Sensors and Actuators B 91 (2003) 211–218
Fast response humidity sensors for a medical microsystem A. Te´telin*, C. Pellet, C. Laville, G. N’Kaoua Laboratoire IXL, CNRS UMR 5818, Universite´ Bordeaux I, 351 cours de la Libe´ration, 33405 Talence Cedex, France
Abstract A fast response humidity sensor was fabricated to equip a medical microsystem for diagnosis of pulmonary diseases. Its main characteristics are reported in this paper. The sensor is based on a capacitor made of a divinyl siloxane benzocyclobutene (BCB) thin film in between parallel plate electrodes. It was fabricated with compatible CMOS technology. It exhibits good linearity, good sensitivity, and a short response time. Equilibrium capacitance variations versus humidity can be considered linear with a linearity error less than 2% of the humidity level. The sensitivity is 0.1 pF by per cent of the humidity level. The device displays a typical adsorption time of 650 ms, a minimum adsorption time of 400 ms, and a desorption time of a few seconds at ambient temperature. Its performances were compared to other types of capacitive humidity sensors fabricated for the same medical use. The static behavior and the dynamic behavior of the device are reported. They are interpreted according to physical processes of gas adsorption and diffusion in and through glassy polymers. The suitability of the conventional Dual-Mode model to explain water sorption in BCB is discussed. This investigation is a starting point in a modeling process to improve the design of the sensor interface circuitry. # 2003 Elsevier Science B.V. All rights reserved. Keywords: Humidity sensor; Sensor modeling; Water sorption; Biomedical microsystem
1. Introduction A humidity sensor was fabricated to equip a multisensor microsystem for pulmonary functions diagnosis [1,2]. It should be able to measure humidity variations in a protracted breathing out in real-time. Hence, it should present good linearity, good sensitivity and above all a short response time. The permittivity of divinyl siloxane benzocyclobutene (BCB) is related to the amount of water adsorbed or desorbed in it [3]. When a BCB film is enclosed in a capacitor, it provides a measure of the relative humidity (RH) level in breath by the resulting change in capacitance. The BCB film is placed between two parallel plate electrodes. The upper electrode is a grid to allow humidity to penetrate the sensitive layer. A heating resistor was included on the chip to solve the problems due to condensation (Fig. 1). Temperature is controlled during measurements, kept close to breath temperature. To point out that the performances of the device suit the considered use, it was compared to other types of capacitive humidity sensors fabricated for the same use. They are coplanar interdigitated electrode capacitors covered by polyimide, etched polyimide, silicon oxide or BCB films. *
Corresponding author. Tel.: þ33-5-56-84-26-33; fax: þ33-5-56-37-15-45. E-mail address:
[email protected] (A. Te´telin).
Mechanisms of water sorption in BCB films were studied as a starting point in a modeling process. The resulting model is due to achieve system-level simulation to design the interface circuitry of the sensor. The relationships that govern adsorption and diffusion of water in BCB were to be understood. This paper provides a characterization of water vapor sorption isotherms for BCB thin films in ambient conditions. It also provides an explanation to the kinetics of water sorption in BCB. The suitability of the conventional Dual-Mode model [4] to describe the equilibrium behavior and the kinetics of the sensor is discussed.
2. Sensor technology and performances 2.1. Technology The sensor is a parallel plate electrode capacitor with the sensitive layer in between (Fig. 1). Electrodes were made of gold–titanium, which does not oxidize in contact with humid environments. The lower electrode is a full plate, the upper electrode is a grid that was deposited on the sensitive layer. The electrodes are 1.4 mm thin. Each bar of the grid is 20 mm wide and 3 mm long. The sensitive layer is made of cyclotene (BCB) 4024-40 from the Dow Chemical Company [3]. It is 1.65 mm thin and its surface area is 9 mm2. The sensor was fabricated with a CMOS compatible process.
0925-4005/03/$ – see front matter # 2003 Elsevier Science B.V. All rights reserved. doi:10.1016/S0925-4005(03)00090-X
212
A. Te´ telin et al. / Sensors and Actuators B 91 (2003) 211–218
2.3. Linearity and sensitivity
Fig. 1. Parallel plate electrode device: cross-sectional view and top view.
