A wireless sensing system for monitoring the workplace environment of an industrial installation

Sensors and Actuators B 224 (2016) 266–274

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb

A wireless sensing system for monitoring the workplace environment of an industrial installation P. Oikonomou a,∗ , A. Botsialas a,c , A. Olziersky a , I. Kazas b , I. Stratakos d , S. Katsikas d , D. Dimas d , K. Mermikli d , G. Sotiropoulos e , D. Goustouridis c , I. Raptis c , M. Sanopoulou a a

Institute of Nanoscience and Nanotechnology, NCSR ‘Demokritos’, Athens 15310, Greece Institute of Nuclear and Particle Physics, NCSR ‘Demokritos’, Athens 15310, Greece c ThetaMetrisis S.A., Athens 12243, Greece d Prisma Electronics S.A., Industrial Area of Alexandroupolis, 68100, Greece e Alfa Beta Roto S.A., Industrial area of Komotini, 69100, Greece b

a r t i c l e

i n f o

Article history: Received 16 March 2015 Received in revised form 31 July 2015 Accepted 14 October 2015 Available online 19 October 2015 Keywords: Chemocapacitors Sensor array Mote Wireless sensing system Industrial application

a b s t r a c t The realization of a wireless sensing system and its sensing performance evaluation, under laboratory conditions, for the monitoring of specific volatile organic compounds (VOCs) present in printed flexing packaging industries is demonstrated. Prior to the utilization of the wireless mote, we present the microfabrication of appropriate sensor array based on chemocapacitors and its integration with appropriate low power consumption read-out electronics meeting the requirements of the application. The sensing unit is an array of interdigitated chemocapacitors (IDCs). The wireless sensing system is tested upon exposure to VOCs, humidity and gaseous mixtures simulating the real industrial environment and the raw data are transmitted via a wireless network and monitored to a front-end software. Results showed that the sensing system is characterized by very good sensing performance with high repeatability and long-term stability. Further data processing with principal component analysis (PCA) highlights the sensing system’s ability to discriminate between gaseous environments with different composition/concentration. Thus the particular wireless sensing system is suitable for remote real-time unattended industrial environment monitoring. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Wireless sensing systems and wireless sensor networks (WSNs) have emerged as a consequence of engineering smaller sized sensing devices characterized by low cost fabrication and low power operation, which enable many applications in diverse fields. Currently, they are very promising in several fields such as environmental science, agriculture, medicine, military surveillance and home health care or assisted living [1–10]. The WSNs are usually composed of a few sinks which in turn communicate with several sensors nodes also referred as motes. These motes can operate

∗ Corresponding author. E-mail addresses: [email protected] (P. Oikonomou), [email protected] (A. Botsialas), [email protected] (A. Olziersky), [email protected] (I. Kazas), [email protected] (I. Stratakos), [email protected] (S. Katsikas), [email protected] (D. Dimas), [email protected] (K. Mermikli), [email protected] (G. Sotiropoulos), [email protected] (D. Goustouridis), [email protected] (I. Raptis), [email protected] (M. Sanopoulou). http://dx.doi.org/10.1016/j.snb.2015.10.043 0925-4005/© 2015 Elsevier B.V. All rights reserved.

in a wide range of environments and provide advantages in cost, size, power, flexibility and distributed intelligence compared to conventional wired sensing solutions. Motes can change place and configuration, be added or removed while the continuous operation of the network remains attainable. Moreover, due to the inherent characteristics of the WSN, a mote can indirectly contact the network coordinator via multiple hops with several sophisticated stream environment scenarios [11]. The key components of an intelligent and smart WSN are usually several motes consisting of three units: sensing, processing and communication [12]. The sensors are analytical devices where a sensing material is applied onto a suitable physical transducer to convert a change in a property of a sensing material into a readable form of energy [13]. The energy-transduction principles that are usually employed for chemical sensing involve radiant, electrical, mechanical and thermal types of energy. The analog signals produced by the sensors are converted to digital signals and are afterwards guided to the processing unit [14]. At this step, the signal obtained from the transducer is processed to provide useful information about the concentration of species in the sample of interest.

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

The motes, with their embedded CPUs, RF communication module and sensor units, gather information from the surrounding environment and communicate with each other via a gateway unit to send the measured data to a base station for further processing [15]. These gateway units can also communicate with other computers via Internet, building Internet of Things (IoT) [16–18]. The monitoring, evaluation and control of the air quality within facilities workspace, with the use of WSN appears to be a promising trend in industrial applications. Thus, on-site measurements in real time, rather than collecting samples for off-line laboratory analysis, are of great importance. In this direction the miniaturization of detection systems is one of the major trends. Toward this objective, a mote is developed operating in Zigbee communication protocol. In this study a hybrid sensor array composed by 8 interdigitated chemocapacitors (IDCs) with the appropriate read-out electronics is used as the analytical device. The design/fabrication of the interdigitated electrodes (IDEs) layout is optimized in terms of sensing performance and is in accordance with the selection of the polymeric materials that will coat the sensing area and the read-out electronics specifications. Then this system is connected with a wireless node that is guided and coordinated by a gateway unit. Evaluation of the wireless sensing system is performed under laboratory conditions upon exposure to gaseous environments, simulating the workspace of a specific printed flexing packaging industry. Effective safety precaution measures, relating both to the flammability of the solvents [45% of the lower explosion limit (LEL)] as well as to the upper limit of safe exposure of the employees (time weighted average, TWA) require the continuous monitoring of the gaseous environment composition in the working areas. In particular, in flexographic or rotogravure printing technologies ethyl acetate is used as the main ink-solvent. Thus for the case of interest, focusing mainly on ethyl acetate vapor, the corresponding values are 9000 ppm and 400 ppm for 45% LEL and TWA respectively [19]. The system is also tested in sensitivity/selectivity to ethanol vapor, since it is reported that at

