IIIIIII

I
I

IIIIIIIII

140
449
THS

 

 

7 LhBRAgY
,. Mic igan tate
3W7 University

 

 

 

This is to certify that the
thesis entitled

PERFORMANCE OF THE ULTRAVIOLET IMAGING
SYSTEM (RUVIS) IN VISUALIZING LATENT FINGERPRINTS
ON VARIOUS NONPOROUS AND SEMIPOROUS
SURFACES

presented by

AGNIESZKA NATALIA STEINER

has been accepted towards fulﬁllment
of the requirements for the

MS. degree in Criminal Justice

 

 

 

W Signature
1554 124/4 2 @2,
Date

MSU is an aﬁirmative-action, equal-opportunity employer

 

—.-I-I-o---c-p-o-.-a—.—.— _ - ._ _ ._ _.

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:/CIRC/DateDue.indd-p.1

 

PERFORMANCE OF THE REFLECTED ULTRAVIOLET IMAGING SYSTEM
(RUVIS) IN VISUALIZING LATENT FINGERPRINTS ON VARIOUS NONPOROUS
AND SEMIPOROUS SURFACES
By

Agnieszka Natalia Steiner

A THESIS
Submitted to
Michigan State University

in partial fulﬁllment of the requirements
for the degree of

MASTER OF SCIENCE
School of Criminal Justice

2007

ABSTRACT

PERFORMANCE OF THE REF LECTED ULTRAVIOLET IMAGING SYSTEM
(RUVIS) IN VISUALIZING LATENT F INGERPRINTS ON VARIOUS NONPOROUS
AND SEMIPOROUS SURFACES
By
Agnieszka Natalia Steiner

The Reﬂected Ultraviolet Imaging System (RUVIS) is an extremely effective tool
for seaming large areas to locate ﬁngerprints, as well as a quick and valuable method for
visualizing individual ﬁngerprints, especially on multicolored or reﬂective backgrounds.
Unfortunately, the complicated interaction between a surface, a ﬁngerprint on that
surface, and RUVIS is not well understood. This study seeks to remedy the lack of
understanding by evaluating RUVIS performance on eight very different surfaces.
Fingerprints were collected on glass, glossy paper, vinyl, metal, wood laminate, CD,
expanded polystyrene foam, and plastic. The effects of various measured surface
characteristics, cyanoacrylate fuming, ﬁngerprint donor gender, and ﬁngerprint exposure
to various environments on RUVIS performance were statistically evaluated. It was
discovered that surface roughness, surface wettability, surface energy, and CA fuming

impact RUVIS performance signiﬁcantly, while donor gender and environmental

exposure affect ﬁngerprint quality but not necessarily RUVIS performance.

ACKNOWLEDGEMENTS

This research was performed under an appointment to the US. Department of
Homeland Security (DHS) Scholarship and Fellowship Program, administered by the
Oak Ridge Institute for Science and Education (ORISE) through an interagency
agreement between the US. Department of Energy (DOE) and DHS. ORISE is managed
by Oak Ridge Associated Universities (ORAU) under DOE contract number DE—ACOS-
06OR23100. All opinions expressed in this paper are the author’s and do not necessarily
reﬂect the policies and views of DHS, DOE, or ORAU/ORISE.

Special thanks to the Michigan State Police Lansing Crime Lab, especially Lab
Director F/Lt. Dunkerley and F/Lt. Michaud and the Latent Prints Unit for access to their
facilities and equipment, and for an education in friction ridge examination.

Thanks to Dr. Reza Loloee, the MSU Center for Advanced Microscopy, Michael
Sinke of Speckin Forensic Laboratories, and everyone who donated their time and
ﬁngerprints.

Thanks also to F/Lt. Michaud, Dr. Hoffman, and especially Dr. Ruth Waddell of

the thesis committee, and family and friends.

iii

TABLE OF CONTENTS

LIST OF TABLES ............................................................................................................. vi
LIST OF FIGURES .......................................................................................................... vii
1. INTRODUCTION .......................................................................................................... l
2. THEORY ........................................................................................................................ 5
2.1 Fingerprints as a forensic tool for identiﬁcation ....................................................... 5
2.2 The Reﬂected Ultraviolet Imaging System ............................................................ 15
2.3 Surface chemistry ................................................................................................... 18
2.3.] Surface topography and roughness .................................................................. 18
2.3.2 Surface wettability and surface energy ............................................................ 21
2.3.3 Elemental composition ..................................................................................... 23

2.4 Statistical cluster analysis ....................................................................................... 26

3. METHODS AND MATERIALS .................................................................................. 29
3.1 Surface selection ..................................................................................................... 29
3.2 Surface chemistry ................................................................................................... 31
3.2.1 Surface topography and roughness .................................................................. 31
3.2.2 Surface wettability and surface energy ............................................................ 32
3.2.3 Elemental composition of surfaces .................................................................. 34

3.3 Fingerprint collection and processing ..................................................................... 35
3.4 Fingerprint analysis ................................................................................................. 37
3.5 Statistical analysis ................................................................................................... 39
4. RESULTS AND DISCUSSION ................................................................................... 41
4.1 Introduction and observations ................................................................................. 41
4.2 Surface characterization .......................................................................................... 43

iv

4.2.1 Surface topography and roughness .................................................................. 43

4.2.2 Surface wettability and energy ......................................................................... 46
4.2.3 SEM-EDS ........................................................................................................ 49

4.3 Surface effects ......................................................................................................... 52
4.3.1 Statistical observations of unprocessed control ﬁngerprints by surface .......... 53
4.3.2 Cluster analysis of unprocessed control ﬁngerprints by surface ..................... 55

4.4 Processing effects ................................................................................................... 60
4.4.1 Statistical observations of processed control ﬁngerprints by surface .............. 60
4.4.2 Cluster analysis of processed control ﬁngerprints by surface .......................... 62

4.5 Gender effects ......................................................................................................... 65
4.6 Environmental effects ............................................................................................. 68

5. CONCLUSIONS AND FUTURE WORK ................................................................... 73
5.1 Conclusions and recommendations ........................................................................ 73
5.2 Future work ............................................................................................................. 76
REFERENCES ................................................................................................................. 79

LIST OF TABLES

Table 1. Proﬁlometry data; all scans combined for each surface. ................................... 45
Table 2. Average contact angles and dyne levels. ........................................................... 47

Table 3. SEM-EDS of eight surfaces. Approximate intensity values are given for the
highest energy x-ray line for the element, averaged over three scans. Values are omitted
in instances when the highest energy line overlapped between two or more elements. 50

vi

LIST OF FIGURES

Figure 1. A cross section diagram of friction ridges and the underlying dermal structures.
Based on descriptions in Champod (3). .............................................................................. 6

Figure 2. Fingerprint pattern types (top) and common minutiae (bottom). Images were
generated using SFinGE Version 2.5 (Biometric Systems Lab, University of Bologna;

2001), available at http://bias.csr.unibo.it/research/biolab/sﬁnge.htrnl. ........................... 12
Figure 3. A basic RUVIS schematic. Based on Saferstein and Graf (7). ....................... 16
Figure 4. Optical sectioning in CLSM. Modiﬁed from (16) ........................................... 18

Figure 5. CLSM topgraphical proﬁle generated for EPF (‘Styrofoam’). Note that the
bead structure is clearly visible in the rise and fall of the proﬁle. .................................... 20

Figure 6. Contact angles on glossy paper (left) and plastic garbage bag (right). The acute
contact angle on paper indicates semiporosity while the obtuse contact angle on plastic

indicates nonporosity. ....................................................................................................... 21
Figure 7. Simple schematic of SEM. Based on Reimer (21). ......................................... 23
Figure 8. X-ray emission after loss of a secondary electron (SE). .................................. 25
Figure 9. SEM-EDS spectrum of glossy paper (magazine cover). .................................. 26

Figure 10. Dendrogram showing clustering of eight surfaces based on unprocessed
ﬁngerprints from ﬁve male donors. Surface key: (1) glass, (2) paper, (3) vinyl, (4)
metal, (5) wood, (6) CD, (7) EPF, (8) plastic. .................................................................. 28

Figure 11. RUVIS setup at MSP lab. ............................................................................... 36

Figure 12. The left thumb print of a single donor on (a) glass, (b) wood, and (c) metal-
stored under water; (d)-(i) Show the effects of processing on (a)-(c) respectively. All

photographs taken through RUVIS ................................................................................... 42
Figure 13. CLSM intensity proﬁles. (a) Glass; (b) Paper; (0) Vinyl. ............................. 44
Figure 14. SEM images of (a) metal, (b) plastic, (c) EPF, and ((1) CD. .......................... 49
Figure 15. Grouping of unprocessed control ﬁngerprint scores by surface ..................... 56

Figure 16. Grouping of processed control ﬁngerprint scores by surface. Surface key: (1)
glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF, (8) plastic. .................. 63

Figure 17. Grouping of female processed control ﬁngerprint scores by surface. Surface
key: (1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF, (8) plastic ..... 67

vii

Figure 18. Grouping of processed light exposed ﬁngerprint scores by surface. Surface
key: (1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF, (8) plastic ..... 69

Figure 19. Grouping of processed water stored ﬁngerprint scores by surface. Surface
key: (1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF, (8) plastic ..... 70

viii

1. INTRODUCTION

During his late 19‘h century stay in Japan, Scottish missionary Dr. Henry Faulds
noticed that ﬁnger impressions in clay pots had distinct patterns (1). He became curious
about these patterns and whether they were unique to the individual and consistent
through the individual’s lifetime. F aulds recruited the workers in his clinic to study these
patterns, collecting thousands of ﬁngerprints. The collection included ﬁngerprints taken
before and after mutilation of the ﬁngertips as well as prints taken from children as they
grew into adulthood. Invariably, the patterns remained unchanged and unique. In 1880,
Faulds noticed his medicinal alcohol supply running low and found ﬁngerprints on an
empty bottle. A search of the clinic’s ﬁngerprint library resulted in a match to one of the
employees (1). Since this ﬁrst forensic use, ﬁngerprints have become a mainstay of
identiﬁcation. In fact, today’s ﬁngerprint laboratories are often backlogged and on a
constant lookout for more efﬁcient solutions to the visualization and processing Of
ﬁngerprints (2).

The Reﬂected Ultraviolet Imaging System (RUVIS) has the potential to become
the newest solution to the ﬁngerprint detection and visualization problem. RUVIS is a
hand-held scope which allows for the visualization of latent ﬁngerprints without the need
for chemical processing or powdering. This means that extremely large surfaces, even
entire crime scenes, can be scanned quickly to locate ﬁngerprints that are otherwise
invisible to the naked eye. More time-consuming processing can then focus on areas
where prints are located in order to preserve the ﬁngerprints for transport, storage, and
identiﬁcation. Further, RUVIS can be used to enhance ﬁngerprints located on

complicated surfaces (photographs, magazines, or painted surfaces) by eliminating

distracting backgrounds and making print detail clearer. Finally, a camera can be
coupled to RUVIS and the visualized ﬁngerprints can be photographed directly. All of
these factors could result in faster case turnover for large ﬁngerprint labs, and a
potentially signiﬁcant reduction in backlogs.

Unfortunately, this revolution is not yet possible. RUVIS is known to work only
on certain surfaces, but no detailed analysis has been performed to determine why. This
means that ﬁngerprint examiners presented with a surface have no way to determine
whether or not RUVIS will work. They can be loath to try a technique that may or may
not be a waste of time. In fact, very little published information is available about
RUVIS. Signiﬁcant texts on modern ﬁngerprint methods mention RUVIS only brieﬂy
and fail to consider its potential advantages over traditional methods, which are of course
discussed in depth (3; 4). Several studies have evaluated RUVIS performance in latent
ﬁngerprint detection, but have been limited to broad observations rather than detailed,
statistical analyses. Cubuk presented a general overview of the technique, concluding
that its non-destructive interaction with ﬁngerprints was an advantage over traditional
methods (5). Keith and Runion noted that RUVIS photography of ﬁngerprints on multi-
colored or reﬂective backgrounds reduced background interference, making ﬁngerprints
more clearly visible (6). Saferstein and Graf placed ﬁngerprints on several nonporous,
semiporous, and porous surfaces and provided qualitative descriptions of the resulting
RUVIS photographs, noting color and level of contrast (7). On a handful of surfaces, the
ﬁngerprints were observed over the course of several weeks as they gradually faded.
However, conclusions in this study were limited to a brief observation that RUVIS is not

useful on porous surfaces; no explanation for the phenomenon was provided.

In all the above-discussed studies, no effort was made to link the inherent
chemical and physical properties of the surfaces studied to RUVIS performance, or to
examine the complex interaction between RUVIS, the surface, and the ﬁngerprint itself.
In fact, analysis was largely observational and sample sizes were small, though no
RUVIS study provided a speciﬁc number of ﬁngerprints used or ﬁngerprint donors
involved. Keith and Runion for example, based their entire study on the handful of
ﬁngerprints present on several photographs and one glass clock which had been
submitted as evidence in two cases (6). These ﬁngerprints were of an unknown age and
from an unknown number of donors. Saferstein and Graf also failed to disclose their
sample Size, or their donor demographics (7). It was precisely in response to weaknesses
like these that Jones discussed the need for more scientiﬁc considerations in ﬁngerprint
studies (8). Speciﬁcally, sample Sizes should be more reﬂective of populations of
interest, from size to balance in gender. Further, the aging and storage of ﬁngerprints in
various environmental conditions must be taken into account. These factors were not
fully considered in the aforementioned studies.

The present study therefore seeks to remedy the lack of understanding about
RUVIS performance by approaching the problem from a chemical and statistical point of
view. Sample size was increased over previous RUVIS studies; twelve ﬁngerprint
donors, ﬁve males and seven females, contributed a total of 1,920 ﬁngerprints. Though
the donor demographics were slightly skewed, both genders were included in sufﬁcient
numbers for statistical analysis. The ﬁngerprints were spread among eight semiporous
and nonporous surfaces, chosen to be representative of those typically encountered in

ﬁngerprint laboratories. These eight surfaces were chemically and physically

characterized to determine their roughness, topography, wettability, surface energy, and
elemental composition. Fingerprints on each surface were photographed with RUVIS,
processed with cyanoacrylate ﬁrming, then photographed with RUVIS again. This
resulted in a total of 3,840 RUVIS images of ﬁngerprints.

Finally, statistical analysis was performed to determine how surface
characteristics affect ﬁngerprint quality as visualized with RUVIS. The connections thus
discovered between surface characteristics and RUVIS performance will allow
ﬁngerprint examiners to make somewhat more efﬁcient use of RUVIS technology.
Further, directions for future RUVIS research were suggested, and could eventually

establish the technique as a standard in ﬁngerprint laboratories.

2. THEORY

2.1 Fingerprints as a forensic tool for identiﬁcation

Though the Skin is a large organ, only several relatively small areas (the soles Of
the feet, the palms of the hands, and the ﬁngers) display regular ridges of skin, called
friction ridges (3). The patterns of these ridges are persistent throughout the lifetime and
unique to individuals, making them forensically useﬁrl for identiﬁcation. This constancy
of friction ridge patterns is due to the morphology of Skin and its mechanism of formation
and regeneration. Friction ridge Skin, like all skin, consists of an inner layer called the
dermis (the bulk of the Skin) and an outer layer called the epidermis. This latter in turn
has ﬁve layers, beginning with the basal layer directly adjacent to the dermis and ending
with the horny comiﬁed layer. Basal cells are permanent, and do not change shape or
location. However, other cells form in the basal layer and migrate outward. These cells
gradually die as they move so that the horny layer consists 15 to 20 layers of dead cells
which are Shed regularly. Because the outward migration occurs in layers (adjacent
cells move together), the pattern of the horny layers continues to reﬂect the shape of the
permanent and well—protected basal layer, which in turn reﬂects the shape of the dermis.

