Article pubs.acs.org/ac
Raman Spectroscopic Analysis of Gunshot Residue Offering Great Potential for Caliber Differentiation Justin Bueno, Vitali Sikirzhytski, and Igor K. Lednev* Department of Chemistry, University at Albany, SUNY, 1400 Washington Avenue, Albany, New York 12222, United States ABSTRACT: Near-infrared (NIR) Raman microspectroscopy combined with advanced statistics was used to differentiate gunshot residue (GSR) particles originating from different caliber ammunition. The firearm discharge process is analogous to a complex chemical reaction. The reagents of this process are represented by the chemical composition of the ammunition, firearm, and cartridge case. The specific firearm parameters determine the conditions of the reaction and thus the subsequent product, GSR. We found that Raman spectra collected from these products are characteristic for different caliber ammunition. GSR particles from 9 mm and 0.38 caliber ammunition, collected under identical discharge conditions, were used to demonstrate the capability of confocal Raman microspectroscopy for the discrimination and identification of GSR particles. The caliber differentiation algorithm is based on support vector machines (SVM) and partial least squares (PLS) discriminant analyses, validated by a leave-one-out cross-validation method. This study demonstrates for the first time that NIR Raman microspectroscopy has the potential for the reagentless differentiation of GSR based upon forensically relevant parameters, such as caliber size. When fully developed, this method should have a significant impact on the efficiency of crime scene investigations. ccording to the Centers for Disease Control, firearmrelated injuries were responsible for over 68% of homicides and were one of the three leading causes of injuryrelated deaths in the United States in 2008.1 Effective ballistic investigations act as a deterrent and reduce the number of firearm-related crimes in the United States.2 The development and innovation of new methods will meet the demand for ballistic investigations which draw more definite conclusions. Therefore, an increasing number of modern analytical techniques have been applied to this field of forensic ballistics and trace evidence analysis in recent years. 3,4 These investigations are designed to match a suspect to a crime but require the recovery of physical evidence in order to make the most confident conclusions. This physical evidence consists of projectiles (bullet), cartridge cases, and firearms. Gunshot residue (GSR) can also be recovered from several locations in the crime scene and may be utilized both as physical and chemical evidence. GSR is a combination of the burnt and unburnt byproducts that result from the combustion process of the original ammunition with the components of the firearm, projectile, and cartridge case. Conventional forensic GSR analysis methods are able to confirm that a shooting incident actually occurred, determining whether a suspect has recently discharged a weapon and estimating shooting distances.5−13 Several methods have suggested altering the state of traditional ammunition by introducing chemical or physical markers to ease GSR identification.14,15 Once identification is achieved, there is no widely accepted method for GSR characterization. Hence, the forensic science community is eager to develop an effective method for the nondestructive identification and discrimination of evidentiary GSR. A specific goal is to identify GSR and discriminate GSR samples based upon forensically relevant parameters, such as ammunition and/or firearm size.