267

industries established at countries with warm climates (e.g. Southern Europe) in order to sustain the fluidity of the ink solutions, ethyl acetate is often mixed for certain inks with low molecular weight alcohols. 2. Realization of the sensing system An overview of a WSN for real time application is illustrated in Fig. 1. Each mote comprises a wireless node connected with the sensing unit developed. The motes communicate with a gateway unit that collects the various data and further transmits them to a computer for monitoring and further processing with appropriate software. In this study, we investigate the realization and the evaluation of the sensing performance of a wireless sensing system, as the core unit of such a network. Details about the sensing unit which consists of the sensor array and the read out electronics are presented below. 2.1. Fabrication of the sensor array The sensor array should be appropriately designed for the targeted application. The use of chemocapacitor type sensors has been chosen due to previous reported promising results of their sensing performance (e.g. selectivity/sensitivity, reproducibility) for complex gaseous environments in conjunction with low power consumption [20–24]. The dielectric layer of the chemocapacitors, which acts as the sensing material, is a polymeric layer with appropriate sorption properties for certain analytes. The configuration of planar IDEs addresses the need of feasible fabrication and appropriate personalization through the application of suitable polymeric material via drop casting or inkjet printing [25–27]. Two interpenetrating comb metallic electrodes with welldefined geometry are fabricated on Si wafers with a thick layer of SiO2 on top. The thickness of the SiO2 layer should fulfill the basic requirement that it should be higher than half the spatial

Fig. 1. Overview of a wireless sensor network (WSN) for real time applications.

268

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

Fig. 2. SEM images of: (a) IDEs with critical dimension 1 ␮m and (b) SU-8 well formed around every sensor of the sensor array.

periodicity ( = 2(W + G)) of the metallic electrodes in order to avoid the propagation of the electric field through the substrate, which results in adverse short-circuit effects and cross-talk between chemocapacitors of the same chip [28]. The selection of the IDEs optimum layout design is based on previous theoretical studies investigating the gain in terms of sensitivity as electrodes critical dimension is miniaturized [29]. Then other parameters were taken into account such as the maximum number of feasible fabricated sensors on the same chip and conformity to the read-out electronics specifications. The latter is inevitably associated with the selection of the polymeric sensing layer. As it is discussed below, the design selection of a sensor array consisting of 8-IDCs with different IDE layout configurations of 2.0 ␮m or 1.0 ␮m critical dimensions on the same chip seems to be the optimum for the targeted application. The fabrication flow-chart employs standard micromachining/microelectronic processing steps. Following previous studies, a bi-layer of PMMA/PEDOT:PSS can be used for the patterning of the IDEs structures via e-beam lithography (EBL) exposure step [26]. The PEDOT:PSS conductive polymeric film acts as charge dissipation layer in order to avoid charging of the highly insulating substrate during beam writing which leads to deflection of the electron beam and results in pattern distortion [30,31]. However, the use of a multi layer resist for patterning the desired IDEs layout increases the process complexity and the probability of defects, due to additional spin coating and thermal processing steps. In this study a fabrication procedure, different to that of Ref. [26], is followed. Only a thick layer of a high resolution chemically amplified positive resist (UV5) with high sensitivity and contrast [32] is used for the patterning of the IDEs structure. Then steps of metallization and lift-off are followed as in Ref. [26]. After the realization of the metallic IDEs on the thick SiO2 layer, a well of 50 ␮m height of negative epoxy resist layer (SU-8 3050 [33]) is formed around the IDE areas through standard I-line photolithographic processing, Fig. 2. The cross-linked SU-8 well serves to define the area where the polymeric layer will be drop casted. Finally the sensing polymeric layers are deposited on the IDEs sensing areas, composing the 8-interdigitated capacitor (IDC) sensor array.

Devices) based data acquisition system is used. The AD7746 is a high resolution, - capacitance-to-digital converter (CDC). The architecture features inherent high resolution (0.5 fF/24bit), high linearity (±0.01%) and high accuracy (±4 fF). It supports I2C-compatible and two-wire serial interface [34]. From the two capacitance channels that are available in the AD7746 chip, only one is used in the present case. Temperature variations in the environment are monitored by the temperature sensor embedded in the AD7746 chip. The AD7746 chip can be brought to a fairly low dissipation level (4.75 mW) at reduced acquisition rate which perfectly matches the particular application, while its stand-by operation is even lower (2.35 mW). For the scanning of all sensors on the chip, a commercially available analog multiplexer (ADG407 (Analog Devices)) is employed along with an 8-bit I/O expander. The communication with the wireless unit is attained since both the capacitance to digital converter and the I/O expander support the I2C protocol. The total footprint of the sensing unit is 10 cm2 and is presented in Fig. 4. The mote is battery-operated.

