- Email: [email protected]
Short range shooting distance estimation using variable pressure SEM images of the surroundings of bullet holes in textiles
Accepted Manuscript Title: Short range shooting distance estimation using variable pressure SEM images of the surroundings of bullet holes in textiles Author: Ruth Hinrichs Paulo Ost Frank M.A.Z. Vasconcellos PII: DOI: Reference:
S0379-0738(16)30565-5 http://dx.doi.org/doi:10.1016/j.forsciint.2016.12.033 FSI 8711
To appear in:
FSI
Received date: Revised date: Accepted date:
18-8-2016 3-11-2016 21-12-2016
Please cite this article as: Ruth Hinrichs, Paulo Ost Frank, M.A.Z.Vasconcellos, Short range shooting distance estimation using variable pressure SEM images of the surroundings of bullet holes in textiles, Forensic Science International http://dx.doi.org/10.1016/j.forsciint.2016.12.033 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Short range shooting distance estimation using variable pressure SEM images of the surroundings of bullet holes in textiles Ruth Hinrichs1,2, Paulo Ost Frank1,3, M. A.Z. Vasconcellos1,4 1
PGCiMat Programa de Pós Graduação de Ciência dos Materiais, Universidade Federal do Rio Grande
do Sul (UFRGS), Av. Bento Gonçalves 9500, 91501-970 Porto Alegre, RS, Brazil 2
Instituto de Geociências, UFRGS
3
Instituto Geral de Perícias do Rio Grande do Sul, Departamento de Criminalística IGP-RS-DC
4
Instituto de Física, UFRGS
Highlights:
Low- vacuum SEM images used to determine GSR coverage Decay length of GSR coverage allows firing distance estimates Gunshot caused fiber modifications at different firing ranges
ABSTRACT Modifications of cotton and polyester textiles due to shots fired at short range were analyzed with a variable pressure scanning electron microscope (VP-SEM). Different mechanisms of fiber rupture as a function of fiber type and shooting distance were detected, namely fusing, melting, scorching, and mechanical breakage. To estimate the firing distance, the approximately exponential decay of GSR coverage as a function of radial distance from the entrance hole was determined from image analysis, instead of relying on chemical analysis with EDX, which is problematic in the VP-SEM. A set of backscattered electron images, with sufficient magnification to discriminate micrometer wide GSR particles, was acquired at different radial distances from the entrance hole. The atomic number contrast between the GSR particles and the organic fibers allowed to find a robust procedure to segment the micrographs into binary images, in which the white pixel count was attributed to GSR coverage. The decrease of the white pixel count followed an exponential decay, and it was found that the reciprocal of the decay constant, obtained from the least-square fitting of the coverage data, showed a linear dependence on the shooting distance. Key-words: textile fiber modification, shooting distance estimate; GSR coverage, variable pressure scanning electron microscope, image segmentation
1
1.
Introduction In many firearm related crimes, the range from which a weapon has been fired is an
important component in the reconstruction of the crime scene. The firing distance estimation is based on the examination of the appearance of the bullet entrance hole and the examination of gunshot residue (GSR) patterns around the hole [1]. The mixture of unburned and partially burnt propellant, amorphous sooty material, incandescent gases, and primer discharge residues expelled from the muzzle will deposit in front of the gun, causing vastly variable discharge residue patterns [2]. From the shape and size of the powder residue distribution (soot stains, powder tattooing) and with the knowledge of the weapon and ammunition, the distance from the target can be elucidated in many instances [3]. These patterns, often visible with the unaided eye, have been used to estimate shooting distances, sometimes using chemical processing with the Modified Griess or the Sodium Rhodizonate tests, to detect the presence of heavy metals [4]. Even though in many circumstances only rough estimates of shooting distance can be obtained, this often suffices to aid police in their inquiries [5]. In most of the cases in which there is a need for shooting distance estimation, the victim or the victims clothing have to be examined. When the cloth is of dark color, the discrimination of soot may be hindered [6], and alternatively chemical analysis of sampled locations around the entrance hole can help to elucidate the firing distance. Usually Pb, Sb and Ba are determined, because these are typical constituents of Sinoxid primers in conventional shotgun ammunition. However these analyses rely on dissolution or incineration of the sampled area, and the most abundant element, lead, which will allow for the most sensitive determinations, can be found as well in textiles that have not been involved in shooting [3]. False positives due to Sb detection have been reported as well, due to the usage of this element as fire retardant [7]. Instead of sample dissolution, microanalytical procedures with an energy dispersive X-ray spectrometer (EDX) in a scanning electron microscope (SEM) can be performed. The SEM has been utilized in forensic science since the early times of commercial electron microscopes [8]. The use of SEM/EDX for the analysis of GSR particles still attached to textile surfaces is relatively uncommon and labor intensive, demanding several dozens of micro analytical points [8]. In a conventional SEM the insulating 2
textile sample has to be coated with a conductive layer, and the entangled texture of the fibers does not allow the establishment of a continuous film. To be analyzed in the SEM, usually the particulate has to be collected from the textile, even though, depending on fiber entanglement, the recovering efficiency can be poor [9-12]. Once recovered, however, the analysis of GSR particles in SEM/EDX has become a standard procedure [13-15]. Morphological analysis of individual fibers can be performed in an uncoated condition, when the fibers are small and fixated on conductive carbon adhesive tape. Several authors have published SEM results on the morphology of fiber fracture. As early as 1971, Goynes and Rollins [16] studied the abrasion of cotton fibers due to successive cycles of washing and drying. In late 1990,
comprehensive books on
forensic fiber examination were published [17, 18] that included SEM analysis of textiles that were struck by a gunshot [19]. High temperature gases produced by the ignition of the propellant can modify the fibers, depending on their composition and on the muzzle to textile distance. But for all that, morphological fabric damage identification cannot be easily attributed to a determined cause, and there is still little research to support conclusions on ballistic impact on clothing [20]. Anyhow the microscopic analysis of fibers and fiber alterations has contributed greatly to elucidate crimes [21]. Since the development of variable pressure SEMs (VP-SEM), their use has been recommended for forensic analysis, due to the possibility to observe uncoated non conducting samples [22-24], preserving the pristine state of the evidence. The avoidance of time-consuming metallization procedures is a important advantage, yet it has to be kept in mind that several artifacts occur in low pressure atmosphere. The gas in the sample chamber, which prevents electrical charging of the sample, brings a severe disadvantage to micro-analytical resolution. A fraction of the beam electrons scatter on impact with gas molecules and reach the sample outside of the focalized beam, forming a "skirt" of electrons, which produce a faint halo on the sample surface, reducing contrast in the images [25], and generating spurious X-rays [26]. The radius (in meters) of the halo can be calculated using equation (1) [27]: 𝑟=
364.19𝑍 𝑝 1/2 𝐸
(𝑇)
𝐿3/2
(1)
3
where Z is the atomic number of the gas, p the pressure, T the temperature, L the traversed homogeneous gas thickness (all in SI units) and E the beam energy (in eV). In the SEM used in this work, the gas limiting aperture was located at the end of the pole piece, thus L equals the working distance (WD). For the conditions used in this work (WD of 18 mm, nitrogen pressure between 10-20 Pa, beam energy between 10 and 20 keV, and room temperature) the skirt radius was between 60 and 150 m. The implication of the skirts existence is that the microanalysis of small particles will always be affected by X-rays from the surroundings, and innocuous particles would appear to be contaminated with heavy elements present elsewhere in the sample. The effect of the skirt on image contrast is not as severe as on microanalysis. During the image acquisition the electrons scattered to the skirt generate a slowly varying background, which reduces the signal to noise ratio in each pixel and diminishes contrast. However the skirt does not impair image resolution, because the primary beam is still focused, even though some beam electrons get scattered into the skirt. The fraction of electrons (I/Io) that are still in the focused beam after traversing the distance between the pressure limiting aperture and the sample surface can be calculated using equation (2) [28]: 𝐼 𝐼0
= 𝑒𝑥𝑝 (−
𝑝×𝐿×𝜎𝑡 𝑘×𝑇
)
(2)
where t is the total scattering cross section of the gas in the sample chamber, k is the Boltzmann constant and p, T, and L have the same meaning as in equation (1). The scattering cross section of the gas depends on electron energy. For 10 keV electrons in nitrogen it amounts to 0.54 x 1021 m2 , and for 20 keV to 0.30 x 1021 m2 [29]. For the conditions used in this work, less than 5% of the beam electrons were scattered to the skirt, allowing image acquisition of uncoated textiles with satisfying levels of contrast. Among the immense variety of textiles, the most frequently found in crime scenes depend on local peculiarities, as dictated by climate, fashion and social surroundings. In Southern Brazil the most common garments on crime scene victims are cotton or polyester T-shirts and denim trousers. These observations guided our choice to perform a series of ballistic experiments on cotton, polyester, and mixed fabrics, in order to evaluate the effects of gun shots fired from various distances with the most common handgun in this region.