A
© 2012 American Chemical Society
Here, we address this crucial forensic aspect through the use of Raman spectroscopy. Vibrational (IR and Raman) spectroscopy in combination with chemometrics applied to forensic analysis is a rapidly growing scientific area.16−19 Raman spectroscopy is well suited for forensic analysis due to its nondestructive nature, limited sample preparation, and range of applications. These applications include the identification of explosives,20 paint,21 textile dyes,22 drugs,23,24 and body fluids.25 Previous investigations of GSR analysis via Raman spectroscopy have identified four different inorganic components (including barium, lead, and iron compounds) of GSR. These conclusions were consistent with the most popular GSR identification method, scanning electron microscopy combined with energy dispersive spectroscopy (SEM/EDS).26 There are several analytical methods that have been used to test for GSR. Bulk and single-particle analyses are used to achieve identification of GSR, but there is no standardized procedure to test for GSR. Bulk methods are based on qualitative detection of specific elements, usually heavy metals. Combinations of lead (Pb), barium (Ba), and antimony (Sb) are considered characteristic to GSR.27 However, it has been shown that environmental contaminants can generate false positives.7 Unfortunately, bulk methods often make conclusions based on detecting these elements, which are not necessarily generated by GSR. This issue leads to a lack of specificity for methods such as flameless atomic absorption (FAA) and neutron activation analysis (NAA),28 which can misclassify environmental contaminants as GSR. Single-particle analysis combines chemical and morphological information to classify a Received: December 22, 2011 Accepted: March 13, 2012 Published: March 13, 2012 4334
dx.doi.org/10.1021/ac203429x | Anal. Chem. 2012, 84, 4334−4339
Analytical Chemistry
Article
suspected particle. The most widely accepted GSR analysis method is SEM/EDS. SEM/EDS is able to identify a sample as GSR by detecting the aforementioned elements in certain concentrations.29 Unfortunately, this test requires excessive time, sampling procedures, and instrumentation. Since this technique relies heavily on detecting lead, the removal of leadcontaining primers by manufacturers (for public health reasons30) has caused an increase in false-positive results.31 Elemental analysis is limited to only identify GSR; these techniques do not target the characterization of forensically relevant parameters. The method proposed here does not rely on detecting a limited number of specific heavy metals. It provides greater selectivity, compared to current techniques, by detecting spectroscopic features originating from both the organic and inorganic components of GSR. Our goal was to spectroscopically characterize and statistically explore variations in the Raman spectra of GSR between two particular firearmammunition combinations. Successful caliber differentiation was achieved by the use of support vector machines (SVM) discriminant analysis of Raman spectra acquired from tan colored GSR particles validated by leave-one-out crossvalidation. This technique offers the potential for a rapid, portable, reagentless, and selective alternative for GSR identification, while providing information related to the original shooting parameters. This information may provide a statistical and chemical link between the suspect and the crime scene.
Figure 1. (A) The selected Raman spectra collected from 9 mm (red line) and 0.38 (green line) caliber GSR of different caliber GSR particles. Shaded areas indicate the contributions of the inorganic (purple) and organic (green) substituent’s from the original ammunition. (B) Illustration of the heterogeneity of GSR at the particle level. The red and blue spectra were collected from different locations on the same GSR particle.
■
EXPERIMENTAL SECTION Gunshot Residue Collection. Low lint cloth wipes (23 cm × 23 cm, obtained from Scientific Instrument Services, Inc.) were used as the substrate for collecting the discharged GSR. The cloth substrate was placed in front of the barrel of the firearm as a target, at a distance of approximately 0.3 m. This shooting distance was selected to maximize the number of GSR particles obtained. The 0.38 caliber samples were produced by discharging Winchester brand “0.38 special” ammunition from a Smith and Wesson Model 10 Revolver. CCI brand “Blazer” 9 mm ammunition was discharged from a Bersa Thunder 9 firearm to generate the 9 mm samples. Two different discharge samples from each caliber were used for the preliminary analysis. In addition, four different discharge samples (two from each caliber) were used as an external validation; thus, eight total sample discharges were analyzed. All the collections were performed with the supervision and support of the New York State Police. Raman Spectroscopic Analysis. The GSR particles were removed from the cloth substrates and placed on an aluminum slide for analysis under a Renishaw inVia confocal Raman microscope, equipped with a research grade Leica microscope, a 50× objective, and WiRE 3.2 software. Several spectra were collected from different points on each GSR particle to account for their heterogeneity. Therefore, each GSR particle was represented by a multispectral data set. Each single spectrum was an average of 5 scans for 35 s over a range of 300−1800 cm−1. The 785 nm excitation laser originated from an Nd:YAG (coherent) laser, with a power of approximately 10% of the maximum. The spectrometer was calibrated before the Raman spectra collections using a silicon reference standard (520 cm−1). All the measurements were performed under identical laboratory conditions, and representative raw spectra are illustrated in Figure 1 (upper left inset). The cosmic ray
contribution was removed from all the Raman spectra using the GRAMS/AI software package. The spectra were imported into MATLAB 7.9.0 for preprocessing and statistical analysis.
