- Email: [email protected]
Environmental monitoring with a multisensor platform on polyimide foil
Sensors and Actuators B 171–172 (2012) 190–197
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
Environmental monitoring with a multisensor platform on polyimide foil A. Oprea a,∗ , J. Courbat b , D. Briand b , N. Bârsan a , U. Weimar a , N.F. de Rooij b a
Institute of Physical and Theoretical Chemistry (IPC), University of Tübingen, Auf der Morgenstelle 15, D-72076 Tübingen, Germany Ecole Polytechnique Fédérale de Lausanne (EPFL), Institute of Microengineering (IMT), Sensors, Actuators and Microsystems Laboratory (SAMLAB), Rue Jaquet-Droz 1, P.O. Box 526, CH-2002 Neuchâtel, Switzerland b
a r t i c l e
i n f o
Article history: Received 16 August 2011 Received in revised form 16 February 2012 Accepted 29 February 2012 Available online 17 May 2012 Keywords: Multisensor platform Environmental monitoring Polyimide foil Metal-oxide sensor Capacitive sensor Calibration curve
a b s t r a c t A multisensor platform on plastic foil for environmental monitoring has been produced and its gas sensing performance, investigated. It is an array of conductometric metal-oxide (MOX) and capacitive polymer gas sensors integrated with a resistive platinum thermometer on a polyimide sheet substrate. The feasibility of simultaneous measurement of oxidizing and reducing gases, volatile organic compounds (VOCs), humidity and temperature has been demonstrated. MOX signals comparable with those of the devices realized on ceramic substrates have been obtained. Due to its structure, the platform is very versatile and, by using different sensor configurations and sensing materials, it allows the detection of a broad spectrum of gaseous analytes over wide concentration ranges. From the raw signals, temperature and humidity-corrected gas responses have been inferred which have been used for the calibration of the platform sensors. All the integrated devices were stable and gave reproducible signals for more than two months of operation, even when the MOXs ran continuously at 300 ◦ C. The performed investigation proved the device concept viability and the reliability of its practical implementation. © 2012 Elsevier B.V. All rights reserved.
1. Introduction The development of autonomous sensing systems gained significant interest within the last years due to the decreasing power consumption of the electronic components and the spread of wireless communication. At present it is a clear trend to integrate new components and expand the features of the older ones in order to make the whole assembly smarter. Monitoring environmental parameters may find relevance in several domains, such as human comfort and health, ambient pollution reduction or perishable goods preservation. The measuring of various parameters – temperature, relative humidity, gas concentration or pressure – with a single chip has already been proven on silicon substrates [1–4]. For example Li et al. [5] presented several types of transducers – calorimetric, capacitive, gravimetric, resistive – for gas sensing. Also, the use of plastic foil for sensors and actuators has been reported in the literature of the last years. Among other, anemometer [6], bolometer [7], thermometer [8–10], and metal-oxide [11,12] or capacitive gas sensors [13,14] have been produced on polyimide (PI) foils. Here, a multisensor platform for environmental monitoring on one polyimide substrate that integrates conductometric
∗ Corresponding author. E-mail address: [email protected] (A. Oprea). 0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2012.02.095
metal-oxide (MOX) and polymeric capacitive gas sensors as well as a Pt thermometer is presented. In comparison with the investigations/implementations of different types of sensors on flexible substrates reported earlier by our group of authors, that is, the design, fabrication, and characterization of metal-oxide gas sensors [12,15], and capacitive sensor arrays [14], the multisensor platform addressed by this contribution represents a significant conceptual and technological advance. That is due to a higher degree of integration and, mainly, to the successful combination of different transducers, different types of sensing materials and different technologies. Moreover, the possibility to correct the temperature and humidity parasitic substrate contributions to the sensor responses allows achieving the same sensing potential as in the case of the Si-based platforms [3].
2. Experimental 2.1. Platform concept, structure and design The demonstrator configuration of the hybrid multisensor platform (6 mm × 6 mm) is presented in Fig. 1. It combines three different sensor types on the same polymeric substrate: two metaloxide and two capacitive gas sensing structures as well as a resistive thermometer. Depending on the concrete application needs, other configurations, with higher complexity, are feasible. The main
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 1. Optical micrograph of the multisensor platform fabricated on polyimide foil (up) and the device packaged on a PCB for its characterization (down).
advantages of combining sensors with different transducing principles are: - the possibility to sense gas mixtures containing different components spread over different ranges of concentrations - the possibility to correct the temperature influence on the sensor responses and calibration curves - the possibility to remove, in principle, the cross sensitivities by using extended arrays - the possibility to make predictions concerning the gas composition through multivariate data analysis The cross section of the demonstrator platform is displayed in Fig. 2. It enables the understanding of different sensor structures and their geometric parameters: • The MOX conductometric sensors are based on Pt interdigitated transducers with an active area of 100 m × 100 m, and an aspect ratio of 164. The aspect ratio indicates how many times the measured resistance of the sensing layer is reduced by the electrode configuration in respect with the standard sheet resistance of the same layer. The electrodes of the transducer are deposited
Fig. 2. Cross-sectional schematic of the multisensor platform on polyimide foil. The power consumption of the MOX sensors can be reduced by using a back-etched membrane.
191
over the heating Pt meander, from which they are electrically insulated by a polyimide layer (see Section 2.2). • Each capacitive transducer covers an area of 1.4 mm × 1.4 mm and has equal electrode width and spacing (10 m). A nominal capacitance lower than 10 pF has been targeted, which allows the direct readout of the capacitance with a commercial, high resolution, capacitance-to-digital converter. One capacitor – addressed from now on as “reference” capacitor – has been left uncoated, as a witness for the substrate residual response to gaseous analytes. The other one – to be addressed as “sensing” capacitor – has been covered with an appropriate polymer, providing the capacitive sensing function toward either a certain VOC or to humidity. Because of the additional polymer layer the capacitance of the sensing capacitor is higher than the capacitance of the reference capacitor with an amount (∼10–20%) that will be referred to as “sensing layer capacitance” (or “nominal sensing layer capacitance” if measured in dry synthetic air). The structure of the capacitive section of the multisensor platform allows differential operation [14], so that the reference capacitor signal and the nominal sensing layer capacitance can be subtracted from the sensing capacitor signal, in order to obtain the net gas response of the sensing layer, eliminating in this way the parasitic contributions of the substrate to the total capacitive response of the platform. The two, reference and sensing, capacitors differentially operated will be referred to as “differential sensor”. More information related to sensor evaluation in the differential operation mode is included in Section 3. • The resistive temperature transducer is a Pt meander realized on an area of 1.4 mm × 1.4 mm and having a targeted nominal resistance of 1 k at 0 ◦ C. At present the platform manufacture implies a certain technological complexity due to complementary technological steps being used. The foreseen “full plastic implementations” exclusively based on printed devices will considerably simplify the production chain premising the next generation of low cost and low power gas sensing platforms. 2.2. Fabrication The platform has been realized on a 50 m thick polyimide (PI) wafer (Upilex-50S from Ube Industries, Ltd), making use of a minimal number of conventional technological steps (see Fig. 2). The conducting elements of the sensors (thermoresistor, heaters, electrodes and pads) required only two levels of platinum (130 nm) deposited by DC sputtering over an adhesion layer of titanium (20 nm) and patterned by lift-off technique. The electrical insulation between the two layers of platinum was made by a 700 nm thick polyimide film (PI-2737 from HD Microsystems). To reduce the power consumption of the MOX sensors, optional 3 m thin membranes could be formed through dry etching from the backside of the polyimide foil. The fabrication process used for the multisensor platform was the one previously developed for single metal oxide gas sensors on PI, and is fully described in reference [15]; no extra step to integrate the capacitive transducers and temperature sensor has been required. Onto the transducers, all functional materials for gas sensing have been drop-coated. In the first step, the metal-oxide layers have been deposited: SnO2 loaded with 0.2 wt% Pd, which will be referred to as SnO2 :0.2%Pd [16] and WO3. They required a subsequent annealing at 400 ◦ C to enable the stabilization of the film properties. Afterwards, the sensing capacitor has been covered with either cellulose acetate butyrate (CAB) for humidity measurement or polyetherurethane (PEUT) for VOCs detection. The polymers were allowed to naturally dry at room temperature. For laboratory
192
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
characterization the multisensor platforms have been packaged (Fig. 1b) on PCBs (printed circuit boards) using conductive epoxy glue (Epotek H20E). Two rows of 8 standard gold-plated pins (with 2.54 mm spacing) soldered to the PCB made the platforms compatible with wide (15.24 mm) DIL sockets. 2.3. Electrical and gas measurement protocols The platforms, packaged on the DIL PCBs and placed in a measuring chamber made of Teflon® , were characterized with a computer controlled gas-mixing and measuring system at different relative humidity (r.h.) levels and gas concentrations with values in the TWA (time weighted average) range or below. CO (20–80 ppm), NO2 (0.1–1 ppm), EtOH (2–20, 20–600, or 300–3000 ppm) and relative humidity levels (0–70%) have been dosed as successive concentration pulses in the carrier gas – synthetic air or synthetic air with 50% r.h. background – at a flow rate of 200 sccm. Between each individual exposure pulse the samples were allowed to recover. The exposure protocols are always depicted in the upper side of the graphics containing raw data in order to facilitate the understanding of the platform response. The voltage applied to the heater of the MOX sensors, the resulting currents and the resistances of their gas sensing layers were monitored with a standard laboratory multichannel multimeter (Agilent 34970A). The readout of the capacitive transducers (used as reference or for the humidity or VOCs sensing) was ensured with a 24 bit capacitance-to-digital converter AD7746 from Analog Devices, Inc. This IC has a very high resolution, theoretically of 4aF, and a fairly low power consumption at low acquisition rates. An additional voltage/resistance input of the converter has been used to monitor the resistance of the temperature sensor. All digital data were transmitted to a PC through an I2 C-to-USB converter using the dedicated software developed in cooperation by IPC Tübingen and JLM-Innovation GmbH, Germany. 3. Results and discussion 3.1. Electrical parameters/characteristics The heaters of the metal oxide gas sensors have a nominal resistance of 150 at 0 ◦ C. To reach an operating temperature of 300 ◦ C in continuous heating mode, they need an individual power consumption of 15.5 mW in the micro-machined variant or 30.1 mW in the standard one. The MOX layer resistance, under operation conditions, ranges from several k to few M and depends on the ambient composition (see Section 3.2.3). The resistive thermometer has a resistance of ∼1.16 k at room temperature (1 k targeted nominal value at 0 ◦ C). After a stabilization period (few days at a temperature higher than 60 ◦ C), its response shows a good linearity (nonlinearity between −20 ◦ C and 150 ◦ C smaller than 0.3%). A 0.002 ± 0.0001 ◦ C−1 thermal coefficient of the resistance (TCR) has been experimentally determined in a calibration oven. This value is smaller than the one of the massive pure platinum (0.00391 ◦ C−1 ) because of the layer dimensions and the limited annealing range allowed by the plastic substrate. The capacitive transducers have a nominal capacitance of ∼2 pF. 3.2. Gas sensing performances The evaluation of the multisensor platforms has been performed with a test mixture containing, synthetic air with different levels of background humidity, as carrier gas, and VOCs vapors or inorganic gases as analytes. The capacitive polymer sensors of the platforms are sensitive only to VOCs and water vapors. Indeed, the sorption of the vapors in the amorphous polymer is a solvation process leading to a significant increasing of the analyte concentration in polymer in respect with its concentration in the ambient [17]. This is not
