OK, so we will now discuss moos and clucks. Which is the basis for the problem with Fourier Transform Infrared Spectroscopy (FTIR) in forensic science. [2] If you close your eyes in a stock yard you can’t tell what moo came from what cow. If you close your eyes in a chicken house you can’t tell what cluck came from what chicken.

Infrared spectra can be seen to be “fingerprints” of organic compounds. However, we seldom if ever see organic compounds by themselves in forensic samples. (We have what can be described as stock yards or chicken houses full of organic cows/chickens.) What we see are complex samples the content of which is unknown. The infrared spectra are based upon absorption of infrared radiation by particular moieties (functional groups) (parts of the) molecule. But different molecules can have the same functional groups. So when we have a mixture of materials we can’t tell what molecule containing what functional group gave rise to the peak on the spectrum.

The forensic analyst in order to cut short his work load simply decides that because he sees, for instance, carbonyl groups together in a particular area of the spectrum, must have cocaine when in fact there are tens of thousands of molecules which have carbonyl groups. We don’t know what is in the alleged cocaine plus diluent mixture so we don’t know what molecules are giving rise to those carbonyl peaks. The forensic analyst ignores the problem and with a magic wave of his or her hand simply assigns the peaks based upon a request from the field to “check for the presence of controlled substances.” The protocol he is using has not been validated nor determined to be reliable. He or she seldom if ever separates the components with chromatography, leaves the data in the file, and upon questioning says he or she knows that he or she is looking at cocaine peaks based upon his or her vast knowledge and experience. Then he or she may run a GC/MS with its separation powers and detect the presence of cocaine.

However, the FTIR and GC/MS must stand alone. The FTIR gives an answer and the GC/MS tests the hypothesis. But with the FTIR data in the file the FTIR data is different from cocaine and proves the null-hypothesis presented by the GC/MS and the government data proves that cocaine is not present.

The proper way to use FTIR in a forensic setting is GC/FTIR or solvent or mechanical separation.

Another trick that the government lab is using everywhere is to present the spectra without the peaks labeled according to their absorption energy. The x-axis on the spectra goes from 400 to 4000 cm-1.  The resolution on that axis is of such poor quality that one can not discern with the naked eye what the actual absorption energies are so two spectra may look alike but can not be compared.

an analysis of an unknown by FTIR (note that the peaks are not labelled)

The way to bring that out is to ask the analyst on the stand to label each peak in the spectrum. If he or she can’t do that, which he or she can’t ever, then you ask how you compare one spectrum on paper to another.  When the spectroscopist uses a resolution of 4 cm-1 on the energy axis and a protocol that demands that peaks be within +/-4cm-1 to match then without the labeled peaks he or she can’t say he or she is within the variance required by the protocol in order to make a match.

Another issue with FTIR is that even very small differences in spectra can mean very large differences in structure of molecules. The government analyst normally ignores small differences. One can see that clearly in the size of the peaks. Two peaks together have not only location but amplitude (size). If you compare the FTIR spectrum to another the ratios of the two peaks or any number of peaks should remain the same or else the analyst must be able to explain why the ratios are the same. Peaks on the edge of larger peaks are often ignored also. These issues can be detected by carefully putting two spectra in front of you and simply comparing them for the slightest differences. You will find those differences in almost all government IR spectra comparisons.

To teach about how this is performed in the real world, one needs to present what the government presents: the spectra without the peaks labeled, and then what the government knows: the same spectra with the peaks labeled. The information concerning the energies absorbed is information the government has and yet does not present, in violation of discovery requests and Brady-like obligations. If energies are not within acceptable range then the information is Brady information when the government opines that two spectra are the same. Attorneys should be able to see how this information is hidden.

The way to label peaks is to simply push the button that says “Label Peaks” either on the face of the screen or in a set of tools in the FTIR handling algorithm. So  what about peak resolution? Do you (the resenting analyst in court) understand the difference between digital and analog data collection? It can make a huge difference  in spectral interpretation. It can mean match or no match.

A way to handle the mixture problem is to do a peak subtraction using the computer stored spectra. Suppose you think you have cocaine and caffeine and lidocaine. You can’t say you have cocaine but you think you have cocaine. So you ask that the spectra of caffeine and lidocaine be subtracted from the total spectrum to see if you get cocaine. But no one knows how much caffeine and lidocaine are present or even if they are present so when you subtract those spectra you will never get a good spectrum for cocaine. Just won’t.

The difference between an academic course and the practical application is that you learn theory in the academic courses but academia has not yet really entered into the extremely complex world of forensic samples. We are at the front lines of applying science to real world situations, finding the government underfunded, undereducated, undermanned, while trying to complete impossible tasks.  So they ignore the small stuff, which in science is always very important.