■
RESULTS AND DISCUSSION Raw Raman Spectroscopic Data. GSR particles were collected from four different firearm discharge samples: two 9 mm and two 0.38 caliber discharges. A visual inspection of the cloth substrate revealed more than 100 GSR particles observable by the naked eye. The majority of the particles selected for the analysis were approximately 1 mm × 1 mm in size. Additionally, bullet wipe deposits (darkening of the cloth due to metal and carbon accumulations) were observed around the bullet entrance hole. GSR particles were found to vary in color (tan, brown, and black), both between the discharge samples and within a particular sample. The most informative Raman spectra were obtained from brown and tan colored particles; therefore, the data reported here comes from these colored particles. Spectra from black GSR particles were determined to be dominated by fluorescence contributions and thus were omitted from data analysis. Raman analysis was performed on individual GSR particles by removing the particles from the cloth substrate and placing them on an aluminum slide under a 50× objective for measurement. Approximately four Raman spectra were collected from arbitrary areas of each particle, and a total of 78 particles were investigated. Figure 1 (upper left inset) illustrates the raw Raman spectra collected from two different GSR samples (0.38 and 9 mm calibers). The fluorescent background was determined to be highly variable, with noticeably lower fluorescence contributions for spectra collected from the 0.38 caliber samples. The fluorescent background contribution was 4335
dx.doi.org/10.1021/ac203429x | Anal. Chem. 2012, 84, 4334−4339
Analytical Chemistry
Article
removed using an adaptive, iteratively reweighted, penalized, least-squares baseline correction algorithm.32 After baseline correction, the majority of the Raman spectra (Figure 1A) had similar profiles. Most spectra, even those collected from GSR particles originating from different calibers, did not show variations to the naked eye. As a result, statistical analysis was performed for discriminate analysis (see Chemometric Analysis). An exception to this pattern is illustrated in Figure 1B, in which spectra collected from the same particle demonstrate different profiles. These variations from within a GSR particle illustrate the depth of heterogeneity displayed by the GSR samples. Our preliminary research attempted to account for this variability using a large data set. Raman Spectra and Firearm Chemistry. Modern, smokeless ammunition propellants often consist of nitrate ester explosives (containing C−O−NO2 groups) and a vast array of deterrents (centralites) and plasticizers (phthalates). Single-based propellants use nitrocellulose (NC) as the explosive, whereas double-based propellants contain NC and nitroglycerine (NG). These explosives are rapidly vaporized, producing the hot gases needed to expel the projectile from the barrel. Experimental Raman spectra of GSR revealed several bands that could be attributed to compounds originating from the ammunition propellant. The bands at approximately 850, 1287, 1346, 1369, and 1655 cm−1 could be attributed to these types of explosives (Table 1). NO2 groups from these
bands found in our spectra. Ethyl centralite, a deterrent found in double-based propellants, was determined to be present in the GSR due to the peak at ∼1005 cm−1.36 Additionally, we assigned peaks to two heavy metal salts: PbSO4 and BaCO3. Both of these metals are commonly used in ammunition primers. Lead complexes often act as initiators, whereas barium complexes (barium nitrate) are oxidizers.27 The initiator is sparked by the firing pin hitting the primer cap, while oxidizers provide oxygen to fuel the combustion process.37 It is our hypothesis that barium nitrate decomposes during this process and thus is free to form complexes with carbonate. Table 1 illustrates the Raman peaks we have assigned to substances characteristic to ammunition, based on literature data. However, several peaks were left unassigned because information was not readily available or was conflicting. Therefore, we consider that an unambiguous assignment cannot be made at this time. It is known that the previously mentioned chemicals are common to ammunition. Therefore, detecting these chemicals in GSR is evidence that the chemical composition of the ammunition has a direct impact on the composition of the resulting GSR particles. Therefore, rapid, nondestructive Raman analysis can