4
After the ballistic experiments, we obtained backscattered electron images (BEI) of the uncoated textiles in a VP-SEM, taking advantage of the atomic number contrast carried by these electrons that allows the discrimination of the high-Z (Pb, Ba, Sb) GSR particles against the background of the low-Z polymeric or organic fibers. Besides the compositional contrast, BEI micrographs show topographic contrast as well. In some cases it was possible to obtain additional information on the shooting distance observing the morphology of broken fiber ends, mainly of the synthetic ones. But independently if synthetic or cotton, the amount of GSR on the fibers and the decrease of GSR coverage as a function of radial distance to the entrance hole presented a systematic correlation to shooting distance for this ammunition, weapon, and firing ranges. So, instead of relying on microanalysis, which is problematic in VP-SEM, we focused on the BEI micrographs to estimate shooting distance. To assess the functional behavior of GSR coverage, several backscattered electron images, with sufficient magnification to discriminate micrometer wide GSR particles, were acquired at different radial distances from the entrance hole. Their color depth was decreased to two (binarization), using systematic adjustments of the segmentation threshold. The white pixel count per area on each image was attributed to GSR coverage ratio. Exponential decay functions were fitted to the GSR coverage profiles: 𝑦 = 𝐼0 exp(−𝜆𝑥)
(3)
where y is the predicted value, I0 is the maximum coverage, is the decay constant and x is the distance from the bullet hole. When the fit parameters obtained on samples hit by gunshot from different distances where compared, a linear relationship between the firing distance and the reciprocal of the decay constant ("decay length") was established. The decay length, by definition, is the distance to reduce the coverage to 1/e of its initial value. We adopted a different scaling, analogous to the half-life of radioactive decay, as the length, after which the coverage has fallen to half of the initial value, defining a GSR½=ln 2/. It has to be noted that the correlation to shooting distance is not established with the absolute white pixel count, which depends on subjective threshold choices in the segmentation procedure, but with the decay length, i.e. it is related to the distance necessary to halve the coverage.
5
The approximately exponential behavior, known as well from optical images of the decrease in sooting [3], can be seen in low magnification backscattered electron images around the bullet hole. However when the high magnification binarized images of the profile are quantified, outliers can occur, due to 10-100 micrometer wide lead blobs, or to tattooed unburned propellant in the image. These spots should be avoided during image acquisition, or eliminated from the white pixel count area, or else outliers have to be eliminated before mathematical fitting. The segmentation procedure performed by different operators will not produce the same coverage percentages due to different opinions on the threshold. However, once the operator has trained sufficient times to develop a procedure to establish the threshold, and follows it systematically in the binarization of the whole sequence of micrographs in the profile, the coverage of the images in this profile show an exponential decrease with roughly the same decay length. This decay length is proportional to the shooting distance in short range shots. Hence this is our proposition: to estimate the shooting distance from the decay length of the coverage, not from the coverage itself. This estimate is aided with clues from the fiber end modifications, observed in different kinds of fibers.
2.
Experimental
2.1. Textile sample perforation by gunshots fired at different ranges Three different textiles (white cotton jersey, blue denim, and red polyester jersey) were stapled to 20 cm x 30 cm cardboard rectangles and submitted to gunshots from different shooting ranges: soft contact, 2.5 cm, 5 cm, 10 cm, and 20 cm. The gun was a six-shot .38 SPL revolver with a 3 inch barrel (Taurus), using Brazilian centerfire .38 SPL cartridges (CBC), with unjacketed ogival lead bullets and conventional Sinoxid primer. 2.2. Sample characterization To identify the organic material of the synthetic fibers, Raman spectra were acquired in a Lab-assembled Raman microprobe, described elsewhere [30]. Even though the synthetic fibers of the red jersey and the denim presented strong fluorescence under the He-Ne laser, after 10 minutes of laser bleaching the fluorescence subsided and characteristic peaks appeared above the background, showing the main peaks of 6
polyester, identified by comparison with spectra from the literature [31]. It has been observed by other authors that polyester fibers are frequently used in denim to enhance its tensile and sensory properties [32]. The samples for SEM analysis were squares of approximately 5 cm x 5 cm, cut from the textiles in the vicinity of the bullet hole, and mounted on the SEM sample-holder with double-sided adhesive tape. The samples were observed in a dual beam low vacuum SEM (VP-SEM Jeol JIB4500), using a pressure of 10-20 Pa nitrogen in the sample chamber to avoid charging of the textiles. Fiber alterations were observed on high magnification images taken at the border of the holes. To establish the profiles of GSR coverage density, micrographs with resolution better than 1 m (1000 X magnification) were taken at determined distances of the bullet hole margin (0, 1, 2, 3, 4, 5, 7, 10, 15, 20, 25 mm and more). In order to establish the area fraction of GSR coverage, the images had their color depth reduced to two (image segmentation). The threshold, to decide if a shade of grey belongs to black or to white in the binary image, can be varied by the operator. The segmentation was performed using as criterion the apparently equal size of GSR particles (of 1-2 m) in the b&w image, when compared to the original micrograph. The procedure, performed with the public domain software Image J 1.44p [33], is demonstrated in Figure 1, showing the original micrograph, the threshold adjustment menu, and the resulting b&w image. The area used for the pixel count ratio was outlined in yellow using the freehand tool, avoiding labels, charged fibers, and unfocused regions of the picture. The white area fraction is established using the appropriate command from the Analyze>Set Measurements menu, and the desired information is registered in a result file. It has to be stressed that segmentation was based on the size comparison of small features (1 micron wide GSR particles) on the original micrograph and on the segmented image, while moving the threshold slider. This criterion failed in areas that were out of focus or presented charging, which had to be removed using the freehand area selection tool of the software.
7
3.
Results and discussion
3.1. Macroscopic aspects of the shot textiles The macroscopic effects around the bullet holes of the gunshots fired from different ranges on different textiles can be observed in Fig. 2.
The differences in sooting can be visualized best on the white cotton jersey. On the blue denim the soot and other residues are barely perceptible, making visual inspection difficult. In our experiment, sooting was strongest at 2.5 and 5 cm muzzle to cloth distances, while tattooing was most observable at the 20 cm shots. The bullet holes showed marked differences between cotton and polyester fabrics and between contact and shots from larger distances. The cotton jersey tore with little deformation in the contact shot, while after shots from 5 and 10 cm the fabric deformed strongly, buckling into the backing cardboard. The denim, being a woven textile, showed the least permanent deformation. Contact shots resulted in a peculiar damage: the cotton threads in the denim remained almost intact, while the synthetic warp gave way, producing a wide horizontal tear. In non-contact shots the fabric was torn at the whole circumference of the small circular hole. The polyester jersey, produced with thin filaments, was most damaged by the experiment. The contact shots and the 2.5 cm shot produced large holes, much wider than the diameter of the bullet. The textile deformed strongly and parts of it seemed to be missing. At larger distances the bullet holes were similar to those in cotton jersey, presenting a sharp border and almost no loss of fabric. The textiles presented little permanent deformation after the 20 cm shots, indicating that the rupture of the fibers occurred in the elastic deformation range of the fabric. 3.2. Microscopic aspects of the fibers BEI were taken around the frayed edges of the bullet holes, emphasizing fiber ends (Fig. 3). At contact shots, the cotton fibers in the jersey mostly broke with thickened ends due to scorching. At non-contact shots, scorching was seldom observed and most of the fibers broke with tapering ends due to mechanical extension. The fibers in the cotton yarns of the denim showed similar behavior as those of the cotton jersey at all shooting distances. The synthetic fibers of the denim behaved 8
markedly different. At contact shots the temperature was high enough to melt the polyester (>250 ºC) and to find fused fiber endings, comprising tens of filaments. The coverage with GSR of the cotton weft and the "shadow" of the synthetic warp on the cotton fibers can be observed on the frayed margin of the bullet hole of a soft contact shot on denim (Fig. 4a). The cotton weft resisted while the synthetic warp ruptured due to melting (Fig. 4b). This explains the cause of the horizontal tearing in contact shots: the polyester fibers were heated above their melting point and gave way, leaving the cotton yarn unbroken, even though scorched. At the 2.5 cm and 5 cm shots the polyester fibers still show thickened ends, indicating the occurrence of scorching. At higher firing distances, the synthetic fibers broke with short, tapered endings, different from the slow tapering observed in broken cotton fibers (Fig. 5).
