Chemical warfare agent detection using MEMS-compatible ...

712

IEEE SENSORS JOURNAL, VOL. 5, NO. 4, AUGUST 2005

Chemical Warfare Agent Detection Using MEMS-Compatible Microsensor Arrays Douglas C. Meier, Charles J. Taylor, Richard E. Cavicchi, Edward White V, Michael W. Ellzy, Kenneth B. Sumpter, and Steve Semancik

Abstract—Microsensors have been fabricated consisting of TiO2 and SnO2 sensing films prepared by chemical vapor deposition (CVD) on microelectromechanical systems array platforms. Response measurements from these devices to the chemical warfare (CW) agents GA (tabun), GB (sarin), and HD (sulfur mustard) at concentrations between 5 nmol/mol (ppb) and 200 ppb in dry air, as well as to CW agent simulants CEES (chloroethyl ethyl sulfide) and DFP (diisopropyl fluorophosphate) between 250 and 3000 ppb, are reported. The microsensors exhibit excellent signal-tonoise and reproducibility. The temperature of each sensor element is independently controlled by embedded microheaters that drive both the CVD process (375 C) and sensor operation at elevated temperatures (325 C–475 C). The concentration-dependent analyte response magnitude is sensitive to conditions under which the sensing films are grown. Sensor stability studies confirm little signal degradation during 14 h of operation. Use of pulsed (200 ms) temperature-programmed sensing over a broad temperature range (20 C–480 C) enhances analyte selectivity, since the resulting signal trace patterns contain primarily kinetic information that is unique for each agent tested. Index Terms—Chemical microsensors, chemical vapor deposition (CVD), chemical warfare (CW) agent, conductometric, metal oxide, thin film.

I. INTRODUCTION

F

OR enhanced perimeter security, improved treaty verification, and protection of both civilians in public places and military personnel on the battlefield, it is desirable to have small, low-cost, low-power detectors for chemical warfare (CW) agents that are suited to widespread deployment. The most effective detectors would be highly sensitive to and selective for dangerous agents at concentrations low enough to allow the threat to be neutralized through appropriate countermeasures. The heart of such a device would be a sensor capable of recognizing the chemical agents without triggering false positives. A variety of detection methods are either in use or under study for detection of CW agents [1]. Ion mobility spectrometry uses Manuscript received February 13, 2004; revised September 17, 2004. This work was supported in part by the the Defense Threat Reduction Agency (DTRA) and in part by the National Research Council and National Institute of Standards and Technology Postdoctoral Research Associateship Program. The associate editor coordinating the review of this paper and approving it for publication was Dr. Timothy Swager. D. C. Meier, C. J. Taylor, R. E. Cavicchi, E. White V, and S. Semancik are with the Chemical Sciences and Technology Laboratory, National Institute of Standards and Technology (NIST), Gaithersburg, MD 20899-8362 USA (e-mail: [email protected]; [email protected]; [email protected]; [email protected]; [email protected]). M. W. Ellzy and K. B. Sumpter are with the U.S. Army Edgewood Chemical Biological Center (ECBC), Aberdeen Proving Ground, Aberdeen, MD 21010-5424 USA (e-mail; [email protected]; [email protected]) Digital Object Identifier 10.1109/JSEN.2005.848139

the ionization of analytes in air, which produce an electric current when they collide with a detector plate. A series of peaks is obtained; their intensities and peak positions correspond to analyte concentration and ion mobility (which is a function of composition), respectively [2]. Hand-held devices based upon ion mobility are widely used [3]. Gas chromatography (GC) separates species prior to detection and identification, which is performed using a variety of techniques. Mass spectrometry (MS), which ionizes analyte molecules, then measures the mass-tocharge ratio of the ions and ion fragments, is either used alone [4] or as the analysis component in other analytical techniques. Application of these methods for CW agent detection has been reviewed [5]. GC–MS usually provides quite accurate, although not real-time, data; it can be used to detect degradation products for site analysis and treaty verification [6]–[9]. Smaller and faster GC systems [10]–[13] are being developed with analysis times reduced to 30 s [14]. Colorimetric sensors in the form of reagent-filled tubes or coated papers indicate the presence of a target analyte by a color change. Fluorescent indicator dyes sensitive to warfare agents have also been reported [15]. While sensitive to relevant concentrations of nerve agent simulants, the indicator reaction is nonreversible and could be prone to false positives generated by contact with other, less hazardous fluorophosphonates, such as some pesticides [16]. In the area of microsensors for real-time data, polymer-based surface acoustic wave [17]–[20] and chemiresistor [21] devices have been used to test for the presence of CW simulants. Tissue-based biosensors based on photosynthetic microorganisms have been reported [14]. A Cu-catalyzed porous silicon interferometer has been shown to be sensitive to fluorophosphonate CW agents, although not reversibly [22]. Devices based upon conductance changes of metal oxides find wide application because of their sensitivity, reversibility, speed, robustness, and low cost. Tin oxide is the most widely used sensing material for this class of devices [23]–[25]. Elevated temperatures (from 200 C–500 C) are typically required to optimize the performance of these materials. Initially motivated by the goal of minimizing the power required for a device to reach these temperatures [26], [27], device research over the past decade has focused on the use of microfabrication and micromachining to make arrays of small devices with integrated heaters, thermometers and sensors [28]–[30]. A single sensor element is essentially a serpentine polysilicon resistor fabricated onto a silicon wafer; electrically conductive contacts are fabricated on top of this resistor. When the silicon surrounding it is etched away, the now-suspended resistor structure forms a “microhotplate” ( hp). This structure provides the