2.2. Sensor read-out electronics The electronics module of the miniaturized system allows for the measurement of the capacitance of all sensors on the chip and for the data transfer to the wireless unit. The layout of the electronics module is illustrated in Fig. 3. An AD7746 (Analog

Fig. 3. Block diagram of the read-out and control electronics module.

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

Fig. 4. Image of the proposed mote: the sensing unit is a hybrid system consisting of the sensor array and the read-out electronics (right image). The sensing unit is connected with the wireless node (left image).

2.3. Selection of polymeric materials The selection of the polymeric-sensing materials depends on the targeted application. In the present study the polymers were selected on the basis of different sensitivity/selectivity to the vapor analytes of interest, i.e. mainly to ethyl acetate (EtOAc) and water and in addition to ethanol (EtOH), as it was estimated by their sorptive capability. This was accomplished by exposing thin films of several polymeric materials to vapor analytes and determining the equilibrium thickness expansion of each film by white light reflectance spectroscopy (WLRS) [35,36]. For the monitoring of VOCs in the presence of humidity, IDCs coated with polymeric materials relative hydrophobic, but sensitive/selective to these particular compounds should be used. For example, studies on the pervaporation of water–ethyl acetate mixtures through the dense poly(dimetylsiloxane) (PDMS) membranes exhibit a large perm-selectivity scale for ethyl acetate [37]. This behavior is also verified by the results of corresponding swelling measurements by WLRS where the relative sensitivity and selectivity to each vapor analyte of interest is estimated, Table 1. The sensitivity is expressed as the ratio of the volume fraction of the sorbed analyte (␸s) over the vapor analyte concentration (Cg), Table 1 Sorption capability of polymeric materials upon exposure to vapor of analytes of interest. Results obtained by swelling measurements via white light reflectance spectroscopy (WLRS). Volume fraction of sorbed analyte/vapor analyte concentration, ␸s/Cg = (dL/Lo)/Cg (ppm−1 ) Polymer

EtOAc

EtOH

Water

PDMS (RTV615) PBMA PHEMA

5.96·10−7 7.21·10−7 6.26·10−8

1.33·10−7 4.88·10−7 1.45·10−6

1.26·10−8 6.81·10−8 2.22·10−6

269

(␸s/Cg). As it is shown in Table 1, room temperature vulcanized poly(dimethylsiloxane) (PDMS(RTV615) [38]) and poly(n-butyl methacrylate) (PBMA) thin film layers exhibit at least one order of magnitude higher sensitivity to ethyl acetate than to humidity. Accordingly, PDMS (RTV615) and PBMA coated sensors are suitable for the targeted application. On the contrary, for monitoring the humidity levels with high accuracy, the IDC sensor response should be ideally inert to the presence of other volatile compounds. For that purpose, highly selective hydrophilic polymers such as poly(2-hydroxy ethyl methacrylate) (PHEMA) are applied. In the case of PHEMA, the sensitivity to humidity is at least one order of magnitude larger than that to ethyl acetate. On the other hand, the read-out electronics module has a measurement range of 0–21 pF. This means, that the initial capacitance value and the sensors responses shifts should fall in this particular range. By employing the simulation tool of the capacitance response of IDCs, induced by dielectric permittivity changes of the sensing layer upon the sorption of analyte molecules [29], it was found that for the majority of the polymeric materials, as PBMA and PDMS, which exhibit a low dielectric permittivity value (εs ∼ 2–3), a configuration of IDCs with 1.0 ␮m and sensing area of 0.7 mm2 is translated to sensor responses in the desirable capacitance range. Further miniaturization of the sensing area contributes only to minor enhancement of the sensitivity while the resulting reduction of the sensing area in order to meet the read-out electronic specifications is restrictive due to fabrication constraints and increased cost. However, for a sensing area of 0.7 mm2 , the use of polymeric materials with relative high dielectric permittivity value or IDCs with predicted shifts beyond the upper limit of the detectable scale, as in the case of PHEMA, should be addressed with the use of IDEs with 2.0 ␮m critical dimension. Accordingly, for the particular application, the following polymeric materials were used: PHEMA, PDMS (RTV615) and PBMA. The PHEMA is coated on IDEs with 2.0 ␮m critical dimension while the other two polymers on IDEs with 1.0 ␮m critical dimension. For stability–reliability tests or even sensor array’s self-calibration, more than one sensors coated with the same polymer are used on the same 8 sensor array chip. In particular, two IDEs with critical dimension 2.0 ␮m are coated with PHEMA while the other six IDEs with critical dimensions 1.0 ␮m are separated in two groups of three same coated polymeric material, PBMA and PDMS (RTV615) respectively. 2.4. Communication unit The wireless node is based on a RISC microcontroller with very low power consumption, creating a flexible portable platform for data collection. The wireless node collects the signals from the sensor read-out electronics via I2C protocol, proceeds to a first level signal processing and transmits the data to a PrismaSense ZigBee to WiFi Gateway wireless communication unit [39]. This communication unit serves as the ZigBee sensor network coordinator, as well as the bridge between the ZigBee and the WiFi network. The communication is bidirectional; sensor data are sent via the Gateway to the server and vice versa. The data can then be available to every PC that runs a front-end software for on-site processing. In order the wireless node to be low powered and autonomous, it can be in a variety of states depending on the operation. The various states are signal collection, signal processing, data transmission, data reception, sleep with the most power demanding state being the data transmission. The power required for transmission is at around 200 mW. So, depending on the requirements of the application; frequency of signal collection, amount of data processing, frequency of data transmission, period of sleep time, battery that powers the unit, the autonomy of the node can be from some days to some months.

