For routine analysis, wet chemical methods are often used to determine food quality. In addition, instrumental methods and human sensory analyses have been utilized for the determination of shelf life, food quality and origin. Unfortunately, apart from the time-consuming aspect of these methods, they are sensitive to misinterpretation. In the last decade, new devices, so-called electronic noses, have been described as both a cost-effective and time-saving substitute. However, most of these chemical sensors showed deficiencies in selectivity, sensitivity and reproducibility, and only a few applications are classified to partly replace the classical methods.

The Complexity of Food Processing
Modern food production involves many complex processes that must be tightly controlled to ensure food safety and quality. The increase in national and international guidelines, ordinances and laws on food quality and safety require stricter controls through governmental agencies and/or in-house quality assurance systems (e.g., Hazard Analysis and Critical Control Points). This frequently involves time-consuming and expensive intermediary steps, since individual production processes must be analytically monitored by their respective lab-oratories.

Such controls generate additional financial burdens that disproportionately affect small and medium-size enterprises. So-called on-line methods, integrated into the production process, save time and money but are currently restricted to measuring only the simplest chemical and physical parameters, such as pH and temperature, for routine applications. More complex systems are under development but very often are not reliable enough for on-line measurement in commercial environments.

In the electronic nose business, one may differentiate between bioelectronic noses, which apply physiological-type components, especially olfactory receptors that maintain the specificity of the system, and “classical” electronic noses that work on a purely chemosensory basis, that is, with microelectronics.

Gas Sensors and Electronic Noses
In general, sensor devices already measure critical parameters like pressure, temperature, relative humidity, water activity and pH, which are some of the factors for controlling the microbial environment in the production plant.

There have been trials for many years to detect odorants by means of artificial sensors. These electronic noses, however, are still far from their biological paradigm in terms of sensitivity, robustness and especially specificity. The general principle, similar to the physiology of smell, is to correlate pattern recognition to odorants to generate a specific sensory fingerprint. Unlike human test subjects, who can test only random control samples, electronic noses provide the possibility for objective and continuous monitoring.

Gas detectors, which sense the presence of gases, may be based on any kind of sensor system: combustible gas sensors, photoionization detectors, infrared point sensors, ultrasonic sensors, electrochemical gas sensors, semiconductor sensors and biosensors.

With an increasing need for advanced monitoring systems for food processes, interest must be focused toward on- or in-line monitoring at the production stage. The food processing industry needs a fast, standardized, objective and cost-effective tool to control and improve quality.

However, there is still a need for rapid sensor methods to assess food safety issues as well as microbial and sensory quality in the production line for detection of bacterial contamination or other aberrations.

Looking to the heart of a sensor system, the sensor element, its task is to respond to a physical stimulus by producing a signal (the sensor is a type of transducer) that provides, either directly or indirectly, related information about the status of a target condition (e.g., the presence or concentration of a substance).

Sensors for volatile detection (vapor-phase analysis) are already used in aviation, automobiles, medicine and laboratory measurements. The response of these solid-state gas sensors is usually not very specific. In the early 1980s, the concept of electronic noses, that is, the detection of volatiles by sensor arrays, was proposed.

In 1993, Gardner and Bartlett defined the electronic nose “as an instrument, which comprises an array of electronic chemical sensors (e.g., solid-state gas sensors) with partial specificity and an appropriate pattern-recognition system, capable of recognizing simple or complex odors.”[1]

Detection of Food Aromas
Aroma substances are volatile compounds, which are perceived by the odor receptor sites of the sensory organ—the olfactory tissue of the nasal cavity. More than 10,000 compounds are believed to be detectable in foods, of which no more than 230 play a role in the perceivable aroma of a given food. These odorants are referred to as key food odorants. However, many more may play a role as useful indicators in food processes or for the presence of malodors as a result of suboptimal processing parameters.

Due to the complexity of the task, such electronic nose applications can be used only for very specific applications with clearly defined target profiles, such as the detection of a malodor in the monitoring of continuous production lines. Moreover, and in contrast to the analytical approach of a gas chromatographic system, a sensor system is confronted with all volatiles of a sample at a single moment. So either the sensor system must have a high specificity or only a few volatiles at the same time can be introduced to the system. Implementation of the sensor system requires special efforts in the qualification of the applied sensing elements and their calibration.

