The author(s) shown below used Federal funds provided by the U.S. Department of Justice and prepared the following final report: Document Title: Forensic Glass Analysis by LA-ICP-MS: Assessing the Feasibility of Correlating Windshield Composition and Supplier Author: Abbegayle J. Dodds, Edward M. “Chip” Pollock, and Donald P. Land Document No.: 232134 Date Received: October 2010 Award Number: 2004-IJ-CX-K007 This report has not been published by the U.S. Department of Justice. To provide better customer service, NCJRS has made this Federally- funded grant final report available electronically in addition to traditional paper copies. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. Forensic glass analysis by LA-ICP-MS: Assessing the feasibility of correlating windshield composition and supplier Award No: 2004-IJ-CX-K007 FINAL TECHNICAL REPORT Abbegayle J. Dodds1-3, Edward M. “Chip” Pollock3, and Donald P. Land1,2 1Department of Chemistry, and 2Graduate Group in Forensic Science; University of California, Davis 3Sacramento County District Attorney’s Laboratory of Forensic Services Contact: Robert A. Jarzen, Director Sacramento County DA Laboratory of Forensic Services 4800 Broadway, Suite 200 Sacramento, CA 95820 (916) 874-9240 tel (916) 874-9620 fax [email protected] This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 2 Literature Review Introduction Glass fragments represent a valuable class of trace evidence. Like other traces materials, they are easily transferred from source to suspect, and are easily unnoticed by the suspect bearing them; further, glass fragments are particularly durable. Most glass products readily shatter when broken, distributing glass fragments to objects and persons in their path. Because there is a limited radius of distribution, glass transfers generally represent primary transfers resulting from contact or close proximity with the broken glass product (1). While secondary and environmental transfers do occur, they are rare (2- 4); this suggests that most individuals bearing glass fragments were near the glass product(s) distributed on their person when the breaking event occurred. The persistence of these transfers is largely dependent on the retention of the material to which the transfer is deposited (1), whether the transfer was passive or forcible, and the ability of glass to withstand environmental effects. Common items of clothing (cotton and woolen materials) show a high retention for glass, passively or forcibly transferred. Certain materials, such as wood, soft polymers and metals, retain glass transfers only if forcible contact is made between the material and the glass source. Because glass fragments are often minute and transparent, it is usually difficult for a suspect to see the evidence and remove it. Glass fragments persist on a suspect’s clothing or in soft materials for extended periods of time since glass is resistant to environmental degradation. Classical methods of forensic glass examination are based primarily on variation within the physical properties of glass. Color, density, surface characteristics and optical properties have been relied upon for the comparison of unknown glass fragments with Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 3 control glasses (5-7). Surface characteristics and optical properties deserve attention in particular. Similarities in peculiar surface contamination or patterns of erosion between reference and questioned items are highly associative; like fracture edge matching, however, comparing original surfaces requires recovery of abnormally large questioned fragments. The most common technique for comparing reference and questioned items is by refractive index (RI) comparison (5-7). RI measurement is excellent for distinguishing glasses by type and limited sample is required for multiple measurements. Historically speaking, RI measurement had limited application for classifying glass fragments because RI varied greatly among and within the traditional classes of glass (e.g., tableware, architectural glass, automotive glass, etc). Modern improvements in glass manufacture have decreased the RI variation within a particular class of glass (8). Modern glass has been observed to have a fairly consistent RI that corresponds to the type of glass in question; this makes RI an excellent tool for classifying glass but limits the utility of RI measurement for forensic individualization (7, 9-11). Some have proposed the measurement of RI at multiple wavelengths (called “dispersion analysis”) to enhance individualization by RI. It is rare that dispersion analysis enhances the discriminating power of RI (6). The first reports of chemical analysis for the forensic discrimination of glass were published in the early 1970s. Initial analyses were made with the intention of classifying glass by type, using a wide variety of instrumental techniques including: neutron activation analysis (NAA), direct current arc source atomic emission spectrometry (AES), atomic absorption spectrometry (AAS), spark-source mass spectrometry, and x-ray Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 4 fluorescence (10-18). The inorganic constituents targeted by these methods were in the part-per-million (or µg g-1) to part-per-hundred (or dg g-1) concentration levels. Over twenty elements were shown to have application in differentiating between sheet, container and tableware classes of glass (10). Due to the cumbersome operation of NAA, the semi-quantitative nature of spark source