Brooke W. Kammrath

	Forensic traces are physical remnants of past events that provide critical information for criminal and civil investigations and adjudications. The scientific examination of traces is an incredibly valuable tool for forensic investigations, because the skilled interpretation of traces yields factual answers to a range of pertinent questions. Not only are traces able to associate a person with an incident, thus answering the “who” question, but they may also be used to answer “what”, “where,” “when,” “why,” and “how” (1). The most significant factor limiting the value of traces for scientific investigations was eloquently captured by Paul Kirk (1953) when he wrote, “Only human failure to find it, study, and understand it, can diminish its value” (2). 

	The combination of microscopy and spectroscopy, or 

, is a powerful tool for the study of traces and enables their understanding and interpretation. Microspectroscopy has a plethora of advantages, most notably providing rapid and often non-destructive (or minimally destructive) analysis of a range of specimens, sometimes with little to no sample preparation, and creating a reviewable record of the examination (3). The chemical identification of traces via microspectroscopy may enable a criminalist to determine provenance, and, depending on the sample, individualize it (that is, successive reductions to a class of one, or the establishment of two items having a common source) (2–4). Microspectroscopy is not just for small samples. Although most people think of traces as being microscopic, this is not necessarily the case. Traces may be large objects, such as car headlights or baseball bats, that are left behind or removed from the scene after an incident. Microspectroscopy provides chemical information on the microscale that is useful for the study of small traces that require magnification (such as single fibers or gunshot residues), and large traces or mixtures where imaging is important for examining its physical or chemical components (such as for microhomogeneity or heterogeneity analyses).

	Forensic scientists use a range of microspectroscopical methods for the evaluation of traces, sometimes being at the forefront of their applications, while other times being more cautious with their implementation in casework. The first pairing of microscopy and spectroscopy was by Henry Clifton Sorby in 1868, who recorded the visible light spectra of natural biological pigments as well as blood, the latter having forensic implications (3). It was not until the 1950s that laboratories further developed visible light microspectrometers, which eventually led to their commercial availability. In the 1970s, its utility for the forensic analysis of traces, specifically for fiber evidence, was fully realized (3). Today, ultraviolet-visible (UV-vis) microspectroscopy is employed to study a range of traces, most commonly fibers, paint, and inks. Molecular microspectroscopy-infrared (IR) and Raman-is widely used in forensic science to detect and identify a variety of specimens, including, but not limited to, fibers, paint, polymers, drugs, soil minerals, cosmetic evidence, and general unknowns. IR microspectroscopy was first developed in the late 1940s and early 1950s, but did not see widespread adoption until the 1970s with the affordable availability of Fourier-transform computing capabilities (2). IR microspectroscopy was recognized at that time by some forensic scientists, notably microscopists at Walter C. McCrone Associates, Inc., to have value for the identification and discrimination of traces, and it was implemented for case examinations and research as soon as this technology became commercially available. Raman microspectroscopy was similarly embraced by the forensic science community via McCrone microscopists shortly after it came on the market in the late 1970s (5,6). One of the major advantages of Raman microscopy for the characterization of forensic traces is its ability to nondestructively analyze a particle through a glass coverslip, thus requiring no additional sample preparation beyond that which is required for traditional microscopy (2,7). The scanning electron microscope equipped with an energy dispersive X-ray spectrometer (SEM-EDX) was the first instrument to facilitate the elemental analysis of microscopic traces. The first ED X-ray spectrometer was added to an SEM by Nuclear Diodes (now Edax) in 1969 (8), and this technology was rapidly applied to the identification of gunshot residue and other traces (9). With the development of polycapillary focusing X-ray optics, micro-X-ray fluorescence spectroscopy (µXRF) became possible in the beginning of the 20th century (10), and was soon used to obtain the elemental profile of many forensic traces (such as glass, paint, inks, and paper). Each of these microspectrometers have advantages and limitations for forensic evidence analysis, which is why criminalists performing analyses of traces need to be well-versed in the methodologies, drawbacks, and significance of these techniques in order to choose the right tool to answer relevant questions for a scientific forensic investigation.

	Four recent developments in combined microscopy and spectroscopy instrumentation show great promise for providing additional information about traces that may translate to improved detection capabilities and increased discrimination. First, atomic force microscopy (AFM) coupled with IR and Raman spectrometers, the latter called 

, enable surface information (such as topography, hardness, adhesion, friction, and conductivity) and molecular chemistry to be elucidated with nanometer resolutions (11). Surface coatings and complex mixtures containing nanoparticles are ideally suited for analysis by AFM-IR and TERS, and these inevitably find their ways to the laboratory bench of a trace evidence examiner. The second innovative technology is optical photothermal IR spectroscopy (O-PTIR) which has been combined with Raman microspectroscopy to offer simultaneous explorations of samples with submicron resolution (11). O-PTIR is a non-contact, non-destructive method that has excellent potential for the analysis of complex mixtures, microfibers, and nanometer-thin layered polymers, paints, and coatings. A third trend in instrumentation is combining advanced imaging with molecular spectroscopy, with a standout being morphologically-directed Raman spectroscopy (MDRS) or particle-driven Raman spectroscopy. This automated instrumentation allows for size, shape, and chemical information to be quickly obtained for thousands (or hundreds of thousands) of particles within a complex mixture, thus supporting a meaningful statistical evaluation and increased sample information, which will undoubtedly advance the forensic interpretation of several samples of forensic interest (12). Fourth, in 2019 Leica Microsystems introduced a laser induced breakdown spectrometer (LIBS) microscope attachment that delivers minimally destructive elemental analysis (5-µm spot size) directly on the stage of a microscope without the need for additional sample preparation, akin to IR and Raman microscopes. I confidently predict that as the “micro-LIBS” instrumentation matures, it will prove valuable for the examination of many microscopic traces, such as glass, minerals, paint, and gunshot residues.

