Capillary gas chromatography/Fourier transform infrared microspectrometry at subambient temperature

Increasing the differentiating power of chromatographic and spectroscopic techniques by interfacing them together has been a long-recognized goal of analytical chemists (1). However, the marriage of gas chromatography (GC) and Fourier transform infrared (FT-IR) spectrometry has not always been synergistic because capillary GC has often required lower detection limits than FT-IR has been able to provide (2,3). The development of the light-pipe interface has been largely responsible for the success of gas chromatography/ Fourier transform infrared spectrometry. The key component of this interface is usually a light-pipe gas cell through which the effluent from a capillary column flows continuously during the IR measurement. Typical sample quantities required to yield identifiable spectra using a light pipe with an internal diameter of 1 mm are between 5 and 25 ng, although a light-pipe interface based on the "counter-Jacquinot advantage" holds promise in reducing these identification limits by a factor of 2 or 3 (4). The need for a further reduction of the injected amount required to yield an identifiable spectrum by at least an order of magnitude has resulted in the development of alternative approaches to the light-pipe interface. Possibly the best approach to high-sensitivity GC/FT-IR has involved trapping each eluate in a matrix of argon at about 12 K (5,6). A device which is based on this principle, known as the Cryolect, was introduced commercially in 1984 (7,8). With this interface, identifiable spectra of compounds with average absorptivities can be obtained in the 200-400-pg range. Such low limits of identification have been achieved primarily because of the very small area over which the sample is deposited, the effective thickness of a given amount of a GC eluate being inversely proportional to the area the sample occupies. Let us compare the band intensities of a given amount of material measured by matrix isolation and using a light pipe. With the Cryolect, each eluate can be deposited in a spot the diameter of which is as small as 0.25 mm (8). Neglecting the effect of the low-temperature matrix on the widths and peak absorbances of spectral bands, it can be readily seen that the intensity of the bands due to a given quantity of a GC eluate trapped in an argon matrix as a 0.25-mm-diameter spot should be 16 times greater than that of bands due to the same amount of the eluate held in a 1-mm-i.d. light pipe. For samples of this size, the - 1-mm-diameter image at the sample focus of a conventional 6X beam condenser is too large and the use of an infrared microscope is indicated (9). In an earlier note (IO), we showed that it is possible to trap a GC eluate of low volatility on a stationary infrared window held at ambient temperature. When this window was transferred to an infrared microscope, spectra of deposited materials were measured with 1-ng detection limits. In this paper, we report the feasibility of constructing an interface for real-time GC/FT-IR measurements based on this principle. Eluates are trapped on an infrared-transparent plate which is either held at ambient temperature or is termoelectrically cooled to temperatures as low as -45 "C. This window is located in the focal plane of an FT-IR microscope so that each eluate passes through the beam shortly after it is deposited. Although the sensitivity of this device for GC/FT-IR is not quite as high as that of the Cryolect, detection limits appear to be approximately 1 order of magnitude lower than those of corresponding spectra measured with a light-pipe interface. The fact that this device operates near ambient conditions means that its cost should be considerably less than that of matrix-isolation GC/FT-IR interfaces