Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Abstract
The paper summarizes a detailed design study of silicon piezoresistive pH sensors, whose operation relies on the pH sensitive swelling of a
polymer hydrogel. It is based on an analytical model of the hyperelastic–viscous behaviour of hydrogel. The pH sensor behaviour was investigated
by simulating and measuring the silicon membrane deflection and the stress inside the membrane. Simulation was performed using finite element
method (ANSYS). A technology to manufacture such hydrogel-based sensors was developed. We found satisfying agreement between simulation
and measured results for the sensor output characteristics.
© 2005 Elsevier B.V. All rights reserved.
0925-4005/$ – see front matter © 2005 Elsevier B.V. All rights reserved.
doi:10.1016/j.snb.2005.10.041
18 Q. Thong Trinh et al. / Sensors and Actuators B 117 (2006) 17–26
Fig. 1. General operational principle of hydrogel-based piezoresistive pH sensors: V volume change, σ silicon membrane stress, R/R relative resistance change,
Uout output voltage.
materials, such as metals, ceramics and plastics, which are dry often adequate for predicting its mechanical response. For hyper-
and hard. To describe the polymer gels’ elastic behaviour, free elasticity, mechanical properties are characterized by a strain
swelling experiments were performed to determine gel swelling energy function, which is a scalar function of either a strain or
and deswelling, respectively, in the pH range of 1–12. The mea- a deformation tensor [8]. The strain energy describes the work
sured hydrogel strain is a function of pH value and varying from stored in the body:
some percentages at low pH levels to 34% at pH 11 [7]. The
results of this investigation showed that hydogel strains are large W = W(λ1 , λ2 , λ3 ), (1)
and recoverable. The hydrogel strain versus pH value relation
where λ1 ,λ2 ,λ3 are principal stretch ratios, λi = Li /L0i (i = 1, 2,
was used for simulating the sensor properties.
3), where Li and L0i are actual and initial lengths, respectively.
Furthermore, the stress–strain characteristic of gels resulting
The energy function is determined by the thermodynamics
from a uniaxial extension test was determined by applying step-
theory as follows [8]:
wise increasing forces and by detecting the related strains, up to
a strain of 250%. The gel samples were prepared by deposition of 1
PVA/PAA solution. Then the dry film was peeled-off and cut into W= NkT (λ21 + λ22 + λ23 − 3) (2)
2
strips. Since the experiment is not intended to fail the specimen,
there is no need to use dogbone- or dumbbell-shaped specimen. where N is the number of chains per unit volume in the network,
The strip-shaped dry gel samples were isothermally annealed in k the Boltzmann constant and T is the absolute temperature. The
an oven at 130 ◦ C for 20 min before being soaked in deionized quantity NkT is equivalent to the shear modulus of the hydorgel
water and then preserved in pH 1 solution. The undeformed gel material.
sample geometry was set to a ratio between length and width of In case of large strains, the thermodynamic equation of state is
10:1 in order to ensure uniaxial stress conditions. The relation- less satisfactory and a phenomenological approach is used. This
ship between stresses (force per unit unstrained cross-section) treatment does not consider the gel’s molecular structure but
and strains (percentage variation in the sample length) is shown uses the observed macroscopic behaviour of the system [9,10].
in Fig. 2. It presents the typical non-linear behaviour of rubber- Using the theory of continuum mechanics, the strain energy
like materials [8]. function for an isotropic material is in the following form [11]:
W = C1 (I1 − 3) + C2 (I2 − 3)
2.2. Material model of polymer hydrogel
= C1 (λ21 + λ22 + λ23 − 3) + C2 (λ21 λ22 + λ21 λ23 + λ22 λ23 − 3)
The development of an appropriate material model of poly- (3)
mer hydrogel materials has been an active research area for many
years. Very recently, a number of constitutive models have been where I1 and I2 are the first and second strain invariants. C1 and
developed addressing different experimental observations. Tak- C2 are the material constants having dimensions of MPa.
ing the large volume change in account, hyperelastic models are If the material is assumed incompressible, I2 can be rewritten
as:
λ1 = λ and λ2 = λ3 = λ−1/2 .
3. Sensor model and finite element simulation Fig. 4. Finite element model of the pH sensor: (1) silicon chip, (2) hydrogel
layer, (3) silicon rigid mesh and (4) channel for solution entering.