The substrate is a p-doped silicon wafer covered by SiO2. The first level of metal was sputtered on the SiO2 layer. It was etched to facilitate adhesion of the polymer. The polymer layer was spin coated. Its surface was plasma etched to facilitate adhesion of the second level of metal. Etching created cavities on the surface of the sensitive layer. 2.2. Characteristics of BCB
Static measurements were carried out in a climatic chamber with a HP 4284A Precision LCRmeter. Capacitances of the sensors were measured at a frequency of 10 kHz at an input level of 1 V. Under these conditions, the measurement accuracy of the LCRmeter is 0.035 pF for 100 pF. The climatic chamber delivers humidity with an accuracy of 1% RH. BCB films performances were compared to three other materials used to design humidity sensors: polyimide, etched polyimide and silicon oxide [2,6,7]. Fig. 2 shows the capacitance increase versus RH level at 30 8C. Tested devices were (i) interdigitated electrode capacitors with BCB, SiO2, polyimide and etched polyimide, and (ii) parallel plate electrode capacitors with BCB. Capacitance varies linearly versus humidity for polyimide and BCB films. An increase in sensitivity appears after 70% RH for etched polyimide and for SiO2. This is probably due to condensation and cluster formation, since SiO2 is a quite porous material, and since etching created cavities on the surface of the etched polyimide film. Etched polyimide is more sensitive than SiO2 at high humidity levels. SiO2 and etched polyimide are not suitable materials for the considered use. The sensitivity of the interdigitated electrode device with polyimide is greater than the sensitivity of the same device with BCB. Experiments show that the sensitivity of the interdigitated/polyimide device is about 10 times the sensitivity of the interdigitated/BCB device. For BCB, the parallel electrode device is more sensitive than the interdigitated device (Table 2), because it provides higher values of capacitances, about 100 pF instead of 45 pF for the interdigitated devices. It is the reason why the parallel electrode device was chosen instead of the interdigitated electrode device although technologically less easy to fabricate.
BCB is known to have a low moisture uptake, less than 0.1 wt.% [3,5] when a step of 60% RH is applied. It does not bind water molecules strongly since it has neither hydroxyl nor carboxyl groups. BCB was chosen to provide short response time sensors. The main features of the BCB films are gathered in Table 1. They are to be compared to the polyimide films [6] (Ultradel 7501 from Amocco) which were employed to fabricate capacitive humidity sensors for the same use. The polyimide sensors were made of interdigitated electrodes covered by a sensitive layer. Their performances are reported elsewhere [2,7,8]. Table 1 Main features of the polyimide and the BCB films
Permittivity Moisture uptake by weight at 50% RH (%) Thickness (mm) Surface area in contact with humidity (mm2)
Polyimide
BCB
2.80 3 2 1
2.65 0.1 1.6 0.05 4.5
Fig. 2. Capacitance variations versus relative humidity level.
A. Te´ telin et al. / Sensors and Actuators B 91 (2003) 211–218 Table 2 Sensitivities of the devices with different structures and materials Electrode/film
Sensitivity (pF/RH%)
Interdigitated/polyimide Interdigitated/BCB Parallel/BCB
0.0246 0.0007 0.0023 0.0007 0.1 0.0007
2.4. Response time Dynamic measurements were carried out with an apparatus reproducing breathing out (Fig. 3). The apparatus expels humid air in a pulse shape. The humidity level, the temperature and the flow of the expelled air are chosen by the operator. The humidity level in the cylinder is checked by a dew point hygrometer (Fig. 3). The accuracy of the humidity level is 1% RH. The accuracy of the flow is 0.05 dm3 s1. The accuracy of the temperature is 1 8C. The experiments were carried out on seven sensors fabricated on the same wafer. Three measurements were done and averaged for each of them. Sensors were enclosed in an oscillator circuit to measure capacitance variations through a period measurement. The output of the oscillator is sampled by a computer. Every 10 periods, the average of the last 10 periods is worked out by the computer. The capacitance of the sensor is worked out from this value. The quantization error and the averaging error together are slightly less than 5% of the final mass uptake value. It corresponds to a 0.05 pF error in the capacitance. The overall accuracy on capacitance dynamic measurements is 0.07 pF. The results presented in Table 3 correspond to the response of the sensors when humidity step variations of 30% RH were applied to them. Table 3 shows minimum and
