Activity Report

Report by Dr. Romeo Beccherelli

Topics

• Biomimetic sensing
• Sensor fusion, integration
• Healthcare
• Security

Six invited lectures and 11 oral contributions were given at the Workshop chair by Prof. Roberto Paolesse, University of Rome "Tor Vergata" and Prof. Krishna Persaud, University of Manchester (UK). These were allocated in four half-day sessions as detailed in the attached programme. Almost 40 participants were present at the workshop.

Summary of the sessions

The workshop opened with the "Biomimetic sensing" session chaired by Prof. Persaud. Dr. Tim Pearce from University of Leicester (UK) gave a description of vertebrates olfaction system and presented a "neuromorphic olfactory system implementation" realized in the AMS 0.6 µm CUP CMOS process. This chip encompasses for the first time an array of 25 chemFET (based on the floating gate concept) and resistive sensors employing carbon black (CB) composite materials. The chip integrates externally programmable signal amplification data acquisition and drift compensation. Also the analogue implementation of a neuromorphic processing network was fabricated with the same silicon process.

Ms Sara Jackson form University of Bristol (UK) described in detailed "Behavioural Strategies of Source Location in Air Trailing Dogs" and the process of training these dogs she has been carrying on for many years.

Dr. Gianfranco Bocchinfuso from Universita' di Roma "Tor Vergata" (Italy) discussed "A selective chemoreceptor" for the detection of aflatoxin in foods by using a sensor based on an antibody molecule. The 18 varieties of aflatoxins are carcinogenic, mutagenic, teratogenic and immunosuppressive substances, which are produced as secondary metabolites by the fungi Aspergillus flavus and Aspergillus parasiticus growing on a variety of food products, mainly in cereals, peanuts, and dried fruits. The chemoreceptor modelled with "semi-empirical models" (AM1) and optimised "Multiple Copy Simultaneous Search" (MCSS) to obtain selectivity and high binding affinity towards the selected target molecule. Soluble tetra-phenyl-porphyrin was then synthesised

Prof. Jerzy P. Lukaszewicz from Nicholas Copernicus University (Poland) presented an overview on "Functionalization of carbon nanotubes and other carbon-based materials for chemical and biological sensing". He described the various phases of carbon (polycrystalline, graphite, carbon black, nanotubes, fullerenes), diamond) and their application to sensor technology. Functionalisation by oxidation, metallization with nanoparticles is possible for all forms of carbon but diamond. Sensitivity to gases is found to be as low as 100 ppb.

Dr. Maximilian Fleischer from Siemens AG, Munich (Germany) described industrial interest in "Chemical gas sensors for medical applications", their challenges and requirements. He discussed some specific applications. Detection of NO₂ is a challenging task. The threshold to consider air "clean" is 21 ppb of NO₂. Sensitivity in the ppb range is achieved by ChemFET (based on the floating gate concept) coated with suitable organic materials. Detection of NO is important for Asthma diagnosis, however sensitivity to NO is poor. Therefore conversion to NO₂ is performed by means of KMnO₄ immobilized on Silica gel 60. Detection of ethanol in vehicle drivers breath is also important and has to be in 500 ppm range. However many interference are found in human breaths. Dr. Fleischer described how ethanol can be detected with a resistive sensor based on interdigitated electrodes and Ga₂O₃ (at 900°C) as sensing material. He also addressed gas detection based on NIR-Absorption by using tunable-IR Laserdiodes (DFB or VCSEL type).

In the "Healthcare and Medical" session, chaired by Dr. Udo Weimar, Giorgio Pennazza, University of Rome "Tor Vergata" (Italy) reported on 6 years of experience detection of lung cancer at Forlanini Hospital in Rome. The volatiles in lungs affected by cancer are mainly alkanes and aromatic compounds. However no single compounds can be consider a clear marker of the disease. It is rather the whole combination of all the compounds that makes possible the disease identification Therefore an electronic nose, composed by eight quartz microbalance (QMB) gas sensors, coated with different metalloporphyrins, was used. These sensors show a good sensitivity towards those compounds previously indicated as possible lung cancer markers in breath. Tests were carried out on 42 volunteers affected by various forms of lung cancer and 18 controls were carried out on healthy patients. The application of a ‘partial least squares discriminant analysis’ (PLS-DA) found out a 100% of classification of lung cancer affected patients, 94% of reference was correctly classified. The class of post-surgery patients were correctly individuated in 44% of the cases, while the other samples were classified as healthy references. Extended studies are necessary to evaluate the resolution of the method, namely the stage at which the disease may be identified in order to use this instrument for early diagnosis.

Dr. Daniel Filippini from Linköping University (Sweden) Sweden reported on the "Possible use of computers and web cameras for distributed medical diagnostics". Computer Screen Photo-assisted Technique (CSPT) are 100% based on low cost already available platform (camera, desktops, laptops, internet browsers). These are cheaper than currently used techniques such as VIS spectroscopy fluorescence, spectroscopy, micro-plate readers and less subjective than Visual inspection

Dr. Arunas Setkus Semiconductor Physics Institute, Vilnius (Lithuania), discussed the use of a electronic nose for the non-invasive identification of bacterial infections in wound. They used various sensitive layers (metal oxides; multilayers of the oxides, combined with biomoleculs) and 8 to 13 sensing elements for the identification of different pathogens. The sensor array allows rapid changes in the sample culture with the living bacteria to be observed within 1 hour. However some problems remains related to the poisoning of the sensor surfaces. Some organic molecules produce a thick layer that screens the surfaces of the sensors. This is visually observed as white colouring of the surface of the sensor after 4-6 months of the periodical tests.