1530-437X/$20.00 © 2005 IEEE

MEIER et al.: CHEMICAL WARFARE AGENT DETECTION

heating required both to drive the chemical vapor deposition (CVD) process that applies a semiconducting metal–oxide (such as tin oxide) bridge between the contacts and to maintain the sensor at the operating temperatures required to sensitize these oxide films to analytes in their environment. This approach provides the additional benefits of addressable sensor film growth and sensor operation. The application of different materials on different elements and simultaneous operation of different sensor elements at different temperatures produce a more nearly orthogonal system response to different gases, overcoming the lack of chemical selectivity from individual elements operated at a single temperature [31]–[36]. The small size (device dimensions as low as 30 m have been used) allows rapid (ms time constants) heating and cooling; temperature modulation has been used to enhance the chemical selectivity of individual elements [37]. Excellent sensitivity to organic analytes has already been observed with these conductometric film sensor devices. For example, tin oxide nanoparticle films have been shown to detect methanol in dry air at concentrations as low as 10 nmol/mol [10 parts per billion (ppb)] [38]. Efficient micro-array-based studies of the process and performance relationships for nanostructured films have also been useful for identifying highly sensitive films for varied applications [39]. These previous results suggest that metal–oxide sensor arrays built upon microelectromechanical systems (MEMS) compatible platforms could provide a sensitive detection capability for CW applications. Results for CW agent monitoring measured using prototype MEMS-compatible gas microsensors are reported here. Signals were generated as changes in conductance of thin SnO and TiO films as they were exposed to the following agents and simulants: ethyl N,N-dimethylphosphoramidocyanidate (tabun, or, by its NATO military designation, GA) [40], isopropyl methylphosphonofluoridate (sarin, or GB) [41], bis(2-chloroethyl) sulfide (sulfur mustard, or HD) [42], 2-chloroethyl ethyl sulfide (CEES, an HD simulant), and diisopropyl fluorophosphate (DFP, a GB simulant), all in zero-grade dry air. Four-element prototype arrays featuring two pairs of identically prepared metal–oxide film sensors were used in the experiments. The studies include those of single arrays dedicated to each agent as well as arrays that have been sequentially exposed to each agent of our test set. We have examined the microsensor responses to the blister agent HD at concentrations as low as 16 ppb, and the nerve agents GA and GB as low as 4 and 26 ppb, respectively, all in dry air backgrounds. We present response data for both fixed-temperature sensing (FTS) and temperature-programmed sensing (TPS). The results we present show high sensitivity and good reproducibility within the pairs of identical sensing films, as well as a temperature-dependent and time-sensitive basis for agent recognition. Stability studies were also conducted in order to assess long-term sensing capabilities. From these studies, the robust characteristics of the sensor material and platform were confirmed. The results presented here help to highlight the potential advantages of this technology, including low power consumption, batch processing, and independent temperature control of individual array elements on the microsecond timescale to develop analytical content for agent

713

Fig. 1. Photograph of a typical 4-element deposition.

hp

array prior to sensing film

recognition. Most significantly, sensitivities to agent concentrations on the order of 10 ppb, which is below established LC (lethal concentration for 50% of the test population, shown later in Table II) levels, demonstrate the potential early-warning capabilities of these devices. Despite not being listed as a CW agent, we include certain CEES results in this report. These were used as a tie-point for simulant measurements made at NIST and warfare agent measurements made at the ECBC surety laboratory, provide comparisons between a simulant and an actual agent, and are included in the data set in a related signal analysis paper that demonstrates the ability of these sensors to differentiate agents from simulants [43]. We likewise include certain DFP results in order to illustrate the degree to which simulant-derived signals may vary from the warfare agents they are intended to simulate. II. EXPERIMENTAL Sensor devices for this project were produced from prototype MEMS hp sensor arrays that consist of four independently controllable and measurable device elements, the design and fabrication of which has been described previously [44]. In summary, each element consists of a suspended polysilicon resistive heater that is electrically insulated from square TiN contacts adhered to the top (Fig. 1). The top contacts allow conductance measurements of films applied to the hp surface. Applying a voltage across a heater raises the hp temperature; measuring the resulting current and subsequently computing the heater resistance provides a temperature measurement. These devices were calibrated up to 500 C prior to use in the agent and simulant studies. The arrays are mounted in 40-pin dual in-line packages (DIPs) to provide electrical connectivity to the contacts and to the heaters. This construction permits convenient device handling and measurement, as both the film growth chamber and the

714

analyte flow apparatus have been equipped with 40-pin zero-insertion-force (ZIF) sockets. These devices were coated with sensing films using CVD in a manner consistent with prior NIST hp sensor work [45], [46]. The mounted sensor arrays were inserted into a low vacuum ( milliTorr ) system equipped with glass bubblers containing volatile precursor reagents. Titanium (IV) tetraisopropoxide [39], [47], [48] and anhydrous tin (IV) nitrate [49] were the single-source precursor materials used in the deposition of titanium oxide films and tin oxide films, respectively. These precursors were chosen for both their high vapor pressures at room temperature and their thermal decomposition properties. They convert directly into their respective metal oxides; no additional oxygen source is required. Titanium (IV) tetraisopropoxide liquid was loaded into the bubbler as received. The tin (IV) nitrate solid was finely ground with a mortar and pestle inside a dry box prior to loading into a bubbler along with a magnetic stir bar; this solid precursor was stirred magnetically before each use in order to refresh the powder surface for more effective and reproducible sublimation. Both precursor bubblers were maintained at room temperature during precursor delivery. Following chamber evacuation, 10 mL/min Ar carrier gas was delivered through a bypass line until steady flow was achieved, after which the bubbler lines were opened and the bypass line closed. The precursor flow was allowed to equilibrate 5 min, after which the target hps were heated, generating self-lithographic film growth at the target hps. Films of both oxides were grown on individual sensor element surfaces by heating the hps to 375 C while simultaneously monitoring film resistance. Growth times weretypicallyfrom3to5minforSnO andTiO films,depending upon the growth rate (monitored as a function of film resistance), until SnO films measuring on the order of 10 M and TiO films on the order of 30 k were formed. Following film growth and prior to measuring CW agents and simulants, the finished devices were removed from vacuum and annealed at 400 C in air for 30 min in order to complete film oxidation and achieve a stable baseline film conductance. Each device was produced with two equivalent SnO sensor elements and two equivalent TiO sensor elements in order to examine the reproducibility of film growth procedures and sensing performance. Representative examples of SnO [Fig. 2(a)] and TiO [Fig. 2(b)] are shown; note the polycrystalline morphology of both films. Testing of completed sensors for CW agent detection was performed at the ECBC surety laboratory. The sensor chips were inserted into a 40-pin ZIF socket. The sensor and socket were then inserted into a flow cell apparatus. A viton gasket pressed between the ceramic sensor package (forming the bottom of the test cell) and a stainless steel plate (forming the top of the test cell) formed the test cell seal for the flow system. Three gas lines converged on to the mounted sensor through this top plate: the diluent line, which provided a positive pressure of zero-grade dry air; the analyte inlet line, which delivered agent and simulant materials generated by temperature-controlled analyte generators; and the exhaust line. This configuration was designed to minimize the surface area contaminated by the agents. A separate test compartment and inlet line was dedicated to each analyte tested in order to eliminate cross-contamination. Delivery


Fig. 2. (Left) Optical and (right) SEM images of (a) SnO and (b) TiO sensing films on hps.