213
Table 3 Adsorption and desorption times for polyimide and BCB in the interdigitated devices in response to humidity steps of 30% RH
Polyimide (ambient) BCB (ambient) BCB (40 8C)
Adsorption time (s)
Desorption time (s)
Minimum
Typical
Minimum
Typical
– 0.4 0.3
1.0 0.65 0.5
– 2.7 2.0
15 5.3 4.5
typical adsorption and desorption times of water in BCB at ambient temperature and at 40 8C. The results are to be compared with polyimide performances. The ‘‘adsorption time’’ is the response time of the sensor when a humidity step with positive amplitude is applied to it. The ‘‘desorption time’’ is the time the sensor takes to recover its initial capacitance value, that is its equilibrium value at the considered ambient conditions. The ‘‘adsorption time’’ is defined as the rising time from 10 to 90% of the final mass uptake of water. Adsorption times are smaller than desorption times because they represent response times to steps. Shorter response times were obtained with the BCB films because the low moisture uptake of BCB is linked to a quite large diffusion coefficient, larger than the diffusion coefficient of polyimide [4–7]. The response time was even shorter when the sensors were heated (Table 3), since diffusion coefficient increases with temperature [4]. Thanks to low moisture uptake, BCB films provide shorter response times than polyimide films. In exchange, they are less sensitive. The lack in sensitivity was compensated in two ways: (i) by the parallel plate electrode structure, and (ii) by a quite large surface area. Another advantage of BCB towards polyimide is its better long-term stability [9].
Fig. 3. Experimental set-up for dynamic measurements.
A. Te´ telin et al. / Sensors and Actuators B 91 (2003) 211–218
214
As BCB takes less water than polyimide, it undergoes less swelling and stress. The sensitivity of the parallel electrode/BCB capacitor is 0.1 pF/RH%. If the capacitance change of the sensor versus the humidity level is modeled by the straight line calculated by regression, the error made when working out the humidity level from a capacitance measurement is less than 2% of the humidity level. Its response time is 500 ms at 40 8C. This sensor meets the requirements for the considered medical use.
3. Sorption model for glassy polymers The conventional sorption model used for glassy polymers is the Dual-Mode model. It relates the concentration of water in the polymer to the partial pressure of water in the gaseous phase. As the sensors are made to give a reflect of the humidity level through a variation of capacitance, the models used to relate capacitance to water concentration are presented first. 3.1. Parallel plate capacitor model The capacitance C of the unequal parallel plate capacitor device was modeled using C¼
erwb e0 A þ Cf l
(1)
They are expressed in SI units. f and g are functions of N, erwb and erbb. They are expressed as follows: 3erwb g¼ (3) 2erwb þ erbb f ¼
2N erwb erbb 3erbb e0 2erwb þ erbb
(4)
As erwb is deduced from capacitance measurements, the number density of water molecules in the parallel plate capacitor can be worked out using Eq. (2). The volume of the polymer film is known and the volume expansion due to water uptake is neglected. Then, the number density of water molecules in the film is easy to relate to the mass and to the sorbed vapor concentration. 3.3. Sorption model for glassy polymers at equilibrium The sorption of gases in polymers is well-documented and it is the subject of current developments in the literature [4,11–14]. The sorption model established for conventional glassy polymers is the Dual-Mode model. The Dual-Mode model is the combination of Henry’s law and Langmuir-type hole-filling model. Henry’s law expresses ideal dissolution of the penetrant in the polymer when the equilibrium is reached between fluid phase and adsorbed phase. It is a linear relation between vapor concentration and partial pressure written as G ¼ k D pi
(5)
where e0 is the permittivity of vacuum, A the area of the smaller plate, l the thickness of the film, erwb the relative permittivity of the matrix water/BCB and Cf the fringing field correction. It was neglected for the calculations presented in this paper. erwb for the parallel plate/BCB device was calculated using Eq. (1) and capacitance measurements. There is no simple formula to calculate erwb from C for the interdigitated electrode device.
