Share PDF

Search documents:
  Report this document  
    Download as PDF   
      Share on Facebook



Low-Cost Capacitive Humidity Sensor for Application Within Flexible RFID Labels Based on Microcontroller Systems

José Pelegrí-Sebastiá, Eduardo García-Breijo, Javier Ibáñez, Tomas Sogorb,

Nicolás Laguarda-Miro, and José Garrigues

Abstract—This paper reports on the fabrication of a capacitive- type relative humidity (RH) sensor using screen printing processes for electrode film deposition. The applied measurement method based on microcontrollers is also reported. In this specific case, the microcontroller is used to measure RH by means of a capacitive sensor with a simple low-cost electronic system. In addition, a comparison between two different types of polyester substrates [Melinex (DuPont) and CG3460 (3M)] is shown. Both polyester substrates have similar properties, and only the thickness is differ- ent (175 µm for Melinex and 100 µm for CG3460). A nonlineal response has been obtained in this type of sensors. In order to linearize the response and reduce the external hardware, an artificial neural network embedded into the microcontroller has been used.

Index Terms—Artificial neural network (ANN), capacitive-type sensor, humidity sensor, microcontroller, radio frequency identifi- cation (RFID) label, screen printing.


DUE TO the new technologies, now it is possible to make radio frequency identification (RFID) labels with a sensing stage. These labels include an antenna, a transmitter/ receiver, and a data memory (in the case of active or semiactive labels, a battery is also integrated). All these compounds can be integrated over a flexible film [1]–[6]. Consequently, it is possible to know the state of a product at any time using these stickers [7]. The sensor takes care of sending all the information to the memory, storing here all the events, such as changes in pressure, temperature, and humidity, according to the sensor


Sensors for specific parameters such as temperature and hu- midity are widely used in environmental monitoring, industrial production, etc. Furthermore, these sensors are being integrated into RFID labels. These sensors must be low cost and low power in RFID applications. There are two types of electri- cal impedance relative humidity (RH) sensors, i.e., capacitive

Manuscript received May 10, 2011; revised June 21, 2011; accepted July 20, 2011. Date of publication October 3, 2011; date of current version January 5, 2012. The Associate Editor coordinating the review process for this paper was Dr. V. R. Singh.

J. Pelegrí-Sebastiá and T. Sogorb are with the Instituto de Investigación para la Gestión Integrada de Zonas Costeras, Universidad Politécnica de Valencia, 46730 Grao de Gandia, Spain (e-mail:

E. García-Breijo, J. Ibañez, N. Laguarda-Miro, and J. Garrigues are with the Instituto de Reconocimiento Molecular y Desarrollo Tecnológico, Universidad Politécnica de Valencia, 46022 Valencia, Spain.

Color versions of one or more of the figures in this paper are available online at

Digital Object Identifier 10.1109/TIM.2011.2164860

[8]–[10] and resistive ones [10]–[12]. Capacitive sensors, as well as resistive ones, can operate in a wide RH range with simple electronics. Polymer films are used in low-cost appli- cations because they are quite cheap and have good moisture absorption.

In this paper, a low-cost low-power RH sensor is shown. The sensor has been developed over a polyester substrate by using thick-film technology in order to integrate it into an RFID label. Polymeric conductive inks with low-temperature cure are used.

In view of an important necessity to apply conductive inks in the sensors manufacture, with the notable advantages in intro- ducing them, several methods have been developed [13], [14]. Nowadays, the most used methods in printing manufacturing are screen printing (printing by serigraphy) and inkjet printing. Moreover, in the sensors’ final cost, these inks present some advantages such as print velocity and environmental care. This last advantage is the most important in order to reduce the final cost because it is not necessary to use toxic and dangerous products that are used in traditional printing methods, and the costs of these kinds of wastes can be saved.

In order to reduce the electronic systems in RFIDs, it is necessary to find basic systems for measuring the signal of the sensors [15]. There are two types of interface circuits to measure a capacitive RH sensor: 1) circuits based on the charge-transfer method [16], [17] and 2) oscillator circuits; however, some authors have proposed direct capacitive sensor- to-microcontroller interfaces, in which the capacitive sensor is directly connected to the microcontroller. Those two classic types of interface circuits are not applied to capacitive RH sensors because there is a lack of knowledge about the effects of the parasitic components of the sensor. A wide study about this effect has been developed in Reverter and Casas [18].