270

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

3. Evaluation of the sensing system The evaluation of each sensor of the sensor array is performed, mainly under dynamic laboratory conditions, by the bubbling technique with the use of a computer controlled, via LabView software, vapor delivering experimental set up [20,21]. In addition, capacitance measurements were performed under static conditions in hermetically closed vessel by injecting through a Hamilton syringe minute quantities of liquid analyte. The equilibrium vapor analyte concentration in the vessel was measured by a commercial detector calibrated for EtOAc vapor (TechnoControl, #TS293PX). The raw data are transferred and monitored in the appropriate interface program. Then further processing is utilized with principal component analysis (PCA). In Fig. 5 typical responses during a dynamic measurement with a wireless mote upon exposure to vapors of pure components and their mixtures is illustrated. Plots of an overall view of the sensor array responses and zooms in selected sensors plots as monitored in the built-in interface program are demonstrated. In particular, the sensing system is exposed to increasing humidity concentrations, then to different concentrations of EtOAc vapor and to mixtures of ethyl acetate vapor and humidity. As it is shown the PHEMA-coated sensors are very sensitive to changes in the humidity concentration and practically inert to concentration changes of the weakly polar EtOAc (down right image – Ports 1 and Port

5). On the contrary the PDMS (RTV615) and PBMA-coated sensors show sensitivity/selectivity to EtOAc vapor as compared to humidity. Moreover, for those relative hydrophobic sensors, the change in EtOAc vapor concentration in the presence of constant humidity level is measurable. Then the equilibrium responses are calculated and plotted vs. the corresponding vapor concentration in Figs. 6 and 7. From the data depicted in Fig. 6, details about the sensitivity/selectivity to the vapor analytes of interest for a PBMA and a PDMS (RTV615) coated sensor are obtained. These sensors are more sensitive/selective to EtOAc vapor rather than humidity or ethanol. This behavior is in good agreement with the results of the swelling measurements as illustrated in Table 1. In particular via the swelling measurement values of Table 1, the ratio of the sorption sensitivities to ethyl acetate and ethanol (ratio of the slopes, ␸s/Cg) for PBMA and PDMS (RTV615) polymeric films is calculated to be 1.5 and 4.5 respectively. Meanwhile, the corresponding values of 1.5 and 2.8 are obtained for the respective sensitivities ratio as estimated by the capacitance measurements, Fig. 6. In the latter case the sensitivities are expressed as ratio values of normalized equilibrium responses (dC/Co, where Co the capacitance value of the dry polymer coated sensor) to analyte concentration (Cg). The equilibrium responses of the PBMA and PDMS (RTV615) coated sensors of the sensor array upon exposure to certain vapor concentrations of EtOAc under dynamic and static measurements

Fig. 5. Recorded data of a dynamic measurement of the wireless mote upon exposure to different concentrations of pure components and their mixtures, at 27 ◦ C: humidity (20–50% R.H.), EtOAc (1600–4800 ppm) and binary mixtures of 50% R.H. with EtOAc: 1600–4800 ppm. Plots in the interface software: an overall view of the sensor array responses (upper left plot) and several zooms in selected sensors. [Port Numbers: 2, 3 and 4 correspond to PBMA-coated sensors, Port Numbers: 1 and 5 correspond to PHEMA-coated sensors and Port Numbers: 6, 7, and 8 correspond to PDMS (RTV615)-coated sensors.].

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

271

Fig. 7. Reproducibility test: (a) equilibrium responses of the groups of (i) PBMA and (ii) PDMS-coated sensors of the sensor array upon exposure to different concentrations of EtOAc with two different gas transfer methods at 27 ◦ C. (b) The low concentration range data in Fig. 7b showing the very good agreement between the static measurement results and the linear fit obtained by the dynamic measurement. Fig. 6. Sensitivity/Selectivity test. Equilibrium responses of: (a) PBMA coated sensor and (b) PDMS (RTV615) upon exposure to different concentrations of EtOAc, EtOH and water at 27 ◦ C. The polymers are coated on IDEs with 1.0 ␮m critical dimension.