potentially be used to differentiate the crime scene GSR produced by chemically different ammunition. To our knowledge, this research is the first to target both the organic and inorganic components of GSR as an identification and chemical characterization technique. Chemometric Analysis. Visual inspection of the samples revealed that the majority of the GSR particles could be tentatively separated into one of two groups, based on the color of their outer surface (brown or tan). Therefore, the normalized spectra were grouped into two data sets. The first data set contained all of the Raman spectra collected from the tan-colored GSR particles, whereas the second data set consisted of spectra collected from the brown particles. Both data sets were treated with statistical analysis. Analyses included a principle component analysis (PCA) method utilizing a knearest neighbor algorithm (k-NN), partial least-squares discriminant analysis (PLSDA), and support vector machines discriminant analysis (SVMDA). Our laboratory has already successfully implemented a similar method for the statistical analysis of intrinsically heterogeneous samples for confirmatory identification of body fluid traces.18,25,50,51 Preliminary analysis demonstrated that the data collected from the tan-colored GSR particles (88 Raman spectra) formed the most informative subset of the complete data set. The most promising results were shown by the SVM method, which achieved the most accurate caliber-weapon pair identification. The variations in Raman spectra between the two calibers were analyzed with PCA (Figure 2A). The two groups are highly scattered and overlapped, producing no distinct separation between Raman spectra collected from GSR particles originating from the different caliber samples (9 mm and 0.38). However, the score plot has regions populated by only spectra from the 0.38 caliber ammunition. Similar patterns were obtained using the fourth, fifth, and subsequent principle components (PCs). A generalized least-squares (GLS) weighting algorithm was used to reduce the scattering within the groups. As a result, we observed a substantial increase in the level of separation between the calibers (Figure 2.B). The PCA scores plot is less scattered; as a result, only one 9 mm caliber Raman spectrum is in close proximity to the 0.38 caliber region.
Table 1. Raman Peak Assignment compound, location in ammunition
Raman shift (cm−1)
BaCO3, primer nitrate ester, propellant PbSO4, primer centralite, propellant PbSO4, primer CO32‑ complex, primer
692 850 983 1005 1065 1080
PbSO4, primer 2,4-dinitrotoluene, propellant nitrate ester, propellant
1155 1207 1287
nitrate ester, propellant
1346
nitrate ester, propellant
1369
2,4-dinitrotoluene, propellant BaCO3, primer
1427 1455
dinitrotoluene, propellant
1592
nitrate ester, propellant
1655
vibrational mode CO3 bending in plane26,38−40 NO2 scissoring12,20,41 ref 26 aromatic ring breathing36 ref 26 CO3 symmetric stretching38,42 refs 26 and 43 C−H (ring breathing)41 NO2 symmetric stretching12,20,44,45 NO2 symmetric stretching41,46,47 CNO2 symmetric stretch12,41,44,45 CH3 umbrella41 CO3 asymmetric stretching26,38,39 NO2 asymmetric stretching41,48,49 NO2 asymmetric stretching12,41,44,45
explosives decompose during the ignition process. Stabilizers and deterrents, such as diphenylamine (DPA) and centralites, respectively, are added to propellant mixtures and act as nitrogen scavengers33 after the ammunition has been fired. To increase shelf life of the ammunition, nitrogen scavengers bind free nitrogen oxides, often nitrogen monoxide (NO) and dioxide (NO2).34 Experiments have shown that DPA is relatively stable during the ignition process,35 allowing for nitration and/or nitrosation of the stabilizer by decomposing the components of the ammunition33 and producing the NO2 4336
dx.doi.org/10.1021/ac203429x | Anal. Chem. 2012, 84, 4334−4339
Analytical Chemistry
Article
Figure 2. (A) PCA of Raman spectra from 9 mm (green stars) and 0.38 (red triangles) caliber GSR particles. The black circle outlines the region occupied by only red triangles. (B) Result of applying the (GLS) weighting algorithm to the data from (A). Each score plot shows the variation among the first three principle components.
Figure 3. Classification of Raman spectra collected from 9 mm (green stars) and 0.38 (red triangles) caliber GSR particles. Classification using (A) k-NN algorithm, (B) SVMDA, and (C) SVMDA and validated using (D) leave-one-out cross-validation of SVMDA results.