Bioelectronic Noses
The major advantage of bioelectronic noses is their high specificity, as physiological, environmental odorant-binding molecules (e.g., mammalian or insect odorant receptors) are activated only by highly specific interactions with their cognate volatile ligands. The challenges for bioelectronic noses are to maintain specificity when employed in nonphysiological environments and to retain durability of the biotech fusion elements and the reversibility of the detection event for repeated usage. For example, the complex biochemical detection system in insect antennae can be used as a selective detection system for volatile compounds in the parts per billion range. Many prototypes have already been designed that work sufficiently in the lab. As mentioned above, the main challenges will be the robustness and stability of such devices as a control system in a food processing environment.

Another current trend is the usage of conductive nanomaterials as field-effect transistors using single-walled carbon nanotubes and carboxylated polypyrrole nanotubes as carriers for odorant-binding molecules. In this fusion technology of semiconductor and biomolecular sciences, the receptors are expressed in large amounts in a biotechnological system, such as Escherichia coli strains purified and assembled on these sensing transistors. Cases have been shown in which these sensors responded to odorants in a concentration-dependent manner and with good sensitivity. However, no industry-relevant applications have yet been reported.

Metal Oxide Semiconductors as Examples of Chemosensors
In the early 1990s, commercial electronic noses appeared on the market and in scientific laboratories; since then, hundreds of papers about possible applications have been written. Advantages of these devices are the relatively low cost for the device itself and its maintenance, as well as its easy handling. Disadvantages are effects like sensor signal shifts and sensor drift (phenomena that might change sensor signals unpredictably) and the device’s often nonspecific detection abilities.

Metal oxide semiconductors (MOS) are the most frequently used sensors for gas sensing and belong to the group of solid-state-based chemosensors. They were first developed in the 1960s as detectors for liquid petroleum gases, consisting of a metal oxide layer on top of a semiconductor. The gas-sensing principle is based on the reaction between adsorbed oxygen on the oxide surface and incoming molecules. The output signal is derived from a change in conductivity of the metal oxide caused by the reaction with the volatile molecule. There are two types of MOS sensors: the n-type (SnO₂, ZnO, Fe2O₃, WO₃), which responds to reducible gases, and the p-type (CuO, NiO, CoO), which in turn responds to oxidizable gases. The sensors operate at temperatures between 300 and 450 °C. The metal oxide in the surface layer determines the selectivity of the sensors; an optional added catalyst (noble metals, mostly platinum) will also influence selectivity as well as their operating temperature. They reach sensitivity ranges from 5 to 500 ppm and are relatively insensitive to water in the range of 30–80 percent relative humidity, which is quite important for food applications.

Nowadays, it is possible to find scientific publications on the use of electronic noses for nearly every food matrix. The most prominent examples are coffee and wine, although both clearly show the limitation of electronic noses.

As practically relevant aroma profiles usually consist of many different volatile compounds, a separation of all substances by high-resolution chromatography is desirable but very expensive and elaborate. A basic goal is thus to increase the specificity and sensitivity of the sensing elements, allowing for the omission of advanced separation techniques. Further work in this field focuses on the development of mobile devices that combine gas chromatography and detectors to simplify the analysis and reduce the analysis time drastically. Hence, the basic goal, of course, is to improve the performance of the sensor to make the chromatographic part of the device unnecessary. Which volatile substances may be appropriate for chemical sensing is further influenced by their suitability within the process, their detectability in an available sensing mode and their expected concentration and background levels. In the case of alcoholic beverages, the major volatile substance—ethanol—must be taken into account as a potentially interfering substance.

To achieve the basic goal described above, new detector materials are being developed and fused with each other. The materials tested are tin dioxide (SnO₂), tungsten oxide (WO₃) and SnO₂ fused with a catalytically active platinum addition or chromium titanium oxide. To ensure the highest amount of information, sensor systems are equipped with an array of sensors of the same or different working principles (so-called hybrid or multisensor systems). The required composition of the sensor array, however, strongly depends on the matrix and selected target molecules. Realizable fields of application are processes in which very specific odorants need to be captured or in which the aroma profile consists of only a few volatile molecules. Potential fields of application for more complex future sensors will include the storage, processing and distribution of foods. Monitoring the ripening of fruit and fresh produce is another potential field.

