000000 0003 000000040. .000 0000.00
0800 00000200 0000000 .00 00000>000 00000000000000 .00 00000w000 0000000000m .0 0050.00

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Ea 88005 05.0 8:305
OSN 030 980 8r. 0 . SON 80w gp 8.0. o .
0000—01-0_«<+4—~000 GO 0044—0000—dqu-d0q0q 80
i «d L no.0
o o o e
0 o 4 2 o o o o o o 03...
.wfim . 500..
r 0.0 80
.. 0.0 v 8.0
. m
FFL — 00—0» 0—0 r 0.0 uprpbbrlpp—rbppv—lrh 0 0'0 m M
u
0
E: W
0 00000005 0E3 0000005 )
L
88 8.... 080 80 o 88 80. 080 80 o
< d — d _ d — d fi S gr 0 1 4 d — < 4 u q — q q d — d 4 4 d g m
0
.N
w
i g 0 O 1 88m (
O
o o 0 00.500 0 o 00:38
o 0m 0..
r g 1 r go
0 C
o 0
.0. 0L0 00. .088... 0r. 0» .0.

 

(un1 0) wow

32

.0m 8 0000000000 00003000 00000 00.0 003m 000 00000> Qmm .00 000
0 0000000 00 00000>000 00000000000000 0.50 00 00000000 00.0 0020.» 9mm 000000000 .m 00000.0.

 

 

 

 

 

 

 

$0.0 $0.0 no
$2. $3 3
$0.0 $0.... 00
$0.0 80.0 .N
.00... 0.0.0 >
0": an:

2323 000.300

 

 

 

 

 

33

errors. In cases where an inclusion is inferred, the association between glass fragments is

greatly improved by reducing the size of the class to which they belong.

Case #1 .' Lansing, MI Homicide

A 26 year-old man from Lansing, Michigan was accused of shooting and killing
his former wife. She was found dead on a sidewalk, several meters from her parked
vehicle and just two blocks from her home. Two samples were submitted: one
questioned fragment removed from the victim’s scalp and multiple known fragments
taken from her vehicle. The questioned fragment, three fragments of the known glass,
and NIST 612 were analyzed in triplicate to demonstrate reproducibility of
measurements. Figure 10 shows one of the diagrams used to conclude that the known
and the questioned fragments were consistent in elemental composition and could have
come from the same source. Replicate analyses for each sample plot directly on top of
one another, and no combination of the six elements (sssr, 89Y, 90Zr, 138Ba, 139La, and

140Ce) provided discrimination between the samples.

Case #2: Saginaw, MI Homicide

A 20 year-old resident of Saginaw, Michigan, was accused of breaking and
entering the home of elderly man in December of 2003, killing the homeowner during the
suspected robbery and then stealing the man’s vehicle. Three samples were submitted for
comparison: one known glass fragment obtained from the south window of the home, one
questioned glass fragment found on the front porch of the home, and a second questioned

fragment removed from the handle of a hammer that was suspected to have been used to

34

0800000000000 000 00000 00005000 800 .00 0000 00000000 00000

000 00000000 00080000 000000 .00 000000000000 00000 0000 00000000 003 0000000 000m
000000 00000 0000 00000 08000300 0>000 00000 0000 000000000 0000 00 00085000 03050
000 0000000000 0000 0000 00000000 00 0000 000009000 0000 .00 000 w0030000 0000 .000 000M000

.N 0m

 

 

0000000m N00 .0.m02 4
5350 0

03050 0

03050 D
0000000000 +

 

 

 

mm

gain entry into the home through the boarded south window. Again, all three fragments
and NIST 612 were ablated three times each demonstrate reproducibility, especially on
the fragment from the hammer handle, which was irregularly shaped and less than 1 mm
in its largest dimension. In this case, one of the questioned fragments could be
distinguished from the known fragment (Q Hammer) while the other (Q Front Porch)
could not (Figure 11). -It was concluded that the questioned fragment from the hammer
handle did not originate from the same source as the known fragment, while the fragment

from the front porch could have come from that source.

Case #3 .° Dundee, M1 Breaking and Entering

A 19 year-old resident of Dundee Village, Michigan, was accused of breaking a
storefront window to gain access into a local convenience store. Two samples were
submitted for analysis: one known glass fragment taken from the window frame at the
point of entry and one questioned glass fragment taken from the toe area of the suspect’s
shoe. As in the previous examples, the questioned fi'agment, known fragment, and NIST
612 were ablated three times each. In this case, the questioned and known fragments
were easily discriminated (Figure 12) and it was concluded the fragments did not share a

common origin.