the case for the gases; at room temperature they are well above the critical isotherm and cannot condense in the polymer. In their turn the MOX chemoresistors are mainly sensitive to gaseous analytes that can undergo surface reaction involving a charge exchange with conduction or valence bands (depending on material conduction type – n or p) of the material. Therefore it is not important if the analyte is in vapour or gas phase. The analytes able to promote an increase in the concentration of the free electrons in the conduction band of an n-type MOX semiconductor or to reduce the number of free holes in the valence band of a p-type MOX semiconductor are usually referred to as “reducing gases” (for a detailed explanation see [18]). If the reverse trend occurs the analytes are referred to as oxidizing. Because the demonstrator platform contains only one differential capacitive sensor (recall Section 2.1) it could be dedicated to either humidity or a certain VOC. The choice, besides humidity, has been ethanol, because EtOH vapors are easily detected by the PETU polymer layer but, in the same time, have a strong reducing character in the interaction with the heated MOX being used (WO3 and SnO2 :0.2%Pd). This choice made it possible to check out how the MOX, very sensitive to ethanol at low concentrations, and the polymer (PEUT) sensors complement each other over a wide range of concentrations (1–3000 ppm). Several multisensor platforms have been systematically tested and evaluated during many weeks of experiments. They were all showing good functionality, comparable sensing parameters and unexpected reliability over long operation periods. 3.2.1. Parameters used to quantify the gas sensor performance In the literature, there are different ways to evaluate the gas sensor performance. To avoid confusion, a short overview of the parameters used in this section is given here. The main experimental data is the sensor signal. In the case of conductometric gas sensors the signal is the sample resistance (R). The baseline of the MOX sensors stands for the sensor signal (sensor resistance R0 ) in the absence of the target gas, that is, when the sensor faces the reference ambient only. The sensor response (SR) is the ratio between the sensor signals in the absence (R0 ) and presence (R) of the analyte: SR = R0 /R for reducing gases but SR = R/R0 for oxidizing gases; that is because, by convention, SR is chosen to be larger than 1 in the presence of the target gas. SR has no unit. By displaying the sensor responses against the gas concentration one obtains the sensor calibration curve and the sensor sensitivity, as the slope of the calibration curve. For non-linear calibration curves the sensitivity is concentration dependent; in such cases double logarithmic scales are frequently used, which facilitate the comparisons of the sensor responses over wide ranges of concentrations and naturally guide to the definition of the logarithmic sensitivity: the sensor response increment over ten times (decade) concentration increment. If the sensor response on the target gas concentration follows a power low (as expected for gas sensing mechanisms involving oxygen ionosorption [18]) the calibration curve will graphically appear as linear and the logarithmic sensitivity will be independent of concentration. The sensitivity unit is the reciprocal of the concentration unit, usually ppm−1 while the unit of the logarithmic concentration is (decade)−1 . The sensor stability is expressed through the stability of the signals, as long as the responses and the sensitivity are related to the baseline, which depends on the composition of the reference ambient. 3.2.2. Differential capacitive measurements of humidity Fig. 3 shows the raw and differential signals obtained with the reference and sensing (CAB-coated) capacitors of a platform (MSP2) dedicated to humidity measurements, which has been exposed to humidity (0–70%) and to the potential interfering gases such as
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 3. The response of the CAB differential capacitive sensor from the MSP-2 platform to carbon monoxide, ethanol, and relative humidity. The upper panel displays the exposure protocol. The raw signals of the sensing and reference capacitors are as well shown, but arbitrary shifted from the nominal capacitance (∼2pF) in order to appear with the same baseline and on the same panel with the CAB differential sensor response. The CAB differential capacitive sensor responds to humidity only.
CO (20–80 ppm) and ethanol (2–20 ppm). The differential response is obtained by subtracting the reference capacitor signal and the nominal capacitance of the CAB layer from the sensing capacitor signal as already explained in Section 2.1. The CAB film response toward humidity was selective. No visible response was obtained during the exposure to CO and low concentrations of ethanol. Table 1 presents the response and recovery times of the reference and sensing capacitors. The use of the differential measurement decreased the response time by a factor of six between 0 and 50% r.h., while the recovery time was reduced twenty-four times in respect to the sensing capacitor alone. Similar values are found in the literature [19]. These significant improvements resulted from the elimination, by subtraction, of the slow humidity response of the compact PI substrate (due to long diffusion times), otherwise included in the overall (sensing layer and substrate) sensor response. 3.2.3. Responses of the whole multisensor platforms to gases and humidity A typical result of the multisensor platform is presented in Fig. 4. The capacitive sensor has been based on PEUT as sensing material because it is suitable for VOCs detection once its cross sensitivity to humidity can be compensated (for example, by using an additional humidity sensor on a more complex multisensor platform or another platform whose sensing capacitor is dedicated to humidity). A simpler variant, suggested in Ref. [14], would be to use the substrate itself as humidity sensor, if the response and recovery times are not critical issues. As expected, the WO3 sensor is more sensitive to NO2 [20] than to EtOH, as shown in Fig. 4c (the slopes of the calibration curves in double logarithmic scales represent the logarithmic sensitivities of the gas sensors); its response to NO2 was practically not affected by the background humidity, while that one to EtOH almost quenched at 50% r.h. The humidity thus improved the selectivity of WO3 toward NO2 . The opposite happened for the SnO2 :0.2%Pd sensor. In both dry and humid air, the sensor exhibited a higher Table 1 Response and recovery time (t90% ) of the capacitive sensors between 0 and 50% of relative humidity.
Uncoated capacitor CAB-coated capacitor Differential measurement
tres,90% (min)
trec,90% (min)
29 25 3.8
113 105 4.3
193
Fig. 4. Gas response of the capacitive and chemoresistive gas sensors to nitrogen dioxide (0.1–1 ppm), ethanol (2–20 ppm), and relative humidity (0–70%). (a) Gas protocol; (b) differential PEUT capacitive sensor response (10 times magnified), PEUT sensing capacitor signal and reference capacitor signal. The PEUT sensing capacitor and reference capacitor signals (in fact in the 2 pF range) have been arbitrary shifted in order to be displayed on the same panel with the differential capacitive sensor response; (c) gas response of the SnO2 :0.2%Pd and respectively WO3 MOX sensors. Test performed on MSP-6 platform.
sensitivity for EtOH than for NO2 . In the presence of the humidity, the response to NO2 was reduced, improving the selectivity toward EtOH. To explore the detection of EtOH with the PEUT-coated capacitive sensor, the same kind of gas exposure sequence has been used for a different EtOH range: 300–3000 ppm. The results show (Fig. 5) a clear response to alcohol, almost independent of the humidity background. On the plot the response to humidity is larger than that to ethanol because of the high humidity levels being used in the experiment (50% r.h. at 20 ◦ C correspond to an absolute humidity value of roughly 11,500 ppm). However, the polymer is, in absolute terms, more sensitive to EtOH than to humidity. As all highly sensitive MOX sensors, the platform MOX channels will saturate with the increase of the target gas concentration, even if still in the TWA range (e.g. ∼500 ppm for EtOH). Fortunately, in the alcohol case, one can still measure by using the polymer (PEUT)
Fig. 5. Capacitive and chemoresistive responses of the multisensor platform to nitrogen dioxide (0.1–1 ppm), ethanol (300–3000 ppm), and relative humidity (0–70%). (a) Gas protocol; (b) the response of the PEUT differential capacitive sensor (10 times magnified) and the signals from the sensing and reference capacitors (see the explanations given in the caption of the Fig. 4b; the background humidity level is designated as B.H.L. on the response curve of the PEUT differential capacitive sensor; (c) Gas response of the MOX sensors coated with SnO2 :0.2%Pd and WO3 . Test performed on MSP-6 platform.
194
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 6. The response (R/R0 ) of the SnO2 :0.2%Pd and WO3 MOX gas sensors to humidity (baseline dependency on the background humidity) at 300 ◦ C operating temperature. Eight measurements have been performed with the same device resulting in a precision better than 95%. Test performed on MSP-2 platform.
differential capacitive sensor. Moreover, the multisensor platform has an important advantage here: a smart driving electronic can switch off the MOX heaters and validate only the capacitive outputs, preventing inaccurate records and saving power. 3.2.4. Dependency of the MOX sensors baseline on the humidity The background humidity influence on the baseline of the SnO2 :0.2%Pd and WO3 gas sensors is presented in Fig. 6, as sensor response referred to dry synthetic air. For the SnO2 :0.2%Pd based sensor the largest effects occur for the 0–10% r.h. interval, namely a factor of ∼2.5. In the much wider range from 10 to 70% r.h. only ∼20% changes of the baseline are recorded. This represents a factor of ∼1.2 in response, small in respect with the usual responses reported later on for the TWA concentrations of ethanol and NO2 . The situation is even better in the case of WO3 sensors, with a factor of 1.1 between 10 and 70% r.h. These values are slightly varying from platform to platform, but do not exceed a factor of 1.6 for humidity range between 20% and 70% r.h., which is the interesting one for, e.g., indoor applications. 3.3. Reproducibility and stability of the sensors The reproducibility of the CAB differential capacitive sensor response has been tested together with the sensor reliability during several weeks of uninterrupted operation. No major degradation is visible in Fig. 7, which presents the results. The highest decrease in the response (and, consequently, in sensitivity) over a twenty-day period, of 0.2%/day, has been observed for 70% r.h. That might be due to the repeated NO2 exposures the sensing capacitors of the platform undergone, as it has not been encountered for the individual capacitive humidity sensors exposed to humidity only.