Fig. 3. Schematic of the flow cell and the analyte delivery and dilution plumbing.

of gases to the sensor flow cell was computer controlled via a system of mass flow controllers and pressure-actuated valves. The testing sequence consisted of changing the relative flow rates through the analyte inlet and balance diluent inlet ports. During an “analyte off” cycle, the analyte inlet flow was directed through a bypass line while the balance diluent line supplied zero-grade dry air to the sample at a given flow rate. During an “analyte on” cycle, the analyte flow rate was set to the desired dilution (by adjustment of a dilution flow rate) while the diluent line was reduced to maintain equal total flow. This protocol, combined with the hardware configuration (Fig. 3), reduced the possibility of there being analyte present during the “analyte off” portion of the tests and also minimized analyte backflow into the diluent lines. Values for the analyte flow program are shown in Table I, resulting in relative concentrations of roughly 1, 2, 4, and 8 times the lowest concentration. GA, GB, HD, and CEES were placed into permeation tubes in the vapor generators and allowed to equilibrate for four weeks prior to starting the microsensor experiments at ECBC. The simulants DFP and CEES were loaded into diffusion bubblers for the tie-point experiments at NIST. Analyte quantitation and identity verification was performed using a gas chromatograph equipped with a mass selective detector and a thermal desorption system feeding into a cooled injection system held


715

TABLE I FLOW LEVEL ASSIGNED TO EACH FLOW CONTROLLER FOR EACH AGENT EXPOSURE LEVEL

TABLE II AGENT CONCENTRATIONS AT EACH AGENT EXPOSURE LEVEL. ALSO LISTED IS THE LC FOR GA [40], GB [41], AND HD [42]

at C on the inlet. Determination of the response factors for each analyte was performed by adding known quantities directly to a sorbent material, which was subsequently desorbed into the gas chromatograph. These response factors were used to find the delivery rates for the analytes, as measured at the analyte inlet into the flow cell. From the analyte inlet and diluent flow rates (Table I), analyte delivery concentrations were calculated. These concentrations are reported along with the LC [40]–[42] for each agent, where applicable (Table II). Note that all measured concentrations are well below the intermediate-term (10- to 30-min exposures) LC values for the agents tested (short-term values are even higher). Impurities were identified from their mass spectra. The only contaminants found in the samples were present at trace levels and were decomposition products that are consistent with degradation in the environment. It is notable that the contaminants resulting from agent decomposition are often also toxic, although usually not as toxic as the agents themselves. The sensors were operated using one of two operational modes: FTS and TPS. Both of these methods have been described in prior work [33], [37], [50]. During FTS, the sensor elements are maintained at a single temperature throughout the entire agent delivery program (2220 s to complete 170 s of

Fig. 4. Schematics of: (a) analyte flow level program for FTS and TPS experiments, (b) TPS temperature program, and (c) analyte flow level program for TPS experiments.

agent flow at each concentration, alternating between periods of pure diluent flow lasting 310 s) before sensor temperature is changed [Fig. 4(a)]. The TPS program used in this work alternates the sensor temperature between a ramp value (from 20 C–480 C in 20 equal steps) and a base value (200 C)

716


Fig. 5. FTS signal from 4-element sensor array including SnO and TiO sensing films for four temperatures upon exposure to GB according to the agent flow program of Fig. 4(a).

every 500 ms. In this way, 40 data points (20 collected at the ramp temperature and 20 at the base temperature) are collected for each sensor in 10 s. After 3.5 s at the base temperature, this temperature control pattern [Fig. 4(b)] is repeated throughout the analyte delivery program. For this type of experiment, two analyte delivery programs were used; the first was equivalent to the delivery program used in the FTS experiments, the second extended the analyte flow cycles to 180 s while eliminating the interceding diluent flow cycles [total program time of 1200 s, Fig. 4(c)]. For both FTS and TPS, a laptop computer was used to control the sensor temperature and measure the conductometric response of the four gas-sensor elements in the array. This was accomplished via a signal-conditioning unit that amplified the voltage and power output of the computer’s analog-to-digital converter to the levels necessary to perform hp heating for each device. The computer allowed for either FTS or TPS with separate heating profiles specified for each sensor element. Two modestly different device platforms were used to fabricate the microsensors: device type A includes hps that are 100 m across and possess an average heater resistance of 3.1 k , while the hps of device type B are 50 m across with average heater resistance of 4.8 k . While signal-conditioning units powered by 15-V power supplies were sufficient to operate the array elements of device type A, 18-V power supplies were required to operate device type B over its entire operating temperature range. Since the performance variations between these device types are largely indistinguishable, these device types were used interchangeably in these experiments. III. RESULTS AND DISCUSSION Fig. 5 shows one complete FTS cycle for on–off exposures to GB. The agent was cycled through all four concentrations at each temperature for four temperatures from 325 C–475 C. Each sensor element exhibits a fast (on the order of seconds), low-noise, positive conductometric response when the analyte is introduced to the sample. The features shown here are typical

Fig. 6. FTS signal for (a) HD and (b) CEES. Spikes in the signal traces are an electronic artifact due to range adjustments.

of the sensors in this study, which exhibit conductance changes of up to an order of magnitude when exposed to agent concentrations of up to 200 ppb. The baseline conductance of the (Siemens, or SnO films typically measure between S, while TiO film conductance is far more variS) and S through S). These able from sample to sample ( substantial variations in baseline conductance can likely be attributed to a combination of events, including microvariations in crystallite size, crack formation in identically produced sensing films, and variations in contact pad oxidation. Development of both improved fabrication process control and oxidation-resistant contact materials (such as platinum) in an effort to reduce these variations is ongoing. In the mean time, we have developed signal-processing techniques that largely compensate for this baseline variability. These techniques are examined later in this manuscript and elsewhere [43]. The sensors register a slight analyte response due to a flow artifact introduced by the opening of a pneumatic valve on the agent flow line. As this valve opens, agent present in the valve’s dead space volume is evolved, causing the response. Since the opening of this valve


717

Fig. 8. FTS comparison of GB measured using TiO sensing films of a series of thicknesses.