are shown in Fig. 7. In the dynamic measurements the minimum concentration was 1600 ppm while lower concentrations down to 400 ppm (TWA value) were achieved in static measurements. The sensors are separated in two groups. Each group comprising three sensors coated with the same polymer. The dC (fF) vs. Cg (ppm) plots are linear in all cases, due to the low Cg range studied, and the sensitivity (expressed as dC/Cg) for the specific polymer–analyte system is calculated from the respective slopes of the plots. As illustrated in Fig. 7 the results of the static measurements, for the tested concentration range: 400–1600 ppm, are in very good agreements with the linear fit obtained by the dynamic measurements. From the dC/Cg values, in conjunction with the acceptable signal to noise ratio (S/N ≥ 3), the limit of detection (LOD) by extrapolation for EtOAc vapor is estimated to be lower than the TWA value (400 ppm). In particular for a noise of 0.2–0.3 fF, a capacitance signal change of 1.0 fF is detectable, which leads to a LOD value of 45 ppm and 88 ppm for the case of PBMA and PDMS (RTV615) coated sensors respectively. Also, for EtOAc vapor, the upper limit of interest corresponding to 45% of LEL value (9000 ppm) is measurable, Fig. 6. The calculated sensitivities also highlight the repeatability of the sensing performance of sensors coated with the same polymer. Since these sensors demonstrate a good performance in terms of repeatability they were also tested for long-term stability demands. This is a first test toward their evaluation for long-term operation

in real industrial environment. Mean values of normalized equilibrium responses at each vapor concentration, calculated from a set of 4 successive different measurements over a six months period are shown in Fig. 8. The results indicate a very good behavior in terms of aging-stability since the maximum observed standard deviation for a given mean value is below 10%. The targeted application is the monitoring of the workspace of a printed flexing packaging industry. Therefore evaluation

Fig. 8. Long-term stability test: normalized equilibrium responses of groups of (i) PBMA and (ii) PDMS, coated sensors of the sensor array upon exposure to different concentrations of EtOAc under same laboratory conditions over a 6 month period.

272

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

Fig. 9. Typical example of a dynamic measurement at 27 ◦ C. The exposure steps include several substeps with increasing humidity level (20, 30, 40, 50% R.H.) (step 1), increasing EtOAc vapor concentration (1600, 3200, 4800 ppm) (steps: 2, 4b, 5b, 6b) and increasing EtOH vapor concentration (1000, 2000, 3000 ppm) (steps: 3, 4a, 5a, 6a).

of the sensing system is carried out in laboratory upon exposure to gaseous environments simulating the corresponding real environments. In flexographic or rotogravure printing technologies EtOAc is the main ink-solvent. So, the sensing system is mainly tested upon exposure to binary mixtures of humidity and EtOAc vapors. Also binary mixtures of humidity and EtOH vapors were tested. A typical dynamic measurement of a PDMS (RTV615) sensor of the sensor array is shown in Fig. 9. For both PDMS (RTV615) and PBMA coated sensors the change in VOCs concentration is measurable, even in the presence of high humidity levels. The sensor response upon exposure to binary mixtures is approximately equal to the sum of the capacitance changes that are induced by the pure components of the mixture (C(mixture) = C(component)1 + C(component)2). Further data post-processing is performed with principal component analysis (PCA). PCA does not provide the direct information on chemical composition of a mixture or concentrations of its compounds, but it represents both, indirectly. The results were autoscaled, in order to prevent high sensor responses from dominating the analysis and loose information from sensors with low responses respectively. Fig. 10 shows the PCA score for data obtained upon exposure to EtOAc vapor over a concentration range up to 55% LEL, humidity concentrations representative of

Fig. 11. PCA score of the sensor array equilibrium responses upon exposure to vapors of pure ethyl acetate, ethanol, humidity and several selected binary mixtures of them with different concentrations.

real industrial workspace and several binary mixtures of EtOAc and humidity. The discrimination between values corresponding to gaseous environments of different composition and different concentration is attainable. The PCA values for pure EtOAc and humidity increase linearly, but with totally different slopes, at the PCA score area. Moreover, in the presence of a certain relative humidity (R.H.) level, the PCA values increase linearly with EtOAc concentration and in a different PCA score area. This linearity of the PCA values is implied by results from the linear responses of the sensors to the analytes of interest. For gaseous environment with a certain EtOAc concentration, when humidity level is changing, the PCA values shift parallel to the slope corresponding to PCA values of pure EtOAc. Additionally for gaseous environment with constant humidity level when EtOAc concentration is increasing, the PCA values shift is parallel to the slope corresponding to values of pure humidity. Additionally, Fig. 11 shows the PCA score for pure EtOAc, EtOH and humidity and several binary mixtures of them. In the latter case a narrower region of VOCs concentration is examined in order to investigate the discriminating capability under more complex conditions. The PCA values for the VOCs, the humidity and their mixtures exhibit the same behavior as in Fig. 10. So, even though the PCA plot is not a quantitatively analytical tool, it can be used for control and evaluation of the gaseous environment of the targeted application. 4. Conclusions

Fig. 10. PCA score of the sensor array equilibrium responses upon exposure to vapors of pure ethyl acetate over a wide concentration range, humidity and several selected binary mixtures of them with different concentrations.

