Capacitive transducer for in-mold monitoring of injection molding.

INTRODUCTION

The measurement and/or monitoring of polymer melt status inside an injection mold, particularly the melt-front velocity and melt-front position during the filling stage, is important to the understanding and mold design of injection molding. Many efforts have been made in this area. Most measurements of the material status inside injection molds have been limited to cavity pressure (1), mold surface temperature (2), mold temperature (2), and mold heat flux (2). Limited progress has been made with respect to the monitoring of melt-front position. Yokoi et al. (3-6) were perhaps the first to succeed in observing the melt flow in a mold by flow visualization through glass inserts fitted on the mold wall. This technique was modified by Bress and Dowling (7) for observing the surfaces of large parts. This kind of visual technique, using glass inserts, allows direct observations of the melt flow status, typically using a high-speed camera. Such a device provides good laboratory information but little information for industrial applications, as glass cannot withstand the high pressure required in the applications. Furthermore, the flow and heat transfer along the glass surfaces may be different from those along the mold metal surfaces.

An ultrasonic pulse-echo technique was developed by Wang et al. (8) for a simple box mold, based on the attenuation difference between the air and polymer melt, to detect the flow front of the melt inside the mold. Multiple sensors of this type were installed at selected locations inside the mold for the detection of flow positions. When the melt front reaches any probe location, it could be reflected by sudden changes in the pulse-echo magnitude. However, this method can measure the melt-front only at the locations where the ultrasonic sensors were installed, giving limited and discrete information of mold filling. In fact, this type of detection of flow position is no different from the detection methods involving temperature and pressure transducers by Gao et al. (1, 2) and optical fiber sensors by Thomas et al. (9, 10).

To eliminate the limitations of the existing technologies, a capacitive transducer is proposed and developed in this project, based on the dielectric property difference between polymer and air. In addition to the fact that the proposed sensor can online and continuously measure the melt-front position in the mold cavity, other potentially useful information on the molding status can be effectively monitored, as demonstrated in the following.

PRINCIPLES OF CAPACITIVE TRANSDUCER

Capacitance is a useful physical phenomenon for sensor design. It has been successfully applied for measurements of distance, area, volume, force, humidity, etc. (11-17), in the industry. Taking a parallel-plate capacitor, for example, the capacitance of the capacitor, [C.sub.0], is:

[C.sub.0] = [[epsilon] * A]/d (1)

where [epsilon] is the dielectric constant of the medium between the plates of area A separated by a distance d. Varying any of the above constituent parameters results in a linear change in the capacitor output, capacitance. Note that the dielectric constant used in this paper refers only to the real part of the complex number.

Based on this principle, we propose to develop a capacitive transducer for the measurements of the melt-front position and melt-front velocity. If two metal plates can be separately installed on the two mold halves and isolated from each other, they can form the two electrodes of a capacitor. The medium of the capacitor is air before the mold filling, and it is gradually replaced by polymer melt during filling, until the cavity is completely occupied by the polymer melt at the end of filling. The dielectric constant of the medium between the capacitor plates, therefore, varies continuously through the filling stage, resulting in changes in the capacitor's capacitance. Figure 1 schematically illustrates the melt flow during the filling stage. Neglecting the temperature effect, the capacitor output varies with the composite medium of air and melt polymer during filling as:

C = [C.sub.0](1 - afl/L + k * afl/L) (2)

where [C.sub.0] is the capacitance before the start of filling, k the relative dielectric constant of the filled polymer, afl the average melt flow length, and L the total length of the plate. The relative melt flow length, afl/L, can be predicted by the measured capacitance as:

afl/L = (C/[C.sub.0] - 1)/(k - 1) (3)

The relative dielectric constant, k, is the ratio of the capacitance of a capacitor made with the polymer to the capacitance of the same capacitor of air. The dielectric constants of many polymers can be found in the literature (18, 19). Alternatively, it can be measured at the end of filling, when afl/L = 1, and

k = [C.sub.end]/[C.sub.0] (4)

[FIGURE 1 OMITTED]

This is possible and convenient, as injection molding is a cyclic, repeatable process.

A preliminary test is first conducted for the feasibility study of constructing a capacitance sensor for measuring melt-front position. A capacitor is formed using two aluminum plates, each 159 mm X 15 mm, as its two electrodes, separated by 4 mm. An ABS plate 4 mm in thickness is placed between the two electrodes with varying length to simulate a varying medium of injection molding filling. Figure 2 shows the influence of the increasing polymer length on the capacitance, measured by a capacitance meter with a resolution of 0.1 pf. The data suggest that the capacitance output has high linearity with the polymer length, showing the good feasibility of such a sensor for melt flow and position detection in injection molding.