3.1. pH sensor design
swelling acts only on the silicon diaphragm. This makes the gel
In this work, a novel pH sensor design was used (Fig. 3). A
deformations similar to the uniaxial extension test.
thin layer of PVA–PAA hydrogel is placed in the cavity of a
The Cauchy stress following Eq. (5) developed due to gel
piezoresistive transducer (silicon chip) between the upper sil-
swelling is similar to the pressure acting on the plate of silicon
icon diaphragm and a lower silicon rigid mesh as shown in
piezoresistive pressure sensor chips leading to their bend and
Fig. 3. The solution itself enters the gel through the mesh chan-
causing the stresses inside. The output voltage can be determined
nels without means of a pump like in our previous designs
[4–6].
Since kinetics of gel swelling is a function of both gel dimen-
sion and diffusion coefficient, the response time is proportional
to the square of its dimension [14]. Therefore, in order to
decrease the reaction time constant, the gel dimension was cho-
sen to be smaller than that of the silicon diaphragm. Because of
the gel is fixed in the hard square frame, the force exerted by gel
Fig. 3. pH sensor design: (1) silicon chip, (2) piezoresistors, (3) silicon mem-
brane, (4) hydrogel layer and (5) silicon rigid mesh. Fig. 5. Process scheme for fabricating silicon rigid mesh.
20 Q. Thong Trinh et al. / Sensors and Actuators B 117 (2006) 17–26
First, the sensor diaphragm deflection was measured using Fig. 10. Finite element simulation: (a) membrane deformation and (b) contour
nanofocus microscan method (NMS). For this measurement, pH plot of membrane stress.
22 Q. Thong Trinh et al. / Sensors and Actuators B 117 (2006) 17–26
solutions were poured into and drawn out of the sensor succes- presented in Fig. 9 shows that the mechanical behaviour of
sively. PVA–PAA gels was corresponding to hyperelastic material prop-
For the measurement of the sensor output voltage, which is erties. Hence, the physical values of the constants C1 and C2
the main parameter of such piezoresistive pH sensors, a supply in Eqs. (3)–(5) were automatically determined. For the fully
voltage of 5 V is applied to the sensor. The output voltage is contracted gel (preserved in pH 1 solution), the calculated
displayed on a multimeter and the measurement data is acquired parameters are C1 = 0.15 MPa and C2 = 0.06 MPa. The mate-
by computer. These measurements have been described in details rial parameters C1 and C2 are related to the shear modulus,
in Ref. [16]. G:
G = 2(C1 + C2 )
5. Results and discussion
In our work, G gets the value of 0.42 MPa, which is in relatively
5.1. Material parameter determination good agreement with reported results in Ref. [12]. Calculations
performed with these values for C1 and C2 delivered the same
The results of uniaxial extensional test were used to verify results. The comparison with Marra et al. [12] shows that our
the material model characterizing the mechanical behaviour of gels are a slightly softer because the hydrogels here were not pro-
PVA–PAA hydrogels. Using available ANSYS codes (TBFT cessed in the same way regarding polymerization temperature
command), a fitting program was performed. The fitting result and time.
Fig. 11. Results of finite element analysis: (a) membrane deflection profile and (b) membrane longitudinal and transverse stress profile.
Q. Thong Trinh et al. / Sensors and Actuators B 117 (2006) 17–26 23
5.2. Simulation results gitudinal stress gradient is higher than the transverse one. The
maximum stresses, corresponding to the hydrogel’s maximum
Determination of deformation and stress distribution inside strain, are two orders of magnitude less than the maximum ten-
the bending plate are main objectives for the static structural sile strength of silicon (σ yield = 7 GPa) [17]. Thus, it gives proof
analysis by FEM. Fig. 10 shows the contour plots for membrane that the pH sensor can operate safely in the pH range 1–11. Based
deformation and stress. The maximum strain case amounts to on the membrane stress calcualtions, the sensor output voltage
34%, which is given for pH 11. The in-plane stresses in the two can be predicted using Eq. (6).
main directions within the silicon plate were found to be exactly
equal to each other due to the symmetry involved in the problem. 5.3. Measurement results
Concretized calculations are presented by profiles shown in
Fig. 11. The maximal deflection was calculated as 15.1 m at Nanofocus microscan measurements were used to determine
the membrane center (Fig. 10a). Fig. 10b shows the membrane’s membrane deformations profiles (Fig. 12). These results demon-
longitudinal and transverse stresses. It can be seen that the lon- strate that the sensor, namely the silicon membrane, exhibits
Fig. 12. Nanofocus microscan measurement: (a) 3D image of membrane bending and (b) membrane deflection profile.