The uniqueness of ﬁction ridge skin is a result of its formation during fetal
development (3). Approximately 10 weeks into gestation the dermis has a wave-like
structure. These waves develop into primary dermal ridges over the next 5 to 6 weeks;
their orientation is dependent on the shape of the structures underlying the skin (such as
volar pads) and the relative rates of development of these structures. This process has

some genetic components that account for the inheritance of general ﬁngerprint pattern

types and the differences between male and female prints (males tend to have more
minutiae). However, the shapes of individual ridges, where and how they start or stop,
are determined by a number of largely random physical factors.

Once primary ridges are formed and the ﬁngerprint pattern is permanently set,
secondary ridges begin to develop (3). Primary ridges, separated from one another by
deep groves, become pairs of secondary ridges separated by Shallow furrows. This lasts
approximately until 24 weeks into gestation, at which point the dermis is complete and
the epidermis begins to form, reﬂecting the underlying dermal shape. Ridges in the
friction ridge skin reﬂect a bridging of the deep groove between two adjacent primary
ridges, while furrows in the ﬁngerprint reﬂect the shallow furrows between secondary

ridges (Figure 1). Sweat glands form in the deep grooves and send ducts outward to

 

 

 

Epidermal (friction)

Sweat pore ridge Epidermal

\ I L l m” -
/ L‘ Epidermis
— Dennis
Sweat duct
I I \ —
Secondary dermal . I Groove
ridge Furrow Prrrnary dermal
ridge

Figure 1. A cross section diagram of friction ridges and the underlying dermal
structures. Based on descriptions in Champod (3).

become sweat pores along the tops of corresponding ridges. Some ridges fail to develop
completely before the dermis is ﬁnalized. These are called incipient ridges, and are
usually relatively short and much narrower than other ridges of the same individual.

While existing incipient ridges are persistent, it has recently been shown that new

incipient ridges can develop during the lifetime. For this reason, some ﬁngerprint
examiners counsel against their use in identiﬁcation.

The ﬁngerprints thus formed would be forensically useless were it not for the fact
that they function as stamps. Any material present on the ridges of the ﬁngertip, be it
sweat, other biological secretions, ink, or soot, is transferred to a touched surface.
Material in the furrows, which cannot reach the touched surface, is not transferred. This
results in a reproduction of the ﬁngerprint ridges on the touched surface. If the
transferred material is visible to the naked eye (paint, blood, ink, etc.), the ﬁngerprint is
“patent.” If the materials are invisible to the naked eye (sweat and oils), the ﬁngerprint is
“latent.” Latent ﬁngerprints are much more common in forensic investigations; the exact
nature of “sweat and Oils” thus becomes important to understand.

The sweat pores that dot ﬁngerprint ridges stem from eccrine glands, which are
present throughout the human body (9). However, secretions from sebaceous glands
(present everywhere but areas of friction ridge skin) also contribute to latent print
residues. Sebaceous secretions are transferred to the ﬁngertips when other areas of skin
(such as the face and scalp) are touched during normal daily activity. The composition of
eccrine sweat has been thoroughly investigated, and it has been found that despite its
99% water make-up, over 300 other components are present. These are grouped into
inorganic components such as sodium and potassium, organic components such as amino
acids and proteins, lipids, and miscellaneous enzymes and immunoglobulins. The
secretions of sebaceous glands, called sebum, have been found to contain glycerides,

fatty acids, wax and cholesterol esters, cholesterol, squalene, and other trace organics.

While these individual secretion types have been well analyzed, the speciﬁc
mixture that makes up latent ﬁngerprint residues has not been extensively studied. The
work done to date is discussed by Champod (3). Early studies by the United Kingdom
Home Ofﬁce reported chloride as the main water-soluble component, and that the non-
water soluble components were largely similar to sebum composition (3). Later studies
focused on the changes in latent ﬁngerprint residue with the age of the ﬁngerprint donor.
It was found that children’s ﬁngerprints contain volatile components in higher
percentages, explaining the low incidence of children’s ﬁngerprints in most forensic
investigations.

Fortunately, interest in detailed analyses of ﬁngerprint residues is beginning to
increase. This is due in large part to a ﬁngerprint community consensus indicating a need
for reliable age determination techniques for latent ﬁngerprints (10). It was previously
thought that older ﬁngerprints could be recognized simply based on appearance or the
results of ﬁngerprint processing reactions. It is now understood that a more quantiﬁable
method is necessary. Many studies are therefore attempting to isolate a single component
of ﬁngerprint residues which is universally present and deteriorates at a predictable rate
(9). As potential candidates are isolated, they are tested in a variety of settings, as it has
been agreed that environmental conditions such as humidity and sunlight can affect
ﬁngerprint aging rates (11). The ultraviolet (UV) components of natural light have been
found to be signiﬁcantly damaging to the lipid components of ﬁngerprint residues,
especially squalene.

The complex nature of latent ﬁngerprint residues also means that not only

environmental effects but also the surface on which a ﬁngerprint is placed can have a

signiﬁcant effect on the recoverability of the ﬁngerprint. Surfaces are generally divided
into three broad categories: porous, semiporous, and nonporous surfaces (3). Porous
surfaces like paper or untreated wood absorb the water soluble components of ﬁngerprint
residue almost immediately, while the lipids remain on the surface somewhat longer.
Because the amino acids which soak into the surface do not migrate signiﬁcantly,
staining techniques can be used to visualize the latent print. Nonporous surfaces like
glass and metals do not absorb the latent residues at all; the mixed residue remains on top
of the surface until it degrades. These types of ﬁngerprints can be wiped away. Finally,
semiporous surfaces are the intermediate category, falling somewhere between nonporous
and porous. These are surfaces like varnished wood and glossy papers which absorb the
water soluble components of the print very slowly. Again, lipid residues remain on the
surface until they degrade or are wiped away.

These differential interactions with surface types have led forensic laboratories to
design a wide range of techniques for the visualization or “processing” of latent
ﬁngerprints (3; 12). In all cases, a surface suspected to contain a latent ﬁngerprint is ﬁrst
visually examined and then photographed. This preserves a record of the original surface
Should it, or the print, be damaged during further processing. Next, a large variety of
techniques is available to the ﬁngerprint examiner in order to enhance the ﬁngerprint.
However, only a small handful is used routinely. For porous surfaces, techniques rely on
staining the residues that have soaked into the surface. Ninhydrin is a colorless solution
that reacts with the amine components of eccrine sweat to turn purple. This is an

inexpensive reagent, simple to use, and relatively stable. Other options (diazaﬂuorene,

physical developer, and multimetal deposition) are more expensive, require Signiﬁcantly
more time, or use unstable solutions.

The oldest and simplest method for latent ﬁngerprint development on nonporous
surfaces is powder dusting (3; 12). Particles of the powder adhere to the ﬁngerprint
residues, making ridges and minutiae stand out from the background surface. This is
Simple and inexpensive, and a wide range of powders are available, from traditional
black, white, and gray, to ﬂuorescent or magnetic powders. However, the technique has
downsides; older ﬁngerprints do not take powder well, while some surfaces can cause the
powders to adhere unifome and hence mask any ﬁngerprints.

A more recent and widely used option for ﬁngerprint processing on nonporous
surfaces is cyanoacrylate ﬁrming (3; 12). Cyanoacrylate (CA) is a colorless liquid widely
available as superglue. This liquid vaporizes when heated, then polymerizes
preferentially on ﬁngerprint residues. The process is improved in a high humidity
setting, and works on both eccrine and lipid components of ﬁngerprint residues, though
high levels of lipids seem to improve results. The result is a hard white coating on the
ﬁngerprint. Perhaps the biggest advantage of this technique is that it preserves the
ﬁngerprint, both from normal aging and evaporation and from removal by wiping. The
ﬁngerprint can be ﬁrrther enhanced with powders or ﬂuorescent dyes for easier
visualization. Again, more expensive and difﬁcult options (such as vacuum metal
deposition) are available, but less commonly used.

Semiporous surfaces belong to a problematic group, along with surfaces that have
been wet and adhesive surfaces. This group requires the ﬁngerprint examiner to make

difﬁcult decisions. Fingerprints on adhesive surfaces can be developed using specialized

10

techniques such as sticky-side powder, but simple CA fuming can potentially result in
clear prints with less time and effort expended. Physical developer can be used on wet
surfaces, but the solution is not stable and thus difficult to have on hand. Allowing the
surface to air dry and then proceeding with standard techniques may be preferable.
Finally, semiporous surfaces must be processed either as porous or as nonporous
surfaces. CA fuming will often work, but only if some residue remains on the surface.
Ninhydrin may be successful, unless insufﬁcient amounts of amino acids have soaked
into the surface. Recommendations vary between texts and laboratories, and choice may
come down to the experience of the ﬁngerprint examiner (3; 12).

Once a latent ﬁngerprint has been processed, it is photographed to create a
permanent record and to simplify handling during the identiﬁcation process. To make a
ﬁngerprint identiﬁcation, 3 qualiﬁed examiner follows the ACE-V protocol: analysis,
comparison, evaluation, and veriﬁcation (3). Analysis involves determining whether the
questioned ﬁngerprint is of sufﬁcient quality to be useful. If so, then individual details in
the ﬁngerprint are located. There are three levels of detail: the overall pattern of the
ﬁngerprint (level one detail), individualizing minutiae (level two detail), and the structure
of the ridges themselves (level three detail). There are multiple classiﬁcation systems for
level one detail, though all are based on three pattern types (Figure 2 a—c): arches (ridge
ﬂow starts on one side of the pattern and ends on the other), loops (ridge ﬂow starts on
one Side of the pattern, then reverses direction and exits on the same side), and whorls
(ridge ﬂow is circular, having no beginning or end). Every classiﬁcation system
describes different subclasses of these patterns. While level one detail is a class

characteristic, level two detail is individualizing; it is based on ridge endings, ridge

11

bifurcations, and ridge dots (Figure 2 d-f). However, combinations or variations of these
(Figure 2 g-k), such as islands, enclosures, bridges, crossovers, and spurs, are possible.
Level two details are called “minutiae” or “points.” Finally, level three detail describes
the shape of individual ridges, the arrangement of sweat pores on the ridges, and similar
detail. This is the most individualizing level of detail, but it is not seen often in crime

scene ﬁngerprints.

      
  

(3) Arch (c) Whorl

(g) Crossover (h) Spur (i) Island (j) Enclosure (k)—Bridge

Figure 2. Fingerprint pattern types (top) and common minutiae (bottom). Images
were generated using SFinGE Version 2.5 (Biometric Systems Lab. University of
I IL‘ I I I 1‘ htm‘

Bologna; 2001), available at http://hias oer nm'hn ’

Once all details in the questioned ﬁngerprint have been analyzed, the print can be
compared to a known print. This can be an inked ﬁngerprint collected from a suspect.
Often, however, no suspect is available and the unknown ﬁngerprint is entered into an
Automated Fingerprint Identiﬁcation System (AF IS). This is a computer database system

that allows a ﬁngerprint examiner to enter a questioned ﬁngerprint in a computer, mark

12

any pertinent minutiae, and launch a search (13). AF IS compares the questioned
ﬁngerprint to all others in its database and provides a list of potential matches with
associated match probabilities. The latent print examiner must then select a likely print
from the AF IS list, designate it as the known print, and continue with the comparison
manually. Though AFIS cannot designate a match, it has exponentially expanded the
database available to any given agency, allowed the integration of all databases into the
federal Integrated AFIS (IAF IS), and reduced the time required for a search from days to
hours or even minutes. As of 2005, IAFIS was the largest biometric database in the
world, containing the criminal records and ﬁngerprints of 47 million individuals (14).

The ﬁngerprint examiner compares the level one pattern of the questioned
ﬁngerprint to that of the known print to determine whether it is Similar (3). If so, several
level two details ﬁom the questioned print are selected and the examiner attempts to
locate these same details, in the same conﬁguration, in the known ﬁngerprint. If there is
correspondence, another point from the unknown print is added to the conﬁguration.
This is repeated until all points in the questioned ﬁngerprint are exhausted. If there is
correspondence of level two detail, any visible level three detail is compared in a similar
fashion.

In the third stage of ACE-V, the comparison results are evaluated to determine the
strength of the match (3). Signiﬁcant discrepancies (those which cannot be attributed to
temporary cuts on the ﬁngertip, distortion due to pressure, or smudging of the ﬁngerprint)
at any level of detail result in an exclusion; the questioned and known ﬁngerprints cannot
be said to come from a common source. However, a lack of discrepancies does not

constitute an identiﬁcation. A match in level one detail alone is not individualizing.

l3

Level two and three detail is more selective, but a single point is also not individualizing.
Only a correspondence of a combination of level two or three details can be considered to
constitute a match. The evaluation stage therefore seeks to determine whether the
combination of level two and three points found in both the questioned and the known
ﬁngerprint is individualizing.

At the outset of ﬁngerprint science, numeric standards were established to
determine ﬁngerprint individuality. In 1911, Locard proposed that 12 corresponding
points were unquestionably individualizing, and 8 to 12 corresponding points were
borderline but could constitute a match if the two ﬁngerprints in question were very clear
and if the level one detail was of a rare subtype (3). This led many countries to require a
certain number of matching minutiae to constitute a legal identiﬁcation. Countries such
as Belgium, France, Israel, Ireland, Poland, Japan, and most South American nations
have adopted a stringent 12-point requirement. Germany, Sweden, Switzerland and
Holland are more in line with Locard’s recommendation and require 8 to 12 points to
match, depending on the ﬁngerprint quality. Other countries have set their own
standards, independent of Locard: South Aﬁ'ica requires only 7 points while Italy
requires 16 to 17.

Many ﬁngerprint experts agree, however, that a numeric standard denies the
variability of ﬁngerprints; a perfectly clear ﬁngerprint with no minutiae, for example,
would be extremely rare and potentially individualizing. Many countries are therefore
adopting a “holistic” approach to ﬁngerprint match evaluation (3). Rather than working
to an arbitrarily established number of points, the examiner continues the comparison

until he is satisﬁed, based on his experience, that there is sufﬁcient correspondence for a

14

common donor to be determined. The United States and Canada have subscribed to this
approach Since 1973, when the International Association for Identiﬁcation published its
ﬁndings that there is no scientiﬁc basis for requiring a minimum number of points for a
match (3). The Scandinavian countries followed somewhat later, and Australia and
Switzerland dropped numeric standards in the late 19905. The most recent adherent to
the holistic approach, the United Kingdom (UK), had a 16-point standard until 2001.
However, the UK had been evaluating the numeric standard since 1988, and experts had
been pushing for the holistic approach since 1996. Appeals as early as 1999 had allowed
identiﬁcations with as few as eight points.

Regardless of how a match is evaluated, it mush be veriﬁed to complete an ACE-
V identiﬁcation (3). Veriﬁcation is required because ﬁngerprint experts acknowledge the
subjective nature of their science. Careﬁrl hiring and training of examiners is a crucial
ﬁrst step of veriﬁcation. Further, frequent in-house proﬁciency testing and stringent

quality control fulﬁll the veriﬁcation requirement.