It is necessary to point out that dielectric constant is material-dependent and temperature-dependent. For a nonpolar polymer, the temperature effect on the dielectric constant is mainly due to the effect of the density change associated with the temperature (18, 19). Therefore, the temperature effect on the dielectric constant, and consequently the capacitance, of a nonpolar material are small, except around the melting temperature, [T.sub.m]. This can be clearly verified by another test with HDPE. A capacitor was formed with the two aluminum plates as its electrodes and HDPE as the medium. After it was heated to 200[degrees]C in an oven, the capacitor was taken out and cooled gradually in the atmosphere for its capacitance and temperature measurements. The capacitance and temperature relation is shown in Fig. 3. A distinct change in capacitance can be clearly seen around the melt temperature of HDPE (about 130[degrees]C). Except for the points around the melt temperature, the temperature effect on the capacitance is relatively small. The nature of capacitance changes around the melt temperature may be explored for solidification study in the cooling stage. During the filling stage, most of the melt has a temperature greater than the melt temperature, so the influence of temperature on the capacitance output is small, and a near linear relation between the capacitance measurement and the melt position can be assumed. For a polar polymer, the dielectric constants hardly change with its molding temperature (19), which is typically much higher than its glass transition temperature.

[FIGURE 2 OMITTED]

[FIGURE 3 OMITTED]

MOLD AND IN-MOLD CAPACITIVE TRANSDUCER DESIGN

Mold Design

A set of molds with different mold geometries will be employed to demonstrate the applications of the proposed sensor. To reduce the installation complexity of fitting a capacitance sensor to each of these molds, a common mold base is developed with a fitted capacitive transducer. The mold geometry changes can be obtained by replacing the mold insert. As shown in Fig. 4, the shape of the mold can be changed by replacing the 150 mm X 100 mm rectangular with any of the shapes shown in Fig. 5. (The dimensional unit in the figures is millimeters.) All four of these mold inserts have a 2-mm mold thickness, sharing the same fan gate system and the capacitor developed in the following. These inserts are referred as Mold 1 to Mold 4.

In-Mold Capacitor Sensor Design

Two plates are needed to form a parallel plate capacitor. In our design, only one plate is specifically designed to reduce the mold complexity as shown in Fig. 6; a 136 mm X 35 mm stainless steel strip, insulated electrically by aluminum oxide ceramic from the rest of the mold base, is installed to the fix mold half. The moving mold half, made from the same stainless steel, forms as the opposite electrode, without any special modification. Theory shows that for such a system with a large area relative to the gap, fringe fields can usually be neglected (17). The moving platen is grounded to further reduce the fringe flux and disturbances. With the coolant flow-rate controlled at a constant, the cooling design and the coolant flow have little effect on the measurement. Figure 6 is a schematic illustration of the electrode. The thickness and area of the insulating layer is minimized to reduce its influences on the molding. The insulation layer is made by a specially selected aluminum oxide ceramic of thermal conductivity (24 W/m[degrees]C) comparable to that of the steel (14 W/m[degrees]C) around the room temperature.

[FIGURE 4 OMITTED]

[FIGURE 5 OMITTED]

[FIGURE 6 OMITTED]

Capacitance Measurement System and Its Calibration

The capacitance output is converted to a electrical voltage signal by an in-house-built circuit based on the charge-and-discharge principle (20-23). The charge-and-discharge frequency is set at 800 Hz. The signal is conditioned by amplification and filtering, before it is read by a data acquisition system consisting of two National Instruments cards (AT-MIO-16X and AT-DIO-32F). The AT-MIO-16X card provides the Digital-to-Analog Conversion (DAC) and Analog-to-Digital Conversion (ADC), and the AT-DIO-32F card performs the digital input/output communication for the process. An additional software filter system is implemented via programming to remove high-frequency noises. The data is acquired at a sampling rate of 5 ms, and the measurement lag is minimized by a properly selected cut-off frequency of 200 Hz.

Four capacitors with capacitances of 220, 273, 333 and 395 pf, respectively, were used to find the voltage-capacitance relation of the measurement system. The CT outputs recorded by the computer acquisition system were 0.4485V, 2.8473V, 5.5444V, and 8.3140V, respectively. Good linearity was found to correlate the capacitance to the voltage output in the following relation:

V = -9.430 + 0.0449 * C (5)

where C is the capacitance in a unit of picofarad and V is the CT output in voltage.