	Criminalists are asked to provide complex analyses of some of the most demanding samples. It is more than just their size, limited quantity, and unknown origins that present challenges, but also the information that traces hold which require sophisticated interpretations. It is for these reasons, and more, that the forensic community requires the most capable modern microspectrometers for the scientific interrogation of traces.

 for 35 years in publishing! Thank you to the 

 team for providing the scientific community with a free and easily accessible resource that delivers the latest information on advances in both instrumentation and applications. I look forward to reading 

P. Buzzini, B.W. Kammrath, and P. De Forest, 

Crime Investigation: Physical Evidence and the Police Laboratory 

(Interscience Publishers, Inc., New York, New York, 1953).

(John Wiley & Sons, Ltd., Hoboken, New Jersey, 2017). 

Principles and Practice of Criminalistics: The Profession of Forensic Science

B.W. Kammrath, A. Koutrakos, P.E. Leary, J. Castillo, J. Wolfgang, and D. Huck-Jones, 

is the Assistant Director of the Henry C. Lee Institute of Forensic Science and an Associate Professor in the Forensic Science Department at the University of New Haven, in West Haven, Connecticut. Direct correspondence to 

Articles

Forensic traces are physical remnants of past events that provide critical information for criminal and civil investigations and adjudications. The scientific examination of traces is an incredibly valuable tool for forensic investigations, because the skilled interpretation of traces yields factual answers to a range of pertinent questions.

Combining Spectroscopy with Microscopy for Advancing the Analysis of Forensically Relevant Traces

Advancements in technology enable deployment of field-friendly gas chromatography–mass spectrometry (GC–MS), so the value of the technique can be realized at the sample site. A primary benefit of deployment to the scene is the ability to analyze trace levels of threats, providing real-time information to support scene safety and other intelligence, so mission-critical decisions using confirmed information may be made. This capability is especially useful in situations where a sample degrades, changes, or is unstable during storage or transport, or when access to a laboratory is not an option. When explosives are encountered on the battlefield, GC–MS may be useful. In this research, a variety of commercial, military, and homemade explosives were analyzed using portable ion-trap GC–MS. Data from both ion-trap and quadrupole mass-spectral libraries were compared to establish whether portable ion-trap GC–MS is useful for the identification of explosives. Ion-trap mass spectral data can be different from quadrupole data due to ion chemistry and other ion interactions, including space charge that may occur during ion trapping.  These results show that, although mass spectral data are sometimes different from quadrupole mass spectral data, portable ion-trap GC–MS is useful for performing confirmatory identification of explosives in the field. The forensic value of this capability is also considered.

	
	Portable gas chromatography–mass spectrometry (GC–MS) instrumentation has enabled the confirmatory analysis of explosive residues in the field. For battlefield forensics, portable GC–MS instruments are used for the detection and confirmatory identification of pre- and post-detonation threats. In addition, these systems provide information regarding the source of explosives based on the detection and identification of trace-level chemicals in the sample. The use of portable GC–MS is valuable for supporting scene safety, gathering intelligence, and providing investigative and adjudicative information about explosives events.

	It is important to recognize that GC–MS in the laboratory has historically been used for some of these same purposes, but there are significant benefits to performing the analysis at the scene. First, on-scene analysis enables the development of render-safe procedures based upon real-time threat identification and assessment. Second, intelligence turnaround time to support offensive operations is significantly improved. Commanders are empowered to make actionable decisions using reliable, confirmed information. Third, GC–MS results in real time can help guide and optimize scene processing. Finally, analysis at the sample site enables the most accurate evaluation of the scene. Results are representative of the scene at the time of analysis, not at some later date when samples are received and analyzed at the laboratory. This timeliness is especially relevant for this type of physical evidence, because, aside from the danger associated with storage and transport of explosive evidence, some explosives have relatively high vapor pressures, and will evaporate or sublimate during transport to the laboratory.

	Portable GC–MS systems were initially introduced to the market almost 25 years ago. Since this time, the design and performance of these systems has evolved (1,2), and there are currently a number of portable GC–MS options available to both scientists and non-scientists performing field analysis. These systems include both quadrupole and ion-trap designs. Quadrupole systems offer the benefit of generating classical mass spectra that are more easily interpreted and directly comparable with spectra in the National Institute of Standards and Technology (NIST) MS database. Ion-trap systems are desirable for field use, because they can run at higher operating pressures. The pumps required to achieve and maintain these higher pressures are more field friendly than the chemical pumps that were historically needed to achieve and maintain the operating pressures required of portable quadrupole systems. However, spectra generated using ion-trap systems may exhibit ion-chemistry events such as dimer formations occurring in the trap (2), and, therefore, are not always directly comparable with mass spectra within the NIST MS database. In addition, space charge is sometimes observed. Space charge occurs when the force of ion-ion repulsions inside the trap become too great, and can result in incorrect 

 assignments. Space charge further complicates spectral library matching. Library search results are frequently critical to end users of portable GC–MS systems who rely on library matching to achieve successful results in the field.