2.2 The Reflected Ultraviolet Imaging System

The detection of latent ﬁngerprints with the use of reﬂected ultraviolet light is a
relatively recent advancement (3). It was noted that, while different surfaces can absorb
or reﬂect UV light to varying degrees and in varying ways, latent ﬁngerprint residues
result in diffuse reﬂection of large portions of UV light (3). This difference between the
print residue and the surface results in contrast. Diffuse reﬂection from ﬁngerprint
residues means that, regardless of the angle of incidence of the UV light, at least some

will be reﬂected upward to the viewer or instrument, and the ﬁngerprint will appear

15

bright. In contrast, surfaces tend to reﬂect light more specularly than diffusely. This
means that the UV light angle of reﬂection will be equal to the angle of incidence.
Unless both the UV light source and the viewer are perpendicular to the surface, very
little UV light is reﬂected to the viewer and the surface appears dark. A reﬂected
ultraviolet imaging system (RUVIS), such as the KrimeSiteTM Scope by Sirchie® and the
Scene ScopeTM by SPEX Forensic Group, takes advantage of this contrast and converts it
into a visible light image. Though different manufacturers introduce their own
innovations, any RUVIS instrument is essentially a combination of a ﬁlter and a UV
image intensiﬁer (7). The ﬁlter allows only UV light to enter the scope to prevent other
light from overpowering it, while the image intensiﬁer uses a photocathode to convert the
UV light to a stream of electrons. These electrons impact a phosphor and result in the
release of a visible light image that is then projected on the eyepiece and photographed

with an attached camera (Figure 3).

 

 

 

 

 

 

 

 

 

Image
Filter Objective intensiﬁer
UV light lens
/ Eyepiece
l
RUVIS

Figure 3. A basic RUVIS schematic. Based on Saferstein and Graf (7).

The major advantages of the RUVIS over traditional methods of latent ﬁngerprint
processing begin with the ability to visualize latents in a non-destructive manner.
Secondly, Space considerations are minimal. CA fuming cabinets can only hold so much
and maintain proper air circulation, and powder can only be applied to so much space
before it becomes impractical. RUVIS on the other hand is able to scan large areas
quickly and efﬁciently. What’s more, RUVIS is Simpler and cleaner to use at a crime
scene than standard methods. It is possible to set up tents or cabinets at a scene and fume
prints on-site, but the logistics can become problematic. Powder is extremely simple to
transport to a scene, but makes an enormous mess, which is not always desirable.
Finally, RUVIS eliminates the color problem of nearly all traditional processing methods.
Powders, CA, ninhydrin, and all other methods result in a ﬁngerprint of a given color.
On a white surface, white CA is invisible. On dark papers, purple ninhydrin does not
contrast. On complex or multicolored surfaces, a Single color powder will only clarify
some portions of the ﬁngerprint. RUVIS, on the other hand, uses ultraviolet light rather
than the visible Spectrum, drastically reducing color interference issues.

The major disadvantage of the RUVIS is that the complex interaction between the
surface, the ﬁngerprint, and the RUVIS is not well understood, nor is there much
available literature in the area. This lack of understanding may be a reason for the
currently limited use of RUVIS in forensic laboratories. However, a better understanding
of surface chemistry effects on RUVIS performance has the potential to make it a part of

the forensic arsenal.

17

2.3 Surface chemistry

This study hypothesized that three main areas of surface chemistry would show
signiﬁcant impact on RUVIS performance by affecting either RUVIS interaction with the
surface or surface interaction with the ﬁngerprint. The physical structure of a surface, its
topography and roughness, could impact both how continuous the ﬁngerprint appears and
how well ultraviolet light is able to reﬂect ﬁ'om the surface. Secondly, surface wettability
and surface energy can affect how well the surface accepts the moisture of a ﬁngerprint
and how well the ﬁngerprint is able to adhere to the surface. Finally, the elemental
composition of the surface may result in reactions with the ﬁngerprint or the ultraviolet

light spectrum.

2.3.] Surface topography and roughness

Surface topography can be measured with confocal laser scanning microscopy
(CLSM), a subtype Of confocal scanning Optical microscopy (SOM) (15; 16). SOM
instruments use pinholes or slits for illumination and imaging, allowing for optical
sectioning. This means a focal plane can be established and all information outside that

focal plane (that is, out of focus) is excluded from the image (Figure 4). In CLSM,a

 

 

 

Laser excitation light
I Achromat
lens Detector
““““ >
3A”.
1 “
_ ____________________ )
i Microscope Primary Aperture
1 lens beamsplrtter
.5313??? :

 

Figure 4. Optical sectioning in CLSM. Modiﬁed from (16).

18

laser hits the sample at the established focal plane and is then reﬂected. The reﬂected
laser light is passed through a pinhole, eliminating all light that was not reﬂected from the
designated focal point. The laser beam is scanned over the sample to produce a full
image.

CLSM can be used to quantify surface topography by measuring the light
intensity of pixels in the image. Higher intensity pixels correspond to projections while
darker pixels represent grooves or depths in the surface. However, a CLSM image can
contain multiple artiﬁcially dark areas. All portions of the image that have been excluded
by optical sectioning appear solid black. The software interprets these as very deep
grooves in the surface. This problem can be remedied by the collection of a z-series,
which is a set of images taken at different focal planes; the focus is adjusted until it is
below the sample surface, then increased in small increments until the focal plane is
above the highest projections in the surface. At each increment, an image is collected.
The z-series images can then be scanned to select a central image in which a long section
of the surface is in focus. A reference line is established along this section, and the light
intensity of the pixels along that line is measured. This results in a topographical proﬁle
with intensity on the vertical axis and distance on the horizontal axis (Figure 5).
Although the intensity measurements may not be exact, the breadth of relative

topographical changes is accurate to the pixel.

l9

 

 

 

 

 

25077777 7777777777
200 1‘
=3 II
0150 . 7 7 7 7 7 7 7 7
1:: II} I
‘II
100 7,i ,l 7 - , , L , ,
5, II, , . [WI
o . .
o 200 400 600 800 1000

Distance (pm)

Figure 5. CLSM topgraphical proﬁle generated for EPF (‘styrofoam’). Note that the
bead structure is clearly visible in the rise and fall of the proﬁle.

Proﬁlometry provides complementary information to CLSM; it accurately
quantiﬁes the vertical variation in a surface, called surface roughness. Though there are
many ways to measure this, use of a mechanical stylus is by far the most common (17).
A sample surface is placed on a stage and a stylus is lowered to contact it with a given
load or pressure. This stylus load must be sufﬁciently small to prevent damage to the
surface, and can range from fractions of a milligram to hundreds of milligrams. The
stage and surface are then slowly moved underneath the stylus, which falls into any
grooves in the surface and rises with any projections. A pick-up point at the opposite end
of the stylus converts the vertical movement of the stylus into electrical signals which can
be quantiﬁed as roughness. This is a robust and effective measurement as long as stylus
load is kept sufficiently small and the stylus tip is chosen to be smaller than any features
of interest in the surface. The instrument is calibrated with sample grids of very exact

roughness.

20

2. 3.2 Surface wettability and surface energy

When a drop of liquid is placed on a solid surface, the solid-liquid interface
results in the droplet taking on a speciﬁc shape and causes the edges of the droplet to
form speciﬁc angles against the surface (18). If that angle is greater than 90°, the liquid
does not wet the surface at all and does not enter any pores on the surface. At an angle of
0° (that is, if no droplet is visible above surface level), the liquid wets the surface fully.
Between these two extremes are grades of wettability, wherein the liquid will wet the
surface incompletely, only some of it entering pores. In a ﬁngerprint context, this means
that deﬁnitions of “nonporous”, “semiporous”, and “porous” are not as clear as they
appear; many surfaces that would be considered nonporous in a ﬁngerprint laboratory
actually have contact angles below 90°.

Contact angle can be measured simply by placing a droplet of liquid on a ﬂat
surface, photographing the droplet at right angles to the surface, and measuring the angle

formed between the tangents along the edges of the droplet and the surface (Figure 6).

 

Figure 6. Contact angles on glossy paper (left) and plastic garbage bag (right). The
acute contact angle on paper indicates semiporosity while the obtuse contact angle on
plastic indicates nonporosity.

Various liquids can be used, and every liquid will have a different contact angle with a

given surface. It is important to maintain consistent droplet size and to work on a level

21

surface. This latter consideration can be a problem with rougher surfaces, as variations in
surface topography may make a very small portion of the surface uneven, resulting in
different angles on either side of the droplet. This can be overcome by repeating the test
for different areas of the surface and averaging the resulting angles.

Surface energy, measured in dynes per centimeter (1 dyne=10'5 N), is related to
wettability. In fact, contact angle measurements can be converted into approximate dyne
levels. However, occasional discrepancies between the two measurements mandate that
both tests be performed for a thorough understanding of the surface. While wettability
measures the amount of liquid entering the surface, surface energy can be used to
measure how well liquid adheres to the surface (18). Various tests are available for
surface tension, among them Accu Dyne TestTM Marker Pens (Diversiﬁed Enterprises,
Claremont NH). This test consists of a series of markers whose inks are composed of
forrnamide and ethyl CellosolveTM (2-ethoxyethanol) (19). The mixture is adjusted so
that each marker has a Speciﬁc dyne level. When the ink is applied to a sample surface,
there is an interaction between the forces of solid surface energy and liquid surface
tension. This causes the ink to bead up or spread out, depending on which forces are
greater; the solid surface energy must be greater than the liquid surface tension in order
for adhesion between the two to be strong (spreading liquid). Consecutive markers can
be used until a boundary is established between heading and Spreading—this is the dyne

level Of the surface.

22

2.3.3 Elemental composition

Scanning electron microscopy-energy dispersive x-ray Spectroscopy (SEM-EDS)
is a combined technique that allows both microscopic observation and the determination
Of elemental composition. Scanning electron microscopy improves on light microscopy
techniques by increasing the depth of view, thus allowing a clearer image of sample
morphology regardless of roughness (20). Samples for SEM must be stable in a vacuum
and must be electrically conductive. For samples that are not inherently conductive, a
sputter coater can be used to cover the sample in a thin layer (10-30 nm) of gold particles.
Because the particles are smaller than the minimum resolution of the microscope, no
topographical changes result from this coating. Samples are inserted into the

microscope’s vacuum chamber through an airlock system and placed underneath an

electron gun such as a lanthanum-hexaboride crystal (LaB6 or “lab-6”), a tungsten loop,

E[;j Electron gun

Condensor lens

or a tungsten wire (Figure 7).

Anode—

9"le

Scan coilsl:[]

l: :IObjective lenses

Aperture —-
' ;Electron beam

[:1 W?SE I:I
X-ray detector O SE detector

 

 

Figure 7. Simple schematic of SEM. Based on Reimer (21 ).

The electron gun is heated, producing a cloud of electrons which can be focused

into a beam and sent towards the sample by an accelerating voltage. When the electron

23

beam strikes the sample, the electrons interact with the sample and collide with its
component atoms, gradually losing energy and spreading out in a teardrop-shaped cloud.
Some of these collisions result in orbital shell electrons within the sample being knocked
free. When this happens close to the sample surface, these low-energy secondary
electrons (SE) can leave the sample. SE are then diverted to a Faraday cage, converted to
photons, and used to produce the sample image. When the electron beam is focused on a
projection in the sample surface, a larger percentage of the electron cloud is near the
surface and more SE are produced, resulting in a brighter point on the image.

SEM-EDS is an extension of SEM that allows the determination of the sample’s
elemental composition by measuring x-rays emitted from the sample (20). Because these
x-rays can be absorbed by gold, gold coating samples for SEM-EDS is
counterproductive. Gold can also release x-rays that mask those of other elements.
Carbon, on the other hand, is a conductive material that does not absorb x-rays. Further,
the x-rays carbon emits are unlikely to mask those of other elements. Samples for SEM-
EDS are therefore carbon coated.

Once the sample is carbon coated, it is inserted into the instrument exactly as for
SEM micrOSCOpy; the electron beam again reacts with the sample, and SE are produced.
The focus now shifts, however, to the impact electron loss has on a sample atom. The
departure Of an SE results in a vacancy in one of the atom’s inner shells (21). This is an
energy imbalance and must be remedied by an outer Shell electron falling to ﬁll the
vacancy. This electron, moving from a higher energy Shell to a lower energy Shell,

releases excess energy in the form of an x-ray (Figure 8).

24

 

Figure 8. X-ray emission after loss of a secondary electron (SE).

Several factors make this x-ray characteristic, and therefore useful for element
identiﬁcation. First, electron shells are quantized, so the possible x-ray energies are also
quantized rather than continuous. Second, this quantum x-ray energy is proportional to
the atomic number of the element and the energy separation of the electron shells
involved. This means that an element can emit several characteristic x-rays of different
energies, depending on which shell was vacated by the SE, and which shell donated the
replacement electron. The combination of x-rays produced by an element is unique to
that element.

The characteristic x-rays released in SEM—EDS are collected by an x-ray detector
and are converted to a charge pulse. The magnitude of this pulse is proportional to the
energy of the x-ray. The pulse is converted into a signal which is then sent to a computer.
Software then places the signal into an energy spectrum. Once x-ray collection is
completed, the resulting spectrum displays a series of peaks or “x-ray lines” at quantized
energies (Figure 9). The energy of the x-ray is characteristic of the element and the x-ray

intensity is proportional to the concentration of the parent element in the sample. Hence,

25

both qualitative and quantitative information regarding the elemental composition of the

sample is possible.

 

 

 

1 500
1000
3‘
§
.5 500
Fe Fe
n l I H‘ V
0 6 7

Energy (keV)

Figure 9. SEM-EDS spectrum of glossy paper (magazine cover).

2.4 Statistical cluster analysis

When multiple factors interact to deﬁne a single object, as surface characteristics
interact to deﬁne a surface, statistical comparison of objects based on individual factors
can be misleading. Instead, it may be better to compare objects based on all their
component factors simultaneously. Cluster analysis is one approach to this type of
analysis, as it groups objects based on similarity (22; 23). For example, some numeric
value designating ﬁngerprint quality can be considered as a variable on a surface. Cluster
analysis can then group different surfaces based on the impact they have on ﬁngerprint
quality. This takes into account all the surface characteristics as well as the relative
importance of their impact on ﬁngerprint quality.

Data for cluster analysis is presented as a matrix Of objects (surfaces, for example)
and variables for each object. The objects are mapped onto an n-dimensional space,

where n is equal to the product of objects and variables. Distance between the objects

26

can then be calculated. Though several different distance calculations are available,

Euclidean distance is the most common choice. The Euclidean distance (d) between two

objects x and y with coordinates (x1, x2, X") and (yl, yg, yn) is calculated as:

“er2309in (1)

There are also several choices for the reference points between which distance is
calculated. Most commonly, single linkage is used. This means that distance is
measured between the nearest elements of neighboring clusters. Distances between all
possible object pairs are calculated and plotted in a distance matrix, which is then used as
the basis for cluster formation.

Hierarchical cluster analysis (HCA) is an agglomerative method of cluster
formation, wherein each object is initially designated as a separate cluster. Distances
within the distance matrix are then compared and clusters separated by the smallest
distance are grouped into a new cluster. This process is repeated with progressively less
similar objects until all objects are in a single cluster; once formed in this method,

clusters are never split. The similarity (s) between objects in a cluster is calculated based

on the distance between those Objects (dij) and the maximum distance in the distance

matrix (dmax):
1—d--
s = 100[—” j (2)
max
. The resulting HCA output is a list of clusters and associated similarity levels.
Graphically, this can be displayed as a tree diagram or dendrogram (Figure 10). The

dendrogram can be ‘cut’ at a given similarity level so that only clusters above that level

27

are considered Signiﬁcant. For example, cutting a dendrogram at 90% similarity
indicates that clusters with similarity 90% or greater are signiﬁcant, while all objects that

cluster with less than 90% Similarity are considered as individual objects.