Fig. 7. Stability of the CAB differential capacitive humidity sensor over a period of 20 days. The periodic instabilities are due to the deactivation of the air conditioning system in the laboratory during the weekends. Test performed on MSP-2 platform.
Fig. 8. The temperature of the MOX sensor heaters and the Pt thermometer readout on the multisensor platform MSP-6. The supplied power was ∼17 mW and the ambient temperature 25 ◦ C. The heaters of both WO3 and SnO2 :0.2%Pd sensors showed a high stability with a temperature variation of less than 1 ◦ C. During the initial operation period (missing from data), the sensors have been heated at other temperatures.
The temperature variation of the MOX sensors when operating at constant heating voltage over 48 days is presented in Fig. 8. The sensors exhibited a very constant behavior after a formation/stabilization period. The temperature variations of the WO3 and SnO2 :0.2%Pd coated sensors were respectively of 0.6 ◦ C and 0.5 ◦ C in the last 44 days. As well as stable was the Pt thermometer; this is, however, not surprising since the operation conditions are much friendly for this device (see Fig. 8). As a representative example for the stability of the MOX devices, the resistance variation of a WO3 sensor over a period of 21 days is shown in Fig. 9 – the heater has been kept at ∼300 ◦ C. The sensors showed a good reproducibility. After a stabilization time of two days, the baseline, the response to 400 ppb NO2 and to 1000 ppm EtOH drifted by 2.2%, 3.4% and 6.5%, respectively (50% r.h. background). Translated into response values, the changes in the sensor resistance accounted for in Fig. 9 represent a factor less than 1.008 for ethanol and less than 1.004 for NO2 . 3.4. Calibration curves From the above-discussed responses of the CAB differential capacitive sensor toward humidity (see Figs. 4 and 5), a calibration curve could be derived, which is displayed in Fig. 10a. It is fairly linear, with a slope (sensitivity) of 2.45 fF/%r.h. The calibration curve of the capacitive humidity sensor can be used to predict the actual humidity but also to compensate the cross sensitivity of the polymer and MOX sensors to humidity, resulting in a more accurate
Fig. 9. Stability of the WO3 sensor over a period of 20 days; the sensor has been operated at 300 ◦ C. Device stabilization occurred after two days of exploitation. The time variations of the sensing layer resistance under constant exposure is less than 7%, independent of the target gas being supplied (in wet environment). Test performed on MSP-7 platform.
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 10. (a) The response of the CAB differential capacitive sensor and that of the reference capacitor (substrate) from MSP-2 platform to humidity. Eight measurements performed with the same device; mean statistical error <4% for CAB and <6% for the substrate. (b) The response of the PEUT differential capacitive sensor from the MSP-6 platform to ethanol and its cross sensitivity to humidity. Five consecutive measurements; mean statistical error <12% for ethanol and 5% for H2 O. Almost no reference capacitor response to ethanol has been observed.
determination of the gas and VOCs concentrations. For the same purpose it is possible to exploit the parasitic substrate response to humidity, as addressed at the beginning of the paragraph in Section 3.2.3 (see Fig. 10a). The time constants of this readout are, however, significantly higher than those of the differential capacitive sensors (see Table 1.). The plots in Fig. 10b account for the linearity and reproducibility of the responses delivered by the PEUT differential capacitive sensor of the platform against EtOH. In a first step the calibration curve of PUET against ethanol in dry synthetic air is extracted from the differential capacitive responses to the five incremental ethanol concentration pulses (300, 600, 1000, 2000, 3000 ppm). By a similar procedure one obtains the calibration curve toward humidity, considered in its turn as analyte. The 10–70% r.h. pulses in the middle of the exposure are used for this purpose. In the presence of background humidity the baseline of the differential response shifts upwards to the BHL level in Fig. 5b. It has to be subtracted in order to obtain the contribution of the ethanol only (curve “humidity subtracted” in the same figure) from which the ethanol calibration is derived (Fig. 10b, “Ethanol; 50% r.h.”). More detailed description of the approach can be found in [14]. Due to the additional interactions with the polymer and target analyte, the humidity effect is not exclusively additive to the EtOH one, resulting in a small the separation of the calibration lines “Ethanol; 0% r.h.” and “Ethanol; 50% r.h.” in Fig. 10b. Fig. 11 shows the calibration curves obtained for the metal-oxide gas sensors based on WO3 and SnO2 :0.2%Pd sensing materials. The calibration curves of both MOX sensors are little influenced by a background humidity in the range of 20% r.h. −70% r.h. Actually, the WO3 response to NO2 was almost independent of the
195
Fig. 11. Calibration curves of the MOX gas sensors based on (a) SnO2 :0.2%Pd and (b) WO3 sensing layers. Test performed on MSP-6 platform.
background humidity level while SnO2 :0.2%Pd response to EtOH displayed only a reduced dependency on the background humidity. As observed above, on the response plots, the presence of the background humidity increases the selectivity of WO3 layer toward NO2 and that of the SnO2 :0.2%Pd layer to ethanol. Numerical evaluations of the cross sensitivities presented by the platform sensors toward the tested gases/vapors, other than the target ones (NO2 for WO3 , ethanol for SnO2 :0.2%Pd and PEUT and humidity for CAB) are included in Table 2. It demonstrates that for environment monitoring purposes, where precisions of 10% from TWA are more than satisfactory, the MOX sensors do not need any correction as well as the CAB capacitive one. The PEUT capacitive sensor only has a high cross sensitivity to humidity. Fortunately this cross sensitivity is almost additive: the responses to ethanol and background humidity are almost independent and therefore they practically superpose to yield the total sensor response. In consequence the background humidity contribution can be removed by using the humidity sensor (simple subtraction). The calibration curves in Fig. 11 also contain subsidiary information about the gas sensing mechanism of the platform MOX devices. As explained in Section 3.2.1 the linear shapes of the calibration curves depicted in the figure are due to the double logarithmic scales and represent power law dependencies of sensor responses on the EtOH and respectively NO2 concentrations. This means that the chemoreceptive and transducing processes [18] in the MOX films coating plastic foils are the same as in the case of MOX films on ceramic substrates; they are not essentially affected by the underlying organic material. 3.5. Operation algorithms and measurement scenarios On the basis of the above reported results one can detail the functional algorithm of the demonstrator-platform when monitoring the ambient atmosphere in locations where, nearby background humidity, NO2 and ethanol are also present (actually the aim of the application). For instance a slowly changing humidity is assumed,
196
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Table 2 Cross sensitivity of the platform sensors. The cross sensitivity is evaluated as the ratio (in percents) of the sensor responses to the test and target gases respectively, both having the TWA concentrations. The humidity has been also included with the formal TWA value of 50% r. h., because it is not harmful (despite the human discomfort above 25 ◦ C at saturation). Below 0 ◦ C the ambient humidity is very reduced (∼9000 ppm at 0 ◦ C and only ∼200 ppm at −20 ◦ C). Analyte
TWA [ppm]
MOX sensor responses
Cross sensitivity [%]
SnO2 :0.2%Pd NO2 CO EtOH H2 O Analyte
NO2 CO EtOH H2 O
5 30 400 11,500 TWA [ppm]
5 30 400 11,500
4 1.3 150.0 1.5
WO3 200.0 1.6 5 2.7
Polymer sensor responses [fF] PEUT
<1 <1 2 102
<0.1 <0.1 4 97
1. All sensors are powered on and allowed to warm-up. 2. All raw signals are recorded and used to obtain: a. the temperature b. the gas responses of the MOXs resistors c. the single capacitance values for reference (uncoated) and Ethanol-sensing (PEUT coated) capacitors 3. In parallel: a. the relative humidity at the given temperature is evaluated from the reference capacitance by using its calibration curve (like the one in Fig. 10a) already stored in the system database or by interpolating the two nearest ones. b. the response of the PEUT differential capacitive sensor at the ambient temperature is estimated from the signals of the sensing (PEUT) and reference capacitors. c. the concentrations of NO2 and EtOH are evaluated with good precision from the individual calibration curves (temperature independent) of the WO3 and respectively SnO2 :0.2%Pd sensors because: i. the influence of the humidity (in the range 30–70% r.h.) is very small (see Sections 3.2.4 and 3.4 and Table 2) ii. the cross sensitivity of WO3 to EtOH and that of SnO2 :0.2%Pd to NO2 with background humidity is reduced (see Section 3.4, Fig. 11 and Table 2). 4. From the response of the PEUT differential capacitive sensor the ethanol concentration is estimated in agreement with Section 3.4. Depending on the availability of the calibration curves at the measuring temperature interpolation might be required. 5. If the ethanol response of the capacitive platform channel overcomes a preset level the electronics switches off the SnO2 :0.2%Pd heating (see Section 3.2.3) reading only the capacitive response to ethanol. 6. All acquired values in one run are displayed or/and compared with threshold ones enabling/disabling an alarm procedure. At this stage the operation is resumed from the step 2 in continuous cycles until the multisensor platform service is no more required and it is let in standby modus or powered of.
WO3
2.7 0.9 – 1.0
– 0.8 2.5 1.4 Cross sensitivity [%]
CAB
which allows the use of the reference capacitor as humidity sensor (accounted for just above, in Section 3.4). Rapid variations of the background humidity have to be evaluated according to the Section 3.2.3. The intelligent system (microcontroller or PC) driving the platform, should already contain the temperature depending calibrations curves of the individual sensors in its own database (the ambient temperature has almost no relevance for the MOX sensors because they are already heated at a constant temperature).
