Evaluation of multitransducer arrays for the determination of organic vapor mixtures

A study of vapor recognition and quantification by polymer-coated multitransducer (MT) arrays is described. The primary data set consists of experimentally derived sensitivities for 11 organic vapors obtained from 15 microsensors comprising five cantilever, capacitor, and calorimeter devices coated with five different sorptive-polymer films. These are used in Monte Carlo simulations coupled with principal component regression models to assess expected performance. Recognition rates for individual vapors and for vapor mixtures of up to four components are estimated for single-transducer (ST) arrays of up to five sensors and MT arrays of up to 15 sensors. Recognition rates are not significantly improved by including more than five sensors in an MT array for any specific analysis, regardless of difficulty. Optimal MT arrays consistently outperform optimal ST arrays of similar size, and with judiciously selected 5-sensor MT arrays, one-third of all possible ternary vapor mixtures are reliably discriminated from their individual components and binary component mixtures, whereas none are reliably determined with any of the ST arrays. Quaternary mixtures could not be analyzed effectively with any of the arrays. A "universal" MT array consisting of eight sensors is defined, which provides the best possible performance for all analytical scenarios. Accurate quantification is predicted for correctly identified vapors.     

Limits of recognition for simple vapor mixtures determined with a microsensor array

The "limit of recognition" (LOR) has been defined as the minimum concentration at which reliable individual vapor recognition can be achieved with a multisensor array, and methodology for determining the LORs of individual vapors probabilistically on the basis of sensor array response patterns has been reported. This article explores the problems of defining and evaluating LORs for vapor mixtures in terms of the absolute and relative component vapor concentrations, where the mixture must be discriminated from those component vapors and from the subset of possible lower-order component mixtures. Monte Carlo simulations and principal components regression analyses of an extant database of calibrated responses to a set of 16 vapors from an array of 6 diverse polymer-coated surface acoustic wave sensors are used to illustrate the approach and to examine trends in LOR values among the 120 possible binary mixtures and 560 possible ternary mixtures in the data set. At concentrations exceeding the LOD, 89% of the binary mixtures could be reliably recognized (<5% error) over some composite concentration range, while only 3% of the ternary mixtures could be recognized. Most binary mixtures could be recognized only if the constituent vapor relative concentration ratio, defined in terms of multiples of the LOD for each vapor, was < or =20. Correlations with the Euclidean distance(s) separating the normalized constituent vapor response vectors allow reasonably accurate predictions of the limiting recognizable mixture composition ranges for binary and ternary cases. Results are considered in the context of using microsensor arrays for vapor detection and recognition in microanalytical systems.     

Organic vapor discrimination with chemiresistor arrays of temperature modulated tin-oxide nanowires and thiolate-monolayer-protected gold nanoparticles

This paper explores the discrimination of organic vapors with arrays of chemiresistors (CRs) employing interface layers of tin-oxide nanowires (NWs) and thiolate-monolayer-protected gold nanoparticles (MPNs). The former devices use contact-printed mats of NWs on micro-hotplate membranes to bridge a pair of metal electrodes. Oxidation at the NW surface causes changes in charge transport, the temperature dependence of which differs among different vapors, permitting vapor discrimination. The latter devices use solvent cast films of MPNs on interdigital electrodes operated at room temperature. Sorption into the organic monolayers causes changes in film tunneling resistance that differ among different vapors and MPN structures, permitting vapor discrimination. Here, we compare the performance and assess the 'complementarity' of these two types of sensors. Calibrated responses from an NW CR operated at two different temperatures and from a set of four different MPN CRs were generated for three test vapors: n-hexane, toluene, and nitromethane. This pooled data set was then analyzed using principal components regression classification models with varying degrees of random error superimposed on the responses via Monte Carlo simulation in order to estimate the rates of recognition/discrimination for arrays comprising different combinations of sensors. Results indicate that the diversity of most of the dual MPN-CR arrays exceeds that of the dual NW-CR array. Additionally, in assessing all possible arrays of 4-6 CR sensors, the recognition rates of the hybrid arrays (i.e. MPN + NW) were no better than that of the 4-sensor array containing only MPN CRs.     