 

 

36.34-

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E 57.564 I j
78.78~
90.00
' 1
100.00 .
1 2 3 5 7 8 6 4
Surfaces

Figure 10. Dendrogram showing clustering of eight surfaces based on
unprocessed fingerprints from ﬁve male donors. Surface key: ( 1) glass, (2) paper,
(3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF, (8) plastic.

28

3. METHODS AND MATERIALS

3.1 Surface selection

Eight surfaces were selected for this study: glass, plastic, glossy paper, CDS,
metal, wood laminate, expanded polystyrene foam (EPF, frequently called ‘styrofoam’),
and vinyl. These surfaces are commonly encountered in ﬁngerprint cases and range
from easy (glass) to challenging (vinyl), both for traditional methods and for RUVIS.
The ﬁrst, glass, is a remarkably good surface for taking and visualizing ﬁngerprints. It
was thus selected as a control for ﬁngerprint quality. Donors unable to leave clear
ﬁngerprints on glass would in general result in lower print quality regardless of surface,
processing, or RUVIS performance. In order to have uniform glass quality for all
ﬁngerprints, pre-cleaned 25 x 75 mm Cosmopolitan microscope slides (VWR Scientiﬁc
Inc., West Chester PA) were used. Due to the limited quantity of slides per box, multiple
boxes were used. However, all were from lot number 010684.

Four smooth, but in some ways problematic, surfaces were then chosen. Polished
metal is a very smooth, nonporous surface. The smooth or ‘Shiny’ side of Reynolds
Wrap® Quality Aluminum Foil (Alcoa Inc., Pittsburgh PA) was used as the metal
surface. One 75 sq. ft. package was more than sufﬁcient for the study. This had the
beneﬁts of being easy to handle and obtain while at the same time being chemically and
physically identical to polished aluminum surfaces of many types. Second, plastic was
represented by Best Buy black plastic trash bags (Quality Transparent Bag Co., Bay City,
MI). Again, only one package of ten 26-gallon bags was required. This type of plastic

was chosen because plastic bags of all types are very commonly submitted to ﬁngerprint

29

labs for processing, often in bulk amounts that would lend themselves well to the
screening capabilities of RUVIS. CDs were chosen as the third smooth surface because
they are typically thought to reveal ﬁngerprints well, but upon closer examination often
cause discontinuous ridges. Memorex® 700 MB CD-R disks (Imation Corp., Oakdale
MN) were used, with prints always placed on the underside of the disk. Finally, glossy
magazine covers were used as the paper surface. Covers from 2004-2006 issues of
Vogue and Cosmopolitan magazines were collected for the study. Glossy papers are
frequently handled, but oﬁen present difﬁculty in ﬁngerprint laboratories due to their
semi-porous nature and the distracting background they provide to ﬁngerprint
identiﬁcation.

Three textured surfaces were also selected. Wood laminate, such as that used for
kitchen cabinets and countertops, as well as various fumiture items, is a semi-porous
surface with a consistent texture rather than a wood grain. Samples of Wilsonart
Laminate (ITW, Glenview IL) were obtained from a Lowe’s hardware store in south
Lansing, MI. The second textured surface chosen was expanded polystyrene foam (EPF).
This is the material commonly used to make disposable plates and cups for hot foods and
liquids, and is generally called ‘Styrofoam.’ However, StyrofoamTM is a registered
trademark for an insulation material manufactured by the Dow Chemical Company. As
Dow is fond of stating, “there isn’t a coffee cup in the world made from Styrofoam”
(24). The term ‘EPF’ will therefore be used throughout this study. Heﬁy® brand 8.875
in. diameter Everyday Plates (Practiv Corp., Lake Forest IL) were used to eliminate the
curvature of most other EPF items. The underside of the plates was used, and all plates

were drawn from a single package. Lastly, the American Class Co. 05 “Spirit” polyvinyl

30

chloride (available in JoAnn stores, Lansing MI) was used to represent the vinyls found
on some furniture, car upholstery, and similar items. Vinyl is a remarkably difﬁcult
surface; though it is not more textured than EPF, it is extremely rare for usable
ﬁngerprints to be found on vinyl surfaces.

Each surface type was cut into samples large enough to easily accommodate ﬁve
ﬁngerprints and all necessary identifying labels. Exceptions to this were the CDs (one
CD constituted one sample), glass Slides (three Slides joined together constituted one
sample), and the wood larrrinate (three pieces of laminate joined together constituted one
sample). One sample of each surface was set aside for surface chemistry analyses, and

the remainder were used for ﬁngerprint collection.

3.2 Surface chemistry

Techniques used to analyze the surfaces were selected after a review of the
literature, and fell into three broad categories (18; 25). First, surface topography and
roughness categorized the physical surface structure. Next, surface wettability and
surface energy were measured to determine surface interaction with a ﬁngerprint.
Finally, each surface was photographed at high magniﬁcation and its elemental

composition was determined.

3. 2.1 Surface topography and roughness
Surface topography was measured with the LSM 5 Pascal laser scanning
microscope (Carl Zeiss Inc., Oberkochen Germany) operated by a trained technician in

confocal reﬂective mode, with a 488 nm laser and a 10x objective. A clean, small sample

31

of each surface was used and a z-series of images was collected at plane intervals of
approximately 5.5 pm. The exact plane interval was set automatically by the software for
each surface, and the number of planes was selected to ensure that every portion of the
visible surface was in focus in at least one plane. The z-series of each sample was then
observed in order to select a plane in which as much of the sample as possible was in
focus, especially along the center horizontal cross-section. This was usually one of the
central planes. The image intensity proﬁle was then mapped along the central horizontal
cross-section. One sample of each surface was analyzed, as pertinent topographical
features spanned only 100-200 um compared to the 1,000 um length of each
topographical proﬁle.

Surface roughness was measured using the Dektak IIA Surface Proﬁle Measuring
System, model SHS-O440 A (Sloan, Santa Barbara CA). A stylus with a 12.5 urn tip was
tracked across 5 mm of the surface with a tracking pressure below 50 pg. A low scan
speed was used in order to maximize resolution. The scan was repeated for three areas of
each surface sample. The sample was then rotated 90° and three more scans were
performed. This was done to account for any potential unidirectional texture in the
surfaces. For each scan, the software calculated an average roughness value (RA) by
determining a mean or center line through the trace and then computing a standard

deviation from that line.

3. 2.2 Surface wettability and surface energy

Surface wettability was measured using the VCA 2000 Video Contact Angle

System (AST Products Inc., Billerica MA) and the VCA Optima software. The system’s

32

micro-syringe was used to dispense a 2 uL droplet of deionized (DI) water onto each
surface sample, which was placed on the leveled stage. An image of the droplet was
captured and ﬁve reference points were set on the droplet manually—the two droplet
corners, the droplet top, and two points along the edges. The software then used these
points to outline the droplet and calculate the tangential angles formed by each edge of
the water droplet against the surface. This process was repeated on ﬁve different areas of
each sample, resulting in a total often angles calculated for each surface. The average of
these ten angles was considered the surface contact angle.

Surface energy was obtained with Accu Dyne TestTM Dyne Level Marker Pens
(Diversiﬁed Enterprises, Claremont NH). A sample of each surface was wiped clean and
placed in the fume hood and a pen was pressed to the surface until its tip was saturated.
The pen was then lightly drawn across the surface in a one-inch stroke. This was
repeated twice more, and a timer was started as soon as the third stroke was completed.
The timer was stopped when the ink swath headed up, tore apart, or shrank to a thin line,
and the time was recorded. If the time was 1 second or less, a lower dyne level marker
was used. If the time was 3 seconds or more, a higher dyne level marker was used. If the
time was between 1 and 3 seconds, the dyne level of the marker matched the surface
energy. The test was repeated until a positive result was Obtained three times with the
same dyne level, and conﬁrmed with the next higher and next lower markers as available.
This procedure was recommended by the test manufacturer, and the time intervals are

based on standard testing of polyethylene and polypropylene ﬁlms.

33

3. 2.3 Elemental composition of surfaces

Samples were prepared for SEM by removing one section measuring
approximately 5 mm square from each surface sample. The sections were then mounted
on individual stubs and gold-coated with the SC 500 sputter coater (Emscope, Ashford
Kent, England). SEM was performed with a JSM-64OO scanning electron microscope
with a LaB6 emitter (JEOL Ltd., Tokyo Japan). Four stubs were inserted at one time, and
images were captured with 15 kV accelerating voltage, a 15 mm working distance, a 20
us pixel time, and medium resolution. Each sample was captured at both 100x and 1000x
magniﬁcation, ensuring that pertinent morphology was visible at both magniﬁcations.

Sample preparation for SEM-EDS involved removing three 5 mm square sections
from different areas of each surface sample and attaching all three replicates of two
surfaces to a large stub. The samples were then carbon coated with a carbon evaporator
(Ladd Research Ind., Burlington VT). Analysis was again performed on the JSM-6400
SEM. Accelerating voltage was set to 20 kV, the working distance was 15 mm, and
magniﬁcation was maintained at 1000x to limit the chance of dust particles being
included in the scan. The scan was performed at 10 eV per channel, from 0 to 20 keV,
with a live time of 120 seconds. Dead time was maintained between 30% and 40%
throughout the scan by adjusting gun alignment and condensor lens settings. The scan
was repeated for each section of sample, resulting in three elemental composition scans

for each surface.

34

3.3 Fingerprint collection and processing

Fingerprints were collected from 12 volunteers in their 20S and 305, 5 male and 7
female (donor participation was approved by UCRIHS IRB #06—245). Each subject was
asked to refrain from washing their hands immediately prior to their session in order that
their prints reﬂected an accumulation of oils and moisture. If hand-washing was
unavoidable, the subject was asked to gently touch their ﬁngertips to their forehead or
scalp in order to replace some of the removed oils. The surface sample was wiped
thoroughly clean to remove all pre-existing ﬁngerprints and as much dust as possible.
The subject then touched one ﬁnger at a time to the indicated area of the surface sample.
An identifying label noted the location of each individual ﬁngerprint. This was repeated
twice for each hand, resulting in 20 ﬁngerprints collected from each subject on four
samples of the surface. This procedure was repeated on eight individual occasions for the
eight different surfaces.

The four samples of surface thus collected were then exposed to different
environmental conditions. One sample was designated a control surface, and was
immediately packaged in a tented manila envelope for transport to the Michigan State
Police (MSP) latent print laboratory, where processing was to occur. The sample was
transported overnight and processed the following morning. The second sample was
placed in a drawer which was then closed to eliminate light. The third sample was placed
approximately 1 foot below a simulated sun lamp (Exo Terra® ReptiGlo 8.0 ﬂuorescent
bulb (Rolf C. Hagen, Mansﬁeld MA), visible light plus 33% UVA and 8% UVB), and the
fourth was suspended in a tank of DI water. The latter three samples were left

undisturbed for one week, then collected, packaged, and transported overnight just as the

35

control sample had been. The water samples were allowed to air dry before being
packaged.

At the MSP latent print laboratory, all samples were treated the same regardless of
environmental condition. Each surface sample was photographed with a Canon
Powershot G5 digital camera, then placed on a stage above which the Sirchie
KrimeSiteTM Imager RUVIS was suspended (Figure 11). The room lights were turned off
to prevent ambient light washout, and a short wavelength (254 nm), 4 amp UV lamp was
used to illuminate the surface. The image of each ﬁngerprint (its labeled location if no
print was visible) was centered in the RUVIS view. The focus and UV lamp placement
were adjusted until the image had as much contrast and clarity as possible. The camera
was then attached to the RUVIS and a photograph was taken. In all cases, the camera
was set to take photographs on the macro setting, with the autofocus function. The
process was repeated for each of the ﬁve ﬁngerprints on the surface sample, and then for

all surface samples.

l'\' light
\IIIII'CL‘

 

Figure 11. RUVIS setup at MSP lab.

36

Once all the samples intended for analysis that day had been photographed, the
surfaces were placed in a Hamilton fuming cabinet (Thermo Fisher Scientiﬁc Inc.,
Waltham MA). A 250 mL beaker of hot water was added to the cabinet, and a quarter
sized dollop of Hot Stuffm cyanoacrylate (Satellite City, Simi Valley CA) was placed in
an aluminum foil ashtray atop a hot plate. Finally, a control print was placed on a sheet
of plastic and added to the cabinet. Fuming was set for 15 minutes, followed by a 10
minute venting period. The surface samples and control print were observed for visible
development. If development was incomplete or invisible, the fuming cycle was repeated
with fresh hot water and cyanoacrylate.

After fuming, the surfaces were once again photographed exactly as before
fuming. Focus and UV light placement were adjusted to each individual print. The
surface samples were then repackaged for transport and storage. The control prints used
for fuming were also photographed for a record, which was especially important in cases

where no development was visible on the test surfaces.

3.4 Fingerprint analysis

The resulting ﬁngerprints were then visually examined. This necessitated several
stages of digital image enhancement, performed in Photoshop® 7.0 (Adobe, San Jose
CA). As the original RUVIS images are in shades of green and thus difﬁcult to see
clearly, each image was converted to grayscale. This resulted in a pale gray to white
ﬁngerprint on a dark gray to black background. In keeping with conventional ﬁngerprint
analysis, these images were inverted so that the print appeared dark on a light

background. Finally, the color saturation, contrast, and brightness of the image were

37

adjusted until the print was as clear as possible. The image was converted back to color
so that minutiae could be marked clearly.

The search for minutiae was performed by visually following three ridges at a
time across the print, moving from top to bottom one ridge at a time. This meant that
each ridge was examined several times. Further, the two neighboring ridges served to
keep the examination from slipping to an adjacent ridge. Standard Galton point minutiae
were marked: ridge endings, bifurcations, islands, dot ridges, enclosures, bridges,
crossovers, and spurs. Care was taken to avoid including incipient ridges. The search for
minutiae was conservative; if there was any doubt as to whether a detail was a Galton
point, it was not marked.

All located minutiae were marked with 9-pixel diameter red dots; this Size was
found to be clearly visible while at the same time not likely to obscure the detail it was
meant to mark. For each image, three items were noted and scored: whether or not the
print was visible at all (one point), whether any ridge structure was visible in the print
(one point), and the number of minutiae marked (one point per detail). The sum of these
factors resulted in a numerical “score” for the ﬁngerprint, with higher scores indicating a
higher quality print. For example, a smudged print with visible ridges but no minutiae
would have a score of 2, while a clear ﬁngerprint with 18 minutiae would have a score of
20. Each ﬁngerprint was photographed and visually examined before and after
processing, resulting in two scores for every print.

A sample of six ﬁngerprints images, from different surfaces, different subjects,
and varying in quality from poor to excellent was sent to a certiﬁed ﬁngerprint examiner

for conﬁrmation that minutiae were marked correctly. His conclusion was that the

38

assignment of minutiae was conservative but correct. That is, all points marked were
reliable minutiae, and those not marked were difﬁcult to assign and therefore

appropriately left out of a conservative analysis.

