Pollution monitoring is the key to air quality management. The concentration of air pollutants is measured at network reference stations using precise analytical instruments consisting of bulky, heavy, and difficult-to-use high energy-consumption equipment. Thus, the number of available stations is limited due to high operation and maintenance costs. The stations are preferentially located in urban areas, although in many cases they are far from the main sources of pollution. However, in rural areas with few inhabitants and in remote or inaccessible areas, the measuring stations and therefore the pollution data are not available.
Currently, the most promising alternative for monitoring atmospheric pollutants is the use of electronic noses formed by a sensor array. The first step in the development of electronic noses for environmental applications is reducing the cost of sensors. These sensors are required, besides their low cost, to be autonomous, easy to use, reliable and accurate. Their size, weight, and energy consumption must also be reduced [1
]. Resistive sensors of metal oxide semiconductors (MOX) are suitable candidates for the development of low-cost, high-performance sensors due to the simplicity of the physical magnitude involved in the measurement (resistance) and the high sensibility to toxic gases. In particular, nanostructured materials are the most appropriate strategy to minimize some of the current problems with gas sensors (lack of sensitivity, power consumption, and stability).
In this work, we present the development of tin oxide nanosensors for electronic noses (e-noses). The two main applications of e-noses in the environment are pollution and odour monitoring. Due to the increased interest in this field and in order to improve potential use of instrumental odour monitoring, including sensors or e-noses, a new working group (WG41) started in 2015 within the framework of the European Committee of Standardization (CEN/TC264 Air Quality). The objective of this group was to propose a new European standard for instrumental odour monitoring [2
]. Applications of electronic noses in the environment can be found in several works, some based on MOX [3
] or amperimetric commercial sensors [4
]. Other types of e-noses are those based on surface acoustic wave (SAW) sensors [5
]. Biomimetic artificial noses, including whole-cell olfactory receptor protein and odorant binding protein (OBP)-based biosensors are also being studied [6
]. Portable devices are being developed for the measurement of urban pollution [7
Gas sensors based on sensitive layers of one-dimensional metal oxide (1D) nanostructures have shown superior performance to bulk sensors due to their large surface area–volume ratio and their dimensions being comparable to the extent of the surface charge region [10
]. Tin oxide is still the most important material used for the detection of atmospheric polluting gases, and its most outstanding characteristics with respect to other semiconductors are its high sensitivity at low temperatures and low cost. One-dimensional SnO2
nanostructures (nanowires, nanobelts, nanoribbons, nanofibres, etc) can be synthesized using several methods like laser ablation, chemical vapour deposition, electro-deposition, thermal evaporation, rapid oxidation and electrospinning [13
Electrospinning is a simple, versatile and economic technique that allows fibres to be obtained at micro and nanometric scales [17
]. The electrospinning process began to be employed in conventional organic polymers of high molecular weight [19
] and in the last decade has been used for the preparation of semiconductor oxide fibres from polymer solutions incorporating metallic precursors [20
]. The process involves the application of an electrostatic field to a polymer solution with a certain viscosity and when the electric field strength is greater than the surface tension, the polymer solution is expelled to a collector in the form of a fibre.
Nitrogen dioxide (NO2
) is one of the major air pollutants, especially in large cities. NO2
is an oxidizing gas whose main emission sources are combustion processes (heating, power generation and engines in vehicles and ships). Its effect on human health can be both short-term (causing significant inflammation of the respiratory tract) and long-term (affecting organs such as the liver and spleen, systems such as the circulatory system and the immune system, which in turn leads to lung infections and respiratory failure) [23
]. In addition, nitrogen oxides alter the environment by contributing to the acidification and eutrophication (excess nitrogen nutrients) of terrestrial and aquatic ecosystems, leading to a loss of life in animals and plants and changes in species diversity [24
exposure limit values recommended by the World Health Organization (WHO) [25
] are shown in Table 1
. These low concentration ranges cannot be detected by commercial sensors at low temperature.
The European Commission [26
] has urged member states to implement air quality management plans that ensure compliance with the standards set by the EU air quality directive [27
] no later than 2020. Air pollution monitoring is a key air quality management task, for which the Air Quality Directive (AQD) opts for a strategy based on a network of a limited number of fixed stations, equipped with precision analytical instruments, which has some drawbacks.
Measuring equipment is bulky, heavy, difficult to use, and consumes a lot of energy. Equipment costs, operation, and maintenance are high. In many cases, the stations are located far away from areas of high traffic density where the greatest local increases in air pollution occur. A small number of these stations dispersed in a city allows data to be obtained with hourly resolution, but at a small number of points. In emergency situations, decisions are based on real-time measurements or, in the absence of such measurements, on predictive models of the distribution of pollutants in the atmosphere, the usefulness of which depends on the degree of validation of the models. Thus, although stations accurately measure air pollution, their spatial representativeness and temporal resolution are insufficient to capture the spatial–temporal variability of air pollution.
Although the AQD does not consider sensors as reference instruments, it does open the door to the use of sensors for indicative measurements, for which it sets less restrictive quality objectives. It is estimated that the use of low-cost, low-consumption sensors that meet AQD quality standards for indicative measurements would allow a 50% reduction in the minimum number of stations [28
]. The new generation of sensors finds application (unregulated) in sectors such as personal and community monitoring of air quality, traffic management, estimation of exposure to air pollution, R&D, and environmental education, in which there are numerous business opportunities.
In this work, two prototypes of electronic noses for environmental applications based on low-cost sensors are described. The low-cost sensors tested were nanostructured tin oxide materials (nanofibres) obtained by an economical and versatile process (electrospinning) and commercial sensors. The sensor responses to low concentrations of NO2
] in controlled air atmospheres are also presented and discussed. We obtained good responses even at room temperature.
These nanofibre-based tin oxide resistive sensors can be incorporated into an electronic nose and could be used for air quality control.
In the detection processes, the resistance changes occurred with the adsorption of gaseous molecules on the sensitive surface. Nanostructures were considered for gas detection applications due to their high surface area–volume ratio. In this work, the nanostructures—porous nanofibre networks—were composed of many nanograins that favoured the adsorption of gases.
The sensor calibrations had low errors, especially around 50–100 °C, and a lower error at room temperature. At higher temperatures, the sensors probably experienced some instability and the measurements had a much higher variability, which reflected the weaker performance. The sensors showed a good linear response in the concentration range tested. The combination of both sensors in a multilinear calibration was validated and the results were better estimated because the stricter validation and the aggregation of both sensors on a single performance was validated. The PLS had low error that tended to increase with the temperature and showed very good performance at 50 °C.
The sensors based on nanofibres had better low-temperature performance than commercial sensors and also better than that reported in the literature (Table 4
). The references showed that NO2
concentrations lower than 0.5 ppm were detected and that the sensors would operate at moderate temperatures generally higher than 150 °C. Most of the references of the sensitive layers corresponded to complex nanostructures prepared by hydrothermal methods (due to difficulty to control the process, and problems of reliability and reproducibility). Although there are usually references for the sensor response (RNO2
), there is no detail of the sensor resistance. The commercial sensors used in this work, did not have any significant response below 250 °C, but they showed a more stable response with lower errors in the calibration for higher temperatures.