Limits of recognition for binary and ternary vapor mixtures determined with multitransducer arrays

The discrimination of simple vapor mixtures from their components with polymer-coated multitransducer (MT) arrays as a function of the absolute and relative concentrations of those components is explored. The data set consists of calibrated responses to 11 organic vapors from arrays of 5 or 8 microsensors culled from a group of 5 cantilever, 5 capacitor, and 5 calorimeter transducers coated with 1 of 5 different sorptive-polymer films. Monte Carlo methods are applied to simulate error-enhanced composite responses to all possible binary and ternary mixtures of the 11 vapors, and principal component regression models are established for estimating expected rates of recognition as a function of mixture composition. The limit of recognition (LOR), defined as the maximum recognizable mixture composition range, is used as the metric of performance. With the optimal 8-sensor MT array, 19 binary and 3 ternary mixtures could be identified (i.e., discriminated from their components) with <5% error. The binary-mixture LORs are shown to decrease with increases in the baseline noise levels and random sensitivity variations of the sensors, as well as the similarity of the vapors. Importantly, most of the binary LOR contours are significantly asymmetric with respect to composition, and none of the mixtures could be recognized with <5% error at component relative concentration ratios exceeding 20:1. Discrimination of ternary mixtures from their components and binary subcomponent mixtures is possible only if the relative concentration ratio between any two of the components is <5:1. In comparing binary LORs for the best five-sensor single-transducer (ST) array to those of the best five-sensor MT array, the latter were larger in nearly all cases. The implications of these results are considered in the context of using such arrays as detectors in microanalytical systems with upstream chromatographic modules.     

Ordered‐Assembly Conductive Nanowires Array with Tunable Polymeric Structure for Specific Organic Vapor Detection

An artificial organic vapor sensor based on a finite number of 1D nanowires arrays can provide a strategy to allow classification and identification of different analytes with high efficiency, but fabricating a 1D nanowires array is challenging. Here, a coaxial Ag/polymer nanowires array is prepared as an organic vapor sensor with specific recognition, using a strategy combining superwettability‐based nanofabrication and polymeric swelling‐induced resistance change. Such organic vapor sensor containing commercial polymers can successfully classify and identify various organic vapors with good separation efficiency. An Ag/polymer nanowires array with synthetic polyethersulfone polymers is also fabricated, through molecular structure modification of the polymers, to distinguish the similar organic vapors of methanol and ethanol. Theoretical simulation results demonstrate introduction of specific molecular interaction between the designed polymers and organic vapors can improve the specific recognition performance of the sensors.     

Dual-chemiresistor GC detector employing monolayer-protected metal nanocluster interfaces

The synthesis and testing of two gold-thiolate monolayer-protected (nano)clusters as interfacial layers on a dual-chemiresistor vapor sensor array are described. Responses (changes in dc resistance) to each of 11 organic solvent vapors are rapid, reversible, and linear with concentration at low vapor concentrations, becoming sublinear at higher concentrations. Limits of detection (LODs) range from 0.1 to 24 parts per million and vary inversely with solvent vapor pressure. When configured as a GC detector and used to analyze 0.5-L preconcentrated samples of the 11-vapor mixture, the array provides LODs of < or = 700 parts per trillion for most vapors, comparing favorably with those from an integrated array of polymer-coated surface acoustic wave sensors configured and tested similarly. This first report on the use of such an array as a GC detector shows that the combination of response patterns and GC retention times improves capabilities for vapor recognition compared to the sensor array alone or to single-detector GC systems. Spray-coated nanocluster thin films can be deposited reproducibly and exhibit response stability in air that ranges from fair to excellent for up to several months. Scaling the active device area down by a factor of 16 has no significant effect on sensitivity. Implications of these results for portable vapor sensing systems are discussed.     

Hydrogen bond acidic polymers for surface acoustic wave vapor sensors and arrays

Four hydrogen bond acidic polymers are examined as sorbent layers on acoustic wave devices for the detection of basic vapors. A polysiloxane polymer with pendant hexafluoro-2-propanol groups and polymers with hexafluorobisphenol groups linked by oligosiloxane spacers yield sensors that respond more rapidly and with greater sensitivity than fluoropolyol, a material used in previous SAW sensor studies. Sensors coated with the new materials all reach 90% of full response within 6 s of the first indication of a response. Unsupervised learning techniques applied to pattern-normalized sensor array data were used to examine the spread of vapor data in feature space when the array does or does not contain hydrogen bond acidic polymers. The radial distance in degrees between pattern-normalized data points was utilized to obtain quantifiable distances that could be compared as the number and chemical diversity of the polymers in the array were varied. The hydrogen bond acidic polymers significantly increase the distances between basic vapors and nonpolar vapors when included in the array.     

