llWlHHWIHHWHIIIHI!WWW!“WM

     

 

374
.THS_

  

3 o7 . '-

LIBRARY
Michigan State 3
University

 

This is to certify that the
thesis entitled

FORENSIC ANALYSIS OF COSMETIC FACE POWDERS

presented by
KELLY GREENOUGH

has been accepted towards fulfillment
of the requirements for the

MS. degree in Forensic Science

 

 

A. afi pfid

 

 

,— Majom Signature

4/27/07

 

Date

MSU is an affinnative-action, equal-opportunity employer

 

—.—.-.----n-o--—u--._ «

_ —--.-.---u--o-u-n--

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

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

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

FORENSIC ANALYSIS OF COSMETIC FACE POWDERS
By

Kelly Greenough

A THESIS

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

MASTER OF SCIENCE
Department of Criminal Justice

2007

ABSTRACT
FORENSIC ANALYSIS OF COSMETIC FACE POWDERS
By
Kelly Greenough

Cosmetic face powders are widely used and a large amount is applied to the facial
region, thus making it possible to be transferred during the course of a crime. Therefore,
powders have potential to be used as forensic evidence. In such a case, it may be useful to
determine if two powder samples originated from a common source. The purpose of this
project is to determine if various samples of cosmetic face powder can be differentiated
from one another.

In this work, three analytical techniques were selected to differentiate between
pairs of cosmetic face powder samples. The chosen techniques included scanning
electron microscopy-energy dispersive spectroscopy (SEM-EDS), attenuated total
reflectance-Fourier transform infrared spectroscopy (ATR-FTIR), and matrix-assisted
laser desorption ionization mass spectrometry (MALDI-MS). Each method analyzes
different components of the face powders, and the data are thus complementary.

Twenty-seven samples were analyzed in this project, and each of the possible
pairs of samples was compared in order to evaluate the discrimination ability of each
technique. The individual techniques and the combined techniques achieve a high level of
discrimination between face powder samples, displaying the usefulness of cosmetic face

powders in forensic investigations.

ACKNOWLEDGEMENTS

I would like to thank several people for their help and support during the time this
project was taking place. First, I would like to thank my advisor, Dr. Ruth Waddell for
her tremendous support and assistance in this project. She gave up much of her own time
working out difficulties with instruments, offered many helpful suggestions to make the
project develop, and guided me through the thesis writing process. I would also like to
thank Dr. Christina DeJong of the School of Criminal Justice for serving on my
committee and being available to answer questions along the way.

I cannot thank Dr. Jay Siege] enough for providing the idea for the project,
continuing to play an active role in my pursuit of the degree, and accepting me into the
Forensic Science program at Michigan State University. Most importantly, I extend my
extreme gratitude to him for being both a friend and mentor to me.

There were several individuals who assisted me in various aspects of the project. I
would like to thank Dr. Dan Jones and Bev Chamberlain of the Mass Spectrometry
Facility who took the time to help me understand the MALDI-MS technique, and also
how to operate the instrument. In addition, Dr. Ewa Danielewicz of the Center for
Advanced Microscopy answered several questions I had regarding SEM-EDS, and I am
extremely grateful for her assistance. I would also like to thank Monique Lapinski for
allowing me to use her SEM beam time.

Throughout the course of this project many attempts at other techniques were

made, and due to instrumental errors, were unable to be used. However, Dr. Kathy

iii

Severin spent a tremendous amount of time with me working on the various instruments,
and I would like to thank her for giving up her own time to help me.

I would like to thank friends and family for their continued support throughout
my years at MSU. I would especially like to say thank you to my mother, Susie Patching,
who not only purchased many of the powder samples for me, but was also there for me
whenever I needed her.

Most importantly, I would like to thank my husband Dan Greenough for being my

best friend and biggest fan.

iv

TABLE OF CONTENTS

LIST OF TABLES .................................................................................. vii
LISTOFFIGURES...................... .......................................................... viii
CHAPTER 1. INTRODUCTION
1.1 Cosmetics ................................................................................. 1
1.2 Cosmetic Face Powders ................................................................ 2
1.3 Forensic Analysis of Cosmetics ....................................................... 3
1.4 Research Objectives ..................................................................... 8

CHAPTER 2. INSTRUMENTAL THEORY
2.1 Scanning Electron Microscopy/Energy Dispersive

Spectroscopy (SEM-EDS) ............................................................. 9
2.1.1 Introduction ................................................................... 9

2.1.2 Scanning Electron Microscopy ............................................. 9

2.1.3 X-ray Production ............................................................ 12

2.1.4 X-ray Detection ............................................................. 14

2.1.5 SEM/EDS .................................................................... 16

2.2 Fourier Transform Infrared Spectroscopy .......................................... 17
2.2.1 Introduction .................................................................. 17

2.2.2 Molecular Interactions with IR Radiation ............................... 17

2.2.3 Fourier Transform Infrared Spectroscopy .............................. 18

2.2.4 Attenuated Total Reflection ............................................... 20

2.3 Matrix-Assisted Laser Desorption Ionization-Mass Spectrometry .............. 23
2.3.1 Introduction .................................................................. 23

2.3.2 MALDI ...................................................................... 23

2.3.3 Interpretation of Mass Spectra ............................................ 26
CHAPTER 3. METHODS AND MATERIALS ................................................ 28
3.1 SEM-EDS .............................................................................. 30
3.2 FTIR Spectroscopy .................................................................... 30
3.3 MALDI-MS ............................................................................ 31
CHAPTER 4. RESULTS AND DISCUSSION ................................................ 34
4.1 SEM-EDS .............................................................................. 34
4.1.1 Sample Differentiation Based on Unique Elements .................. 36

4.1.2 Sample Differentiation Based on Common Elements ................. 37

4.1.2.1 Cluster Analysis ................................................. 37

4.2 FTIR Spectroscopy .................................................................... 40
4.3 MALDI-MS ............................................................................ 45
4.4 Blind Samples .......................................................................... 53
4.5 Combined Techniques ................................................................ 54

5. CONCLUSIONS AND FUTURE RESEARCH ............................................. 57

5.1 Conclusions ............................................................................ 57

5.2 Future Research ........................................................................ 59
APPENDICES

Appendix A. Abbreviated Table of Group Frequencies for Organic Groups ...... 63

Appendix B. Listed Ingredients in Each F ace Powder Sample ...................... 64
REFERENCES ...................................................................................... 66

vi

LIST OF TABLES

Table 3.1.1 Face powder samples ................................................................. 28
Table 4.1.1 Samples that could be identified based on

unique characteristics (SEM-EDS) ................................................. 37
Table 4.1.2 Resulting clusters with 90% similarity (SEM-EDS) ............................. 40
Table 4.2.1 Nine groups based on presence or absence of major peaks (FTIR) ............ 43
Table 4.3.1 Initial grouping of MALDI-MS data ............................................... 49
Table 4.3.2 Classification of group D based on presence of m/z 320 ........................ 50

Table 4.4.1 Pairs of samples that could not be distinguished by SEM—EDS, FTIR, or
MALDI-MS ............................................................................ 54

vii

LIST OF FIGURES

Figure 2.1.1 - Diagram of SEM ................................................................... 10
Figure 2.1.2 - Inelastic scatter resulting in an x-ray ............................................ 12

Figure 2.1.3 - An electron from the N shell fills a vacancy in L shell, resulting in

the emission of an Lp x-ray ......................................................... 14
Figure 2.1.4 - The EDS detector ................................................................... 15
Figure 2.1.5 - Example of an EDS x-ray spectrum .............................................. 16
Figure 2.2.1 - A Michelson interferometer ....................................................... 19
Figure 2.2.2 - Diagram of a single-bounce ATR accessory .................................... 21
Figure 2.3.1 - The principle of MALDI ........................................................... 24
Figure 2.3.2 - Time-of-flight mass analyzer ..................................................... 25
Figure 2.3.3 - Example of a mass spectrum ...................................................... 26
Figure 4.1.1 - EDS spectrum of sample 6 ........................................................ 35
Figure 4.1.2 - Resulting dendrogram from cluster analysis of atomic
percentages at a similarity level of 90% .................................................... 39
Figure 4.2.1 - IR spectrum of talc ................................................................. 41
Figure 4.2.2 - IR spectrum of sample ............................................................ 41
Figure 4.2.3 - Classification of samples based on presence and absence
of major peaks in IR spectra ..................................................... 44
Figure 4.3.1 - Mass spectrum of sample 13 ...................................................... 45

Figure 4.3.2 - Scheme for differentiating powder samples in Group A based
on MALDI-MS data ............................................................... 47

Figure 4.3.3 - Scheme for differentiating powder samples in Group B based
on MALDI-MS data ............................................................... 48

Figure 4.3.4.- Scheme for differentiating powder samples in Group C based
on MALDI-MS data ............................................................... 49

viii

Figure 4.3.5 - Mass spectrum of sample 6 ...................................................... 50

Figure 4.3.6 - Scheme for differentiating powder samples in Group D, based
on MALDI-MS data ............................................................... 51

Figure 4.3.7 - Scheme for differentiating powder samples in Group Du based
on MALDI-MS data. . . ._ ........................................................... 52

ix

1. INTRODUCTION

1.1 Cosmetics

Ancient Egyptians used cosmetics as early as 4000 BC, a trend that was continued
throughout history by several differentcivilizations. Ancient Greeks and Romans applied
makeup in order to attract attention and to signify class standing. In our society today,
cosmetics continue to be used, applied to the eye area, fingernails, lips, or overall face,
typically to increase attractiveness.

The term “cosmetic” is defined by the Food and Drug Administration (FDA) as:

(1) articles intended to be rubbed, poured, sprinkled, or sprayed on,
introduced into, or otherwise applied to the human body or any part thereof for
cleansing, beautifying, promoting attractiveness, or altering the appearance, and

(2) articles intended for use as a component of any such articles; except
that such term shall not include soap. '

Cosmetics that are blended into the skin to even out skin tone have persisted
throughout history. Currently, a variety of cosmetics can help achieve the look of
flawless face, including foundations, concealers, and powders. Foundations are typically
liquid based, formulated to match skin tone, and are rubbed into the entire facial region to
reduce uneven skin tone. Concealers are also liquid based and skin colored, but are
generally used only to cover blemishes, rather than the whole face. Powders are also
matched to skin color, but their main purpose is not to cover redness or blemishes, but to
decrease sheen that may occur due to oils in the skin. As with foundations, powders are
also generally applied over the entire face, although are more easily reapplied throughout

the day.

Today, there are thousands of foundations, concealers, and powders produced by
different manufacturers and available in numerous shades to match the natural skin color.
Due to their widespread use in our culture, these foundations and face powders can easily
be transferred onto cloth or skin during the course of a crime and hence are an important
type of trace evidence. The aim of this research is to characterize a number of
commercially available face powders using a variety of analytical techniques, in order to
assess the potential to differentiate powders according to manufacturer and brand.

1.2 Cosmetic Face Powders

Cosmetics may contain hundreds of different components, and federal regulations
require ingredients to be listed on product labels in descending order by quantity.l
However, the federal government relies heavily on industry self-regulation.l
Unfortunately, the exact formulation is often proprietary and therefore ingredients within
a cosmetic may not be listed on the product label. The label might contain only a partial
list of ingredients, or a special listing of what “may” be contained in a product. Fatty
acids, waxes, and oils are typically present in cosmetics. Oils present include lanolin,
aloe, castor oils”, while waxes commonly include beeswax, paraffins, and candelilla.2’3
The color of cosmetics is provided by pigments and dyes such as D&C Red dyes,
dihydroxyacetone, titanium dioxide, mica, and iron oxidesl.