Fig. 7. FTS signal for (a) GA, (b) GB, and (c) DFP. Note contrast between positive conductometric response to the CW agents opposed to the negative response to the simulant.

precedes adjustment of the agent flow control by 10 s, and this signal from the noncalibrated agent concentration largely dis) sipates within that time frame, there is no significant ( interference with the signal due to the calibrated analyte flow. Two types of events are expected to cause a conductometric response in metal–oxide films: adsorption on the surface (ei-

ther to fully coordinated surface sites or defects) or chemical reaction with the surface (whether to form a stable surface compound, or to decompose via reaction with surface oxygen followed by product desorption). At the low-temperature limit, adsorption events are expected to dominate; the elevated temperature regime favors reaction and decomposition events. Events of either category must result in a modification in the position or population of electronic states in and near the conduction band in order to elicit a measurable response. For reducible metal oxides in an oxidizing atmosphere, such as the TiO and SnO used in this study, reducing analytes should produce positive conductometric responses while oxidizing analytes should produce negative conductometric signals during any reaction or decomposition events. Alternatively, adsorption events affect the conductivity of a sample based upon the band gap change or population changes in the conduction band owing to electronic interactions between the adsorbate and the substrate. These interactions are expected to be more complex, since adsorption can often result in subtle transfers of electron density between molecular orbitals and surface electronic states, as well as shifts in the energy levels of those states. In general, however, adsorbates that are electron withdrawing (increasing surface work function) should reduce conductivity while electron-donating adsorbates should increase it. For FTS traces, agent responses to different films at different temperatures can be compared using two parameters: response time and response magnitude. Response times and response magnitudes are governed concomitantly by both adsorption and reaction events at the surface; however, response magnitudes should be concentration dependent at a given temperature while response time constants should be concentration independent. Since both of these parameters are a function of the analytes’ interactions with the sensing films, both may be useful in identification and quantification. FTS data was collected from microsensor arrays dedicated to use with a single agent. Fig. 6 shows the FTS signal for a single temperature from both the blister agent HD and its simulant, CEES. The logarithmic

718


Fig. 9. FTS comparison of single concentrations of (a) HD and (b) CEES at the four temperatures for which measurements were collected. Spikes in the signal traces are electronic artifacts due to range adjustments.

Fig. 10. FTS comparisons of single concentrations of (a) GA and (b) GB at the four temperatures for which measurements were collected.

response of both the SnO and the TiO films to these analytes is positive and increases roughly exponentially as a function of the exponent of the analyte concentration (thus the use of log plots for data presentation). Note the sharp signal artifacts in several of the traces. Spikes of this kind occasionally occur due to the autoranging function of the data-acquisition software and signal-conditioning electronics, not from chemical effects. Fig. 7 shows single-temperature FTS data for the nerve agents, GA and GB, and for the GB simulant, DFP. While the response of both films to GA and GB is clearly positive with increasing analyte concentration, it is most striking that a negative signal response is observed upon sensor exposure to DFP, a polarity reversal from responses yielded by the actual nerve agents. Minor differences between molecules can result in enormous deviations in their surface interactions; clearly, DFP interactions with both oxide films are more oxidative than that of the dry air carrier. Also notable (Fig. 7) is that only the SnO films exhibit a concentration-dependent signal magnitude change. The TiO films, alternatively, exhibit a signal response magnitude that is related in part to the sensing film thickness. Fig. 8 shows results from

a study of response characteristics of TiO under GB exposure for films with a series of thicknesses characterized as a function of the CVD deposition time. Four individual TiO films were deposited on a single 4-element sensor array for times ranging from 2 to 8 min. Films grown for 4 min and above exhibit a concentration-independent signal response magnitude for the range studied, like those films shown in Fig. 7. This result suggests that the agent concentrations tested are above the necessary level to produce the maximum saturation signal for TiO films. The film grown for 2 min, however, exhibits a logarithmic response to increasing concentration, much in the same manner as SnO films. Thinner TiO films apparently possess a greater dynamic range with respect to GB concentration. The drawback of the thinner films is a decrease in film conductance to values approaching the lower limit of the signal-conditioning electronics used in this study, which results in increased noise in the measurement. Certain device geometries featuring more closely spaced film contacts (such as interdigitated comb contacts [44]) can serve to increase film conductance for equivalently produced films, thereby overcoming this obstacle. TiO films do not exhibit these saturation effects when exposed to


719

Fig. 11. TPS traces (two cycles) for all tested concentrations of GB exposure to SnO and TiO sensing films. Data collected at ramp temperatures are shown separated from those collected at the base temperature (200 C) for clarity.

HD or CEES; this property could provide a useful analytical discrimination function. The response variation with changing sensor temperature is also of interest for purposes of identification. Fig. 9 shows sections of the signals produced by blister agent HD and the simulant CEES at four different temperatures but equivalent agent concentrations (all at twice the minimum tested concentration, see Table II); Fig. 10 shows equivalent data for the nerve agents GA and GB. Temperature-dependent response characteristics are evident in the HD and CEES traces. The signal response time shortens considerably for both analytes as temperature increases. Furthermore, signal levels are reduced slightly for both analytes with increasing temperature for the TiO sensing film; on SnO , HD does not appear to exhibit such a reduction. Nerve agent response behavior does not change appreciably in either film over the sensor temperature range studied. A complete tabulation of response times and discussion of their value as analytical tools are reported elsewhere [43]. A paramount consideration for conductometric sensors of this type is the inherent nonselective equilibrium response characteristics of the oxide films (all chemicals tested in this study, as well as many others [33], [37], [50], produce a response of some magnitude). We have demonstrated that while equilibrium (or thermodynamic) responses are nonselective, the transient (or

kinetic) response due to a given analyte is distinguishable from that of another. Furthermore, a temperature change in the hotplate element results in a transient response in the sensor films’ conductance. Measurement of these transitions in an unused sensor array tested in dry air show that the response times observed for establishing an equilibrium baseline at a new temperature range from 0.5 to 2 s, depending upon the film temperature and the magnitude of the transition. We exploit this basis by employing TPS. Rapid temperature -ms time constants [33], [37]) of the hp platform control ( permits the collection of data containing mainly kinetic information. As already outlined in the experimental section, the TPS experiment sets the hotplate temperature to a ramp value between 20 C–480 C for 200 ms, then returns to the baseline temperature (200 C) for 300 ms, after which it applies the next ramp temperature [Fig. 4(b)]. Since these values are less than the response times measured using FTS [43], the data collected at each of the 20 steps in the temperature program is kinetic data, resulting in 80 nonredundant “virtual sensor” measurements in 13.5 s. The high structural specificity of kinetic information contained within that time frame provides a basis for identification. Fig. 11 shows a representative set of TPS signal traces for all four levels of analyte exposure (GB, in this case) and dry air. Each element is shown separately, as are the base and ramp tem-