	To evaluate the potential use of portable ion-trap GC–MS analysis for the on-scene evaluation of explosive evidence, explosive standards were analyzed and compared with the proprietary library of the system and with the NIST MS database spectra to establish effectiveness of search results for these explosive standards. In addition, explosives samples received from international law-enforcement organizations were analyzed using portable ion-trap GC–MS, and results generated were reviewed to evaluate capabilities of the technology for analyzing the explosive-related components in these recovered samples.

	The portable GC–MS system used in this research directly links a low thermal mass resistively heated capillary column with a miniaturized toroidal ion-trap mass spectrometer (3–5). The GC column enables fast separation and thermal recovery (3-min analysis time) resulting in high throughput (approximately 5 min between injections). High throughput is always a desirable feature of analysis, but especially for field applications where scene-processing time may be limited or the scene itself may be dangerous. The toroidal ion-trap design is intended to increase trapping volume and, therefore, to minimize ion-ion repulsions. Regardless, the impact of both space charge and ion chemistry on the mass spectra generated may still be observed when using this system.

	The following equipment was used in this research:

		Smiths Detection Guardion portable GC–MS system with a GC capillary column; 5 m-long, Restek Corporation MXT-5 resistively heated, crossbond diphenyl-dimethylpolysiloxane, inner diameter 0.1 mm, film thickness 0.4 μm

		Leland Limited Incorporated disposable helium cartridge, 95 cc x 5/8 in helium filled, 2.4 g at 2500 psi

		Torion Technologies Custodian solid-phase microextraction (SPME) holder

		Supelco 23-gauge, 65-μm polydimethylsiloxane/divinylbenzene (PDMS/DVB) SPME fiber assembly

		Smiths Detection Guardion performance validation mixture sample

	Explosive standards were purchased from AccuStandard, Inc., in New Haven, Connecticut, and include:

		2,3-dimethyl-2,3-dinitrobutane (DMNB), 100 µg/mL in acetonitrile, CAS#3964-18-9

		2,4-dinitrotoluene (2,4-DNT), 1000 µg/mL in methanol, CAS#121-14-2

		ethylene glycol, dinitrate (EGDN), 100 µg/mL in acetonitrile:methanol (95:5), CAS#628-96-6

		hexamethylene triperoxide diamine (HMTD), 100 µg/mL in acetonitrile, CAS#283-66-9

		pentaerythritol tetranitrate (PETN), 1000 µg/mL in methanol, CAS#78-11-5

		1,3,5-trinitro-1,3,5-triazinane (RDX), 1000 µg/mL in methanol:acetonitrile (50:50), CAS#121-82-4

		triacetone triperoxide (TATP), 0.1 mg/mL in acetonitrile, CAS#17088-37-8

		1,3,5-trinitrobenzene (TNB), 1000 µg/mL in methanol:acetonitrile (50:50), CAS# 99-35-4

		2,4,6-trinitrotoluene (TNT), 1000 µg/mL in methanol:acetonitrile (50:50), CAS#118-96-7.

	Law-enforcement explosive evidence samples were received from various international law-enforcement organizations. These samples are detailed in Table I. The 20 samples are each from a different origin. The explosive type, as well as the sample description for each, was provided by the contributing organization. Aside from the details shown in Table I, no further information about the history of these samples was provided to the authors.

	Prior to all sampling, the SPME fiber was verified to be clean by performing an analysis using the GC–MS method described below. If chemicals were detected, the SPME fiber was continually cleaned by heating, until no chemicals were detected on the fiber. Explosive standard solutions were sampled using the direct-deposition method. Law-enforcement explosive samples were analyzed using either SPME headspace sampling, direct deposition, or both methods.

	Between 100 and 500 mg of sample was transferred to a GC–MS headspace vial. The vial was closed and then stored at 22 °C for at least 2 h. The headspace was then sampled by piercing the vial cap with the needle end of the SPME holder containing the SPME fiber and exposing the fiber for a specified period of time. Sampling times were variable and ranged from 10 min to 40 min.

	For explosive standards, explosive amounts between 20 and 200 ng were directly deposited using a microsyringe onto the coated portion of the SPME fiber, and allowed to dry for up to 5 min to allow evaporation of the solvent. Although attempts were made to control the amount of sample deposited onto the fiber, no attempts to provide quantitative analysis of these data were made. For law-enforcement samples, between 100 and 500 mg of sample was dissolved in 10 mL of acetone and stored at 22 °C for at least 2 h. Then, 10 µL of the sample solution was directly deposited using a microsyringe onto the coated portion of the SPME fiber, and allowed to air dry for up to 5 min, to allow evaporation of the solvent.

	Performance of the system was tested at the start of every work day. Performance testing evaluates GC, MS, and library search performance. To perform this testing, a standard containing 13 chemicals is analyzed. These 13 chemicals have expected retention-time values between 0 and 90 s. GC performance acceptance criteria requires that the retention times for all 13 chemicals must be the stated value ±2 s. MS performance includes tests for spectral quality, mass calibration, mass resolution, ion statistics, space charge, signal-to-noise ratio, and sensitivity. Library- search testing verifies the system is matching substances as expected against the system’s proprietary library. The system always passed all performance-test criteria prior to use. The performance-test procedure used was the procedure recommended by the instrument manufacturer.

