B5.2 - Evaluation of an E-Nose System for Objective Smell Assessment of Shoe/Socks Systems by Comparison with a Sensory Panel

Abstract

In this paper we report on the evaluation of an E-nose system for objective evaluation of the smell of sweat in shoe/socking systems by comparison with a human sensory panel. The test system is based on temperature cycled semiconductor gas sensors to achieve good stability for field use of the system without the requirement for frequent recalibration. The ultimate goal is to provide a tool for developing improved shoe/sock systems with optimized materials. The main approach to achieve this goal is to find a correlation between the assessment of a human sensory panel and the complex sensor response patterns of an E-nose system to appraise the smell of sweat in shoes and socks. Therefore a range of test persons wear shoes and socks under defined ambient conditions in a controlled test environment as well as during everyday use. The evaluation is done by comparison of the data measured by the E-nose system with two independent sensory panels, each consisting of six persons. The human sensory panels evaluate the smell on worn shoes and socks and assess the intensity of the smell of sweat and the olfactory sensation/unpleasantness for each test object. The paper describes the setup of the E-nose system and presents results of a first larger field test showing the identified correlation between our system and the human assessment of the smell of sweat. Motivation Suppression or mitigation of body odors is one of the key issues for shoe and clothes manufacturers in Europe to achieve a competitive advantage over low cost suppliers especially from Far East. Specifically the smell of sweat in shoes and socks is a major issue for consumers and manufacturers alike and will eventually be an essential aspect for customer acceptance. Today, only very complex studies with human sensory panels allow an estimation of the impact of different construction characteristics of the shoe/socks system, i.e. the materials used and their combinations. In this respect shoes present a major problem as they usually are not or cannot be washed, so that substances and microbes can accumulate and lead to quite severe smells. A collaborative project between the lab for measurement technology and two research institutes specialized on shoes and clothing, respectively, aims at developing a mobile test system for objective evaluation of the strength and unpleasantness of the smell of sweat to be used in development of shoes and socks, in quality control and for settling customer complaints. Preliminary studies have shown a correlation between the signals of an electronic nose based on semiconductor sensors and human perception [1]. The key will be to identify characteristic sensor response patterns for correlation with the assessment by a human sensory panel, a typical approach for developing an application specific electronic nose [2]. Experimental methods The E-nose system is based on a small test chamber in which the shoe and/or socks are placed for evaluation, Fig. 1 [3]. Gas is pumped from inside the chamber using a probe which can be placed inside the shoe or socks and passes through a sensor chamber with up to eight different semiconductor gas sensors (SGS). To suppress unwanted influence by humidity and interfering gases, the chamber is flushed with pressurized clean air, which is humidified to almost saturation. In addition, the test chamber can be placed in a temperature chamber to suppress the influence of ambient temperature. Temperature and humidity in the test chamber and in the chamber for gas sensors are measured, which will allow a correction of the influence of ambient conditions at a later stage, i.e. by taking the temperature and r.h. values into account for the signal interpretation. As many electronic noses based on sensor arrays show a lack of stability and therefore require frequent recalibration, we chose to use a system based on temperature cycling which has proven highly stable for the detection of organic substances over several months [4]. Each sensor is controlled by an individual electronics board for temperature cycling and data acquisition [5] and is connected via USB to a laptop computer with graphical user interface based on LabVIEW, NI, for signal evaluation and