EXPERIMENTAL

All experiments were conducted on a Chen Hsong reciprocating screw injection molding machine, MK-III J88, fitted with a Moog MPC2000 controller for the machine sequencing and barrel temperature control. The data acquisition and computer control system used for this project was developed in-house. It provides a real-time platform for monitoring and controlling all key machine and process conditions. Semicrystalline and amorphous materials used for the experiments include high-density polyethylene (HDPE) (Marlex, HMN6060), polypropylene (PP) (HMC, Profax6331), and polystyrene (PS) (Dow, 685D). All experiments are conducted with well-controlled injection speed profile, packing pressure, and mold and barrel temperatures, with mold geometry given in the last section. The melt flow indexes of the materials are 6.5 g/10 min, 14 g/10 min, and 1.6 g/10 min, respectively. The details on the injection velocity and packing pressure conditions will be given in the Results section; the barrel temperatures were set at 230[degrees]C for HDPE, 240[degrees]C for PP, and 250[degrees]C for PS. The mold temperature for all cases was controlled with a coolant temperature of 50[degrees]C. For purposes of brevity, only results obtained with HDPE are presented, unless otherwise specified.

RESULTS AND DISCUSSION

Detection of Filling Start

During injection, polymer melt passes through the nozzle, sprue, runner, and gate before finally entering the mold cavity. The time at which the melt first enters the cavity is important for the monitoring and control of the process. The developed capacitive transducer (CT) can detect this starting point easily. From the description in the previous sections, it is clear that the CT output is proportional to the polymer melt length along the capacitor electrode installed in the mold. The CT output, therefore, does not change until after the melt enters the capacitor gap. This information can be used for the detection of the start of filling for the cavity. The CT output during filling of Mold 1, a rectangular cavity, at a constant screw injection velocity of 15 mm/s, is shown in Fig. 7. It clearly shows that the CT output starts to increase after a period of about 1.25 s, corresponding to the filling of the sprue, runner, and gate, before entering the cavity. The starting time of CT output deviating from zero is clearly the starting time of mold filling, as the capacitor electrode is placed right at the end of the fan gate.

Detection of V-P Transfer Point

The injection process switches from the filling stage to the packing-holding stage when the mold cavity is filled. Correspondingly, the control of the process should be changed from velocity control in the filling stage to pressure control at this critical time, known as the V-P transfer point. A too-early transfer may result in an incomplete filling of the mold (short shot), while a too-late transfer may result in overfill (flash). Accurate and timely detection of a proper V-P transfer time is critical to the successful operation of the process. Gao and Zhao (24) have developed a technology to detect this switch-point by looking for rapid increase of nozzle pressure at the end of filling.

The proposed capacitive transducer can be used for such detection as well. It is understood that the measured capacitance should increase with the melt development, and eventually it becomes a near constant after the melt reaches the end of the electrode, if the temperature influence is neglected. When the electrode is designed to cover the last-to-fill point of the mold. V-P transfer can be easily detected by observing the CT output changes. In addition to the detection of filling start. V-P transfer can also be observed in Fig. 7, which plots together the measurements of the nozzle pressure and the CT output during filling of Mold 1 at a constant injection velocity. Minor over-fill has been purposely set with an injection stroke of 49 mm. It clearly shows that CT output starts to flatten at the time of rapid nozzle pressure increase, corresponding to the material compression at the end of complete mold filling. This suggests that the CT output can be used for the proper detection of the V-P transfer, if the capacitor electrode is designed to cover the last point to fill of the cavity.

[FIGURE 7 OMITTED]

Melt Flow Position and Velocity Measurement During Injection

The primary design objective for the capacitive transducer is to measure the melt-front position (velocity) during filling. The flow length is used to represent the distance that the melt-front has traveled in the mold from the gate. As many points are needed to characterize the melt-front, an average melt-flow length is used to describe this distance. The average flow length could be difficult to calculate for a complex mold with multigates. This project is limited to relatively simple molds, as the objective of this project to demonstrate the feasibility of measuring flow-position using a capacitive sensor. The following may be used to describe the relationships among injection velocity (IV), average-flow-length (afl), and melt-front-area ([A.sub.mf]) during mold filling

IV * [A.sub.b] = [[dafl]/[dt]] * [A.sub.mf] (6)

where [A.sub.b] is the barrel area, a machine constant.