CONCLUSIONS AND FUTURE WORK
With the benefits of efficient analyses and essentially no sample preparation,
LA-ICP-MS is an ideally suited for the discrimination of glass fragments. As an

additional step in traditional analytical schemes involving measurement of refractive

36

000000 00000 00000000 003 000800 0000 .000w< 000009000

030 08000 800 00 000000 00000 05 0000.0 w000000 00 00000000 000000 00 0000000 000000000
800 0000.0 00000w000 00000000000 05 00000 802 .000000 8000 0000 0000.0 08000w000 300
00000 300003 000000 800 0000.0 00000900 030000 0000 000 0800 0000.0 800 0000.0 00000w000
0000000000 0000 0000 00000000 00 0000 00000w000 800 .00 000 w003000 0000 .00 000»:—

> 00

 

 

   

0000005 N00 0.22 4
>0600000, 500m 0 0
000000 00000 O D
58800 0 +

 

 

 

0N

37

.0000: 00000 0003000

003 030000 0000 .00000 0003000 00 0< .000000 00000 05 0000.0 0005300 000 000 300005
0000 00000002000 05 0000.0 00000m00m 0305. 000 000 0000 0.0000000 05 0000.0 0000030
000000000 000 0000 000—0000 00 0000 00000w0€ 05 .00 000 303000 003 .N— 000m:—

5 mm

 

38

 

0:00:00 N00 0002 4
0300M O
8000 o D

 

 

 

0N

index, trace elemental analyses can help reduce Type II errors (or false inclusions) and
improve the associative value of comparisons when samples cannot be discriminated.
The range of refractive index values has become increasingly small in float glasses since
about the early 1960’s (Koons and Buscaglia, 2001, and references therein). However,
the class size is significantly smaller for trace element composition of glasses, as
demonstrated by the shrinking number of indistinguishable pairs of unknowns for a given
sample suite when using chemical composition in addition to refractive index (Trejos et
al., 2003; Duckworth et al., 2002).

There are several goals for maximizing the discriminatory and associative
power of LA-ICP-MS glass comparisons by this method. The first is the general aim of
all LA applications: improve the efficiency of sample ablation and delivery of the sample
to the plasma. It has been demonstrated that newer laser ablation systems can improve
laser-sample coupling, and therefore analytical precision (e.g. CETAC LSX-200® vs.
LSX-500®, Trejos and Almirall, 2004). Variables such as crater shape, carrier gas types,
mixture ratios, and flow rates have also been studied to improve performance (Russo et
al, 2002, and references therein). As forensic glass samples are ofien microscopic, these
parameters must often be optimized for the smallest spot size possible to obtain the most
information from the sample.

For relatively short ablation times, the precision of an analysis decreases with a
smaller amount of sample delivered to the ICP-MS (i.e. smaller spot size). Similarly, for
the same amount of material, precision will decrease with a longer list of elements in the
menu, as less time is spent scanning for any given mass-to-charge ratio (m/z). Therefore,

the selection of discriminatory elements is particularly important for comparisons.

39

Maximizing the number of abundant elements in glasses while ensuring adequate dwell
time for each element is a second goal for improving the technique. For example, the
menu of six elements in the present work (”Sr, 89Y, 90Zr, 138Ba, 139La, and 140Ce) is
currently being modified to include three additional, highly discriminatory isot0pes (46Ti,
55Mn, me, 208Pb) based partly on the work of Trejos and Almirall (2004) and Tejos et
al. (2003).

In addition to demonstrating the wide range of trace element ratios observed in
various float glass fragments, this study demonstrates the usefulness of LA-ICP-MS
analysis in routine casework. Glass fragments have been shown to be homogeneous with
respect to trace element composition, and within the precision of the technique.
Although the cost of operation and maintenance has been cited as a drawback for LA-
ICP-MS, the lack of sample preparation and short analysis time (even with replicates)
compensates for the expense. The method is designed to be an efficient comparative tool
for the forensic analyst, and by no means are the data used to quantitatively compare
samples using robust statistical tests. The graphical technique of ratio comparisons is

exactly that: a demonstrably precise semi-quantitative comparative tool.

40

REFERENCES

Aeschliman, D.B., Bajic, S.J., Baldwin, DR, and Houk, R.S., Multivariate pattern
matching of trace elements in solids by laser ablation inductively coupled plasma-

mass spectrometry: source attribution and preliminary diagnosis of fractionation,
Analytical Chemistry, 2004, 76: 3119-3125.

