Hans Sundgren

Performance of an electronic nose for quality estimation of ground meat

An electronic nose is described, which consists of a gas sensor array combined with a pattern recognition routine. The sensor array used consists of ten metal-oxide-semiconductor field effect transistors with gates of catalytically active metals. It also contains four commercially available chemical sensors based on tin dioxide, so-called Taguchi sensors. In some studies, a carbon dioxide monitor based on infrared absorption is also used. Samples of ground beef and pork, stored in a refrigerator, have been studied. Gas samples from the meat were then led to the sensor array, and the resulting patterns of sensor signals were treated with pattern recognition software based on an artificial neural network as well as with an algorithm based on an abductory induction mechanism. When using all sensors for learning, the two nets could predict both type of meat and storage time quite well. Omitting the carbon dioxide monitor, both nets could predict type of meat, but storage time not so well. Finally, it is also shown how a net based on unsupervised training could be used to predict storage time for ground beef.

Artificial neural networks and gas sensor arrays: quantification of individual components in a gas mixture

A very promising way of increasing the selectivity and sensitivity of gas sensors is to treat the signals from a number of different gas sensors with pattern recognition (PARC) methods. A gas sensor array with six metal-oxide-semiconductor field-effect-transistors (MOSFETs) operating at elevated temperatures was exposed to two types of multiple-component gas mixture, one containing 5-65 ppm of hydrogen, ammonia, ethanol and ethylene in air and the other containing hydrogen and acetone in air. The signals from the sensors were analysed with both conventional multivariate analysis, partial least-squares (PLS), and artificial neural network (ANN) models. The results show that both hydrogen and ammonia concentrations can be predicted with PLS models; the predictions were even better with ANN models. The predictions for ethanol and ethylene concentrations were, however, poor for both types of model. Hydrogen and acetone, from the two-component mixture, were best predicted from an ANN model.

Electronic nose for microbial quality classification of grains

The odour of grains is in many countries the primary criterion of fitness for consumption. However, smelling of grain for quality grading should be avoided since inhalation of mould spores or toxins may be hazardous to the health and determinations of the off-odours are subjective. An electronic nose, i.e. a gas sensor array combined with a pattern recognition routine might serve as an alternative. We have used an electronic nose consisting of a sensor array with different types of sensors. The signal pattern from the sensors is collected by a computer and further processed by an artificial neural network (ANN) providing the pattern recognition system. Samples of oats, rye and barley with different odours and wheat with different levels of ergosterol, fungal and bacterial colony forming units (cfu) were heated in a chamber and the gas in the chamber was led over the sensory array. The ANN could predict the odour classes of good, mouldy, weakly and strongly musty oats with a high degree of accuracy. The ANN also indicated the percentage of mouldy barley or rye grains in mixtures with fresh grains. In wheat a high degree of correlation between ANN predictions and measured ergosterol as well as with fungal and bacterial cfu was observed. The electronic nose can be developed to provide a simple and fast method for quality classification of grain and is likely to find applications also in other areas of food mycology.

Evaluation of a multiple gas mixture with a simple MOSFET gas sensor array and pattern recognition

Abstract The properties of a gas sensor array can be improved by the use of pattern recognition (PARC) methods. A gas sensor array with three pairs of Pd-gate MOSFETs and Pt-gate MOSFETs is exposed to a multiple-component gas mixture. Each pair is operated at a different temperature. The signals from the six sensors are analysed with both linear and nonlinear PLS (partial least square) models. The calculations of the PLS models are based on sensor signals obtained from calibration experiments. By means of combining a pattern recognition method with a semiconductor-based gas sensor array, we show that hydrogen concentrations can be predicted well in the presence of three other interfering gases.

From hydrogen sensors to olfactory images — twenty years with catalytic field-effect devices

Abstract A personal description of the history of gas-sensitive field-effect devices is given. It is shown how the originally described palladium-gate metal-oxide-semiconductor field-effect transistor has developed into sensing surfaces enabling the production of response images to odours. Images obtained for the odour from different cheeses are presented as examples of such ‘artificial olfactory images’.

Quartz crystal gas monitor with a gas concentrating stage

Abstract A quartz crystal monitor is described which enables a reproducible detection of certain hydrocarbons at a concentration level of parts per million in varying backgrounds of relative humidity. The background signal due to water vapour and the influence of base line drift were eliminated by the use of a gas absorption/desorption unit and water-permeable tubing in front of the detector crystal. The absorption and desorption were obtained through cooling and heating a layer of a silicone oil with Peltier elements. The quartz crystal detector, coated with the same type of silicone oil, has a fast and reversible response to gases like halothane and toluene. The requirements on the absorption/desorption unit are studied in some detail.

A low-power micromachined MOSFET gas sensor

This paper reports on the design, fabrication, and characterization of the first low-power consumption MOSFET gas sensor, The novel MOSFET array gas sensor has been fabricated using anisotropic bulk silicon micromachining. A heating resistor, a diode used as temperature sensor, and four MOSFETs are located in a silicon island suspended by a dielectric membrane. The membrane has a low thermal conductivity coefficient and, therefore, thermally isolates the electronic components from the chip frame. This low thermal mass device allows the reduction of the power consumption to a value of 90 mW for an array of four MOSFETs at an operating temperature of 170/spl deg/C. Three of the MOSFETs have their gate covered with thin catalytic metals and are used as gas sensors. The fourth one has a standard gate covered with nitride and could act as a reference. The sensor was tested under different gaseous atmospheres and has shown good gas sensitivities to hydrogen and ammonia. The low-power MOSFET array gas sensor presented is suitable for applications in portable gas sensor instruments, electronic noses, and automobiles.

Recent developments in field-effect gas sensors

Some of the ongoing studies at our laboratory of gas-sensitive field-effect devices with catalytic metal gates are reviewed. More particularly, we discuss the use of such devices in so-called electronic noses due to the possibility of changing the selectivity patterns of the devices by the choice of catalytic metal and operation temperature. Several examples of the application of electronic noses consisting of field-effect devices in combination with metal oxide-based sensors are given. Finally, a summary is given of some remaining scientific problems and studies related to the understanding and development of gas-sensitive field-effect devices.

Neural networks and abductive networks for chemical sensor signals : a case comparison

Abstract Both artificial neural networks (ANNs) and the abductory induction mechanism (AIM) have been proven to be suitable for the evaluation of signals from chemical sensors with strong interactions of gas components on the catalytically active surface. The algorithms allow the calculation of calibration curves for a multisensor. AIM yields a good mean approximation within a very short time; ANN covers a broader concentration range with an adequate approximation. The signal evaluation of a set of two pellistors and another set of six MOSFETs is used for illustration.

Thermally isolated MOSFET for gas sensing application

This work reports on thermally isolated electronic components for gas sensing applications. The device is composed of an array of 4 MOSFETs, a diode and a semiconductor resistor integrated on a micro-hotplate, which is fabricated using bulk micromachining of silicon. The thermal efficiency of the device is 2/spl deg/C/mW with a thermal constant less than 100 ms. Holes are made in the passivation film over the gates of the MOSFET and gas sensitive films deposited on top of the gate insulator. The low thermal mass device realized allows new modes of operation for MOSFET gas sensors such as a combination of the field and thermal effects and a pulsed temperature mode of operation.