720


Fig. 12. (a) GA, (b) GB, (c) HD, and (d) CEES TPS ramp/base signal ratio derivatives (two cycles). Compare the simple concentration dependence of the traces to that of the raw data of Fig. 11, which has its concentration dependence obscured by the baseline drift. This feature provides a basis for agent quantification.


perature data, for clarity. Each graph provides two complete TPS cycles of each analyte concentration, one near the beginning of the analyte exposure cycle and one near the end, to demonstrate signal pattern reproducibility. Note that while signal patterns are quite reproducible, signal magnitudes vary somewhat between TPS cycles at constant analyte concentration. This feature may largely be attributable to baseline drift; the trace patterns do not differ substantially except in position. The patterns shown in Fig. 11 (as well as similar results for the other agents) exhibit small differences as concentration is increased. These subtleties, however, are difficult to discern with the unaided eye; computer-aided analysis was used to resolve these similar patterns. Recursive training and validation of artificial neural networks (ANN) was used to both identify analytes and estimate their concentrations using the signal patterns in this TPS study [43]. Another method for clarifying these patterns is to compute the ratio of the data collected at each ramp temperature to the data collected at the corresponding base temperature. Physically, this reduced data pattern represents conductance change in the film upon sensor temperature change; visually, it makes the patterns arising from analyte exposure more obvious. Fig. 12(a)–(d) shows the time derivatives of these ramp/base data ratios for GA, GB, HD, and CEES (the derivatives more clearly show the distinguishable features in the data), which demonstrate the potential analytical value of TPS patterns. This treatment of the data eliminates the effects of baseline drift. Instead, the derivatives at the higher ramp temperatures C) decrease smoothly with increasing concentration ( for all analytes measured using the SnO sensing film, thereby simplifying concentration estimation regardless of raw data baseline position. Furthermore, the derivative trace for each agent contains features that distinguish it not only from dry air but also from equal concentrations of other agents (Fig. 13). These variations in chemical interaction between analyte and substrate provide a means by which analytes can be identified. In practice, a library of response patterns could be established, then constant comparison of sensor response traces to this CW agent library would allow the sensor to identify the presence of CW agents and estimate their concentrations. Sensor lifetime at operating temperatures has been addressed by performing FTS studies of the sensor signal during programmed exposures [Fig. 4(a)] to GB [Fig. 14(a)] and HD [Fig. 14(b)]. These measurements indicate that hp sensors are still quite sensitive with excellent signal-to-noise through 14 h of analyte measurement (insets). The sensors used to measure GB exhibit baseline drift [and in Fig. 14(a), possess the greater dynamic range indicative of thinner TiO films]; the HD sensors stabilize quite quickly following a “burn-in” period of about 1.2 h. In either case, it was found that both the GB and HD signal-to-baseline ratios exhibit far less drift than the values themselves, indicating that effective compensation for baseline drift through a combination of better sensing film processing and signal processing can be accomplished. IV. CONCLUSION Gas microsensors based upon hp sensing technology show promise as a low cost, widely deployable technology for de-

721

Fig. 13. Ramp/base signal derivatives for similar concentrations of GA, GB, and HD (approximately 27 ppb), as well as the trace obtained from dry air measurements. The differences between the traces is a basis for identification.

tecting and identifying a number of CW agents. The NIST microsensors are based on conductometric detection principles that have been used for a variety of analytes (mainly reducing gases). However, these “new generation” microsensors are fabricated on MEMS-compatible platforms using nanostructured thin films as the sensing medium. The single-analyte results reported here from microhotplate sensor arrays demonstrate stable, high signal-to-noise responses to relevant concentrations of CW agents and simulants. We have also demonstrated a novel method for agent and simulant identification based upon kinetic properties of the analyte/film interaction by monitoring the TPS temperature ramp/base temperature response ratio under constant analyte concentration. In addition, our results show that metal–oxide fabrication conditions, particularly the growth time, can have a significant impact on the dynamic range of the sensor films to certain analytes. While it was found that simulant testing of sensor technology can be valuable (for example, film growth conditions and agent flow programs were optimized for simulants prior to agent testing at a surety laboratory), no simulant testing program can replace the information derivable from studies using the genuine warfare agents. DFP, used as a nerve agent simulant, provides an example of how distinctly different simulant responses can be from those of the agents they are intended to simulate (the conductometric responses of GA/GB and DFP

722


Fig. 14. A 14-h FTS sensor signal/baseline stability test for SnO and TiO sensors at 375 C under (a) GB and (b) HD exposure. The final three FTS traces (inset) are shown to demonstrate signal-to-noise in a sensor operated for an extended time period. The four concentrations shown correspond to those in Table II.