mass spectrometry and x-ray fluorescence, and the limitation of single element quantitation inherent to AA, forensic researchers incorporated inductively coupled plasma source AES in the late 1970s and 1980s. Using this technology, Catterick and Hickman showed the potential to discriminate glasses by type having sample sizes of 500 µg or less (9). They also reported that no correlation existed between chemical composition and RI, indicating that chemical data can be used in conjunction with RI for increased distinction of glasses by type. The increased discrimination of elemental data used in conjunction with RI data was also suggested by Koons, Peters and Rebbert (17). In a separate report, Koons and Buscaglia (19) estimated the random occurrence of two fragments being indistinguishable in RI and elemental composition to be 10-13 – 10-15. This would indicate that chemical data in tandem with RI measurement could facilitate individualization of glass with a high degree of certainty. Discrimination among glasses using chemical data alone was first suggested by the results of Catterick and Hickman; they showed the discrimination potential of certain elements in discriminating glass samples that fell into the same general class (9). Koons, Fiedler and Rawalt (20) demonstrated the differentiation of sheet glass produced by separate manufacturing plants using six elements. Zurhaar and Mullings (21) argue that the quantitation of 15 – 25 elements provides a unique elemental profile for a particular Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 5 glass sample. They further report that 85 – 95% of window glass samples manufactured in the US and Australia are easily distinguished when a greater suite of elements are analyzed. While the uniqueness of a particular elemental profile can be argued (22), there has been a marked increase in the discrimination of glasses within a class by increasing the number of elements quantified and targeting trace elements (≤ µg g-1 in concentration, 23). Until the advent of ICP-source mass spectrometry (ICP-MS), the ability to perform simultaneous, multielement quantitation was not available for the forensic analysis of glass. While ICP-AES is capable of performing multielement quantitation, this technique does not offer simultaneous, multielement data collection. Nor is ICP-AES able to detect low-level elements, especially following acid digestion and sample dilution. Zurhaar and Mullings (21) were the first to apply ICP-MS to the glass matrix for forensic analysis. Parouchais, et al (24) used the principles of analysis set forth by Zurhaar and Mullings to propose improved sample preparation protocol for glass analysis by ICP-MS. In the work following, Suzuki, et al (25) were able to show the superior discriminatory capabilities of elemental data collected by ICP-MS for bottle glass; Montero, et al (26) similarly showed a high level of discrimination available for vehicle float glass using ICP-MS. Forensic Glass Analysis by ICP-MS The ICP-MS is a highly sensitive instrument capable of performing rapid, simultaneous, multielement analysis. This technique offers exceptionally low detection limits (< pg mL-1) compared to other techniques for elemental analysis, and can be used Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 6 to detect over 70 isotopes. The ICP-MS has three main components: (1) the sample introduction mechanism, which is variable to accommodate gaseous, liquid, and solid samples, (2) the plasma and MS interface region, and (3) the mass analyzer and detector. Following sample introduction, the sample is injected into the plasma and undergoes desolvation and atomization. The resulting atoms are then ionized in the high-energy environment of the plasma; ions are transported through the MS interface region due to a sequential decrease in pressure. They are then mass-filtered and detected by a quadrupole or time-of-flight (TOF) mass spectrometer. Most commercial instruments are equipped with a quadrupole MS. One of the many benefits of ICP-MS is the number of sample types that may be accommodated. The ICP-MS has been adapted for gaseous, liquid and solid samples, though the original design was intended for liquid sample introduction (called solution nebulization, SN). Early applications of ICP-MS to glass analysis involved lengthy dissolution protocols so that glass could be introduced using SN (21, 23-27). Advances in solid sampling for ICP-MS have been realized in the past decade, and now several reports regarding solid sampling for ICP-MS exist. The majority of forensic applications involve the use of laser ablation (LA) sample introduction (8, 28-30). SN is the most common sample introduction technique for ICP-MS (31-35) and has wide application in forensic science (21, 36-40). It was the first method of introduction for the forensic analysis of glass (21, 23, 24). In fact, it is the only sampling technique for which an American Standards for Materials and Testing (ASTM) method exists (27). A multitude of sample types are appropriate for SN; matrix-matched calibration and quality control standards are easily obtained. One practical benefit of SN Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 7 is the decreased cost over other introduction techniques. Further, current instrumental configurations facilitate highly automated and rapid analysis. Automated instrument optimization and analysis are available, enabling high sample throughput with little analyst intervention. By comparison to