where G is the total sorbed vapor concentration at equilibrium, kD the Henry’s law solubility coefficient, and pi the vapor partial pressure above the surface of the film. The Langmuir-type hole-filling contribution is attributed to sorption in cavities associated with the glassy state. It is expressed by:
3.2. Sorbed vapor concentration
where G is the sorbed vapor concentration, CH 0 reflects the capacity of the polymer for the penetrant, b reflects the affinity of the polymer for the penetrant, and pi the vapor partial pressure. b has units of inverse pressure. The Dual-Mode model isotherm equation is the combination of Eqs. (5) and (6)
The concentration of water in the sensors is worked out from the value of erwb by using Onsager’s model [10]. The model relates molecular polarizability and number density to permittivity for condensed phases with permanent dipoles. It is suitable since water molecules are polar and condensable. It was applied to water molecules in BCB instead of vacuum Ng m2 aþ e0 ðerwb erbb Þ ¼ (2) 1 af 3½ð1 af ÞkT where erbb and erwb are, respectively, the relative permittivity of BCB alone and the relative permittivity of the matrix water/BCB. N is the number density of water molecules in BCB, a the molecular polarizability of water, including electronic, ionic and orientational polarizabilities, m the dipole moment, k the Boltzmann constant, T the temperature.
G ¼ CH 0
bpi bpi þ 1
G ¼ k D pi þ C H 0
(6)
bpi bpi þ 1
(7)
G is expressed in cm3(STP)/cm3 (poly) (standard conditions for temperature and pressure), it is the ratio of the equivalent volume of gas of the sorbed vapor in standard conditions of temperature and pressure and the volume of polymer. At constant temperature, RH is directly proportional to water partial pressure pi RHð%Þ ¼ 100 (8) psat where psat is the saturated vapor pressure.
A. Te´ telin et al. / Sensors and Actuators B 91 (2003) 211–218
3.4. Kinetics of sorption In the Dual-Mode model, the penetrant is distributed in two types of sorption phenomena. The first one corresponds to free diffusion (Henry’s law) and the second one corresponds to immobilization in specific sites (Langmuir mode). The solution of the general diffusion law (or Fick’s second law of diffusion) (Eq. (9)) is given by Eq. (10) [4]
215
is not the sum of the solution of Eq. (9) and the solution of the transfer mode. But, the solution is nevertheless equivalent to the combination of Fickian diffusion and a first-order relaxation term [15].
4. Results and discussion 4.1. Static results
@G @2G ¼D 2 @t @x
(9)
1 Mt 8X 1 Dðð2n þ 1Þ2 p2 tÞ ¼1 2 exp 2 p n¼0 ð2n þ 1Þ l2 M1
!
(10) where Mt is the mass sorbed from the beginning of the experiment until time t, M1 the equilibrium mass uptake of the experiment, l the thickness of the film, and D the diffusion coefficient. The Langmuir kinetic theory is based on the following equilibrium [11,14]: kads
sorbent þ sorbate @ sorbent=sorbate kdes
where kads and kdes are, respectively, the kinetic constants of adsorption and desorption of the dynamic equilibrium. They depend on the energy of the sites available for adsorption. The kinetics of this equilibrium are described by: dy ¼ kads pi ð1 yÞ kdes y dt
(11)
where y(t) is the fractional coverage, yðtÞ ¼ NðtÞ=N , N(t) is the concentration of the adsorbed molecules on the solid surface at time t, and N is the concentration of total available adsorption sites. y(t) is proportional to Mt. The solution of Eq. (11) is given by h t i yðtÞ ¼ A 1 exp (12) t
The adsorption isotherms of water in BCB films were derived from experiments with the parallel plate capacitor using Eqs. (1) and (2) at three different temperatures (Fig. 4). Sorption isotherms for water vapor in BCB are linear functions of partial pressure at considered temperatures and pressures. Consequently, separate estimation for kD, CH 0 and b was not possible. To model adsorption isotherms, the following approximation of Eq. (7) was used: G ¼ k pi
(14)
where k is a constant depending on kD, CH 0 and b. To explain the shape of the adsorption isotherms, two hypotheses were put forward and discussed to find the right one. The first hypothesis was bpi ! 1 under the conditions of the experiments. In this case k would be the infinite dilution solubility, expressed by k ¼ kD þ CH 0 b
(15)
The second hypothesis lied in the fact that Langmuir adsorption could be negligible. In that case k would be the Henry solubility coefficient. The second hypothesis seemed to be the more plausible since the low moisture uptake behavior of BCB implies weak water/BCB interactions and then ideal dissolution. But etching created cavities on the surface of the polymer, so a Langmuir adsorption mode may exist. To determine k and then to determine which hypothesis is the more plausible, the variations of k with temperature
where t is a constant depending on kads and kdes. kads and kdes depend on adsorption energy level of the sites. A is y(1) A¼
bpi 1 þ bpi
(13)
The model describing the kinetics corresponding to the Dual-Mode model is called the diffusion-reaction model [15]. The transfer between the two modes of the Dual-Mode model is assumed to be a first-order reversible reaction: k1
diffusing species @ immobilized species k2
This equilibrium leads to a relaxation term alike Eq. (12). Of course, the solution of the combination of Eq. (9) and the transfer between the two modes of the Dual-Mode model
Fig. 4. Water sorption isotherms of water in BCB at 23, 30 and 40 8C.