A.Sensor Design

The design is based on the measurement of capacitance vari- ations. The two most important parameters in sensor printing are geometrical design and the material of the active element. In order to configure the sensor, it is very important to know the best shape and the behavior of the material used in the development of the sensor because the shape modifies the sensor capacitance and the material modifies both conductivity and resistance.

0018-9456/$26.00 © 2011 IEEE



Fig. 1. (a) Design of a rectangular double layer. (b) Design with holes.

Fig. 3. Design of sensors “flag” type.

Fig. 2. Interdigitated sensor.

As far as sensor geometric design is concerned, there are two models that could modify the sensor behavior. They are called double-layer design and one-layer design, and there are differences between them. The double-layer design has an active material between two conductor ink layers. On the contrary, in the one-layer design, the air is usually the active material, although there are some exceptions (e.g., interdigi- tated electrodes).

With reference to sensors’ dimensions [20], it is necessary to pay attention to the size of the design because a variation in the conductive area involves a variation in the total capacitance and, therefore, a variation in humidity measurement.

Equation (1) explains the behavior of the sensor capacitance [1], where eo is the dielectric constant for vacuum, er is the relative dielectric constant of polyester, A is the conductive area, and a is the thickness of the active layer, i.e.,

C = eoer










because it is not a flat capacitive sensor. To analyze the behavior of the layer, the Maxwell equation must be used [1].

B. Sensor Fabrication

This equation shows that capacitance is directly proportional to the relative dielectric constants, vacuum dielectric constant, and sensor area and inversely proportional to thickness. Only one of these four variables changes with humidity: It is the relative dielectric constant of the polyester. As the relative dielectric constant is directly proportional to capacitance, it is possible to measure changes in humidity through capacitance changes.

1)Double Layer: Fig. 1(a) shows the rectangular design of a sensor. Capacitance, in this case, responds, as shown in (1). In order to increase the exposure to humidity, holes are made in the conductive layers. If these holes are made, a sensor in the shape of a flag or a grid is obtained [see Fig. 1(b)].

2)One Layer: This design has only one conductive layer, using air as dielectric among the track of the conductive layer. Fig. 2 shows this design. It is called interdigitated sensor.

The behavior of the one-layer design is not as easily calcu- lated as the double layer design one. In this case, (1) is not valid

Fig. 3 shows the design of the double-layer sensor (called “flag” type due to its form) and its dimensions (in millimeters), as well as a track to increase the area. The conductive area is obtained from the total area, the track area, and the hole. In this case, the conductive areas are 218 mm2 for the first one and 259 mm2 for the second one.

Two polyester films have been used as substrates, i.e., Melinex ST726 (DuPont) with 175 µm in thickness (Table I shows its properties) and CG3460 (3M) with similar properties but only 100 µm in thickness.

Knowing A, a, eo, and er,, the capacitance value can be obtained with (1). The theoretical capacitance is around 38 pF for the Melinex-based sensor and around 62 pF for the CG3460- based sensor.

A polymeric ink has been used, i.e., 5033 (DuPont). This ink is based on silver and has been developed for polyester substrates. Main ink parameters are 14 mΩ/sq./25 µm of resistivity and 15–30 Pa · s of viscosity.



Fig. 5. Basic diagram of electronic system.

Fig. 4. Sensor with DuPont 5033 over ST726.




Fig. 6. Direct sensor for microcontroller interface.

This ink is deposited onto the substrate, using screen printing by a screen printing machine AUREL, model 900 (Italy). It is printed through PET1000-230/48W (inches) mesh polyester screens with 30-µm photosensitive masking. Each layer is dried in an oven at 130 ◦C for 5 min. Fig. 4 shows the humidity sensor with two conductive layers and a dielectric layer between them.