solid sampling, less instrument maintenance is required using liquid sample introduction. This is because liquid samples leave fewer deposits on the sampler and skimmer cones and ion lens. While there are a few exceptions, liquid samples tend be “cleaner” overall (31). The main drawback of SN introduction is that it is difficult to adapt to solid sample types (31-35). This is especially true of the glass matrix. For the forensic analysis of glass, a costly, time-intensive and potentially hazardous digestion using hydrofluoric acid (HF) is required. This digestion is open-vessel and is followed by two days of sample preparation (27). This process provides many opportunities for contamination and dilution errors; worse, it is a destructive technique. The existing digestion protocol facilitates only a narrow range of sample masses because the final dilution volumes are relatively small. The minimal suggested sample size, 500 µg, is atypical of glass fragments received as evidence. Glass fragments of only several micrograms in mass are more frequently recovered than those of several hundred micrograms. Having trace elements at 1 – 100 parts-per-million, routine casework fragments push the lower sample size limits for SN-ICP-MS using the established forensic methodology. Finally, the resulting sample volumes might prevent the analyst from performing replicate analyses with certain devices for SN sample introduction. Nonetheless, SN introduction remains the most frequently used sample introduction technique for forensic glass analysis by ICP-MS. Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 8 Only a handful of forensic applications exist for LA-ICP-MS, including methods for glass, paint, and bulk metal materials (8, 28-30). Glass analysis stands out as the prominent and best-developed forensic application of both ICP-MS and LA-ICP-MS. Since the application of the LA sampling technique to ICP-MS, forensic analysts have been able to push the lower limits regarding sample size while providing comparable or greater statistical information than was previously available (29, 30). Due to the diminutive sample size requirements for LA sampling, typical casework sample sizes are easily accommodated. For example, to perform triplicate analyses of a single questioned glass fragment an optimal sample volume of 3 x 106 µm3 is preferred; this corresponds to a fragment 300 µm in length, 100 µm in width and 100 µm in depth. In terms of mass, such a fragment is approximately 7 µg. To examine glass by SN-ICP-MS, many would argue that the minimum sample mass is 500 µg but some agencies require 2000 µg – of which 100% is consumed by digestion and analysis. If a traditional-flow (1 mL min-1) nebulizer is used, only one analysis can be performed for a particular digest. This presents a limitation in that statistics cannot be applied to the result. Alternatively, only 0.9 µg glass is consumed during a triplicate analysis using LA. This is roughly 12% of a 7-µg fragment. The limited sample consumption of this technique enables the analyst to perform replicate analyses while preserving the majority of the sample. A benefit to the minimal sample consumed during LA analysis is that when fragments larger than the minimum are recovered (15 µg or more), there is enough sample for additional analyses to be conducted indepdently. To analyze fragments by SN- ICP-MS, the criminalist performing elemental analysis would be required to obtain permission prior to digestion and analysis; additional scientific experts would be then Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. 9 able to view the analysis conducted by the criminalist but they would not be able to conduct an independent analysis. Sample preparation requirements for LA introduction are greatly reduced over those required for SN introduction. For glass analysis by LA-ICP-MS, the samples need only to be cleaned and mounted on a glass slide. There is little opportunity for contamination and no dilution error – both advantages of analyzing the sample “as it is.” This benefit was much lauded by Lundell in 1933 (41), who asserts that valuable analyte information can be lost when removed from the original sample matrix. The field of criminalistics similarly fosters the ideology of in situ analysis – physical evidence examinations are always performed in such a manner as to provide highly discriminating data while preserving the character and quantity of the sample to the extent possible. Currently, the primary analytical limitation of LA is calibration; for some applications, internal standardization is an equally limiting factor. The issue of calibration stems from the fact that well characterized matrix-matched calibration standards are not readily available for many sample matrices. To overcome this issue many have attempted liquid calibration by SN or LA, while many others use the solid National Institute of Standards (NIST) standard reference glasses 610 and 612 for single point calibration (31, 42-46). The use of a single calibrant has been done following the validation of linear response in analyte: internal standard. Alternatively, it is done under the assumption that LA sampling does not alter the many decades of linear response available from ICP-MS. However, this assumption may not be valid for some analytes in certain matrices (47). As Akbar Montaser aptly put, calibration remains the “Achilles’ heel of the laser ablation technique” (31). Dodds, Pollock and Land Final Technical Report (Draft): 2004-IJ-CX-K007