are opposite in sign due to differences in the chemical interactions between these analytes and the oxide films). Even CEES exhibits minor, yet measurable differences in signal compared to that of HD. Investigation of simulants and warfare agents in the same study can provide important insights of relevance to both sensing mechanisms and avoidance of false positives. Simulant studies also permit a valuable assessment of potential false positive detection events. The sensors described here can differentiate between similar organophosphate compounds and chlorinated organic sulfides, even those that differ by a single chlorine atom. This feature bodes well for applications wherein detection of CW agents is desired in environments that contain similar molecules, such as pesticide or fuel vapors. Microscale devices that depend upon conductometric effects of analyte adsorption and decomposition have been demonstrated here to produce unique signals for exposure to relevant, sublethal concentrations of GA, GB, HD, DFP, and CEES. Building upon this foundation through the improvement of signal processing, production and evaluation of more chemically varied sensing films, and the application of these improvements to the challenge of differentiating CW agent signals from those produced by benign environmental interferences, is the subject of future work. ACKNOWLEDGMENT The authors would like to thank J. Melvin, M. Carrier, and C. Montgomery for the excellent technical support provided in the fabrication of test hardware and the design and preparation of microsensor devices. They would also like to thank the W. L. Gladfelter Laboratory, University of Minnesota, for the provision of the anhydrous tin nitrate precursor used to fabricate sensing films for this study. REFERENCES [1] H. H. Hill and S. J. Martin, “Conventional analytical methods for chemical warfare agents,” Pure Appl. Chem., vol. 74, no. 12, pp. 2281–2291, 2002. [2] D. C. Collins and M. L. Lee, “Developments in ion mobility spectrometry-mass spectrometry,” Anal. Bioanal. Chem., vol. 372, no. 1, pp. 66–73, 2002. [3] R. B. Turner and J. L. Brokenshire, “Hand-held ion mobility spectrometers,” Trac-Trends Anal. Chem., vol. 13, no. 7, pp. 275–280, 1994. [4] W. A. Bryden, R. C. Benson, H. W. Ko, and M. Donlon, “Universal agent sensor for counterproliferation applications,” Johns Hopkins Appl. Tech. Dig., vol. 18, no. 2, pp. 302–308, 1997. [5] C. E. Kientz, “Chromatography and mass spectrometry of chemical warfare agents, toxins and related compounds: State of the art and future prospects,” J. Chromatogr. A, vol. 814, no. 1–2, pp. 1–23, 1998. [6] G. L. Gresham, G. S. Groenewold, A. D. Appelhans, J. E. Olson, M. T. Benson, M. T. Jeffery, B. Rowland, and M. A. Weibel, “Static secondary ionization mass spectrometry and mass spectrometry/mass spectrometry (MS2) characterization of the chemical warfare agent HD on soil particle surfaces,” Int. J. Mass Spectrom., vol. 208, no. 1–3, pp. 135–145, 2001. [7] J. E. Melanson, B. L. Y. Wong, C. A. Boulet, and C. A. Lucy, “High-sensitivity determination of the degradation products of chemical warfare agents by capillary electrophoresis-indirect UV absorbance detection,” J. Chromatogr. A, vol. 920, no. 1–2, pp. 359–365, 2001. [8] A. E. F. Nassar, S. V. Lucas, W. R. Jones, and L. D. Hoffland, “Separation of chemical warfare agent degradation products by the reversal of electroosmotic flow in capillary electrophoresis,” Anal. Chem., vol. 70, no. 6, pp. 1085–1091, 1998. [9] W. E. Steiner, B. H. Clowers, L. M. Matz, W. F. Siems, and H. H. Hill, “Rapid screening of aqueous chemical warfare agent degradation products: Ambient pressure ion mobility mass spectrometry,” Anal. Chem., vol. 74, no. 17, pp. 4343–4352, 2002.

723

[10] L. S. Riter, Y. A. Peng, R. J. Noll, G. E. Patterson, T. Aggerholm, and R. G. Cooks, “Analytical performance of a miniature cylindrical ion trap mass spectrometer,” Anal. Chem., vol. 74, no. 24, pp. 6154–6162, 2002. [11] V. D. Berkout, R. J. Cotter, and D. P. Segers, “Miniaturized EI/Q/oa TOF mass spectrometer,” J. Amer. Soc. Mass Spectrom., vol. 12, no. 6, pp. 641–647, 2001. [12] H. C. Trap and J. P. Langenberg, “Semi-continuous high speed gas analysis of generated vapors of chemical warfare agents,” HRC-J. High Res. Chromatogr., vol. 22, no. 3, pp. 153–158, 1999. [13] G. Matz, T. Hunte, and W. Schroeder, “Hand-portable gas-detector array (GDA) for rapid field detection and identification of chemical threat,” Field Anal. Chem. Technol., vol. 4, no. 4, pp. 195–203, 2000. [14] C. A. Sanders, M. Rodriguez, and E. Greenbaum, “Stand-off tissuebased biosensors for the detection of chemical warfare agents using photosynthetic fluorescence induction,” Biosens. Bioelectron., vol. 16, no. 7–8, pp. 439–446, 2001. [15] S. W. Zhang and T. M. Swager, “Fluorescent detection of chemical warfare agents: Functional group specific ratiometric chemosensors,” J. Amer. Chem. Soc., vol. 125, no. 12, pp. 3420–3421, 2003. [16] J. F. Weimaster, W. T. Beaudry, P. C. Bossle, M. W. Ellzy, L. G. Janes, D. W. Johnson, J. M. Lochner, S. G. Pleva, J. H. Reeder, D. K. Rohrbaugh, T. E. Rosso, L. J. Szafraniec, L. L. Szafraniec, T. G. Albro, W. R. Creasy, J. R. Stuff, P. B. Smith, and I. R. Stewart, “Chemical analysis of environmental samples collected in Iraq—Analysis for the presence of chemical warfare agents,” J. Chem. Technol. Biotechnol., vol. 64, no. 2, pp. 115–128, 1995. [17] M. S. Nieuwenhuizen and J. L. N. Harteveld, “Studies on a surface acoustic wave (SAW) dosimeter sensor for organophosphorous nerve agents,” Sens. Actuators B, vol. 40, no. 2–3, pp. 167–173, 1997. , “A surface-acoustic-wave gas sensor for organophosphorus com[18] pounds,” Sens. Actuators B, vol. 19, no. 1–3, pp. 502–505, 1994. [19] J. W. Grate and B. M. Wise, “A method for chemometric classification of unknown vapors from the responses of an array of volume-transducing sensors,” Anal. Chem., vol. 73, no. 10, pp. 2239–2244, 2001. [20] J. W. Grate, S. J. Patrash, S. N. Kaganove, M. H. Abraham, B. M. Wise, and N. B. Gallagher, “Inverse least-squares modeling of vapor descriptors using polymer-coated surface acoustic wave sensor array responses,” Anal. Chem., vol. 73, no. 21, pp. 5247–5259, 2001. [21] A. R. Hopkins and N. S. Lewis, “Detection and classification characteristics of arrays of carbon black/organic polymer composite chemiresistive vapor detectors for the nerve agent simulants dimethylmethylphosphonate and diisopropylmethylphosponate,” Anal. Chem., vol. 73, no. 5, pp. 884–892, 2001. [22] H. Sohn, S. Letant, M. J. Sailor, and W. C. Trogler, “Detection of fluorophosphonate chemical warfare agents by catalytic hydrolysis with a porous silicon interferometer,” J. Amer. Chem. Soc., vol. 122, no. 22, pp. 5399–5400, 2000. [23] A. Balasubramanian, “Tin oxide gas sensors,” Solid State Phenomena, vol. 55, pp. 54–58, 1997. [24] N. Barsan, M. Schweizer-Berberich, and W. Gopel, “Fundamental and practical aspects in the design of nanoscaled SnO gas sensors: A status report,” Fresenius J. Anal. Chem., vol. 365, no. 4, pp. 287–304, 1999. [25] N. Barsan, R. Grigorovici, R. Ionescu, M. Motronea, and A. Vancu, “Mechanism of gas-detection in polycrystalline thick-film SnO sensors,” Thin Solid Films, vol. 171, no. 1, pp. 53–63, 1989. [26] V. Demarne and A. Grisel, “An integrated low-power thin-film CO gas sensor on silicon,” Sens. Actuators B, vol. 13, no. 4, pp. 301–313, 1988. [27] U. Dibbern, “A substrate for thin-film gas sensors in microelectronic technology,” Sens. Actuators B, vol. 2, no. 1, pp. 63–70, 1990. [28] J. S. Suehle, R. E. Cavicchi, M. Gaitan, and S. Semancik, “Tin oxide gas sensor fabricated using CMOS microhotplates and in situ processing,” IEEE Electron Device Lett., vol. 14, no. 2, pp. 118–120, Feb. 1993. [29] N. R. Swart and A. Nathan, “Design optimization of integrated microhotplates,” Sens. Actuators A, vol. 43, no. 1–3, pp. 3–10, 1994. [30] W. Y. Chung, C. H. Shim, S. D. Choi, and D. D. Lee, “Tin oxide microsensor for LPG monitoring,” Sens. Actuators B, vol. 20, no. 2–3, pp. 139–143, 1994. [31] H. E. Endres, W. Gottler, R. Hartinger, S. Drost, W. Hellmich, G. Muller, C. Bosch von Braunmuhl, A. Krenkow, C. Perego, and G. Sberveglieri, “A thin-film SnO sensor system for simultaneous detection of CO and NO with neural signal evaluation,” Sens. Actuators B, vol. 36, no. 1–3, pp. 353–357, 1996. [32] B. Panchapakesan, D. L. Devoe, M. R. Widmaier, R. Cavicchi, and S. Semancik, “Nanoparticle engineering and control of tin oxide microstructures for chemical microsensor applications,” Nanotechnol., vol. 12, no. 3, pp. 336–349, 2001.