While liquid based makeup products, such as the creams and lotions mentioned
previously, contain an abundance of oils and waxes, cosmetic face powder has a different
formulation which results in a different texture. The most common ingredient in face
powders is talc (Mg33Si40m(OH)2), which is not typically found in liquid products. In

addition, powder lacks many of the oily components and preservatives found in

foundations. Other ingredients commonly found in powder samples include: titanium
dioxide, zinc stearate, parabens, octyldodecyl stearoyl stearate, palmitate, aloe extracts,
mica, calcium carbonate, silica, dimethicone, and mineral oil. Titanium dioxide, mica,
and calcium carbonate are used as opacifying agents, for their covering power, brilliance,
and reflectivity.4 Parabens are a group of preservatives added to cosmetics for an
antimicrobial effect, while dimethicone, or polydimethylsiloxane (PDMS) is a silicon-
based organic polymer that gives a smooth feel to a cream, lotion, or powder.1
Octyldodecyl stearoyl stearate is an ester that functions as a skin-conditioning agent, and
a viscosity increasing agents The listed ingredients are found in nearly all powders, but
each powder contains many more compounds, some that are listed on the product label,
and others that are not. In addition, there is typically no listing of the proportion of each
ingredient contained within that powder. Therefore, the exact composition of cosmetic
face powders is not easily determinable from the product labeling; in fact, the exact
composition is often known only by the manufacturer.
1.3 Forensic Analysis of Cosmetics

Several studies in the forensic science field have focused on the evidentiary value
of cosmetics. To date, the primary focus has been on lipsticks, eye shadows, and nail
polish, however, studies involving mascaras, sunscreens, and foundations have also been
conducted. 2’6'” These studies have used a variety of instrumental techniques to either
differentiate or categorize cosmetics based on differences in their chemical composition.
Choudhry and colleagues examined lipstick by microspectrophotometry (MSP) in the
visible region and scanning electron microscopy-energy dispersive spectroscopy (SEM-

EDS). Both techniques offered a rapid means of sample discrimination.7 MSP was able to

differentiate lipstick samples of similar color but with minor shade differences that were
invisible to the naked eye. SEM-EDS provided information regarding the elemental
composition of the samples, and provided images of pigment particle morphology.7 Both
the elemental compositions and particle morphology of lipsticks provided discrimination
ability. Ehara and Marumo performed a large study of 174 lipstick samples using purge-
and-trap gas chromatography (P&T-GC) to analyze lipstick smears and identify them
based on differences in oil composition.12 Each sample was first observed under white
light, and those samples that visually could not be separated based on color were
classified in the same group.” The fluorescence of each sample was next observed using
light at wavelengths of 350, 445, and 515 nm, P&T-GC.12 The measurement of
fluorescence at these wavelengths enabled discrimination of dyes within the lipstick
samples. P&T-GC was used as a direct analytical technique for lipstick smears on filter
paper, and the results were based on chromatographic peaks characteristic of the oils in
each sample.12 Using this technique, 174 samples could be classified into 48 groups.12
The lipsticks that originated from the same manufacturer often produced chromatograms
that could not be distinguished from one another.12 Combining this technique with
fluorescence and visual observation, the researchers were able to discriminate 15,038
lipstick pairs of samples out of 15,051, which corresponds to 99.9% of samples
analyzed.12

Thin-layer chromatography (TLC) is a simple and economic technique that has
proven to be useful when studying cosmetics, particularly lipstick, due to its ability to
separate dye components.”ls Singh and Jasuja analyzed 15 liquid lipstick samples of

different brands applied to a variety of substrates such as glass, filter paper, tissue paper,

and skin.14 The dry samples were lifted from the surfaces with cotton swabs moistened
with alcohol, and the stains were then dissolved in toluene before being spotted on a TLC
plate.M The authors were able to develop a solvent system that allowed separation of dye
components and differentiation of the lipstick samples.l4 TLC is a valuable screening
technique, simple and rapid, yet it has limitations in its selectivity, and thus cannot be
used as a confirmatory test.

Several thesis projects performed at Michigan State University have characterized
different cosmetic types for forensic purposes.2’6’lo Mascaras were analyzed using
microscopy, Fourier transform infrared spectroscopy (FTIR), SEM-EDS, and MSP to
determine an appropriate protocol for analysis of such samples.6 Microscopy grouped the
24 mascara samples into 4 groups based on particle colors, and presence or absence of
fibers that are sometimes added to mascara to increase the length of eyelashes when
applied.6 However, visual observation and determination of color is subjective, and fiber
presence may be due to transfer. It was therefore concluded that microscopy was useful
as a preliminary screening tool.6 Microscopy was then followed by FTIR spectroscopy,
which can identify functional groups in samples and allow for the comparison of samples
based on chemical composition.6 FTIR was able to distinguish 21 of the 24 samples from
each other based on the presence of major peaks in the IR spectra.6 The third technique,
SEM-EDS, allowed differentiation of samples based on elemental concentrations.6 Based
on the comparison of elemental weight percentage of the samples, SEM-EDS could
distinguish 99.67% of samplesf’ The last technique, MSP, did not provide any

discriminatory data for the analyzed samples, due to the black color of mascara.6

In addition to the mascara research, a project to determine an appropriate protocol
for analysis of nail polish was also performed.10 Romero studied 20 nail polishes varying
in color and manufacturer using microscopy, MSP, and FTIR spectroscopy.10 Microscopy
was able to distinguish between each of the nail polishes, with the exception of five.
These five were from the same manufacturer, but different lots.10 Three of the five
samples could not be distinguished by MSP, but were distinguished by microscopy. '0
When compared to a IR spectral library of lipstick samples, the correct lipstick was
identified 96% of the time.10 This study determined that, in order to determine a match
between a questioned and known sample, microscopy should be the first step in analysis,
followed by MSP, then reflectance F TIR for samples where quantity is limited.10 This
combination of techniques allows more confidence in discriminating samples compared
to a single technique which, as shown in this study, has limited discriminating power.10

A thesis project performed by Stark combined the techniques of MSP, laser
desorption ionization mass-spectrometry (LDI-MS), and FTIR spectroscopy to classify
5 7 lipsticks.6 MSP grouped the lipsticks into 5 spectral classes based on their wavelength
of maximum absorbance. IR spectra of the lipstick samples were compiled to create a
library of spectra, and then the spectra were re-collected, and compared to the library. For
this method, the top ten possible matches were listed.2 Forty-eight lipsticks yielded
accurate search results, where the correct lipstick was the number one match, while the
other samples matches were all in the top ten.2 Mass spectra of the lipstick samples were
obtained by LDI-MS and classification was performed based on presence or absence, as
well as the identity, of predominant peaks.2 The initial assessment grouped the samples

into five general groups, then further classification into smaller sets was performed

within each of the groups. Within each group, distinctions between most of the lipsticks
could be made based on additional peaks present within the spectra.2 Some of the peaks
found in the mass spectra could be identified as components of castor oil, a common
ingredient in lipsticks, and known dyes such as D&C Red 27 .2 Other peaks within the
spectra could not be identified as known ingredients.

While the characterization of various cosmetics has been published in the
literature, few studies deal with cosmetics that have the specific function to even out skin
tone, such as foundations, concealers, and powders. Gas chromatography-mass
spectrometry (GC-MS) was used in an early study of castor oil-based cosmetic powders,
creams, and lotionsg, and a recent study examined the ability to discriminate between
different types of foundations using a combination of methods.8 A study performed by
Gordon and Coulson in 2004 focused specifically on liquid foundations. Foundations
were smeared on three different types of fabric, and the smears were analyzed by FTIR,
GC with flame ionization detection (FID), and SEM-ED85. A total of 53 samples were
considered. Twenty-three pairs out of the 1378 possible pairs could not be distinguished
with F TIR, yielding a discrimination power of 98.3%.8 GC-FID alone yielded a
discrimination power of 82%, or 1130 pairs distinguished out of 1378‘, and SEM-EDS
discriminated all but 83 pairs out of the possible 1378, for a discrimination power of
94%.8 The three techniques when combined, resulted in a discrimination power of

99.7%.8

1.4 Research Objectives

The proposed research aims to firstly characterize cosmetic face powders using
SEM-EDS, FTIR, and matrix-assisted laser desorption ionization mass spectrometry
(MALDI-MS) and then assess the discrimination ability of each technique in
differentiating the face powders. Currently, no studies are available that analyze and
characterize face powders and thus it is believed that this work will contribute to the
significance of cosmetics as forensic evidence.

A total of 27 face powders were examined by SEM/EDS, F TIR, and MALDI MS.
Each of these methods provide complementary information about cosmetic samples.
SEM-EDS can produce images so the morphology of samples can be compared, and can
provide an elemental profile of the samples, yielding the atomic percentage of each of the
various elements within the samples. These atomic percentages can be statistically
compared among samples for comparison. FTIR spectroscopy can provide information
about the chemical functional groups present in the samples which can be used to identify
components present in the samples. The presence or absence of major peaks in the
spectra can be used to visually assess the similarities and differences between samples.
MALDI-MS can provide information about qualitative and quantitative composition of
both organic and inorganic components in the face powders, and can yield structural
information of the species within and each sample.16 The ability of the techniques to
discriminate the face powders will be investigated, individually and in various
combinations. By characterizing the chemical composition of the face powders using a
variety of techniques, successful discrimination of samples may be possible increasing

the significance‘of such evidence in forensic casework.

CHAPTER 2 INSTRUMENTAL THEORY
2.1 Scanning Electron Microscopy/Energy Dispersive Spectroscopy (SEM-EDS)
2.1.1 Introduction
Scanning electron microscopy-energy dispersive spectroscopy has proven to be a useful
tool in the field of forensic science, as it can provide both qualitative and quantitative
information of the inorganic components of samples.17 The scanning electron microscope
(SEM) in particular is often found in forensic laboratories and provides powerful
magnification (10X to 100,000X), a large depth of field, and high resolution when
compared to a light microscope.17 In addition, when combined with energy dispersive
spectroscopy (EDS), SEM has the capability to identify and quantify elements within a
sample.
2.1.2 Scanning Electron Microscopy

The magnification and imaging ability of the SEM allows the morphology of
samples to be visually examined, allowing sample differentiation based on morphology.

Samples must be prepared for the SEM so that they contain no volatiles, are
firmly mounted, and are electrically conductive. Samples are mounted in aluminum stubs,
attached by either carbon cement, epoxy glue, or adhesive tabs, then sputter coated,
typically with gold or carbon to make the samples conductive. Once conductive, the
samples are placed into the instrument chamber.

An SEM is composed of the following basic parts”: the electron gun, condenser
lens, scan coils, objective lens, final aperture, scintillator, detector, computer, and cathode

ray tube (Figure 2.1.1)

 

 

 

I U I ] Electron gun

Condenser lens

   
  
 

 

  

Scan coils

\
Objective lens \
Aperture\

\ -

Scintillator

  

Scan
generator

 

 
 
 

 

Computer

  

Cathode-ray
tube

— Preamplifier

 

  

 

\
—_

 
 

Sample

 

 

 

Figure 2.1.1. Diagram of SEM

The electron gun produces a beam of electrons that is condensed by the condenser lens.
The scan coils within the objective lens create a magnetic field that deflects the electrons
towards the sample. The electrons are focused and the electron beam then strikes the
sample.

When the electron beam interacts with the sample, a series of interactions occur,
producing secondary electrons (SE). When the beam interacts with the orbital shells of
atoms in the sample, the electron within the beam are scattered. When this occurs, the
path of the beam is deflected by a small angle, resulting in energy loss. This type of
interaction is called inelastic scatter and results in a low energy (3-5 eV) electron being

ejected from the atomic orbital. Any electron detected as a result of sample-beam

10

interactions that has energy of less than 50 eV is a secondary electron. Most secondary
electrons do not have enough energy to escape from the sample, unless they are located
near the surface of the sample, within approximately 50 nm.

