Sensors for Environmental Control

Foreword

CONTENTS

This paper deals with problems of both molecular compounds and fine particles present as pollutants in surrounding environment. Criteria classification as defined by Italian law are presented in terms of exposure limit as the relevant parameter. Also a list of detection methods and detection limits is presented. This reference may help in the sensor strategy assessment for environmental control. Finally a block diagram indicating the environmental evolution due to the possible energy transformation is here presented and discussed in order to emphasise that pollutants do represent by-products of industrial processes. Our duty is to reduce as much as possible the concentration of these pollutants. Whatever the solution taken, sensors will always play a fundamental role for monitoring purposes.

This paper discusses the Framework directive on Air Quality which has been recently enforced in Europe to set limit values and assessment methods for several important atmospheric pollutants. The Directive allows to use methods characterised by accuracy and precision less than those required by the reference methods or by equivalent methods, providing that such candidate methods are used for indicative assessment and not for mandatory measurements. Examples on methods which seem to be equivalent to the reference but which are not meeting the minimum requirements for equivalence are given. Finally, general and practical recommendations for running equivalence testing are given.

Recent European directives concerning the outdoor air quality control and monitoring in large urban areas ask for a continuous monitoring of pollutant gasses usually released by vehicle engines, such as C₆H₆, CO and NO_x. At the present time, the pollutant concentrations are evaluated by standard analytical tools, i.e. gas chromatography (GC) with PID or FID as detectors for C₆H₆, non-dispersive infrared spectroscopy for CO and chemoluminescence techniques for NO_x, respectively. The high initial investments, the high maintenance costs and the big size, make the installation of large and distributed networks of outdoor air monitoring stations impracticable.

The drastic reduction of investments and maintenance costs along with the miniaturization of the overall system size, still keeping the typical performances of std. analytical tools, is one of the most challenging issues for outdoor air monitoring equipment as well as for those applications where GC techniques are involved. In particular, since the early ’90, several innovative approaches aiming for the integration of the GC functional units in a miniaturized pneumatic motherboard have been pursued. Furthermore, a big step toward the integration and the miniaturization of the different parts has been made possible by using the Si-micromachining techniques. In this scenario, the layout of the detection unit has been extensively redesigned and sensing devices other than PID, FID or GCMS have been successfully investigated.

This paper will review the most recent advancements in the miniaturization of the major GC components, aiming to realize a new generation of small, reliable and low-cost gas monitoring system tailored for a large variety of applications. The usage of new materials suitable to adsorb and release specific gas compounds in a controllable way, new system layout integrating self calibrating units, low-power consumption detectors and preconcentration and/or separation columns will be presented. In particular, the activity carried out by the IMM-CNR - Section of Bologna for the realization of a monitoring system suitable for the detection of Benzene at sub-ppb level in urban areas will be discussed.

The development of two voltammetric sensors that allow real-time, in-situ monitoring of trace elements Cu(II), Pb(II), Cd(II), Zn(II), Mn(II), and Fe(II) in natural aquatic ecosystems is described. Both are based on agarose gel-integrated Ir(Hg) microdisk arrays either interconnected or individually addressable.

Up to now La doped Tin Oxide gas sensors have pointed out as one of the most reliable options for CO₂ detection. Over the last years, several authors have presented their headway in this direction but some clues have to be sorted out before this option will become a solid alternative. In this work, several samples with different La over SnO₂ content have been prepared to give responses to some discrepancies reflected in previous studies as well as to analyse the influence of humidity and CO in the sensors performance.

MoO₃-TiO₂ and MoO₃-WO₃ thin films were synthesised by the sol-gel process and exposed to gasoline exhaust gas. A microprocessor based system was designed and developed which allowed the comparison of the fabricated sensors with commercial sensors from Figaro (Japan), MicroChemical (Switzerland) and Capteur (U.K) companies. The laboratory tests showed that the MoO₃ based sensors possess comparable gas sensing properties. The best fabricated sensor was the MoO₃(75%)-TiO₂(25%) with a response of 74% relative to the best commercial sensor tested. This sensor had a fast response time of 27 seconds and a decay time of 17 seconds.