A. Te´ telin et al. / Sensors and Actuators B 91 (2003) 211–218
216
Table 4 Infinite dilution solubility of water in BCB at different temperatures Temperature (8C)
k (cm3(STP)/cm3 (poly) atm)
23 30 40
93 5 51 5 18 5
Fig. 6. Adsorption kinetics: fractional mass uptake of the sensor versus time, in response to a step from 33 to 67% RH at 40 8C.
Fig. 5. Van’t Hoff plot of k for water sorption in BCB.
were studied. The results are gathered in Table 4. They are in agreement with low moisture uptake in BCB: kD for water in methyl methacrylate (MMA) is 710 cm3(STP)/(cm3 (poly) atm) and kD for water in vinyl acetate (VAc) is 860 cm3(STP)/(cm3 (poly) atm) [12]. Van’t Hoff type equations describe the dependence on temperature of kD, b and CH 0 [13]. We have noticed that the variation of k versus temperature obeys a Van’t Hoff-type equation (Eq. (16)) since the plot of ln(k ) as a function of 1/T is linear with correlation r 2 ¼ 0:998 (Fig. 5) a k ¼ k0 exp (16) T where k0 and a are constants. The Van’t Hoff-type dependence of k on temperature tend to support the hypothesis that k is equal to Henry’s law constant. The second hypothesis is the more plausible. k is modeled by: 9036:9
12 k ¼ 5 10 exp (17) T
measurements. The curve is characterized by a sharp increase followed by a rather slow approach to equilibrium. The adsorption kinetic curves were fitted quite well by Eqs. (18) and (19) with a correlation coefficient r 2 0:980 (Fig. 7). The Fickian diffusion equation alone (Eq. (10)) does not fit well. t t Mt ¼ P1 1 exp þ ð1 P1Þ 1 exp P2 P3 M1 (18) 1 Mt t 8X 1 ¼ 1 P10 exp 0 ð1 P10 Þ 2 P2 p n¼0 ð2n þ 1Þ2 M1 ! P30 ðð2n þ 1Þ2 p2 tÞ exp (19) l2 where P1, P2, P3, P10 , P20 and P30 are fitted parameters. Eqs. (18) and (19) are almost equivalent under the experimental conditions (Fig. 7). Eq. (18) is the sum of two first-order relaxation terms and Eq. (19) is the sum of Fickian diffusion and a relaxation term.
It corresponds to a difference in the molar enthalpy of water between the gaseous phase and the sorbed phase close to 18 kcal mol1. 4.2. Dynamic results Fig. 6 shows the kinetics of water adsorption in BCB. Humidity steps from 33 to 67% RH were applied to the sensors at 40 8C. The flow was 1 dm3 s1. Mass uptake in the sensors was derived from Eq. (2) and real-time capacitance
Fig. 7. Adsorption kinetics: experiments and fit.