	The system used in this research is deployed with a single GC–MS method. This method is linked to a proprietary library, as well as a condensed version of the NIST MS database. It is required that, if the user wants to search these libraries at the time of analysis, the deployed GC–MS method must be used. Although there is some deviation from that method that is tolerated, the use of some GC settings, including ramp rate, begin temperature, and begin hold time are required. This is because library search criteria use retention times when searching the proprietary library, and retention index when searching the condensed NIST MS database. It is possible to change these settings, but this would require either library development (if automatic library searches were required), or manual interpretations of GC–MS results.

	The method’s standard GC settings (which were used for this research) areas follows:

	A manual review of all data from the explosive standard samples was performed, and chromatographic peaks and characteristic mass spectra for seven of the nine samples were reliably detected. For these seven samples, detections were possible at both low and high concentrations, and at all inlet temperatures. For two of the nine standards, no chromatographic peak for the explosive was detected. In addition, expected target ions were actively searched in the region of their expected retention and across the entire chromatogram in each data file to confirm the negative findings of the chromatographic data review. Table II summarizes the detection results from the manual review of the GC–MS data from the explosive standards.

	RDX was not detected in the standard sample. As is described later, it also was not detected in any law-enforcement samples expected to contain RDX. Although no RDX was detected, it is assumed that the system in its current configuration is capable of detecting RDX because this substance is programmed into the proprietary library for the system. This library is purported to contain library data generated using the system with its deployed method. The library entry describes GC retention values and MS fragments consistent with those reported in the NIST MS database for RDX, so it is assumed the system is capable of detecting RDX, not just its degradation products, impurities, or other contaminants. The failure to detect this compound using the portable GC–MS may be due the inlet temperature or transfer line temperatures being too high. As has been reported by other researchers, the thermally sensitive nature of some explosives can cause them to degrade when introduced to the high temperatures of the inlet (6). To address this potential expectation, the systems inlet temperature was lowered to 210 °C, but no RDX was detected. It is likely that a method for this substance at even lower temperature would be required for its detection.

	HMTD was also not detected in the standard sample under any analysis conditions. It was also not detected in any of the law-enforcement samples, although none of the law-enforcement samples were expected to contain this explosive. In accordance with the NIST MS database, HMTD has an estimated non-polar retention index value of 1619, and its three most intense mass fragments at 

 values in descending order of intensity of 208, 88, and 45. As such, this explosive would be expected to exhibit a retention time in this portable GC–MS under these analysis conditions of approximately 101.5 s. No chromatographic peaks were observed at or around this retention time. It is also important to note that the entry for this explosive in the proprietary library of the portable GC–MS system listed the anticipated retention time of 34.4 s for HMTD (RI value of approximately 770), with its most intense mass fragments in descending order of intensity of 73, 74, and 44. The significant discrepancy between the retention values reported in the NIST MS database and the proprietary library entry of the portable GC–MS system indicate the entry in the proprietary library is not representative of HMTD, and is potentially a degradation product, impurity, or other contaminant.

	In addition to performing a manual review of the data, all GC–MS data files for the seven detected explosives were searched against the library of the portable GC–MS system. This search uses its own proprietary search algorithm. Of the seven samples detected, only six were in the library. For the six explosives in the library, an automatic identification of the explosive sample was reliably made. The automatic identification occurred at both low and high concentrations, and at all inlet temperatures.

	A search of each GC–MS data file for the seven detected explosives was also performed by comparing the mass spectrum at the chromatographic peak maximum of the explosive against the NIST MS database using the NIST MS database search software. For six of the seven samples, the highest probability hit always correctly identified the explosive substance. However, TATP was only identified as the top hit about 50% of the time. The ability to correctly identify TATP as the top hit against the NIST MS database was not attributable to sample load. It is important to note that searches against the NIST MS database only consider mass spectral data when generating a hit list. A summary of library searching results is shown in Table III.

	Finally, the mass-spectral data from the ion-trap system were manually compared with NIST MS database quadrupole spectra. In most cases, as is evidenced in the automatic library search results, no significant differences were observed between the two types of spectra. Figure 1 shows the mass spectrum from the portable ion-trap system, along with the NIST MS database spectrum of TNT. Although small differences in peak intensities were observed across the spectrum, these spectra are generally consistent with each other. The most significant difference between the ion-trap and quadrupole spectra were observed in the spectra of DMNB. The ion-trap spectrum of DMNB had an additional ion fragment at 

 130 that was not observed in the NIST MS database spectrum. This fragment had an intensity of about 13% of the most intense fragment in the spectrum (

41 was not observed in the ion-trap data. However, the ion-trap system is only specified to detect fragments of 

43 to 500. Therefore, not too much should be interpreted from its absence in this data. The ion-trap spectrum of DMNB from the portable ion-trap system and the NIST MS database spectrum of DMNB are shown in Figure 2.

	The TATP data was also reviewed to investigate the cause of the missed identifications when searched against the NIST MS database. It is likely that the missed identifications are due to the simplicity of the mass spectrum of this explosive. The most intense fragment in the NIST MS database spectrum is 

43, followed by fragments at 58, 59, and 70, which have relative intensities to 43 of 13%, 11%, and 10%, respectively. There are spectra of many chemicals within the NIST MS database that have very similar spectra to TATP. The incorrect identifications of TATP when searched against the NIST MS database were all from chemicals that had very similar spectra to TATP. It was possible to sometimes, but not always, exclude the missed identifications due to inconsistencies between the retention time observed in the chromatogram from the ion-trap GC–MS system and the retention index reported in the NIST MS database.