24 Q. Thong Trinh et al. / Sensors and Actuators B 117 (2006) 17–26
good response due to PVA–PAA gel swelling. For the case of sociation constant pKa of PAA. The sensitivity value of this novel
pH 11, the measured value was about 14.3 m. pH sensor is twice of those in our previous studies [4–6].
Due to the plate bending and hence stresses developed in it, The sensor response time is illustrated in Fig. 15a for pH 5.5
the changes in the sensor output voltage in accordance with pH and 11 solutions (measurements started at pH 1). It is probable
were detected. The measured long-term behaviour of the sen- that the much larger amount of hydroxyl groups at high pH
sor output voltages in pH range from 1 to 12 and vice versa values leads to the stronger and faster ionization process and
is shown in Fig. 13. In this measurement, the pH 1 solution results in the faster response time at pH 11.
was used as reference. The sensor output voltage shows a repro-
ducible response to pH value changes. Increase and following
decrease of pH causes a slight hysteresis. With increasing pH,
the ionization degree is going up and hence the output voltage
increases. However, as the pH further increases, the gel swelling
tends to decrease due to the osmotic pressure difference of free
ions inside and outside decreases. As a result, when the pH value
exceeds 11, the effect of ionic strength surpassed the effect of
the electrostatic repulsion between the polycarboxylate groups
making the gel deswelling and consequently leading to the lower
output voltage at pH 12 [13].
6. Conclusions
Acknowledgements
References
[1] I.S. Han, M.-H. Han, J.W. Kim, S. Lew, Y.J. Lee, F. Horkay, J.J. Magda,
Fig. 15. Time dependent sensor output voltage: (a) measurement and (b) finite
Constant-volume hydrogel osmometer: a new device concept for minia-
element calculation.
ture biosensors, Biomacromolecules 3 (2002) 1271–1275.
[2] S. Herber, W. Olthuis, P. Bergveld, A swelling hydrogel-based PCO2
Because of the rate-limiting of diffusion processes, the sensor, Sens. Actuators B 91 (2003) 378–382.
[3] S. Herber, W. Olthuis, P. Bergveld, A. van den Berg, Exploitation of
mechanical readjustments of gel depend on the pH range. As a pH-sensitive hydrogel disk for CO2 detection, Sens. Actuators B 103
a result, the response time at pH 5.5 is longer than at pH (2004) 284–289.
11. [4] G. Gerlach, M. Guenther, G. Suchaneck, J. Sorber, K.-F. Arndt, A.
The calculated results are compared with the measured data Richter, Application of sensitive hydrogels in chemical and pH sensors,
in Fig. 15b. The deformation time history measured by free Macromol. Symp. 210 (2004) 403–410.
[5] M. Guenther, J. Sorber, G. Suchaneck, G. Gerlach, T.Q. Thong, K.-F.
swelling experiments represents the input of finite element Arndt, A. Richter, Piezoresistive chemical sensors based on hydrogel,
analysis and the output voltage needed to realize this deforma- Technisches Messen 72 (2005) 93–102.
tion history is output of the calculation. It can be seen that for [6] G. Gerlach, M. Guenther, G. Suchaneck, J. Sorber, K.-F. Arndt, A.
both cases the measured response time is evaluated about four Richter, Chemical and pH sensors based on the swelling behavior of
times longer than the calculated one. Possibly, this difference is hydrogels, Sens. Actuators B 112 (2005) 555–561.
[7] Q. Thong Trinh, J. Jorber, G. Gerlach, Modeling silicon piezoresistive
caused by the fact that the time constants used for calculations pH sensors behaviour under swelling of polymer hydrogel, in: Proceed-
were determined using free-swelling experiments. It should ings of 5.GMM/ITG/GI—Multi-Nature Systems Workshop, Dresden,
be reminded that in these experiments the gel was immersed 2005, pp. 41–45.
alone into the solution. Hence, both the diffusion process and [8] L.R.G. Treloar, The Physics of Rubber Elasticity, third ed., Clarendon
the response could take place more quickly. In addition, the gel Press, Oxford, 1975.