SnO2 :0.2%Pd
CAB <1 <1 2 –
PEUT <3 <3 – 2200
4. Conclusion The integration of different sensing devices for environmental monitoring on the same polyimide substrate toward a multifunctional platform has been demonstrated for dissimilar detection principles, transducer types and sensing materials. During the performed investigations, all platform components exhibited good sensitivity, a fair selectivity and a satisfactory stability throughout the test period. It has been demonstrated that by joint operation of the Pt thermo-resistive, polymer-based capacitive and metaloxide conductometric sensors, several interfering gases can be detected and, to some extent, separated at hardware level. Thus, the capacitive structures monitored the humidity and ethanol (as representative VOC), while the metal-oxide ones monitored the oxidizing and reducing gases (NO2 , CO) as well as the reducing ethanol vapors. Beyond the achievements in terms of sensing parameters, the compensation of the substrate intrinsic (parasitic) response to the target gases, mainly to humidity, through the differential readout, can be regarded as a significant step forwards. Additionally, one has to remark the stable behavior of the MOX layers, even if deposited over double sandwiched plastic/Pt areas and continuously heated at 300 ◦ C. Due to the high versatility of the platform, the number of its elements can be further increased in order to extend the detection spectrum toward other classes of analytes and to expand the functional capabilities. Moreover, because it is possible to produce several dissimilar sensors on a single substrate, the assembly step toward hybrid arrays can be eliminated. A considerable improvement is foreseen at mini-system level where the humidity and temperature corrected responses delivered by the smart platform could be further deconvoluted through multivariate data analysis or other appropriate mathematical algorithms/methods. Also, the power consumption issues due to the MOX operation conditions, critical for autonomous or supply free (RFID) applications, can be circumvented if the substrate variant with back-etched hot plates is employed. Further fabrication simplicity and cost reduction are feasible by using only additive fabrication techniques, such as printing. The main challenge now is the platform improvement toward a smart sensing system with embedded electronics and signal processing facilities, based on nonconventional and cheap materials and technologies.
Acknowledgments We would like to thank the staff of the clean room facility from Center of Micronanotechnology (CMI), Ecole Polytechnique Fédéral de Lausanne (EPFL), for their help in the processing of the devices.
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
References [1] C. Hagleitner, D. Lange, A. Hierlemann, O. Brand, H. Baltes, Cmos single-chip gas detection system comprising capacitive, calorimetric and mass-sensitive microsensors, IEEE Journal of Solid-State Circuits 37 (2002) 1867–1878. [2] A.D. DeHennis, K.D. Wise, A wireless microsystem for the remote sensing of pressure, temperature, and relative humidity, Journal of Microelectromechanical Systems 14 (2005) 12–22. [3] M. Hautefeuille, C. O’Mahony, B. O’Flynn, K. Khalfi, F. Peters, A MEMS-based wireless multisensor module for environmental monitoring, Microelectronics Reliability 48 (2008) 906–910. [4] J. Hayes, S. Beirne, K.T. Lau, D. Diamond, Evaluation of a low cost wireless chemical sensor network for environmental monitoring, in: Proceedings of the IEEE Sensors 2008 Conference, 2008, pp. 530–533. [5] Y. Li, C. Vancura, D. Barrettino, M. Graf, C. Hagleitner, A. Kummer, M. Zimmermann, K.U. Kirstein, A. Hierlemann, Monolithic CMOS multi-transducer gas sensor microsystem for organic and inorganic analytes, Sensors and Actuators, B: Chemical 126 (2007) 431–440. [6] U. Buder, R. Petz, M. Kittel, W. Nitsche, E. Obermeier, Aeromems polyimide based wall double hot-wire sensors for flow separation detection, Sensors and Actuators, A: Physical 142 (2008) 130–137. [7] A. Mahmood, D.P. Butler, Z. Celik-Butler, Micromachined bolometers on polyimide, Sensors and Actuators, A: Physical 132 (2006) 452–459. [8] T. Kinkeldei, C. Zysset, K. Cherenack, G. Troester, Development and evaluation of temperature sensors for textile integration, in: Proceedings of IEEE Sensors 2009, Christchurch, New Zealand, 2009, pp. 1580–1583. [9] D.J. Lichtenwalner, A.E. Hydrick, A.I. Kingon, Flexible thin film temperature and strain sensor array utilizing a novel sensing concept, Sensors and Actuators; A: Physical 135 (2007) 593–597. [10] C.Y. Lee, S.J. Lee, G.W. Wu, Fabrication of micro temperature sensor on the flexible substrate, nanotechnology, in: 7th IEEE Conference on Nanotechnology, Hong Kong, China, 2007, pp. 1050–1053. [11] D. Briand, S. Colin, J. Courbat, S. Raible, J. Kappler, N.F. de Rooij, Integration of MOX gas sensors on polyimide hotplates, Sensors and Actuators, B: Chemical 130 (2008) 430–435. [12] J. Courbat, D. Briand, L. Yue, S. Raible, N.F. de Rooij, Ultra-low power metal-oxide gas sensor on plastic foil, in: Proc. of the Transducers’09 Conference, Denver, USA, 2009, pp. 584–587. [13] A. Oprea, N. Barsan, U. Weimar, M.L. Bauersfeld, D. Ebling, J. Wollenstein, Capacitive humidity sensors on flexible RFID labels, Sensors and Actuators, B: Chemical 132 (2008) 404–410. [14] 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, Sensors and Actuators, B: Chemical 140 (2009) 227–232. [15] J. Courbat, M. Canonica, D. Teyssieux, D. Briand, N.F. de Rooij, Design and fabrication of micro-hotplates made on polyimide foil: electrothermal simulation and characterization to achieve power consumptions in the low mW range, Journal of Micromechanics and Microengineering 21 (2011) 015014. [16] J. Kappler, N. Barsan, U. Weimar, A. Dieguez, J.L. Alay, A. Romano-Rodriguez, J.R. Morante, W. Gopel, Correlation between XPS, Raman and TEM measurements and the gas sensitivity of Pt and Pd doped SnO2 :0.2%Pd based gas sensors, Fresenius Journal of Analytical Chemistry 361 (1998) 110–114. [17] J.W. Grate, M.H. Abraham, Solubility interactions and the design of chemically selective sorbent coatings for chemical sensors and arrays, Sensors and Actuators B 3 (1991) 85–111. [18] N. Barsan, U. Weimer, Conduction model of metal oxide gas sensors, Journal of Electroceramics 7 (2001) 143–167. [19] E. Zampetti, S. Pantalei, A. Pecora, A. Valletta, L. Maiolo, A. Minotti, A. Macagnano, G. Fortunato, A. Bearzotti, Design and optimization of an ultra thin flexible capacitive humidity sensor, Sensors and Actuators, B: Chemical 143 (2009) 302–307.
197
[20] D.-D. Lee, D.-S. Lee, Environmental gas sensors, IEEE Sensors Journal 1 (2001) 214–224.
Biographies Alexandru Oprea received the diploma in physics from the University of Bucharest in 1976 and the PhD in solid state physics from the Central Institute of Physics, Bucharest, Romania in 1996. Since 2001 he is senior researcher in the Gas Sensor Group of the University of Tübingen, Germany. The research fields: thin films solar cells, high field electroluminescent devices, polymer and metal oxide gas sensors. Jérôme Courbat received his M.Sc. degree in microengineering from the Swiss Federal Institute of Technology, Lausanne (EPFL) in 2005 and his Ph.D. in the field of microchemical systems in 2010 at the same institution. He is currently a post-doc in the Sensors, Actuators and Microsystems Laboratory (SAMLAB) at EPFL in Neuchâtel, Switzerland. His research activities include the development and integration of low power and low-cost sensors on plastic foils and optical gas sensors. Danick Briand received his B.Eng. degree and M.A.Sc. degree in engineering physics from École Polytechnique in Montréal, Canada, in collaboration with the Institut National Polytechnique de Grenoble (INPG), France, in 1995 and 1997, respectively. He obtained his Ph.D. degree in the field of micro-chemical systems from the Institute of Microtechnology (IMT), University of Neuchâtel, Switzerland in 2001. He is currently a team leader at EPFL IMT Samlab in the field of EnviroMEMS, Energy and Enviromental MEMS. He has been awarded the Eurosensors Fellowship in 2010. He has been author or co-author on more than 150 papers published in scientific journals and conference proceedings. His research interests in the field of sensors and Microsystems include polymeric and Power MEMS, printed sensors and smart sensing systems on plastic and flexible foil, environmentally conscious Microsystems, green microtechnologies and microfabrication, and the development of microsystems for environmental and energy applications. Nicolae Bârsan received in 1982 his diploma in Physics from the Faculty of Physics of the Bucharest University and in 1993 his PhD in Solid State Physics from the Institute of Atomic Physics, Bucharest, Romania. Since 1995 he is a senior researcher at the Institute of Physical Chemistry of the University of Tübingen where, currently, is leading together with Udo Weimar the Gas Sensor research group. Udo Weimar received his diploma in physics 1989, his PhD in chemistry 1993 and his Habilitation 2002 from the University of Tübingen. He is currently a full Professor at the Faculty of Science of the University of Tübingen. His research interest focuses on chemical sensors as well as on multicomponent analysis and pattern recognition. Nico F. de Rooij received his Ph.D. degree from Twente University of Technology, The Netherlands, in 1978. From 1978 to 1982, he worked at the Research and Development Department of Cordis Europa N.V., The Netherlands. In 1982, he joined the Institute of Microtechnology of the University of Neuchâtel, Switzerland, as professor and head of the Sensors, Actuators and Microsystems Laboratory. Since 1987, he has been a lecturer at the Swiss Federal Institute of Technology, Zurich (ETHZ), and since 1989, he has also been a professor at the Swiss Federal Institute of Technology, Lausanne (EPFL). He is now the director of the Institute of Microengineering at EPFL. His research activities include microfabricated sensors, actuators and microsystems. He is head of the Sensors, Actuators and Microsystems Laboratory, with more than 30 staff members, more than 300 referred publications and several patents. In 2007, he was the recipient of the IEEE Junichi Nishizawa Medal for “Pioneering contributions to microsystem technology and effective transfer into industrial products and applications”.