3.5 Statistical analysis

First, basic statistics (average score, standard deviation, etc.) were established for
each surface based on the control data set, both before and after processing. Signiﬁcant
differences were identiﬁed using one-way balanced ANOVA and Tukey’s comparisons.
Next, the complex interactions of factors that contribute to a ﬁngerprint score were
assessed by hierarchical cluster analysis (HCA). Matrices were designed with surfaces as
objects (rows) and individual ﬁngerprints as variables (columns). Initially, one 8x60
matrix was created for all unprocessed data, and another identical matrix for processed
data. The data was then Split into four smaller matrices based on gender: female
unprocessed, female processed, male unprocessed, and male processed. MinitabTM 15
(Minitab Inc., State College PA) was used to perform HCA on the matrices in order to
assess similarity among surfaces in terms of the ﬁngerprint quality observed with RUVIS
before and after processing. Single linkages and Euclidean distance were used.

Next, the unprocessed and processed data were compared using a paired t-test or
Wilcoxon Signed Rank test. For each surface, one data set was established with control
prints from all 12 donors, and another for the processed equivalents. The difference
(unprocessed subtracted from processed) was calculated and the normality of the
resulting data set was tested. If the differences were normally distributed, a paired t-test

was performed for that surface. If the differences were not normally distributed, a

39

Wilcoxon Signed Rank test was performed. The difference between male and female
ﬁngerprints was tested for signiﬁcance in a similar manner. However, since the data sets
were unbalanced (seven females and ﬁve males), two female subjects were randomly
eliminated. Because this last comparison was performed only to indicate potential
differences and determine the necessity of running separate cluster analyses for the two
genders, the comparison was not repeated for every possible combination of eliminated
females.

Finally, data from the four environmental conditions was evaluated. HCA was
performed for each environment as for control data. The matrices included both male
and female donors. Two surfaces were selected for further analysis: glass and wood.
The former represented surfaces on which ﬁngerprint scores tended to be high while the
latter represented surfaces with poorer ﬁngerprint scores. Neither of these two surfaces
was an extreme, however. One-way balanced ANOVA was carried out to compare all
four possible conditions: control, light exposed, dark stored, and wet stored ﬁngerprints,
and to determine the effect of environment on the print as observed by RUVIS. If
ANOVA indicated signiﬁcant differences were present, Tukey’s comparisons were

performed to identify which differences were signiﬁcant.

40

4. RESULTS AND DISCUSSION

4.] Introduction and observations

Despite the explorative nature of this study, several hypotheses were made prior
to its beginning. These were based on the previous qualitative studies of RUVIS and on
general ﬁngerprint texts (3-7). It was expected that higher surface porosity and higher
surface roughness would negatively impact RUVIS performance. That is, the semiporous
and rough surfaces (paper, wood, EPF, and vinyl) were expected to result in
discontinuous RUVIS visualized ﬁngerprints with few or no minutiae. The nonporous
and smooth surfaces (glass, metal, plastic, and CD) were expected to result in high
minutiae counts and hence high ﬁngerprint scores. All three environmental conditions
(exposure to light, storage in the dark, and storage under water) were also expected to
reduce ﬁngerprint quality and RUVIS performance, though to different extents. Storage
in darkness was hypothesized to be the least damaging, as only the time lapse between
deposition and processing would impact the ﬁngerprint. Simulated sunlight was expected
to be more damaging due to ultraviolet radiation, which has been shown to deteriorate the
lipid components of ﬁngerprints (11). Finally, storage under water was expected to wash
away water soluble components of the ﬁngerprint, thus reducing ﬁngerprint quality most
drastically. Conversely, cyanoacrylate (CA) fuming was expected to improve RUVIS
performance for all surfaces and all environmental conditions, compared to the
unprocessed ﬁngerprints.

Cursory visual observations made throughout the ﬁngerprint collection and

analysis process seemed for the most part to bear out these expectations. Rougher

41

surfaces like wood or more porous surfaces like paper tended to result in ﬁngerprints with
less visible detail than smooth and nonporous surfaces like glass (Figure 12 a-b). Both
pictured ﬁngerprints were donated by the same individual, yet while the ﬁngerprint on

glass (Figure 12 a) shows clear ridges and minutiae, the print on wood (Figure 12 b)

b c
e f

Figure 12. The left thumb print of a single donor on (a) glass, (b) wood, and
(c) metal-stored under water; (d)—(f) show the effects of processing on (a)—(c)
respectively. All photographs taken through RUVIS.

shows no ridge structure.

    

(1

Further, simple visual examination implied that most ﬁngerprints visualized with
RUVIS were improved with CA processing (Figure 12 d-f). While the unprocessed print
on glass is clear (Figure 12 a), that same print aﬁer processing (Figure 12 d) has more
contrast against the glass surface. The print on wood underwent an even more
remarkable transformation; CA fuming was able to develop ridge structure (Figure 12 e)
in the previously blank print (Figure 12 b). Finally, the faint water submerged print

(Figure 12 0) showed increased contrast after processing (Figure 12 f). The

42

environmental factors were somewhat more difﬁcult to interpret without a numerical
measure of quality. Fingerprints stored under light and in the dark appeared to be of very
Similar quality. Both tended to Show somewhat less contrast than control ﬁngerprints,
but did retain numerous minutiae. However, ﬁngerprints stored under water certainly
seemed to have faint ridge structure (Figure 12 c). It is precisely this sort of
observational uncertainty that makes statistical analysis critical when assessing the
performance of RUVIS on different surfaces and under different conditions, both before

and after CA fuming.

4.2 Surface characterization

4. 2.1 Surface topography and roughness

Confocal laser scanning microscopy (CLSM) surface intensity proﬁles served to
quantify the observable topographical variations of each surface. Based on these
variations, the eight surfaces fell into three broad topographical categories. The ﬁrst
category consisted of glass and CD, both of which resulted in relatively smooth and ﬂat
proﬁles (Figure 13 a). However, there were occasional features in these proﬁles not
explainable by surface structure. These always consisted of a sharp rise followed by a
large, Sharp fall within 5 pm and were likely caused by dust particles that had settled on
the surface between cleaning and the initiation of the scan. A dust particle illuminated by
the laser would appear extremely bright, and any roughness in the dust would result in a
miniscule Shadow adjacent to it. The software would interpret this type of lighting as a

topographical rise immediately followed by a groove.

43

 

O 200 400 600 800 1000
Distance (pm)

Figure 13. CLSM intensity profiles. (a) Glass; (b) Paper; (c) Vinyl.

The second topographical category consisted of paper and wood, both of which
showed large surface variation with no discernible pattern (Figure 13 b). These two
surfaces were remarkably similar in breadth of topographical changes, but the intensity
range of any given topography in wood was approximately twice that of paper. The ﬁnal
category consisted of metal, vinyl, plastic, and EPF. These four surfaces showed
topographicalvariations characterized by rolling projections spanning 100-200 pm at
irregular intervals (Figure 13 c). On plastic and metal, these projections most likely
corresponded to the striae left on the surfaces by the manufacturing process. These were
widely and randomly spaced on plastic, as only three such projections were observable
throughout the 1000 um proﬁle, separated by distances varying ﬁom 100-400 pm. On
metal, the projections were relatively regular and closely packed, separated by 50 pm or
less. Finally, EPF and vinyl exhibited the rolling projections that characterize their

observable surfaces. The beaded structure of EPF was clearly visible in its topographical

proﬁle (see Chapter 2, Figure 5) while the leathery or scaly texture of vinyl resulted in
peaked topographical features.

These topographical proﬁles were further quantiﬁed by the roughness
measurements obtained with proﬁlometry; while CLSM illustrated the breadth and
spacing of topographical features, proﬁlometry quantiﬁed their height as a measure of
roughness. Topographically, glass and CD were similar, paper and wood formed a
second group, and vinyl, EPF, metal, and plastic fell into a third group. This third group
was further subdivided into two sub-groups: metal and plastic in one and vinyl and EPF
in the other. However, when the eight surfaces were ranked in order from lowest to
highest average roughness, the groups previously observed with CLSM changed (Table
1). Glass and CD were both much smoother than the remaining six surfaces, but CD was

Table l. Proﬁlometry data; all scans combined for each surface.

 

 

 

 

 

 

 

 

 

Surface Average Max. Average Roughness
projection projection (RA)*
height* height* n=6
Glass 0.098 0.28 0.072 i 0.004
CD 0.44 0.92 0.21 i 0.08
Plastic 0.46 7.3 1.2 i 0.2
Paper 2.0 8.3 1.7 :1: 0.9
Metal 3.7 13 2 :t 1
Wood 1.7 110 4.3 :h 0.7
Vinyl 13 45 10 :l: 2
EPF 7.0 49 18 :t 2

 

 

 

 

 

 

* All values given in um.
nearly three times rougher than glass. Further, proﬁlometry illustrated that, while plastic,
metal, vinyl and EPF showed Similar topographical formations, the differences in size
and spacing of these features resulted in plastic and metal being much less rough. It is

worth noting that the large amount of variation in the metal roughness variations was due

45

to the highly unidirectional nature of the striae. These contrasts between topographical
and roughness data Showed that surface structure is quite complex, and that both the
height and breadth of surface features must be taken into account in order to provide a
clear understanding of surface structure.

The topographical features and roughness of a surface impact both ﬁngerprints
and RUVIS interactions with the surface. Rougher surfaces scatter more ultraviolet light
than smoother surfaces. AS more UV light is scattered by the surface rather than
reﬂected, more noise will be present in the RUVIS image. Topography complicates this;
if roughness is present but only in the form of widely Spaced topographical features (such

as seen here with plastic) it may not affect enough of a ﬁngerprint to reduce image

quality.

4. 2.2 Surface wettability and energy

Contact angle measurements resulted in four groups of surfaces, though these did
not correspond to the topographical or roughness groupings (Table 2). Instead, these
classes were based on contact angle deﬁnitions of porosity. Contact angles of 90° or
higher indicate a nonporous surface, and contact angles of 0° indicate a completely
porous surface. Technically, semiporous surfaces can have any contact angle between
these two values, but this is a wide range and can be subdivided into levels of

semiporosity.

46

Table 2. Average contact angles and dyne levels.

 

 

 

 

 

 

 

 

 

 

Surface Average contact angle Dyne level (dynes/cm)*
n=10
Wood 55° :1: 2 52 :1: 2
Glass 59° :1: 3 48 d: 2
EPF 74° at 2 36 i 2
CD 76° i 2 34 :t 2
Vinyl 77° :1: 3 34 d: 2
Paper 80° :t 3 32 i 2
Metal 88° :l: 2 32 a: 2
Plastic 94° i 4 30 i 2

 

 

 

 

*Standard deviation provided by test manufacturer.

Plastic had the highest contact angle measurements; all of the ten angles measured
were at 90° or higher. This was the only surface that, according to contact angle
deﬁnitions, was truly nonporous. Metal followed closely, with an average contact angle
of 88°, just under the nonporous threshold. These two surfaces were allotted to a
nonporous category. Paper, CD, vinyl and EPF fell between 72° and 84°, indicating that
they had some level of porosity but were for the most part impermeable to liquids. These
four surfaces formed the nearly nonporous category. The glossiness of the magazine
cover paper was sufﬁcient to make it nearly nonporous rather than nearly porous, as had
been expected. Finally, wood and glass had the lowest contact angles, from 53° to 65°.
While wood was expected to be semiporous, the inclusion of glass in the lowest contact
angle category was surprising, as such low angles indicate a semiporous surface.
However, glass is generally considered nonporous. Additionally, surface topographical
and roughness measurements indicated that glass was by far the smoothest surface and
therefore unlikely to cause ambiguous or misleading contact angle measurements.

Clearly, a factor other than porosity or wettability was causing the water to spread across

47

the glass rather than bead up in a distinct droplet; surface energy is most likely to account
for this effect.

Measurements of surface energy did not result in clear surface grouping, but
rather in a continuous range (Table 2). Glass had the highest surface energy, with a
measurement of 52 dynes/cm. This high surface energy meant that liquids adhered
strongly to glass, and explains why a water droplet was forced to Spread across the
surface rather than maintaining a contact angle consistent with a nonporous surface. The
remaining surfaces ranged from 30 dynes/cm to 48 dynes/cm in the following order:
metal, plastic and CD, paper and vinyl, EPF, and wood. Though there were several
exceptions, this trend was a reverse of the contact angle groupings, indicating that contact
angles tended to be low for surfaces with high surface energy. This reﬂects the close
relationship between surface wettability and surface energy.

Surface energy and wettability results for the eight studied surfaces indicated that
these two surface characteristics impact RUVIS performance only indirectly, by affecting
ﬁngerprint quality. High contact angle measurements, indicating low porosity and low
wettability, mean that a surface rejects moisture, such as that in ﬁngerprint residues.
Enough of the ﬁngerprint remains on the surface and exposed that evaporation is likely to
occur quickly and physical removal of the ﬁngerprint is likely. Low surface energy
compounds this problem, as it means that the ﬁngerprint residues do not adhere to the
surface. Not only can the ﬁngerprint be easily wiped away, but it may also drift on the
surface, much in the same way as droplets of water Slide off a waxed car. If this type of
surface characteristic reduces ﬁngerprint quality, RUVIS will be unable to visualize a

highly detailed ﬁngerprint.

48

4. 2.3 SEM-EDS

Imaging of the surfaces with SEM allowed examination of surface morphology at
high magniﬁcations (in this case 1000x) and visualization of the topographical features
that were quantiﬁed by CLSM and proﬁlometry. For example, CLSM topographical
proﬁles indicated that plastic and metal both had striae, but proﬁlometry indicated that
metal was rougher than plastic. SEM showed that this was due to the close spacing of the
striae on metal (Figrre 14 a) as opposed to the occasional striae on plastic (Figure 14 b).
The exact structure of the broad topographical features observed by CLSM in EPF and
vinyl was also apparent. SEM images showed the beaded structure (Figure 14 c) Of EPF
particularly well. Finally, the presence of dust on glass and especially CD was conﬁrmed

by the observation of small, randomly oriented particles (Figure 14 d).

a b
c d

Figure 14. SEM images of (a) metal, (b) plastic, (c) EPF, and (d) CD.

49

The more interesting aspect of SEM was the EDS capability, which allowed an
elemental analysis of the surfaces (Table 3). The carbon peaks detected in all surfaces
were discounted, since the large amounts of carbon coating the surface would mask any
carbon inherent to the surface. Carbon coating can trap some atmospheric oxygen atop
the surface. Oxygen peaks were therefore also disregarded, unless they were
signiﬁcantly above background intensity levels. As most surfaces exhibited an oxygen
peak with intensity of 500-600, this level was designated as contamination. Other

elements found, however, were attributed to their respective surfaces and several were

Table 3. SEM-EDS of eight surfaces. Approximate intensity values are given for the
highest energy x-ray line for the element, averaged over three scans. Values are omitted
in instances when the highest energy line overlapped between two or more elements.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Glass Paper Vinyl Metal Wood CD EPF Plastic
N 1500
O 3000 600 500 100 2000 1300 600 500
Na 1500 200 100
Mg 500 200 200
A1 100 500 20000 500
Si 9000 700 2200 100 300 200
S 700 300
C1 X
K X
Ca X X X X
Ti 100 100 150
Fe 100 100
Sn 200

 

 

 

 

 

 

 

 

 

 

 

worth noting. Four elements were unique to their surfaces: nitrogen (N) was found only
in wood, potassium (K) only in glass, and chlorine (Cl) and tin (Sn) only in vinyl.
Nitrogen was likely an inherent component of wood, and chlorine an inherent component

of vinyl (polyvinyl chloride). The potassium and tin were likely added to glass and vinyl

50

during the manufacturing process. Conversely, CD was the only surface that had no
peaks aside from carbon and oxygen, though the oxygen peak in this case was
approximately three times larger than contamination levels (intensity 1,300) and can thus
be considered a valid component of the surface.