This paper has reported the development and characterization of a wireless sensing system to be used for the monitoring of the workplace environment of an industrial installation. The sensor unit of the system has been appropriately designed with the use of chemocapacitor type sensors which were selected due to previous promising results on their sensing performance (e.g. selectivity/sensitivity, reproducibility) in complex gaseous environments [20–23]. Several requirements and specifications were taken into account: the demand for low and practicable fabrication process of the sensing unit, the need for interconnection with miniaturized read-out electronics and the communication unit (wireless node), as well as the need for sensitive/selective polymeric materials for the targeted application. The wireless sensing system was tested under laboratory conditions upon exposure to certain analytes of interest simulating the workspace environment of the real application. The results

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

showed that the sensing system is characterized by very good sensing performance with high repeatability and long-term stability. Additionally, the estimated LOD values by extrapolation for EtOAc vapor are 45 and 88 ppm for PBMA and PDMS (RTV615) coated sensors respectively, which are considerably lower than the corresponding TWA value (400 ppm). Furthermore PCA was used for data post-processing. The PCA values increase linearly with both VOCs and humidity concentrations, but are located at different PCA score area. Due to this behavior, the discrimination between values corresponding to gaseous environments of different composition and different concentration is attainable. Concluding, the wireless sensing system, due to its low power consumption, has the potential to be used for remote real-time unattended monitoring of specific industrial environments. Acknowledgements This work was supported by the research project entitled “Autonomous and Integrated system for in situ and continuous contaminant gases in industrial environments (ALEPOY)”, co-financed by the European Union (European Regional Development Fund – ERDF) and Greek national funds through the Operational Program “Competitiveness and Entrepreneurship” of the National Strategic Reference Framework (NSRF) – National Action “Support of New Enterprises & SMEs” (MIA – RTDI, Contract No. 19SMEs2010). References [1] L. Ruiz-Garcia, L. Lunadei, P. Barreiro, I. Robla, A review of wireless sensor technologies and applications in agriculture and food industry: state of the art and current trends, Sensors 9 (2009) 4728–4750. [2] M.D. Steinberg, I. Zˇ urab, I. Murkovic Steinberg, Wireless smart tag with on-board conductometric chemical sensor, Sens. Actuators B 196 (2014) 208–214. [3] R.A. Potyrailo, N. Nagraj, C. Surman, H. Boudries, H. Lai, J.M. Slocik, N. Kelley-Loughnane, R.R. Naik, Wireless sensors and sensor networks for homeland security applications, TrAC – Trends Anal. Chem. 40 (2014) 133–145. ´ M.D. Steinberg, I.M. Steinberg, Low-cost conductometric [4] I. Zˇ ura, D. Babic, transducers for use in thin polymer film chemical sensors, Sens. Actuators B 193 (2014) 128–135. [5] S. De Vito, P. Di Palma, C. Ambrosino, E. Massera, G. Burrasca, M.L. Miglietta, G. Di Francia, Wireless sensor networks for distributed chemical sensing: addressing power consumption limits with on-board intelligence, IEEE Sens. J. 11 (2011) 947–955. [6] R. Shepherd, S. Beirne, K.T. Lau, B. Corcoran, D. Diamond, Monitoring chemical plumes in an environmental sensing chamber with a wireless chemical sensor network, Sens. Actuators B 121 (2007) 142–149. [7] J.V. Capella, A. Bonastre, R. Ors, M. Peris, A step forward in the in-line river monitoring of nitrate by means of a wireless sensor network, Sens. Actuators B 195 (2014) 396–403. [8] R.A. Potyrailo, C. Surman, W.G. Morris, S. Go, Selective detection of chemical species in liquids and gases using radio-frequency identification (RFID) sensors, in: Solid-State Sensors, Actuators and Microsystems Conference, TRANSDUCERS, 2009, pp. 1650–1653. [9] L. Lamont, M. Toulgoat, M. Deziel, G. Patterson, Tiered wireless sensor network architecture for military surveillance applications, in: The Fifth International Conference on Sensor Technologies and Applications, SENSORCOMM, 2011, pp. 288–294. [10] S.C. Mukhopadhyay, Wearable sensors for human activity monitoring: a review, IEEE Sens. J. 15 (2015) 1321–1330. [11] G.-J. Horng, T.-Y. Chang, S.-T. Cheng, An effective node-selection scheme for the energy efficiency of solar-powered WSNs in a stream environment, Expert Syst. Appl. 41 (2014) 3143–3156. [12] I.F. Akyildiz, W. Su, Y. Sankarasubramaniam, E. Cayirci, Wireless sensor networks: a survey, Comput. Netw. 38 (2002) 393–422. [13] J. Janata, Principles of Chemical Sensors, 2nd ed., Springer Science & Business Media, 2009. [14] Z. Huang, J. Zhu, L. Lu, An AD7746-based data acquisition system for capacitive pressure sensor in weather detection application, Key Eng. Mater. 483 (2011) 461–464. [15] W.-S. Jang, W.M. Healy, M.J. Skibniewski, Wireless sensor networks as part of a web-based building environmental monitoring system, Autom. Constr. 17 (2008) 729–736. [16] R. Piyare, S.R. Lee, Towards internet of things (IoTs): integration of wireless sensor network to cloud services for data collection and sharing, Int. J. Comput. Netw. Commun. (IJCNC) 5 (2013) 59–72.