In₂O₃ thin films were prepared by the sol-gel process, by using an indium oxide sol obtained through a chemical complexation modified process. XRD measurements performed on the film deposited onto glass substrates show that the crystallization of the oxide takes place after heating at 300°C in air and that the films prepared in this way have a smaller grain size than the films prepared with the traditional process. The gas sensing properties of the samples towards CO, NO₂ and O₃ were also analysed. The In₂O₃ thin films showed high response to NO₂ and O₃ and no cross sensitivity with CO.

In this work, we present some preliminary results on the exploitation of porous silicon microcavities in chemical and bio-chemical sensing. The resonating microcavities are obtained by confining a low refractive index zone between two distributed Bragg reflectors. These structures are much more sensitive than single layer resonant devices. Small variations of the average refractive index of the porous silicon microcavity, due to its interaction with chemical species and biomolecular complexes, result in marked shift of the reflectivity spectrum. Experimental results carried out in vapor- and liquid-phase chemical species (acetone, isopropanol, ethanol, methanol and chlorobenzene) and DNA detection are reported.

A new generation of ion-selective chalcogenide-based membranes of non-trivial composition and stoichiometry such as:Cu_2−xSe (x=0.65), CuAgSe, Ag_2+δSe, Ag₂Se_1−xTe_x, Ag_2+δSe_1−xTe_x and Ag₂Se, have been developed and their performance as flow-injection potentiometric detectors (FIPDs) for Cu(II), Hg(II), CN⁻ and Ag(I) assessed. The basic operation principles and the major advantages of the electrochemical approach are outlined. The most important analytical results from implementing the new generation chalcogenide membranes as FIPDs and some unique features are discussed. The effect of the membrane composition factors such as stoichiometry (intrinsic defects factor) and introduction of dopant (extrinsic defects factor) on the profile of the output FI-signal has been examined on the example of Hg(II)-FIPD.

Small gas sensors on alumina substrate with different NiO thin films were fabricated and then were investigated their physical and sensing properties for application to nitrogen oxide and carbon monoxide. NiO thin films (= 150 nm) were prepared by dc reactive magnetron sputtering from a nickel metal target in an Ar + O₂ mixed atmosphere in two sputtering modes. The oxygen content in the gas mixture varied from 20 to 60 %. Only as-deposited films in the metal-sputtering mode are crystalline. Annealing in dry air gives rise to crystalline phases in all samples. During the annealing process, changes in the lattice occur. A promising sensitivity and response was found towards NO₂ concentrations in the range 1÷10 ppm at low temperature (160÷200 °C). The different NiO-based sensors showed similar gas sensing properties to NO₂. Moreover, we found that humidity improves the NO₂ sensing properties of the NiO thin films. The NiO thin films were also tested in the presence of CO in the range of 50-200 ppm at various temperatures (250-420°C). The response increases as the operating temperature rises. The maximum value of CO response (30% to 200 ppm) was obtained at an operating temperature of 420°C.

In this work the effect of additives on the catalytic properties of In₂O₃ nanopowders is investigated. Results are compared with the electrical response of the sensors fabricated by the printing of the materials on alumina substrates. For the purpose of this study, Pd and Pt were introduced in sol-gel obtained tin oxide nanopowders. Additives nominal weight concentration used was 2%. After additive introduction, nanopowders were calcined at 600°C. While catalytic activity of the tin oxide in the oxidation of CO is seen to be improved by Pt and Pd introduction, no clear additive effect is shown for CH₄ conversion. Comparison of the catalytic and electrical results is discussed and the additive role on the sensing scheme proposed.

The suitability of TiO₂ as gas sensor depends, among others, on its structural form, which can be brookite, anatase and rutile. Several processes able to hide the anatase-to-rutile transformation are described. One of them is the introduction of additives (Nb and Cr in our case). Another one is based on stabilisation process (hydrothermal treatment). Bare TiO₂, Nb-doped TiO₂, Cr-doped TiO₂ and hydrothermally treated TiO₂ have been prepared and compared in order to study the influence of stabilisation processes on the titania crystallographic structure as well as on the electrical behaviour and on the gas sensor performances.