Contents lists available at SciVerse ScienceDirect
Sensors and Actuators B: Chemical journal homepage: www.elsevier.com/locate/snb
Environmental monitoring with a multisensor platform on polyimide foil A. Oprea a,∗ , J. Courbat b , D. Briand b , N. Bârsan a , U. Weimar a , N.F. de Rooij b a
Institute of Physical and Theoretical Chemistry (IPC), University of Tübingen, Auf der Morgenstelle 15, D-72076 Tübingen, Germany Ecole Polytechnique Fédérale de Lausanne (EPFL), Institute of Microengineering (IMT), Sensors, Actuators and Microsystems Laboratory (SAMLAB), Rue Jaquet-Droz 1, P.O. Box 526, CH-2002 Neuchâtel, Switzerland b
a r t i c l e
i n f o
Article history: Received 16 August 2011 Received in revised form 16 February 2012 Accepted 29 February 2012 Available online 17 May 2012 Keywords: Multisensor platform Environmental monitoring Polyimide foil Metal-oxide sensor Capacitive sensor Calibration curve
a b s t r a c t A multisensor platform on plastic foil for environmental monitoring has been produced and its gas sensing performance, investigated. It is an array of conductometric metal-oxide (MOX) and capacitive polymer gas sensors integrated with a resistive platinum thermometer on a polyimide sheet substrate. The feasibility of simultaneous measurement of oxidizing and reducing gases, volatile organic compounds (VOCs), humidity and temperature has been demonstrated. MOX signals comparable with those of the devices realized on ceramic substrates have been obtained. Due to its structure, the platform is very versatile and, by using different sensor configurations and sensing materials, it allows the detection of a broad spectrum of gaseous analytes over wide concentration ranges. From the raw signals, temperature and humidity-corrected gas responses have been inferred which have been used for the calibration of the platform sensors. All the integrated devices were stable and gave reproducible signals for more than two months of operation, even when the MOXs ran continuously at 300 ◦ C. The performed investigation proved the device concept viability and the reliability of its practical implementation. © 2012 Elsevier B.V. All rights reserved.
1. Introduction The development of autonomous sensing systems gained significant interest within the last years due to the decreasing power consumption of the electronic components and the spread of wireless communication. At present it is a clear trend to integrate new components and expand the features of the older ones in order to make the whole assembly smarter. Monitoring environmental parameters may find relevance in several domains, such as human comfort and health, ambient pollution reduction or perishable goods preservation. The measuring of various parameters – temperature, relative humidity, gas concentration or pressure – with a single chip has already been proven on silicon substrates [1–4]. For example Li et al. [5] presented several types of transducers – calorimetric, capacitive, gravimetric, resistive – for gas sensing. Also, the use of plastic foil for sensors and actuators has been reported in the literature of the last years. Among other, anemometer [6], bolometer [7], thermometer [8–10], and metal-oxide [11,12] or capacitive gas sensors [13,14] have been produced on polyimide (PI) foils. Here, a multisensor platform for environmental monitoring on one polyimide substrate that integrates conductometric
∗ Corresponding author. E-mail address: [email protected] (A. Oprea). 0925-4005/$ – see front matter © 2012 Elsevier B.V. All rights reserved. doi:10.1016/j.snb.2012.02.095
metal-oxide (MOX) and polymeric capacitive gas sensors as well as a Pt thermometer is presented. In comparison with the investigations/implementations of different types of sensors on flexible substrates reported earlier by our group of authors, that is, the design, fabrication, and characterization of metal-oxide gas sensors [12,15], and capacitive sensor arrays [14], the multisensor platform addressed by this contribution represents a significant conceptual and technological advance. That is due to a higher degree of integration and, mainly, to the successful combination of different transducers, different types of sensing materials and different technologies. Moreover, the possibility to correct the temperature and humidity parasitic substrate contributions to the sensor responses allows achieving the same sensing potential as in the case of the Si-based platforms [3].
2. Experimental 2.1. Platform concept, structure and design The demonstrator configuration of the hybrid multisensor platform (6 mm × 6 mm) is presented in Fig. 1. It combines three different sensor types on the same polymeric substrate: two metaloxide and two capacitive gas sensing structures as well as a resistive thermometer. Depending on the concrete application needs, other configurations, with higher complexity, are feasible. The main
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 1. Optical micrograph of the multisensor platform fabricated on polyimide foil (up) and the device packaged on a PCB for its characterization (down).
advantages of combining sensors with different transducing principles are: - the possibility to sense gas mixtures containing different components spread over different ranges of concentrations - the possibility to correct the temperature influence on the sensor responses and calibration curves - the possibility to remove, in principle, the cross sensitivities by using extended arrays - the possibility to make predictions concerning the gas composition through multivariate data analysis The cross section of the demonstrator platform is displayed in Fig. 2. It enables the understanding of different sensor structures and their geometric parameters: • The MOX conductometric sensors are based on Pt interdigitated transducers with an active area of 100 m × 100 m, and an aspect ratio of 164. The aspect ratio indicates how many times the measured resistance of the sensing layer is reduced by the electrode configuration in respect with the standard sheet resistance of the same layer. The electrodes of the transducer are deposited
Fig. 2. Cross-sectional schematic of the multisensor platform on polyimide foil. The power consumption of the MOX sensors can be reduced by using a back-etched membrane.
191
over the heating Pt meander, from which they are electrically insulated by a polyimide layer (see Section 2.2). • Each capacitive transducer covers an area of 1.4 mm × 1.4 mm and has equal electrode width and spacing (10 m). A nominal capacitance lower than 10 pF has been targeted, which allows the direct readout of the capacitance with a commercial, high resolution, capacitance-to-digital converter. One capacitor – addressed from now on as “reference” capacitor – has been left uncoated, as a witness for the substrate residual response to gaseous analytes. The other one – to be addressed as “sensing” capacitor – has been covered with an appropriate polymer, providing the capacitive sensing function toward either a certain VOC or to humidity. Because of the additional polymer layer the capacitance of the sensing capacitor is higher than the capacitance of the reference capacitor with an amount (∼10–20%) that will be referred to as “sensing layer capacitance” (or “nominal sensing layer capacitance” if measured in dry synthetic air). The structure of the capacitive section of the multisensor platform allows differential operation [14], so that the reference capacitor signal and the nominal sensing layer capacitance can be subtracted from the sensing capacitor signal, in order to obtain the net gas response of the sensing layer, eliminating in this way the parasitic contributions of the substrate to the total capacitive response of the platform. The two, reference and sensing, capacitors differentially operated will be referred to as “differential sensor”. More information related to sensor evaluation in the differential operation mode is included in Section 3. • The resistive temperature transducer is a Pt meander realized on an area of 1.4 mm × 1.4 mm and having a targeted nominal resistance of 1 k at 0 ◦ C. At present the platform manufacture implies a certain technological complexity due to complementary technological steps being used. The foreseen “full plastic implementations” exclusively based on printed devices will considerably simplify the production chain premising the next generation of low cost and low power gas sensing platforms. 2.2. Fabrication The platform has been realized on a 50 m thick polyimide (PI) wafer (Upilex-50S from Ube Industries, Ltd), making use of a minimal number of conventional technological steps (see Fig. 2). The conducting elements of the sensors (thermoresistor, heaters, electrodes and pads) required only two levels of platinum (130 nm) deposited by DC sputtering over an adhesion layer of titanium (20 nm) and patterned by lift-off technique. The electrical insulation between the two layers of platinum was made by a 700 nm thick polyimide film (PI-2737 from HD Microsystems). To reduce the power consumption of the MOX sensors, optional 3 m thin membranes could be formed through dry etching from the backside of the polyimide foil. The fabrication process used for the multisensor platform was the one previously developed for single metal oxide gas sensors on PI, and is fully described in reference [15]; no extra step to integrate the capacitive transducers and temperature sensor has been required. Onto the transducers, all functional materials for gas sensing have been drop-coated. In the first step, the metal-oxide layers have been deposited: SnO2 loaded with 0.2 wt% Pd, which will be referred to as SnO2 :0.2%Pd [16] and WO3. They required a subsequent annealing at 400 ◦ C to enable the stabilization of the film properties. Afterwards, the sensing capacitor has been covered with either cellulose acetate butyrate (CAB) for humidity measurement or polyetherurethane (PEUT) for VOCs detection. The polymers were allowed to naturally dry at room temperature. For laboratory
192
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
characterization the multisensor platforms have been packaged (Fig. 1b) on PCBs (printed circuit boards) using conductive epoxy glue (Epotek H20E). Two rows of 8 standard gold-plated pins (with 2.54 mm spacing) soldered to the PCB made the platforms compatible with wide (15.24 mm) DIL sockets. 2.3. Electrical and gas measurement protocols The platforms, packaged on the DIL PCBs and placed in a measuring chamber made of Teflon® , were characterized with a computer controlled gas-mixing and measuring system at different relative humidity (r.h.) levels and gas concentrations with values in the TWA (time weighted average) range or below. CO (20–80 ppm), NO2 (0.1–1 ppm), EtOH (2–20, 20–600, or 300–3000 ppm) and relative humidity levels (0–70%) have been dosed as successive concentration pulses in the carrier gas – synthetic air or synthetic air with 50% r.h. background – at a flow rate of 200 sccm. Between each individual exposure pulse the samples were allowed to recover. The exposure protocols are always depicted in the upper side of the graphics containing raw data in order to facilitate the understanding of the platform response. The voltage applied to the heater of the MOX sensors, the resulting currents and the resistances of their gas sensing layers were monitored with a standard laboratory multichannel multimeter (Agilent 34970A). The readout of the capacitive transducers (used as reference or for the humidity or VOCs sensing) was ensured with a 24 bit capacitance-to-digital converter AD7746 from Analog Devices, Inc. This IC has a very high resolution, theoretically of 4aF, and a fairly low power consumption at low acquisition rates. An additional voltage/resistance input of the converter has been used to monitor the resistance of the temperature sensor. All digital data were transmitted to a PC through an I2 C-to-USB converter using the dedicated software developed in cooperation by IPC Tübingen and JLM-Innovation GmbH, Germany. 3. Results and discussion 3.1. Electrical parameters/characteristics The heaters of the metal oxide gas sensors have a nominal resistance of 150 at 0 ◦ C. To reach an operating temperature of 300 ◦ C in continuous heating mode, they need an individual power consumption of 15.5 mW in the micro-machined variant or 30.1 mW in the standard one. The MOX layer resistance, under operation conditions, ranges from several k to few M and depends on the ambient composition (see Section 3.2.3). The resistive thermometer has a resistance of ∼1.16 k at room temperature (1 k targeted nominal value at 0 ◦ C). After a stabilization period (few days at a temperature higher than 60 ◦ C), its response shows a good linearity (nonlinearity between −20 ◦ C and 150 ◦ C smaller than 0.3%). A 0.002 ± 0.0001 ◦ C−1 thermal coefficient of the resistance (TCR) has been experimentally determined in a calibration oven. This value is smaller than the one of the massive pure platinum (0.00391 ◦ C−1 ) because of the layer dimensions and the limited annealing range allowed by the plastic substrate. The capacitive transducers have a nominal capacitance of ∼2 pF. 3.2. Gas sensing performances The evaluation of the multisensor platforms has been performed with a test mixture containing, synthetic air with different levels of background humidity, as carrier gas, and VOCs vapors or inorganic gases as analytes. The capacitive polymer sensors of the platforms are sensitive only to VOCs and water vapors. Indeed, the sorption of the vapors in the amorphous polymer is a solvation process leading to a significant increasing of the analyte concentration in polymer in respect with its concentration in the ambient [17]. This is not
