IMPURITIES IN WATER
Infrared absorption spectroscopy has not been especially successful at measuring organic compounds dissolved in liquid water. The infrared radiation is attenuated so strongly by the liquid water that the sample of solution being studied is not allowed to be more than a few micrometers thick. At that small thickness the concentration of the solutes must be relatively high for them to be detected. Another drawback is that compounds in aqueous solution do not have much detail in their spectra. Instead of showing peaks and lines, the spectra of the solvent and solutes are broad and without much structure. This creates difficulties in the quantitative analysis, especially when there are several different solute molecules.
In the gas phase, things are very much different and sensitive measurements of complex mixtures are feasible. The infrared spectrum of water vapor is not really very intense, compared to the spectra of other molecules. In the direct analysis of vaporized water samples, the water vapor spectrum will appear even weaker than it appears in air, because the width of the spectral lines goes down nearly in proportion to the sample pressure. The absorption is then concentrated at the line centers, and regions of transmission open up between lines that normally are overlapping. There is then enough energy transmitted to reveal molecular lines and bands beneath the water spectrum. The water absorption is removed easily enough by subtraction methods.
Handling gases is somewhat more complicated than handling liquids, which may be one reason for the relative neglect of vapor phase infrared studies. The extra work, however, is not really very great. The figure here is a diagram of a gas handling apparatus. Only a small droplet of aqueous solution, a few microliters in volume is needed for a measurement. This droplet is placed in an open-ended U- tube attached to the gas handling manifold. The U-tube is then closed and the droplet is frozen with liquid nitrogen. Next the U-tube, the manifold and the gas absorption cell are all evacuated. Then the liquid nitrogen is removed and the droplet is allowed to evaporate with most of the vapors going into the long path cell. The total pressure is measured on the manometer and the spectrum is recorded. For the background spectrum, a similar sample of pure water is used.
To demonstrate the operation of the detection method, a drop of methanol and two drops of acetone were added to about 0.5 liters of tap water. This was well stirred, and then, by syringe, 0.01 milliliters of the tainted water was placed in the sample isolation U-tube. After freezing and pumping, the sample was vaporized and allowed to flow into the 6 meter absorption cell, yielding about 10 Torr. of pressure. The sample spectrum was recorded and referenced to a background spectrum of pure water vapor, with the results shown in the second figure.
From the reference spectra, the calculated partial pressures were calculated as follows: methanol, 2.1 x 10-6 atm.; acetone, 5.8 x 10- 6 atm. Since 10 Torr. of water is about 13,000 x 10-6 atm., the calculated mole fractions were 0.00016 for methanol and 0.00044 for acetone. When comparing the bands in the spectrum to the spectrum noise level, it is estimated that mole fractions 10 times smaller could have been measured.
By lengthening the optical path and further reducing the noise level in the spectrum, the sensitivity of the method may be increased to allow detection of solutes with mole fractions of 10-7 or smaller. This corresponds to about one droplet of impurity in a bathtub full of water.
Finally, it is expected that non-aqueous liquid and even solid samples may be investigated with the same sample-handling system used for aqueous solutions.