A ﬁnal aspect of SEM-EDS was an examination of peak heights to indicate which
elements were major components of the surface and which were only present in trace
amounts. Silicon was by far the most common element in glass (intensity 9,000), while
magnesium, aluminum, and potassium were minimal (intensity under 500). These latter
three elements are added to glass during the manufacturing process to vary its properties
(26). A similar phenomenon held true for metal. Aluminum was of course the main
component (intensity 20,000), while all others were at or below detection limits (intensity
100). Vinyl had a large silicon peak (intensity 2,200); it is likely that the chlorine peak
was large as well, but its combination with the carbon peak made it difﬁcult to quantify
without a more in-depth analysis. Wood had a large oxygen peak (intensity 2,000)
accompanied by a Slightly smaller nitrogen peak (intensity 1,500). These types of
combinations implied that the major surface component was not a single element but a
compound, in this case likely a nitrate or nitrite. Finally, surfaces like paper showed no
major peaks; most elements were present in similar amounts though exact ratios varied
between distinct areas of the surface. In the case of paper, it is likely that inks present on
the paper were contributing to the elemental proﬁle. This accounts both for the complex
chemical proﬁle and for the uneven distribution over the paper surface.

EDS was done to determine whether certain elements reacted with ﬁngerprint

residues or UV light to impact ﬁngerprint quality and RUVIS performance. However,

51

the lack of patterns observed between the surfaces made the EDS data very difﬁcult to
apply to RUVIS performance. It iS likely that elemental composition does not have an

important impact on RUVIS visualization of ﬁngerprints.

4.3 Surface effects

Surface effects were assessed statistically based on ﬁngerprint scores. Each score
was the sum of marks assigned to the ﬁngerprint. A ﬁngerprint was given one point if it
was visible and a second if any ridge structure was clear. Further points were assigned
for each minutiae point in the ﬁngerprint. Thus a score of 1 indicated a visible smudge
with no detail and a score of 2 indicated a ﬁngerprint with visible ridge structure but no
useful details. Scores this low (2 points or less) were considered poor. Scores from 3 to
7 were considered fair, as a handful of minutiae were present but the ﬁngerprint was
unlikely to yield an identiﬁcation. Scores of 8 or higher were categorized as good. These
were ﬁngerprints with six or more minutiae; the number is low, especially in countries
which ascribe to the numeric standard. However, identiﬁcations are possible at this level,
especially considering the conservative assignment of minutiae in this study.

Because the main goal of this study was to explore surface effects on RUVIS
performance, statistical analysis focused ﬁrst on control ﬁngerprints from all donors.
These were ﬁngerprints placed on various surfaces, transported overnight, and processed
the following morning. Control ﬁngerprints were not subjected to environmental
conditions (exposure to light, storage in darkness, and storage under water). Control
ﬁngerprints were processed, and each had a score before CA fuming and another after.

Statistical analysis focused ﬁrst on the unprocessed ﬁngerprints to evaluate surface

52

effects on RUVIS performance, then moved on to the processed ﬁngerprint scores to
examine the effects of processing on RUVIS performance.

A ﬁnal stage of statistical analysis focused on factors secondary to this study:
gender of ﬁngerprint donors and environmental conditions to which the ﬁngerprints were
subjected. The control ﬁngerprint data set was split according to gender and the resulting
data sets were compared to determine whether gender was a signiﬁcant factor affecting
the quality of ﬁngerprints visualized by RUVIS. Fingerprint scores for each gender were
examined both before and after processing. Attention was then turned to the three
environmental conditions to which ﬁngerprints had been subjected: exposure to light,
storage in darkness, and storage under water. Each environment constituted a separate
data set, which was analyzed in the same manner as the control data set. The three
environment data sets and the control data set were then compared to one another. Both
the analysis and the comparison between environments was carried out for unprocessed
and processed ﬁngerprints.

Regardless of which aspect of RUVIS performance was being evaluated,
statistical tests were carried out with 95% conﬁdence (0I=0.05). The only exception to
this rule was hierarchical cluster analysis, where clusters with similarity higher than 75%

were considered as grouping.

4. 3.1 Statistical observations of unprocessed control ﬁngerprints by surface
Basic statistical analysis of all unprocessed control ﬁngerprints (calculation of
mean scores and standard deviations, etc.) showed that RUVIS performance with

ﬁngerprints on glass was consistently good (mean score 8.95). However, RUVIS

53

performance on metal was Slightly stronger (mean score 9.87). The difference between
glass and metal was not statistically signiﬁcant. With approximately half the mean score
of glass, CD and paper fell into the fair performance category. The remaining four
surfaces (wood, plastic, EPF, and vinyl) fell into the poor performance group. The mean
score for wood was just above 1 and scores on the remaining three surfaces were below
1. RUVIS performance on vinyl was weakest, with a mean score of 0.30 and a maximum
score of 2.00. ANOVA and Tukey’s comparisons showed that the three groups (good,
fair, and poor) were statistically different from one another (p=0.00) and that there were
not signiﬁcant differences between surfaces within groups.

The groups of surfaces formed based on average ﬁngerprint scores did not
directly match any of the surface groupings based on surface characteristics. The two
surfaces which resulted in good ﬁngerprint scores (glass and metal) had no pertinent
surface characteristics in common. CD and paper, which resulted in fair ﬁngerprint
scores, had similar contact angles and surface energies. However, neither surface
characteristic distinguished these two surfaces from others. Wood, EPF, and vinyl
resulted in poor ﬁngerprint scores, and all three had high surface roughness values.
However, none of these three surfaces shared characteristics with plastic, which also
resulted in poor ﬁngerprints.

While most of these ﬁngerprint visualization results were unsurprising, three
surfaces deﬁed all expectations. First, glass had been selected as the best available
medium for ﬁngerprints, based on personal ﬁngerprint laboratory experience. It was thus
surprising that RUVIS performed as well on metal, especially given metal’s roughness

and striated topography which would be expected to interfere with the UV light reﬂection

54

required for effective RUVIS performance. Fingerprint scores on paper also Showed
more detail than expected, given that paper was expected to be among the more porous
surfaces. In fact, contact angle measurements Showed paper to be one of the less porous
surfaces in the study. Finally, RUVIS was expected to perform quite well on plastic
because there was little roughness on this surface to interfere with UV light reﬂection and
no porosity to absorb the ﬁngerprint. Further, previous RUVIS studies had concluded
that smooth and nonporous surfaces like plastic Should result in clear RUVIS visualized
ﬁngerprints (7). However, this study found that RUVIS visualized ﬁngerprints on plastic
had poor scores (low quality). This is precisely why a more speciﬁc understanding of
surface effects on RUVIS performance is required; the general “smooth and nonporous”

characterization is simply insufﬁcient.

4. 3.2 Cluster analysis of unprocessed control ﬁngerprints by surface

Fingerprint scores obtained from all 12 donors for each surface were listed, then
hierarchical cluster analysis (HCA) was used to assess similarity among surfaces based
on the scores. Since distinct groupings of the surfaces were Observed, HCA Showed that
there were some surface effects on RUVIS performance when visualizing control
environment ﬁngerprints before processing (Figure 15). At 95% similarity, vinyl and
EPF were grouped, with wood included at 93% similarity and plastic at 91% Similarity.
The remaining surfaces had at most 57% similarity and hence each of these surfaces was

considered an individual group.

55

 

36.34-I

 

 

 

 

57.56‘

swam (“/0)

78.78-

 

 

93.00
9500 I
T

 

}h
U'l-I—
m-t—-I

 

 

 

 

 

 

 

 

 

100.00

i“—
t

Figure 15. Grouping of unprocessed control ﬁngerprint scores by surface.
Surface key: (1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF, (8) plastic.

The question thus arises: what made vinyl, EPF, wood, and plastic surfaces so
similar in the ‘eyes’ of RUVIS? Vinyl and EPF constitute the most similar statistical
grouping at 95% similarity. These were the only two surfaces with average roughness
values over 10 um. Further, they both had surface topographies following a pattern of
100-200 um wide projections. Both fell into the nearly nonporous grouping with contact
angles only 3° apart, and their dyne level error ranges overlapped (32-36 dynes/cm for
vinyl and 34-38 dynes/cm for EPF). In fact, the only way in which these two surfaces
were not similar was in elemental composition. Aside from carbon and oxygen, they
shared only one element in common: Silicon. However, Silicon was present in Six of the
eight surfaces and was therefore unlikely to be a factor in this level of similarity. This
elemental dissimilarity is of little consequence, as elemental composition was determined

to be the least important surface characteristic in this study.

56

At 93% similarity, wood was also grouped with EPF and vinyl. However, there
were far fewer common surface characteristics in this group of surfaces than between
vinyl and EPF. Although wood was the third roughest surface after EPF and vinyl, the
average roughness value was less than half that of vinyl (4.3 pm compared to 10 um).
Wood had a random topography compared to the patterned topography of vinyl and EPF.
Wood was also far more porous than either vinyl or EPF with a contact angle
approximately 20° lower (48°). Wood had a much higher dyne level than vinyl or EPF
(48 dynes/cm compared to 34 and 36), meaning that ﬁngerprint adhesion would have
been much stronger than on vinyl or EPF. Aside from carbon and oxygen, wood Shared
no common elements with EPF and only one common element (sodium) with vinyl.
Further, even the oxygen commonality fell away, as wood was one of the rare surfaces
with an oxygen peak above than contamination levels (intensity 2,000).

At 91% similarity, plastic was also grouped with vinyl, EPF, and wood. Here,
even fewer surface characteristics coincided. While vinyl, EPF, and wood were the three
roughest surfaces, plastic was the third smoothest surface with an average roughness less
than a quarter of the wood value (1.2 pm compared to 4.3 pm). Plastic topography did
fall into the category of patterned surfaces with vinyl and EPF. However, the striae on
plastic were widely spaced rather than closely packed like the blister structures of EPF or
the leathery scales of vinyl. Plastic was the only truly nonporous surface and had a
contact angle almost 20° higher than vinyl (94°). The only potential source of similarity
is that plastic had a surface energy of only 2 dynes/cm less than vinyl (32 dynes/cm).
Finally, it was interesting that the elemental composition of plastic contained only

elements present in wood or EPF; carbon, oxygen, magnesium, and silicon were present

57

in EPF, while calcium and titanium were present in wood. Again, however, elemental
composition was the least important surface characteristic.

Based on the grouping of the three roughest surfaces with 93% similarity,
roughness was Obviously an important factor in RUVIS performance. As was discussed
in Chapter 2, RUVIS is able to visualize ﬁngerprints because, while the prints scatter UV
light to the instrument, surfaces reﬂect the light away from the instrument and appear
darker. Above a certain roughness threshold, the surface will scatter as much UV light as
the ﬁngerprint, no contrast will be created, and no print visualized. This roughness
threshold was illustrated by the three grouped surfaces: wood, vinyl, and EPF, which
were the only surfaces in this study with RA values greater than 3 pm. Since RUVIS
performance on all three of these surfaces was poor, 3 pm was established as the
threshold at which surface roughness begins to result in UV light scattering similar to the
light scattering from ﬁngerprint residues. Performance on wood, which had RA just over
this threshold at 4.3 pm, was poor but marginally (though not signiﬁcantly) better than
performance on vinyl or EPF. As roughness increased to the 10 um RA of vinyl, RUVIS
performance became somewhat poorer. This indicates that, at approximately 10 um RA,
the UV light scattering ability of the surface matches the UV light scattering ability of
ﬁngerprint residues, and nearly no contrast between surface and ﬁngerprint is observable
through RUVIS. When roughness becomes still higher, such as the 18 um RA of EPF,
the surface begins to scatter more UV light than the ﬁngerprint residues and contrast is
once again possible with RUVIS, though the relative brightness of the surface and
ﬁngerprint will now be reversed. However, at roughness this high, the surface

topography may interfere with the ﬁngerprint itself and ridges may appear discontinuous.

58

These two competing effects of high roughness may explain why RUVIS performance on
EPF was marginally but not Signiﬁcantly higher than on vinyl.

The inclusion of plastic in the poor performance category complicated matters.
The average roughness of plastic (1.2 pm) placed it between CD and paper, both of which
resulted in fair RUVIS performance. The surface energy of plastic (32 dynes/com) fell
just below that of vinyl (34 dynes/cm), which appeared to be a common surface
characteristic. However, CD and paper both had surface energies within the same range.
The one characteristic that did set plastic apart, not only ﬁ'om CD and paper but from all
other surfaces studied, was its high contact angle. Plastic was the only surface studied
with a contact angle above 90°. This level of nonporosity means that the surface rejects
moisture. It is possible that this makes it easier for a ﬁngerprint to evaporate or be
physically removed from the surface. In this case, it may not be RUVIS performance that
is adversely affected by the surface, but rather the inherent quality of the ﬁngerprint that
is reduced by interaction with the surface.

The statistical analysis of unprocessed control ﬁngerprints indicated that surface
roughness and surface wettability had the largest impact on RUVIS performance.
Ideally, ﬁngerprint examiners seeking to make RUVIS use more efﬁcient Should screen
surfaces for roughness and energy. This can be done either by performing a surface
chemistry test in the ﬁngerprint laboratory or by consulting published data for an
approximate roughness or contact angle. Unprocessed ﬁngerprints on surfaces with
average roughness greater than 3 pm and contact angles greater than 90° should not be
visualized with RUVIS, as results are likely to be poor. Instead, these ﬁngerprints can be

visualized with other methods, or processed and then visualized with RUVIS.

59

4.4 Processing effects

4. 4.1 Statistical observations of processed control ﬁngerprints by surface

Basic statistical analysis (calculation of means, standard deviation, etc.) was
carried out for processed control ﬁngerprints, and the results were compared to
unprocessed ﬁngerprint scores using the paired t-test or Wilcoxon Signed Rank Test, as
appropriate given normality of the distribution of scores on each surface. These
comparisons showed that processing changed the performance rankings of the surfaces.
Before processing, RUVIS visualization showed good ﬁngerprints (mean scores 8 or
higher) on metal and glass, fair ﬁngerprints (mean scores 3-7) on CD and paper, and poor
ﬁngerprints (means scores 2 or lower) on wood, plastic, vinyl, and EPF. After
processing, RUVIS visualized good ﬁngerprints on metal, CD, and plastic, fair
ﬁngerprints on glass, paper, wood, and EPF, and poor ﬁngerprints only on vinyl.

This rearrangement of the performance rank meant that mean scores on seven of
the eight surfaces had changed signiﬁcantly; only scores on metal had no signiﬁcant
change after processing. Of the remaining surfaces, only scores on glass decreased after
processing (72.0). This decrease was not only statistically signiﬁcant, but practically
meaningful as well, since it moved glass from the good category to the fair. This was
surprising, as personal experience in crime labs indicated that ﬁngerprints on glass tend
to be clear both before and after processing. This numerical evidence also contradicted
the cursory visual observations made during ﬁngerprint processing, when it had appeared

that ﬁngerprint quality improved for all surfaces after processing.

60

The remaining surfaces Showed Signiﬁcant increases in RUVIS visualized
ﬁngerprint scores after processing. Fingerprint scores on plastic increased most
dramatically (+6.0), resulting in plastic moving from the poor category to the good
category. This large change may have resulted from the high contact angle that made
plastic a difﬁcult surface for RUVIS visualization of unprocessed prints. It was
postulated in the previous section that the non-wettable nature of plastic caused much of
the water-soluble ﬁngerprint residue to evaporate, thus concentrating the lipid residues.
High lipid concentrations have been shown to improve CA fuming results (3; 12) and
resulting in very clear processed ﬁngerprints. Fingerprint scores on CD and wood
showed the next highest increase (+1.0 to 1.5), promoting CD from fair to good and
wood from poor to fair. Fingerprint scores on paper and vinyl showed the least
improvement after processing (+1.0). Though score improvements on these last two
surfaces were statistically signiﬁcant, they were meaningless in practice, as both surfaces
remained in their previous performance categories.