724

[33] S. Semancik and R. Cavicchi, “Kinetically controlled chemical sensing using micromachined structures,” Accounts Chem. Res., vol. 31, no. 5, pp. 279–287, 1998. [34] C. Cane, I. Gracia, A. Gotz, L. Fonseca, E. Lora-Tamayo, M. C. Horrillo, I. Sayago, J. I. Robla, J. Rodrigo, and J. Gutierrez, “Detection of gases with arrays of micromachined tin oxide gas sensors,” Sens. Actuators B, vol. 65, no. 1–3, pp. 244–246, 2000. [35] S. Capone, P. Siciliano, N. Barsan, U. Weimar, and L. Vasanelli, “Analysis of CO and CH gas mixtures by using a micromachined sensor array,” Sens. Actuators B, vol. 78, no. 1–3, pp. 40–48, 2001. [36] Y. W. Mo, Y. Okawa, M. Tajima, T. Nakai, N. Yoshiike, and K. Natukawa, “Micro-machined gas sensor array based on metal film micro-heater,” Sens. Actuators B, vol. 79, no. 2–3, pp. 175–181, 2001. [37] R. E. Cavicchi, J. S. Suehle, K. G. Kreider, M. Gaitan, and P. Chaparala, “Fast temperature-programmed sensing for micro-hotplate gas sensors,” IEEE Electron Device Lett., vol. 16, no. 3, pp. 286–288, Mar. 1995. [38] R. E. Cavicchi, R. M. Walton, M. Aquino-Class, J. D. Allen, and B. Panchapakesan, “Spin-on nanoparticle tin oxide for microhotplate gas sensors,” Sens. Actuators B, vol. 77, no. 1–2, pp. 145–154, 2001. [39] C. J. Taylor and S. Semancik, “Use of microhotplate arrays as microdeposition substrates for materials exploration,” Chem. Mater., vol. 14, no. 4, pp. 1671–1677, 2002. [40] GA MSDS (2003). [Online]. Available: http://hazard.com/msds/tox/tf/q97/q222.html [41] GB MSDS (2003). [Online]. Available: http://hazard.com/msds/tox/tf/q96/q884.html [42] HD MSDS (2003). [Online]. Available: http://hazard.com/msds/tox/tf/q136/q562.html [43] Z. Boger, D. C. Meier, R. E. Cavicchi, and S. Semancik, “Rapid identification of chemical warfare agents by artificial neural network pruning of temperature-programmed microsensor databases,” Sens. Lett., vol. 1, no. 1, pp. 86–92, 2003. [44] S. Semancik, R. E. Cavicchi, M. C. Wheeler, J. E. Tiffany, G. E. Poirier, R. M. Walton, J. S. Suehle, B. Panchapakesan, and D. L. Devoe, “Microhotplate platforms for chemical sensor research,” Sens. Actuators B, vol. 77, no. 1–2, pp. 579–591, 2001. [45] R. E. Cavicchi, J. S. Suehle, K. G. Kreider, B. L. Shomaker, J. A. Small, M. Gaitan, and P. Chaparala, “Growth of SnO films on micromachined hotplates,” Appl. Phys. Lett., vol. 66, no. 7, pp. 812–814, 1995. [46] S. Semancik, R. E. Cavicchi, K. G. Kreider, J. S. Suehle, and P. Chaparala, “Selected-area deposition of multiple active films for conductometric microsensor arrays,” Sens. Actuators B, vol. 34, no. 1–3, pp. 209–212, 1996. [47] J. P. Lu and R. Raj, “Ultra-high vacuum chemical vapor deposition and in situ characterization of titanium oxide thin films,” J. Mater. Res., vol. 6, no. 9, pp. 1913–1918, 1991. [48] C. J. Taylor, D. C. Gilmer, D. G. Colombo, G. D. Wilk, S. A. Campbell, J. Roberts, and W. L. Gladfelter, “Does chemistry really matter in the chemical vapor deposition of titanium dioxide? Precursor and kinetic effects on the microstructure of polycrystalline films,” J. Amer. Chem. Soc., vol. 121, no. 22, pp. 5220–5229, 1999. [49] D. G. Colombo, D. C. Gilmer, V. G. Young, S. A. Campbell, and W. L. Gladfelter, “Anhydrous metal nitrates as volatile single source precursors for the CVD of metal oxide film,” Chem. Vapor Depos., vol. 4, no. 6, pp. 220–222, 1998. [50] R. E. Cavicchi, J. S. Suehle, K. G. Kreider, M. Gaitan, and P. Chaparala, “Optimized temperature-pulse sequences for the enhancement of chemically specific response patterns from micro-hotplate gas sensors,” Sens. Actuators B, vol. 33, no. 1–3, pp. 142–146, 1996.