273

[17] P.A.C.D.S. Neves, J.J.P.C. Rodrigues, Internet protocol over wireless sensor networks, from myth to reality, J. Commun. 5 (2010) 189–196. [18] S.D.T. Kelly, N.K. Suryadevara, S.C. Mukhopadhyay, Towards the implementation of IoT for environmental condition monitoring in homes, IEEE Sens. J. 13 (2013) 3846–3853. [19] http://www.osha.gov. [20] A. Botsialas, P. Oikonomou, D. Goustouridis, T. Ganetsos, I. Raptis, M. Sanopoulou, A miniaturized chemocapacitor system for the detection of volatile organic compounds, Sens. Actuators B 177 (2013) 776–784. [21] P. Oikonomou, I. Raptis, M. Sanopoulou, Monitoring and evaluation of alcoholic fermentation processes using a chemocapacitor sensor array, Sensors 14 (2014) 16258–16273. [22] T.E. Mlsna, S. Cemalovic, M. Warburton, S.T. Hobson, D.A. Mlsna, S.V. Patel, Chemicapacitive microsensors for chemical warfare agent and toxic industrial chemical detection, Sens. Actuators B 116 (2006) 192–201. [23] S.V. Patel, W.K. Tolley, Developments toward a low-cost approach for long-term, unattended vapor intrusion monitoring, Analyst 139 (2014) 3770–3780. [24] A. Oprea, J. Courbat, N. Bârsan, D. Briand, N.F. de Rooij, U. Weimar, Temperature, humidity and gas sensors integrated on plastic foil for low power applications, Sens. Actuators B 140 (2009) 227–232. [25] A.M. Kummer, A. Hierlemann, Configurable electrodes for capacitive-type sensors and chemical sensors, IEEE Sens. J. 6 (2006) 3–10. [26] P. Oikonomou, A. Botsialas, A. Olziersky, D. Goustouridis, A. Speliotis, I. Raptis, M. Sanopoulou, Chemocapacitive sensor arrays on Si substrate: towards the hybrid integration with read-out electronics, Microelectron. Eng. 119 (2014) 11–15. [27] U. Altenberend, F. Molina-Lopez, A. Oprea, D. Briand, N. Bârsan, N.F. De Rooij, U. Weimar, Towards fully printed capacitive gas sensors on flexible PET substrates based on Ag interdigitated transducers with increased stability, Sens. Actuators B 187 (2013) 280–287. [28] R. Igreja, C.J. Dias, Analytical evaluation of the interdigital electrodes capacitance for a multi-layered structure, Sens. Actuators A 112 (2004) 291–301. [29] P. Oikonomou, G.P. Patsis, A. Botsialas, K. Manoli, D. Goustouridis, N.A. Pantazis, A. Kavadias, E. Valamontes, Th. Ganetsos, M. Sanopoulou, I. Raptis, Performance simulation, realization and evaluation of capacitive sensor arrays for the real time detection of volatile organic compounds, Microelectron. Eng. 88 (2011) 2359–2360. [30] R. Abargues, U. Nickel, P.J. Rodriguez-Canto, Charge dissipation in e-beam lithography with Novolac-based conducting polymer films, Nanotechnology 19 (2008), 125302. [31] K. Mohamed, M.M. Alkaisi, R.J. Blaikie, Surface charging suppression using PEDOT/PSS in the fabrication of three dimensional structures on a quartz substrate, Microelectron. Eng. 86 (2009) 535–538. [32] S. Thoms, D. Macintyre, Process optimisation of DUV photoresists for electron beam lithography, Microelectron. Eng. 46 (1999) 287–290. [33] http://www.microchem.com. [34] http://www.analog.com/static/imported-files/data sheets/AD7745 7746.pdf. [35] K. Manoli, D. Goustouridis, S. Chatzandroulis, I. Raptis, E. Valamontes, M. Sanopoulou, Vapor sorption in thin supported polymer films studied by white light interferometry, Polymer 47 (2006) 6117–6122. [36] K. Manoli, D. Goustouridis, I. Raptis, E. Valamontes, M. Sanopoulou, Vapor-induced swelling of supported methacrylic and siloxane polymer films: determination of interaction parameters, J. Appl. Polym. Sci. 116 (2010) 184–190. [37] N. Stafie, D.F. Stamatialis, M. Wessling, Effect of PDMS cross-linking degree on the permeation performance of PAN/PDMS composite nanofiltration membranes, Sep. Purif. Technol. 45 (2005) 220–231. [38] http://www.momentive.com. [39] http://www.prismaelectronics.eu.

Biographies Dr. Petros Oikonomou received the B.Sc. degree in Chemistry from the Chemistry Department of Ioannina University (Greece) in 2003. He received the M.Sc. and the Ph.D. degree in “Polymers and their Applications” from the Chemistry Department of University of Athens in 2005 and 2012, respectively. Both M.Sc. and Ph.D. thesis research were performed at the Institutes of Microelectronics and Physical Chemistry at National Center for Scientific Research – NCSR, “DEMOKRITOS” – Greece. Currently, he is working as a Research Fellow at Institute of Nanoscience and Nanotechnology at NCSR “DEMOKRITOS” Greece. His research is focusing on the field of chemical microdevices with emphasis in gas sensing applications. His research interests include microfabrication, polymers, (bio)chem sensing applications and physical chemistry. Athanasios Botsialas received the B.S. degree in Electronics Engineering from the TEI of Piraeus (2005) and the M.Sc. in Microelectronics from the National and Kapodistrian University of Athens (2008). From 2006 he is working as Electronics Engineer at the Institute of Microelectronics, NCSR “Demokritos” and from 2008 he is working as laboratory associate at the Electronic Engineering Department TEI of Piraeus. His interests include instrumentation development, software development, experimental device characterization, design of data acquisition and automation applications for prototype measurement setups.