C. Calibration

Humidity sensors have been calibrated by mixing solid salt and its saturated aqueous solution at a fixed temperature to ob- tain a specific RH value [6], [21], [22]. Each RH concentration required specific salt (see Table II). The saturated salt solution, which is a mixture of distilled water and chemically pure salt, is enclosed in a sealed metal or a glass chamber. A humidity- measuring instrument has been used to compare the results.

Electrical testing of the sensors was accomplished by con- necting wires to the electrode contacts using a silver loaded conductive adhesive.

First, capacitance was measured using a Promax MZ705 LCR meter [1 kHz/2 V/8–160 pF ± (3% + 2 pF)] and our own electronic system (see Fig. 5).

The capacitance response of the sensors was studied over the range of 10%–90% RH (experimentally 9.2%–98.5% RH). Humidity was measured using a RS-1366 humidity/temperature meter (range 1%–99% RH, resolution 0.1% RH, and accuracy

±3% RH).

D. Measurement Electronic System

Generally, microcontrollers include timers to make fine charge and discharge time measures in resistor–capacitor (RC) circuits. Charges and discharges give exponential signals with a RC time constant. This signal is timed until it reaches the pin threshold voltage of a microcontroller input port. The input port includes a Schmitt trigger (ST) buffer with a low threshold voltage (VTL) and a high one (VTH). The noise voltage super- imposed on VTL is less than the one superimposed on VTH; hence, discharge time measures should suffer fewer variations than the charge time ones [19]. Due to this, discharge-time- based circuits are studied in this paper [18].

1) Interface Circuits for Capacitive Sensors: Fig. 6 shows the basic microcontroller interface of a direct capacitive sensor. Cx is the capacitance sensor, Cc is the calibration capacitor, and R is the discharging resistor. The “three-signal method” [19] can be used in this design with no additional components. This design includes a two-point calibration. In order to do the calibration, the microcontroller has to do three measures: 1) a sensor measure (to obtain Nx); 2) a reference measure (to obtain Nc); and 3) a zero error measure (to obtain Noff ). Because of stray capacitances, Noff is different to zero. With this, an estimation of sensor capacitance value Cx can be done by means of the following expression [18]:

Cx =

Nx - Noff




Nc - Noff


The reference capacitance value must be well known to do the measure. This value is the average value of the sensor capacitance range. For example, if we have a sensor that has



Fig. 7. Flowchart for capacitance measurement.

to variate from 60 to 100 pF, the reference capacitance value should be approximately 80 pF. In our case, the closer standard value (82 pF) was chosen. A 20-MΩ resistor was used to measure values in the order of picofarad.

In order to do the necessary measurements, specific soft- ware has been developed for the applied microcontroller (PIC16F877). This software has been done attending to the flowchart shown in Fig. 7.



Fig. 8. Multilayer feedforward ANN.

This process is run three times to get arithmetic average of the three obtained values and to avoid possible fluctuations. A computer receives the results by a RS-232 serial port to capture them and save them in a file. Later, a specific application developed with LabVIEW is used.


As subsequently explained, the two sensors obtained do not present a lineal response. A nonlineal response complicates the measuring process; therefore, a linearization procedure is needed. In order to satisfy this need, a sensor linearizing method based on feedforward neural networks multilayer perceptron (MLP) is shown. Our low-cost microcontroller allows us to program a linearization procedure. With this, a mathematical model can be applied to design software routines that help us improve the sensor characteristic. In fact, this is quite usual when working with microprocessor-based data acquisition sys- tems [29]. Depending on the final realization, the mathematical model can be programmed or numerically implemented as a simple lookup table [19], [23]. However, more sophisticated techniques have been recently proposed for sensor signal con- ditioning, such as artificial neural networks (ANNs) [24]–[26]. Due to its self-learning ability based on examples [27], ANNs are very useful tools to model sensors in case of absence of appropriate (or accurate enough) mathematical models [26] for sensor calibration [25] or temperature compensation [27]. The applied neural model is very simple that it can be pro- grammed into any low-cost microcontroller, such as those used in embedded applications for data loggers or industrial controls, including the one used in this paper.

