A complete field-deployable microfabricated gas chromatograph (μGC) is described, and its adaptation to the analysis of low- and subparts-per-billion (ppb) concentrations of trichloroethylene (TCE) vapors in complex mixtures is demonstrated through laboratory testing. The specific application being addressed concerns the problem of indoor air contamination by TCE vapor intrusion. The μGC prototype employs a microfabricated focuser, dual microfabricated separation columns, and a microsensor array. These are interfaced to a nonmicrofabricated front-end pretrap and high-volume sampler module to reduce analysis time and limits of detection (LOD). Selective preconcentration and focusing are coupled with rapid chromatographic separation and multisensor detection for the determination of TCE in the presence of up to 45 interferences. Autonomous operation is possible via a laptop computer. Preconcentration factors as high as 500 000 are achieved. Sensitivities are constant over the range of captured TCE masses tested (i.e., 9-390 ng), and TCE is measured in a test atmosphere at 120 parts-per-trillion (ppt), with a projected LOD of 40 ppt (4.2 ng captured, 20 L sample) and a maximum sampling + analytical cycle time of 36 min. Short- and medium-term (1 month) variations in retention time, absolute responses, and response patterns are within acceptable limits.

Use of electron-beam induced crosslinking (EBIX) to pattern films of thiolate-monolayer-protected gold-nanoparticles (MPNs) on chemiresistor (CR) vapor sensors is described. MPNs with alkyl, cyanoalkyl, phenoxyalkyl, and hydroxyfluoroalkyl thiolate tail groups were patterned on integrated arrays of interdigital electrodes using electron doses of 500-750 C=cm 2 . The dc resis- tances of solvent cast films of these MPNs decrease and the baseline-normalized changes in resistance to each of five organic vapors increase to different degrees with increasing electron-beam dose. Relative responses patterns from an array of MPN-coated CR sensors for the test vapors change after EBIX patterning and the diversity of responses is diminished, on average, but it is still projected to be sufficient for the discrimination of most of the individual test vapors and binary mixtures. Results are rationalized in terms of expected changes in ligand structures and film properties following EBIX patterning using known models of electronic conduction, and vapor-induced changes of conduction, through MPN films. The implications of the results for creating arrays of densely packed MPN-coated CRs as detectors for microanalytical systems are considered.Index Terms-Chemiresistor, electron beam, nanoparticle, sensor array, vapor sensor.

Results are presented of inaugural field tests of two identical prototype microfabricated gas chromatographs (μGC) adapted for the in situ determination of trichloroethylene (TCE) in indoor air in support of vapor intrusion (VI) investigations. Each μGC prototype has a pretrap and partially selective high-volume sampler of conventional design, a micromachined-Si focuser for injection, dual micromachined-Si columns for separation, and an integrated array of four microscale chemiresistors with functionalized gold nanoparticle interface films for multichannel detection. Scrubbed ambient air is used as the carrier gas. Field-generated calibration curves were linear for injected TCE masses of 26-414 ng (4.8-77 ppb·L; r(2) > 0.98) and the projected single-sensor detection limit was 0.052 ppb for an 8-L air sample collected and analyzed in 20 min. Consistent performance between the prototypes and good medium-term stability were shown. Above the mitigation action level (MAL) of 2.3 ppb for the field-test site, μGC TCE determinations fell within ±25% of those from the reference method for 21 of 26 measurements, in the presence of up to 37 documented background VOCs. Below the MAL, positive biases were consistently observed, which are attributable to background VOCs that were unresolvable chromatographically or by analysis of the sensor-array response patterns. Results demonstrate that this type of μGC instrument could serve the need for routine TCE determinations in VI-related assessment and mitigation efforts.

The design, fabrication, and preliminary testing of a micromachined-Si passive vapor preconcentrator/injector (μPPI) are described. Intended for incorporation in a gas chromatographic microsystem (μGC) for analyzing organic vapor mixtures, the μPPI captures vapors from the air at a known rate by means of passive diffusion (i.e., without pumping) and then desorbs the vapor sample thermally by means of an integrated heater and injects it downstream (with pumping). The μPPI chip comprises a 1.8 μL deep reactive-ion-etched (DRIE) Si cavity with a resistively heated membrane floor and a DRIE-Si cap containing >1500 parallel diffusion channels, each 54 × 54 × 200 μm. The cavity is packed with 750 μg of a commercial graphitized carbon adsorbent. Fluidic and heat-transfer modeling was used to guide the design process to ensure power-efficient sample transfer during thermal desorption. Experiments performed with toluene at concentrations of ~1 ppm gave a constant sampling rate of 9.1 mL min(-1) for up to 30 min, which is within 2% of theoretical predictions and corresponds to a linear dynamic mass uptake range of ~1 μg. The cavity membrane could be heated to 250 °C in 0.23 s with 1 W of applied power and, with 50 mL min(-1) of suction flow provided by a downstream pump, yielded >95% desorption/injection efficiency of toluene samples over an 8-fold range of captured mass.

We demonstrate the use of two prototype Si-microfabricated gas chromatographs (μGC) for continuous, short-term measurements of indoor trichloroethylene (TCE) vapor concentrations related to the investigation of TCE vapor intrusion (VI) in two houses. In the first house, with documented TCE VI, temporal variations in TCE air concentrations were monitored continuously for up to 48 h near the primary VI entry location under different levels of induced differential pressure (relative to the subslab). Concentrations ranged from 0.23 to 27 ppb by volume (1.2-150 μg/m(3)), and concentration trends agreed closely with those determined from concurrent reference samples. The sensitivity and temporal resolution of the measurements were sufficiently high to detect transient fluctuations in concentration resulting from short-term changes in variables affecting the extent of VI. Spatial monitoring showed a decreasing TCE concentration gradient with increasing distance from the primary VI entry location. In the second house, with no TCE VI, spatial profiles derived from the μGC prototype data revealed an intentionally hidden source of TCE within a closet, demonstrating the capability for locating non-VI sources. Concentrations measured in this house ranged from 0.51 to 56 ppb (2.7-300 μg/m(3)), in good agreement with reference method values. This first field demonstration of μGC technology for automated, near-real-time, selective VOC monitoring at low- or subppb levels augurs well for its use in short- and long-term on-site analysis of indoor air in support of VI assessments.