Silver surfaces have been covered with Langmuir Blodgett (LB) films of meso-octaethylcalix[4]pyrrole (PF). Such films have been investigated as sensing material for alcohol vapors recognition by using Surface Plasmon Resonance (SPR) transduction methodology. In particular, SPR tests have been performed by exposure of PF films at fast sequential alcoholic insets in the optical cell. The results highlight that PF provide a clearly recognition of alcoholic species employed in the inset sequences. Reproducibility of measures has been also observed.

Meso-tetra(methyloctaethyl)calix[4]pyrrole (1) has been synthesised and fully characterised. Thin films of 1 have been deposited onto quartz substrates by using Langmuir-Schafer (LS) method. Preliminary sensing tests performed by using Induced Circular Dichroism have shown that, although not chiral by themselves, such films exhibit induced chirality due to reversible binding with chiral alcohol vapours.

Based on our previous experience in microelectronic device fabrication, we made a test process run to investigate the influence of the stress of the different multilayer films such as SiO₂ and Si₃N₄ as masking material which are commonly used in MEMS fabrication processes.

The stress of the films was measured by using a profilometer. We evaluated the different structures using anisotropic TMAH etching with different concentrations and temperatures to demonstrate the influence of the thin film stress on the etching rate of different crystallographic orientation. Optical microscope was used to measure both the underetching rate, the etch depth and the inclination angle. Morphology and surface roughness were studied by using a SEM.

The results of our investigation demonstrate that it exists a significative influence of the superficial masking layer stress on the etching behaviour. This has to be taken in consideration in order to get a better control in wet etching processes.

The paper is devoted to description of the use of the spin-on glass films prepared by the sol-gel method in a cycle of manufacturing of the gas sensor. Tetraethoxysilane (TEOS)-derived sols doped by diverse elements, e.g. B, P, Sb, Pt, Pd, Mn serve as precursors to form the thin glassy films formed on the surface of semi-conductor materials. Features and merits of the films obtained as diffusion sources to dope both the monocrystalline and polycrystalline materials are considered. As a result the resistive layers of polycrystalline Si and SnO₂ doped by P or Sb as well as rare earths have been prepared. After the doping the layers have been utilized as the sensor heater and gas-sensitive element. The attention has also been focused on the application of the spin-on glass films as catalytic coatings for detection of some tested gases, e.g. CO and NO₂. The use of the films doped by Pt or Pd allows of increasing the sensitivity of the gas sensor on the basic of SnO₂ to CO and CH₄ as well as decreasing the destabilizing influence of air humidity. Selective NO₂ sensitivity of the gas sensor prepared is reached due to using the spin-on glass films doped by Mn. Besides the application of the spin-on glass films at different stages of technological process of semi-conductor sensor manufacturing creates the uniform technological basis for realization of technological services.

Doped Fe₂O₃ and Ga₂O₃ thin films have been prepared by chemical methods and tested as low temperature oxygen sensors. Ce and Zn were chosen as dopants. Microstructural and electrical characterisations have been carried out in order to study the effect of the dopants on the oxygen sensitivity. Based on the results obtained, a discussion on the role played by these dopants on the O₂ sensing properties of iron and gallium oxide thin films is presented.

The microstructural and electrical characteristics of ZnFe₂O₄ thin films prepared by liquid phase deposition (LPD) have been investigated. The performance of thin film ZnFe₂O₄-based sensors (sensitivity, selectivity and response/recovery time) and the sensing mechanism in the monitoring of nitrogen dioxide (NO₂) are presented and discussed. The interferent effects of CO and water are also investigated. The results obtained indicate these sensors as promising devices for the use in car ventilation system control.

Starting from the idea that energies of adsorption, dissociation and surface reactions, involved in gas interaction with SnO₂ chemoresistive sensors, are gas-specifics, we present a new method for improving selectivity, which consist in combining the variations of two physical parameters: thermal effect ΔT and electrical resistance R.

The interaction between gas molecules and the active surface of thin organic film can generate variation in the physical properties of the layer. In this paper we report spin coated layers of four metallophthalocyanines which have been used as optochemically interactive materials together with a discussion concerning the preliminary data obtained with optical absorbance measurements in UV-VIS spectral range onto these optical sensitive sensors. Glass substrates coated with metallophthalocyanine show good sensitivity and selectivity depending both on the metal and the peripheral substituents. A certain sensitivity towards tert-butylamine, diethylammine, dibutylamine, 2-butanone and acetic acid has been found.