Almirall, J ., Elemental analysis of glass fragments, in Forensic Examination of Glass and
Paint: Analysis and Interpretation, Caddy, B., ed., Taylor and Francis, 2001: 65-83.

Becker, 8., Chemometric classification of float glass and new developments in the use of
laser ablation-ICP-MS in the field of glass analysis, Abstracts of the Second
European Academy of Forensic Sciences Meeting, 2000.

Curran, J.M., Triggs, C.M., Almirall, J.R., Buckleton, J.S., and Walsh, K., The
interpretation of elemental composition measurements from forensic glass evidence:
1, Science & Justice, 1997a, 241-244.

Curran, J .M., Triggs, C.M., Almirall, J.R., Buckleton, J .S., and Walsh, K., The
interpretation of elemental composition measurements from forensic glass evidence:
11, Science & Justice, 1997b, 245-249.

Duckworth, D.C., Morton, S.J., Bayne, C.K., Montero, S., Koons, RD, and Almirall,
J .R., Forensic glass analysis by ICP-MS: A multi-element assessment of
discriminating power via analysis of variance and pair-wise comparisons, Journal of
Analytical and Atomic Spectrometry, 2002, 17: 662-668.

Duckworth, D.C., Bayne, C.K., Morton, S., and Almirall, J .R., Analysis of variance in
forensic glass analysis by ICP-MS: variance within the method, Journal of Analytical
and Atomic Spectrometry, 2000, 15: 821-828.

Durrant, S.F., Laser ablation inductively coupled plasma mass spectrometry:
achievements, problems, prospects, Journal of Analytical and Atomic Spectrometry,
1999, 14: 1385-1403.

Hickman, D.A., Glass types identified by chemical analysis, Forensic Science
International, 1987, 33: 23-46.

Koons, RD, and Buscaglia, J ., Interpretation of glass composition measurements: The
effects of match criteria on discrimination capability, Journal of Forensic Sciences,
2002, 47: 505-512.

Koons, RD, and Buscaglia, J ., Distribution of refractive index values in sheet glasses,
Forensic Science Communications, 2001, 3.

41

Montero, S., Hobbs, A.L., French, T.A., and Almirall, J .R., Elemental analysis of glass
fragments by ICP-MS as evidence of association: analysis of a case, Journal of
Forensic Sciences, 2003, 48: 1-7.

Parouchais, T., Warner, I.M., Palmer, L.T., and Kobus, H., The analysis of small glass
fragments using inductively coupled plasma mass spectrometry, Journal of Forensic
Sciences, 1996, 41: 351-360.

Russo, R.E., Mao, X., Liu, H., Gonzales, J., and Mao, S.S., Laser ablation in analytical
chemistry — a review, Talanta, 2002, 57: 425-451.

Suzuki, Y., Sugita, R., Suzuki, 8., and Maruma, Y., Forensic discrimination of bottle
glass by refractive index measurement and analysis of trace elements with ICP-MS,
Analytical Sciences, 2000, 16: 1195-1198.

Szymanski, D., Patino, L., Bommarito, C., and Siegel, J., Trace element profiles of float
glass fragments determined by laser ablation inductively coupled plasma mass
spectrometry (LA-ICP-MS), Abstracts of the 56"I Annual Meeting of the American
Academy of Forensic Sciences, 2004, Dallas, TX.

Szymanski, D., Patino, L., Bommarito, C., and Siegel, J., Use of laser ablation
inductively coupled plasma mass spectrometry (LA-ICP-MS) for float glass fragment
discrimination by elemental analysis: preliminary results, Abstracts of the 32’”
Annual Meeting of Midwestern Association of Forensic Scientists, 2003, Columbus,
OH.

Trejos, T. and Almirall, J .R., Effect of fractionation on the forensic elemental analysis of
glass using laser ablation inductively coupled plasma mass spectrometry, Analytical
Chemistry, 2004, 76: 1236-1242.

Trejos, T., Montero, S., Almirall, J.R., Analysis and comparison of glass fragments by
laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS) and ICP-
MS, Analytical and Bioanalytical Chemistry, 2003, 376:1255-1264.

Watling, R.J., Novel application of laser ablation inductively coupled plasma mass
spectrometry in forensic science and forensic archaeology, Spectroscopy, 1999, 14:
16-34.

Watling, R.J., Lynch, BR, and Herring, D., Use of laser ablation inductively coupled
plasma mass spectrometry for fingerprinting crime scene evidence, Journal of
Analytical and Atomic Spectrometry, 1997, 12: 195-203.

42

     

     
 

IIIIIIII

llllllllllllllllill“will

3 1293 O 504 030

  

h~._ _

/ ‘I—l.f