Our research is focused on the basic and applied aspects of chemical sensor science and technology in the following areas:
- The development and evaluation of nanoscale materials, e.g., nanowires and nanotubes, for high-performance chemical sensing of gas-phase species
- Novel approaches for low-cost, multi-analyte chemical monitoring technologies
- The signal processing approaches needed to identify and quantify in near real time a broad range of gas-phase chemical species in varying backgrounds
The research effort seeks to develop and integrate modular components of high-performance chemical sensors that include nanostructured sensing materials, multi-element microsensor arrays, advanced signal processing algorithms and novel sensor operational modes to yield a viable, generic and tunable sensor technology for broad application.
In materials research, we employed template-directed synthesis methods to prepare metal oxide nanotubes and nanowires. A novel sensor architecture based upon the nanotubes was prepared by using the template as a scaffold to support the nanotubes in a parallel alignment (Figure 1). Oxide nanowires, on the other hand, were released from their template and integrated with a microsensor array to compare them with a film-configured sensor as a function of sensor operating temperature and analyte. To compare the morphologically varied materials, we developed mathematical approaches involving cross-correlation techniques to visualize material- and temperature-dependent data that yield orthogonal information from chemical microsensor arrays.
Inspired by biological approaches for processing high-density data streams, we have developed a hierarchical method for acquiring and analyzing sensor data. Unlike "all-at-once" methods of signal- processing, the new approach selects optimal portions of the data to answer a series of relatively simple analytical questions that become increasingly precise (Figure 2a). We have also developed an "Event-Detection" method that employs a two-step approach using high dimensionality data to first detect whether a chemical event (such as the introduction of a trace amount of a dangerous chemical) has occurred via correlation calculations, and then apply recognition algorithms to identify the chemical, if necessary. Linear discriminant analysis (LDA) demonstrates the separability of three target analytes in untrained backgrounds (Figure 2b).
Higher sensitivity, stability, speed and reproducibility of sensing materials are critical to next-generation chemical sensing devices for gases. Furthermore, as sensors become more sophisticated and provide higher data density, advanced signal processing techniques become imperative for effectively making use of response data. Our research has yielded high sensitivity nanomaterials (nanotube assembly) used to fabricate sensors that gain sensitivity by a factor of 1000 over standard film-configured sensors. In signal processing, we have demonstrated two approaches for dealing with unknowns: analytes, which may be classified by their chemical functionality, and backgrounds, which may be ignored by using Event-Detection analyses. The synergistic use of these successes in nanomaterials, microsensor arrays, signal processing and operational modes are expected to impact many application areas, including alarm triggers for homeland security, trace gas detection in space exploration and the monitoring of gaseous, biologically generated compounds for medical diagnostics. Advanced microsensor arrays also hold the promise of enabling networked, low-cost microsensors to be used to detect and monitor low-concentration analytes in complex, dynamic backgrounds for a range of applications where the use of expensive instrumentation is not viable.