Law-Enforcement Explosive Evidence Samples

	Table IV details the explosive-related compounds included in the proprietary library of the portable ion-trap GC–MS system; Table V details the automatic library search results for the explosives-related identifications from the 20 law-enforcement samples. A manual review of all data was also performed. As previously mentioned, the system did not detect RDX. This is important because a number of samples contain RDX, including Semtex and C4. In addition, the system failed to automatically identify tetryl, even though tetryl was in the library. This is because this sample was slightly overloaded, and does not chromatograph well on this column. The poor chromatography coupled with column overload caused a shift in the observed retention time. This observed retention time was outside of the retention-time window of the library entry. Therefore, the substance was not identified.

	Aside from missing RDX and tetryl, the system returned a few false automatic identifications including for EGDN in two of the three samples of C4 and one of the six samples of Semtex, PETN in one of the three dynamite samples, dinitroglycerin in four of the six samples of Semtex, and RDX in both TNT samples. Identifications were considered false if no chromatographic peak was present, but the software identified the substance. Such a result may occur when spectra are noisy and the chemical for identification has few or common mass fragments. For instance, the proprietary ion-trap library entry for dinitroglycerin contains only three fragments with 

 values of 46, 76, and 64, at 100%, 8%, and 1% intensities, respectively. It is reasonable to expect that these three fragments will be present in noisy data.

	A review of the results also shows that there is value to analyzing explosive samples using both headspace and direct-deposition sampling. For instance, dinitroglycerin in dynamite samples was always observed in the headspace samplings, but never in the direct-deposition sampling. For TNT, DNT degradation products were observed in the headspace sampling, but not in the direct-deposition sampling. The reverse was true for the TNT that was not observed in the headspace sampling, but was reliably observed in the direct-disposition samples. The different results that occur from using the two sampling methods are not contradictory to each other; they enable a more complete assessment of the sample. Headspace sampling prior to sample dilution is useful for identifying the components of the sample with higher vapor pressures. Vapor-pressure values of explosives reported in the literature, for a number of reasons, are variable and sometimes contradictory (7–9), but Figure 3 summarizes the ranking order of vapor-pressure values of many explosives at 25 ËC, calculated by Ewing and associates after an extensive review of values reported in the literature. (8)

When explosives are encountered on the battlefield, the use of portable GC–MS is valuable for the detection and confirmatory identification of pre- and post-detonation threats. In addition, this technique provides information about the source of explosives based on the detection and identification of trace-level chemicals in the sample. The data presented here confirm this capability.

Optimized Explosives Analysis Using Portable Gas Chromatography–Mass Spectrometry for Battlefield Forensics

Raman spectroscopy is fast becoming a popular technique in the forensic science discipline, and more recently its focus has turned to biological samples. This study reveals the ability of Raman spectroscopy to identify some forensically relevant body fluids, both individually and within mixed samples, that can be crucial in some forensic investigations. This study also further demonstrates the capabilities of Raman as a means for human blood identification in simulated crime scene samples to include bloodstains on a variety of fabrics, at varying dilutions, following laundering, and with the use of various blood-enhancement reagents. The impact of washing and blood-enhancement reagents reveals the importance of the choice of method and its bearing on subsequent Raman analysis.

	In forensic investigations, determining the identity of an unknown stain can aid both in reconstruction and, if it is a biological stain, identification of an individual. Many suspected body fluid stains are first presumptively identified at the scene, then confirmed in a laboratory setting. However, many of the tests used, both presumptive and confirmatory, consume the sample in question, preventing further analysis, namely DNA profiling. These methods can also be time- and labor-intensive, while producing variable results. Raman spectroscopy has been gaining interest as a new method of body fluid identification, partly because of its nondestructive nature. Prior research has demonstrated that Raman spectroscopy gives a unique spectrum for several body fluids regularly found at crime scenes, while also preserving the sample for DNA analysis (1,2).

	Many of the presumptive tests for biological stains are reagent-based, targeting specific components in the body fluid being tested. These are then followed by a second test used to confirm the result of the initial test. In blood, the main component tested for is hemoglobin (3), and if a presumptive positive is found, the Takayama crystal test can be performed (3), although "card" methods are being used more (4). Presumptive testing for semen relies on the presence of seminal acid phosphatases, although these tests can create false positives because of the presence of acid phosphatases in other body fluids (5). These presumptive tests are then checked for the presence of spermatozoa to confirm the presence of semen. The sperm are stained using techniques such as Christmas tree staining (6). If the sample is azoospermic, prostate-specific antigen (PSA) can also be tested for, but this technique is also nonspecific (7). For saliva, presumptive tests are mainly color reaction tests checking for the presence of α-amylase (8). Confirmatory tests for saliva consist of enzyme-linked immunosorbent assays (ELISAs), targeting amylase (9) or statherin (10). Presumptive testing for urine is nonspecific and insensitive (3), but it can prove useful in forensic investigations because DNA has been recovered from urine samples (11). Methods are being developed using messenger RNA (mRNA) (12) and micro RNA (miRNA) (13) as a means of identifying body fluids, both individually and in mixtures. However, there is still a need for a sensitive and specific test to identify biological fluids found on evidence.