3B,C shows the results of PLSDA and SVMDA. The plots are an illustration of the predicted class (caliber) for each experimental spectrum. All spectra were assigned to the correct class, green stars as 9 mm and red triangles as 0.38. Table 2 combines the classification rates from the k-NN, PLSDA, and SVMDA classification models. Each confusion matrix illustrates the rate (proportions) of positive cases that were correctly identified (TP, true positives), negative cases that were incorrectly classified as positive (FP, false positives), negative cases that were classified correctly (TN, true negatives), and positive cases that were incorrectly classified as negative (FN, false negatives). It is evident from the presented data that there are no false positive or false negative results obtained by PLSDA and SVMDA. Leave-one-out cross-validation was used to test the experimental classification results. The results illustrate that only one 0.38 caliber Raman spectrum was misclassified as a spectrum of 9 mm (Figure 3.D). Therefore, our studies demonstrate that the origin of GSR particles can be
PCA scores calculated with the help of GLS were subjected to further classification analysis. Classification analysis was performed using k-NN, PLSDA, and SVMDA methods. The k-NN algorithm assigns an object to the class most common to its nearest neighbors (small positive integer, k), it is considered one of the simplest machine learning algorithms. Figure 3A shows the results of classification using the k-NN algorithm. Only one spectrum was misclassified. A more accurate classification was obtained by PLSDA and SVMDA methods. These two methods are useful for dealing with data sets with nonlinear relationships between variables. SVMDA is an example of supervised classification, in which patterns (recognized in a training set) are used to help differentiate several classes. SVM-based algorithms are capable of dealing with calibration issues, which fails Hadamard’s criteria for the uniqueness and stability of mathematical solutions, in addition to producing robust models for nonlinear spectral variations.52−54 Figure 4337
dx.doi.org/10.1021/ac203429x | Anal. Chem. 2012, 84, 4334−4339
Analytical Chemistry
Article
ammunition. We hypothesize that a Raman spectroscopic library could potentially be developed to link GSR to a specific ammunition and/or firearm. In a simpler version, the method proposed here could be used to determine whether GSR recovered in a crime scene could be linked to a specific firearm without the use of an extended spectroscopic library. In the latter case, a law enforcement agency will possess a suspect’s weapon and ammunition and will need to determine whether they were discharged at a crime scene. A firearm examiner can make this determination by creating a GSR sample set by discharging several rounds of the ammunition in a controlled environment. These test samples can be analyzed using the Raman method proposed here and compared with the crime scene GSR. The conclusion that a specific firearm was not discharged at the crime scene could be reached without an extensive preliminary study of the numerous available firearms. In contrast, the extensive preliminary study would be necessary to prove that the firearm was discharged and to indicate the confidence interval for such conclusions. Nondestructive Raman spectroscopic analysis of GSR to determine and/or differentiate forensically relevant parameters (such as caliber) can be considered a novel advance in crime scene forensics. Our method has the potential to greatly reduce the time, expense, and bias of GSR identification.
Table 2. Classification Rates from k-NN, PLSDA, and SVMDA Classification Modelsa TP class 1: 0.38 class 2: 9 mm class 1: 0.38 class 2: 9 mm class 1: 0.38 class 2: 9 mm class 1: 0.38 class 2: 9 mm
FP
k-NN 0.98592 0.00000 1.00000 0.01408 PLSDA 1.00000 0.00000 1.00000 0.00000 SVMDA 1.00000 0.00000 1.00000 0.00000 cross-validated SVMD 0.98592 0.00000 1.00000 0.01408
TN
FN
1.00000 0.98592
0.01408 0.00000
1.00000 1.00000
0.00000 0.00000
1.00000 1.00000
0.00000 0.00000
1.00000 0.98592
0.01408 0.00000
a
TP, true positives; FP, false positives; TN, true negatives; FN, false negatives.