Finally, ANOVA and Tukey’s comparisons indicated that, after processing, the
three performance categories (good, fair, and poor) were no longer as distinct as they had
been prior to processing. For the unprocessed ﬁngerprint scores, signiﬁcant differences
between surfaces had been aligned with the category delineations. After processing,
there continued to be no signiﬁcant differences between mean ﬁngerprint scores on
surfaces within categories. However, there were also few Signiﬁcant differences between
categories. Vinyl, the only surfaces with poor scores, was signiﬁcantly different from
surfaces with good scores, but could not be distinguished from surfaces with fair scores.

Similarly, all surfaces with scores in the good category were indistinguishable from at

61

least one surface in the fair category. The clearest indication of this category merging
was paper; scores on paper were indistinguishable from scores on all surfaces aside from
metal, in all three performance categories. Because the performance groups overlapped

so much, it was difﬁcult to compare them based on surface characteristics without HCA.

4. 4.2 Cluster analysis of processed control ﬁngerprints by surface

Hierarchical cluster analysis of processed control ﬁngerprint scores Showed that
processing changed groupings of surfaces (Figure 16). The three roughest surfaces
(wood, EPF, and vinyl) were clustered with greater similarity than any others, just as they
had been prior to processing. However, processing reduced their similarity from 93% to
60%. While this low similarity cluster can be noted, it is no longer statistically
signiﬁcant. Processing also caused plastic scores to change sufﬁciently that plastic was
no longer the fourth member of the cluster. This change in clustering was a direct result
of the different changes each surface experienced after processing. RUVIS performance
on plastic improved more than it had on any other surface, resulting in the removal of
plastic from the low performance cluster. The three surfaces remaining in the cluster
became less similar to one another because the improvement each experienced was of a
different magnitude. RUVIS performance improved by approximately +1.0 on vinyl,
+1.5 on wood, and +2.0 on EPF. Though a difference of only half a point between
surfaces seems small, it was sufﬁcient to reduce similarity within the cluster by over

30%.

62

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

19.19-
g 46.131 I
r I
g 60.00 ” ‘ I
73.06‘
100.00
1 2 3 5 7 6 8 4

Figure 16. Grouping of processed control ﬁngerprint scores by surface. Surface key:
(1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF, (8) plastic.

The changes in RUVIS performance with processing were likely due to the
interactions of CA with ﬁngerprints and surfaces. CA fumes settle and polymerize on
ﬁngerprint residues, enhancing and preserving them. However, some CA fumes do settle
on the surface itself. This means that when CA fuming is performed, both the surface
and the ﬁngerprint are affected. An ideal processing would result in sufﬁcient CA
polymerization on the ﬁngerprint to Show but not overwhelm detail and minimal
polymerization on the surrounding surface. However, CA coverage can vary from trace
to heavy amounts on both the surface and the ﬁngerprint. This means that CA fuming
can obscure detail in a ﬁngerprint as well as enhancing it. For example, too much
polymerization on the ﬁngerprint would hide ﬁne detail and too much polymerization on
the surface would reduce contrast with the ﬁngerprint.

Based on the overall decrease in RUVIS performance on glass after processing, it

appears that some surface characteristic of glass made it more prone to the type of

63

overdevelopment that can reduce ﬁngerprint quality. This was most likely due to surface
energy. Glass had the highest dyne level of all surfaces studied (52 dynes/cm), and was
therefore the surface most amenable to adhesion. This may have caused higher than
normal levels of CA to build up on the glass surrounding the ﬁngerprint. This would
give the glass and the ﬁngerprint Similar textures and reﬂective capabilities, reducing the
contrast between glass and ﬁngerprint when observed with RUVIS.

While glass was the one surface on which RUVIS performance decreased after
processing, plastic resulted in the largest performance increase of all eight surfaces. It
was previously postulated that unprocessed ﬁngerprints on plastic resulted in poor
RUVIS performance because the high contact angle and nonporosity of plastic caused
this surface to reject the moisture of the ﬁngerprint. However, plastic also had a low
surface energy (32 dynes/cm). This would prevent CA from adhering effectively to the
surface, and could possibly increase the preferential polymerization of CA on the
ﬁngerprint residues remaining on plastic. This would not only hugely increase any
contrast between the ﬁngerprint and the plastic surface, it would also protect the
ﬁngerprint from evaporation and serve to anchor it to the surface.

CA fuming also produced a slight but signiﬁcant improvement in RUVIS
performance on the three rough surfaces (wood, EPF, and vinyl) that were clustered prior
to processing. This can also be attributed to the differential accumulation of CA on the
surface and the ﬁngerprint. It was previously established that these surfaces were
problematic for RUVIS performance because their roughness caused them to scatter light
in the same way as the ﬁngerprint residue. Processing can remedy this if CA polymerizes

at a different level on the surface than it does on the ﬁngerprint, causing the two to reﬂect

64

UV light differently and restoring contrast for RUVIS. It is most likely that some CA
settled on the wood, EPF, and vinyl surfaces and slightly reduced their roughness while

much more CA polymerized on the ﬁngerprint, enhancing contrast.

4.5 Gender effects

Initial visual comparisons of male- and female-donated ﬁngerprints for each
surface had indicated that there was a difference between RUVIS performance for the
two groups. Statistical comparison of ﬁngerprint scores between genders (paired t-test or
Wilcoxon Signed Rank test) conﬁrmed that, regardless of processing, RUVIS
performance with male-donated ﬁngerprints was either equal to or better than
performance with female-donated ﬁngerprints. For unprocessed control ﬁngerprints,
scores on metal and CD showed the greatest difference between genders. Scores on
glass, plastic, vinyl, and wood followed with male prints scoring Slightly but signiﬁcantly
higher than female ﬁngerprints. Finally, RUVIS performance was equal for the genders
on EPF and paper. These two surfaces had no characteristics in common to differentiate
them from the other surfaces, and it is possible that the reason for their lack of
differentiation between genders was not based on surface characteristics.

The trend was rather different for processed ﬁngerprints. RUVIS performance
remained much better for male ﬁngerprints than female ﬁngerprints on metal (+7.0) and
somewhat better on glass and plastic (+4.0). However, male ﬁngerprint scores on EPF
were slightly higher (+1.0) than female scores, while no difference was evident on CD,
paper, vinyl or wood. It is worth noting that, after processing, there was a smaller

difference between male and female ﬁngerprints on vinyl and wood than before

65

processing and a larger difference on EPF. This means that, when ﬁngerprint scores from
all donors were considered, processing resulted in tighter clusters of scores on vinyl and
wood (the two donor subgroups became more similar) and broader clusters on EPF. This
would partially account for the reduced Similarity after processing for the three surfaces.

Since signiﬁcant differences between male and female scores were discovered,
HCA was repeated for unprocessed and processed control ﬁngerprints, now with the data
set separated by gender. Similarity among surfaces was assessed as before, although this
time considering males and females as separate data sets. For unprocessed ﬁngerprints,
the differences were minimal between the male and female data sets and the original
combined data set. Female data clustered exactly as the combined data set had (Figure
15), though the Similarity scores were 2% to 3% lower (vinyl was grouped with EPF at
90.2% Similarity, with wood at 90% similarity, and with plastic at 89% similarity). Male
data resulted in the same ﬁnal cluster of surfaces (vinyl, wood, EPF and plastic), but the
order of Similarity was different. Wood and EPF were the most similar grouping at 95%
similarity, vinyl joined them at 94% similarity, and plastic at 91% similarity. The gender
effects on RUVIS performance for unprocessed ﬁngerprints were therefore signiﬁcant,
but did not outweigh surface effects, as ﬁnal surface clusters were unchanged regardless
of whether genders were considered separately or together.

The difference between male, female, and combined processed prints was more
striking than the differences observed with unprocessed ﬁngerprints. The combined
processed data set had shown the same wood, vinyl, and EPF clusters as all unprocessed
ﬁngerprints, though at similarity levels of 60% and lower. The male processed data

showed wood and EPF clustering at 85% similarity and vinyl joining them at 76%

66

similarity. The female processed data Showed yet another grouping (Figure 17). Not
only were there no clusters above 48%, but the familiar wood, vinyl, and EPF cluster was

not present.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

32.16- I
r

g 54.77-
E

77.39-

100.00
1 2 3 4 5 7 8 6
Surfaces

Figure 17. Grouping of female processed control fingerprint scores by surface.
Surface key: (1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF,
(8) plastic.

These gender-based differences in RUVIS visualized ﬁngerprints were not quite
large or consistent enough to be useful in a forensic laboratory. They would not help a
ﬁngerprint examiner decide whether an unknown ﬁngerprint was more likely male or
female. However, the differences were signiﬁcant and therefore sufﬁcient to create
doubts about the validity of unisex studies and to indicate that further exploration of
gender differences in ﬁngerprints may be useful. This conclusion is consistent with
studies that indicate genetically-based differences between minutiae counts in males and
females (3), as well as the general differences in ﬁnger and therefore also print size

between the genders.

67

4.6 Environmental effects

The ﬁnal stage Of this study examined whether the environment to which a
ﬁngerprint was exposed could affect the ability of RUVIS to visualize that print. Three
environmental conditions were considered: week-long exposure to light, storage in
darkness, and storage under water, all compared to the control ﬁngerprints which were
taken to the laboratory within a day of collection. Initial expectations had been that aging
ﬁngerprints for one week in darkness would have a minimal effect on ﬁngerprint
residues, while exposure to light would damage the lipid components of residues and
storage under water would remove water soluble components. Water was expected to be
the most damaging both to ﬁngerprints and to RUVIS performance.

Prior to processing, ﬁngerprints from all three environmental conditions clustered
exactly as control ﬁngerprints had: vinyl and EPF clustered ﬁrst, followed by wood and
ﬁnally plastic. It was somewhat surprising that storage under water did not damage the
more porous surfaces sufﬁciently to change their interaction with RUVIS. For light
exposed prints, clustering was most similar to control ﬁngerprints: vinyl and EPF
clustered at 93% similarity (compared to 95% for control), wood joined the group at 92%
(compared to 93% for control), and plastic joined the group at 88% (compared to 91% for
control). The similarity values for groupings of surfaces stored in dark and wet
conditions were somewhat different, though the ﬁnal groupings remained the same. For
dark stored ﬁngerprints, vinyl and EPF clustered with 94% Similarity, wood was added at
93%, and plastic was added at only 77% similarity. For wet stored ﬁngerprints, all three
of the rough surfaces clustered at 94% similarity and plastic was added at 87% similarity.

However, even these differences in Similarity values between environmental conditions

68

were not critical to RUVIS performance, as the surface groupings themselves remained
the same as for control ﬁngerprint scores. This indicates that, while environmental
factors may affect ﬁngerprint quality, they are unlikely to affect RUVIS performance.
When the exposed ﬁngerprints were processed, the changes in clustering were
more signiﬁcant. Clusters for light exposed ﬁngerprints were rearranged completely.
CD and plastic were 79% similar, joined by EPF at 78% similarity and wood and vinyl at
70% similarity (Figure 18). This was the only case where the CD surface was involved
in a cluster. It is possible that the highly reﬂective surface Of the CD caused similar
evaporation of water soluble ﬁngerprint residues to that observed in plastic. Fingerprints
on both surfaces would have had high concentrations of lipid residues and similar
reactions to CA fuming. Dark stored ﬁngerprints maintained the vinyl, EPF, and wood

cluster at 72% Similarity but plastic was completely removed from the cluster.

 

42.80-

61.86- I

75.00 I

80.93- ‘

 

 

 

 

 

Shifty (We)

 

 

 

 

 

 

 

 

 

 

 

100.00
1 2 3 6 8 7 5 4

Surfaces

Figure 18. Grouping of processed light exposed fingerprint scores by surface.
Surface key: (1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF,
(8) plastic.

69

Finally, the order of clustering was slightly rearranged when wet stored
ﬁngerprints were processed. Vinyl and wood clustered ﬁrst at 94% Similarity, followed
by the addition of plastic at 93% and EPF at 92% (Figure 19). This was the only case in
which processed ﬁngerprints clustered at similarity levels above 90%. It is likely that
storage under water removed most water soluble components from the ﬁngerprint
residues on all surfaces, which would have made the ﬁngerprints more Similar to one
another regardless of surface. All ﬁngerprints would thus have reacted similarly to CA
fuming. On surfaces which had already been clustering prior to processing, this would

have increased similarity values.

 

 

 

 

 

 

 

 

 

 

 

 

 

15.437
E 43.62-
71.er I l
100.00 . . . .
r 2 3 5 8 7 6 4
Surfaces

Figure 19. Grouping of processed water stored fingerprint scores by surface.
Surface key: (1) glass, (2) paper, (3) vinyl, (4) metal, (5) wood, (6) CD, (7) EPF,
(8) plastic.

To investigate differences between the environmental conditions further, one-way
ANOVA was carried out for all ﬁngerprints collected on glass and wood, both before and
after processing. These two surfaces were selected to represent surfaces on which

RUVIS had visualized good and poor ﬁngerprints, respectively. ANOVA included

70

control ﬁngerprints as well as light exposed, dark stored, and wet stored ﬁngerprints.
Because these data did not follow normal distributions, they were log-transformed. The
only exception was the data set of unprocessed ﬁngerprints on wood, which was
sufﬁciently normal to satisfy the relatively robust ANOVA technique.

ANOVA results showed that, for unprocessed ﬁngerprints on glass, there was at
least one signiﬁcant difference between the four environments (p=0.001). Closer
examination Showed that, as expected, control ﬁngerprints had the highest log-
transforrned score and wet stored ﬁngerprints the lowest. Tukey’s comparisons indicated
that the only signiﬁcant difference was between control and wet stored ﬁngerprints. For
processed ﬁngerprints on glass, there was no signiﬁcant difference between environments
(p=0.345). However, the log-transformed score for wet prints was slightly lower than for
the remaining environments. Finally, direct comparisons of unprocessed and processed
data for the three environmental conditions were performed and showed that processing
had no signiﬁcant impact for scores on glass. This was surprising, given that processing
had resulted in lower scores on glass for control ﬁngerprints. It is possible that the week-
long aging of ﬁngerprints on glass (regardless of speciﬁc environment) sufﬁciently
changed ﬁngerprint residues to affect their reaction with CA.

Differences between environments for both unprocessed and processed
ﬁngerprints on wood were signiﬁcant (ANOVA p=0.000 in both cases). For unproceSsed
prints, Tukey’s comparisons indicated that control ﬁngerprint scores were signiﬁcantly
higher than scores from all three environments. Within the three environments, light
exposed ﬁngerprint scores were signiﬁcantly higher than water stored ﬁngerprint scores.

For processed ﬁngerprints on wood, scores for water stored ﬁngerprints were

71

signiﬁcantly lower than scores for both other environments and for control ﬁngerprints.
Direct comparison Showed that processing improved light exposed and dark stored
ﬁngerprint scores on wood slightly, but did not signiﬁcantly impact wet stored
ﬁngerprints scores. It is likely that improvement on light and dark stored prints was
Similar to improvement on control prints; CA polymerization increased contrast between
surface and ﬁngerprint by changing their relative roughness. The lack of improvement
for water stored ﬁngerprints is surprising, as the water should have increased lipid
concentrations of the residues and improved CA fuming results. However, it is possible
that water also soaked into the relatively porous wood surface (contact angle 55°) and
sufﬁciently changed the surface to counteract any changes made by processing.