In the "Security" session, chaired by Dr. Romeo Beccherelli, Dr. Enrico Verona from CNR-IDAC, Rome, gave an overview on "Mass sensors for security applications" based on bulk acoustic waves (BAW) and surface acoustic waves (SAW) and their resolution. For instance, commonly BAW structures are based on 20 MHz QMB. These have a sensitivity of 4.8 Hz/ng. The higher the frequency and the higher the acoustic velocity, the better is the sensitivity. However, the higher the thickness, the thinner and the more fragile is the device. This prevents use of much thinner and sensitive BAW QMB based on quartz. SAW devices can easily operate at much higher frequency and can be based on materials with higher acoustic velocity. An device operating at 402 MHz for the detection in the 100 ppb range of nerve gas were demonstrated. Dr. Verona described an alternative route to increase sensitivity that employ a thin film bulk acoustic resonator (TFBAR) based on high acoustic velocity AlN on a suspended micromachined membrane. The manufacturing process was thorough fully described for a 1.5 GHz device.

Matteo Falasconi from University of Brescia and CNR-INFM Brescia (Italy) gave an overview food pathogens in food and addressed the "Screening of mycotoxins in corn by an Electronic Nose" based on metal oxides. The problem addressed is quite relevant as 10-20% of the worldwide annual production of maize is spoiled by filamentous fungi (Aspergillus, Eurotium, Penicillium and Fusarium), with an estimated economic loss in excess of 1B€/year (source: FAO). The fungi produce mycotoxins which determines health problems upon ingestion by humans and animals. Experiments have been carried out on maize grains contaminated by Fusarium spp. The EOS835 electronic nose seems able to correctly classify contaminated maize samples. However works remain to be done on the effects of multiple contaminations.

Dr. Zafer Ziya Öztürk from TUBITAK (Turkey) reported on "Detection Of Explosive Chemicals and Chemical Warfare Agents Simulants With Chemical Sensors". Use of simulants is needed because of limitation on the use actual warfare agents. Dimethyl methylphosphonate (DMMP) was used and its detection was investigated with QCM coated with Phthalocyanines (NiPc, ZnPc and CoPc). Also Phtalocyanines were also use for detection the detection of nitro compounds such as 2,6 dinitrotoluene (DNT) in the ppb range in air at room temperature by coating SAW devices. The sensor behaviour is completely reversible.

Andrea Lixi from Sartec, Cagliari (Italy) discussed the "Analysis of Odours from Explosives using an Electronic Nose". Conventional equipment use ionisation mass spectroscopy (IMS). He stated that this technique is very sensitive but has the problem that it can be calibrated for a limited number of energetic compounds and has a limitation in a lack of coverage of the background chemicals in odours. On the other hand, electronic noses can be design to respond to a broad range of compounds. Sartec has developed a commercial product (zNose) based on a new type of ultra-fast gas chromatograph a and a SAW detector that separates and quantifies the organic chemistry of odours, from C5 to C30 by means of ultra-high speed chromatography (10s response time) with ppt sensitivity. Detectable explosive include nitroglycerine, dinitrotoulene, trinitrotoluene pentaerythritoltetranitrate, cyclotrimethylenetrinitramine, trinitrophenyl-n-methylnitramine and real world compounded explosives (PETN, RDX, Datasheet, C4, triacetone triperoxide (TATP).

In the "Sensor fusion and Integration" session, chair by Prof. Ingemar Lundström from Linköping University (Sweden), Prof. Corrado di Natale University of Rome "Tor Vergata" described some applications of "Data Fusion". He pointed out the necessity to ensure that the sample does not change going through the different instruments. He discussed low-abstraction level techniques where data are merged together before the analysis and high-abstraction level techniques where data are separately analyzed and the most prominent features are merged. This mimics human perception where sensorial perception involves many senses, so that sensorial analysis is actually a data fusion. Application employing an electronic nose and evaluation of skin appearance for the identification of ripeness of apples and freshness of cod fish were described.

Dr. Matteo Pardo from INFM-CNR Brescia (Italy) "Feature Selection", aiming at reduction of dimensionality in general for (i) improved classification, (ii) simpler graphical representation (in 2D, e.g. PCA), (iii) easier interpretation and (iv) reduced effort in term of number of measurements and computing equipment. In the case of an electronic nose this corresponds to identifying which are the sensors that give the largest contribution to discrimination. He pointed out that this is not by necessity the most sensitive.

Dr. Fredrik Winquist from Linköping University Sweden reported on two types of "electrodes arrays in sensor systems", namely a voltammetric and a potentiometric electronic tongues.

Dr. JaeHo Sohn from Queensland State Government, Australia, presented "Machine based measurement of odours emission from anaerobic pond" filled with piggery manure. The instruments used are an AromaScan A32S with Artificial Neural Network (ANN) and relives human tester from the unpleasant task.