Common detectors for SEM include Everhart-Thornley (ET) detectors which are
composed of a Faraday cage, scintillator, and photomultiplier tube (PMT). Secondary
electrons emitted from the sample are attracted to the Faraday cage, which is maintained
at ~ +300 V. Behind the Faraday cage is the scintillator, which typically has a thin
coating of conductor material to maintain the positive voltage.l8 Electrons are accelerated
to a +12000 V charge on the scintillator and, upon striking the scintillator, photons of
light are produced that are then directed into the PMT. Inside the PMT, the photons’
interactions with electrodes causes an amplified emission of electrons that are sent to the
cathode ray tube to produce 3D images of the sample.

The images obtained by SEM result from the differences in amounts of secondary
electrons from flat portions of the sample versus projections on the sample. The flat
regions are seen as dark images on the cathode-ray tube, while projections yield brighter
spots. This is a direct result of secondary electrons’ ability to escape the sample surface.
When topographical features are present on the sample, a greater volume is close to the
surface. In such an area, the secondary electrons are close to the surface, and can escape,
thus creating a brighter image.

Although secondary electrons account for the majority of interactions that
produce the SEM image, other types of interactions occur between the sample and the
electron beam. Elastic scatter, which can occur between incident electrons and the

nucleus of atoms in the sample,17 is characterized by a large-angle deflection of the

11

incident electron beam, as well as little energy loss by the incident electrons.l7 Some
elastic scatter produces backscattered electrons (B SE), which are those electrons
eventually scattered at ~180° to the incident electrons. Unlike secondary electrons,
backscattered electrons have energies on the order of kV, similar to the energy of the
incident beam, allowing escape from much greater depths within the sample. They can
provide an atomic number image as well as a depth image.
2.1.3 X-ray Production

Along with secondary and backscattered electrons, interaction of the electron
beam with the sample can also result in the emission of X-rays, as electrons from higher
energy shells fall to fill the vacancy created by the initial interaction (Figure 2.1.2). Due
to the quantized nature of atomic energy shells, the energy of the x-ray emitted is
characteristic of the atom and related by Equation 2.1, where A is the wavelength in
nanometers, and E is the energy in keV.”

zl=-l-%9—8 (eqn. 2.1)

incident electron beam

 

Figure 2.1.2 Inelastic scatter resulting in an x-ray.

12

With the SEM coupled to an energy dispersive spectrometer (EDS), the x-rays
emitted following interaction of the electron beam with the sample are detected. Since the
energy of the emitted x-ray is characteristic of the atom, elemental composition of the
sample can be determined.

The critical excitation energy, also known as the absorption edge, is the beam
energy needed to eject an electron from an atomic shell. With an electron ejected, a
vacancy now exists in the shell, and the atom is unstable. To restore stability, an electron
from a higher shell falls to fill the vacancy, emitting an x-ray as it does so. Each type of
X-ray is called a line, and if sufficient x-rays of a given line are generated, they produce
an x-ray peak in the spectrum. If the critical excitation energy is not sufficient to remove
an electron from a given shell, then no x-ray lines are seen.17 The difference in critical
excitation energies for different elements gives rise to differences in the analytical spatial
resolution when analyzing for specific elements.'7

X-rays are named according to the atomic shell where the vacancy exists, and the
number of jumps made by the electron to fill that vacancy. The atomic shells are named
K, L, M, and N, where K is the inner shell, closest to the nucleus, followed by L, M, and
N.

If the electron falls one shell to fill the vacancy, the emitted x-ray is called an a x-
ray, falling two shells results in a B x-ray being emitted, and falling three shells to fill the
vacancy would be termed a y x-ray. For example, if there was a vacancy in the L shell of
an atom and an electron from the N shell falls to fill the void, then the electron has fallen
two shells. Since the vacancy existed in the L shell and involved an electron falling two

shells to fill the Vacancy, the emitted x-ray is an LB x-ray (Figure 2.1.3).

13

 

 

Figure 2.1.3 An electron from the N shell fills a vacancy in the L shell, resulting in
the emission of an LB x-ray.
2.1.4 X-ray Detection

The energy of the emitted x-rays can be detected via energy dispersive
spectroscopy (EDS). Since the x-ray energies are characteristic of the elements present,
EDS allows elemental identification.

The EDS detector is designed to convert the energy of each individual x-ray
emitted into a voltage signal of a proportional size.’9 The EDS detector (Figure 2.1.4)
consists of the following main components: collimator, crystal, field-effect transistor

(FET), pulse processor, analog-to-digital converter, multichannel analyzer.

14

 

 
  
    
 

 

 
 
 

 

 

 

  

I:
stray X-ray
BSE electron
beam
window
sample
\ I
collimator \
-ray from
field-effect "5's“!
transistor sample

 

 

 

Figure 2.1.4 The EDS detector

The carbon-lined collimator contains internally mounted magnets that absorb stray
radiation. Following the collimator is a thin membrane window that serves as a barrier
and prevents contamination of the detector crystal, which is composed of silicon and
lithium.'7 The crystal, which has a 1000 V bias, converts each x-ray into a charge by the
ionization of atoms in the semiconductor crystal.19 The charge is then converted into a
voltage signal by the FET pre-amplifier.19 The voltage signal is sent to a pulse processor
for measurement.17 The number of pulses produced corresponds to the energy of the
incoming x-ray. The analog-to-digital converter takes each pulse from the pulse processor
and converts it into a series of pulses of equal height.17 The number of pulses corresponds
to the height of the original pulse and to the energy of the x-ray.17 The multichannel
analyzer places the pulses into channels of varying energy which, through the computer
and cathode ray tube, displays the energy spectrum. The resulting spectrum is plotted as

intensity or counts on the y-axis and x-ray energy (in keV) on the x-axis (Figure 2.1.5).

15

 

 

 

 

 

 

 

 

 

 

4000« Si
3000
‘3
82000 Mg "I
O
1000
Na K
Zn
.10 .e LA .43.: a .
0 1 2 3 4 5 6 7 8 9 10

Figure 2.1.5 Example of an EDS x-ray spectrum
2.1.5 SEM/EDS

SEM combined with EDS is an effective analytical technique. The secondary
electrons produced from the sample make it possible to collect high resolution images,
giving detailed sample morphology, while the EDS detector allows qualitative and
quantitative analysis of elemental composition within the sample.

The combination of SEM images and EDS make this technique particularly
effective when analyzing cosmetic samples since complementary information can be
obtained in a single analysis. In this work, the morphology of the face powder samples
was compared by SEM. Elemental composition was determined by EDS and the element

weight percentages were calculated for comparison of samples.

16

 

2.2 FOURIER TRANSFORM INFRARED SPECTROSCOPY
2.2.1 Introduction

Infrared (IR) spectroscopy is a technique in which infrared radiation is directed
through a sample and, depending on the chemical structure of the sample, various
wavelengths of IR light are absorbed, allowing sample identification. The infrared region
of the electromagnetic spectrum is between 12800 cm"1 and 10 cm”, and is further
divided into the near IR (12800 cm" -— 4000 cm"), mid IR (4000 cm" — 200 cm“), and
far IR (200 cm'l— 10 cm!) regions. The mid IR region is the most useful for identifying
functional groups in the sample and to provide both qualitative and quantitative
information. Within the mid-IR region, functional groups display absorbances in the
4000-1500 cm'1 region, while the region from 1200 cm'1 to 600 cm'l, called the
“fingerprint region” displays peaks characteristic of certain molecules, potentially
allowing definitive identification of the sample.
2.2.2 Molecular Interactions with IR Radiation

Infrared light is low energy and therefore excites vibrational modes of a
molecule. There are two types of vibrations within a molecule: bending modes and
stretching modes. For a substance to absorb IR radiation, the molecule in question must
undergo a net change in dipole moment as a consequence of the incited vibrational
motion. Therefore, the bonds within the molecule must have a permanent dipole moment
to absorb IR light. Molecules including homonuclear diatomics and some simple
inorganic salts do not absorb in the mid IR region, but many organic functional groups do

have the required dipole moment and will absorb infrared light.

17

Every molecule has natural vibrational frequencies based on the functional
groups present in the molecule. Because vibrational energy levels are quantized, there are
particular frequencies associated with transitions between energy levels for different
types of bonds and functional groups (Appendix A). When the molecule absorbs infrared
light with a frequency that corresponds to the energy between two levels, the molecule is
vibrationally excited, an increase in the amplitude of absorbance occurs, and peaks are
seen in those regions on the infrared spectrum. Typically, the presence or absence of
major peaks in the region from ~4000 cm'1 to ~1500 cm“1 can yield information about the
functional groups present in a particular samples, and differentiation of samples can be

made based on such visual comparison of the corresponding IR spectra.

The “fingerprint” region of an IR spectrum is between ~1200 cm”1 and ~600 cm".
This region of an IR spectrum is often more complicated than in the region above 1200
cm'1 because these absorptions are mainly due to bending modes rather than stretching
modes within the molecule. Different molecules have unique vibrations in this region,
allowing definitive identification of the molecule.20 However, when classifying groups of
compounds based on IR, it is often easier to first examine the “cleaner” region from 4000
cm'1 to ~1200 cm'1 and determine whether samples can be grouped based on the presence

or absence of major peaks.
2.2.3 Fourier Transform Infrared Spectroscopy

Fourier transform infrared (FTIR) spectrometers now constitute the majority of
the instrument market compared to the older type dispersive instruments. The F TIR

spectrometer offers the advantages of improved signal-to-noise, speed, resolution,

18

accuracy, and reproducibility. The main advantages of F TIR is that it has the ability to
scan all wavelengths simultaneously, resulting in rapid analysis, and it converts data from
the time domain to the frequency domain.16 Rapid scanning of all wavelengths is

accomplished by the use of a Michelson interferometer, as shown in Figure 2.2.1.

 

 

 

 

 

 

 

 

 

detector

N

beam A _ %

incident 7 F %

beam beam .\
splitter mrrr or 1

(stationary)
beam B
i mirror 2
(moveable)

Figure 2.2.1 A Michelson interferometer

The incident IR beam encounters a beam splitter where the beam is divided, with half of
the beam traveling to a stationary mirror and the other to a moveable mirror. The beam is
reflected from both mirrors back to the beam splitter, where it recombines and passes to
the detector. However, due to the moving mirror, the two half beams have traveled
different pathlengths, resulting in interference patterns as the two recombine. When the
distance from the beam splitter to the moveable mirror is equal to the distance between
the beam splitter and the stationary mirror the pathlengths are the same, resulting in

constructive interference when recombined.'6’20’2' When the moveable mirror is moved a

19

distance of M4 from the beam splitter, the beam traveling towards this mirror must travel
a distance of M2 greater than the beam traveling to the stationary mirror”21 The result is
destructive interference when the beams are recombined, and no signal is detected. Due
to the constructive and destructive interference on recombining the beam, an
interferogram is produced when the moveable mirror is scanned at a constant velocity.
The recombined beam passes through the sample, which absorbs at all the frequencies
characteristic of the functional groups present. The frequencies absorbed are subtracted
from the interferogram and the beam passes to the detector, which records variation in
energy versus time for all wavelengths simultaneously.

A mathematical function, the Fourier transform, is used to convert the
interferogram, which is measured in the time domain, into the absorbance spectrum,
which is in the frequency domain. The period of the interferogram is related to the center
frequency that is seen in the spectrum and the interferogram’s exponential decay is
related to the linewidth of the peaks in the spectrum.

2.2.4 Attenuated Total Reflection

Attenuated total reflectance (ATR-FTIR) uses a sampling accessory that has been
shown to be reproducible for qualitative and quantitative analysis of liquid, paste, solid,
film, and powder samples that range from organic to inorganic in nature.22 While a
transmission IR measurement collects a spectrum that is an average of the bulk properties
of the sample, ATR measurements only measure surface properties of the sample because
the IR beam only penetrates a short distance into the sample.