The polyester jersey, with thin synthetic fibers, showed the most evident microscopic damage, as expected form visual inspection of the textiles. The yarns, unprotected from the cotton weft as in the denim, fused in contact shots and in shots fired from 2.5 and 5 cm. The fibers showed scorching, even at 10 cm and 20 cm shots (Fig. 6).
Fiber alterations on the border of the bullet holes are shown in Fig. 7, organized in columns of cotton/synthetic fibers and rows of different shooting ranges. 3.3. GSR coverage of the fibers A mosaic of nine micrographs (BEI) from the cotton jersey submitted to a 5 cm shot shows the distribution of GSR over the knitted yarn in Fig. 8. The fraying of the threads and the tearing of the cotton fibers on the margin of the bullet hole can be seen. A radial decrease in the GSR coverage can be seen and unburned propellant particles are indicated with arrows. As has been described by other authors [3], the macroscopic powder-soot distribution around the bullet hole can be roughly approximated by an exponential distribution. This is observed as well for the high Z particles that can be discriminated with the aid of the 9
backscattered electron detector. The main deviation of this decay is due to bigger blobs of high Z material that can be easily be recognized on the images. It is expected to find irregularities in the microscopic sampling on random radial profiles, considering that the bullet drag is turbulent. Different densities of the GSR deposits as a function of the distance from the bullet hole border from a contact shot on the denim sample are shown in Fig. 9.
The grid of images in Fig. 10a shows the distribution of GSR at different radial distances from the bullet holes, produced by gunshots fired from different ranges at polyester jersey. Fig. 10b displays the same images after the segmentation procedures.
The percentage of GSR coverage, obtained by a single operator with a consistent threshold choice from the images in Fig. 10b, as a function of radial distance from the entry holes is exemplified in the five plots of Fig. 11. To be able to fit an exponential curve on the data, some points had to be considered outliers and are indicated with lighter symbols. Frequently the first point (zero mm) had to be ignored, and a zero % coverage point at higher radial distance (60 mm) had to be added, to avoid a constant background fit. The parameters A and obtained from the exponential fits shown in (y = A exp-x) are summarized in Table 1, as well as the length GSR½, after which the coverage has fallen to half of the initial value.
When other operator choices for the threshold segmentation were used, the obtained decay length fell within ± 15% of the values of Table 1. This uncertainty was added to the standard deviation of the fit in the plot against the firing distance, shown in Fig. 12. The GSR½ length presented an almost linear relationship to the firing distance. It can be seen that the distance from the bullet hole border at which the GSR coverage had fallen to half is approximately a tenth of the shooting distance (muzzle to cloth distance). As a rule of thumb it can be said that if the coverage divides in half every 10 10
mm, the shot was fired from approximately 10 cm distance. A helpful feature from exponential decay is that the decay constant can be determined anywhere on the radial profile, i.e. it is not necessary to start at the center of the bullet hole (usually torn away), but can begin at any point on the radial profile. It can be seen that at 20 cm shots the uncertainty reaches a level that implies that the estimation should be restricted to short range shots. In the literature the decrease in GSR coverage has been estimated at the surroundings of bullet holes using the antimony content determined with neutron activation analysis [34]. There the shots were fired with a .22 rifle and the author's conclusion was that the concentration profiles obeyed an exponential decay function of the square of the radial distance from the edge. Our findings showed an exponential dependence of the nonsquared radial distance, but the differences in weapon, firing ranges, and analysis method might account for that. The same author mentioned in a later paper [35] that at firing distances smaller than 30 cm the exponential decay dependence of the squared radial distance was not followed.
4. Conclusions The modifications of the textiles that allowed to estimate firing ranges were of two kinds: i) a qualitative one, based on different alterations of the fibers on the edge of the holes; ii) a quantitative estimation, determining different characteristic lengths of the exponential decrease of GSR coating densities on radial profiles around the holes. Molten and fused polyester fibers in mixed textiles were observed only at contact shots, resulting in a peculiar horizontal tear around the bullet hole. In purely synthetic textiles, mainly with thin filaments, fusing of fibers was still observed at 2.5 cm shots. At this distance the fibers did not melt immediately, as in the contact shots, but the knitwear deformed strongly before melting, presenting the most visible damage. Above 5 cm muzzle-to-textile distance, no fusing of fibers was observed anymore, but scorched fiber ends could always be found. In the cotton jersey the microscopic damage was least discernible, and only at contact shots scorched fiber ends could be detected. In the noncontact shots only mechanical breakage was observed. The method to quantitatively estimate firing distance using the reciprocal of the exponential decay constant (GSR1/2), extracted from the GSR bright contrast on 11
backscattered electron images obtained on radial profiles, was independent of the type of fibers in the textile. For the handgun and ammunition type used in this study, the decay length of the coverage presented a linear relationship with the firing distance and allowed an estimation of the latter.
Acknowledgements The authors thank the Laboratório de Conformação Nanométrica and Laboratório de Microanálise, Physics Institute, UFRGS, Porto Alegre, RS, Brazil, for the use of the SEM facilities. The authors acknowledge funding from the Brazilian agencies FINEP, CAPES, CNPq, and FAPERGS.
References
[1] A. Zeichner, B. Glattstein, Recent Developments in the Methods of Estimating Shooting Distance, The Sci World J 2 (2002) 573-585. doi: 10.1100/tsw.2002.140 [2] H. Ditrich, Distribution of gunshot residues - the influence of weapon type, Forensic Sci. Int. 220 (2012) 85-90. doi:10.1016/j.forsciint.2012.01.034 [3] K. Sellier, Forensic Science Progress Volume 6 Shot range determination, Springer, Heidelberg, 1991. ISBN-13: 978-3-642-76723-4 [4] A.M. Cimino, J.E. Debney, Distance of Shot from Discharge Residues of Rifled Weapons, J. Forensic Sci. 8 (1968) 8-11. [5]. Collaborative Testing Services, Inc. Forensic Testing program. GSR-Distance Determination Test No. 14-530 Summary report; 2014. https://www.ctsforensics.com/assets/news/3430_Web.pdf [6] K.L. Kersh, J.M. Childers, D. Justice, G. Karim, Detection of Gunshot Residue on Dark-Colored Clothing Prior to Chemical Analysis; J Forensic Sci 59(3) (2014) 754-762. doi: 10.1111/1556-4029.12409 [7] C. Aksoy, T. Bora, N. Şenocak, F.Aydın, A new method to reduce false positives due to antimony in detection of gunshot residues, Forensic Sci. Int. 250 (2015) 8790. doi: 10.1016/j.forsciint.2015.03.006 [8] V.J.M. DiMaio, Gunshot Wounds Practical Aspects of Firearms, Ballistics, and Forensic Techniques, CRC Press, Boca Raton, 1999. ISBN 0-8493-8163-0 [9] J. Andrasko, S. Petterson, A simple method for collection of gunshot residues from clothing, J. Forensic Sci. 31 (1991) 321-330. 12