In this kind of networks, the example-based learning ability is very useful, particularly when using sensors without an exact mathematical representation of their input–output characteris- tic. As shown [29], MLP can linearize a nonlinear sensor from samples, considering samples as data pairs consisting of sensor

Fig. 9. Hyperbolic tangent sigmoid transfer function.

outputs and their corresponding desired linear behaviors (or corrections that linearize the outputs). Moreover, the flexibility of ANN allows its easy adaptation to changes in the sensor response due to different causes such as aging, variations in RH, drift, and sensor replacement. Retraining the system with new samples allows this adaptation [28].

1) Implementation of FF-MLP: Several tests have been done in order to obtain the best neural network architecture. These tests have been done by changing the number of layers, the number of nodes, and activation functions. The best result has been obtained using a feedforward MLP (FF-MLP) with three layers, one input node corresponding to capacitor value, five hidden nodes, and one output node (see Fig. 8).

Tansig (hyperbolic tangent sigmoid transfer function) activa- tion functions are used for neurons in the hidden nodes and the output nodes. This function is showed in Fig. 9 and

a =

en - e-n




en + e-n


Data are normalized in the range of [-1, 1] to enhance the neural network algorithm [28]. Weights and biases of the neural network are obtained from the personal computer, where the Levenberg–Marquardt backpropagation algorithm imple- mented in MATLAB is used to train this network.

- xmin)



Fig. 10. ANN MATLAB-based training flowchart.

a)Implementation in MATLAB: Development and train- ing of the neural network are explained in the flowchart showed in Fig. 10.

The data necessary for microcontroller programming are weights and biases, which can be obtained using NET commands.

b)Implementation in the microcontroller: The software was coded in C language and consists of two routines: 1) the overall system control in which capacitance calculation is ac- cording to point Section II-D1 and 2) implementation of the neural network. The capacitances data are normalized using the same MATLAB function [29] as

y =

(ymax - ymin) · (x - xmin)

+ ymin.



Fig. 11. Flowchart for the whole process.



Weights Wji and biases Bj of the trained ANN are obtained from the memory of the microcontroller. Using coefficient data Xi from registers and the weights and the biases, the microcontroller calculates the output for each of the five hidden nodes by using the following expression:


yj = F Xi · Wji + Bj (5)


where F is the tansig activation function, i is the input node (i: 1 to 5), and j is the hidden node (j: 1 to 5).

By using this yj data, weight Vkj , and biases Bk values, the values of the output node are obtained by using the following expression:


The output data are denormalized using the same MATLAB function. These data are displayed on the liquid crystal display panel.

This routine is coded in C language and is converted to HEX code using the cross compiler. The HEX file is downloaded into the microcontroller Flash memory. The ANN has been programmed in 5908 B of program memory and 221 B of data memory and is executed in less than 1 ms.

The whole process can be translated into a flowchart, as shown in Fig. 11.

yk = Ψ

yj · Vkj + Bk









where Ψ is the tansig function, k is the output node (k: 1), and j is the hidden node (j: 1 to 5).

With this, the tansig function can be defined by the following:

Tansig =


- 1.


1 + e 2a






Capacitance has been measured by an impedance meter (1 kHz) at 20 â—¦C and 50% RH. The nominal capacitance obtained in Melinex-based sensor is 57 pF and in the CG3460- based sensor is 93 pF (see Table III).

Differences between real and theoretical values may be due to a substrate thickness reduction when manufacturing them. In fact, significant thickness variations have been observed by



Fig. 12. Cross-sectional SEM photograph of the Melinex ST-726 film.

Fig. 13. Relation between capacitance and humidity (20 â—¦C) for Melinex ST726-based sensor.

using secondary electron microscopy (SEM) with a JEOL JSM- 5410 scanning microscope (see Fig. 12). This reduction may be influenced by some factors such as temperature and/or ink diffusion inside the substrate, generating effective dielectric thickness depletion.

Fig. 13 shows the capacitance response measures with the LCR meter and our own microcontroller system for MELINEX ST-726 (125 µm) based sensor.

Fig. 14 shows the capacitance response measures with the LCR meter and our own microcontroller system for CG3460 (100 µm) based sensor.

As shown in Figs. 13 and 14, the response of our own mea- surement system is quite similar to the LCR meter response. On the other hand, the observed hysteresis was nearly void.