	Raman spectroscopy has several biological uses outside of forensic science, and it has been used to characterize different biochemical processes in cells (14,15) and identify bacteria to the species level (16). It is currently emerging as a technique in the forensic disciplines because of its nondestructive nature and reagent-free sample preparation (17). Raman also requires minimal amounts of a sample and receives little interference from water molecules, separating it from other forms of vibrational spectroscopy, specifically infrared (IR) spectroscopy (18). Portable Raman spectrometers have further increased the popularity of the technique because it has allowed for testing at a crime scene (19). Raman spectroscopy has been used in the forensic sciences to identify a variety of different materials chemically, including pen inks (20), condom lubricants (21), controlled substances (22), and fabrics (23). For the analysis of biological fluids, Raman spectroscopy has been shown to identify blood (17), semen (24), and saliva (25). Blood and semen were analyzed further, and it was revealed that both could be detected on several different substrates (26,27). It was also shown that blood and semen could be differentiated from each other in mixtures (28).

	This research had several aims. First, the aim was to confirm the generation of unique spectra for each of the individual body fluids using Raman spectroscopy. Second, the aim was to determine the ability of the technique to differentiate between two biological fluids in mixed samples. Third, the study aimed to investigate the effects of laundering and enhancement reagents on the Raman spectroscopic signature of human blood in simulated crime scene samples. Finally, the study aimed to investigate the impact of extracting bloodstains from fabric before Raman analysis.

	All body fluid samples used in this study were collected after Institutional Review Board (IRB) approval and with the informed consent of the volunteers. Venous blood was collected from the volunteers by a registered phlebotomist. Saliva, semen, and urine were collected by the volunteers themselves using sample collection kits. After the volunteers collected and returned the samples, the kits were stored at 4 °C.

Individual and Mixed Body Fluid Preparation

	First, 10 µL of each body fluid sample to be analyzed was placed either directly onto an aluminum foil–covered glass microscope slide or onto a piece of a fabric substrate that was taped to such a substrate. All four body fluids were tested on black cotton and white cotton, with blood being further analyzed on white polyester, black polyester, and denim.

	All possible two-component mixtures of the four body fluids were tested, for a total of six different mixtures. Four samples of each mixture were prepared: two samples with the fluids in equal ratios, and one sample of a 1:3 ratio for each fluid in the mixture (that is, 5 µL of blood in 15 µL of semen, and the reverse). Body fluids were mixed before being placed onto the slide. All samples were allowed to dry overnight.

Laundering and Blood-Enhancement Reagents

	For the testing involving the blood enhancement reagents, 150 µL of blood was placed in triplicate on six swatches of white cotton fabric and allowed to dry. The swatches were then analyzed using Raman spectroscopy. Afterwards, half of the swatches were washed in the hot water cycle of a washing machine without detergent for 26 min. Those swatches were reanalyzed using Raman spectroscopy. Following this, one unwashed and one washed swatch were treated with one of three blood enhancement reagents: leuco crystal violet (LCV), Coomassie blue, or luminol. After drying overnight, these samples were once more analyzed using Raman spectroscopy.

	The evaluation of the extraction method involved the testing of six volumes of blood (10, 50, 100, 150, 200, and 250 µL) on five substrates (white cotton, black cotton, white polyester, black polyester, and denim). After drying, the stains were cut from the fabric and placed into a centrifuge tube. Deionized water was added to the tube based on the size of the stain, and the tube was shaken until the fabric was completely wet. The fabric was then placed into a spin basket, and the tube was centrifuged at 14,000 rpm (20,817 rcf) for approximately 2 min. For the 200- and 250-µL volumes, the stains were cut in half, prepared in separate tubes, and recombined after being centrifuged. Two dilutions, 1:10 and 1:100, were also tested, and were prepared in the same manner on all five substrates. The stains treated with the enhancing reagents were extracted in the same manner. Then 10 µL of all the resulting samples was placed on microscope slides as described above.

	All samples were analyzed using a DXR Raman microscope (Thermo Scientific) equipped with a 10× objective and Thermo Scientific OMNIC software. A 780-nm laser was used, and kept at a constant power of 10 mW. The spectrometer aperture was a 50-µm slit and a 400-line/mm grating was used.

	Five spectra were collected from different areas of each sample, with each analysis consisting of five 20-s exposures. Areas of neat substrate were also analyzed to collect spectra representative of the fabrics used

	The five runs from a single sample were then averaged using the OMNIC software to create the sample average. Any spectra requiring baseline corrections were manually corrected, also using the software. Spectral subtractions were performed by exporting to GRAMS/AI 7.01 and using that software for the subtraction. The extracted samples were compared to a library reference created in OMNIC, using a correlation algorithm that gave "match" scores.

Identification of Individual and Mixed Body Fluids

	The individual body fluids tested were differentiated from each other, with the body fluid average spectra shown in Figure 1. Three of the four body fluid spectra contained peaks that were not present in the other fluids, as shown in Table I. Saliva could not be identified by its Raman spectrum because there was only one peak (at 1349 cm

) detected in the four different saliva samples. This peak was also found in the spectra of urine and semen. This result is not in agreement with previously published research (25). Identification of a material by one peak alone is not acceptable; thus classification of this body fluid by Raman spectroscopy is not possible. Consequently, saliva was removed from further data analyses. It was noted that although the urine samples gave characteristic spectra, there was a larger amount of variability within samples than with the other body fluids.