A. Te´ telin et al. / Sensors and Actuators B 91 (2003) 211–218
Dynamics cannot be interpreted according to Fickian diffusion alone. Four hypotheses can be set out to explain the shape of the dynamic response. The first one is that adsorption is more significant than diffusion at the times considered, diffusion occurs at longer times. It is then suggested that there should be two energetically different kinds of adsorption sites for water on BCB, since the fit (Eq. (18)) is the sum of two terms of the form of Langmuir kinetics (Eq. (12)) [14]. The second hypothesis is that diffusion is as significant as adsorption, that is to say they occur at the same time. In this case, Eq. (19) may be the result of the combination of Fickian diffusion (Eq. (10)) and a first-order relaxation term (Eq. (12)) due to the transfer between the two modes of the Dual-Mode model [15] (when immobilized molecules become diffusing molecules and vice-versa). But if we consider that the two modes of the Dual-Mode model are completely independent phenomena (the part of the water molecules which take part in a Langmuir adsorption equilibrium only take part in this equilibrium and will never diffuse), Eq. (19) may be the result of the combination of Fickian diffusion (Eq. (10)) and Langmuir adsorption (Eq. (12)). In this case, a part P10 of the water molecules take part in a surface adsorption equilibrium, and a part ð1 P10 Þ of the molecules diffuse independently. The third hypothesis lies in the fact that the first-order term in Eq. (19) corresponds to structural relaxation in the polymer [12,15]. Generally, relaxation is a long-time phenomenon at ambient temperature, but the uptake of water may induce stress on the thin film, and then quick rearrangements and flexibility in the polymer chains. The last hypothesis is that the relaxation term in Eq. (19) results from the fact that the diffusion coefficient, D, was considered constant although it could depend on concentration. The correlation of the fit of Eq. (10) with experiments is enhanced if D is a linear function versus time (i.e., D increases when concentration increases), but the fit is not so good as those of Eqs. (18) and (19). The fit of Eq. (19) provides a diffusion coefficient equal to 4.08 mm2 s1. This value is close to the value given by Pranjoto and Denton [5] for BCB: 4.5 mm2 s1. The values of the other fitting parameters of Eq. (19) are P10 ¼ 65:3% and P20 ¼ 0:26 s. To know which of the hypotheses is correct, an investigation has to be made on the dependence of the fitting parameters on the humidity level, on the temperature and on the flow applied to the sensor. For the modeling of the sensor, as Eqs. (18) and (19) are equivalent, we may choose Eq. (18), because it has a simpler form than Eq. (19).
5. Conclusion The parallel plate/BCB sensor is suitable for real-time breathing out measurements thanks to sufficiently short response times. Short response times are provided by the low moisture uptake of BCB in exchange of a lack in sensitivity.
217
To compensate this, the parallel plate electrode structure was chosen instead of an interdigitated electrode one. The sensitivity of the sensor is 0.1 pF/RH% and its adsorption time is about 650 ms at ambient temperature and about 500 ms at 40 8C. Its desorption time is less than 6 s. Its error towards linearity is less than 2% RH. Water sorption mechanisms in the device were investigated as a starting point to model the sensor. Sorption equilibrium concentration versus vapor partial pressure was described by a linear function whose slope was less than 100 cm3(STP)/(cm3 (poly) atm) at ambient temperature. It is in accordance with the theoretical low moisture uptake of BCB films. It seems that Henry dissolution is the predominant phenomenon to describe the sorption equilibrium of water in BCB. The dependence of the slope on temperature was modeled by a Van’t Hoff-type equation. Hence, an equilibrium model for water sorption in the sensor was put forward. The kinetics of water sorption were adequately modeled by the sum of two first-order terms. The study of the dependence of the parameters of the first-order terms on the humidity level, on the temperature and on the flow will lead to a complete model of the sensor.