An ATR accessory typically consists of a sampling plate in close contact with a

high refractive index crystal such as diamond, zinc selenide, or germanium. The solid

20

sample is placed on top of the sample plate and pressure is applied to the sample to

ensure good contact between the sample and the crystal (Figure 2.2.2) .16

ATR crystal

/

/ ,.
/ \
/ \ mirror
:.\ /
‘ V.
1R beam in / \ 1R beam out

Figure 2.2.2 Diagram of a single-bounce ATR accessory

     

IR radiation is directed into the crystal at an angle exceeding the critical angle for total
internal reflection.22 The high refractive index of the crystal allows the radiation to reflect
within the ATR element rather than be transmitted. The fraction of the beam that is
reflected increases with the angle of incidence, 9,. When the incident angle is greater than
the critical angle, the beam is completely reflected. When an absorbing sample is brought
into contact with the crystal, the incident beam penetrates a small distance into the
sample before reflection occurs.'6’2' The distance the beam protrudes beyond the surface
of the crystal and into the sample is the penetration depth which is dependent on the
wavelength of the incident light. With IR radiation, the penetration depth is typically only

a few microns deep. However, the penetration depth is also dependent on the refractive

21

 

index of the medium, the angle of incidence, and the wavelength of the incident beam of
light. On penetrating into the sample, frequencies of IR radiation corresponding to
energies of vibrations of chemical bonds in the sample are absorbed. A spectrum of
absorption versus frequency is produced in the same manner as FTIR, allowing sample
identification. The absorbances, although dependent upon the angle of incidence, are
independent of sample thickness because the radiation penetrates only a few micrometers
into the sample.'6

A major concern when using ATR is that, while the same peaks are observed on
an ATR spectrum as with transmission FTIR, their intensities are often weaker.
Therefore, some of the detail may be obscured in ATR-IR spectra. This is most likely due
to the penetration depth. Since penetration depth is dependent on the wavelength of the
incident light, absorbance bands at shorter wavelengths (infrared region) will be of lower
intensity that those of longer wavelength.

A major benefit of using ATR is that there is no sample preparation; the sample
can simply be placed upon the ATR crystal which is cleaned thoroughly between
samples. In this research, the sample introduction method is particularly useful because it
is difficult to make a transparent KBr disk with opaque samples and samples like face
powders that contain anti-caking agents. In addition, ATR is a technique commonly used
in forensic labs. In this work, ATR-FTIR was used to qualitatively determine the organic

components present in the face powder samples.

22

2.3 MATRIX-ASSISTED LASER DESORPTION IONIZATION-MASS
SPECTROMETERY
2.3.1 Introduction
The field of mass spectrometry blossomed in the second half of the twentieth century.”23
The basic principle behind the technique was that a sample solution was evaporated into
the gas phase, the molecules were then ionized, and the resulting ions separated and
analyzed based on their mass/charge (m/z) ratio. Since the introduction of mass
spectrometry as an analytical technique, a number of different sample ionization methods
have been developed, for example electron impact ionization (EI), chemical ionization
(C1), or electrospray ionization (ESI). However, larger molecules ( >1000 Da) were
difficult to analyze utilizing CI and EI, mainly due to difficulties in evaporating the
sample solution; on heating, the sample would decompose rather than evaporate. To
overcome this obstacle, instantaneous evaporation must occur. A solution to this problem
was discovered in the 19805, with the introduction of matrix-assisted laser desorption
ionization (MALDI).
2.3.2 MALDI

MALDI is a laser ionization method in which a short, intense laser pulse is used
to ionize the analyte molecules. This is achieved in two steps. First, the compound that is
to be analyzed is mixed with an organic matrix on a solid sampling plate.23 Several
matrices are available for MALDI analyses, with the choice of matrix dependent on the
laser wavelength used. The molar ratio of analyte to matrix in the mixture is typically
~l :10,000. The sample-matrix mixture is allowed to dry on the plate, resulting in a spot

of analyte-doped matrix.23 The plate is then placed in the sample chamber, which is

23

evacuated to the operating pressure of the instrument. The laser is then fired at the
sample-matrix spot in short pulses, resulting in the ablation of bulk portions of the
sample-matrix spot. The laser pulses are in the order of ~10 u] and are focused on a small
diameter spot (~100 um) resulting in intense heat on the surface of the sample spot. This
rapid heating causes rapid sublimation of the matrix crystals, leaving the analyte

molecules intact in the expanding matrix (Figure 2.3.1).23

 

 

 

 

Irradiation
(~337 nm)
o-
“+-
e *8 o —-> o e o 3’ °
Desorption desolvatron ° 0 o
proton
transfer
0 Matrix
O Analyte

Figure 2.3.1 The principle of MALDI

Although the exact mechanism of ionization is not well understood, the process produces
gaseous molecules of matrix and analyte, ion pairs, and ions of both matrix and analyte.
Both positive and negative ions of the initial analyte are produced and hence, either the
positive ion mode or negative ion mode of detection can be selected.

The ions produced are then focused using a series of electrostatic lenses into the
mass analyzer. A time—of-flight (TOF) mass analyzer is typically used for pulsed ion
sources, and is therefore often utilized with MALDI-MS.23 The linear TOF analyzer has
been used since the 19505, and consists of a flight tube, approximately 1 meter in length

(Figure 2.3.2).23

24

Analyzer
field-free region
.I. I Positive ions I
3. - Detector
Source . uuuuuuuuuuu I IIIIIIIII-III...III-IIIIIIIIIII)l -20 kV
+20,000 V L i
: d —

 

time-of-flight tube

Figure 2.3.2 The time-of-flight mass analyzer

The ions are accelerated from the sample chamber into the TOP tube by an applied

electric field and are imparted with kinetic energy (KE):

, 2

KEzmv

 

= ze Va“. (equation 2.3.1)

where m = mass (kg), v = velocity (m/s), 2 = charge, Vacc = accelerating voltage (V), and
e = elementary charge (1.6 x lO'19 C). The ions then enter the flight tube and are separated
according to the time taken to drifi down the tube in a field-free region. The drifi time of
the ions is dependent on their size, or mass-to-charge ratio (m/z) as shown in equation

2.3.2:

2ze V

__ acc
v _ .—

(equation 2.3.2)
m

Larger, heavier ions take longer to migrate the length of the tube than smaller, lighter

ions. Therefore, ions become separated according to their flight time.

25

At the end of the tube the ions strike the photographic plate detector, and the mass
spectrum is obtained by measuring the signal as a function of time.23 The time it takes for
the ions to reach the detector, t, is the distance traveled, d, divided by the velocity,
(t=d/v). By substitution and rearrangement, the m/z can be determined by knowing the

accelerating voltage, flight tube distance, and flight time, as shown in equation 2.3.3:

. 2t2e V
m _. . ace '
A 2 (equatron 2.3.3)

2.3.3. Interpretation of Mass Spectra
The mass spectrum is a graph of ion intensity as a function of m/z (Figure
2.3.3).24 This record of ions and their intensities serves to establish the molecular weight

and structure of the compound being analyzed.24

1250i
10003

750

intensity

500 i

250

    
   

 

 

 

   

1000 1200 1400

1600 1800 2000
m/z

Figure 2.3.3 Example of a mass spectrum.25

26

 

Within a mass spectrum are several notable peaks. The base peak is the most
intense peak in the spectrum and is assigned a relative intensity of 100%. The molecular
ion results from ionization of the analyte molecule and appears at an m/z value
numerically equal to the nominal molecular weight, or at one mass unit higher or lower
than the molecular weight. The ionization process breaks up the analyte molecule into
smaller fragments at certain bonds within the molecule. Once the analyte molecule breaks
into fragments, each fragment has a different, lower m/z value. The fragments themselves
will break up further in a similar manner. The pattern of fragmentation can often be
determined by the peaks seen in the mass spectrum. In these cases, it may be possible to
determine the structure of the analyte molecule from the mass spectrum based on the
molecular ion and fragment ions.

MALDI-MS is typically utilized for analysis of complex biological samples that
often include proteins or peptides and a great number of contaminants.23 MALDI allows
determination of the molecular weight of molecules up to 500 kDa.23 The contaminants
may include buffers, salts, detergents, and many other compounds of unknown origin.23
Due to the complexity of samples often analyzed by MALDI-MS, the resulting spectrum
is typically complicated with hundreds of different peaks, making structural identification
difficult. Therefore, such samples are often first separated by chromatography, or tandem
MS is performed. However, in this work neither of these options were available and face
powder samples were compared based on a visual comparison of corresponding mass

spectra.

27

CHAPTER 3 METHODS AND MATERIALS
A total of 27 face powders were examined in this research, totaling 351 different

pairs of samples, where:

n(n -— 1)

 

pairs = ’ (eqn. 3.1)

Samples were either purchased or donated for the project, and varied in manufacturer,
type of powder, and shade (Table 3.1.1). Due to the high cost of these products, a limited
number of sample powders could be obtained. While 27 face powders represent only a
small fraction of the products available, the samples obtained for this work provided
seventeen different brands of powders to be compared. A complete ingredient listing for
each powder sample is shown in Appendix B.

Table 3.1.1 Face powder samples

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Sample Brand Type of powder (as Color Condition
described on label)

1 Almay Luxury Finish loose ivory new
2a Revlon Skinlights golden light new
2b Revlon Skinlights golden light new
2c Revlon Skinlights golden light new
2d Revlon Skinlights golden light new
3 Neutrogena Skinclearing pressed medium

Oil-Free
4 Almay Clear Complexion pressed translucent

Light & Perfect
5 Loreal True Match Super Neutral

Blendable
6 Covergirl Clean ivorL Used
7 Max Factor Colour Adapt pressed Light, pale
8 J afra Translucent Blushing Used

beige

9 Prestige Perfectly Matte Honey

translucent
10 Covergirl Clarifying Natural ivory
11 NYX Twin Cake Honey beige New
12 Le Mirador Instant Used

 

 

 

 

 

28

 

Table 3.1.1, continued

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Infusion
l3 Covergirl Smoothers pressed Translucent
14 Bonnie Bell Glimmer loose Gold ‘11 glitz new
Bronze Luminous
1 5 Jordana pressed natural New
16 Maybelline Finish Matte pressed Medium
beige
17 Jane Bronzing Sahara
18 Wet ‘n Wild pressed Ivory
19 Estee Lauder Moisture Beige
Balanced Translucent
20 Motives Color Swirl Sun storm Used
21 Covergirl (unknown) used
22 Covergirl TruBlend pressed Translucent used
fair
23 Wet ‘11 Wild Ultimate pressed bare Used
touch
24 Wet ‘11 Wild Ultimate pressed natural new
touch

 

 

Some of the powders were labeled as either “pressed” powder or “loose” powder, while
others had no designation. Samples 2a, 2b, 2c, and 2d were separate powders within the
same compact. All of the powder samples were very fine, uniform powders, which could
not be easily differentiated by the naked eye. Each sample was analyzed by SEM-EDS
for morphology and elemental composition, and FTIR and MALDI-MS for information
regarding organic components. The data from these techniques will be compared in order
to determine whether these methods can be beneficial in distinguishing between the
various sample powders. In addition, three samples (samples 2b, 21, and 22) were
analyzed by each technique in triplicate to assess reproducibility of the techniques.
Samples from three of the powders (samples 12, 17, 18) were re-analyzed and treated as

“blind” samples to assess the discrimination and classification ability of each technique.

29

 

3.1 SEM-EDS

Adhesive tabs were attached to aluminum stubs, and approximately 1 mg of each
powdered sample was placed on an individual stubs. The stubs were then coated with
carbon using an Ernest F. F ullam Inc. EFF A MKII Carbon Coater. Although SEM
samples are most often coated with gold, carbon was used instead because it has a very
low absorption factor for x-rays.l7 Once the samples were prepared, the stubs were placed
in a sample holder that held up to 4 samples at one time. The sample holder was then
inserted into the SEM for analysis.