evaluated with high confidence using advanced statistical analysis of their Near-infrared (NIR) Raman spectra. Method Validation. To test the quality of the calculated classification models, two additional discharge samples (Figure 4, red triangles and blue squares for 9 mm and green circles and
■
CONCLUSIONS Particles containing Pb, Ba, and Sb are considered to be characteristic to GSR, and detecting these elements is required for current GSR identification techniques. A new approach reported here is based on a spectroscopic analysis of both organic and inorganic compounds. The preliminary Raman band assignment for the experimental spectra includes vibrational modes from nitro (NO2), carbonate (CO3), and aromatic and inorganic salt groups. These results are consistent with the components found in both ammunition propellants and primers. The majority of the GSR particles from experimental samples can be tentatively grouped as brown or tan-colored. Statistical analysis demonstrated that Raman spectra of tan-colored GSR particles (88 Raman spectra) subjected to classification analysis can provide a highly accurate ammunition caliber−firearm pair identification. We believe that the accuracy of identification can be increased using automatic algorithms targeting particle color identification. It is important to emphasize that, although the results of this study are very promising, they should be considered as preliminary due to a use of a limited number of ammunition−firearm combinations. We demonstrated the capability of our method to differentiate GSR produced by the discharge of two different ammunition−firearm combinations, with high statistical confidence. At this time, it is unknown what specific characteristic(s) of these combinations result in the differentiation. Nevertheless, the impact of caliber on the chemical nature of GSR can be rationalized. Both the combustion process and chemical composition of specific ammunition is dependent upon caliber. Different caliber ammunition will undergo physically different discharge processes when being expelled from the firearm. Additionally, the chemical composition of the propellant will vary based upon the size of the ammunition. This variation involves not only the composition of specific components in the ammunition but also their quantitative amounts. Therefore, we believe that the caliber should have an important impact on the chemical nature of GSR. Our future plans include
Figure 4. External validation of the developed PLS-DA-based method by Raman spectroscopic data acquired from additional discharge samples. Y-axis corresponds to the predicted “class values” (9 mm) of Raman spectra. The horizontal red dashed line represents the threshold of GSR classification, estimated using Bayes’ theorem. Each colored symbol corresponds to a single experimental spectrum.
blue triangles for 0.38) were added for each caliber (external validation). All new Raman spectra were correctly assigned to the corresponding caliber. Figure 4 demonstrates the results of the external validation using PLS-DA analysis. The Y-axis in Figure 4 corresponds to the predicted “class values” of a particular Raman spectrum or the propensity of that spectrum to be identified as collected from a GSR particle originating from a 9 mm caliber ammunition. A value of 1 indicates that a spectrum belongs to the class (9 mm), while a 0 indicates that the spectrum is not classified as 9 mm. Error bars show the uncertainty in the reported values. The horizontal red dashed line represents the prediction threshold, estimated using Bayes’ theorem and the available data in order to minimize the total error. The threshold is selected at the Y-value at which the number of false positives and false negatives are minimal (PLS toolbox, eigenvector Research, Inc., Wenatchee). It is evident from Figure 4 that all 9 mm spectra were correctly classified. Potential Impact on Practical Forensics. The most obvious application of the proposed method is in determining the origin of specific evidence, i.e., as a tool to link recovered crime scene GSR to a specific type of a firearm and/or 4338