Gallium oxide – zinc oxide (Ga₂O₃-ZnO) metal oxide thin films have been prepared by the sol-gel process and their oxygen gas sensing performance has been investigated. These films were deposited on alumina transducers with interdigital electrodes for gas sensing measurements and on single crystal silicon substrates for microcharacterization. X-ray Photoelectron Spectroscopy showed that the actual concentrations in Ga-Zn oxide thin films differ from the nominal values in the prepared solutions as a function of annealing temperature. Furthermore, the concentration of Zn decreases with a rise of sample annealing temperature. It was found that by increasing the amount of ZnO in the thin film sensors, their operating temperature decreased as well as the base resistance. In addition, the sensors prepared with 40 at. % of Zn showed the largest response and preferred oxygen sensing at operating temperatures from 380 to 420°C.

Novel side chain substituted Polythiophenes (PTs) are employed as active layers in sensing Organic Thin Film Transistors (OTFTs). Side chains functionalities on PT backbones have been chosen ad hoc in order to confer broad specificity towards analytes bearing homologous moieties. The electrical and sensing properties towards alcohol vapors are investigated. The sensors responses are evaluated from transient source-drain current and correlations between the chemical nature of the polymer side chain and the analyte functionality are discussed.

The resistance response to a steep change in composition of the atmosphere was studied in metal oxide thin film gas sensors. Deposition temperature and the doping with impurity metals were intentionally varied for the sensor parameters could be modified. Chemical composition and the morphology of the modified sensors were determined by XPS and SEM. Time dependence of the resistance change was measured in the sensors included in an array after the steep change in the composition of the atmosphere. An influence of the surface properties on the sensitivity was interpreted by a phenomenological model in which a relationship between the response kinetics and the rates of the surface catalytic reaction was defined. Based on the response kinetics of the sensor array, features of the target smell was visualised by an original method in the two-dimensional multilayer plot. The smell of bacteria infected meat was recognised by this method.

The formation and characterization of thin biofilms on solid substrate have been studied intensely for the past decade due to the potential for many technological applications including biosensor. Various concepts to control the specific adsorption of a particular protein and to prevent the nonspecific adsorption of other biological molecules onto the different substrates has been extensively studied in order to gain an understanding of the optimal arrangement of surface-coupled receptor centers. The general aspects of bioanalytical systems based on most promising affinity sensors are discussed, especially, regarding the peculiarities of interfacial architecture for sensors based on evanescent wave phenomena.

A state-of-art in acoustic wave liquid sensing is reviewed on the basis of sensing mechanisms and wave type employed. The presence of liquid at liquid-solid interface and changes in liquid properties are detected as velocity and/or amplitude changes of the wave propagating along the interface. Proper selection of the probing wave and sensing mechanism are key points for any application. Mass loading mechanism is attractive for studying interaction between different biological substances (protein binding, etc) and for measurements of the substances’ concentrations at ng/mL level. Viscoelastic interaction mechanism is useful for measurements of the liquid viscosity, density, solidification, association and dissociation kinetics and their constants values. By use of electroacoustic mechanism dielectric permittivity, electric conductivity and ^ion concentration in liquids can be measured. Thermal mechanism is fruitful for calorimetric measurements for detecting exo/endothermal (bio)chemical reactions and obtaining such information, as temperature, density, rate, heat and kinetics of a droplet evaporation. Techniques for choosing proper sensing mechanism, advantages and drawbacks as well as perspectives of acoustic wave sensing are also discussed.

Recent developments in silicon micromachining and fabrication have led to the realization of miniature sensors and actuators, many of which are finding applications in the field of medicine. In clinics and hospitals, measurement systems using chemical and physical sensors integrated with interface electronics has become the preferred method of operation. In this paper a simple non-invasive micro system specifically aimed at evaluation of urinary pressure and flow rate is proposed.

Environmental applications of screen-printed sensors and biosensors are reported. The disposable electrochemical strips have been applied as electrochemical sensors for the detection of heavy metals, and as transducers in the development of an immunosensor for PCB and of DNA biosensors.

