C. Analytical Spectroscopy: Almost from the beginning of microwave spectroscopy, its potential for the chemical analysis of gases has been recognized (e. g. Chapter 18 of the classic monograph by Townes and Schawlow "The Use of Microwave Spectroscopy for Chemical Analysis").38 However, while techniques based on the ultraviolet, optical, and infrared regions of the electromagnetic spectrum have become standard analytical tools, this early promise has not been realized at longer wavelength. In the late '50s and early '60s a commercial instrument, operating in the general frequency region between 10 and 40 GHz was developed and marketed by Hewlett-Packard. This instrument experienced some commercial success, being sold primarily to spectroscopists with interest in molecular structure. In hind sight, its modest success with the analytical community was due to the instrument's size, cost, and complexity. However, analytical systems based on the FASSST concept described in Section III overcome these limitations. In addition, they offer significantly greater generality, speed, and sensitivity. This has been achieved by the use of (1) the Submillimeter Terahertz rather than the more classic microwave spectral region, (2) a fast scanning 'optical' technique rather than the more traditional microwave phase-lock methodology, and (3) modern data acquisition, signal processing, and computing.
In the THz spectral region fingerprints arise from the rotational energy levels of molecules. These fingerprints are ordinarily complex and unique because many rotational levels are thermally populated. Additionally, as discussed in Section II, in the THz the underlying physical interactions between radiation and matter are strong and the small Doppler broadening in the THz in comparison to the IR allows resolution of the complex rotational signature.
As an example, consider Figure V.C-1. It shows the FASSST spectra obtained with an ISTOK OB-80 BWO in an ~ 30 GHz region centered near 500 GHz that results from first adding 10mTorr of Pyrrole (C4H5N) to the sample cell (upper trace), then 20mTorr of pyridine (C5H5N) (middle trace), and finally 20mTorr of SO2 (lower trace). Figure V.C-2 shows an expansion of the ~ 1 GHz shaded region near 512 GHz in Fig. V.C-1. Finally, Fig. V.C-3 shows an expansion of the ~ 0.2 GHz shaded region near 511.8 GHz in Fig. V.C-2. The spectral region in Fig. V.C-3 is ~0.1% of the BWO bandwidth and represents ~0.01 second of data acquisition. In each figure the sensitivity is such that no noise can be displayed on the graph. It is clear even in the small expanded spectral interval displayed in the last figure that each has a unique signature, as they would in almost any other randomly chosen small interval throughout the submillimeter. One way of viewing the information content and power of FASSST as an analytical tool is to recognize that if Fig. V.C-1 were expanded to show all of its resolution elements in the horizontal and 1 mm of noise in the vertical, the resulting graph would be approximately 10 m high by 100 m long for each 1 second of data acquisition.
The technique's sensitivity, specificity, and quantitative properties can be summarized as:
1. Sensitivity: THz frequencies are typically 10 - 100 times higher than the region historically referred to as the microwave. The distinction between systems operating in the microwave and the THz regions is very important. As discussed above, molecular absorption strengths increase with frequency with a functional dependence between ν2 - ν3, peaking somewhere in the THz. As a result, spectroscopic systems in this region are between 102 and 106 more sensitive than the much more common microwave systems. Additionally, the larger spectral region results in virtually universal coverage of molecular species, subject only to the requirement that the species have a dipole moment or a large amplitude vibration in the THz. Sensitivity comparisons with infrared systems are more complex because of the diversity of infrared technologies that have been used for analytical purposes, as well as the variety of analytical applications. However, because (1) the FASSST system is based on powerful electronic oscillators which are fundamentally very bright and very quiet, (2) at long
wavelength it is possible to build very quiet detectors, (3) at long wavelengths noise due to microphonics, etc. is much reduced, and (4) rotational transition moments are related to the total electric dipole moment, the FASSST system is inherently very sensitive. 
2. Specificity and Rotational Spectroscopy: The large number of resolvable channels in the THz, the large number of thermally excited rotational lines which populate these resolvable channels, the high signal to noise attainable even with very short integrating times, and the absolute measurement of absorption coefficients available in the submillimeter make it possible in virtually all scenarios to set false alarm rates that are so low as to be considered absolutely specific. Because of the larger Doppler width in the infrared, even Doppler limited instruments are not capable of resolving the rotational fingerprints of molecules much larger than ClONO2.
3. Quantitative Analysis: Because the strengths of the molecular rotational transitions observed in the THz depend upon the permanent molecular dipole moment (which can be measured to great accuracy via the Stark Effect) and angular momentum quantum mechanics (for which 'exact' solutions are obtainable), measured fractional absorption can be translated into absolute concentration by straightforward calculation. Most importantly, because of the high resolution of the FASSST spectra, any of many isolated rotational lines can ordinarily be used for quantitative purposes; thus providing massive redundancy and eliminating the possibilities of contributions from interfering and overlapping lines.
A more detailed discussion of a THz FASSST system used as an analytical instrument can be found in an A-pages article in Analytical Chemistry.39
References
[38] C. H. Townes and A. L. Schawlow, Microwave Spectroscopy. New York: McGraw-Hill Dover Publications, Inc., 1955.
[39] S. Albert, D. T. Petkie, R. P. A. Bettens, S. P. Belov, and F. C. De Lucia, "FASSST: A new Gas-Phase Analytical Tool," Anal. Chem., vol. 70, pp. 719A-727A, 1998.