274

P. Oikonomou et al. / Sensors and Actuators B 224 (2016) 266–274

Antonis Olziersky received his B.Sc. in Physics from the Aristotle University of Thessaloniki, Greece, in 2000. He joined the Institute of Microelectronics of NCSR “Demokritos” in 2000 where he received both his M.Sc. and Ph.D. in Microelectronics from the Department of Informatics of the National University of Athens, Greece in 2003 and 2006, respectively. He has worked in the field of fabrication and characterization of nanocrystal memory devices and electron beam lithography. In 2007 he obtained a “Juan de la Cierva” postdoctoral position at the Electronics Department of the University of Barcelona in Spain. He has worked on characterization of amorphous metal oxide materials for large area electronics, inkjet printing of functional materials, focused ion beam patterning. In 2010 he moved back to the Institute of Microelectronics of NCSR “Demokritos” to work as a postdoctoral research associate in electron beam lithography. In November 2014 he joined the IBM Research Laboratory in Zurich, Switzerland. Ioannis Kazas is currently pursuing M.Sc. in Microelectronics at National and Kapodistrian University of Athens. Main research interests include solid state detectors and associated electronics, for spectroscopic and digital imaging applications. Ioannis Stratakos received his B.Sc. in Electrical & Computer Engineering from the National Technical University of Athens in 2013. He participated in the development of Embedded Software in Prisma Electronics. He is currently doing his M.Sc. in Information Systems at the Technische Universität of München. His interests include mainly embedded systems development. Serafim Katsikas received his B.Sc. in Electrical & Computer Engineering from the National Technical University of Athens in 1998. He is the Head of Prisma Electronics R&D department. He has participated in a variety of research programs in Greece and internationally. He is member of the board of directors of the Hellenic Semiconductor Industry Association (Hellenic-SIA) while participates in various International Congresses. Dimitrios Dimas received the B.Sc. degree in Physics and M.Sc. in Electronics from the Physics Department of Aristotle University of Thessaloniki (Greece) in 2004 and 2008, respectively. He has participated in a variety of research projects mainly focused in environmental sensing applications. Currently, he works as manager of the Installations Department of Prisma Electronics. Konstantina Mermikli received her B.Sc. in Electrical & Computer Engineering of the Faculty of Engineering of the Aristotle University of Thessaloniki in 2008. She currently works as the manager of the Software Development Department of Prisma Electronics. Her interests are focused on embedded systems and backend software development. George Sotiropoulos currently works as technical manager at Alfa Beta Roto S.A. His main topic is Environmental Management System specifications focusing on

the: solvent recovery, distillation of used ink, rinsing rotogravure tanks, rotogravure cylinder recycling and waste minimization. Dr. Dimitris Goustouridis received the B.S. in 1992 from the Department of Physics of the University of Patras. In 2002 he received the Ph.D. degree in microelectronics from the Department of Applied Sciences of National Technical University of Athens for his work on capacitive type pressure sensors. From 2000 until 2010 he was working as research associate at the Institute of Microelectronics of NCSR “Demokritos”. His current position is Lecturer in Department of Electronics of TEI Piraeus. His interests include silicon micromachining, capacitive pressure sensors, optical sensors, chemical and biological sensing devices and measurements set-up for sensors characterization. He is/was Key Researcher in several research projects in the areas of micro-fabrication and bio(chem) sensors funded by EU (FP6, FP7), national funding agencies and industries. He is author more than 50 publications in international journals and holder of 3 patents. Ioannis Raptis received his Ph.D. on e-beam lithography (1996) from Physics Department, University of Athens. The experimental part of his Ph.D. thesis was carried out at IESS-CNR (Rome, Italy). From 2003 he works at NCSR ‘Demokritos’ as researcher on the implementation of technologies and electronic/photonic devices in the micro/nano scale for bio(chem) sensing applications. At the end of 2008 he co-founded Theta Metrisis as the first spin-off company of the Institute of Microelectronics NCSR ‘Demokritos’ exploiting technologies and methodologies that have been developed in his team. He is/was Key Researcher and Coordinator for several national and EU (FP6, FP7) funded research projects in the areas of nanotechnology and microsystems. He is program and steering committee member in several international conferences and serves regularly as reviewer in scientific journals and as invited speaker in conferences/summer schools. He is author of more than 130 publications in international journals and holder of 4 patents. Dr. Merope Sanopoulou received her B.Sc. degree in Chemistry from the University of Athens in 1977 and her Ph.D. in Physical Chemistry from the same institution in 1984. She then joined the Institute of Physical Chemistry, NCSR “DEMOKRITOS”, currently being Head of the Laboratory of Transport of Matter Phenomena in Polymers. She has been engaged in extensive modeling and experimental studies of sorption and diffusion in polymer films, focusing in transport-related applications of polymers, such as controlled release devices, transport properties in polymeric hydrogels, characterization of polymeric membranes for separation processes, and performance evaluation of polymer-based capacitive sensors. She is the author/co-author of more than 70 publications in peer-reviewed journals and has been key researcher or project leader in various National and EU research projects.