Fig. 15 shows the capacitance change with time of CG3460- based sensor when subjected to a step change in humidity of 25%–60% RH (absorption) and 60%–25% RH (desorption).

Time response is high compared with the one from commer- cial sensors as Humirel HS1101 (5 s from 33% to 76%RH).

A. Linearity

The maximum nonlinearity error is equal to 12.2% full-scale span (FSS) for CG3460-based sensor (see Fig. 16). This error is considerable compared with the one from a commercial sensors such as HS 1101 (2% FSS). For the ST726-based sensor, the response is completely nonlinear. Due to these nonlinear re- sponses, we have decided to use a neural network, as described before.

Fig. 14. Relation between capacitance and humidity (20 â—¦C) for CG3460- based sensor.

Fig. 15. Response time for CG3460-based sensor.

Fig. 16. Linearity response of the interface circuit when measuring the CG3460-based sensor.

In the end, results of the neural network application to the sensors’ response are presented: The normalized relationship between the estimated humidity values and the measured ones are shown in Fig. 17 (Melinex ST726-based sensor). In this figure, we also present the line fit and the R2 value, i.e., 0.998. Fig. 18 shows the estimated (via ANN) and measured (via sensor) humidity values versus capacitances.


Low-cost capacitive relative-humidity sensors have been de- veloped. These sensors are formed by screen printing conduc- tive tracks on opposing sides of a moisture-sensitive polyester substrate. The sensors adhered well to the polyester, and they were stable in humid environments.



Fig. 17. Relationship between estimated and measured humidity values.

Fig. 18. Neural Network response for Melinex ST726-based sensor.

In order to measure relative humidity, a microcontroller has been used to develop and implement a capacitive impedance measuring system. The design of the system is based on a direct sensor–microcontroller interface. The “three-signal method,” which includes a self-calibration measuring system, has been used with no need of conditioners. With this, low-power and low-cost devices can be developed. Both low power and low cost are important requirements for RFID sensors and their applications such as traceability of food products.

In the end, due to the sensors’ nonlineal response, a neural network has been developed and applied. This ANN has been embedded into the microcontroller.


The authors would like to thank the Electronic Microscopy Service of the Polytechnic University of Valencia for sup- port and suggestions while working with secondary electron microscopy.


[1]J. Laundgren, Investigation of Paperbased RFID With Integrated Sensors. Demonstration and Analysis of Printed Sensors. Examensarbete.

[2]N. D. Sankir, Flexible Electronics: Materials and Device Fabrication. Blacksburg, VA: Virginia Polytech. Inst. State Univ., Dec. 2005.

[3]K. Finkenzeller, RFID Handbook, 2nd ed. Hoboken, NJ: Wiley, 2004.

[4]“RFID white paper: Technology, systems and applications,” German Assoc. Inf. Technol., Commun. New Media, Berlin, Germany, 2005.

[5]A. Ferrer-Vidal, A. Rida, S. Basat, L. Yang, and M. M. Tentzeris, “In- tegration of sensors and RFID’s on ultra-low-cost paper-based substrates for wireless sensor networks applications,” in Proc. IEEE WiMesh, 2006, pp. 126–128.

[6]A. Oprea, N. Barsan, U. Weimar, J. Courbat, D. Briand, and N. F. de Rooij,

“Integrated temperature, humidity and gas sensors on flexible substrates for low-power applications,” in Proc. IEEE Sens. Conf., 2007,

pp. 158–161.

[7]A. Rida, R. Vyas, S. Basat, A. Ferrer-Vidal, L. Yang, S. K. Bhattacharya, and M. M. Tentzeris, “Paper-based ultra-low-cost integrated RFID tags for sensing and tracking applications,” in Proc. IEEE-ECTC Symp., San Diego, CA, 2007, pp. 1977–1980.

[8]L. K. Baxter, Capacitive Sensors: Design and Applications. New York: IEEE Press, 1997.

[9] C. Roman, O. Bodea, N. Prodan, A. Levi, E. Cordos, and I. Manoviciu, “A capacitive-type humidity sensor using crosslinked poly(methyl methacrylate-co-(2 hydroxypropyl)-methacrylate),” Sens. Actuators B, Chem., vol. 25, no. 1–3, pp. 710–713, Apr. 1995.

