The two main Research activities:

Chemical Sensors Science and Technology

In order to understand their interaction with the gases, two types of sensitive materials are investigated - semiconducting metal oxides (SMOX) and polymers. In both cases, all investigation techniques are used in sensor operation conditions (operando) meaning that the samples are fabricated in the same way in which the real sensors are (e.g. the SMOX are deposited as polycrystalline, thick, porous films onto substrates provided with electrodes and heaters) and are studied in real-life like conditions (heated between 200 and 400°C for the SMOX, at normal pressure, in the presence of humidity and of the target gases). Most of the operando experiments are performed in-house; for SMOX they include: DC resistance, AC impedance spectroscopy, work function changes (Kelvin probe), catalytic conversion and Diffuse Reflectance Infrared Fourier Transformed Spectroscopy (DRIFTS) measurements.

On the side of technology, the group invented, developed and transferred to the industry - by spinning off a company, AppliedSensors GmbH - SMOX microsensor technology that combines thick film sensing layers with micro hotplates. To date, more than 6 million sensors were commercialized. The current development are focused on: a novel SMOX sensor fabrication technology - in cooperation with the University of Bremen - based on Flame Spray Pyrolysis (FSP) one step synthesis and layer deposition; and on integration of polymer and SMOX sensing layers onto flexible/plastic substrates - in cooperation with EPF Lausanne.

Operando Methods

In order to understand the interaction of semiconducting metal oxides (SMOX) with different gases, several investigation techniques can be used - all in sensor operation conditions. The generic term operando is used to describe that the samples are fabricated in the same way in which the real sensors are (the SMOX are deposited as polycrystalline, thick, porous films onto substrates provided with electrodes and heaters) and studied in real-life like conditions (heated between 200 and 400°C, at normal pressure, in the presence of humidity and of the target gases). Most of the operando experiments are performed in-house: DC resistance, AC impedance spectroscopy, work function changes (Kelvin probe), catalytic conversion and Diffuse Reflectance Infrared Fourier Transformed Spectroscopy (DRIFTS) measurements.

Applications of Chemical Sensors

This type of activity capitalizes on the two strengths of our group: the analytical chemistry expertise and available analytical instrumentation and the deep understanding of the way in which chemical sensors work. The former allows us to first investigate the application and identify its boundary conditions: target gases, possible interferents, detection speed needs, etc. The latter makes it possible to select the most appropriate chemical sensors - if available, if not to develop them - combine those devices into arrays and, if necessary, with purging/enriching/separation elements. The collected data are used for establishing evaluation algorithms and the whole, hardware and software, are prototypes of Application Specific Chemical Sensor Systems (ASCSS) that can be transferred to the industry. There is already a very good track record in various fields: automotive, packaging, food quality, etc.

The R&D activities of the group are funded by a variety of third party projects financed by the European Commission, DFG, BMBF and directly by the industry.

Example Application: Monitoring Hand Hygiene

To improve hand hygiene in healthcare settings and reduce healthcare-associated infections, electronic noses can monitor these practices. Using portable and stationary commercially available SMOX sensors, we successfully classified hand disinfection incidents as correct or incorrect, employing various machine-learning algorithms such as multi-layer perceptrons.

Example Application: Predicting Pollutant Gas Concentrations

The World Health Organization sets exposure limits for pollutant gases like ozone, making it crucial to know current concentrations. Over 1700 hours of data, encompassing random gas mixtures (CO, NO2, O3, SO2, H2O), were recorded using temperature-modulated commercially available gas sensors. This data trained a convolutional neural network, achieving a mean relative error (MRE) of 8% for the considered gases.