dx.doi.org/10.1021/ac203429x | Anal. Chem. 2012, 84, 4334−4339
Analytical Chemistry
Article
(24) Ali, E. M. A.; Edwards, H. G. M.; Hargreaves, M. D.; Scowen, I. J. Anal. Chim. Acta 2008, 615, 63−72. (25) Virkler, K.; Lednev, I. K. Forensic Sci. Int. 2008, 181, e1−e5. (26) Stich, S.; Bard, D.; Gros, L.; Wenz, H. W.; Yarwood, J.; Williams, K. J. Raman Spectrosc. 1998, 29, 787−790. (27) Schwoeble, A. J.; Exline, D. L. Current Methods in Forensic Gunshot Residue Analysis; CRC Press: New York, 2000. (28) Wallace, J. S.; McQuillan, J. J. Forensic Sci. Soc. 1984, 24, 495− 508. (29) Nesbitt, R. S.; Wessel, J. E.; Jones, P. F. J. Forensic Sci. 1976, 21, 595−610. (30) Steffen, S.; Otto, M.; Niewoehner, L.; Barth, M.; Brozek-Mucha, Z.; Blegstraaten, J.; Horvath, R. Spectrochim. Acta, Part B 2007, 62, 1028−1036. (31) Burleson, G. L.; Gonzalez, B.; Simons, K.; Yu, J. C. C. J. Chromatogr., A 2009, 1216, 4679−4683. (32) Zhang, Z. M.; Chen, S.; Liang, Y. Z. Analyst 2010, 135, 1138− 1146. (33) Espinoza, E. O.; Thornton, J. I. Anal. Chim. Acta 1994, 288, 57− 69. (34) Laza, D.; Nys, B.; De Kinder, J.; Mesmaeker, A. K. D.; Moucheron, C. J. Forensic Sci. 2007, 52, 842−850. (35) Dahl, D. B.; Lott, P. F. Microchem. J. 1987, 35, 347−359. (36) Louden, J. D.; Kelly, J.; Phillipson, J. J. Appl. Polym. Sci. 1989, 37, 3237−3250. (37) Wallace, J. S. Chemical Analysis of Firearms, Ammunition, and Gunshot Residue; CRC Press: Boca Raton, 2008. (38) Burgio, L.; Clark, R. J. H. Spectrochim. Acta, Part A 2001, 57, 1491−1521. (39) Pasierb, P.; Komornicki, S.; Rokita, M.; Rekas, M. J. Mol. Struct. 2001, 596, 151−156. (40) Miller, F. A.; Carlson, G. L.; Bentley, F. F.; Jones, W. H. Spectrochim. Acta 1960, 16, 135−235. (41) Ramos, C. M.; Alzate, L. F.; Hernández, N. M.; Hernández, S. P.; Mina, N. J. Mol. Struct. Theochem. 2006, 769, 69−76. (42) Urmos, J.; Sharma, S. K.; Mackenzie, F. T. Am. Mineral. 1991, 76, 641−646. (43) Bullock, K. R.; Trischan, G. M.; Burrow, R. G. J. Electrochem. Soc. 1983, 130, 1283−1289. (44) Botti, S.; Cantarini, L.; Palucci, A. J. Raman Spectrosc. 2010, 41, 866−869. (45) Roos, B. D.; Brill, T. B. Combust. Flame 2002, 128, 181−190. (46) Sylvia, J. M.; Janni, J. A.; Klein, J. D.; Spencer, K. M. Anal. Chem. 2000, 72, 5834−5840. (47) Lopatin, B. V. J. Appl. Spectrosc. 1985, 43, 1137−1139. (48) López-López, M.; Ferrando, J. L.; García-Ruiz, C. Anal. Chim. Acta 2012, 717, 92−99. (49) Moros, J.; Lorenzo, J. A.; Lucena, P.; Miguel Tobaria, L.; Laserna, J. J. Anal. Chem. 2010, 82, 1389−1400. (50) Virkler, K.; Lednev, I. K. Forensic Sci. Int. 2009, 193, 56−62. (51) Virkler, K.; Lednev, I. K. Forensic Sci. Int. 2009, 188, 1−17. (52) Vapnik, V. Statistical Learning Theory; John Wiley & Sons: New York, 1998. (53) Wang, L., Ed. Support vector machines: theory and applications; Springer-Verlag: New York, 2008. (54) Wang, B.; Kong, X.; You, X. In Advances in Digital Forensics V; Peterson, G., Shenoi, S., Eds.; Springer-Verlag: New York, 2009.
examining parameters other than caliber size, to determine their effects (if any) on the chemical nature of GSR. These factors include the chemical composition of the propellant, primer, projectile and cartridge case, the age and location of the collected GSR sample, the type of firearm and firing mechanism used in the incident, and the condition and age of the discharging firearm.
■
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. Phone: (518) 591 8863. Fax: (518) 442-3462. Notes
The authors declare no competing financial interest.
