Surface channel based Coriolis flow sensors with integrated piezoelectric transducers

Measuring flow rate is critical for many applications. Coriolis flow measurement has drawn much attention thanks to its ability to measure fluid mass flow regardless of the properties of the fluid. A Coriolis flow sensor fabricated with MEMS technology can measure fluid flow as low as 1 g/h. Our group has developed a method called surface channel technology that allows fabrication of suspended channels on the surface of the wafer, which can operate as a micro Coriolis flow sensor. A state of the art micro Coriolis flow sensor uses Lorentz force actuation or electrostatic actuation. The displacement of the suspended channel is detected with a capacitive readout structure or an optical method. Piezoelectric actuation and detection can be an option for actuation and sensing because piezoelectric transducers require low current and moderate voltage for actuation. Furthermore, piezoelectric transducers are relatively robust and easy to integrate.

PZT was chosen as the material for the piezoelectric transducers used in this thesis because of its ferroelectricity and high piezoelectric coefficient. The polarization direction of PZT can be switched by applying an electric field. Puled laser deposition (PLD) was used to deposit PZT thin films. The PZT active layer was deposited on top of silicon nitride thin films or surface channels to form unimorph actuators. Parallel plate electrodes and interdigitated electrodes were used to apply an electric field to the thin film.

We designed, fabricated and tested a micro Coriolis mass flow sensor driven by piezoelectric actuators. The sensor consists of a rectangular loop of the microfluidic channel, two integrated piezoelectric actuators and a capacitive readout structure. The integrated actuators can actuate the sensor into both swing mode and twist mode. Both actuation modes were evaluated for different fluids. In the case of swing mode actuation, a laser Doppler vibrometer was used to detect the resulting Coriolis movement, resulting in a relatively low sensitivity. In the case of twist mode actuation, a capacitive readout was used and, in comparison to previously published devices, a high mass flow sensitivity was achieved thanks to an optimized readout design. The measurement results show that the read-out noise has a standard deviation of 0.15% of the full scale of 1.8 g/h when actuated in the twist mode. The measured phase shift sensitivity is different for water and nitrogen because of the difference in density. However, the time delay calculated from the phase shift is almost independent of the fluid density.

In comparison with Lorentz force actuation, piezoelectric actuation is more flexible in exciting different vibration modes. We took this advantage and designed a device in which the eigenfrequencies of the actuation mode and Coriolis mode are very close together. The amplitude of the Coriolis mode is then enhanced by resonance. We designed and fabricated a micro Coriolis mass flow sensor with an S-shaped suspended channel. The measurement results show that the phase shift sensitivity is higher with twist mode actuation than with swing mode actuation. The measured phase shift sensitivity with swing mode actuation is higher than the corresponding results in Chapter 3, thanks to the resonance effect of the Coriolis mode. The measured phase shift sensitivity with twist mode actuation is slightly lower than the corresponding results in Chapter 2 because a less efficient capacitive readout structure was used.
A theoretical model was proposed to calculate the influence of fluid density on the measured phase shift values. The measurement results shows that the model can compensate the influence of fluid density. The measurement results show that the read-out noise has a standard deviation of 0.4% of the full scale of 0.8 g/h when actuated in the twist mode.

Piezoelectric transducers can be used to both actuate the actuation mode and detect the Coriolis mode. Chapter 6 presents a micro Coriolis flow sensor with piezoelectric transducers for both actuation and readout. The sensor consists of a rectangular loop of the microfluidic channel, six integrated piezoelectric transducers for actuation and two piezoelectric transducers for readout. The measurement results show that the output signal of a piezoelectric readout is dependent on both the rotation and the displacement of the moving end of the transducer. The measured sensor response to fluid flow shows that the phase shift sensitivity of this device does not depend much on which mode is used for actuation. For both actuation modes the phase shift sensitivity is a lot smaller than the twist mode actuation results presented in Chapter 4 and 5. The measurement results show that the read-out noise at zero flow has a standard deviation of 1.1% of the full scale of 1.6 g/h when actuated in the twist mode. The read-out noise at zero flow has a standard deviation of 3.1% of the full scale when actuated in the swing mode.

A large vibration amplitude is desired for a Coriolis flow sensor as the Coriolis forces due to mass flow and, accordingly, the signal-to-noise ratio, are directly proportional to the vibration amplitude. Therefore, it is important to maximize the quality factor Q so that a large vibration amplitude can be achieved without requiring high actuation voltages and high power consumption. We performed an investigation of the Q factor of different devices in different resonant modes. Q factors were measured both at atmospheric pressure and in vacuum. The measurement results are compared with theoretical predictions. In the atmospheric environment, the Q factor increases when the resonance frequency increases. When reducing the pressure from 1 bar to 0.1 bar, the Q factor almost doubles. At even lower pressures, the Q factor is inversely proportional to the pressure until intrinsic effects start to dominate, resulting in a maximum Q factor of approximately 7200.