The scanning electron microscope used in this work was located in the Center for
Advanced Microscopy on the Michigan State University campus. A JEOL (Japan
Electron Optics Laboratories) 6400 V scanning electron microscope with a LaB6 emitter
(N oran EDS) coupled with a Thermo Electron Corporation X-ray Detector (Waltham,
MA) was used for the analysis of all samples, using an accelerating voltage of 20.0 kV
and a working distance of 15 mm.

For three samples, x-rays were collected at three different spots to ensure
homogeneity, and x-rays from one spot was collected on the other samples. In addition,
three samples (samples 12, 17, and 18) were selected as blind samples to test
discrimination ability of this technique. The computer software program Noran System
Six X-ray microanalysis system (Thermo Fisher Scientific, Inc., Waltham, MA) was used
to identify the elements within the sample, and to provide quantitative results.

3.2 FT IR Spectroscopy
A Perkin Elmer Spectrum One FTIR spectrometer equipped with a mercury

cadmium telluride (MCT) detector (Perkin Elmer, Shelton, CT) was used in this study.

30

The spectrometer was fitted with an attenuated total reflectance (ATR) sampling
accessory containing a ZnSe crystal (Perkin Elmer, Shelton, CT). The spectrometer was
operated and spectra were analyzed using the software program Spectrum (version 5.0.1,
Perkin Elmer, Shelton CT).

The crystal and sample plate were cleaned with methanol prior to placing ~50 mg
of powder sample directly on the sample plate. The exact mass of the powder used was
not measured, but enough was used to completely cover the crystal with a measureable
depth. Pressure was applied using the pressure arm, and the software program provided a
pressure bar that measured the actual pressure imparted on the samples. For every
sample, > 110 psi was recorded, sufficient to ensure a flat, uniform sample, yet not
sufficient to damage the crystal. Scans were collected from 4000 cm'1 - 400 cm".
Background scans were run prior to each sample. A total of 8 scans were collected per
sample. Each sample was analyzed in replicate (n=5) to ensure reproducibility, and the
automatic baseline correction tool was used on each spectrum. The five spectra were
averaged for each sample, and the average spectrum was used in subsequent sample
comparisons. The peaks were labeled using the software program using a threshold for
peak detection of 1.00% transmittance. Along with the known samples, three blind
samples (samples 12, 17, and 18) were also analyzed by F TIR-ATR to assess the
discriminating ability of the technique.

3.3 MALDI-MS

A Voyager DE STR Biospectrometry Workstation with a MALDI-TOF Mass

Spectrometer (Applied Biosystems, Foster City, CA) was used in this work. This

instrument is located in the Mass Spectrometry Facility at Michigan State University.

31

Each of the 27 samples were prepared by extracting in ~0.1 mL dichloromethane (DCM),
a solvent that dissolves many organic substances. Saturated solutions of each powder
were prepared in DCM then filtered with an ultra-filter (0.2 um) syringe to produce the
sample solutions. A 0.5 uL aliquot of u-cyano-4-hydroxy-cinnamic acid matrix provided
by the Mass Spectrometry Facility was placed in 34 separate wells of a lOO-well stainless
steel plate and allowed to dry. A 0.5 uL aliquot of each sample solution was placed in a
particular well, and wells were also prepared containing DCM only (solvent blank),
matrix only (matrix blank), a calibration standard containing the peptides bradykinin,
angiotensin, and neurotensin (with masses of 1059.6, 1671.9, and 1295.7 amu
respectively), and the three blind samples (samples 12, 17, 18).

The software provided a method to mass calibrate the instrument, a process that
was done prior to collecting data. To calibrate the mass spectrometer, a mass spectrum of
the calibration standard was first obtained. The reference file in the computer for the
calibration standard being used was opened and compared to the mass spectrum obtained.
The peaks in the initial mass spectrum were matched to the peaks in the reference
spectrum with a mass tolerance of i4 m/z. On matching peaks, the spectrum was saved
and used as the calibration file for all samples.

The software program Voyager Control Panel (v. 5.1, Applied Biosystems, Foster
City, CA) was utilized to set instrument parameters and to collect spectra, and the
program Data Explorer 4.0 (Applied Biosystems) was used to visualize mass spectra.
The instrument used a nitrogen laser (337 nm) and was operated in linear positive ion
mode with 100 shots/spectrum and analyzed over an m/z range of 10-1000 amu. A

spectrum was obtained for five different spots on each sample, and the five spectra were

32

averaged. The guide wire setting was at 0.05, and the delay time was 100 ns. The peak
threshold for all the mass spectra was set at 0.1% of the intensity of the most abundant

peak.

33

4. RESULTS AND DISCUSSION
4.1 SEM-EDS

The weight percentage of elements was calculated using Vontage 1.5.1 software
(N oran Instruments). Calibration of the EDS is performed by the staff at the Center for
Advanced Microscopy periodically using known metal standards. When determining
values of atomic/weight percentages, a correction factor that contained atomic number
effects (Z), absorption effects (A), and fluorescence effects (F), known as ZAF correction
was applied in order to compensate for matrix effects and provide accurate
quantitationf"7 These factors are described in detail below.

The two factors affecting characteristic x-ray emission that are dependent on
atomic number are stopping power and backscattering.26 Stopping power is the ability of
the sample material to reduce the energy of an incident electron by inelastic scattering,
which is related to the ratio of atomic mass number to the atomic number. The ratio
increases with increasing atomic number.26 The atomic number effect is the difference
between the average atomic number of the sample and a calculated 100% pure standard.
Therefore, if the mean atomic number of the standard differs from the sample, it must be
accounted for in the Z correction.

The absorption factor occurs as a result of the neighboring atoms. If an atom
emits an x-ray, the x-ray must pass through a certain distance within the sample. Within
the distance traveled are surrounding atoms. These other atoms may absorb that energy,

resulting in fewer x-rays observed in the spectrum.

34

Fluorescence occurs when the x—rays emitted from certain elements are absorbed
by other elements, resulting in subsequent fluorescence. The ZAF correction factor
compensates for all three of these effects.

During analysis, images of the powder samples were observed to be similar for all
samples. Thus, morphology as a discrimination tool was not a useful means of
classification. An EDS spectrum was collected for each of the samples (Figure 4.1.1).
The tabulated atomic percentages for all samples were compared to distinguish samples
based on elemental composition. The most common elements detected in the samples
were oxygen (O), silicon (Si), aluminum (Al), magnesium (Mg), and iron (Fe). The high
levels of O, Si, and Mg relative to other elements were not surprising as talc, a primary
component of most face powders, has the chemical formula Mg33Si4Om(OI-I)2. Another
possible source of the silicon was polydimethylsiloxane (PDMS), typically present in

cosmetics as a skin sofiener and emollient.

 

 

 

 

 

1500,
Si
1000
El
5
8
Mg l
500
O
0 Al Ti Fe
0 1 i 3 4 '5 6 7
keV

Figure 4.1.1. EDS spectrum of sample 6.

35

4.1.1 Sample Differentiation Based on Unique Elements

Along with these commonly found elements, there were also elements in some
samples that were uncommon which allowed differentiation of these samples from the
majority, as shown in Table 4.1.1. These elements included potassium (K), chlorine (Cl),
sodium (Na), bismuth (Bi), titanium (Ti), fluorine (F), silver (Ag), zinc (Zn), and
phosphorus (P) with weight percentages ranging from 1% to 8%. The major source of
the fluorine in these samples is most likely Teflon®. Another possible contributor of
fluorine is mica. Mica is a mineral sometimes found in cosmetics that contains F, and a
period 1 element such as K, Na, or Ca, and either Al, Mg, or Fe.

Sodium could originate from common cosmetic ingredients such as sodium
benzoate, or sodium laureth sulfate. Titanium dioxide, a white pigment, could account for
the titanium in some samples. Zinc stearate, calcium carbonate, and iron oxides are also
listed as possible ingredients for some samples.

The presence of these rare elements made several samples unique. For example,
sample 3 was the only sample that did not contain oxygen, and it was also the only
sample that contained silver. Samples 7 and 22 were the only ones that contained fluorine
and could thus be distinguished from all other samples. The sixth ingredient listed on
Sample 7 was PTFE, or Teflon®, while sample 22 had no ingredient list although is
likely to also contain Teflon® or mica due to the presence of fluorine. Sodium was only
found in samples 13 and 14, phosphorus was only present in sample 8, and zinc was
found only in samples 3, 13, and 24. Samples 8 and 9 were the only powders containing
calcium, while only sample 12 contained bismuth (Bi) and chlorine (Cl). However, no

ingredient list was available for sample 12, so the ingredient that contributed these

36

elements is unknown. A possible ingredient containing both bismuth and chloride is
bismuth oxychloride. This was listed under “may contain” on sample 15. However, the
EDS spectrum of sample 15 did not contain either Bi or CI, but since no product list label
was located on sample 12, this ingredient could be a possible source of Bi and C1 in the
spectrum of sample 12.

Table 4.1.1 Samples that could be identified based on unique characteristics

 

 

 

 

 

 

 

 

 

 

Sample number Unitme element(s)

3 no oxygen, silver, zinc
8 phosphorus, calcium

9 calcium

12 bismuth, chlorine

13 sodium, zinc

14 sodium

24 zinc

7 and 22 fluorine

 

 

 

While the presence of the less common elements could differentiate nine of the 27
samples (Table 4.1.1), the remaining 18 samples contained similar elements and hence
could not be distinguished in a similar manner.
4.1.2 Sample Differentiation Based on Common Elements
All samples except for sample 3 contained 0, Mg, Al, Si, and Fe at atomic percentages
ranging from 1 to 66%. A hierarchical cluster analysis (HCA) was performed using the
statistical software program, Minitab (version 14, Minitab, Inc., State College, PA) in an
attempt to differentiate all samples based on differences in levels of 0, Mg, Al, Si, and Fe
present.
4.1.2.1 Cluster Analysis

Cluster analysis is a method for dividing a group of objects into classes so that

similar objects are in the same class.27 Samples are plotted in multi-dimensional sample

37

space and cluster analysis searches for objects which are close together in the variable

space.27 One of the most common distances used in cluster analysis is the Euclidian

distance, d, that is computed from the sample data using equation 4.1.1:27

 

d = \/(xl —y,)2 +(x2 —y2)2 +---+(xn —y,,)2 (eqn. 4.1.1)

where x is a data set, y is a data set, and n is the number of data points in a set. When
applying this to the SEM-EDS data, each face powder sample is a data set, and within
each data set are six variables corresponding to the weight percentages of 0, Mg, Al, Si,
and Fe within that sample. The Euclidian distance can then be calculated between each
sample data set. Once the distances between each of the samples are calculated, the data
sets are clustered using the single linkage method which is based on the minimum
distance between each cluster.27 If two neighboring data sets are separated by a small
Euclidian distance, they will most likely be clustered together. The similarity between
samples is defined as:27

s = 100(1— dxy /dmax) (eqn. 4.1.2)

where dxy is the distance between two points, and dmax is the maximrun separation
between any two points.27 The results of cluster analysis are often shown visually as a
dendrogram, with similarity displayed on the y-axis and samples on the x-axis. Grouping
of samples can be assessed at various levels of similarity; the higher the level of
similarity, the more similar the samples are within that group.

The SEM-EDS data were inputted into the Minitab 14 software program; each
sample was placed in a column, and six rows were present for each sample. Cluster
analysis was performed using the Euclidean distance and single linkage method, and

selecting a 90% similarity level. The resulting dendrogram was obtained (Figure 4.1.1):

38

SEM-EDS
Atomic percentages

90% similarity

35‘ —

 

57 ‘
similarity

 

90
100

 

 

 

 

 

Figure 4.1.2 Resulting dendrogram from cluster analysis of atomic percentages at a
similarity level of 90%. Tick marks on the x-axis correspond to each sample and each
color represents a final clustering of samples.