[10] Z. Brożek-Mucha, Distribution and properties of gunshot residue originating from a Luger 9 mm ammunition in the vicinity of the shooting gun, Forensic Sci. Int. 183 (2009) 33-44. doi:10.1016/j.forsciint.2008.10.010 [11] M.R. Rijnders, A. Stamouli, A. Bolck, Comparison of GSR Composition Occurring at Different Locations Around the Firing Position, J. Forensic Sci.55 (2010) 616-623. doi 10.1111/j.1556-4029.2009.01292.x [12] S. Charles, M. Lannoy, N. Geusens, Influence of the type of fabric on the collection efficiency of gunshot residues, Forensic Sci. Int. 228 (2013) 42-46. doi: 10.1016/j.forsciint.2013.02.022 [13] L. Niewoehner, H. W. Wenz, J. Andrasko, R. Beijer, L. Gunaratnam, ENFSI Proficiency Test Program on Identification of GSR by SEM/EDX*, J. Forensic Sci. 48(4) (2003) 1-8. Paper ID JFS2002396_484 www.astm.org. [14] American Society for Testing and Materials, ASTM E1588-10e1: Standard Guide for Gunshot Residue Analysis by Scanning Electron Microscopy/Energy Dispersive X-Ray Spectrometry, American Society for Testing and Materials, West Conshohocken, PA, 2010. [15] A. Zeichner, Recent developments in methods of chemical analysis in investigations of firearm-related events, Anal Bioanal Chem 376 (2003) 1178– 1191. doi: 10.1007/s00216-003-1994-y [16] W.R. Goynes, M.L. Rollins, A scanning electron microscope study of washer dryer abrasion in cotton fibers, Textile Res. J. 3 (1971) 226-231. [17] J.W.S. Hearle, B. Lomas, W.D. Cooke, Atlas of fiber fracture and damage to textiles, 2nd ed., CRC Woodhead Publishing, Boca Raton, 1998. ISBN 9781855733190 [18] J. Robertson, M. Grieve, Forensic examination of fibers, Taylor&Francis, London, 1999. ISBN 9780203484517 [19] W. Pelton, Use of SEM in textile forensic work, In: J.W.S. Hearle, B. Lomas, W.D. Cooke, Atlas of fiber fracture and damage to textiles, 2nd ed., CRC Woodhead Publishing, Boca Raton, 1998. ISBN 9781855733190. [20] F. Poole, M. Pailthorpe, Comparison of bullet and knife damage, In: J.W.S. Hearle, B. Lomas, W.D. Cooke, Atlas of fiber fracture and damage to textiles, 2nd ed., CRC Woodhead Publishing, Boca Raton, 1998. ISBN 9781855733190. [21] K. De Wael, L. Lepot, K. Lunstroot, F. Gason, 10 years of 1:1 taping in Belgium — A selection of murder cases involving fibre examination, Sci. Justice 56 (2016) 18-28. http://dx.doi.org/10.1016/j.scijus.2015.07.002 [22] H. Krüsemann, SEMs in forensic science, Problems Forensic Sci. 47 (2001) 110121. [23] Z. Brożek-Mucha, Variation of the chemical contents and morphology of gunshot residue in the surroundings of the shooting pistol as a potential contribution to a 13
shooting incidence reconstruction, Forensic Sci. Int. 210 (2011) 31-41. doi:10.1016/j.forsciint.2011.01.031 [24] M. Chohra, B. Beladel, L.B. Ahmed, M. Mouzai, D. Akretche, A. Zeghdaoui, A. Mansouri, M.E.A. Benamar, Study of gunshot residue by NAA and ESEM/EDX using several kinds of weapon and ammunition, J. Rad. Res. App. Sci., 2015, 8(3), 404.-410. doi 10.1016/j.jrras.2015.02.01 [25] D.J. Stokes, Principles and Practice of Variable Pressure/Environmental Scanning Electron Microscopy (VP-ESEM), John Wiley & Sons, Chichester, 2008. ISBN 978-0-470-06540-2 [26] D.E. Newbury, X-Ray Microanalysis in the Variable Pressure (Environmental) Scanning Electron Microscope, J. Res. Natl. Inst. Stand. Technol. 107 (2002) 567– 603. Available online: http://www.nist.gov/jres [27] G.D. Danilatos, Implication of the figure of merit in environmental SEM, Micron 44 (2013) 143-149. doi:10.1016/j.micron.2012.06.001 [28] L. Kouchaf, C. Mathieu, A. Kadoun, Electron microbeam changes under gaseous environment: CP-SEM case and microanalysis limits, in A.Mendez-Vilas, J. Diaz: Microscopy: Science, Technology, Applications and Education, Microscopy Series nº4 vol. 3. FORMATEX , Badajoz, 2010. ISBN-13: 978-84-614-6191-2 [29] A. Kadoun, R. Belkorissat, B. Khelifa, C. Mathieu, Comparative study of electron beam gas interactions in an SEM operating at pressures up to 300 Pa, Vacuum 69 (2003) 537-543. [30] R. Hinrichs, M.T. Brown, M.A.Z. Vasconcellos, M.V. Abrashev, W. Kalkreuth, Simple procedure for an estimation of the coal rank using micro-Raman spectroscopy, Int. J. Coal Geol. 136 (2014) 52-58. doi: 10.1016/j.coal.2014.10.013 [31] Jochem, G., Lehnert, R.J.; On the potential of Raman microscopy for the forensic analysis of coloured textile fibres; Sci & Justice 42 (2002) 215-221. [32] De Wael, K., Baes, C., Lepot, L., Gason, F.; On the frequency of occurrence of a peculiar polyester fi bre type found in blue denim textiles; Sci. Justice 51 (2011) 154-162. doi:10.1016/j.scijus.2011.02.001 [33] Rasband, W.; National Institute of Health, USA, http://imagej.nih.gov/ij [34] Bhattacharyya, C.N.; Dispersion of firing discharge residues using a Maxwellian model; Forensic Sci. Int. 42 (1989) 271-277. [35] Bhattacharyya, C.N.; Dispersion of firing discharge residues using a modified Maxwellian model; Forensic Sci. Int. 47 (1990) 31-37.
14
a)
b)
Fig. 1: a) BEI micrograph taken from the polyester jersey at 3 mm from the bullet hole of a soft contact shot. The horizontal arrow indicates a "charged" fiber, the vertical arrow an "out of focus" area (inset: threshold menu of ImageJ); b) binary image obtained with the "threshold" command. The yellow outline on both images indicates the area used to measure the white/black pixel count ratio.
Fig. 2: Sooting and tattooing of the textile samples after gunshots from different ranges (distances indicated on top row). Top to bottom: white cotton jersey, blue polyester/cotton denim, red polyester jersey. Scale bars are 5 cm long.
15
a)
b)
Fig. 3: Yielding of cotton fibers at a) contact shot, with thickened, scorched fiber ends; b) at a firing distance of 10 cm, showing tapered fiber endings. Scale bar on both images 50 m).
a)
b)
Fig. 4: a) BEI of the bullet hole (soft contact) in the denim, showing GSR coverage of the cotton threads, except for the shadow of the synthetic warp (scale bar 1 mm); b) molten and fused filaments of the synthetic warp (scale bar 100 m).
16
Fig. 5: Yielding of polyester fibers at a) 5 cm shot, with thickened, scorched fiber ends; b) at a firing distance of 20 cm, showing tapered fiber endings. Scale bar on both images 50 m).
a)
b)
c)
d)
Fig. 6: Damage on thin polyester fibers at a) contact shot, with fused fiber ends; b) firing distance of 2.5 cm, with thickened fibers and limited fusing; c) firing distance of 5 cm with thickened fibers due to scorching; d) at a firing distance of 20 cm with minor scorching and tapering of the fibers. Scale bar on all images 50 m.
17
Fig. 7: BEI of the fibers on the border of bullet holes: a) cotton (jersey); b) cotton (denim); c) polyester (denim); d) polyester (jersey). The rows show different muzzle to textile distances (HC: hard contact, SC: soft contact, other distances in mm; all images have a width of 570 m).
18
Fig. 8: BEI mosaic (uncoated sample observed in low vacuum of 15Pa), showing frayed fibers at the margin of the bullet hole from a 5 cm shot on cotton jersey. GSR on the fibers (white contrast) and unburned propellant particles (arrows) can be discriminated.
a)
b)
c)
d)
Fig. 9: Density of GSR coverage in a contact shot on denim at: a) 3 mm; b) 5 mm; c) 7 mm; and d) 9 mm from the border of the bullet hole (image widths 95 m).
19
a)
soft contact
2.5cm
5 cm
10 cm
20 cm
soft contact
2.5cm
5 cm
b)
10 cm
20 cm
Fig. 10: GSR distribution on polyester jersey as a function of radial distance from the bullet hole (in mm, indicated in first column), and as a function of the firing range (indicated in first row): a) BEI micrographs; b) binarized images. Width of all images is 128 m.
20
a)
b)
d)
c)
e)
Fig. 11: GSR coverage as a function of radial distance from the bullet holes for different firing ranges: a) soft contact; b) 2.5 cm; c) 5 cm; d) 10 cm; e) 20 cm. The bold data points were used for the exponential decay fit (grey lines). Error bars were estimated to be 10% of the pixel count statistic.
21
Fig. 12: GSR½ as a function of shooting distance, linear fit, 95% confidence and 95% prediciton levels. Inset is the equation of the error weighed linear fit (uncertainties in parentheses).