the case for the gases; at room temperature they are well above the critical isotherm and cannot condense in the polymer. In their turn the MOX chemoresistors are mainly sensitive to gaseous analytes that can undergo surface reaction involving a charge exchange with conduction or valence bands (depending on material conduction type – n or p) of the material. Therefore it is not important if the analyte is in vapour or gas phase. The analytes able to promote an increase in the concentration of the free electrons in the conduction band of an n-type MOX semiconductor or to reduce the number of free holes in the valence band of a p-type MOX semiconductor are usually referred to as “reducing gases” (for a detailed explanation see [18]). If the reverse trend occurs the analytes are referred to as oxidizing. Because the demonstrator platform contains only one differential capacitive sensor (recall Section 2.1) it could be dedicated to either humidity or a certain VOC. The choice, besides humidity, has been ethanol, because EtOH vapors are easily detected by the PETU polymer layer but, in the same time, have a strong reducing character in the interaction with the heated MOX being used (WO3 and SnO2 :0.2%Pd). This choice made it possible to check out how the MOX, very sensitive to ethanol at low concentrations, and the polymer (PEUT) sensors complement each other over a wide range of concentrations (1–3000 ppm). Several multisensor platforms have been systematically tested and evaluated during many weeks of experiments. They were all showing good functionality, comparable sensing parameters and unexpected reliability over long operation periods. 3.2.1. Parameters used to quantify the gas sensor performance In the literature, there are different ways to evaluate the gas sensor performance. To avoid confusion, a short overview of the parameters used in this section is given here. The main experimental data is the sensor signal. In the case of conductometric gas sensors the signal is the sample resistance (R). The baseline of the MOX sensors stands for the sensor signal (sensor resistance R0 ) in the absence of the target gas, that is, when the sensor faces the reference ambient only. The sensor response (SR) is the ratio between the sensor signals in the absence (R0 ) and presence (R) of the analyte: SR = R0 /R for reducing gases but SR = R/R0 for oxidizing gases; that is because, by convention, SR is chosen to be larger than 1 in the presence of the target gas. SR has no unit. By displaying the sensor responses against the gas concentration one obtains the sensor calibration curve and the sensor sensitivity, as the slope of the calibration curve. For non-linear calibration curves the sensitivity is concentration dependent; in such cases double logarithmic scales are frequently used, which facilitate the comparisons of the sensor responses over wide ranges of concentrations and naturally guide to the definition of the logarithmic sensitivity: the sensor response increment over ten times (decade) concentration increment. If the sensor response on the target gas concentration follows a power low (as expected for gas sensing mechanisms involving oxygen ionosorption [18]) the calibration curve will graphically appear as linear and the logarithmic sensitivity will be independent of concentration. The sensitivity unit is the reciprocal of the concentration unit, usually ppm−1 while the unit of the logarithmic concentration is (decade)−1 . The sensor stability is expressed through the stability of the signals, as long as the responses and the sensitivity are related to the baseline, which depends on the composition of the reference ambient. 3.2.2. Differential capacitive measurements of humidity Fig. 3 shows the raw and differential signals obtained with the reference and sensing (CAB-coated) capacitors of a platform (MSP2) dedicated to humidity measurements, which has been exposed to humidity (0–70%) and to the potential interfering gases such as
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 3. The response of the CAB differential capacitive sensor from the MSP-2 platform to carbon monoxide, ethanol, and relative humidity. The upper panel displays the exposure protocol. The raw signals of the sensing and reference capacitors are as well shown, but arbitrary shifted from the nominal capacitance (∼2pF) in order to appear with the same baseline and on the same panel with the CAB differential sensor response. The CAB differential capacitive sensor responds to humidity only.
CO (20–80 ppm) and ethanol (2–20 ppm). The differential response is obtained by subtracting the reference capacitor signal and the nominal capacitance of the CAB layer from the sensing capacitor signal as already explained in Section 2.1. The CAB film response toward humidity was selective. No visible response was obtained during the exposure to CO and low concentrations of ethanol. Table 1 presents the response and recovery times of the reference and sensing capacitors. The use of the differential measurement decreased the response time by a factor of six between 0 and 50% r.h., while the recovery time was reduced twenty-four times in respect to the sensing capacitor alone. Similar values are found in the literature [19]. These significant improvements resulted from the elimination, by subtraction, of the slow humidity response of the compact PI substrate (due to long diffusion times), otherwise included in the overall (sensing layer and substrate) sensor response. 3.2.3. Responses of the whole multisensor platforms to gases and humidity A typical result of the multisensor platform is presented in Fig. 4. The capacitive sensor has been based on PEUT as sensing material because it is suitable for VOCs detection once its cross sensitivity to humidity can be compensated (for example, by using an additional humidity sensor on a more complex multisensor platform or another platform whose sensing capacitor is dedicated to humidity). A simpler variant, suggested in Ref. [14], would be to use the substrate itself as humidity sensor, if the response and recovery times are not critical issues. As expected, the WO3 sensor is more sensitive to NO2 [20] than to EtOH, as shown in Fig. 4c (the slopes of the calibration curves in double logarithmic scales represent the logarithmic sensitivities of the gas sensors); its response to NO2 was practically not affected by the background humidity, while that one to EtOH almost quenched at 50% r.h. The humidity thus improved the selectivity of WO3 toward NO2 . The opposite happened for the SnO2 :0.2%Pd sensor. In both dry and humid air, the sensor exhibited a higher Table 1 Response and recovery time (t90% ) of the capacitive sensors between 0 and 50% of relative humidity.
Uncoated capacitor CAB-coated capacitor Differential measurement
tres,90% (min)
trec,90% (min)
29 25 3.8
113 105 4.3
193
Fig. 4. Gas response of the capacitive and chemoresistive gas sensors to nitrogen dioxide (0.1–1 ppm), ethanol (2–20 ppm), and relative humidity (0–70%). (a) Gas protocol; (b) differential PEUT capacitive sensor response (10 times magnified), PEUT sensing capacitor signal and reference capacitor signal. The PEUT sensing capacitor and reference capacitor signals (in fact in the 2 pF range) have been arbitrary shifted in order to be displayed on the same panel with the differential capacitive sensor response; (c) gas response of the SnO2 :0.2%Pd and respectively WO3 MOX sensors. Test performed on MSP-6 platform.
sensitivity for EtOH than for NO2 . In the presence of the humidity, the response to NO2 was reduced, improving the selectivity toward EtOH. To explore the detection of EtOH with the PEUT-coated capacitive sensor, the same kind of gas exposure sequence has been used for a different EtOH range: 300–3000 ppm. The results show (Fig. 5) a clear response to alcohol, almost independent of the humidity background. On the plot the response to humidity is larger than that to ethanol because of the high humidity levels being used in the experiment (50% r.h. at 20 ◦ C correspond to an absolute humidity value of roughly 11,500 ppm). However, the polymer is, in absolute terms, more sensitive to EtOH than to humidity. As all highly sensitive MOX sensors, the platform MOX channels will saturate with the increase of the target gas concentration, even if still in the TWA range (e.g. ∼500 ppm for EtOH). Fortunately, in the alcohol case, one can still measure by using the polymer (PEUT)
Fig. 5. Capacitive and chemoresistive responses of the multisensor platform to nitrogen dioxide (0.1–1 ppm), ethanol (300–3000 ppm), and relative humidity (0–70%). (a) Gas protocol; (b) the response of the PEUT differential capacitive sensor (10 times magnified) and the signals from the sensing and reference capacitors (see the explanations given in the caption of the Fig. 4b; the background humidity level is designated as B.H.L. on the response curve of the PEUT differential capacitive sensor; (c) Gas response of the MOX sensors coated with SnO2 :0.2%Pd and WO3 . Test performed on MSP-6 platform.
194
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 6. The response (R/R0 ) of the SnO2 :0.2%Pd and WO3 MOX gas sensors to humidity (baseline dependency on the background humidity) at 300 ◦ C operating temperature. Eight measurements have been performed with the same device resulting in a precision better than 95%. Test performed on MSP-2 platform.
differential capacitive sensor. Moreover, the multisensor platform has an important advantage here: a smart driving electronic can switch off the MOX heaters and validate only the capacitive outputs, preventing inaccurate records and saving power. 3.2.4. Dependency of the MOX sensors baseline on the humidity The background humidity influence on the baseline of the SnO2 :0.2%Pd and WO3 gas sensors is presented in Fig. 6, as sensor response referred to dry synthetic air. For the SnO2 :0.2%Pd based sensor the largest effects occur for the 0–10% r.h. interval, namely a factor of ∼2.5. In the much wider range from 10 to 70% r.h. only ∼20% changes of the baseline are recorded. This represents a factor of ∼1.2 in response, small in respect with the usual responses reported later on for the TWA concentrations of ethanol and NO2 . The situation is even better in the case of WO3 sensors, with a factor of 1.1 between 10 and 70% r.h. These values are slightly varying from platform to platform, but do not exceed a factor of 1.6 for humidity range between 20% and 70% r.h., which is the interesting one for, e.g., indoor applications. 3.3. Reproducibility and stability of the sensors The reproducibility of the CAB differential capacitive sensor response has been tested together with the sensor reliability during several weeks of uninterrupted operation. No major degradation is visible in Fig. 7, which presents the results. The highest decrease in the response (and, consequently, in sensitivity) over a twenty-day period, of 0.2%/day, has been observed for 70% r.h. That might be due to the repeated NO2 exposures the sensing capacitors of the platform undergone, as it has not been encountered for the individual capacitive humidity sensors exposed to humidity only.