	Figure 1: Average Raman spectra of blood (blue), saliva (purple), semen (green), and urine (red).

	Eight of the nine mixture samples, made from three concentrations of the two body fluid components (blood–semen, blood–urine, and semen–urine), were identifiable. Identification was made by the presence of the characteristic peaks for each body fluid (Table I). For the blood–semen mixtures at 1:1 and 1:3 ratios, both components were identified although semen's major peaks were masked by the blood spectra, however the minor peaks were present. Semen was not identified in the blood–semen mixture at the 3:1 dilution. For the blood–urine mixture at all concentration, both components were identified. Similar to the blood–semen mixtures, urine's major peaks were masked by the blood spectra; however, the minor peaks were visible. Both components of the semen–urine mixture were identified from the mixture spectra at all three dilutions. It is important to note that these identifications took place under ideal laboratory conditions, and the identities of the components in the mixtures were known. For identification of unknown samples, the creation of a thorough database would be required.

	The substrate testing revealed that semen and urine were not identified on black cotton or white cotton. The signals from the substrates were too strong for the signals of the body fluids to be detected, resulting in indeterminate results of spectral subtraction. Furthermore, none of the mixtures were identified on either of the substrates tested for the same reasons.

	Blood was tested on five different substrates, and was identified on two: white cotton (Figure 2) and white polyester. The darker dyes in the denim, black cotton, and black polyester created a large amount of fluorescence and interference that masked the blood signal. The signal retrieved using spectral subtraction on the two lighter substrates was rather weak, but possessed several of the major peaks associated with venous blood.


	Figure 2: Raman spectral subtraction results for blood on white cotton.

	The Raman analysis of the wet and dry dilution series revealed that blood diluted beyond a 1:100 ratio could not be detected, and the diluted blood could be detected only when dry (Figure 3). With each successive dilution, the peaks in the spectra became less intense and began to lose their characteristic shape. Only the neat blood sample was detected while wet, and the signal had significantly more noise than the dried samples.


	Figure 3: The average Raman spectra of the dried dilution series of blood stains, including whole blood (blue), 1:10 dilution (purple), 1:10

The Impact of Laundering and Blood-Enhancement Reagents

	The results of the washing and subsequent enhancement of blood stains revealed that laundered blood stains do not produce a Raman signal strong enough to be detected, and the further enhancement of the blood stains, while giving presumptive positives for the laundered samples, do not aid in recovering the Raman signal. All the initial stains produced spectra consistent with venous blood after spectral subtraction was used to remove the underlying fabric. All laundered samples did not produce any signals consistent with the venous blood reference spectrum. This result is consistent with the dilution results because anything less than 1:100 would not have produced a detectable Raman spectrum. Post-enhancement, only the unwashed samples treated with luminol returned Raman spectra consistent with the venous blood reference (Figure 4). Any laundered samples or samples treated with the LCV or Coomassie blue possessed too much interference or too weak of a blood signal to allow for detection. The samples extracted from the unwashed blood stains treated with LCV and luminol were shown to have spectra consistent with venous blood, allowing for the identification of more samples than the subtraction method.


	Figure 4: Raman spectral subtraction results for an unwashed bloodstain on white cotton treated with luminol.

The Impact of Extracting Bloodstains from Fabric Before Analysis

	The extraction method proved to be moderately successful, in that the venous blood library reference spectrum was provided by the correlation library search as the first match for 87 of the 90 neat blood samples that were tested on all five substrates at all six volumes. Of the three samples not identified, each had something other than venous blood as the highest match, and two had blood as the second most likely substance. Furthermore, two of the three false negatives occurred in samples extracted from black polyester, and the third in a sample from denim. Thus, the misidentification by the library may have been caused by the substrates they were extracted from, because it was noted under the microscope that small fibers of the fabrics had passed through the spin basket and were present in the sample. These fibers may have introduced some interference, leading to the incorrect identifications. In addition, the hit quality itself was quite poor for all 90 spectra, with hit qualities ranging from 29.32 to 44.68. The low match scores likely resulted from the decreased intensity of peaks caused by the inherent dilution required for the extraction to occur.

	This research highlights the potential of Raman spectroscopy for body fluid identification, and further demonstrates the importance of taking into consideration the impact certain crime scene variables have on the ability of the method to correctly identify body fluids. This research has shown that Raman spectroscopy is capable of the nondestructive identification of some forensically relevant body fluids both individually, and in mixed samples, prepared in laboratory conditions. In forensic investigations, however, body fluids are rarely found in pristine conditions and it is crucial that the method is tested on simulated crime scene samples. Body fluids are typically recovered on various substrates, such as clothing, where the body fluids may be diluted or the fabric washed or laundered. This study revealed the impact that various fabric substrates and certain colored substrates have on the ability of Raman to correctly identify certain body fluids. Evidence items that are suspected to have latent bloodstains are typically sprayed with a presumptive blood-enhancement reagent. This study tested Raman's sensitivity by assessing diluted blood samples, revealing blood dilutions up to 1:100 to be detectable using Raman spectroscopy. All samples tested after laundering were not detectable, suggesting the blood had been diluted greater than 1:100 in the laundering process. The application of commonly used presumptive blood-enhancement reagents revealed luminol to have no impact on the ability of Raman to correctly identify unwashed blood samples. This is important information to understand when determining an analytical scheme for the analysis of a suspected body fluid both at a crime scene and in the forensic laboratory. Furthermore, it was also shown that although extracting the body fluids from the substrates may be an alternative to spectral subtraction, additional research is needed to find a method to obtain more robust signals for library searching or another method for comparison and subsequent identification. The use of Raman spectroscopy for body fluid identification is indeed a hot topic and keen area of research and it is crucial that the research, such as the study discussed in this article, addresses the real-world applications of the method.