22
Table 1: Mean and uncertainty () of the fit parameters I0 and GSR½ of the exponential fit to the GSR coverages obtained on radial profiles on polyester jersey, fired at from different distances Firing distance (cm) Soft contact 2.5 5 10 20
I0 (%) Mean 22.4 14.1 16.5 9.7 6.1
GSR½= ln2/(mm)
2 1.9 1.9 0.8 2
Mean 0.6 2.8 4.1 8.8 14.7
0.04 0.9 1.1 1.9 10.6
23
S0379-0738(16)30565-5 http://dx.doi.org/doi:10.1016/j.forsciint.2016.12.033 FSI 8711
To appear in:
FSI
Received date: Revised date: Accepted date:
18-8-2016 3-11-2016 21-12-2016
Please cite this article as: Ruth Hinrichs, Paulo Ost Frank, M.A.Z.Vasconcellos, Short range shooting distance estimation using variable pressure SEM images of the surroundings of bullet holes in textiles, Forensic Science International http://dx.doi.org/10.1016/j.forsciint.2016.12.033 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Short range shooting distance estimation using variable pressure SEM images of the surroundings of bullet holes in textiles Ruth Hinrichs1,2, Paulo Ost Frank1,3, M. A.Z. Vasconcellos1,4 1
PGCiMat Programa de Pós Graduação de Ciência dos Materiais, Universidade Federal do Rio Grande
do Sul (UFRGS), Av. Bento Gonçalves 9500, 91501-970 Porto Alegre, RS, Brazil 2
Instituto de Geociências, UFRGS
3
Instituto Geral de Perícias do Rio Grande do Sul, Departamento de Criminalística IGP-RS-DC
4
Instituto de Física, UFRGS
Highlights:
Low- vacuum SEM images used to determine GSR coverage Decay length of GSR coverage allows firing distance estimates Gunshot caused fiber modifications at different firing ranges
ABSTRACT Modifications of cotton and polyester textiles due to shots fired at short range were analyzed with a variable pressure scanning electron microscope (VP-SEM). Different mechanisms of fiber rupture as a function of fiber type and shooting distance were detected, namely fusing, melting, scorching, and mechanical breakage. To estimate the firing distance, the approximately exponential decay of GSR coverage as a function of radial distance from the entrance hole was determined from image analysis, instead of relying on chemical analysis with EDX, which is problematic in the VP-SEM. A set of backscattered electron images, with sufficient magnification to discriminate micrometer wide GSR particles, was acquired at different radial distances from the entrance hole. The atomic number contrast between the GSR particles and the organic fibers allowed to find a robust procedure to segment the micrographs into binary images, in which the white pixel count was attributed to GSR coverage. The decrease of the white pixel count followed an exponential decay, and it was found that the reciprocal of the decay constant, obtained from the least-square fitting of the coverage data, showed a linear dependence on the shooting distance. Key-words: textile fiber modification, shooting distance estimate; GSR coverage, variable pressure scanning electron microscope, image segmentation
1
1.
Introduction In many firearm related crimes, the range from which a weapon has been fired is an
important component in the reconstruction of the crime scene. The firing distance estimation is based on the examination of the appearance of the bullet entrance hole and the examination of gunshot residue (GSR) patterns around the hole [1]. The mixture of unburned and partially burnt propellant, amorphous sooty material, incandescent gases, and primer discharge residues expelled from the muzzle will deposit in front of the gun, causing vastly variable discharge residue patterns [2]. From the shape and size of the powder residue distribution (soot stains, powder tattooing) and with the knowledge of the weapon and ammunition, the distance from the target can be elucidated in many instances [3]. These patterns, often visible with the unaided eye, have been used to estimate shooting distances, sometimes using chemical processing with the Modified Griess or the Sodium Rhodizonate tests, to detect the presence of heavy metals [4]. Even though in many circumstances only rough estimates of shooting distance can be obtained, this often suffices to aid police in their inquiries [5]. In most of the cases in which there is a need for shooting distance estimation, the victim or the victims clothing have to be examined. When the cloth is of dark color, the discrimination of soot may be hindered [6], and alternatively chemical analysis of sampled locations around the entrance hole can help to elucidate the firing distance. Usually Pb, Sb and Ba are determined, because these are typical constituents of Sinoxid primers in conventional shotgun ammunition. However these analyses rely on dissolution or incineration of the sampled area, and the most abundant element, lead, which will allow for the most sensitive determinations, can be found as well in textiles that have not been involved in shooting [3]. False positives due to Sb detection have been reported as well, due to the usage of this element as fire retardant [7]. Instead of sample dissolution, microanalytical procedures with an energy dispersive X-ray spectrometer (EDX) in a scanning electron microscope (SEM) can be performed. The SEM has been utilized in forensic science since the early times of commercial electron microscopes [8]. The use of SEM/EDX for the analysis of GSR particles still attached to textile surfaces is relatively uncommon and labor intensive, demanding several dozens of micro analytical points [8]. In a conventional SEM the insulating 2
textile sample has to be coated with a conductive layer, and the entangled texture of the fibers does not allow the establishment of a continuous film. To be analyzed in the SEM, usually the particulate has to be collected from the textile, even though, depending on fiber entanglement, the recovering efficiency can be poor [9-12]. Once recovered, however, the analysis of GSR particles in SEM/EDX has become a standard procedure [13-15]. Morphological analysis of individual fibers can be performed in an uncoated condition, when the fibers are small and fixated on conductive carbon adhesive tape. Several authors have published SEM results on the morphology of fiber fracture. As early as 1971, Goynes and Rollins [16] studied the abrasion of cotton fibers due to successive cycles of washing and drying. In late 1990,
comprehensive books on
forensic fiber examination were published [17, 18] that included SEM analysis of textiles that were struck by a gunshot [19]. High temperature gases produced by the ignition of the propellant can modify the fibers, depending on their composition and on the muzzle to textile distance. But for all that, morphological fabric damage identification cannot be easily attributed to a determined cause, and there is still little research to support conclusions on ballistic impact on clothing [20]. Anyhow the microscopic analysis of fibers and fiber alterations has contributed greatly to elucidate crimes [21]. Since the development of variable pressure SEMs (VP-SEM), their use has been recommended for forensic analysis, due to the possibility to observe uncoated non conducting samples [22-24], preserving the pristine state of the evidence. The avoidance of time-consuming metallization procedures is a important advantage, yet it has to be kept in mind that several artifacts occur in low pressure atmosphere. The gas in the sample chamber, which prevents electrical charging of the sample, brings a severe disadvantage to micro-analytical resolution. A fraction of the beam electrons scatter on impact with gas molecules and reach the sample outside of the focalized beam, forming a "skirt" of electrons, which produce a faint halo on the sample surface, reducing contrast in the images [25], and generating spurious X-rays [26]. The radius (in meters) of the halo can be calculated using equation (1) [27]: 𝑟=
364.19𝑍 𝑝 1/2 𝐸
(𝑇)
𝐿3/2
(1)
3
where Z is the atomic number of the gas, p the pressure, T the temperature, L the traversed homogeneous gas thickness (all in SI units) and E the beam energy (in eV). In the SEM used in this work, the gas limiting aperture was located at the end of the pole piece, thus L equals the working distance (WD). For the conditions used in this work (WD of 18 mm, nitrogen pressure between 10-20 Pa, beam energy between 10 and 20 keV, and room temperature) the skirt radius was between 60 and 150 m. The implication of the skirts existence is that the microanalysis of small particles will always be affected by X-rays from the surroundings, and innocuous particles would appear to be contaminated with heavy elements present elsewhere in the sample. The effect of the skirt on image contrast is not as severe as on microanalysis. During the image acquisition the electrons scattered to the skirt generate a slowly varying background, which reduces the signal to noise ratio in each pixel and diminishes contrast. However the skirt does not impair image resolution, because the primary beam is still focused, even though some beam electrons get scattered into the skirt. The fraction of electrons (I/Io) that are still in the focused beam after traversing the distance between the pressure limiting aperture and the sample surface can be calculated using equation (2) [28]: 𝐼 𝐼0
= 𝑒𝑥𝑝 (−
𝑝×𝐿×𝜎𝑡 𝑘×𝑇
)
(2)
where t is the total scattering cross section of the gas in the sample chamber, k is the Boltzmann constant and p, T, and L have the same meaning as in equation (1). The scattering cross section of the gas depends on electron energy. For 10 keV electrons in nitrogen it amounts to 0.54 x 1021 m2 , and for 20 keV to 0.30 x 1021 m2 [29]. For the conditions used in this work, less than 5% of the beam electrons were scattered to the skirt, allowing image acquisition of uncoated textiles with satisfying levels of contrast. Among the immense variety of textiles, the most frequently found in crime scenes depend on local peculiarities, as dictated by climate, fashion and social surroundings. In Southern Brazil the most common garments on crime scene victims are cotton or polyester T-shirts and denim trousers. These observations guided our choice to perform a series of ballistic experiments on cotton, polyester, and mixed fabrics, in order to evaluate the effects of gun shots fired from various distances with the most common handgun in this region.