The amount of raw data from up to 30 single sensors requires complex statistical data processing. Evaluation software of sensor systems often includes functions for these sophisticated statistical approaches, like principal component analysis, partial least squares or artificial neural network calculations. Importantly, these mathematical postmeasurement steps require a high degree of experimental repeatability and technical standardization to be applied by automatic algorithms without the need for manual expert evaluation.

Desired Industrial Applications and Future Trends
Sensors have a great potential for use in the food processing industry for process and quality control. The application areas are in the fields of pathogen detection in raw materials, processing and quality control of final product.

So far, most of the sensors that have been implemented in the food industry on the production line have been for the environmental monitoring of hazardous gases like hydrocarbons, ammonia and hydrogen sulfide that may occur during production. However, these kinds of sensors, including physical sensors, provide information on the performance of the process, which may indirectly contribute to controlling the hygienic quality of the process but does not provide direct information on the quality of the product being processed. For this purpose, chemical and biosensors are helpful with regard to their on-/in-line options. Since most research has been conducted under laboratory conditions, these options strongly depend upon the application and the process surroundings.

In a typical bioprocess, cells are grown under strictly controlled monocultural conditions in fermenters or bioreactors in liquid media that provide essential nutrients, vitamins, etc. The products from the bioprocesses range from enzymes to biopharmaceuticals or valuable food odors, which naturally all impose high demands on product quality and safety. Gas sensor array systems have proven very useful for both quantitative and qualitative bioprocess monitoring, which allows the real-time determination of cell status, growth rates and product concentration. Another advantage of this technology is that it can be used to discover contamination with foreign microorganisms on-line in real time in the bioreactor tank after only a few hours of processing, which is a significant gain over traditional microbiological methods. The application of noninvasive, on-line monitoring methods like gas sensor arrays could therefore certainly contribute to improvements in the quality of bioprocessed products. This has also been documented by several studies.[2-4]

Two basic trends will have the strongest influence on applied research in the field of sensor techniques: automation and miniaturization, which are strongly connected. The necessary instrumentation, which can work without supervision, can be implemented in an existing process technology to avoid disturbing the ongoing process. Furthermore, these systems should be integrated into the overall quality assurance and traceability system.

Currently, existing methods cannot serve these needs of regulatory agencies and food producers. Future real-time testing with reliable sensor technology will provide value to food producers by reducing treatment costs and product recalls. As the demands for food safety increase, the requests for fast sensing technologies will only rise with them.

Because of ever-increasing regulations and transparency demands by customers, the food industry must know more about its products than ever before. While traditional food processors would like to control the process and thus measure process parameters at the production site to obtain maximum certainty, many more diagnostic devices are needed or highly desirable. At the production site, this means that these devices—in contrast to external commercial analysis—need to work at least at-line or by-line, and better on-line or in-line. There will be a market for devices that

•    measure in a given range (e.g., within EU regulations) everything that helps comply with the preceding requirements;

•    can be implemented in an existing quality assurance system, also with respect to data management;

•    work accurately, reliably and for a long time;

•    have a suf?ciently short response time;

•    are hygienic (do not adversely affect the run time of the process line or the quality of the product);

•    are easy to operate, maintain, dismantle and reassemble; and

•    are affordable for the food industry.  

Mark Bücking, Ph.D., is head of the Department for Environmental and Food Analyses at Fraunhofer Institute for Molecular Biology and Applied Ecology IME, Schmallenberg, Germany, and managing director of the Fraunhofer Food Chain Management Alliance.

Matthias Kotthoff, Ph.D., is head of the laboratory for food and environmental analysis at the Fraunhofer Institute for Molecular Biology and Applied Ecology IME.

References
1. Gardner, JW and PN Bartlett. 1994. “A Brief History of Electronic Noses.” Sens Actuators B18:211–220.
2. Lelieveld, HLM et al. Handbook of Hygiene Control in the Food Industry (Elsevier, 2005).
3. Pearce, TC et al. Handbook of Machine Olfaction: Electronic Nose Technology (John Wiley & Sons, 2006).
4. Loutfi, A et al. 2015. “Electronic Noses for Food Quality: A Review.” J Food Engin 144:103–111.