Filling with a constant screw injection velocity of Mold 1, a rectangular flat mold, results in near constant melt-front-area throughout the filling except the beginning and the end of the cavity due to the flow development. The average-flow-length (afl) during filling under a constant injection velocity (IV) should, therefore, increase nearly linearly with time according to Eq 6, as shown in Fig. 7. The CT outputs, representing the average-flow-length or flow position, increase linearly during mold filling. To further demonstrate that the capacitive transducer can quantitatively measure the in-mold flow position, a series of experiments were conducted using different velocity profiles, mold inserts and materials in the following.

Tests With Different Velocity Profiles

Experiments were conducted using HDPE and Mold 1 with different constant injection velocities of 10, 15, 20, 25 and 30 mm/s, respectively. The CT outputs for these cases are given in Fig. 8, illustrating that different injection velocity results in different time of melt entering the cavity. The CT outputs of all cases appear as straight lines for filling with a constant injection velocity of the mold with the constant cross-sectional area. Least square regression was performed to analyze the experimental data for the relation between the CT output and injection velocity. The CT output after the start of filling of cavity is fitted with ramp functions with results shown in Table 1. The correlation coefficients (R-value), a measure of how well the data are fitted by the regressing function, are very close to 1, indicating a high linearity of the CT output. The regression slope of each ramp function is given in the third column of the table. The ratios of the slopes over injection velocities are given in the last column. The near same value of the last column suggests that Eq 6 fits well with the experimental data. The melt-front velocity in the mold can be easily obtained by differentiating the CT output with time.

[FIGURE 8 OMITTED]

Experimental tests with step change profiles were conducted next for the capacitive transducer. Four different step change profiles, including both the step-up and step-down types, were tested, with a step change from 10 to 30 mm/s, from 15 to 25 mm/s, from 25 to 15 mm/s, and from 30 to 10 mm/s. All the step changes were introduced around the time the cavity was half filled. The resulting CT outputs are shown in Fig. 9. Unlike the cases with constant injection velocities, obvious inflection can be found in all the CT output curves as a result of the step changes. Furthermore, the step-up and step-down changes in injection velocity result in opposing inflection shapes--concave for the step-up, and convex for the step-down. The inflection magnitude is clearly associated with the size of the step change. The CT outputs of injection velocity of 10-30 mm/s step-up and 30-10 mm/s step-down show a more pronounced inflection than those of the other two step changes of smaller step sizes. 15-25 mm/s step-up and 25-15 mm/s step-down. Quantitative analysis can be conducted with the step change experimental data. In the 10-30 mm/s step-up profile, for example, the curve can be divided into two separate parts by the inflection point marked in the figure. Each part can be well fitted as a ramp function with fitting data shown in Table 2. The injection velocity is also calculated and listed in the second column. The slope over velocity ratio is given in the last column of the table, with a near same value before and after the inflection (or the step-change). This suggests that the in-mold flow velocity changes caused by the step change in injection velocity can be well measured by the developed capacitive transducer.

[FIGURE 9 OMITTED]

In addition to the constant and step change profiles, the ramp injection velocity profiles were also tested with CT outputs shown in Fig. 10. Inspection of the figures indicates that the CT outputs are parabolic and can be well fitted by parabolic curves. The ramp-up and ramp-down result in opposing curvature directions of the two CT output curves. Differentiating the CT outputs can clearly give a ramp function, a linear increase of melt flow velocity, as expected for the filling of a rectangular cavity with a ramp screw velocity.

Tests With Different Mold Inserts

The capacitive transducer was tested with molds of varying cross-sectional areas. Four different mold inserts shown in Fig. 5 were used with HDPE at a constant injection velocity of 20 mm/s. The CT outputs are given in Fig. 11. As indicated earlier, all inserts share the same sprue, runner and gate system, resulting in the same start of filling at the same constant velocity, as shown in the figure. The melt flow development in different molds, however, is dependent on the shape of the insert, as clearly reflected by the CT outputs. The fillings of different inserts not only end at different times, but also result in different melt flow-rate histories, corresponding to the melt-front-areas resulting from different insert shapes.