Thick film gas sensors have been fabricated by screen-printing technology starting from different nanostructured semiconducting oxide powders both n-type and p-type. The temporal evolution of the conductivity changes in the thick-films have been studied in comparison with the concentrations of CO, NO, NO₂, and O₃ measured by the analytical equipment. An atmospheric pollutant monitoring in an historical area through arrays of nanostructured thick film sensors is described. Quantitative responses of all arrays have been obtained by calibrating only one of them in the field and the others on the basis of the latter one.

The Automated Water Analyser Computer Supported System (AWACSS) project intends to help meet the needs of both today’s and tomorrow’s water managers. The goal of this consortium is to develop a cost-effective, on-line, water-monitoring biosensor that will measure a variety of surface water pollutants in real-time with remote control and surveillance. The instruments being developed will be based on immunochemistry technology with detection via total internal reflection fluorescence (TIRF). Water samples are incubated with pre-mixed solutions of fluorescently labelled antibodies. In addition, our wave guide surface is spotted with analyte derivatives. After the incubated samples are presented to the wave guide surface, the instruments measure a fluorescence signal that reveals the concentration of the different water contaminants. A previous EU-funded project entitled, “River ANAlyser (RIANA)” successfully utilised similar technology. AWACSS is building on the achievements of the RIANA device with major improvements in three critical areas: 1) expanded multi-analyte analysis capability allowing for up to 30 simultaneous analyte measurements, 2) novel design approaches to the optical detection and fluidics including miniaturized integrated optics and microfluidics, and 3) intelligent remote surveillance and control that will allow for unattended continuous monitoring. Immunochemical reagents are being developed for simultaneous measurement of a host of small organic water pollutants, such as, Simazine, Atrazine, Carbofuran, Alachlor, and Isoproturon.

An electronic nose (Cyranose 320) and a preconcentration system (Enrichment Desorption Unit) have been used in field conditions for the detection of fugitive odours. Two sets of field trials were undertaken: at a municipal wastewater treatment plant and at a petrochemicals additive plant. Results from the waste treatment site were very promising; not only was the electronic nose able to distinguish between two different odorous areas of the site, but also, using these data, a downwind odour was classified. Success was limited on the other trial and a number of factors causing difficulty are discussed.

The air pollution in urban and industrial environments represents one of the main concerns of today’s developed societies. Densely populated areas or areas in which high concentrations of pollutants are likely to be released can profit from the implementation of dense monitoring networks. The ability of providing real-time data through easily accessible channels of information such as internet web sites, can re-establish people’s confidence in the air they breathe. The main obstacle to the implementation of this kind of systems is the huge cost of monitoring stations. A significant part of this price comes from the high cost of the analytical equipment for pollutant substances. This work introduces new approaches to the monitoring of pollution, such as the employment of an electronic nose with a baseline correction through the use of an activated charcoal filter. Other solutions are proposed for the problem of long-term drift. An accurate mobile monitoring station can be integrated with the network in such a way that, a dynamic calibration of static monitoring stations is performed. With this kind of equipment, a cheaper solution for the implementation of high-density networks is presented.

In this paper an innovative electronic nose, based on Quartz Crystal Microbalances (QCMs), is presented. The system comprises a dedicated fully digital front-end hardware that measures the QCM resonance frequency shift using a technique proposed by Cantoni [1]. This technique allows reducing the measurement time while maintaining a high frequency resolution. The developed system allows measuring variations of the sensor resonance frequency shifts during chemical transients obtained with abrupt changes in gas concentrations. Hence the reaction kinetics can be exploited to differentiate among different compounds. In this paper the application of the developed electronic nose to the detection of CO in presence of ethanol is presented. The results obtained by an array of 4 sensors with different sensitive layers are presented. An exponential fitting of the transient responses is used for feature extraction.

In this paper an experimental apparatus is described for automatic measurement of Cr(VI) concentration in water. Measuring section is a miniaturized fiber optic probe, whose working principle is light absorption of the complex Chromium (VI) – Diphenylcarbazide. Device design and set up guidelines of an automatic measuring device are described in order to get a reliable, effective and user friendly configuration and to reduce measurement uncertainty depending on interfering quantities of different type, fluid dynamic, optical, environmental, taking into account “in situ” measuring conditions.

Experimental results are discussed with reference to both manual and automatic working conditions in order to get information on performance which can be achieved in automatic configuration and useful suggestion for system improvement.