Characterization of dense arrays of chemiresistor vapor sensors with submicrometer features and patterned nanoparticle interface layers

The performance of arrays of small, densely integrated chemiresistor (CR) vapor sensors with electron-beam patterned interface layers of thiolate-monolayer-protected gold nanoparticles (MPNs) is explored. Each CR in the array consists of a 100-μm(2) interdigital electrode separated from adjacent devices by 4 μm. Initial studies involved four separate arrays, each containing four CRs coated with one of four different MPNs, which were calibrated with five vapors before and after MPN-film patterning. MPNs derived from n-octanethiol (C8), 4-(phenylethynyl)-benzenethiol (DPA), 6-phenoxyhexane-1-thiol (OPH), and methyl-6-mercaptohexanoate (HME) were tested. Parallel calibrations of MPN-coated thickness-shear-mode resonators (TSMR) were used to derive partition coefficients of unpatterned films and to assess transducer-dependent factors affecting responses. A 600-μm(2) 4-CR array with four different patterned MPN interface layers, in which the MPN derived from 7-hydroxy-7,7-bis(trifluoro-methyl)heptane-1-thiol (HFA) was substituted for HME, was then characterized. This is the smallest multi-MPN array yet reported. Reductions in the diversity of the collective response patterns are observed with the patterned films, but projected vapor discrimination rates remain high. The use of such arrays as ultralow-dead-volume detectors in microscale gas chromatographic analyzers is discussed.     

The fractional free volume of the sorbed vapor in modeling the viscoelastic contribution to polymer-coated surface acoustic wave vapor sensor responses

Surface acoustic wave (SAW) vapor sensors with polymeric sorbent layers can respond to vapors on the basis of mass loading and modulus decreases of the polymer film. The modulus changes are associated with volume changes that occur as vapor is sorbed by the film. A factor based on the fractional free volume of the vapor as a liquid has been incorporated into a model for the contribution of swelling-induced modulus changes to observed SAW vapor sensor responses. In this model, it is not the entire volume added to the film by the vapor that contributes to the modulus effect; it is the fractional free volume associated with the vapor molecules that causes the modulus to decrease in a manner that is equivalent to free volume changes from thermal expansion. The amplification of the SAW vapor sensor response due to modulus effects that are predicted by this model has been compared to amplification factors determined by comparing the responses of polymer-coated SAW vapor sensors with the responses of similarly coated thickness shear mode (TSM) vapor sensors, the latter being gravimetric. Results for six to eight vapors on each of two polymers, poly(isobutylene) and poly(epichlorohydrin), were examined. The model predicts amplification factors of the order of about 1.5-3, and vapor-dependent variations in the amplification factors are related to the specific volume of the vapor as a liquid. The fractional free volume factor provides a physically meaningful addition to the model and is consistent with conventional polymer physics treatments of the effects of temperature and plasticization on polymer modulus.     

Extending the longevity of fluorescence-based sensor arrays using adaptive exposure

Fluorescent microbead sensor arrays were prepared to determine sensor array longevity. Sensor longevity is limited by photobleaching of the dyes attached to the microbeads and presents one of the biggest drawbacks of most fluorescent dye-based arrays. Responses of an array of organic vapor sensors were acquired for 2 weeks to evaluate the sensor performance over time. Photobleaching effects were overcome in two ways: (1) by limiting the excitation light power and gradually increasing the power at a rate comparable to the sensor photobleaching rates and (2) by illuminating subsections of the array through an optical slit. Both approaches extended the longevity of a sensor array. During the longevity study, the sensor arrays were employed to test their ability to correctly distinguish between responses to seven vapors. A high classification accuracy (99.8%) was obtained after 17,700 exposures for vapor responses collected over two weeks using only approximately 8% of the array's surface area.     