[10]Y. Sakai, Y. Sadaoka, and M. Matsuguchi, “Humidity sensors based on polymer thin films,” Sens. Actuators B, Chem., vol. 35, no. 1–3, pp. 85– 90, Sep. 1996.

[11]A. Bearzotti, D. Caliendo, E. Verona, and A. D’Amico, “Integrated optic sensor for the detection of H2 concentrations,” Sens. Actuators B, Chem., vol. 7, no. 1/3, pp. 685–688, Mar. 1992.

[12]G. Gusmano, G. Montesperelli, E. Traversa, and A. Bearzotti, “Humidity-

sensitive electrical properties of MgAl2O4 thin films,” Sens. Actuators B, Chem., vol. 14, pp. 525–527, 1993.

[13]N.-B. Cho, T.-H. Lim, Y.-M. Jeon, and M.-S. Gong, “Humidity sensors fabricated with photo-curable electrolyte inks using an ink-jet printing technique and their properties,” Sens. Actuators B, Chem., vol. 130, no. 2, pp. 594–598, Mar. 2008.

[14]P. M. Harrey, B. J. Ramsey, P. S. A. Evans, and D. J. Harrison, “Capacitive-type humidity sensors fabricated using the offset lithographic printing process,” Sens. Actuators B, Chem., vol. 87, no. 2, pp. 226–232, Dec. 2002.

[15]F. Reverter and R. Pallás-Areny, Direct Sensor-to-Microcontroller Inter- face Circuits. Barcelona, Spain: Marcombo S.A., Sep. 2005.

[16]P. H. Dietz, D. Leigh, and W. S. Yerazumis, “Wireless liquid level sensing for restaurant applications,” in Proc. IEEE Sens., Orlando, FL, Jun. 12–14, 2002, pp. 715–720.

[17]J. E. Gaitán-Pire, M. Gasulla, and R. Pallàs-Areny, “Direct interface for capacitive sensors based on the charge transfer method,” in Proc. IEEE IMTC, Warsaw, Poland, May 1–3, 2007, pp. 1–5.

[18]F. Reverter and O. Casas, “Direct interface circuit for capacitive humid- ity sensors,” Sens. Actuators A, Phys., vol. 143, no. 2, pp. 315–322, May 2008.

[19]R. Pallás-Areny and J. G. Webster, Sensors and Signal Conditioning. New York: Wiley, 1991.

[20]J. Laconte, V. Wilmart, D. Flandre, and J. P. Raskin, “Capacitive humidity Sensor Using 3-Layer Patterned Polyimide Sensing Film,” in Proc. IEEE Sens. Conf., 2003, pp. 372–377.

[21]O. McGaughey, J. V. Ros-Lis, A. Guckian, A. K. McEvoy, C. McDonagh, and B. D. MacCraith, “Development of a fluorescence lifetime-based sol- gel humidity sensor,” Anal. Chimica Acta, vol. 570, no. 1, pp. 15–20, Jun. 2006.

[22]D. B. Papkovsky, G. V. Ponomarev, S. F. Chernov, A. N. Ovchinnikov, and I. N. Kurochkin, “Luminescence lifetime-based sensor for relative air humidity,” Sens. Actuators B, Chem., vol. 22, no. 1, pp. 57–61, Oct. 1994.

[23]B. Martín-del-Brío and N. Medrrano-Marqués, “Microcontroller makes linear temperatura measurement simple,” Electron. Design, vol. 48, no. 22, pp. 133–134, 2000.

[24]J. Dias Pereira, O. Poastolache, and P. Silva Girão, “A temperature- compensated system for magnetic field measurements based on artificial neural networks,” IEEE Trans. Instrum. Meas., vol. 47, no. 2, pp. 494– 498, Apr. 1998.

[25]D. Massicotte, S. Legendre, and A. Barwicz, “Neural network based method of calibration and measure reconstruction for a high-pressure



measuring system,” IEEE Trans. Instrum. Meas., vol. 47, no. 2, pp. 362– 370, Apr. 1998.

