The first illustration below is that of a typical FTIR "single-beam spectrum". We recieve thousands of these spectra, in this form, each month from the refinery. Before these spectral files can be analyzed for the presence of chemicals, each one is converted into an "absorbance spectrum" by means of a mathematical comparison to a (clean air) background reference spectrum.
Identifying and choosing appropriate background spectra is one of the more challenging tasks in this type of work, one made more difficult by the fact that these background spectral files tend to "go out of date" and need to be changed regularly. For instance, a single-beam spectrum which has been generated on a hot sunny day might not work well as a background reference for spectral files generated on a cool cloudy day.
In this type of FTIR analysis, frequency is normally referred to in terms of "wavenumber" (inverse wavelength).
The illustration below is of two infrared absorbance spectra across the entire near-infrared spectral range between about 700 and 3,500 wavenumers. The top spectral file or "trace" was generated at the Tosco North fenceline at 3:54am on 8/7/97. The bottom trace is the library reference spectrum for the chemical known as MTBE. MTBE is clearly identified in the top trace.
FTIR spectroscopy is a well-established laboratory technique used for the identification and measurement of chemical gases. The biggest challenge faced in applying this technique to "open-path" measurement is the overwhelming presence of water vapor in the open atmosphere.
Water vapor absorbs light energy throughout the infrared range, completely obscuring the signal in many spectral regions. Open-path FTIR analysis is generally limited to those regions where the effects of water vapor are relatively less prevalent. Three regions are primarily used, these are: the "fingerprint region" centered near 1,000 wavenumbers, the "C-H Stretch" centered near 2,900 wavenumbers, and a small region near 2,000 wavenumbers.
The next illustration represents a closer look at the fingerprint region for the same two spectra. Visual analysis of this type is used to confirm "hits" reported by the analytical software. Unfortunately, confirming the presence of absorbing compounds by visual analysis is not always quite so simple as in this example.
The next illustration represents what is probably the "dirtiest" infrared absorbance spectrum we have seen from this site so far. This spectrum was generated in August of '96 during the "interim phase" of the FTIR monitoring (which occured after the "6-month test period" and before the final installation). The FTIR contractor for this "interim" monitoring was Dr. Judy Zwicker with RSA of St. Louis. At that time, the refinery was still owned and operated by Unocal. In the next few illustrations, we will examine three regions of this spectral file.
This is a closer look at the fingerprint region of the same file.
Four compounds are positively identified in this region.
This is a look at the area surrounding 2,000 wavenumbers in the same file. The detection of carbonyl sulfide (at around 80ppb) came as somewhat of a surprise here, and this chemical has not been detected since this time. Carbon monoxide, on the other hand, can usually be detected and tracked at normal "background" levels. In fact, when the FTIR monitors go for extended periods without reporting fluctuations in carbon monoxide levels, this can be an indication that the instruments are not operating properly.
The C-H Stretch region of this absorbance spectrum appears to be "saturated". As you will see in the following illustrations, any number of hydrocarbon compounds display very similar features in this region. When several hydrocarbons are present simultaneously in sufficient quantities, saturation of this type can occur, and the identification of individual compounds may become difficult or impossible. In a case like this, an alert FTIR operator might notice extremely poor detection limits being reported for any compounds monitored in this region.
Note: none of the chemicals listed below are identified in this spectrum, they are included here only as examples of compounds displaying similar absorbance features in the "C-H Stretch". There are many more like this.
Methane is another atmospheric constituent which can readily be detected at normal background levels (1-3ppm) by the FTIR monitors. In this case, when the C-H stretch is saturated, we had to move way out to the "wing" regions to correctly identify methane, which is indicated to be present here at approximately 30ppm.
All of the infrared reference spectra depicted in the above illustrations come from the Phil Hanst Library available from:
1334 North Knollwood Circle
Anaheim, CA 92801
(714) 236-8900 http://fbits.com/infrared/
These guys are doing some very exciting stuff in this field. You might want to check out their website!