4
After the ballistic experiments, we obtained backscattered electron images (BEI) of the uncoated textiles in a VP-SEM, taking advantage of the atomic number contrast carried by these electrons that allows the discrimination of the high-Z (Pb, Ba, Sb) GSR particles against the background of the low-Z polymeric or organic fibers. Besides the compositional contrast, BEI micrographs show topographic contrast as well. In some cases it was possible to obtain additional information on the shooting distance observing the morphology of broken fiber ends, mainly of the synthetic ones. But independently if synthetic or cotton, the amount of GSR on the fibers and the decrease of GSR coverage as a function of radial distance to the entrance hole presented a systematic correlation to shooting distance for this ammunition, weapon, and firing ranges. So, instead of relying on microanalysis, which is problematic in VP-SEM, we focused on the BEI micrographs to estimate shooting distance. To assess the functional behavior of GSR coverage, several backscattered electron images, with sufficient magnification to discriminate micrometer wide GSR particles, were acquired at different radial distances from the entrance hole. Their color depth was decreased to two (binarization), using systematic adjustments of the segmentation threshold. The white pixel count per area on each image was attributed to GSR coverage ratio. Exponential decay functions were fitted to the GSR coverage profiles: 𝑦 = 𝐼0 exp(−𝜆𝑥)
(3)
where y is the predicted value, I0 is the maximum coverage, is the decay constant and x is the distance from the bullet hole. When the fit parameters obtained on samples hit by gunshot from different distances where compared, a linear relationship between the firing distance and the reciprocal of the decay constant ("decay length") was established. The decay length, by definition, is the distance to reduce the coverage to 1/e of its initial value. We adopted a different scaling, analogous to the half-life of radioactive decay, as the length, after which the coverage has fallen to half of the initial value, defining a GSR½=ln 2/. It has to be noted that the correlation to shooting distance is not established with the absolute white pixel count, which depends on subjective threshold choices in the segmentation procedure, but with the decay length, i.e. it is related to the distance necessary to halve the coverage.
5
The approximately exponential behavior, known as well from optical images of the decrease in sooting [3], can be seen in low magnification backscattered electron images around the bullet hole. However when the high magnification binarized images of the profile are quantified, outliers can occur, due to 10-100 micrometer wide lead blobs, or to tattooed unburned propellant in the image. These spots should be avoided during image acquisition, or eliminated from the white pixel count area, or else outliers have to be eliminated before mathematical fitting. The segmentation procedure performed by different operators will not produce the same coverage percentages due to different opinions on the threshold. However, once the operator has trained sufficient times to develop a procedure to establish the threshold, and follows it systematically in the binarization of the whole sequence of micrographs in the profile, the coverage of the images in this profile show an exponential decrease with roughly the same decay length. This decay length is proportional to the shooting distance in short range shots. Hence this is our proposition: to estimate the shooting distance from the decay length of the coverage, not from the coverage itself. This estimate is aided with clues from the fiber end modifications, observed in different kinds of fibers.
2.
Experimental
2.1. Textile sample perforation by gunshots fired at different ranges Three different textiles (white cotton jersey, blue denim, and red polyester jersey) were stapled to 20 cm x 30 cm cardboard rectangles and submitted to gunshots from different shooting ranges: soft contact, 2.5 cm, 5 cm, 10 cm, and 20 cm. The gun was a six-shot .38 SPL revolver with a 3 inch barrel (Taurus), using Brazilian centerfire .38 SPL cartridges (CBC), with unjacketed ogival lead bullets and conventional Sinoxid primer. 2.2. Sample characterization To identify the organic material of the synthetic fibers, Raman spectra were acquired in a Lab-assembled Raman microprobe, described elsewhere [30]. Even though the synthetic fibers of the red jersey and the denim presented strong fluorescence under the He-Ne laser, after 10 minutes of laser bleaching the fluorescence subsided and characteristic peaks appeared above the background, showing the main peaks of 6
polyester, identified by comparison with spectra from the literature [31]. It has been observed by other authors that polyester fibers are frequently used in denim to enhance its tensile and sensory properties [32]. The samples for SEM analysis were squares of approximately 5 cm x 5 cm, cut from the textiles in the vicinity of the bullet hole, and mounted on the SEM sample-holder with double-sided adhesive tape. The samples were observed in a dual beam low vacuum SEM (VP-SEM Jeol JIB4500), using a pressure of 10-20 Pa nitrogen in the sample chamber to avoid charging of the textiles. Fiber alterations were observed on high magnification images taken at the border of the holes. To establish the profiles of GSR coverage density, micrographs with resolution better than 1 m (1000 X magnification) were taken at determined distances of the bullet hole margin (0, 1, 2, 3, 4, 5, 7, 10, 15, 20, 25 mm and more). In order to establish the area fraction of GSR coverage, the images had their color depth reduced to two (image segmentation). The threshold, to decide if a shade of grey belongs to black or to white in the binary image, can be varied by the operator. The segmentation was performed using as criterion the apparently equal size of GSR particles (of 1-2 m) in the b&w image, when compared to the original micrograph. The procedure, performed with the public domain software Image J 1.44p [33], is demonstrated in Figure 1, showing the original micrograph, the threshold adjustment menu, and the resulting b&w image. The area used for the pixel count ratio was outlined in yellow using the freehand tool, avoiding labels, charged fibers, and unfocused regions of the picture. The white area fraction is established using the appropriate command from the Analyze>Set Measurements menu, and the desired information is registered in a result file. It has to be stressed that segmentation was based on the size comparison of small features (1 micron wide GSR particles) on the original micrograph and on the segmented image, while moving the threshold slider. This criterion failed in areas that were out of focus or presented charging, which had to be removed using the freehand area selection tool of the software.
7
3.
Results and discussion
3.1. Macroscopic aspects of the shot textiles The macroscopic effects around the bullet holes of the gunshots fired from different ranges on different textiles can be observed in Fig. 2.
The differences in sooting can be visualized best on the white cotton jersey. On the blue denim the soot and other residues are barely perceptible, making visual inspection difficult. In our experiment, sooting was strongest at 2.5 and 5 cm muzzle to cloth distances, while tattooing was most observable at the 20 cm shots. The bullet holes showed marked differences between cotton and polyester fabrics and between contact and shots from larger distances. The cotton jersey tore with little deformation in the contact shot, while after shots from 5 and 10 cm the fabric deformed strongly, buckling into the backing cardboard. The denim, being a woven textile, showed the least permanent deformation. Contact shots resulted in a peculiar damage: the cotton threads in the denim remained almost intact, while the synthetic warp gave way, producing a wide horizontal tear. In non-contact shots the fabric was torn at the whole circumference of the small circular hole. The polyester jersey, produced with thin filaments, was most damaged by the experiment. The contact shots and the 2.5 cm shot produced large holes, much wider than the diameter of the bullet. The textile deformed strongly and parts of it seemed to be missing. At larger distances the bullet holes were similar to those in cotton jersey, presenting a sharp border and almost no loss of fabric. The textiles presented little permanent deformation after the 20 cm shots, indicating that the rupture of the fibers occurred in the elastic deformation range of the fabric. 3.2. Microscopic aspects of the fibers BEI were taken around the frayed edges of the bullet holes, emphasizing fiber ends (Fig. 3). At contact shots, the cotton fibers in the jersey mostly broke with thickened ends due to scorching. At non-contact shots, scorching was seldom observed and most of the fibers broke with tapering ends due to mechanical extension. The fibers in the cotton yarns of the denim showed similar behavior as those of the cotton jersey at all shooting distances. The synthetic fibers of the denim behaved 8
markedly different. At contact shots the temperature was high enough to melt the polyester (>250 ºC) and to find fused fiber endings, comprising tens of filaments. The coverage with GSR of the cotton weft and the "shadow" of the synthetic warp on the cotton fibers can be observed on the frayed margin of the bullet hole of a soft contact shot on denim (Fig. 4a). The cotton weft resisted while the synthetic warp ruptured due to melting (Fig. 4b). This explains the cause of the horizontal tearing in contact shots: the polyester fibers were heated above their melting point and gave way, leaving the cotton yarn unbroken, even though scorched. At the 2.5 cm and 5 cm shots the polyester fibers still show thickened ends, indicating the occurrence of scorching. At higher firing distances, the synthetic fibers broke with short, tapered endings, different from the slow tapering observed in broken cotton fibers (Fig. 5).
