XStream Systems, Inc






EDXRD Technology in more detail

by Dr. William Mayo - Co-Founder, Veracity Network, Inc.
Chairman, Scientific Advisory Board

Energy Dispersive X-ray Diffraction (EDXRD), sometimes referred to as Coherent X-ray Scattering (CXRS), was first proposed in the 1960's as a simple and high speed way to analyze remotely the crystal structure of material within a vessel.

Among its first uses was the examination of material within a high pressure vessel or furnace. Later, the method was adapted for real-time analysis of materials being moved on high speed conveyor belt such as those at cement or chemical plants.

In the mid to late 1980's, the method was investigated for possible use as a detection tool for explosives hidden within luggage. The first work took place at Philips Medical Systems in Hamburg, Germany, where a simple, non-tomographic device was built. In 1988, Philips teamed with researchers at Rutgers University to develop the first commercial scale unit, which then became the model for the full tomographic unit produced by Yxlon Gmbh (now marketed through its parent company - Invision). Later, in 2002, the same researchers at Rutgers collaborated with L-3 Communications to develop a similar, non-tomographic diffraction-based system.

In both the Invision/Yxlon and L-3 machines, the XRD method is used as a secondary scanner to investigate suspicious regions that were previously identified by a primary instrument such as a CT scanner. By having the XRD unit look at only a single spot (or at most a few spots), the speed problem of the XRD can be reduced. In return, the XRD unit offers high detection rates with single digit false alarm rates. When combined, the CT scanner and XRD promise to provide good throughput with good detection capabilities.

All XRD instruments function by matching the diffraction pattern obtained by the instrument against a library of patterns previously obtained.

For comparison, the XRD patterns of C-4 explosive and Sucrose are, even to an untrained eye, markedly different. Although there may be a few features that are common to both patterns, they are nonetheless very different. In a testing environment, the two diffraction patterns would each be compared against the library and a best match would be found in order to identify an unknown material. Of course, the human eye does this almost instanteously. But a machine-based detection device must rely on mathematical procedures to extract a unique feature set from the pattern and use them to identify the material giving rise to that pattern.

Several methods have been developed to analyze the XRD patterns, but two seem to have become dominant:

1.) Table lookup - the positions and intensities of the prominent peaks in the pattern are located and then compared to similar data prepared from the reference materials in the library.


2.) Transform methods - a transform, such as a Fourier, Discrete Cosine, Cepstral, or similar transform is performed on the pattern to extract the key features. These are then analyzed with a neural network to ascertain a match.


One common component of all analysis methods is that the strongest peaks in the diffraction pattern tend to dominate the process, while the weaker peaks that are easily lost in the background are often ignored. While this approach discards a lot of information contained in the pattern, it does offer several advantages. Specifically, it allows us to ignore the noise and the background, which can easily obscure the weaker peaks. Also, these methods allow us to reduce the effect of absorption, which may cause the low energy end of the diffraction pattern to be preferentially absorbed while the rest of the pattern is less affected. Thus, the methods highlighted above seem to be sufficiently robust to allow us to identify the material with high accuracy.

In general, the XRD methods are highly material specific, since the diffraction patterns of crystalline materials are unique. Thus, C-4 is different from Semtex, dynamite, TNT, black powder, etc. Equally important is the fact that the patterns of the explosives are very different from the other materials in a suitcase. For example, chocolate, shoe leather, or cotton would never be interpreted as Semtex. This high selectivity of XRD has two important ramifications - high detection rates and low false alarm rates. Under ideal conditions (i.e. - low absorption and high test mass), the detection rate is close to 100% and the false alarm rate is 0.1% or lower.







2013 Veracity Network, Inc.  * Specifications, terms and pricing subject to change.