Fig. 7. Stability of the CAB differential capacitive humidity sensor over a period of 20 days. The periodic instabilities are due to the deactivation of the air conditioning system in the laboratory during the weekends. Test performed on MSP-2 platform.
Fig. 8. The temperature of the MOX sensor heaters and the Pt thermometer readout on the multisensor platform MSP-6. The supplied power was ∼17 mW and the ambient temperature 25 ◦ C. The heaters of both WO3 and SnO2 :0.2%Pd sensors showed a high stability with a temperature variation of less than 1 ◦ C. During the initial operation period (missing from data), the sensors have been heated at other temperatures.
The temperature variation of the MOX sensors when operating at constant heating voltage over 48 days is presented in Fig. 8. The sensors exhibited a very constant behavior after a formation/stabilization period. The temperature variations of the WO3 and SnO2 :0.2%Pd coated sensors were respectively of 0.6 ◦ C and 0.5 ◦ C in the last 44 days. As well as stable was the Pt thermometer; this is, however, not surprising since the operation conditions are much friendly for this device (see Fig. 8). As a representative example for the stability of the MOX devices, the resistance variation of a WO3 sensor over a period of 21 days is shown in Fig. 9 – the heater has been kept at ∼300 ◦ C. The sensors showed a good reproducibility. After a stabilization time of two days, the baseline, the response to 400 ppb NO2 and to 1000 ppm EtOH drifted by 2.2%, 3.4% and 6.5%, respectively (50% r.h. background). Translated into response values, the changes in the sensor resistance accounted for in Fig. 9 represent a factor less than 1.008 for ethanol and less than 1.004 for NO2 . 3.4. Calibration curves From the above-discussed responses of the CAB differential capacitive sensor toward humidity (see Figs. 4 and 5), a calibration curve could be derived, which is displayed in Fig. 10a. It is fairly linear, with a slope (sensitivity) of 2.45 fF/%r.h. The calibration curve of the capacitive humidity sensor can be used to predict the actual humidity but also to compensate the cross sensitivity of the polymer and MOX sensors to humidity, resulting in a more accurate
Fig. 9. Stability of the WO3 sensor over a period of 20 days; the sensor has been operated at 300 ◦ C. Device stabilization occurred after two days of exploitation. The time variations of the sensing layer resistance under constant exposure is less than 7%, independent of the target gas being supplied (in wet environment). Test performed on MSP-7 platform.
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Fig. 10. (a) The response of the CAB differential capacitive sensor and that of the reference capacitor (substrate) from MSP-2 platform to humidity. Eight measurements performed with the same device; mean statistical error <4% for CAB and <6% for the substrate. (b) The response of the PEUT differential capacitive sensor from the MSP-6 platform to ethanol and its cross sensitivity to humidity. Five consecutive measurements; mean statistical error <12% for ethanol and 5% for H2 O. Almost no reference capacitor response to ethanol has been observed.
determination of the gas and VOCs concentrations. For the same purpose it is possible to exploit the parasitic substrate response to humidity, as addressed at the beginning of the paragraph in Section 3.2.3 (see Fig. 10a). The time constants of this readout are, however, significantly higher than those of the differential capacitive sensors (see Table 1.). The plots in Fig. 10b account for the linearity and reproducibility of the responses delivered by the PEUT differential capacitive sensor of the platform against EtOH. In a first step the calibration curve of PUET against ethanol in dry synthetic air is extracted from the differential capacitive responses to the five incremental ethanol concentration pulses (300, 600, 1000, 2000, 3000 ppm). By a similar procedure one obtains the calibration curve toward humidity, considered in its turn as analyte. The 10–70% r.h. pulses in the middle of the exposure are used for this purpose. In the presence of background humidity the baseline of the differential response shifts upwards to the BHL level in Fig. 5b. It has to be subtracted in order to obtain the contribution of the ethanol only (curve “humidity subtracted” in the same figure) from which the ethanol calibration is derived (Fig. 10b, “Ethanol; 50% r.h.”). More detailed description of the approach can be found in [14]. Due to the additional interactions with the polymer and target analyte, the humidity effect is not exclusively additive to the EtOH one, resulting in a small the separation of the calibration lines “Ethanol; 0% r.h.” and “Ethanol; 50% r.h.” in Fig. 10b. Fig. 11 shows the calibration curves obtained for the metal-oxide gas sensors based on WO3 and SnO2 :0.2%Pd sensing materials. The calibration curves of both MOX sensors are little influenced by a background humidity in the range of 20% r.h. −70% r.h. Actually, the WO3 response to NO2 was almost independent of the
195
Fig. 11. Calibration curves of the MOX gas sensors based on (a) SnO2 :0.2%Pd and (b) WO3 sensing layers. Test performed on MSP-6 platform.
background humidity level while SnO2 :0.2%Pd response to EtOH displayed only a reduced dependency on the background humidity. As observed above, on the response plots, the presence of the background humidity increases the selectivity of WO3 layer toward NO2 and that of the SnO2 :0.2%Pd layer to ethanol. Numerical evaluations of the cross sensitivities presented by the platform sensors toward the tested gases/vapors, other than the target ones (NO2 for WO3 , ethanol for SnO2 :0.2%Pd and PEUT and humidity for CAB) are included in Table 2. It demonstrates that for environment monitoring purposes, where precisions of 10% from TWA are more than satisfactory, the MOX sensors do not need any correction as well as the CAB capacitive one. The PEUT capacitive sensor only has a high cross sensitivity to humidity. Fortunately this cross sensitivity is almost additive: the responses to ethanol and background humidity are almost independent and therefore they practically superpose to yield the total sensor response. In consequence the background humidity contribution can be removed by using the humidity sensor (simple subtraction). The calibration curves in Fig. 11 also contain subsidiary information about the gas sensing mechanism of the platform MOX devices. As explained in Section 3.2.1 the linear shapes of the calibration curves depicted in the figure are due to the double logarithmic scales and represent power law dependencies of sensor responses on the EtOH and respectively NO2 concentrations. This means that the chemoreceptive and transducing processes [18] in the MOX films coating plastic foils are the same as in the case of MOX films on ceramic substrates; they are not essentially affected by the underlying organic material. 3.5. Operation algorithms and measurement scenarios On the basis of the above reported results one can detail the functional algorithm of the demonstrator-platform when monitoring the ambient atmosphere in locations where, nearby background humidity, NO2 and ethanol are also present (actually the aim of the application). For instance a slowly changing humidity is assumed,
196
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
Table 2 Cross sensitivity of the platform sensors. The cross sensitivity is evaluated as the ratio (in percents) of the sensor responses to the test and target gases respectively, both having the TWA concentrations. The humidity has been also included with the formal TWA value of 50% r. h., because it is not harmful (despite the human discomfort above 25 ◦ C at saturation). Below 0 ◦ C the ambient humidity is very reduced (∼9000 ppm at 0 ◦ C and only ∼200 ppm at −20 ◦ C). Analyte
TWA [ppm]
MOX sensor responses
Cross sensitivity [%]
SnO2 :0.2%Pd NO2 CO EtOH H2 O Analyte
NO2 CO EtOH H2 O
5 30 400 11,500 TWA [ppm]
5 30 400 11,500
4 1.3 150.0 1.5
WO3 200.0 1.6 5 2.7
Polymer sensor responses [fF] PEUT
<1 <1 2 102
<0.1 <0.1 4 97
1. All sensors are powered on and allowed to warm-up. 2. All raw signals are recorded and used to obtain: a. the temperature b. the gas responses of the MOXs resistors c. the single capacitance values for reference (uncoated) and Ethanol-sensing (PEUT coated) capacitors 3. In parallel: a. the relative humidity at the given temperature is evaluated from the reference capacitance by using its calibration curve (like the one in Fig. 10a) already stored in the system database or by interpolating the two nearest ones. b. the response of the PEUT differential capacitive sensor at the ambient temperature is estimated from the signals of the sensing (PEUT) and reference capacitors. c. the concentrations of NO2 and EtOH are evaluated with good precision from the individual calibration curves (temperature independent) of the WO3 and respectively SnO2 :0.2%Pd sensors because: i. the influence of the humidity (in the range 30–70% r.h.) is very small (see Sections 3.2.4 and 3.4 and Table 2) ii. the cross sensitivity of WO3 to EtOH and that of SnO2 :0.2%Pd to NO2 with background humidity is reduced (see Section 3.4, Fig. 11 and Table 2). 4. From the response of the PEUT differential capacitive sensor the ethanol concentration is estimated in agreement with Section 3.4. Depending on the availability of the calibration curves at the measuring temperature interpolation might be required. 5. If the ethanol response of the capacitive platform channel overcomes a preset level the electronics switches off the SnO2 :0.2%Pd heating (see Section 3.2.3) reading only the capacitive response to ethanol. 6. All acquired values in one run are displayed or/and compared with threshold ones enabling/disabling an alarm procedure. At this stage the operation is resumed from the step 2 in continuous cycles until the multisensor platform service is no more required and it is let in standby modus or powered of.
WO3
2.7 0.9 – 1.0
– 0.8 2.5 1.4 Cross sensitivity [%]
CAB
which allows the use of the reference capacitor as humidity sensor (accounted for just above, in Section 3.4). Rapid variations of the background humidity have to be evaluated according to the Section 3.2.3. The intelligent system (microcontroller or PC) driving the platform, should already contain the temperature depending calibrations curves of the individual sensors in its own database (the ambient temperature has almost no relevance for the MOX sensors because they are already heated at a constant temperature).
