Chemiresistor Arrays

Overview
Sandia National Laboratories' Microsensors S&T department is developing a flexible chemical sensor "micro-lab" for detecting volatile organic compounds (VOCs) with high chemical selectivity based on large (~ 25 - 50 sensors) arrays of sorption-based resistors (chemiresistors). This µ-lab will comprise a "massively parallel" microsensor array using inexpensive, easily-fabricated polymer-coated planar interdigitated resistors; polymers are rendered conductive through admixture of conductive colloidal particles. Sensor-array resistance variations due to VOC exposure can be monitored inexpensively and in real time. The response of each sensor is rapid, reversible, and repeatable; 10s of polymers with distinct sorptive characteristics are under study. Limits of detection should reach ppb levels. Our approach is analogous to spectroscopy: the rich "spectrum" of resistor responses enables high chemical selectivity through multivariate analyses without separations; sensor redundancy provides improved system robustness.

Motivation and Operational Concept
The detection of VOCs is a key national security concern: detection of the proliferation of weapons of mass destruction (WMD). Particular combinations of chemical precursors, solvents, and byproducts signal the production of nuclear, chemical, and biological WMD. Massively parallel chemiresistor arrays are being evaluated for WMD compounds, including various multicomponent VOC mixtures. In addition, environmental monitoring and remediation applications, as well as industrial waste minimization, require the development of effective, inexpensive VOC µ-labs; key species, many of which are potential interferants for WMD detection and must therefore be included in any case, are being addressed.

Figure 1

The low cost of the chemiresistor results from its structure: two interdigitated, lithographically defined metal "combs" on a planar substrate provides a simple means to monitor thin-film resistance (Figure 1). The sensitive materials, instrumentation, and the computational hardware for PR are inexpensive as well. The potential for high sensitivity is a consequence of the ease of adapting the chemiresistor structure. Using modest lithographic technology, our sensor has one hundred interdigitated fingers separated by several- µm gaps. These structures enable measurement of very high resistivity thin films with an ordinary ohmmeter, a key to high sensitivity and low cost. High-resistivity capability also allows the use of thinner films (10s of nm), minimizing response time.
A Sandia-invented pattern-recognition technique based on human visual perception is well suited for handling arbitrary sensor responses (e.g. nonlinear and non-additive responses to mixture component concentrations). Using SAW sensor systems, we recently demonstrated the ability to distinguish and quantify small sets of VOCs and simple VOC mixtures using small sensor arrays; we also demonstrated the ability to automatically, without user threshold selection or adjustment, correctly reject (rather than misidentify) chemicals not in the calibration database. Many popular analysis techniques have difficulties with unexpected chemicals not in the database, and often incorrectly identify them as an expected species. The " massively parallel " array is the key to extending our analysis abilities to handle a wide variety of mixtures over wide concentration ranges.

Sensing films are easily obtained without sacrificing the chemical diversity requisite for effective PR: we are using thin films of many different "cataloged" polymers, inexpensive commercial materials that have been well characterized for their affinity to different classes of materials. Polymer films are rendered electrically conductive by adding a few percent by weight of dispersed conductive particles, such as colloidal silver or graphite, prior to film deposition. Swelling of the film as a consequence of the VOC sorption changes particle-particle spacing, thereby increasing film resistance.
Some applications require the measurement of lower concentrations than can be easily observed on a chemiresistor. To lower the limits of detection, we have developed an integrated preconcentrator-chemiresistor array. The preconcentrator is built on a micro-hotplate and because of the close proximity of the pair, no pumps and valves are required for the preconcentration to lower the limits of dection by 100 to 1000. Electrical timing signals determine the length of time over which the preconcentration occurs. This low power, small package should find use in first responder badges and environmental monitoring.

Results

• Approximately 50 polymers have been used to construct chemiresistors.
• We have explored colloidal conductive particle loading ranging from 5 - 60% by weight relative to the polymer.
• Six different types of conductive colloidal particles have been examined.
• We have also shown that ultrasonic agitation and the addition of surfactants and dispersants improve the conductive particle distribution within the polymer film.
• Both individual VOCs and binary mixtures have been examined.
• Integrated preconcentrator lowers limits of detection by 100 to 1000 times