At a similarity level of 90%, all samples (27) were divided into four clusters (A,
B, C, and D) (Table 4.1.2). Cluster A contained only sample 3. It was expected that this
sample would be in a separate group than the others, as it was the only sample that
contained no oxygen as determined by SEM/EDS. It was therefore considered unique
when looking at unique characteristics (Table 4.1.1). Cluster B contained sample 12 and
one of the blind samples, which was also sample 12. Again, sample 12 was considered
unique due to the presence of bismuth and chlorine. However, the blind sample 12 being
clustered correctly with sample 12 provided further evidence of the accuracy of the
technique. Cluster C contained samples 2a, 2c, 2d, and each of the triplicate samples for
sample 2b. It was not surprising that 2a, 2b, 2c, and 2d were grouped together because the
four samples were within the same compact, with similar ingredients listed for each, and

were thus indistinguishable by SEM-EDS. However, within this cluster was also sample

39

14, but this was one of the samples previously determined unique based on the presence
of sodium. It is unknown why sample 14 was grouped with samples 2, because no
ingredient list was available for sample 14. The final cluster, cluster D, contained the
remaining samples, which limited discrimination ability based on O, Si, Mg, AI, and Fe.
These samples could not be distinguished from each other due to their large and similar
amounts of O, Si, Mg, Al, and Fe. The presence of these major elements possibly masked
the elements present at lower concentrations that may have been unique to such samples.

Table 4.1.2 Resulting clusters with 90% similarity

 

Cluster Samples

 

3

 

12, 12 blind

 

2a,2b,2b,2b,2c,2d,14

 

 

U0w>

4, 5, 6, 7, 8, 9,10,11,13,15,16,17,18,19,20, 21,21, 21, 22, 22, 22, 23, 24,
17 blind, 18 blind

 

 

It is important to note that for the three samples analyzed in triplicate, all were grouped
within the same cluster, thus providing confidence in the reproducibility of SEM-EDS.
Based on the presence of unique elements and subsequent cluster analysis, 267
pairs out of 351 total pairs of samples could be distinguished, for a discrimination power
of 76%. The discrimination power was limited by the large number of samples in cluster
D, which could not be differentiated based on levels of O, Si, Mg, Al, and Fe.
4.2 FT IR Spectroscopy
IR spectra for each sample were compared visually, basing the comparison on the
presence or absence of major peaks in the spectra. Many of the samples had peaks
common with talc (Figure 4.2.1) which was expected because talc is a major component

in face powders. A peak at ~3 675 cm'1 was seen in all samples, and a trio of peaks at

40

 

2848, 2916, and 2957 cm" were seen in most samples. An exception to these common

peaks was sample 12 which showed none of these, allowing differentiation of sample 12

from the remaining samples.

99.3

 

 

2500 2000 1500 1000 650.0
wavenumber (cm")

Figure 4.2.1 IR spectrum of talc

100

80

60

%T

20

 

w

4000 3500 3000 2500 2000 1500 1000 550

wavenurnber (crn‘l)

Figure 4.2.2 IR spectrum of sample 6

41

Although the samples shared many common peaks, there were a few prominent
peaks that allowed classification of the samples into smaller groups. Samples were
classed into two groups depending on the presence or absence of an absorbance at 1740
cm", which is a characteristic carbonyl stretch found in aldehydes, ketones, carboxylic
acids, or esters. Within each group, firrther classification was possible based on an
absorbance at 1260 cm", characteristic of a C-O stretch found in alcohols, ethers,
carboxylic acids, or esters. The study by Gordon and Coulson also used peaks at ~1740
cm'land ~1640 cm'1 as discriminating characteristics for their foundation samples.8

Samples were further classified based on the presence or absence of peaks at 1538
cm'l and 1638 cm”. The 1538 cm'1 peak is characteristic of the double bond in an
aromatic ring, and the 1638 cm'1 peak is characteristic of a C=C in an alkene or C=O
bond in an aldehyde, ketone, carboxylic acid, or ester.

The listed ingredients could account for several of the functional groups
responsible for the peaks on the IR spectra. Stearic acid and palmitate, both compounds
listed on several powders, contain a carboxylic acid. Allantoin, also a main ingredient
contains several amide bonds. Parabens contain both an aromatic ring and ester bonds.

Using the four major peaks (1740, 163 8, 153 8, and 1260 em"), a method for
discrimination was developed to classify the samples. The classification scheme is shown
in figure 4.2.1. By comparing the IR spectra visually, based on the presence or absence of

these four major peaks, the 27 samples were classed into nine groups (Table 4.2.1).

42

Table 4.2.1 Nine groups based on presence or absence of major peaks (1740 cm“,
1260 cm", 1538 cm", and 1638 cm")

 

C3
'1
8
”U
in

Sample

 

2a, 2b, 2c, 2d, 10, 21

 

9,11

 

16,19

 

12,14

 

13

 

3,4, 5, 6

 

7, 22

 

15, 17,18, 23, 24

 

 

 

 

“EQ'TJITJUOUU>

1,8, 20

 

It was possible to discriminate 313 pairs of samples out of a total 351 pairs, for a
discrimination ability of 89%. In comparison, FTIR analysis of foundation samples by

Gordon and Coulson yielded a discrimination ability of 98.3%.8

43

Figure 4.2.3. Classification of samples based on presence and absence of major
peaks in IR spectra.

   

PEAK AT 1740 cm'1

 

YES
1. 3, 4, 5, 6, 7, 8, 13,
15,17, 18,20, 22,
23, 24

NO

Za, 2b, 2c, 2d, 9, 10,
11,12, 14, 16, 19, 21

PEAK AT 1260 cm'1 PEAK AT 1260 cm”1
YES NO

NO YES
23, 2b, 2C, 2d, 9, 10, l, 8, 15, 17, 18, 20,
11’ 21 1,14,16,19 3.4,5,6,7, 13,22 23’ 24

PEAK AT 1260 cm" PEAK AT 1260 cm'1

PEAK AT 1533 cm‘1 PEAK AT 1538 cm“ PEAK AT 1538 cm'1 PEAK AT 1538 cm-1
YES YES NO YES

2a, 2b, 2c, 2d, 10, 21 15,19 7, 22 15, 17, 1s, 23, 24

PEAK AT 1538 cm'1 pEAK AT 1533 cm-1 PEAK AT 1538 cm" PEAK AT 1538 cm'1
NO NO YES
9, 11 12,14 3:41516; 13

PEAK AT 1538 cm'1
YES
13

PEAK AT 1638 cm"

NO
3,4,5,6

44

4.3 MALDI-MS

Mass spectral data for each sample were imported into a Microsoft Excel
(Microsoft Corporation, Redmond, WA) spreadsheet to plot the mass spectrum. The
spreadsheets had columns for the centroid mass and the relative intensity of each peak in
the spectrum. A plot was then constructed with relative intensity as the dependent
variable (y-axis), and centroid mass as the independent variable (x-axis) (Figure 4.3.1).
Most of the peaks occurred at masses of less than 800 Da, so in order to obtain more
detail in the plot, the x-axis was adjusted for a maximum value of 800. The resulting

spectra were compared visually to assess similarities and differences among samples.

100. 7 , , . 1 — Ml“— _, ~ _.-_ ___.___-._ —»— ,__1_ ——————— _. ,,

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

relative ,0 __ _.-________,, E. _-fi. ‘ l
abundance » l i
so * *-—— v»#~g——~ +e—4 [
L
.0. _-__ __ _ m 2*. __ “—2..._ ______._____.;
.. _-_-__ - E , l l
t I > i ‘l
20 __r_,_-_ __ - -_._ 11 e- .- l
,0 .1 E, II. ? II
I t‘
O
0 1m 31) ID 403 SD 000 7CD KI)
m/z

Figure 4.3.1 Mass spectrum of sample 13

When performing the experiment and preparing the samples on the MALDI plate,
matrix solution was first pipeted onto each plate well followed by a small amount of
sample. Due to this method, peaks on each mass spectrum could be attributed to the

matrix. Therefore, when preparing the plate, there was a well reserved for matrix only,

45

and a well reserved for the extraction solvent (DCM) only, so a spectrum for each was
obtained. The mass spectrtun for each sample was first compared to the matrix spectrum,
then the dichloromethane spectrum. Any peaks that were not present in the matrix or
DCM spectra but were present in the sample spectra were recorded and used to compare
samples.

There were three samples with obvious repeating patterns in their mass spectra.
Sample 22 had peaks at m/z 440, 468, 496, and 524, as well as another pattern at m/z
662, 690, 718, 746, and 774. The difference in each of these groups is 28 mass units. In
sample 7, this pattern of peaks with a difference of 28 mass units was also seen at m/z
690, 718, and 746. That pattern is also seen in sample 13 (m/z 705, 733, 761), and
another pattern is seen with peaks separated by 14 mass units (m/z 467, 481, 495, 509,
523, 537, 551). The pattern of 14 and 28 mass unit differences is characteristic of
fragmentation of an alkane chain, where 14 mass units represents a loss of —CH3 and the
addition of a proton, while 28 mass units represents a 103 of —CH2CH3 and the addition of
a proton. Hence, the pattern of peaks observed in these samples corresponds to losses of
methyl and ethyl groups with the simultaneous gain of a proton. However, since the
actual masses are different, these groups are being lost from ions with a different
molecular weight, and are thus different compounds.

These three samples (samples 22, 13, and 7) were classified into one group
(Group A) that was different from all other samples because of the presence of these
repeating patterns. However, further differentiation between the samples in Group A was
possible. For example, sample 13 is different from both sample 7 and 22 because its

pattern is in the 500 m/z range while the others’ patterns are in the 400 and 700 m/z

46

range, respectively (Figure 4.2.3). This is most likely due to an alkane chain attached to

different compounds or functional groups. Samples 7 and 22 could not be distinguished

from one another based on this comparison.

Group A
7, 13, 22

 

r——'-—1

Repeating patterns in Repeating patterns in
m/z 500 range: m/z 400, 700 range:

13 7,22
Group A = repeating patterns of 14 and 28 mass units

Figure 4.3.2. Scheme for differentiating powder samples in Group A based on
MALDI-MS data
There were prominent peaks seen in the mass spectra of the remaining samples that
would allow for further classification. The peak at m/z 615 was used as a characteristic to
further classify the samples. This peak was observed only in the mass spectra of samples
9, 18, 23, 24, and the blind sample 18. Therefore, these were placed in Group B. Within
this group, it was determined that these samples could be distinguished further. Sample 9
was the only one of the 5 samples in Group B that had a peak at m/z 538 but no peak at
m/z 593. A peak at m/z 504 was present in samples 23 and 24, but not in 9 or 18. Sample
18 was different from the others in that it had no peak at m/z 538 or 504, yet did have a
peak at m/z 504. Within group B, samples 9 and 18 were unique, and samples 23 and 24
could not be distinguished from each other (Figure 4.2.4). The m/z values discussed

above could not be attributed to any of the ingredients listed on the products.

47

Group B:
9.18, 18 blind, 23,
24

 

No peak at m/z 538,
peak at m/z 593:

18, 23, 24

Peak at m/z 538:
9

 

Peak at m/z 504: No peak at m/z 504:
23, 24 18

 

Group B = Mass spectra contains a peak at m/z 615

Figure 4.3.3. Scheme for differentiating powder samples in Group B based on
MALDI-MS data

The remaining samples that were not already grouped into Group A or Group B
were then examined, with a focus on whether or not a peak at m/z 538 was present, as it
was one of the more intense peaks in the spectra. Spectra from samples 1, 2a, 2b, 2c, and
2d contained a peak at m/z 538. These were designated as belonging to Group C (Figure
4.2.5). Upon closer examination, sample 1 could be distinguished from the others from its
peaks at m/z 328, 356, 494, 510, 522, and 550 that were not found in any of the powders
from sample 2. However, samples 2a, 2b, 2c, and 2d could not be distinguished from

each other, which was not surprising as they all came from the same compact.