The polyester jersey, with thin synthetic fibers, showed the most evident microscopic damage, as expected form visual inspection of the textiles. The yarns, unprotected from the cotton weft as in the denim, fused in contact shots and in shots fired from 2.5 and 5 cm. The fibers showed scorching, even at 10 cm and 20 cm shots (Fig. 6).
Fiber alterations on the border of the bullet holes are shown in Fig. 7, organized in columns of cotton/synthetic fibers and rows of different shooting ranges. 3.3. GSR coverage of the fibers A mosaic of nine micrographs (BEI) from the cotton jersey submitted to a 5 cm shot shows the distribution of GSR over the knitted yarn in Fig. 8. The fraying of the threads and the tearing of the cotton fibers on the margin of the bullet hole can be seen. A radial decrease in the GSR coverage can be seen and unburned propellant particles are indicated with arrows. As has been described by other authors [3], the macroscopic powder-soot distribution around the bullet hole can be roughly approximated by an exponential distribution. This is observed as well for the high Z particles that can be discriminated with the aid of the 9
backscattered electron detector. The main deviation of this decay is due to bigger blobs of high Z material that can be easily be recognized on the images. It is expected to find irregularities in the microscopic sampling on random radial profiles, considering that the bullet drag is turbulent. Different densities of the GSR deposits as a function of the distance from the bullet hole border from a contact shot on the denim sample are shown in Fig. 9.
The grid of images in Fig. 10a shows the distribution of GSR at different radial distances from the bullet holes, produced by gunshots fired from different ranges at polyester jersey. Fig. 10b displays the same images after the segmentation procedures.
The percentage of GSR coverage, obtained by a single operator with a consistent threshold choice from the images in Fig. 10b, as a function of radial distance from the entry holes is exemplified in the five plots of Fig. 11. To be able to fit an exponential curve on the data, some points had to be considered outliers and are indicated with lighter symbols. Frequently the first point (zero mm) had to be ignored, and a zero % coverage point at higher radial distance (60 mm) had to be added, to avoid a constant background fit. The parameters A and obtained from the exponential fits shown in (y = A exp-x) are summarized in Table 1, as well as the length GSR½, after which the coverage has fallen to half of the initial value.
When other operator choices for the threshold segmentation were used, the obtained decay length fell within ± 15% of the values of Table 1. This uncertainty was added to the standard deviation of the fit in the plot against the firing distance, shown in Fig. 12. The GSR½ length presented an almost linear relationship to the firing distance. It can be seen that the distance from the bullet hole border at which the GSR coverage had fallen to half is approximately a tenth of the shooting distance (muzzle to cloth distance). As a rule of thumb it can be said that if the coverage divides in half every 10 10
mm, the shot was fired from approximately 10 cm distance. A helpful feature from exponential decay is that the decay constant can be determined anywhere on the radial profile, i.e. it is not necessary to start at the center of the bullet hole (usually torn away), but can begin at any point on the radial profile. It can be seen that at 20 cm shots the uncertainty reaches a level that implies that the estimation should be restricted to short range shots. In the literature the decrease in GSR coverage has been estimated at the surroundings of bullet holes using the antimony content determined with neutron activation analysis [34]. There the shots were fired with a .22 rifle and the author's conclusion was that the concentration profiles obeyed an exponential decay function of the square of the radial distance from the edge. Our findings showed an exponential dependence of the nonsquared radial distance, but the differences in weapon, firing ranges, and analysis method might account for that. The same author mentioned in a later paper [35] that at firing distances smaller than 30 cm the exponential decay dependence of the squared radial distance was not followed.
4. Conclusions The modifications of the textiles that allowed to estimate firing ranges were of two kinds: i) a qualitative one, based on different alterations of the fibers on the edge of the holes; ii) a quantitative estimation, determining different characteristic lengths of the exponential decrease of GSR coating densities on radial profiles around the holes. Molten and fused polyester fibers in mixed textiles were observed only at contact shots, resulting in a peculiar horizontal tear around the bullet hole. In purely synthetic textiles, mainly with thin filaments, fusing of fibers was still observed at 2.5 cm shots. At this distance the fibers did not melt immediately, as in the contact shots, but the knitwear deformed strongly before melting, presenting the most visible damage. Above 5 cm muzzle-to-textile distance, no fusing of fibers was observed anymore, but scorched fiber ends could always be found. In the cotton jersey the microscopic damage was least discernible, and only at contact shots scorched fiber ends could be detected. In the noncontact shots only mechanical breakage was observed. The method to quantitatively estimate firing distance using the reciprocal of the exponential decay constant (GSR1/2), extracted from the GSR bright contrast on 11
backscattered electron images obtained on radial profiles, was independent of the type of fibers in the textile. For the handgun and ammunition type used in this study, the decay length of the coverage presented a linear relationship with the firing distance and allowed an estimation of the latter.
Acknowledgements The authors thank the Laboratório de Conformação Nanométrica and Laboratório de Microanálise, Physics Institute, UFRGS, Porto Alegre, RS, Brazil, for the use of the SEM facilities. The authors acknowledge funding from the Brazilian agencies FINEP, CAPES, CNPq, and FAPERGS.
References
[1] A. Zeichner, B. Glattstein, Recent Developments in the Methods of Estimating Shooting Distance, The Sci World J 2 (2002) 573-585. doi: 10.1100/tsw.2002.140 [2] H. Ditrich, Distribution of gunshot residues - the influence of weapon type, Forensic Sci. Int. 220 (2012) 85-90. doi:10.1016/j.forsciint.2012.01.034 [3] K. Sellier, Forensic Science Progress Volume 6 Shot range determination, Springer, Heidelberg, 1991. ISBN-13: 978-3-642-76723-4 [4] A.M. Cimino, J.E. Debney, Distance of Shot from Discharge Residues of Rifled Weapons, J. Forensic Sci. 8 (1968) 8-11. [5]. Collaborative Testing Services, Inc. Forensic Testing program. GSR-Distance Determination Test No. 14-530 Summary report; 2014. https://www.ctsforensics.com/assets/news/3430_Web.pdf [6] K.L. Kersh, J.M. Childers, D. Justice, G. Karim, Detection of Gunshot Residue on Dark-Colored Clothing Prior to Chemical Analysis; J Forensic Sci 59(3) (2014) 754-762. doi: 10.1111/1556-4029.12409 [7] C. Aksoy, T. Bora, N. Şenocak, F.Aydın, A new method to reduce false positives due to antimony in detection of gunshot residues, Forensic Sci. Int. 250 (2015) 8790. doi: 10.1016/j.forsciint.2015.03.006 [8] V.J.M. DiMaio, Gunshot Wounds Practical Aspects of Firearms, Ballistics, and Forensic Techniques, CRC Press, Boca Raton, 1999. ISBN 0-8493-8163-0 [9] J. Andrasko, S. Petterson, A simple method for collection of gunshot residues from clothing, J. Forensic Sci. 31 (1991) 321-330. 12