■
ACKNOWLEDGMENTS We are grateful to Dr. Barry Duceman, Director of Biological Science, New York State Police Forensic Investigation Center, and John Hicks, former Director of North East Regional Forensic Institute (NERFI), for continuous support and Lieutenant Heller and Sergeant D’Allaird for providing the GSR samples.
■
REFERENCES
(1) Minino, A. M.; Murphy, S. L.; Xu, J.; Kochanek, K. D. Natl. Vital Stat. Rep. 2011, 59, 132−133. (2) Casey, W. M.; Gagliardi, P. Crime Justice Int. 2007, 23, 27−29. (3) Brettell, T. A.; Butler, J. M.; Almirall, J. R. Anal. Chem. 2011, 83, 4539−4556. (4) Dalby, O.; Butler, D.; Birkett, J. W. J. Forensic Sci. 2010, 55, 924− 943. (5) Silva, M. J.; Cortez, J.; Pasquini, C.; Honorato, R. S.; Paim, A. P. S.; Pimentel, M. F. J. Braz. Chem. Soc. 2009, 20, 1887−1894. (6) Romolo, F. S.; Margot, P. Forensic Sci. Int. 2001, 119, 195−211. (7) Garofano, L.; Capra, M.; Ferrari, F.; Bizzaro, G. P.; Di Tullio, D.; Dell’Olio, M.; Ghitti, A. Forensic Sci. Int. 1999, 103, 1−21. (8) Dockery, C. R.; Goode, S. R. Appl. Opt. 2003, 42, 6153−6158. (9) Houck, M. M.; Siegel, J. A. Fundamentals of Forensic Science; Academic Press: London, 2006. (10) Santos, A.; Magalhaes, T.; Vieira, D. N.; Almeida, A. A.; Sousa, A. V. Am. J. Forensic Med. Pathol. 2007, 28, 24−30. (11) Capannesi, G.; Ciavola, C.; Sedda, A. F. Forensic Sci. Int. 1993, 61, 75−78. (12) Sharma, S. P.; Lahiri, S. C. Sci. Justice 2009, 49, 197−204. (13) Muller, D.; Levy, A.; Vinokurov, A.; Ravreby, M.; Shelef, R.; Wolf, E.; Eldar, B.; Glattstein, B. J. Forensic Sci. 2007, 52, 75−78. (14) Weber, I. T. v.; de Melo, A. J. G.; Lucena, M. A. d. M.; Rodrigues, M. O.; Alves Junior, S. Anal. Chem. 2011, 83, 4720−4723. (15) Krivosta, G. G. AFTE J. 2006, 38, 41−47. (16) Bartick, E. G. Handbook of Vibrational Spectroscopy; John Wiley & Sons Ltd: Chichester, 2002. (17) McLaughlin, G.; Lednev, I. K. Anal. Bioanal. Chem. 2011, 401, 2511−2518. (18) Sikirzhytski, V.; Sikirzhytskaya, A.; Lednev, I. K. Appl. Spectrosc. 2011, 65, 1223−1232. (19) Kocak, A. Forensic Chemistry Handbook; John Wiley & Sons: Hoboken, NJ, 2012. (20) Ali, E. M. A.; Edwards, H. G. M.; Scowen, I. J. Talanta 2009, 78, 1201−1203. (21) Suzuki, E. M.; Carrabba, M. J. Forensic Sci. 2001, 46, 1053− 1069. (22) Abbott, L. C.; Batchelor, S. N.; Smith, J. R. L.; Moore, J. N. Forensic Sci. Int. 2010, 202, 54−63. (23) Hodges, C. M.; Akhavan, J. Spectrochim. Acta, Part A 1990, 46, 303−307.
■
NOTE ADDED AFTER ASAP PUBLICATION This paper was published on the Web on March 28, 2012. A correction was made to the Acknowledgment, and the corrected version was reposted on April 26, 2012.
4339
dx.doi.org/10.1021/ac203429x | Anal. Chem. 2012, 84, 4334−4339