Differentiation of chemical components in a binary solvent vapor mixture using carbon/polymer composite-based chemiresistors

We demonstrate a "universal solvent sensor" constructed from a small array of carbon/polymer composite chemiresistors that respond to solvents spanning a wide range of Hildebrand solubility parameters. Conductive carbon particles provide electrical continuity in these composite films. When the polymer matrix absorbs solvent vapors, the composite film swells, the average separation between carbon particles increases, and an increase in film resistance results, as some of the conduction pathways are broken. The adverse effects of contact resistance at high solvent concentrations are reported. Solvent vapors including isooctane, ethanol, diisopropylmethylphosphonate (DIMP), and water are correctly identified ("classified") using three chemiresistors, their composite coatings chosen to span the full range of solubility parameters. With the same three sensors, binary mixtures of solvent vapor and water vapor are correctly classified; following classification, two sensors suffice to determine the concentrations of both vapor components. Poly(ethylenevinyl acetate) and poly(vinyl alcohol) (PVA) are two such polymers that are used to classify binary mixtures of DIMP with water vapor; the PVA/carbon particle composite films are sensitive to less than 0.25% relative humidity. The Sandia-developed visual-empirical region of influence (VERI) technique is used as a method of pattern recognition to classify the solvents and mixtures and to distinguish them from water vapor. In many cases, the response of a given composite sensing film to a binary mixture deviates significantly from the sum of the responses to the isolated vapor components at the same concentrations. While these nonlinearities pose significant difficulty for (primarily) linear methods such as principal component analysis, VERI handles both linear and nonlinear data with equal ease. In the present study, the maximum speciation accuracy is achieved by an array containing three or four sensor elements, with the addition of more sensors resulting in a measurable accuracy decrease.     

Organic vapor sensors based on SAW resonator and organic films

The majority of investigations of SAW devices used as chemical sensors are based on delay line oscillators. However, SAW resonator oscillator offers some advantages over the SAW delay line oscillator for its higher stability. In the incipient stage of fabricating gas sensors based upon SAW resonator, taking detection of organic vapor as an example, the analysis method that combines the SAW theory with organic film technology has been adopted to give an intensive insight into the responses of two-port SAW resonator coated with LB-film and cast-film after exposure to organic vapors.     

Array-to-array transfer of an artificial nose classifier.

This paper describes the use of a microsphere sensor technology that allows simple fabrication of vapor sensor arrays with reproducible response patterns. Microsphere sensor fabrication protocols are uncomplicated and yield billions of highly reproducible sensors. Microsphere sensor arrays combined with a generalized Whitney-Mann-Wilcoxen (GWMW) classifier were used to discriminate between the presence and absence of nitroaromatic compounds in high background vapor mixtures. The classifier was trained on one sensor array and then used to obtain 98.2 and 93.7% correct classification rates with data collected using two subsequent arrays made up to six months after the initial training was performed. These results represent an advance in the ability to transfer training data between multiple sensor arrays with a fluorescence-based artificial nose.     

Rapid analyte recognition in a device based on optical sensors and the olfactory system

We report here the development of a new vapor sensing device that is designed as an array of optically based chemosensors providing input to a pattern recognition system incorporating artificial neural networks. Distributed sensors providing inputs to an integrative circuit is a principle derived from studies of the vertebrate olfactory system. In the present device, primary chemosensing input is provided by an array of fiber-optic sensors. The individual fiber sensors, which are broadly yet differentially responsive, were constructed by immobilizing molecules of the fluorescent indicator dye Nile Red in polymer matrices of varying polarity, hydrophobicity, pore size, elasticity, and swelling tendency, creating unique sensing regions that interact differently with vapor molecules. The fluorescent signals obtained from each fiber sensor in response to 2-s applications of different analyte vapors have unique temporal characteristics. Using signals from the fiber array as inputs, artificial neural networks were trained to identify both single analytes and binary mixtures, as well as relative concentrations. Networks trained with integrated response data from the array or with temporal data from a single fiber made numerous errors in analyte identification across concentrations. However, when trained with temporal information from the fiber array, networks using "name" or "characteristic" output codes performed well in identifying test analytes.     