SnO2 :0.2%Pd
CAB <1 <1 2 –
PEUT <3 <3 – 2200
4. Conclusion The integration of different sensing devices for environmental monitoring on the same polyimide substrate toward a multifunctional platform has been demonstrated for dissimilar detection principles, transducer types and sensing materials. During the performed investigations, all platform components exhibited good sensitivity, a fair selectivity and a satisfactory stability throughout the test period. It has been demonstrated that by joint operation of the Pt thermo-resistive, polymer-based capacitive and metaloxide conductometric sensors, several interfering gases can be detected and, to some extent, separated at hardware level. Thus, the capacitive structures monitored the humidity and ethanol (as representative VOC), while the metal-oxide ones monitored the oxidizing and reducing gases (NO2 , CO) as well as the reducing ethanol vapors. Beyond the achievements in terms of sensing parameters, the compensation of the substrate intrinsic (parasitic) response to the target gases, mainly to humidity, through the differential readout, can be regarded as a significant step forwards. Additionally, one has to remark the stable behavior of the MOX layers, even if deposited over double sandwiched plastic/Pt areas and continuously heated at 300 ◦ C. Due to the high versatility of the platform, the number of its elements can be further increased in order to extend the detection spectrum toward other classes of analytes and to expand the functional capabilities. Moreover, because it is possible to produce several dissimilar sensors on a single substrate, the assembly step toward hybrid arrays can be eliminated. A considerable improvement is foreseen at mini-system level where the humidity and temperature corrected responses delivered by the smart platform could be further deconvoluted through multivariate data analysis or other appropriate mathematical algorithms/methods. Also, the power consumption issues due to the MOX operation conditions, critical for autonomous or supply free (RFID) applications, can be circumvented if the substrate variant with back-etched hot plates is employed. Further fabrication simplicity and cost reduction are feasible by using only additive fabrication techniques, such as printing. The main challenge now is the platform improvement toward a smart sensing system with embedded electronics and signal processing facilities, based on nonconventional and cheap materials and technologies.
Acknowledgments We would like to thank the staff of the clean room facility from Center of Micronanotechnology (CMI), Ecole Polytechnique Fédéral de Lausanne (EPFL), for their help in the processing of the devices.
A. Oprea et al. / Sensors and Actuators B 171–172 (2012) 190–197
References [1] C. Hagleitner, D. Lange, A. Hierlemann, O. Brand, H. Baltes, Cmos single-chip gas detection system comprising capacitive, calorimetric and mass-sensitive microsensors, IEEE Journal of Solid-State Circuits 37 (2002) 1867–1878. [2] A.D. DeHennis, K.D. Wise, A wireless microsystem for the remote sensing of pressure, temperature, and relative humidity, Journal of Microelectromechanical Systems 14 (2005) 12–22. [3] M. Hautefeuille, C. O’Mahony, B. O’Flynn, K. Khalfi, F. Peters, A MEMS-based wireless multisensor module for environmental monitoring, Microelectronics Reliability 48 (2008) 906–910. [4] J. Hayes, S. Beirne, K.T. Lau, D. Diamond, Evaluation of a low cost wireless chemical sensor network for environmental monitoring, in: Proceedings of the IEEE Sensors 2008 Conference, 2008, pp. 530–533. [5] Y. Li, C. Vancura, D. Barrettino, M. Graf, C. Hagleitner, A. Kummer, M. Zimmermann, K.U. Kirstein, A. Hierlemann, Monolithic CMOS multi-transducer gas sensor microsystem for organic and inorganic analytes, Sensors and Actuators, B: Chemical 126 (2007) 431–440. [6] U. Buder, R. Petz, M. Kittel, W. Nitsche, E. Obermeier, Aeromems polyimide based wall double hot-wire sensors for flow separation detection, Sensors and Actuators, A: Physical 142 (2008) 130–137. [7] A. Mahmood, D.P. Butler, Z. Celik-Butler, Micromachined bolometers on polyimide, Sensors and Actuators, A: Physical 132 (2006) 452–459. [8] T. Kinkeldei, C. Zysset, K. Cherenack, G. Troester, Development and evaluation of temperature sensors for textile integration, in: Proceedings of IEEE Sensors 2009, Christchurch, New Zealand, 2009, pp. 1580–1583. [9] D.J. Lichtenwalner, A.E. Hydrick, A.I. Kingon, Flexible thin film temperature and strain sensor array utilizing a novel sensing concept, Sensors and Actuators; A: Physical 135 (2007) 593–597. [10] C.Y. Lee, S.J. Lee, G.W. Wu, Fabrication of micro temperature sensor on the flexible substrate, nanotechnology, in: 7th IEEE Conference on Nanotechnology, Hong Kong, China, 2007, pp. 1050–1053. [11] D. Briand, S. Colin, J. Courbat, S. Raible, J. Kappler, N.F. de Rooij, Integration of MOX gas sensors on polyimide hotplates, Sensors and Actuators, B: Chemical 130 (2008) 430–435. [12] J. Courbat, D. Briand, L. Yue, S. Raible, N.F. de Rooij, Ultra-low power metal-oxide gas sensor on plastic foil, in: Proc. of the Transducers’09 Conference, Denver, USA, 2009, pp. 584–587. [13] A. Oprea, N. Barsan, U. Weimar, M.L. Bauersfeld, D. Ebling, J. Wollenstein, Capacitive humidity sensors on flexible RFID labels, Sensors and Actuators, B: Chemical 132 (2008) 404–410. [14] 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, Sensors and Actuators, B: Chemical 140 (2009) 227–232. [15] J. Courbat, M. Canonica, D. Teyssieux, D. Briand, N.F. de Rooij, Design and fabrication of micro-hotplates made on polyimide foil: electrothermal simulation and characterization to achieve power consumptions in the low mW range, Journal of Micromechanics and Microengineering 21 (2011) 015014. [16] J. Kappler, N. Barsan, U. Weimar, A. Dieguez, J.L. Alay, A. Romano-Rodriguez, J.R. Morante, W. Gopel, Correlation between XPS, Raman and TEM measurements and the gas sensitivity of Pt and Pd doped SnO2 :0.2%Pd based gas sensors, Fresenius Journal of Analytical Chemistry 361 (1998) 110–114. [17] J.W. Grate, M.H. Abraham, Solubility interactions and the design of chemically selective sorbent coatings for chemical sensors and arrays, Sensors and Actuators B 3 (1991) 85–111. [18] N. Barsan, U. Weimer, Conduction model of metal oxide gas sensors, Journal of Electroceramics 7 (2001) 143–167. [19] E. Zampetti, S. Pantalei, A. Pecora, A. Valletta, L. Maiolo, A. Minotti, A. Macagnano, G. Fortunato, A. Bearzotti, Design and optimization of an ultra thin flexible capacitive humidity sensor, Sensors and Actuators, B: Chemical 143 (2009) 302–307.
197
[20] D.-D. Lee, D.-S. Lee, Environmental gas sensors, IEEE Sensors Journal 1 (2001) 214–224.
Biographies Alexandru Oprea received the diploma in physics from the University of Bucharest in 1976 and the PhD in solid state physics from the Central Institute of Physics, Bucharest, Romania in 1996. Since 2001 he is senior researcher in the Gas Sensor Group of the University of Tübingen, Germany. The research fields: thin films solar cells, high field electroluminescent devices, polymer and metal oxide gas sensors. Jérôme Courbat received his M.Sc. degree in microengineering from the Swiss Federal Institute of Technology, Lausanne (EPFL) in 2005 and his Ph.D. in the field of microchemical systems in 2010 at the same institution. He is currently a post-doc in the Sensors, Actuators and Microsystems Laboratory (SAMLAB) at EPFL in Neuchâtel, Switzerland. His research activities include the development and integration of low power and low-cost sensors on plastic foils and optical gas sensors. Danick Briand received his B.Eng. degree and M.A.Sc. degree in engineering physics from École Polytechnique in Montréal, Canada, in collaboration with the Institut National Polytechnique de Grenoble (INPG), France, in 1995 and 1997, respectively. He obtained his Ph.D. degree in the field of micro-chemical systems from the Institute of Microtechnology (IMT), University of Neuchâtel, Switzerland in 2001. He is currently a team leader at EPFL IMT Samlab in the field of EnviroMEMS, Energy and Enviromental MEMS. He has been awarded the Eurosensors Fellowship in 2010. He has been author or co-author on more than 150 papers published in scientific journals and conference proceedings. His research interests in the field of sensors and Microsystems include polymeric and Power MEMS, printed sensors and smart sensing systems on plastic and flexible foil, environmentally conscious Microsystems, green microtechnologies and microfabrication, and the development of microsystems for environmental and energy applications. Nicolae Bârsan received in 1982 his diploma in Physics from the Faculty of Physics of the Bucharest University and in 1993 his PhD in Solid State Physics from the Institute of Atomic Physics, Bucharest, Romania. Since 1995 he is a senior researcher at the Institute of Physical Chemistry of the University of Tübingen where, currently, is leading together with Udo Weimar the Gas Sensor research group. Udo Weimar received his diploma in physics 1989, his PhD in chemistry 1993 and his Habilitation 2002 from the University of Tübingen. He is currently a full Professor at the Faculty of Science of the University of Tübingen. His research interest focuses on chemical sensors as well as on multicomponent analysis and pattern recognition. Nico F. de Rooij received his Ph.D. degree from Twente University of Technology, The Netherlands, in 1978. From 1978 to 1982, he worked at the Research and Development Department of Cordis Europa N.V., The Netherlands. In 1982, he joined the Institute of Microtechnology of the University of Neuchâtel, Switzerland, as professor and head of the Sensors, Actuators and Microsystems Laboratory. Since 1987, he has been a lecturer at the Swiss Federal Institute of Technology, Zurich (ETHZ), and since 1989, he has also been a professor at the Swiss Federal Institute of Technology, Lausanne (EPFL). He is now the director of the Institute of Microengineering at EPFL. His research activities include microfabricated sensors, actuators and microsystems. He is head of the Sensors, Actuators and Microsystems Laboratory, with more than 30 staff members, more than 300 referred publications and several patents. In 2007, he was the recipient of the IEEE Junichi Nishizawa Medal for “Pioneering contributions to microsystem technology and effective transfer into industrial products and applications”.