[9] K.S. Anseth, C.N. Bowman, L. Brannon-Pepper, Mechanical proper-
extension was assumed happening instantaneously. However, ties of hydrogels and their experimental determination, Biomaterials 17
when placed in the silicon chip cavity, the penetration of (1996) 1647–1657.
solution ions into the polymer network is much more difficult, [10] I.M. Ward, D.W. Hadley, An Introduction to the Mechanical Properties
therefore resulting in the longer response time. Due to the of Solid Polymers, Wiley, New York, 1993.
reasons mentioned in Section 3.1, the sensor response time can [11] P. Kohnke (Ed.), ANSYS Theory Reference, Rel. 5.7, ANSYS Inc.,
Cannonsburg, 1999.
be improved by using thinner hydrogel layers. However, it must [12] S.P. Marra, K.T. Ramesh, A.S. Douglas, The actuation of a biomimetic
correlate to the silicon membrane thickness for sufficiently good poly (vinyl alcohol)–poly (acrylic acid) gel, Philos. Trans. R. Soc. Lond.
sensitivity. A 360 (2002) 175–198.
26 Q. Thong Trinh et al. / Sensors and Actuators B 117 (2006) 17–26
[13] K.-F. Arndt, A. Richter, S. Ludwig, J. Zimmermann, J. Kressler, D. worked in research and development in the field of sensors and measuring
Kuckling, H.-J. Adler, Poly(vinyl alcohol)/poly(acrylic acid) hydrogel: devices and sensors at several companies. In 1993, he became a profes-
FT-IR spectroscopic characterization of crosslinking reaction and work sor at the Department of Electrical Engineering at Dresden University of
at transition point, Acta Polym. 50 (1999) 383–390. Technology. Since 1996 he has been the head of the Institute for Solid-State
[14] P. Chiarelli, D. De Rossi, Determination of mechanical parameters Electronics. His research is focused on sensor and semiconductor technology,
related to the kinetics of swelling in an activated contractile gel, Prog. simulation and modelling of micromechanical devices and the development
Colloid Polym. Sci. 78 (1988) 4–8. of solid-state sensors, especially pyroelectric infrared sensors and piezore-
[15] G. Gerlach, W. Doetzel, Grundlagen der Mikrosytemtechnik, Carl sistive humidity and gas sensors. From 1994 to 2000 Prof. Gerlach served
Hanser Verlag, Munich, Wien, 1997. as Deputy Dean and Dean of the Electrical Engineering Department at TU
[16] Q. Thong Trinh, J. Jorber, G. Gerlach, Design study of silicon piezore- Dresden, respectively. Since 2000 he has been vice president of the German
sistive pH sensors based on polymer hydrogel, in: Proceedings of Inter- Society for Measurement and Automation Technology (GMA) and since 2002
national Sensor 2005 Conference, vol. 2, Nuremberg, 2005, pp. 133– member of the Presidency of the VDE (German Association of Engineers in
138. Electrical Engineering, Electronics, Information Technology).
[17] Marc J. Madou, Fundamentals of Microfabrication, first ed., CRC Press,
Joerg Sorber received his PhD degree in electrical engineering from Dres-
Boca Raton, Florida, 1997.
den University of Technology, Germany, in 1983. From 1978 to 1995 he
worked in the field of advanced driving systems at the Institute for Preci-
Biographies sion Engineering at TUD. Since 1996 he has been Senior Scientist at the
Institute for Solid-State Electronics at TUD. His current research interests is
Quang Thong Trinh received the BS degree in physics from Hanoi Uni- focused on the design of sensors and sensor systems including the simulation
versity in 1984 and the MSc degree in materials science from International of their components as well as of complex systems, applications of sensitive
Trainning Institute for Materials Science (ITIMS) in 1999. He is currently hydrogels in chemical and pH sensors.
working towards the PhD degree in electronics engineering at the Institute
Karl-Friedrich Arndt studied physics at Dresden University of Technology.
for Solid-State Electronics, Dresden University of Technology. His current
In 1975, he got the PhD and in 1982 the Dr.rer.nat.habil degree at Technische
interests include technology development for applications of piezoresistive
Hocschule Merseburg. In 1990, he became full professor of physical chem-
sensors, design development by simulation and modelling of MEMS.
istry of polymers at TU Dresden. The main focus of his work is synthesis
Gerald Gerlach received MSc and PhD degrees in electrical engineering and characterization of soluble polymers and polymer networks with sensor
from Dresden University of Technology in 1983 and 1987, respectively. He and actuator properties.