[FIGURE 10 OMITTED]

Tests With Different Materials

In addition to HDPE used in the previous experiments, other materials, PP and PS, were also tested for the capacitive transducer developed. Figure 12 compares the CT outputs of different materials with Mold 1 at a constant injection velocity of 20 mm/s. The CT outputs of HDPE and PP are similar, lower in both the output and its increasing rate than those of PS. This can be explained by the following equation, derived from Eqs 2 and 5:

[FIGURE 11 OMITTED]

[FIGURE 12 OMITTED]

Slope = [[dV]/[dt]] = 0.0449 X [[dC]/[dt]] = 0.0449 * (k - 1) * [[C.sub.0]/L] * [[dafl]/[dt]] (7)

where [C.sub.0] is the capacitance before the start of filling, k the relative dielectric constant of the filled polymer, afl the average-flow-length, and L the total length of the plate. In these cases, [C.sub.0], L, and [dafl]/[dt] are the same, the differences lay in k. As shown by Blythe (18), the dielectric constants of HDPE and PP are close to 2.3 and lower than that of PS, 2.5.

The apparent small difference in the time reaching the gate for different materials under the same velocity and mold delivery system may be caused by differences in drooling during plasticizing of the different materials and the difference in the compressibility of the materials. The material compressibility difference leads to different injection strokes, under the same molding conditions. The total screw displacement during filling is 48.1 cm, 47.1 cm, and 45.6 cm for HDPE, PP, and PS, respectively.

The above experiments show that the capacitive transducer developed can not only qualitatively show the flow development in mold filling, but can also quantitatively measure the melt-front velocity and position. In a typical molding, the polymer solidification during filling is not significant. A thin-wall molding is typically accompanied by high injection speed; the solidification in this case will not be significant, the melt-front-velocity can thus also be reasonably measured by the proposed sensor. In an exceptional case, for example, a thin-wall molding coupled with slow injection and high cooling rate, significant solidification can result in errors in the measurements of melt-front position or velocity, as dielectric property of polymer solid and melt can be different. For the cases of complex mold geometry, the placement of the capacitive transducer needs to be adjusted for the flow position measurement. For example, multiple capacitive sensors can be installed in the mold to monitor the flow situations with a mold with different thickness.

Packing Pressure Setting

After injection, the process switches to packing-holding stage, in which additional material is "packed" under pressure control into the mold to compensate for part shrinkage associated with the solidification. Proper setting of the packing pressure is important for producing a quality part. One of the difficulties of injection molding is the lack of a convenient means of monitoring the material status during this stage. The capacitive transducer developed may be used to provide such a means.

Experiments were conducted here to study how the CT outputs could be affected by packing pressure and packing time. A number of constant packing pressures were set for the experiments with different packing durations. Because of space limitation, only CT outputs of three pressure levels, 400, 500, and 600 bar, and with a fixed 7 second packing for material HDPE, are given in Fig. 13. In all cases, the filling of the cavity was conducted under the same condition, resulting in the over-lapping CT outputs before packing. The CT output after the filling, however, is pressure dependent, which can be clearly seen from the figure. The CT output generally increases with the pressure level, as a denser part is molded under a higher packing pressure, except that CT output of 600 bar packing pressure is significantly lower than the others during packing-holding. The exception is due to mold separation caused by over-packing at 600 bar. Over-packing at this pressure results in a slight opening of the mold, consequently a larger electrode separation distance and a smaller capacitance. This was confirmed by the observation of part flash at the 600 bar packing pressure. Parts molded at other pressures have no flash.

The CT signals of this experiment can have the following two applications for the packing-holding: 1) the end of packing CT signal for possible part weight monitoring under normal operation without mold flash, and 2) a lower CT at a higher packing pressure may indicate a mold flash.

[FIGURE 13 OMITTED]

[FIGURE 14 OMITTED]

Detection of Gate Freezing-Off Time

Proper detection of gate freezing-off is important to injection molding packing setting. Too short packing may result in back flow, while excessively long packing will waste energy as the material status in the mold can no longer be influenced by packing after the gate is frozen. Packing should terminate shortly after gate is frozen. The gate freezing-off time may be detected by the capacitive transducer developed. As also shown in Fig. 13, the packing starts at about 2.6s, and the CT output curves level off after about 8.6s, indicating that the CT outputs reach the peak after 6s of packing, though the packing-holding lasted for seven seconds. After the gate is frozen, melt can no longer be packed into mold, leaving the mold with a constant amount of material, and the polymer solidification will start to create voids between the electrodes, resulting in a gradual reduction of CT output. Thus, the peak value of CT should correspond well with the gate frozen time during normal operations. This is in good agreement with the gate-frozen-time results obtained by the off-line packing time detection, which examines the weight variations with different packing times shown in Fig. 14. Little additional material can be packed into the mold after 6 seconds of packing. The exception is again with the 600 bar packing case, which resulted in flash as stated previously.