Preliminary experimental results are encouraging with the purpose of realizing a low cost device prototype for “in situ” operation.

The researches of new ISFET constructions require repeating of many electrochemical measurements. Manual preparation of test samples, setting bath temperatures and performing ISFET characteristic measurements is time consuming. The paper presents new fully automated measurement stand for ISFET sensor evaluation. The measurement head of ISFET back side contats matrix for water pollution monitoring is briefly described. The construction and organization of stand measurement and control system are shown. Finally the stand measurement data are reported.

The use of phthalocyaninc gas sensors for urban pollution monitoring and a methodology of measurements devoted to the detection of oxidising gases in low concentrations in air are related. A thin film of copper phthalocyanine, controlled in temperature, constitute the sensitive layer and is submitted to various concentrations of ozone or nitrogen dioxide in the range of those present in atmosphere. The long response time, due to the diffusion of gaseous molecules in the layer, the difficult desorption of this material and the lack of stability, are the main problems to have a good pollutants evaluation in real time. To improve these characteristics of detection, a technique using the adsorption at low temperature and then, desorption at higher temperature, is developed for these sensors. Variations of conductivity are recorded during the time of adsorption and compared to gas concentrations. Results of laboratory experiments made with ozone and nitrogen dioxide show the good correlations between conductivity variations of the sensitive layer and gaseous concentrations present near the sensor. The reproducibility is also improved. The study of kinetics of reaction between ozone and nitrogen dioxide is also investigated.

An amperometric dissolved oxygen microsensor, for integration into a flow-through environmental monitoring system, has been developed. The sensor is comprised of a working microelectrode array, an Ag|AgCl reference electrode and a nickel thermal resistor for temperature compensation. The effect of varying the individual microelectrode radius and the electrodeposition of platinum black were investigated in order to improve the signal-to-noise ratio (SNR) of the sensor. For a densely packed platinum black microelectrode array, a SNR of 61dB was obtained.

Although manufacturers of portable electronic noses often list ‘environmental monitoring’ as a typical application, there are significant difficulties in such work. Convincing published evidence of success is hard to find. A research project at the University’s QUESTOR Centre is focussed on the detection and identification of fugitive odours from a variety of industrial, waste water treatment and agricultural sources. A computer model has been developed which can predict the likelihood of humans detecting an odour downwind of some source. The program has been used to investigate the difficulties, and to develop protocols for the collection and analysis of samples. Issues such as the size and shape of the source, its elevation above ground level, wind speed, wind direction, atmospheric turbulence, humidity and rainfall all affect the probability of detection. Because portable electronic noses are less sensitive than the human nose, it has been necessary to use a preconcentrating device. This also has implications for sampling protocols and for the interpretation of the analyses. The work has been extended to produce an outline specification for an ‘ideal’ portable electronic nose for the detection of fugitive emissions.

A coupled Flow Injection-Quartz Crystal Microbalance (FI-QCM) system has been developed and validity of Sauerbrey and Kanazawa-Gordon equations has been verified in this case. The observed increase of ca. 1 order of magnitude for mass sensitivity has been attributed to the peculiar hydrodynamics of the flow cell. Langmuir-Blodgett films made of phthalocyanines have been grown as recognition elements on crystals. Preliminary results on such FI-QCM sensors for total phenolics in water are promising.

Initial investigation has shown that an electronic nose can be used for localizing gas sources in a similar way as the biological nose can. The electronic nose, however, can detect a much wider range of gases (especially electronic noses based on the metal oxide principle, like the Karlsruhe micro nose KAMINA). Furthermore, the electronic nose KAMINA is more stable in terms of sensitivity and can also be employed in hazardous environments without running any risks, and its resolution can be adapted for the task. Localizing gas sources the size of a centimeter was easily possible, so that even smaller objects should be able to be located.