This analysis demonstrated that RUVIS performance was largely consistent for
the eight studied surfaces regardless of the environment to which a ﬁngerprint was
exposed. This is very beneﬁcial for an instrument used to visualize ﬁngerprints. The
prints submitted to a forensic laboratory could have been exposed to any number of
environmental conditions, and the ﬁngerprint examiner very rarely knows precisely what
these conditions were. Because RUVIS performance is not affected by environment, no

screening is necessary and all ﬁngerprints, regardless of exposure, can be visualized.

72

5. CONCLUSIONS AND FUTURE WORK

To date, RUVIS has not been used in ﬁngerprint laboratories to its full potential.
This has mainly been due to limited research on the complex interaction between a
ﬁngerprint, the surface on which it is deposited, and the effects on RUVIS performance.
This study was therefore designed to explore the impact of surface chemistry on RUVIS
performance. Surface chemistry can interact directly with the RUVIS technique (for
example by affecting UV light reﬂection) or can interact with ﬁngerprints and thus

indirectly inﬂuence RUVIS results.

5.1 Conclusions and recommendations

Of the surface characteristics studied, roughness had the most signiﬁcant impact
on RUVIS performance. This may not be surprising, since visualization of images by
RUVIS depends on the scattering or reﬂection of UV light from a surface. The impact of
roughness was discovered when HCA Showed clusters of the roughest surfaces (EPF,
vinyl, and wood) with similarity greater than 90%. ANOVA and Tukey’s comparisons
further indicated that, with 95% conﬁdence, the rough surfaces had signiﬁcantly poorer
RUVIS visualized ﬁngerprint scores than smoother surfaces.

Above average roughness values of 3 pm, the light scattering ability of the surface
began to approach the light scattering ability of ﬁngerprint residues, decreasing the
contrast in RUVIS images, which relies on ultraviolet light reﬂection or scattering. At
average roughness values of approximately 10 pm, the surface and ﬁngerprint scattered
light so similarly that no visualization of ﬁngerprints by RUVIS was possible. As surface

roughness increased ﬁrrther, some contrast was reestablished.

73

Paired t-tests and Wilcoxon Signed Rank tests were used to compare RUVIS
performance on rough surfaces before and after processing. These tests Showed a
signiﬁcant improvement in ﬁngerprint scores, with 95% conﬁdence. Cyanoacrylate
fuming was thus able to improve RUVIS performance on rough surfaces, though the
resulting ﬁngerprints only occasionally had enough detail to be of identiﬁable quality.

The realization that surface roughness has a signiﬁcant impact on RUVIS
performance can increase the efﬁciency of RUVIS use in ﬁngerprint laboratories.
Fingerprint examiners could screen surfaces for roughness prior to RUVIS use; portable,
easy to use and relatively inexpensive proﬁlometry instruments for measuring surface
roughness, such as the TR200 (Micro Photonics Inc., Allentown PA) are readily
available. These instruments can be used in the ﬁngerprint laboratory or at a crime scene,
on ﬂat or curved surfaces, and provide an average roughness measure within seconds.
The only potential problem is the contact nature of this test. Contact proﬁlometry would
have to be performed on an area of the surface unlikely to contain ﬁngerprints, though the
area required to perform the test is generally so small that this should not be a problem.

Because HCA indicated that plastic, a surface with an extremely high contact
angle, clustered with the rough EPF, vinyl, and wood at more than 90% similarity,
surface wettability was also found to impact RUVIS performance. On surfaces with very
low wettability (contact angle of 90° or greater), RUVIS visualized faint ﬁngerprints with
very little detail. However, a Wilcoxon Signed Rank test indicated with 95% conﬁdence
that RUVIS performance on this type of surface did improve Signiﬁcantly after CA
fuming. The resulting ﬁngerprints tended to be clear and Show sufﬁcient detail for

identiﬁcation. Unfortunately, tests of wettability are less amenable to the forensic setting

74

than tests of surface roughness. Instead, ﬁngerprint examiners would have to rely on
literature values. Surfaces with low wettability Should be processed as soon as possible
to limit ﬁngerprint residue evaporation and to result in the clearest possible visualization
with RUVIS.

Surface energy was also found to have impact on RUVIS performance, albeit
indirect, by affecting ﬁngerprints rather than UV light reﬂection. Because this impact
was limited to processed ﬁngerprints, it was discovered when a paired t-test comparison
of unprocessed and processed ﬁngerprints on glass indicated with 95% conﬁdence that
processing had a signiﬁcant negative impact on RUVIS performance. Unprocessed prints
on surfaces with high energy (glass) were visualized clearly with RUVIS. After
processing, however, surfaces with energy greater than 50 dynes per centimeter showed
excess cyanoacrylate deposition around, rather than on, a ﬁngerprint. The result was
reduced contrast between ﬁngerprint and surface during RUVIS visualization after
processing. Again, literature values could be used to identify surfaces with high surface
energy so that any ﬁngerprints present could be visualized with RUVIS prior to
processing.

Processing in general had an important impact on RUVIS performance. For the
most part, CA fuming allowed better ﬁngerprint visualization with RUVIS. Aside from
surfaces with high surface energy, Wilcoxon Signed Rank tests showed with 95%
conﬁdence that processing improved ﬁngerprint scores or maintained high scores for
RUVIS visualized ﬁngerprints. Fingerprint scores on most rough surfaces were
improved by processing, though these improvements were insufﬁcient to yield

identiﬁable prints (less than six clear minutiae). However, ﬁngerprints with high

75

concentrations of lipid residue compared to water soluble residue were improved by CA
fuming from having no minutiae to having more than six minutiae.

Elemental composition did not show a statistically Signiﬁcant impact on RUVIS
performance. Similarly, though both gender of the ﬁngerprint donor and the environment
to which a ﬁngerprint was exposed affected ﬁngerprint quality (based on HCA and
ANOVA and Tukey’s comparisons), neither was found to signiﬁcantly affect RUVIS
performance. These factors are therefore important considerations for future studies, but

not particularly helpful for a better understanding of RUVIS.

5.2 Future work

While this study has provided a ﬁrst step towards improving the understanding of
surface effects on RUVIS performance, the work is by no means complete. The eight
surfaces studied here have served to provide glimpses of pertinent surface characteristics.
Each of these surfaces should be examined in more depth, however. A study of different
types or brands of glass, paper, vinyl, metal, wood, CD, EPF, and plastic would
determine whether surface characteristics are consistent across a surface type. That is,
does all glass have sufﬁciently high surface energy to cause strong CA adhesion and
decrease RUVIS performance on processed ﬁngerprints? Is aluminum foil representative
of heavier aluminum, or even other metals? If surface characteristics are found to be
consistent within each surface class, ﬁngerprint examiners would simply have to
familiarize themselves with a limited number of classes rather than running a test on

every surface submitted to the laboratory. Unfortunately, there is a large number of

76

different types of surface within each class; it is unlikely that the pertinent surface
characteristics will be consistent.

The eight surface classes discussed so far hardly exhaust the possibilities of
surfaces encountered at crime scenes and in ﬁngerprint laboratories; ﬁrrther examination
needs to expand the list. Ceramics, polished stone, drywall with gloss or semi-gloss
paint, shined leather, and a multitude of other surfaces are candidates. Particular
attention should be paid to including surfaces with characteristics that ﬁll the gaps in the
present study. Only one of the eight surfaces studied (plastic) had a contact angle in a
range pertinent to RUVIS performance. Only glass had surface energy high enough to
affect CA fuming and, therefore, post-processing RUVIS performance. Including other
surfaces with high surface energies or high contact angles would validate the conclusions
drawn in this study. Further, a selection of rough surfaces with RA values increasing
gradually above 3 pm could establish more speciﬁc thresholds for effective RUVIS
performance.

All of these surfaces lend themselves to myriad processing methods. While this
study only considered CA ﬁrming, semiporous and nonporous surfaces can be processed
with vacuum metal deposition, physical developer, ﬁngerprint powders, ﬂuorescent dyes,
and many other techniques. Fingerprint laboratories which use these methods would
beneﬁt from exploring the speciﬁc effects of each processing method on RUVIS
performance to determine whether any given method is best in combination with RUVIS.
It is also possible that these other techniques may improve the RUVIS visualization of
ﬁngerprints on rough surfaces sufﬁciently for comparison and identiﬁcation, as CA was

unable to do.

77

The conclusions drawn in this study regarding environmental effects can also be
explored further. This study was limited by time and resources to expose ﬁngerprints to
three environmental conditions for time periods of one week. Future studies would
beneﬁt from extending this time period to weeks or months of uninterrupted exposure.
This would allow a determination of how well RUVIS is able to visualize truly ‘old’
ﬁngerprints, and may even yield a technique for ﬁngerprint age determination using
RUVIS. More diverse environmental conditions can also be considered, taking into
account humidity, wind, precipitation, and other similar factors.

Finally, the demographics of ﬁngerprint donors Should be considered in more
detail. Though gender was not found to have a signiﬁcant impact on RUVIS
performance in this study, the sample Size was small relative to the population Size
forensic laboratories deal with. Larger studies of gender effects are therefore in order.
However, other demographic factors can also be considered: age, race, occupation, and
even health may have signiﬁcant impacts on the quality of ﬁngerprint a donor leaves
behind.

RUVIS has true potential to become an effective tool for ﬁngerprint examiners,
and its use should increase as understanding of surface effects on its performance
increases. Once that understanding is achieved, ﬁngerprint examiners will be able to
conﬁdently use RUVIS to scan large crime scenes, photograph on-site ﬁngerprints that
would be difficult to transport to the laboratory, and in the laboratory enhance prints

without risking damage or erasure.

78

REFERENCES

1. Beavan C. Fingerprints. New York: Hyperion, 2001.

2. Peterson JL, Hickman MJ. Census of Publicly Funded Forensic Crime Laboratories,
2002. US Department of Justice, Ofﬁce of Justice Programs; Bureau of Statistics
Bulletin. 2005. www.thecfso.org/advocacy/bjs__study.pdf. Cited 14 April 2007.

3. Champod C, Lennard C, Margot P, Stoilovic M. Fingerprints and Other Ridge Skin
Impressions. J Robertson, editor. Boca Raton, FL: CRC Press, 2004. (J Robertson editor.
International Forensic Science and Investigation Series).

4. Lee HC, Gaensslen RE, editors. Advances in Fingerprint Technology, 2nd edn. Boca
Raton, FL: CRC Press, 2001. (BAJ Fisher editor. CRC Series in Forensic and Police
Science).

5. Cubuk MC. Utilisation Of Ultraviolet Light for Detection and Enhancement of Latent
Prints. ENF SI European F ingerprintworking Group Second Annual Meeting; September
25th-27th, 2002. Institute of Forensic Research, Cracow, Poland, 2002.

6. Keith LV, Runion W. Short-wave UV Imaging Casework Applications. Journal of
Forensic Identiﬁcation 1998; 48(5):563-9.

7. Saferstein R, Graf SL. Evaluation of a reﬂected ultraviolet imaging system for
ﬁngerprint detection. Journal of Forensic Identiﬁcation 2001; 51(4):385-93.

8. Jones NE, Davies LM, Russell CAL, Brennan J S, Bramble SK. A systematic approach

to latent ﬁngerprint sample preparation for comparative chemical studies. Journal of
Forensic Identiﬁcation 2001 ; 51(5):504-15.

9. Ramotowski RS. Composition of Latent Print Residue. In: HC Lee, RE Gaensslen,
editors. Advances in Fingerprint Technology. 2nd edn, CRC Series in Forensic and
Police Science. Boca Raton, FL: CRC Press, 2001; 63-104.

10. Wertheim K. Fingerprint age determination: Is there any hope? Journal of Forensic
Identiﬁcation 2003; 53(1):42-9.

11. Archer NE, Charles Y, Elliott JA, Jickells S. Changes in the lipid composition of
latent ﬁngerprint residue with time after deposition on a surface. Forensic Science
International 2005; 154(2-3):224-39.

12. Lee HC, Gaensslen RE. Methods of Latent Fingerprint Development. In: HC Lee, RE

Gaensslen, editors. Advances in Fingerprint Technology. 2nd edn, CRC Series in
Forensic and Police Science. Boca Raton, FL: CRC Press, 2001; 105-76.

79

13. Leadbetter MJ. Use of automated ﬁngerprint identiﬁcation systems to process, search,
and identify palm prints and latent palm marks. Journal of Forensic Identiﬁcation 1999;
49(1):18-36.

14. IAF IS (Integrated Automated Fingerprint Identiﬁcation System). Department of
Justice, FBI Criminal Justice Information Services. 2005.
www.fbi.gov/hq/cjisd/iaﬁs.htm#main. Cited April 2007.

15. McCormick NJ. Confocal scanning Optical microscopy. In: D Brune, R Hellborg, HJ
Whitlow, O Hunderi, editors. Surface Characterization: A User's Sourcebook. Oslo:
Scnadinavian Science Publisher, 1997; 57776.

16. Image processing in ﬂuorescence microscopy. Wei Shu Hu's Group, Chemical
Engineering and Materials Science, University of Minnesota. 2006.
http://hugroup.cems.umn.edu/Research/imageproc.htrnl. Cited May 2007.

17. Mattsson L. Surface roughness and microtopography. In: D Brune, R Hellborg, HJ
Whitlow, O Hunderi, editors. Surface Characterization: A User's Sourcebook. Oslo:
Scandinavian Science Publisher, 1997; 82-100.

18. Adamson AW, Gast AP. Physical Chemistry of Surfaces. 6th ed. New York: John
Wiley & Sons, Inc., 1997.

19. Using AccuDyne TestTM Marker Pens to measure substrate surface energy. 2003.
http://www.accudynetest.com/pentest.htrnl. Cited March 2007.

20. Kristiansen K. Scanning electron microscope with energy- and wavelength-dispersive
spectrometry. In: D Brune, R Hellborg, HJ Whitlow, O Hunderi, editors. Surface
Characterization: A User's Sourcebook. Oslo: Scandinavian Science Publisher, 1997;
111-28.

21. Reimer L. Scanning Electron Microscopy: Physics of Image Formation and
Microanalysis. 2nd ed. P Hawkes, editor. Berlin: Springer-Verlag, 1998. (P Hawkes
editor. Springer Series in Optical Sciences ).

22. Miller JN, Miller JC. Statistics and Chemometrics for Analytical Chemsitry. 4th ed.
Harlow, England: Pearson Education Limited, 2000.

23. Massart DL, Kaufman L. The Interpretation of Analaytical Chemical Data by the Use
of Cluster Analysis. P Elving, J Winefordner, editors. New York: John Wiley and Sons,
1983. (P Elving, J Winefordner editors. Chemical analysis: a series of monographs on
analytical chemistry and its applications).

24. What is STYROFOAM? Dow Chemical Company. 2007.
www.dow.com/styrofoam/what.htrn. Cited April 2007.

80

25. Brune D, Hellborg R, Whitlow HJ, Hunderi O, editors. Surface Characterization: A
User's Sourcebook Oslo: Scandinavian Science Publisher, 1997.

26. Koons RD, Buscaglia J, Bottrell M, Miller E. Forensic glass comparisons. In: R

Saferstein, editor. Forensic Science Handbook. Upper Saddle River, NJ: Pearson
Education, Inc., 2002.

81

 

 

IIIIIIIIIIIIIIIIIIII

 

_ A ..A“ a‘