CONCLUSIONS

A capacitive transducer has been successfully developed for continuous and real-time measurements of in mold flow front position and velocity. In addition, the sensor may also be used to detect start of mold filling, velocity-to-packing transfer point, and gate freezing-off point. Over-packing and molded part weight may also be correlated with the sensor outputs. Note that the signal change in the post-filling phases is repeatable, despite the fact that the changes over the post-filling phases are relatively small. Better signal conditioning circuit is under development for better detection of capacitive sensor output changes, and for a correlation study of the capacitive signal with other post-filling phases properties such as the cooling rate and/or crystallization rate of the injection molded parts.

Table 1. Comparison of Different Constant Velocity Profiles.

   Injection                       Slope/
Velocity (mm/s)  R-Value  Slope   Velocity

      10         0.9999   0.3836  0.0384
      15         0.9998   0.5780  0.0385
      20         0.9998   0.7679  0.0384
      25         0.9998   0.9643  0.0386
      30         0.9998   1.1531  0.0384

Table 2. Data Analysis for Step-Up Profile.

                   Averaged Injection  Regressed  Slope/
                   Velocity (mm/s)     Slope      Velocity

Before inflection       10.01          0.395      0.0395
After inflection        29.69          1.164      0.0392

ACKNOWLEDGMENT

We thank Moog Japan Ltd. for financial support and for making the mold for this project.

REFERENCES

1. F. Gao, W. I. Patterson, and M. R. Kamal, Polym. Eng. Sci., 36, 1272 (1996).

2. F. Gao, W. I. Patterson, and M. R. Kamal, Intern. Polymer Processing, VIII. 147 (1993).

3. H. Yokoi, T. Hayashi, K. Tado, and N. Morikita, SPE ANTEC Tech. Papers, 34, 329 (1988).

4. H. Yokoi, S. Kamata, and T. Kanematsu, SPE ANTEC Tech, Papers, 37, 358 (1991).

5. H. Yokoi, S. Kamata, and T. Kanematsu, SPE ANTEC Tech, Papers, 37, 367 (1991).

6. H. Yokoi and Y. Inagaki, SPE ANTEC Tech, Papers, 38, 457 (1992).

7. T. J. Bress and D. R. Dowling, J. Reinforced Plastics & Composites, 17, 1374 (1998).

8. H. Wang, B. Cao, C. K. Jen, K.T. Nguyen, and M. Viens, Polym. Eng. Sci., 37, 363 (1997).

9. C. L. Thomas, A. O. Adebo, and A. J. Bur, SPE ANTEC Tech, Papers, 40, 2236 (1994).

10. A. J. Bur and C. L. Thomas, Polym. Eng. Sci., 37, 1430 (1997).

11. R. Puers, Sensors & Actuators A, 37. 93 (1993).

12. W. C. Heerens, J. Phys. E: Sci. Instrum., 19, 897 (1986).

13. H. Kobayashi and Y. Hosokawa, Sensors & Actuators A. 24, 27 (1990).

14. R. K. Pearson, IEEE Trans. Instrum, & Meas., 39, 421 (1990).

15. D. D. Denton, C. N. Ho, and S. He, IEEE Trans. Instrum & Meas., 39, 502 (1990).

16. J. Fraden, AIP Handbook of Modern Sensors. Physics, Designs & Applications, American Institute of Physics, New York (1993).

17. L. K. Baxter, Capacitive Sensors Design & Applications, IEEE Press (1997).

18. A. R. Blythe, Electrical Properties of Polymers, Cambridge University Press (1979).

19. T. Osswald and G. Menges, Material Science of Polymers for Engineers, Hanser Publishers (1996).

20. D. A. Seanor, Electrical Properties of Polymer. Academic Press (1982).

21. J. Davidse, Analog Electronic Circuit Design. Prentice-Hall (1991).

22. J. Sigdell, IEEE Trans. Instrum, & Meas., 21, 60 (1972).

23. J. C. Lotters, W. Olthuis, P. H. Veltink, and P. Bergveld, IEEE Trans. Instrum & Meas., 48. 89 (1999).

24. F. Gao and C. H. Zhao, U.S. Patent 6.309.571 B2 (2001).

XI CHEN ([dagger]), GUOHUA CHEN, and FURONG GAO*

Department of Chemical Engineering

Hong Kong University of Science & Technology

Clear Water Bay, Kowloon

Hong Kong

([dagger]) Dr. Chen is currently with Institute of Systems Engineering. Zhejiang University. Hangzhou, China.

*To whom correspondence should be addressed. E-mail: [e-mail]