48

Group C:
1, 2a, 2b, 2c, 2d

 

   

     
 

No identifying
characteristics:

2a, 2b, 2c, 2d

Peaks at m/z 328,
356, 494, 522, 550:

  
   

1

Group C = Peak at m/z 538

Figure 4.3.4. Scheme for differentiating powder samples in Group C based on
MALDI-MS data

The remaining samples were placed into Group D. A summary of the initial
groupings is outlined in table 4.3.1. However, these classifications were based on very
obvious peaks observed in the spectrum.

Table 4.3.] Initial grouping of MALDI-MS data

 

1 1719

Group D had the most samples in it, and none had the high intensity peaks that
were used to differentiate previous samples into groups (Figure 4.2.6). However, there
were some small peaks (< 10% relative intensity) that were present that allowed further

discrimination of the remaining samples.

49

100 * *

 

 

80

 

 

60

 

40 ‘

 

 

relative abundance

20

 

 

 

 

 

 

 

 

 

 

 

 

 

100 200 300 400 500 600 700 800
m/z

Figure 4.3.5. Mass spectrum of sample 6

Group D was broken down into two groups: Group D, that included samples with a peak
at m/z 320, and Group D“ that included those samples with no peak at m/z 320 (Table
4.3.2).

Table 4.3.2 Initial classification of Group D based on presence of m/z 320

 

Group Samples

 

D, 3, 4, 5, 6, 10, 12, 12 blind

 

 

 

 

Di, 8,11,14,15,]6,17,l9, 20, 21,17 blind

 

Within Group D, samples 3 and 10 both had peaks at m/z 568 and 221 while the
others did not. Also, sample 3 had a prominent peak at m/z 431 that all of the others did
not have. Therefore, 3 and 10 could be distinguished from all other samples in Group D,
as well as each other. Sample 4 contained a peak at m/z 634 and samples 5 and 6
displayed a peak at m/z 282. Upon inspection of the spectra at m/z values greater than
800 Da, samples 4 and 12 (and blind sample 12) contained a peak at m/z 845 while 5 and

6 did not (Figure 4.2.7).

50

Group Di:

3, 4, S, 6, 10, 12,
12 blind

 

 

I
Peak at m/z 221, No peak at 221,
568: 568:

3, 10 4, 5, 6, 12, 12b|ind

No peak at m/z
845:

5,6

Peak at m/z 431: No pef3k1at m/z Peak at m/z 845:

3 10 4, 12, 12b|md

Peak at m/z 634:
4

No peak at m/z
634:

12, 12b|ind
Group Di = does not fit into categories A, B, and C, and has peak at m/z 320.

Figure 4.3.6. Scheme for differentiating powder samples in Group D. based on
MALDI-MS data

Within Group D“ further classification and identification could take place. For
example, a peak at m/z 845 was seen in samples 1 1, 16, 17, 21, and blind sample 17, and
within this subgroup 11 could be distinguished by a peak at m/z 551. Therefore, samples
16, 17, 21, and blind 17 could not be distinguished from each other. For those samples in
Group Dii, (samples 8, 14, 15, 19, and 20) with an absence of the 845 peak, an m/z 568
peak was characteristic of sample 15, and m/z 612 was characteristic of sample 8.

Samples 14, 19, and 20 could not be distinguished from each other (Figure 4.2.8).

51

Group Dii:
8, 11.14, 15, 16, 17,

17 blind, 19, 20, 21

 

 

Peak at m/z 84.5: No peak at m/z 845:
11, 16, 1;,117 blind, 8, 14’ 15' 19' 20

 

F—l—_-l I i l

Peakatm/1568: No peak aim/z 5680
612:
15 14,19, 20

   

Peak at m/z 551: No peak at m/z 551: Peak at m/z 612:
11 16, 17, 21 8

Group D“ = does not fit into categories A, B, and C, and has no peak at m/z 320

 

Figure 4.3.7. Scheme for differentiating powder samples in Group D“ based on
MALDI-MS data

Using this classification of the MALDI-MS spectra, there were 14 pairs of
samples that could not be distinguished from each other, for a discrimination ability of

96%.

52

4.4 Blind Samples

Prior to instrumental analysis, three samples were chosen at random to serve as
blind samples. The samples selected were sample 12, 17, and 18, and these were the blind
samples utilized for each technique to assess the accuracy of both the technique and the
methods of sample comparisons. SEM-EDS classified sample 12 and blind sample 12 in
the same group by cluster analysis. This particular sample was also one of the nine
samples that could be grouped based on unique characteristics, as it was the only sample
of the twenty seven that contained bismuth and chlorine. Both samples 17 and 18 could
not be identified by unique characteristics. Based on the cluster analysis, samples 17 and
18 were grouped into group D, as were their corresponding blind samples. Using the
F TIR method of discrimination based on the presence and absence of major peaks, each
of the blind samples was grouped with their corresponding known sample. Both blind
samples 17 and 18, and known samples 17 and 18 were also grouped with samples 15,
23, and 24. MALDI-MS also correctly classified each blind sample with its known
sample. Sample 18 and blind sample 18 were uniquely identified, sample 12 and blind
sample 12 were grouped with sample 4, and sample 17 and 17 blind were grouped with
sample 16 and 21.

While the classification techniques based on SEM-EDS, FTIR, and MALDI-MS
did not necessarily uniquely identify each sample, all of the blind samples were grouped
correctly with their known counterparts. These classifications give firrther confidence to

the discrimination abilities of each of the techniques.

53

4.5 Combined Techniques

Table 4.4.1 is a summary of the pairs of samples that could not be distinguished
from each other using the three techniques individually. However, it is possible to
combine the results of each technique to yield a higher discrimination ability than by
using the individual results.
Table 4.4.1 Pairs of samples that could not be distinguished by SEM-EDS, ATR-

FTIR, and MALDI-MS. Each row represents pairs that were indistinguishable by the
technique in each column.

 

 

 

 

 

 

 

 

 

 

 

SEM-EDS FTIR MALDI-MS
2a, 2b, 2c, 2d 2a, 2b, 20, 2d, 10, 21 2a, 2b, 2c, 2d
4, 5, 6,10,11,15, 16,17,18, 9,11 5,6
19, 20, 21, 23
7,22 16,19 16,17,21
3, 4, 5, 6 7, 22
7, 22 14, 19, 20
15, 17,18, 23, 24
l, 8, 20

 

For example, while samples 9 and 11 cannot be distinguished by FTIR, by using SEM-
EDS and/or MALDI-MS they can be discriminated from each other based on elemental
composition and mass spectra. Another example is samples 15 and 17 that can only be
distinguished by their mass spectra, while their FTIR and SEM-EDS results could not be
differentiated from each other. The final groupings of indistinguishable samples was
different for each technique. This is because each technique examines different sample
components and thus yields different information.

Together, SEM-EDS and FTIR could not distinguish between 18 pairs (333 out
of 35 1 , or 94.9% discrimination ability), SEM-EDS and MALDI-MS could not

distinguish between 12 pairs (339 out of 351, or 96.9 % discrimination ability), and FTIR

54

 

combined with MALDI-MS could not distinguish between 8 pairs (343 out of 351, or a
97.7% discrimination ability).

There were some samples that could not be distinguished from each other by any
of the methods used. Samples 7 and 22, while different from all other samples, were too
similar to be distinguished from each other and were always grouped together by the
three techniques. Sample 7 was Max Factor Colour Adapt face powder, while sample 22
was Covergirl TruBlend. Even though they were from different manufacturers, they most
likely share a common major ingredient. Inspection of the labels provided no insight as to
which ingredients because, on one of the samples (sample 22) no ingredients were listed.
In addition, there were other Covergirl products that were analyzed and easily
distinguished from sample 22; having the same manufacturer does not necessarily
indicate that those samples could not be distinguished.

Each of the techniques was also unable to discriminate between samples 2a, 2b,
20, and 2d. This was not completely unexpected, as these samples, while all separate
samples, were in the same compact. There was a listing of ingredients, and also a list of
ingredients that “may” have been contained in the powders. However, the list did not
describe which ingredients were in which sample. It is possible that the ingredients that
were listed were contained in each of the samples.

Samples 5 and 6 were the other pair that could not be differentiated. There was no
manufacturer connection, as sample 5 was Loreal, and sample 6 was Covergirl, and a
label containing the list of ingredients was not found on sample 6 for ingredient

comparison.

55

Out of 351 total pairs of samples, SEM-EDS could distinguish 267 pairs (76%
discrimination ability), FTIR could distinguish 313 pairs (89% discrimination ability),
and MALDI-MS could distinguish 337 pairs (96% discrimination ability). The combined

techniques can distinguish between 343 pairs. This is a discrimination ability of 97.7%.

56

5. CONCLUSIONS AND FUTURE RESEARCH
5.1 Conclusions

For cosmetic face powder samples, it is useful to determine whether two samples
could have originated from the same source. Therefore, it was determined whether the
techniques, SEM-EDS, F TIR, and MALDI-MS, could discriminate between pairs of
samples. The three techniques can individually provide a high degree of discrimination
when analyzing cosmetic face powders. However, because each technique provides
different information regarding the samples of interest, when the three methods are
combined, the resulting discrimination ability is high.

SEM-EDS provided elemental composition profiles of samples, and yielded the
lowest discrimination ability, 76%. This was mainly a result of the common ingredient,
talc, resulting in high atomic percentages of Mg, Si, and 0. However, several samples
contained more rare elements, for example bismuth, chlorine, or sodium that made these
samples unique withinthe set.

A major issue of the SEM-EDS data in this work was that for each sample, only
one spectrum was obtained. Ideally, at least three would be obtained and the resulting
atomic percentages would be averaged. However, due to time constraints as well as cost,
only one spectrum was generated, and therefore one atomic percentage value was
determined for each element. However, triplicate spectra were obtained for three samples,
all of which were reproducible. Blind samples were also analyzed and were correctly
classified with the corresponding sample.

A method of determining similarity, cluster analysis, was performed on the EDS

data for the samples. Each of the triplicate samples were grouped together. In addition,

57

the blind samples were all clustered within the same groups as their known counterparts.
However at 90% similarity, only four groupings of the 27 samples were obtained.

Using ATR-FTIR, 89% of possible sample pairs were able to be distinguished.
Many of the sample spectra had several peaks in common. However, a classification
scheme based on the presence or absence of four major peaks allowed the samples to be
grouped into nine groups. Each of these peaks was due to characteristic vibrations of
functional groups within the samples. However, it was impossible from the IR spectra to
determine which ingredient was responsible for IR absorption due to more than one
possible ingredient containing the functional groups that vibrated at those particular
frequencies. Based on the peaks at 1740, 1260, 1538, and 1638 cm'1 there were 313 pairs
that could not be distinguished. The inability to distinguish between powders by F TIR
was most like due to common, shared ingredients such as stearic acid, palmitate, or
parabens which contain functional groups characteristic of the four peaks that were used
for classification.

The final technique used was MALDI-MS, which resulted in a discrimination
ability of 96%. MALDI is a powerful technique for elucidating structural information of
samples. However, due to the complexity of samples, little information of the sample’s
molecular structure could be obtained, yet MALDI-MS provided a high degree of
discrimination based on the presence or absence of major peaks in the mass spectra.

The MALDI technique requires the use of a matrix. The matrix itself contains
numerous peaks in the resulting mass spectrum. This complicated visual analysis of the
sample mass spectra. First, each sample spectrum had to be compared to the pure matrix

spectrum , to discount any peaks in the sample that could be attributed to the matrix.

58

Most of the peaks found in the sample spectra were due to the matrix, leaving only a few
that were useful for classification purposes. It would have simplified the process, and
made it less time consuming, if it were possible for the software to subtract the matrix
spectral peaks from the sample spectra.

