Home About Us Laboratory Services Forensic Science Communications Back Issues July 2000 Use of an X-Ray Spectral Database in Forensic Science by Ward...
Info
This is archived material from the Federal Bureau of Investigation (FBI) website. It may contain outdated information and links may no longer function.

Use of an X-Ray Spectral Database in Forensic Science by Ward (Forensic Science Communications, July 2000)

Use of an X-Ray Spectral Database in Forensic Science by Ward (Forensic Science Communications, July 2000)
fsc_logo_top.jpg
fsc_logo_left.jpg

July 2000 - Volume 2 - Number 3


Use of an X-Ray Spectral Database in Forensic Science

Dennis C. Ward Chemist
Materials Analysis Unit
Federal Bureau of Investigation
Washington, DC

Introduction | Traditional Use of SEM/EDS in Forensic Science
Limitations of Commercial Software Applied to Forensic Applications
Historical Efforts | Present Design
Storage of Standard Spectra and Associated Information | Data Retrieval
Fundamental Assumptions | Additional Applications | Future Development
Conclusion References

Introduction

The scanning electron microscope (SEM) with associated energy dispersive spectrometer (EDS) is commonly used to characterize the structure and elemental composition of a wide variety of materials of forensic significance. When these materials are extremely limited in size, SEM/EDS is often the only suitable method for characterization available in forensic science laboratories.

Because SEM/EDS is unable to access data from large numbers of materials, the Federal Bureau of Investigation (FBI) Laboratory believes that this discipline has not realized its full potential. The usefulness of SEM/EDS has been limited by its inability to archive spectra within a utility having a true database architecture. The FBI Laboratory has designed an X-ray database consisting of storage, query, and display utilities unique to X-ray spectroscopy.

It is not the intent of this paper to describe SEM, EDS, or applications except in relation to database development.

Traditional Use of SEM/EDS in Forensic Science

Traditional analysis with SEM/EDS can be quantitative, qualitative, structural, or comparative.

The objective of quantitative analysis is to determine elemental composition as accurately as possible. Sophisticated data analysis converts X-ray intensity from each element into element concentration. The precision and accuracy of this technique are affected by sample homogeneity, sample size, sample preparation, sample/instrument geometry, instrument stability, counting statistics, and sample matrix. In general, a sample must be homogeneous to the micrometer level, flat, polished, and analyzed with a high take-off angle. The electron beam must be focused on a small area of the sample.

Qualitative methods are used simply to determine the presence of certain elements regardless of absolute concentration.

Various structural imaging methods are used to create pictures of structure, which can be either topographical or compositional.

When any of the above methods are applied to the analysis of several samples, the exam is considered comparative. When comparisons are required, the analyst must analyze each sample under identical conditions.

Many commercial suppliers provide excellent application utilities to perform these traditional analyses. Often they are bundled into a sophisticated analytical suite. Many specific application utilities, such as those for gunshot primer residue analysis, are also available.

Limitations of Commercial Software Applied to Forensic Applications

In addition to the traditional uses of SEM/EDS indicated above, forensic science has specialized requirements, including the following:

  • The need to compare the spectrum of a questioned material to that of reference standards. Frequently it is necessary to identify a material. Many samples requiring compositional comparison, however, are not suitable for traditional quantitative analysis by SEM/EDS by nature of their size, matrix composition, or homogeneity. In these instances a method other than quantitative comparison is needed. One such method involves the simple matching of spectra. Even when elemental composition is unlikely to uniquely identify a material, the composition can associate the material with a particular class of material.
  • The ability to retrieve spectra by nature of a text-based search. It is often advantageous to select materials using criteria other than spectral similarity. Alternative criteria may include color, material class (such as lipstick), or manufacturer.
  • The need for multiple spectral displays. Displaying spectra from numerous samples simultaneously may assist in the critical comparison of composition. It is, therefore, useful to display them in a fashion that permits critical comparison.
  • The association of a sample image with the stored spectrum.
  • The need for an evaluation of the significance of association. If a questioned material and a known material are determined to be similar in elemental composition, it is important to know how significant that association is. This could be achieved by a determination of the variation of composition within that material class.

To achieve this level of information management, a true database architecture is required. The analytical advantages of a database are numerous, and although database management of spectral data has evolved with most other spectroscopic methods, including Fourier transform infrared spectroscopy and mass spectroscopy, such a capability within X-ray spectroscopy has not.

Historical Efforts

McCrone and associates made the first attempt to organize materials to include spectra and images in The Particle Atlas.1 Although limited for X-ray spectral data, it has been a valuable reference for the SEM analyst.

In 1994, a prototype database for X-ray spectra was begun at the FBI to evaluate the utility of an X-ray database and to determine the features that ultimately would be required. The integrated net peak intensities of one peak from each element detected were summed, and the individual peaks detected were ratioed to the sum. This value, representing the percentage of X-ray counts, was entered into the database. Also entered were a description of the standard, a unique number identifier, the standard preparation method, the color, the date, and the peak/background ratio of the tallest peak detected. Several hundred spectra representing a variety of materials categories were collected. These included duct tape adhesives, cosmetics, alloys, paints, and more.

Identifications were accomplished by database queries using percentage X-ray counts of each element in a questioned material. Text searches revealed compositional relationships of descriptively similar materials. Compositional searches could be made using relative peak heights if the sample was particularly small or contaminated. The database structure was satisfactory but cumbersome because data entry was manual, spectra could not be downloaded into reports, and spectra could not be interactively overlapped. The design satisfied the original analytical and comparative needs, however, and demonstrated the utility of the database concept for forensic science applications.

