Fighting Counterfeiting with the Basics
The best way to spot a fake is to check its true colors. Identifying counterfeit commercial goods may be easier in the future, thanks to Effendi Widjaja and Marc V. Garland of the Singapore-based Institute of Chemical and Engineering Sciences, who have shown that band-target entropy minimization analysis of Raman microscopy spectral data reveals the constituents of an object — in their case, a postage stamp.
An analytical method helps discern the real from the fake via spectral fingerprints. The Raman signal from a spot on a stamp (black square) yields a different spectrum at each location (upper left and right), which makes authentication or determining composition a challenge. However, the basic spectral components are revealed with band-target entropy minimization analysis (lower left and right). Comparing the spectral components of an object to a known good example can determine whether it is a forgery. Courtesy of Marc V. Garland, Institute of Chemical and Engineering Sciences.
Garland, a principal scientist at the institute, noted that the analytical method does not need spectral libraries or spectral references. Instead, the authentication involves comparing the individual spectral patterns uncovered by the technique to those of a known genuine item. “If the sets of spectra are the same, then there is a basis to conclude that the test object is also genuine,” he said. “If, however, the sets are dissimilar, then there is reason to believe that the test object is not genuine.”
The algorithm has its roots in a specific research need. Years ago, Garland recalled, he and others wanted to analyze spectra collected from compounds created via complex organic synthesis. The intermediary compounds in this synthesis could not be readily isolated and, thus, their spectral characteristics could not be measured. Moreover, many likely never would have been seen before. The researchers, therefore, needed to resolve the underlying spectral patterns, hopefully producing the signature of the pure component species present.
After a great deal of effort, they developed a technique that they dubbed band-target entropy minimization analysis and the accompanying code to implement it. It is a blind source separation algorithm that searches for the simplest patterns needed to produce what is observed. Often, the spectra that it uncovers are those of the component species, even if they contribute only parts per ten thousand to the total signal.
A typical analysis of a complex reaction may yield 10 to 20 underlying patterns. As few as 100 to as many as 10,000 spectra are used as input, and the algorithm works best if there is a high degree of variation in the data.
The technique initially was applied to the group’s research focus, organometallic materials and homogeneous catalysis. “Authentication is an area far removed from our original interests,” Garland said.
He added, however, that the problem of determining fake from genuine seemed to be well suited to the power of the algorithm. As described in the Feb. 1 issue of Analytical Chemistry, the researchers employed the technique to analyze spectra acquired with Raman microscopy. They used a system from Renishaw plc of Gloucestershire, UK, equipped with a 576 × 384-pixel thermoelectrically cooled CCD array detector.
They irradiated their sample, a postage stamp, with a 785-nm near-infrared diode laser and collected the backscattered light. They performed point-by-point mapping of the Raman spectra in 2.5-μm steps in X and Y over an area that measured 100 μm on a side. The scan at each point covered the spectral window from 1052 to 1700 cm–1, and the acquisition time for each spectrum was about 15 s.
As might be expected, various spots on the stamp had varying Raman signals. This was true even when the two locations appeared visually similar and were located ∼0.1 mm apart. The researchers analyzed data from a small blue section of the stamp after first removing spikes generated by cosmic rays randomly passing by.
They recovered three pure spectral patterns, which they identified from known spectral libraries as copper phthalocyanine blue, a material similar to calcite and a yellow organic dye.
The group plans to consult with the Singapore Post authorities about these results. However, the outcome of such discussions is not important insofar as using the technique for authentication is concerned. All that is needed is the algorithm, a microscope capable of measuring the Raman spectra at various locations, a known authentic object and a test object of unknown origin.
Then, as Garland noted, the test for a forgery can be very thorough. “Since band-target entropy minimization can retrieve spectra from trace components, the issue of authenticity can be probed to a very detailed level.”
He added that a version of the algorithm with a graphical user interface is being commercialized by the research division of the Singapore government. This will be for general applications, although it also could be used for authentication work.
Contact: Marc V. Garland, Institute of Chemical and Engineering Sciences, Singapore; e-mail: email@example.com.
MORE FROM PHOTONICS MEDIA