	The authors would like to thank the Summer Undergraduate Research Fellowship program at the University of New Haven for supporting this work.

	(1) V. Sikirzhytski, K. Virkler, and I.K. Lednev, 

http://doi:10.1016/j.forsciint.2008.08.004

cientific Protocols for Forensic Examination of Clothing

 (CRC Press, Boca Raton, Florida, 2010), pp. 123–143.

Cold Case Homicides: Practical Investigative Techniques, 

 R.H. Walton (CRC Press, Boca Raton, Florida, 2006), pp. 317–335.

	(5) R.E. Gaensslen, National Institute of Justice, 

Sourcebook in Forensic Serology, Immunology, and Biochemistry. 

 p. 692 (U.S. Department of Justice, National Institute of Justice, Washington, D.C.). Available at: 

https://www.ncjrs.gov/pdffiles1/pr/160880_unit_3.pdf

, Volume II, 2nd Edition, R. Saferstein, Ed. (Prentice-Hall, Upper Saddle River, New Jersey, 2002), pp. 525–613.

	(7) S. Schmidt, M. Franke, J. Lehmann, T. Loch, M. Stöckle, and K. Weichert-Jacobsen, 

http://doi.org/10.1016/S0090-4295(00)01093-1

http://doi.org/10.1111/j.1556-4029.2008.00755.x

	(9) L. Quarino, Q. Dang, J. Hartmann, and N. Moynihan, 

	(10) T. Akutsu, K. Watanabe, Y. Fujinami, and K. Sakurada, 

http://doi.org/http://dx.doi.org.unh-proxy01.newhaven.edu:2048/10.1007/s00414-009-0391-2

	(11) T. Nakazono, S. Kashimura, Y. Hayashiba, K. Hara, A., Matsusue, and C. Augustin, 

http://doi.org/10.1111/j.1556-4029.2008.00675.x

	(12) Y. Xu, J. Xie, Y. Cao, H. Zhou, Y. Ping, L. Chen, and Z. Zhao, 

http://doi.org/10.1371/journal.pone.0100123

	(13) M.L. Uchimoto, E. Beasley, N. Coult, E.J. Omelia, D. World, and G. Williams, 

http://doi.org/10.1016/j.fsigen.2013.04.006.

 F.S. Ligler and J.S. Kim, Eds. (Pan Stanford Publishing, Singapore, 2010), Chap. 15.

	(15) F. Pascut, A. Downes, and I Notingher, in 

Optical Techniques in Regenerative Medicine,

 S.P. Morgan, F.R.A.J. Rose, and S.J. Matcher, Eds. (Taylor & Francis Group, Boca Raton, Florida, 2013), pp. 205–236.

	(16) P.C. Buijtels, H.F. Willemse-Erix, P.L. Petit, H.P. Endtz, G.J. Puppels, H.A. Verbrugh, A. van Belkum, D. van Soolingen, and K. Maquelin, 

	(18) J.G. Grasselli, M.K. Snavely, and B.J. Bulkin, 

Chemical Applications of Raman Spectroscopy

 (John Wiley and Sons Ltd., New York, New York, 1981).

	(19) B.A. Eckenrode, E.G. Bartick, S.D. Harvey, M.E. Vucelick, B.W. Wright, and R.A. Huff, 

Portable Raman Spectroscopy Systems for Field Analysis

 (Forensic Science Communications). Retrieved July 28, 2015, from 

https://www.fbi.gov/about-us/lab/forensic-science-communications/fsc/oct2001/eknrode.htm

	(20) I. Geiman, M. Leona, and J.R. Lombardi, 

http://doi.org/10.1111/j.1556-4029.2009.01058.x

	(22) V. Rana, M.V. Cañamares, T. Kubic, M. Leona, and J.R. Lombardi, 

http://doi.org/10.1111/j.1556-4029.2010.01562.x

http://doi.org/10.1016/j.forsciint.2009.09.005

	(26) G. McLaughlin, V. Sikirzhytski, and I.K. Lednev, 

http://doi.org/10.1016/j.forsciint.2013.04.033

	(28) V. Sikirzhytski, A. Sikirzhytskaya, and I.K. Lednev, 

http://doi.org/10.1016/j.forsciint.2012.07.002

Tyler Schlagetter, Brooke W. Kammrath, and Claire L Glynn 

are with the Department of Forensic Science in the Henry C. Lee College of Criminal Justice and Forensic Science at the University of New Haven in West Haven, Connecticut. Direct correspondence to: 

This study reveals the ability of Raman spectroscopy to identify some forensically relevant body fluids, both

Brooke W. Kammrath

Articles

Combining Spectroscopy with Microscopy for Advancing the Analysis of Forensically Relevant Traces

Optimized Explosives Analysis Using Portable Gas Chromatography–Mass Spectrometry for Battlefield Forensics

The Use of Raman Spectroscopy for the Identification of Forensically Relevant Body Fluid Stains