To develop a contemporary, full-featured utility, spectroscopy and database services were obtained through the standard procedures for federal procurement.2

Present Design

This database was designed as a Windows®-based application intended to function with a contemporary EDS system resident on a PC. Spectra are converted to the EMSA format that has emerged as an industry standard. Two functional spectral display windows are used—an analysis window and a search window. Imported spectra are processed, analyzed, and stored from the analysis window. The search window displays the spectral hits resulting from a search.

Storage of Standard Spectra and Associated Information

Upon importing, the spectrum is displayed in the upper portion of the spectrum analysis page (Figure 1). Typical qualitative utilities are available for analysis of the spectrum. In the lower left field a list of imported spectra is displayed. An image of the sample can be displayed at the bottom. The lower right field includes information regarding sample analysis and information regarding the specimen, grouped under tabs. This information (Figure 2) includes sample preparation, manufacturing details, composition, analysis parameters, laboratory details, color, material class, and additional notes. Material classes are organized in a hierarchical tree (Figure 3). Prompts and drop-down menus aid data entry.

Data Retrieval

Display of a single record includes all information stored with that entry, including spectrum, text, composition, and image (Figure 1).


Figure 1: Spectrum analysis window



Figure 1. Spectrum analysis window. Click on image for enlarged view.


Figure 2: Details of spectrum analysis window


Figure 2. Details of spectrum analysis window. Click on image for enlarged view.


Figure 3: Spectrum analysis window and material category

Figure 3. Details of spectrum analysis window and material category “tree.” Click on image for enlarged view.


Database queries can be made by selecting individual or multiple criteria, including keyword, material classification, best fit, or composition (Figure 4).

Keyword—Text searches may be made from any words in the text fields that are associated with the spectrum, including the notes page.

Material classification—All entries within the materials category are selected.

Best fit—A search may include the entire spectrum or a selected portion of it (e.g., to exclude lines of low atomic number elements). Spectra within the materials category searched are sorted according to fit (Figure 5). The time for a search is .02s per spectrum.

Composition—A relational peak search permits searches based on the presence and ratio of specific peaks or the presence of a specific element. Queries use logical operators, such as greater than or less than.

Figure 4: Search criteria options page

 Figure 5: Search results window


Figure 4. Search criteria options page. Click on image for enlarged view.


Figure 5. Search results window.
Click on image for enlarged view.

 

Fundamental Assumptions

In order for a database to function effectively, materials of similar composition and structure must produce a similar spectrum. Therefore, default analysis conditions are applied to the analysis of all materials to be included in the database. These default conditions are 25kv, a mid-value pulse processor time constant, and 100s live time with a beam current selected to yield a dead time of 30 percent. The samples are prepared in such a manner as to yield as flat an analysis surface as possible. A constant SEM-sample-detector geometry is maintained. Magnification is selected in order to provide an analysis area sufficient to produce a spectrum indicative of the average composition of the sample. Although these conditions are not necessarily optimal for each sample, they will consistently return a spectrum that reflects the average composition. As a result, any spectral differences are the result of actual compositional differences between samples.

Additional Applications

In addition to the most obvious uses of the database stated above, it may also be possible to apply this database to assessments of frequency of occurrence and compound identification.

The strength of an association can be assessed by evaluating the variation of composition within a group of similar materials. If a similarity is unique to the group, then the associative value is significant. Conversely, if the similarity is common, then the value of association is limited—that is, an association between materials that are compositionally common is less significant than one that occurs infrequently. The immediate goal is to include each compositionally distinct member within a materials category. If, however, materials are routinely included that have been randomly acquired, the ability to determine the frequency of occurrence may be considered as well. The significance of compositional variations can be determined only subsequently to the collection of vast numbers of individual items.

Attempts at compound identification appear promising. Elemental analysis can never provide definitive compound identification, as X-ray diffraction can. SEM/EDS of a compound can only yield an elemental profile. If, however, that profile is matched against the profiles of other compounds, a list of possible identifications could be produced. The success of this method would depend on the variability of peak ratios within a series of compounds consisting of the same elements. Morphological analysis might also provide for additional discrimination between compounds.

Future Development

Development of this database consists of two phases. The first will establish a stand-alone database capability resident on the SEM used at the FBI for general forensic applications. Use of this database would permit overall design evaluation, define deficiencies, and generate significant numbers of spectra within specific materials classes of interest.

The second phase will involve the development of remote access to permit data sharing between law enforcement agencies. An agency with a satellite database utility would be permitted to query the master database, as well as download and upload individual spectra. Because all X-ray detectors used to collect spectra differ in response and resolution, a channel by channel correction would be applied to permit all spectra to simulate a universal detector response.

Conclusion

With this effort, a spectral database that includes the utilities necessary to perform all of the specialized functions listed in the limitations of commercial software section above and has proven to be an effective utility for SEM in forensic science has been implemented.

The ability to query the spectra of vast numbers of standard materials has enhanced the ability of the FBI Laboratory to use SEM/EDS in a comprehensive manner. Each of the mentioned utilities has been used in casework to provide the advantage of easy to use spectral manipulation and to provide investigative direction from previously unmanageable associations.

As with any database resource, the full advantage will become more significant as the number of records increases. It is anticipated that the importance of this utility for identifications and comparisons in forensic science will continue to grow.

References

1. McCrone, W. C. and Delly, J. G. The Particle Atlas, Edition Two. Ann Arbor Science Publishers, Ann Arbor, Michigan, 1973.

2. The contract for development was awarded to xk, Inc., Phoenix, Arizona.