[26] N. J. Medrano Marqués, R. Del Hoyo, and B. Martín del Brío, “A ther- mocouple model based on neural networks,” in Proc. IWANN, vol. 2085,

Lecture Notes in Computer Science, Jun. 2001, pp. 531–538.

[27] S. Haykin, Neural Networks, A Comprenhensive Foundation. Upper Saddle River, NJ: Prentice Hall, 1999.

[28]N.J. Medrano-Marques and B. Martin-del-Brio, “Implementing neural networks onto standard low-cost microcontrollers for sensor signal processing,” in Proc. 10th IEEE Conf. ETFA, 2005, vol. 2, pp. 967–972.

[29]J. Ibáñez Civera, E. Garcia Breijo, N. Laguarda Miró, L. G. Sánchez, J. G. Baixauli, I R. Gil, R. Masot Peris, and M. A. Fillol, “Artificial neural network onto eight bit microcontroller for Secchi depth calculation,” Sens. Actuators B, Chem., vol. 156, no. 1, pp. 129–132, Aug. 2011.

José Pelegrí-Sebastiá received the M.Sc. degree in physics, the M.S. degree in electronic engineering, and the Ph.D. degree from the University of Valencia, Valencia, Spain, 1994, 1998, and 2002, respectively. His thesis topic involved the study of giant magne- torresistance (GMR) sensors and its applications.

He is currently an Associated Professor with the Instituto de Investigación para la Gestión Integrada de Zonas Costeras, Universidad Politécnica de Valencia, Grao de Gandia, Spain. His research in- terests include instrumentation systems, characteri-

zation sensors (particularly magnetic sensors and its industrial applications), and network sensors.

Eduardo García-Breijo received the M.S. degree in electronic engineering from the Universitat de València, València, Spain, in 1997 and the Ph.D. degree from the Universidad Politécnica de Valencia (UPV), Valencia, in 2004.

He is an Assistant Professor of electronic tech- nologic with UPV. He is a member of the Instituto de Reconocimiento Molecular y Desarrollo Tecnol- ógico, UPV. His main areas of interest are the devel- opment of multisensors in thick-film technology.

Tomas Sogorb was born in Alicante, Spain, in November 1967. He received the M.S. degree in telecommunication engineering and the Ph.D. degree from Universidad Politécnica de Valencia (UPV), Valencia, Spain, in 1997 and 2003, respectively.

Since 1999, he has been a member of the research and teaching staff at the Department of Electronic Engineering, UPV. He is also with Instituto de Inves- tigación para la Gestión Integrada de Zonas Costeras, UPV. His fields of interest are related to characteri- zation sensors and network sensors with harvesting

energy and their applications.

Nicolás Laguarda-Miro received the M.S. degree in environmental sciences and the Ph.D. degree from the Universidad Politécnica de Valencia (UPV), Valencia, Spain, in 2000 and 2005, respectively.

He is currently a member of the Instituto del Agua y del Medio Ambiente and a Teacher of environmen- tal subjects in the Escuela Técnica Superior de Inge- niería del Diseño, UPV. He is also currently with the Instituto de Reconocimiento Molecular y Desarrollo Tecnológico, UPV. His main areas of interest are the environmental quality control, environmental impact

assessment, and technology applications to these areas.

Javier Ibáñez received the Maitrise degree in power electronic and control from the Universite Pierre et Marie Curie (Paris VI), Paris, France, in 1994 and the Ph.D. degree from the Universidad Politécnica de Valencia (UPV), Valencia, Spain, in 2009.

He is an Assistant Professor of electronic tech- nologic with the Department of Electronic Engi- neering, UPV. He is a member of the Instituto de Reconocimiento Molecular y Desarrollo Tecno- lógico, UPV. His main areas of interest are water organic contamination devices.

José Garrigues received the M.S. degree in indus- trial engineering from the Universidad Politécnica de Valencia (UPV), Valencia, Spain, in 1983.

Since 1990, he has been a member of the staff at the Department of Electronic Engineering. He is also a member of the Instituto de Reconocimiento Molec- ular y Desarrollo Tecnológico, UPV. His main areas of interest are artificial neural network applications by microcontrollers. He is currently developing his thesis in these applications.