For rapid determination of a wider spread of concentration, for instance of some 10 cm up to some meters, as e.g. required in soil analysis for residual waste treatment of volatile and slightly volatile organic pollutants, raster scanning is carried out with a insertion probe (measurement time approx. one minute per raster position). At present the IFIA is continuously working on gas or odor mapping techniques and processes that allow investigation of larger areas with LENA. LENA is a radio-controlled airship designed for identification of gas emitting objects and equipped with a KAMINA, a GPS system and a video camera. Initial tests have shown that soil-near sampling with the shortest possible gas way to the KAMINA sensor chip is essential for preventing chromatography effects in the gas insertion tube. Localization of a passive source with a resolution smaller than one meter was realized in combination with a gas source recognition test performed in a hall, all samples were taken 30 cm above the ground. The active model source with a methane emission of 8 l/min could still be detected at a flight speed of some km/h, so that raster-scanning of larger areas should not be too time intensive. It will be interesting to see how far these promising results can also be applied to outer air conditions.

Freedom from restrictive assumptions that underlie many quantitative techniques make neural networks attractive for image processing and data classification. The potential of the self-organizing feature map neural network proposed by Kohonen (SOM) for image processing was investigated. An adapted SOM methodology (Spectral Self-Organizing Map, s-SOM) has been developed and applied to hyper-spectral MIVIS data to attain two distinct objectives: to realize an image segmentation in similar spectral clusters as a support for supervised training and processing and to produce an unsupervised classification map. The s-SOM results were compared with classical unsupervised techniques showing an higher feature extraction ability for images with across scene illumination variability. The s-SOM output was shown, also, to correspond closely to classifications obtained from supervised procedures used as reference data. MIVIS scanner provides a large amount of data about object spectral signatures and s-SOM neural networks proved to be a suitable method to interpret its informational richness.

This paper discusses the design challenges in creating an effective and robust distributed measurement and control architecture for multisensor systems, based on the recent and not all yet approved IEEE 1451 Smart Transducer Interface family of standards, that provides the common interface and enabling technology for the connectivity of transducers to microprocessors, control and field networks, data acquisition and instrumentation systems. The goal of this standard is to make available a mechanism to facilitate interoperability between transducers and networks, so reducing the complexities designers face in establishing digital communication with transducers, providing additional benefits such as self-identification of sensors, self-configuration and documentation, easy maintenance and increased data and system reliability.

Mathematical dispersion models play a central role in both scientific investigation of how pollutants form and evolve in the atmosphere as well as in the interpretation of experimental measurements. In this work, an integrated modeling system for photochemical pollution is presented. The modeling system consists of meteorological models, emission preprocessors and the photochemical model. The system is applied over Salento Peninsula, located in the south-east corner of Italy, where summertime atmospheric conditions, typical of the Mediterranean area, make possible the creation and persistence of photochemical pollution.

The paper investigates the problem of odour field estimation by a community of autonomous agents. The agents can move in space and can have different levels of chemical sensing capabilities, from one to several different sensing elements. The objective is to estimate the diffusion field of each different odour present in the environment. To this community of distributed agents it is called herein: SpreadNose. This work deals with three areas of research: environmental monitoring, electronic noses and autonomous mobile robots. This work can easily be adapted to city buses in order to monitor the pollution inside cities.

Hetero-associative and modular auto-associative memories were tested for electronic nose applications and compared with backpropagation neural networks (NN) with one hidden layer. It was shown that associative memories provide approximately the same recognition quality of chemical images as feed-forward neural networks. Moreover, non-iterative nature of neural associative memories makes them quite attractive: both high-speed learning and possible random network initialisation are essential. Even a low generalisation ability of associative memories may be turned into advantage; using stable chemical sensors they will be able to recognise thousands sorts of chemical substance.

A sensor array based on metal oxides has been used to detect primary pollutants (NO₂, CO, toluene and octane). The array is composed by eight different oxide thin films. The films are formed by tin or tin-titanium oxides deposited by r.f. reactive sputtering. Some tin oxide sensors are doped with Pt by sputtering in order to achieve selectivity to reducing gases. Detections are carried out with single gas (CO, NO₂, toluene and octane) and gas mixtures (two and three gases) in dry air at 250 °C. High sensibility and reproducibility have been obtained for all sensors. The data sets collected from the sensor array were analysed with Principal Component Analysis (PCA) and Artificial Neural Networks (ANNs) in order to discriminate the gases. The PCA allows a good discrimination between gases. ANNs is necessary to discriminate single gases in a mixture. A probabilistic neural network identifies the whole set of single gases and mixtures. We use different pre-processing techniques to improve the classification results.

Author Index