Acknowledgements The authors would like to thank Prof. C. Coulon of the Centre de Recherche Paul Pascal for his sensible advice. They would like to thank N. Fabre and V. Conedera of the Laboratoire d’Analyse et d’Architecture des Syste`mes for the fabrication of the sensors. They are also grateful to C. Hainaut and J.L. Lachaud for their technical help to carry out measurements. References [1] A.F.P. Van Putten, et al., Multisensor microsystem for pulmonary function diagnostics for COPD and asthma patients, in: Proceedings of the First Annual International IEEE-EMBS Special Topic Conference on Microtechnology and Medical Biology, Lyon, France, October 12–14, 2000. [2] C. Laville, C. Pellet, Comparison of three humidity sensors for a pulmonary function diagnostic microsystem, IEEE Sens. J. 2 (2002) 96–101. [3] The Dow Chemical Company, 1995–2002. http://www.dow.com/ cyclotene/index.html. [4] W.R. Vieth, Diffusion in and Through Polymers, Carl Hanser, Munich, 1991. [5] H. Pranjoto, D.D. Denton, Moisture uptake of bisbenzocyclobutene (BCB) films for electronic packaging applications, Mater. Res. Soc. Symp. Proc. 203 (1991) 295–302. [6] H. Kazuyuki, Photosensitive Polyimides—Fundamentals and Applications, Technomic Publishing Cie, 1995. [7] C. Laville, J.Y. Dele´ tage, C. Pellet, Humidity sensors for a pulmonary function diagnostic microsystem, Sens. Actuators B 76 (2001) 304–309. [8] C. Laville, C. Pellet, Interdigitated humidity sensors for a portable clinical microsystem, in: Proceedings of the First Annual International IEEE-EMBS Special Topic Conference on Microtechnology and Medical Biology, Lyon, France, October 12–14, 2000.
218
A. Te´ telin et al. / Sensors and Actuators B 91 (2003) 211–218
[9] R. Buchhold, A. Nakladal, G. Gerlach, K. Sahre, K.J. Eichhorn, M. Herold, G. Gauglitz, Influence of moisture uptake on mechanical properties of polymers used in microelectronics, Mater. Res. Soc. Symp. Proc. 511 (1998) 359–364. [10] R. Coelho, Physics of Dielectrics, Elsevier, Amsterdam, 1979. [11] D.M. Ruthven, Principles of Adsorption and Adsorption Processes, Wiley/Interscience, New York, 1984. [12] S.P. Russell, D.H. Weinkauf, Vapor sorption in plasma polymerized vinyl acetate and methyl methacrylate thin films, Polymer 42 (2001) 2827–2836. [13] V. Detallante, D. Langevin, C. Chappey, M. Me´ tayer, R. Mercier, M. Pine´ ri, Water vapor sorption in naphthalenic sulfonated polyimide membranes, J. Membr. Sci. 190 (2001) 227–241. [14] H. Hu, M. Trejo, M.E. Nicho, J.M. Saniger, A. Garcia-Valenzuela, Adsorption kinetics of optochemical NH3 gas sensing with semiconductor polyaniline films, Sens. Actuators B 82 (2002) 14–23. [15] Y.-M. Sun, Sorption/desorption properties of water vapour in poly(2-hydroxyethyl methacrylate). 2. Two-stage sorption models, Polymer 37 (1996) 3921–3928.
Biographies A. Te´ telin received the engineering degree from the Ecole Nationale Supe´ rieure d’Electronique, d’Informatique et de Radiocommunications de
Bordeaux (ENSEIRB) in 2000. She is currently a PhD student at the IXL Microelectronics Laboratory. Her research interests include modeling of chemical sensors and microsystem simulation. C. Pellet was a researcher at the ‘‘Institut d’Electronique Fondamentale’’ of the University Paris XI-Orsay, from 1982 to 1993, where he studied the deposition of thin film by ion beam sputtering. He joined the IXL Microelectronics Laboratory as a full professor in 1993. His works are now focusing on microtechnology and on the development of microsystems. He is involved in different operations: evaluation of assembly technology, reliability of microsystems, microsystems for medical applications. C. Laville was graduated in electrical engineering from the University of Bordeaux in 1996. She received the PhD degree in electronics from the same University in 2001 for her work about the fabrication of capacitive humidity sensors. She has joined Thale`s Communications since September 2001. G. N’Kaoua works for the National Center for Scientific Research since 1984 as an electronics engineer. He joined the IXL Microelectronics Laboratory in 1993. He is responsible for the clean room. He studies thin film sputtering depositions and laser photolithography. He also works on multi-chip modules.