Douglas C. Meier received the B.A. degree in chemistry from Northwestern University, Evanston, IL, and the Ph.D. degree in chemistry from Texas A&M University, College Station, where he studied the chemical physics of model catalyst systems under the guidance of Prof. D. W. Goodman. He was subsequently awarded a National Research Council Postdoctoral Research Associateship in the Process Sensing Group at the National Institute of Standards and Technology (NIST), Gaithersburg, MD. Now a NIST Research Chemist, he applies surface chemistry and thin-film science in the development of advanced chemical microsensor arrays.


Charles J. Taylor received the B.A. degree in chemistry from Macalester College, St. Paul, MN, and the Ph.D. degree in chemistry from the University of Minnesota, Minneapolis, in 1999. His graduate research focused on the role of precursor chemistry in the development of microstructures in CVD aluminum and titanium-dioxide thin films. From 2000 to 2002, he was a National Research Council Postdoctoral Associate at the National Institute of Standards and Technology (NIST), Gaithersburg, MD. He is currently an Assistant Professor of chemistry at Pomona College, Claremont, CA, and he held a NASA/ASEE Summer Faculty Fellowship while working at the Jet Propulsion Laboratory, Pasadena, CA, in 2003. His current research interests are in the field of chemical sensing and microsensor design.

Richard E. Cavicchi received the B.S. degree in physics, with a thesis on laser light scattering from colloidal crystals, from the Massachusetts Institute of Technology, Cambridge, in 1980 and the Ph.D. degree in physics, with a thesis on electron tunneling in metal nanoparticles at low temperatures, from Cornell University, Ithaca, NY, in 1987. He is a Physicist at the National Institute of Standards and Technology (NIST), Gaithersburg, MD. While a Postdoctorate at AT&T Bell Laboratories, Murray Hill, NJ, (1986 to 1988), he investigated carrier transport in quantum well devices. He joined NIST in 1989 where he has worked in the area of chemical sensors. This work includes the surface characterization of sensor interfaces, studies of sensor materials, micromachined device design, and novel sensing strategies. He is a co-inventor on several patents relating to microhotplates and temperature-programmed sensing for gas detection.

Edward White V received the B.S. degree in chemistry from the Massachusetts Institute of Technology, Cambridge, in 1964 and the Ph.D. degree in chemistry from the University of New Hampshire, Durham, in 1968. He was Robert A. Welch Postdoctoral Fellow at the Baylor College of Medicine, Houston, TX, from 1968 to 1970. In 1970, he joined the SmithKline Corporation, Philadelphia, PA, as a Senior Analytical Chemist, and, in 1976, he joined the National Institute of Standards and Technology (NIST), Gaithersburg, MD, as a Research Chemist. His work has included structure determination by mass spectrometry, high-precision quantitation of clinical analytes by mass spectrometry, and the photodissociation of ions generated by soft ionization techniques. His recent activities include the development of field methods for the detection of chemical weapons-related materials and assisting the Organization for the Prohibition of Chemical Weapons in building a database of relevant mass spectra. He is the author of 52 papers and holds one patent. Dr. White is a member of the American Association for the Advancement of Science, the American Chemical Society, the American Society for Mass Spectrometry. He served on the National Science Foundation’s Biological Instrumentation Panel from 1987 to 1991, NIST’s Advanced Technology Program Source Evaluation Board in 1990, the Toxic Substances Control Act Interagency Testing Committee from 1991 to 1997, and the Organization for the Prohibition of Chemical Weapons’ Validation Group for the Updating of the Central Analytical Database since 1996.

Michael W. Ellzy received the B.A. degree in chemistry from Temple University, Philadelphia, PA, and the M.S. degree in chemistry from Drexel University, Philadelphia. He has served as a Principal Investigator, Program Manager, and Chemist, providing the U.S. Army and Department of Defense solutions to critical problems in the area of protecting the soldier in the chemical battlefield theater for over 20 years. His current emphasis is chemical vapor generation and the array of technology and techniques used to generate trace components (parts per billion to tenths of parts per trillion) in the vapor phase. In concert with this emphasis is the application and modification of analytical instrumentation for trace analysis required to verify and validate chemical concentrations generated. Prior to the current interest, he performed chemistry investigations into analytical techniques for the detection and identification of chemical warfare agents, their degradation products, precursors, and other antagonistic chemicals.


Kenneth B. Sumpter received the B.S. degree in chemistry from the University of Maryland, College Park. He has served as a Principal Investigator and Chemist, providing the U.S. Army and Department of Defense solutions to critical problems in the area of chemical warfare agent compounds for over 15 years. His current emphasis is chemical agent environmental fate, regardless of the route of entry into the environment, utilizing an array of analytical instrumentation, chemical technology, and techniques to measure and identify the agent analyte and all degradation products. In concert with this emphasis is the development of techniques to mimic and expose agent analytes to environmental condition such compounds experience when released into the uncontrolled environment. Prior to the current interest, he performed chemistry investigations into analytical techniques for the detection and identification of chemical warfare agents, their degradation products, precursors, and other antagonistic chemicals.

725

Steve Semancik received the B.S. degree in physics from Rensselaer Polytechnic Institute, Troy, NY, in 1974 and the M.Sc. and Ph.D. degrees in physics from Brown University, Providence, RI, in 1976 and 1980, respectively. He was awarded a National Research Council Postdoctoral Associateship to do experimental studies in the Surface Science Division at the National Institute of Standards and Technology (NIST), Gaithersburg, MD. In late 1982, he joined the Process Measurements Division, NIST, as a Research Physicist and became Project Leader of NIST’s Program in Solid State Chemical Sensing. His work has included research on oxide surfaces, thin-film growth, model catalytic systems, surface structural transitions, and the kinetics of fundamental surface reactions. His recent activities have been focused on developing nanostructured materials for chemical sensing and combining such materials with micromachined structures to realize advanced microsensor devices and operating modes. He has also developed methods for using MEMS structures as novel and efficient research tools. He is an Editorial Board member for two journals, the author of approximately 115 papers, and holds five patents. Dr. Semancik is a Fellow of the American Vacuum Society.