The design space for a SMM/THz sensor is very large. Accordingly it must be bounded by some interaction of scientific and technical limits with requirement specifications. In this case we sought to have a complete sensor packaged within a 1 ft 3 (28 L) box that was able to identify the components in a mixture of more than 30 gases with very high specificity. For a portion of these gases it was desired to be able to detect them at a concentration of 100 ppt or less.
V.D.1.1. The Doppler Limit: The requirement for specificity in mixtures helped drive an important design criterion, a desire to operate well into the Doppler limit so that signal processing with narrow, well-defined lineshapes would be possible. If the system were not in the Doppler limit [1, 2], then the pressure broadening among the gases in the mixture would have made the linewidth problem intractable; the collision broadening cross sections for different pairs of gases can vary by a factor of five [1, 2]. However, in the Doppler limit, the lineshape in each gas’ reference library was the same as in the mixture. While the complexity of the 32 gas mixture considered here did not require this level of analytical power, we wanted to develop a system that could deal with as complex a mixture as possible, so we preserved this requirement.
This choice had three immediate consequences: (1) a SMM/THz technology was chosen whose spectral purity and absolute frequency calibration are much better than defined by the Doppler width, (2) because of the smaller number density, the lower pressure reduces the absorption strengths of the gases, and (3) it also causes the gases to saturate at relatively low power levels. The first of these leads to the choice of a cw electronic source technology and the latter two reduce the ppx sensitivity of the system.
An important design consequence of this decision is that we chose to use a heterodyne detection system because its sensitivity at the low power levels dictated by the low pressure, small volume cells is better than that of square law detectors, although both are limited in principle by Townes noise [1].
V.D.1.2. Gas Sample Handling and Preconcentration: The use of preconcentration is a common strategy to increase the sensitivity of gas sensors [3-5]. Indeed, this strategy is so well established that it is often part of quantitative measurement methods approved by the EPA [6]. Typically with these methods, the sorbent material is used to collect material in the field for subsequent analysis by laboratory instruments. By integrating a sorbent collection system into the spectroscopic sensor, in combination with low cell pressures, the system can realize sensitivity gains of up to five orders of magnitude over direct collection of the gas into the spectrometer cell. These sensitivity gains are due to the removal of the common atmospheric components such as nitrogen, oxygen, carbon dioxide, argon, and water, which serve only to dilute the trace gases of interest. While as defined by the EPA procedures, the sorbent process in reasonably robust against false positives that result from chemical reactions, the specificity of the sorbent based system does not have the ‘absolute’ (PFA << 10-10) specificity of rotational spectroscopy. Indeed, the specificity of any sensor is limited by reactions, desorption, etc. in its gas handling system.
The demands for high sensitivity and selectivity have a strong impact on the design of the gas handling system. Because Doppler width is proportional to frequency, the optimum sensor pressure, for which the Doppler broadening and pressure broadening are similar [1, 2], is also proportional to frequency and can be very low (10-5 – 10-6 atm) in the SMM/THz, two to three orders of magnitude lower than in the Visible/IR. Therefore, the gas sample handling system requires the use of vacuum technology to achieve ultimate pressures of no more than 1 mtorr. The choice of vacuum technology components for the gas handling system was a critical engineering decision with significant impact on the system size, power consumption, and operational timing sequence. The specifications driving vacuum equipment selection were mainly cell pressure and overall system size (1 ft3).
A particular advantage of SMM/THz systems is that a significantly smaller air sample volume is required to achieve a detectable result than systems that operate in the visible or infrared regions of the spectrum. For technical reasons, sensors in both the SMM/THz [7-10] and the Visible/IR [11, 12] often are optimized at pressures that significantly increase their linewidths beyond the Doppler limit. As a result, these sensors typically use 103 – 105 higher operating pressures than the system described here. Thus, for a cell volume of 1 L (which is typical for these systems) and a preconcentrator gain of 105, a cw electronic SMM/THz systems operating at 10-5 atm requires the processing of 1 L of atmospheric air, in comparison to the 103 – 105 L required by the sensors with higher optimal pressures.
V.D.1.3. Libraries: Because we seek to make maximal use of the information in the rotational signatures, we need reference libraries such that the noise and clutter in the spectra of the sensor spectra dominate the recovery process. While it is true that the spectrum of a species is in general redundant, analysis procedures that in effect subtract overlapping lines require knowledge of the spectra at every frequency used for the analysis of a mixture, not just at the best fingerprint regions of each analyte. Moreover, because many low lying vibrational states of even common molecules have not been analyzed and included in catalogs, spectral libraries based on quantum mechanical calculations from these constants can be massively incomplete [13]. Accordingly, we have used the sensor system described here to make experimental libraries for each of the 32 gases. While in principle intensity calibration is not necessary as long as the response function of the spectrometer does not change between recording the library spectra and the observation of the mixture, we have developed absolute calibration methods and used them both for signature library development and sample mixture analysis [14]. This, combined with the aforementioned Doppler linewidths, resulted in libraries that are useful on any similarly calibrated instrument.
[2] W. Gordy, R.L. Cook, Microwave Molecular Spectra, Third ed., John Wiley & Sons, New York, 1984.
[3] C.J. Lu, E.T. Zellers, Analyst, 127 (2002) 1061-1068.
[5] A.M. Parkes, R.E. Lindley, A.J. Orr-Ewing, Anal. Chem., 76 (2004) 7329-7335.
[7] M. van Exter, C. Fattinger, D. Grischkowsky, Opt. Lett., 14 (1989) 1128-1130.
[9] A. Majewski, Laser Focus World, (2008).
[11] M.J. Thorpe, K.D. Moll, R.J. Jones, B. Safdi, J. Ye, Science, 311 (2006) 1595-1599.
[12] S.A. Diddams, L. Hollberg, V. Mbele, Nature, 445 (2007) 627-630.
[13] S.M. Fortman, I.R. Medvedev, C.F. Neese, F.C. De Lucia, Astrophys. J., 714 (2010) 476-486.
[14] S.M. Fortman, I.R. Medvedev, C.F. Neese, F.C. De Lucia, Chem. Phys. Lett., 493 (2010) 212-215.