[10] Z. Brożek-Mucha, Distribution and properties of gunshot residue originating from a Luger 9 mm ammunition in the vicinity of the shooting gun, Forensic Sci. Int. 183 (2009) 33-44. doi:10.1016/j.forsciint.2008.10.010 [11] M.R. Rijnders, A. Stamouli, A. Bolck, Comparison of GSR Composition Occurring at Different Locations Around the Firing Position, J. Forensic Sci.55 (2010) 616-623. doi 10.1111/j.1556-4029.2009.01292.x [12] S. Charles, M. Lannoy, N. Geusens, Influence of the type of fabric on the collection efficiency of gunshot residues, Forensic Sci. Int. 228 (2013) 42-46. doi: 10.1016/j.forsciint.2013.02.022 [13] L. Niewoehner, H. W. Wenz, J. Andrasko, R. Beijer, L. Gunaratnam, ENFSI Proficiency Test Program on Identification of GSR by SEM/EDX*, J. Forensic Sci. 48(4) (2003) 1-8. Paper ID JFS2002396_484 www.astm.org. [14] American Society for Testing and Materials, ASTM E1588-10e1: Standard Guide for Gunshot Residue Analysis by Scanning Electron Microscopy/Energy Dispersive X-Ray Spectrometry, American Society for Testing and Materials, West Conshohocken, PA, 2010. [15] A. Zeichner, Recent developments in methods of chemical analysis in investigations of firearm-related events, Anal Bioanal Chem 376 (2003) 1178– 1191. doi: 10.1007/s00216-003-1994-y [16] W.R. Goynes, M.L. Rollins, A scanning electron microscope study of washer dryer abrasion in cotton fibers, Textile Res. J. 3 (1971) 226-231. [17] J.W.S. Hearle, B. Lomas, W.D. Cooke, Atlas of fiber fracture and damage to textiles, 2nd ed., CRC Woodhead Publishing, Boca Raton, 1998. ISBN 9781855733190 [18] J. Robertson, M. Grieve, Forensic examination of fibers, Taylor&Francis, London, 1999. ISBN 9780203484517 [19] W. Pelton, Use of SEM in textile forensic work, In: J.W.S. Hearle, B. Lomas, W.D. Cooke, Atlas of fiber fracture and damage to textiles, 2nd ed., CRC Woodhead Publishing, Boca Raton, 1998. ISBN 9781855733190. [20] F. Poole, M. Pailthorpe, Comparison of bullet and knife damage, In: J.W.S. Hearle, B. Lomas, W.D. Cooke, Atlas of fiber fracture and damage to textiles, 2nd ed., CRC Woodhead Publishing, Boca Raton, 1998. ISBN 9781855733190. [21] K. De Wael, L. Lepot, K. Lunstroot, F. Gason, 10 years of 1:1 taping in Belgium — A selection of murder cases involving fibre examination, Sci. Justice 56 (2016) 18-28. http://dx.doi.org/10.1016/j.scijus.2015.07.002 [22] H. Krüsemann, SEMs in forensic science, Problems Forensic Sci. 47 (2001) 110121. [23] Z. Brożek-Mucha, Variation of the chemical contents and morphology of gunshot residue in the surroundings of the shooting pistol as a potential contribution to a 13
shooting incidence reconstruction, Forensic Sci. Int. 210 (2011) 31-41. doi:10.1016/j.forsciint.2011.01.031 [24] M. Chohra, B. Beladel, L.B. Ahmed, M. Mouzai, D. Akretche, A. Zeghdaoui, A. Mansouri, M.E.A. Benamar, Study of gunshot residue by NAA and ESEM/EDX using several kinds of weapon and ammunition, J. Rad. Res. App. Sci., 2015, 8(3), 404.-410. doi 10.1016/j.jrras.2015.02.01 [25] D.J. Stokes, Principles and Practice of Variable Pressure/Environmental Scanning Electron Microscopy (VP-ESEM), John Wiley & Sons, Chichester, 2008. ISBN 978-0-470-06540-2 [26] D.E. Newbury, X-Ray Microanalysis in the Variable Pressure (Environmental) Scanning Electron Microscope, J. Res. Natl. Inst. Stand. Technol. 107 (2002) 567– 603. Available online: http://www.nist.gov/jres [27] G.D. Danilatos, Implication of the figure of merit in environmental SEM, Micron 44 (2013) 143-149. doi:10.1016/j.micron.2012.06.001 [28] L. Kouchaf, C. Mathieu, A. Kadoun, Electron microbeam changes under gaseous environment: CP-SEM case and microanalysis limits, in A.Mendez-Vilas, J. Diaz: Microscopy: Science, Technology, Applications and Education, Microscopy Series nº4 vol. 3. FORMATEX , Badajoz, 2010. ISBN-13: 978-84-614-6191-2 [29] A. Kadoun, R. Belkorissat, B. Khelifa, C. Mathieu, Comparative study of electron beam gas interactions in an SEM operating at pressures up to 300 Pa, Vacuum 69 (2003) 537-543. [30] R. Hinrichs, M.T. Brown, M.A.Z. Vasconcellos, M.V. Abrashev, W. Kalkreuth, Simple procedure for an estimation of the coal rank using micro-Raman spectroscopy, Int. J. Coal Geol. 136 (2014) 52-58. doi: 10.1016/j.coal.2014.10.013 [31] Jochem, G., Lehnert, R.J.; On the potential of Raman microscopy for the forensic analysis of coloured textile fibres; Sci & Justice 42 (2002) 215-221. [32] De Wael, K., Baes, C., Lepot, L., Gason, F.; On the frequency of occurrence of a peculiar polyester fi bre type found in blue denim textiles; Sci. Justice 51 (2011) 154-162. doi:10.1016/j.scijus.2011.02.001 [33] Rasband, W.; National Institute of Health, USA, http://imagej.nih.gov/ij [34] Bhattacharyya, C.N.; Dispersion of firing discharge residues using a Maxwellian model; Forensic Sci. Int. 42 (1989) 271-277. [35] Bhattacharyya, C.N.; Dispersion of firing discharge residues using a modified Maxwellian model; Forensic Sci. Int. 47 (1990) 31-37.
14
a)
b)
Fig. 1: a) BEI micrograph taken from the polyester jersey at 3 mm from the bullet hole of a soft contact shot. The horizontal arrow indicates a "charged" fiber, the vertical arrow an "out of focus" area (inset: threshold menu of ImageJ); b) binary image obtained with the "threshold" command. The yellow outline on both images indicates the area used to measure the white/black pixel count ratio.
Fig. 2: Sooting and tattooing of the textile samples after gunshots from different ranges (distances indicated on top row). Top to bottom: white cotton jersey, blue polyester/cotton denim, red polyester jersey. Scale bars are 5 cm long.
15
a)
b)
Fig. 3: Yielding of cotton fibers at a) contact shot, with thickened, scorched fiber ends; b) at a firing distance of 10 cm, showing tapered fiber endings. Scale bar on both images 50 m).
a)
b)
Fig. 4: a) BEI of the bullet hole (soft contact) in the denim, showing GSR coverage of the cotton threads, except for the shadow of the synthetic warp (scale bar 1 mm); b) molten and fused filaments of the synthetic warp (scale bar 100 m).
16
Fig. 5: Yielding of polyester fibers at a) 5 cm shot, with thickened, scorched fiber ends; b) at a firing distance of 20 cm, showing tapered fiber endings. Scale bar on both images 50 m).
a)
b)
c)
d)
Fig. 6: Damage on thin polyester fibers at a) contact shot, with fused fiber ends; b) firing distance of 2.5 cm, with thickened fibers and limited fusing; c) firing distance of 5 cm with thickened fibers due to scorching; d) at a firing distance of 20 cm with minor scorching and tapering of the fibers. Scale bar on all images 50 m.
17
Fig. 7: BEI of the fibers on the border of bullet holes: a) cotton (jersey); b) cotton (denim); c) polyester (denim); d) polyester (jersey). The rows show different muzzle to textile distances (HC: hard contact, SC: soft contact, other distances in mm; all images have a width of 570 m).
18
Fig. 8: BEI mosaic (uncoated sample observed in low vacuum of 15Pa), showing frayed fibers at the margin of the bullet hole from a 5 cm shot on cotton jersey. GSR on the fibers (white contrast) and unburned propellant particles (arrows) can be discriminated.
a)
b)
c)
d)
Fig. 9: Density of GSR coverage in a contact shot on denim at: a) 3 mm; b) 5 mm; c) 7 mm; and d) 9 mm from the border of the bullet hole (image widths 95 m).
19
a)
soft contact
2.5cm
5 cm
10 cm
20 cm
soft contact
2.5cm
5 cm
b)
10 cm
20 cm
Fig. 10: GSR distribution on polyester jersey as a function of radial distance from the bullet hole (in mm, indicated in first column), and as a function of the firing range (indicated in first row): a) BEI micrographs; b) binarized images. Width of all images is 128 m.
20
a)
b)
d)
c)
e)
Fig. 11: GSR coverage as a function of radial distance from the bullet holes for different firing ranges: a) soft contact; b) 2.5 cm; c) 5 cm; d) 10 cm; e) 20 cm. The bold data points were used for the exponential decay fit (grey lines). Error bars were estimated to be 10% of the pixel count statistic.
21
Fig. 12: GSR½ as a function of shooting distance, linear fit, 95% confidence and 95% prediciton levels. Inset is the equation of the error weighed linear fit (uncertainties in parentheses).
22
Table 1: Mean and uncertainty () of the fit parameters I0 and GSR½ of the exponential fit to the GSR coverages obtained on radial profiles on polyester jersey, fired at from different distances Firing distance (cm) Soft contact 2.5 5 10 20
I0 (%) Mean 22.4 14.1 16.5 9.7 6.1
GSR½= ln2/(mm)
2 1.9 1.9 0.8 2
Mean 0.6 2.8 4.1 8.8 14.7
0.04 0.9 1.1 1.9 10.6
23