IMPURITIES IN INFRARED ACTIVE GASES
Oxygen and nitrogen are transparent in the infrared, but most other gases are not. Analysis of polluted air has therefore been a major application of infrared absorption spectroscopy. Although the water vapor in air causes some interference in measuring trace gases, subtraction of the water lines will allow most impurities to be measured down to parts-per-billion mixing ratios.
The infrared technique can also be used to measure impurities in gases other than air. This category of gas measurement is usually more difficult than measurements in air, however, because of absorption by the gas under analysis. What can be seen as an impurity in an infrared-absorbing gas depends on the relative strengths of the infrared absorption by the principal component and by the impurities. The measurements also depend on differences in the details of the spectra. It is easy to see one compound in the presence of another if the main bands of the two compounds do not overlap and if there is much detailed structure in the two spectra. It is difficult to measure one compound in the presence of another if their bands overlap and do not have much detail.
Simple molecules (diatomic, triatomic) will absorb only in restricted portions of the infrared spectrum and measurement of their impurities will be easy. In many cases the detection limits will be the same as those seen for the same impurities in air. Absence of water vapor may allow some impurities to be measured with even lower detection limits than in air because bands in the region 1800 cm-1 to 1400 cm-1 may be used more freely.
Complex molecules, like high molecular weight hydrocarbons, will absorb widely and strongly, making the impurity measurement more difficult. Reference spectra for large amounts of gas are not usually available, so each case of impurity measurement will require the preparation of reference samples of the main component.
The ability to measure impurities will be greatly enhanced if an absolutely pure sample of the principal component is available for reference. When a low-noise spectrum of the pure compound is subtracted from the spectrum of the impure compound, the bands of the impurities are left standing alone. For success, it is required that enough infrared energy penetrates the sample to allow the recording of the spectrum. 10 meter-atmospheres of air is quite transparent in the fundamental infrared region, while 10 meter-atmospheres of a compound like butane would be almost totally opaque. To measure impurities in butane one would have to insure sufficient infrared transmission. Generally, the measurement sensitivity will be greatest when the main component absorbs about two-thirds of the infrared energy in the region being considered.
Three methods of creating the needed infrared transmission come to mind: (1) shortening the optical path, (2) diluting the gas being studied with an infrared-transparent gas, such as nitrogen, and (3) reducing the total pressure of the sample while lengthening the optical path. In methods (1) and (2), one simply removes molecules from the optical path. In method (3), one can maintain the number of molecules in the optical path while the pressure reduction opens up regions of the spectrum. This happens because the spectral lines are narrower at lower pressure and absorbance shifts from the line wings to the line centers. The result is more measurement capability with no loss of sensitivity.
A ten meter optical path through one-tenth of an atmosphere of butane will give much more infrared transmission that a one meter path through one atmosphere of butane.
For most gaseous compounds encountered in industry and in environmental studies, the infrared absorption is not as strong as in the case of butane. Measurement of trace impurities is then practical, even when they are at the parts-per-million mixing ratio.
We consider an example of the measurement of trace impurities in CO2. Figure 1 shows an absorbance spectrum was made through one atmosphere of carbon dioxide in a 50 meter cell. Although the principal CO2 bands show very high absorbance in this spectrum, the spectral regions 3500 to 2400 and 2200 to 800 have relatively low absorbances and are therefore open to the measurement of trace impurities. Although the CO2 was supposed to be pure, there are two regions of absorption in its spectrum that appear to be due to impurities. First, there is the water band between 1800 and 1300. This water may have come off the cell walls. Secondly, there is a small absorption feature near 3000 which we have identified as being due to about one part-per-million of ethane. See figure 2 for a comparison of sample spectrum and an ethane reference spectrum.
Figure 3 shows the absorbance spectrum of the 50 meter-atmospheres of CO2 after a small amount of auto exhaust was mixed in. One can readily see in this spectrum the C-H bands of a number of compounds (near 3000), the overtone band of CO (near 4200) and the absorption by water (1800 to 1300). Smaller bands are not so obvious, but they are revealed when the CO2 bands are subtracted away. The spectrum of clean CO2 was used to subtract the CO2 bands from the spectrum of the polluted CO2 (figure 3). Portions of the spectrum after CO2 subtraction are seen in figures 4, 5 and 6. These figures show that we
have made PPM-level measurements of acetylene methane, hexane, NO2, NO, nitrous acid, isobutylene, propylene and ethylene.
One additional question we have addressed is whether we can measure sulfur-containing impurities in CO2. The answer is YES. SO2, COS, CS2 and mercaptans are easy to measure directly, even when at the PPM level.. H2S cannot be measured directly at the PPM level because its infrared bands are very weak. However , we can measure H2S by the PAPA technique, as discussed in our topic PHOTOLYSIS-ASSISTED POLLUTION ANALYSIS. To show this, we mixed 1.5 PPM of H2S into the pure CO2 in the 50 meter cell and then irradiated one minute with a PAPA ultraviolet lamp. We then measured about one PPM of SO2, which represented about a 70% conversion of the H2S. The SO2 band and the ozone band are shown in Figure 7.