Overall, the three analytical techniques provide powerful discrimination ability
when analyzing cosmetic face powders, able to distinguish 343 out of 351 possible pairs
of samples. The inability to distinguish some pairs was most likely the result of
formulation similarities, as many samples shared common ingredients. There was no
clear pattern of the inability to discriminate between certain pairs and their manufacturer.
All of the sub-samples from sample 2 (2a, 2b, 2c, and 2d) were unable to be distinguished
by all techniques. These samples were all in the same compact, and therefore from the
same manufacturer, and had the same product ingredient lists. Sample 5 and 6, another
pair that could not be distinguished by the combined techniques, were from different
manufacturers, as were samples 7 and 22, another pair that could not be distinguished.
5.2 Future Research

As a preliminary study, this research has highlighted the potential of SEM/EDS,
F TIR, and MALDI-MS for the discrimination of cosmetic face powders. However, before
such techniques are commonly employed in crime laboratories, significant further
research remains.

Cosmetics are prevalent in our society, and may therefore be encountered as
forensic evidence at crime scenes. The ability to accurately determine type of powder, or
to discriminate between powder samples can be useful in the field of forensic science. To

date, little research has been performed dealing specifically with cosmetic face powders.

59

While this research provides some information regarding the ability of three techniques to
discriminate between powder samples, there is much more that would prove beneficial in
the analysis of cosmetic face powders.

Cosmetic face powder is applied over the entire facial region, and thus a large
amount of transfer could occur. The transfer would most likely occur between skin and
some type of substrate, most often clothing. Therefore, a useful experiment would be to
apply the cosmetics to several different types of fabric, develop an extraction procedure
to remove the powder from the clothing, followed by instrumental analysis. In addition,
the interference effects of certain fabrics could be determined. Also, the existence of oils
on skin would most likely have an effect on the results of the experiment. Obtaining
wipings of powder directly from skin would be useful in ruling out interferences from
naturally occurring skin excretions.

The mass spectra obtained in this work were extremely complex, making
identification of compounds nearly impossible. Hence, it may be beneficial to employ a
chromatographic technique to separate sample components prior to detection by MS.
Identification of sample components based on fragmentation patterns in the mass spectra
would be considerably simplified due to the initial separation of components. Another
type of mass spectrometry such as GC-MS may be a more suitable method of analysis
because MALDI-MS, while used extensively in research facilities, is not readily available
in most forensic laboratories. GC-MS is routinely used in forensic analysis, however,
extraction of face powder components would need to be performed before the sample

could be injected into the instrument.

60

Another technique that would allow definitive identification of sample
components is tandem MS, where each peak and its resulting fragmentation patterns are
analyzed, making identification much simpler.

In this work, 27 samples were analyzed. However, on the market today there are
several hundred kinds of face powders. Obviously, using a larger sample set would prove
useful in determining the capability of these techniques to distinguish between samples.
Due to cost of these types of products, the sample set was limited in number. In addition
it may be useful to collect several powder samples from only one manufacturer to analyze
and compare differences between the powders of one brand.

This research project was the first to examine the applicability of SEM-EDS,
FTIR, and MALDI-MS to the analysis of face powders as forensic evidence. Overall, the
three methods have shown to be complementary analytical techniques that provide a

powerful means of discrimination among cosmetic face powder samples.

61

APPENDICES

62

APPENDIX A. Abbreviated Table of IR Group Frequencies for Organic Groups16

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Frequency Range (cm'l)
Bond Type of Commd
C—H Alkane 2850-2970
C-H Alkene 1340-1470
C-H Alkyne 3010-3095
C-H Aromatic rings 675-995
O-H Monomeric alcohols, phenols 3590-3650
hydrogen-bonded alcohols, phenols 3200-3600
Monomeric carboxylic acids 3500-3650
Hydrogen-bonded carboxylic acids 2500-2700
N-H Amines, amides 3300-3 500
C=C Alkenes 1610-1680
C=C Aromatic rings 1500-1600
CEC Alkynes 2100-2260
C-N Amines, amides 1180-1360
CE Nitriles 2210-2280
C—O Alcohols, ethers, carboxylic acids, 1050-1300
esters
C=O Aldehydes, ketones, carboxylic 1690-1760
acids, esters
N02 Nitro compounds 1500-1570
1 300-1 3 70

 

63

 

APPENDIX B. Listed ingredients in each sample

 

Sample

Ingredients contained in sample

Ingredients that may be
contained

 

1

Not listed

 

2a
2b
2c
2d

Talc, dimethicone, silica, zinc stearate,
polyehylene, pentahydrosqualene, diisostearyl
malate, lecitin, calcium silicate, nylon-12, retinyl
palmitate, tocopheryl acetate, isopropyl
isostearate, cyclomethicone, magnesuium
ascorbyl phosphate, gingko biloba extract, panax
ginseng root extract, camellia sinesis leaf (green
tea), centaurea cyanus flower extract, vitis
vinifera (grape) seed extract, alumina, calcium
carbonate, conchiolin powder, pulverized rose
quartz, pulverized topaz, polymethyl
methacrylate, polyglycceryl-3 diisostearate,
saccharomyces zinc ferment, saccharomyces
copper ferment, saccharomyces manganese
ferment, saccharomyces iron ferment,
saccharomyces silicon ferment, saccharomyces
potassium ferment, fragrance, diazolidnyl urea,
ethylparaben, methllparaben, propylparaben

Mica, titanium dioxide,
iron oxides, yellow 5, red
6, red 7, red 30,

ultramarines, carrnine

 

b)

Salicylic acid

 

4:.

None listed

 

Talc, zea mays/corn starch, dimethicone, zinc
stearate, pentaerthrityl tetraisostearate,
octyldodecyl stearoyl stearate, zeolite, sorbic
acid, methylparaben, propylparaben,
tocopherylacetate, tetrasodim EDTA,
butylparaben, BHT, panthenol

 

None listed

 

Talc, mica, ocyldodecyl stearoyl stearate,
dimethicone, hydrogenated coco-glycerides,
PTFE, methicone, lecithin, methylparaben,
propylparaben, sodium dehydroacetate,

Titanium dioxide, iron
oxides

 

None listed

 

None listed

 

10

None listed

 

11

Talc, mica (and) titanium dioxide,
hydroxyapatite, calcium carbonate, silica,
dimethicone, methylparaben, butylparaben,
mineral oil

Titanium dioxide, iron
oxides, fragrence

 

12

None listed

 

13

None listed

 

 

14

 

None listed

 

 

64

 

 

 

 

 

 

 

 

 

 

 

 

15 Talc, ethylhexyl palmitate, zinc stearate, Ultrarnarines, iron oxides,
caprilic/capric triglyceride, lauroyl lysine, red 7 lake, carmine,
magnesium myristate, methylparaben, sodium bismuth oxychloride,
dehydroacetate, fragrance (parfum), manganese violet, titanium
prpylparaben, ascorbyl palmitate, glycine soja dioxide
(soybean) oil, tocopherol, butylparaben, mica,
iron oxides

16 None listed

17 Talc, ethylhexyl palmitate, zinc stearate, Titanitun dioxide, iron
juniperus communis (juniper) extract, aloe oxides, ultramarines, mica
barbadensis extract, urtica dioica (nettle) extract,
sambucus nigra (elder) flowers, propylene glycol,
tocopherol, methylparaben, propylparaben,
diazolidinyl urea, butylgaraben ascorbyl palmitate

l8 Talc, zinc stearate, octyldodecyl stearoyl stearate, Titanium dioxide
mineral (paraffinum liquidum) oil, ethylhexyl
palmitate, diazolidinyl urea, methylparaben,
isostearyyl palmitate, aloe barbadensis leaf
extract, prOpylparaben, allantoin, tocopheryl
acetate, iron oxides

19 None listed

20 Synthetic wax, aluminum starch, Mica, titanium dioxide,
octenylsuccinate, hydroxylated lanolin, isopropyl iron oxides, carmine,
lanolate, petrolatum, cyclomethicone, cetyl ultramarines, manganese
acetate, acetylated lanolin alcohol, violet, D&C Red No. 7,
methylparaben, propylparaben calcium latke, FD&C red

No. 40, chromium oxide
greens

21 None listed

22 None listed

23 Talc, zinc stearate, octyldodecyl stearoyl stearate, Titanium dioxide
mineral (paraffinum liquidum) oil, ethylhexyl
palmitate, diazolidinyl urea, methylparaben,
isostearyyl palmitate, aloe barbadensis leaf
extract, propylparaben, allantoin, tocopheryl
acetate, iron oxides

24 Talc, zinc stearate, octyldodecyl stearoyl stearate, Titanium dioxide

 

mineral (paraffinum liquidtun) oil, ethylhexyl
palmitate, diazolidinyl urea, methylparaben,
isostearyyl palmitate, aloe barbadensis leaf
extract, propylparaben, allantoin, tocopheryl
acetate, iron oxides

 

 

65

 

REFERENCES

(1) h_ttp://www.cfsag.fda.gov/~dms/cos-toc.html.

(2) Stark, C. A. A Basic Study of Lipsticks Utilizing Microspectrophotometry,
Laser Desorption/Ionization Mass Spectrometry, and Fourier Transform Infrared
Spectroscopy, Michigan State University, 2005.

(3) Williams, D. F. a. S., W.H. Chemistry and Technology of the Cosmetics and
Toiletries Industry; Chapman & Hall: London, 1992.

(4) httpz//www.azco.com/mica/micainfo.php.

(5) International Journal of Toxicology 2005, 24, 65.

(6) Bilek, J. I. Forensic Analysis of Black Mascara via Microscopy, Fourier
Transform Infrared Spectroscopy (FT IR), Scanning Electron Microscopy-Energy
Dispersive Spectroscopy (SEM-EDS), and Visible Microspectrophotometry (MSP),
Michigan State University, 2005.

(7) Choudhry, M. Y. Journal of Forensic Sciences 1991, 36, 366.

(8) Gordon, A.; Coulson, S. Journal of Forensic Sciences 2004, 49, 1244.

(9) Keagy, R. L. Journal of Forensic Sciences 1983, 28, 623.

(10) Romero, L. A. A Protocol for the Analysis of Nail Polish, Michigan State
University, 2003.

(11) Russell, L. W.; Welch, A. E. Forensic Science International 1984, 25, 105.
(12) Ehara, Y. a. M., Y. Forensic Science International 1998, 96, l.

(13) Russell, L. W. a. W., A.E. Forensic Science International 1984, 25, 105.
(14) Jasuja, O. P., and Singh, R. Journal of Forensic Identification 2005, 55, 28.
(15) Rendle, D. F. Chem. Soc. Rev. 2005, 34, 1021.

(16) Douglas A. Skoog, J. J. L. Principles of Instrumental Analysis, 4th ed.;
Saunders College Publishing, 1992.

66

(17) Fleger, S. L., Heckman, John W. Jr., Klomparens, Karen L. Scanning and
Transmission Electron Microscopy An Introduction; Oxford University Press: New
York, 1993.

(18) www.chems.msu.edu/curr.stud/mse.sops/sem.det.htm.
(l9) l_rttp://www.x-raymicroggalysisxom.
(20) James Ingle, S. C. Spectrochemical Analysis; Prentic Hall: New Jersey, 1988.

(21) Smith, A. L. Applied Infrared Spectroscopy Fundamentals, Techniques, and
Analytical Problem-solving; John Wiley & Sons: New York, 1979.

(22) l_rttp://www.forensicmag.com/articles.asp?pid=95.

(23) Edmond de Hoffmann, V. S. Mass Spectrometry Principles and Applications,
Second ed.; Wiley & Sons, Ltd.: England, 2002.

(24) www.asms.org.

(25) www.(_Irnelson.utmem.edu.

(26) www.geo.ucalga;y.ca.

(27) James Miller, J. M. Statistics and Chemometrics for Analytical Chemistry;
Pearson Prentice Hall: England, 2000.

67

llllllllllllllllllllllllllllll

IIIIIIIIIIIIIIIIIIILIIILIIIIIIIIIIIIIII ‘