Portable gas chromatograph with tunable retention and sensor array detection for determination of complex vapor mixtures

A prototype portable gas chromatograph that combines a multiadsorbent preconcentrator/focuser, a tandem-column separation stage with individual column temperature control and junction point pressure modulation, and a detector consisting of an integrated array of polymer-coated surface acoustic wave microsensors is described. Using scheduled first-column stop-flow intervals and independent temperature programming of the two columns, it is possible to adjust the retention of eluting analyte vapors to maximize vapor recognition with the microsensor array and minimize the time of analysis. A retention window approach is combined with Monte Carlo simulations to guide retention tuning requirements and facilitate pattern recognition analyses. The determination of a 30-vapor mixture of common indoor air contaminants in < 10 min is demonstrated using ambient air as the carrier gas. Detection limits of < 10 ppb are achieved for the majority of compounds from a 1-L air sample on the basis of the most sensitive sensor in the array. Performance is assessed in the context of near-real-time indoor air quality monitoring applications.     

Surface acoustic wave gas sensor of the sorption type sensitive to the thermal properties of gases

The basic principles of a new surface acoustic wave (SAW) gas sensor are described. Being essentially a sensor of the sorption type, the proposed device possesses certain features of the thermometric SAW sensors and is not only sensitive to the vapors of volatile substances, but capable of detecting gases by their thermal properties as well. In contrast to the known thermometric SAW sensors, the proposed sensor is characterized by high temperature stability and fast response. A variant of the sensor based on a LiNbO₃ SAW delay line is described and some results of the test for detecting propane-butane mixtures are presented.     

High-speed fluorescence detection of explosives-like vapors

In this paper, we report on the preparation of novel cross-reactive optical microsensors for high-speed detection of low-level explosives and explosives-like vapors. Porous silica microspheres with an incorporated environmentally sensitive fluorescent dye are employed in high-density sensor arrays to monitor fluorescence changes during nitroaromatic compound (NAC) vapor exposure. The porous silica-based sensor materials have good adsorption characteristics, high surface areas, and surface functionality to help maximize analyte-dye interactions. These interactions occur immediately upon vapor exposure, i.e., in less than 200 ms and are monitored with a high-speed charge-coupled device camera to produce characteristic and reproducible vapor response profiles for individual sensors within an array. Employing thousands of identical microsensors permits sensor responses to be combined, which significantly reduces sensor noise and enhances detection limits. Normalized response profiles for 1,3-dinitrobenzene (1,3-DNB) are independent of analyte concentration, analyte exposure time, or sensor age for an array of one sensor type. Explosives-like NACs such as 2,4-dinitrotoluene and DNB are detected at low part-per-billion levels in seconds. Sensor-analyte profiles of some sensor types are more sensitive to low-level NAC vapor even when in a higher organic vapor background. We show that single-element arrays permit the detection of low-level nitroaromatic compound vapors because of sensor-to-sensor reproducibility and signal averaging.     

A portable, high-speed, vacuum-outlet GC vapor analyzer employing air as carrier gas and surface acoustic wave detection

Vacuum-outlet GC with atmospheric-pressure air as the carrier gas is implemented at outlet pressures up to 0.8 atm using a low-dead-volume polymer-coated surface acoustic wave (SAW) detector. Increases in the system outlet pressure from 0.1 to 0.8 atm lead to proportional increases in detector sensitivity and significant increases in column efficiency. The latter effect arises from the fact that optimal carrier gas velocities are lower in air than in more conventional carrier gases such as helium or hydrogen due to the smaller binary diffusion coefficients of vapors in air. A 12-m-long, 0.25-mm-i.d. tandem column ensemble consisting of 4.5-m dimethyl polysiloxane and 7.5-m trifluoropropylmethyl polysiloxane operated at an outlet pressure of 0.5 atm provides up to 4 x 10(4) theoretical plates and a peak capacity of 65 (resolution, 1.5) for a 3-min isothermal analysis. At 30 degrees C, mixtures of vapors ranging in vapor pressure from 8.6 to 76 Torr are separated in this time frame. The SAW detector cell has an internal volume of < 2 microL, which allows the use of higher column outlet pressures with minimal dead time. The sensor response is linear with solute mass over at least 2-3 decades and provides detection limits of 20-50 ng for the vapors tested. The combination of atmospheric-pressure air as carrier gas, modest operating pressures, and SAW sensor detection is well-suited for field instrumentation since it eliminates the need for support gases, permits smaller, low-power pumps to be used, and provides sensitivity to a wide range of vapor analytes.