1. Introduction
Micro pressure sensors that are important components can be applied in biomedical and various industries. For instance, a micro capacitive pressure sensor, presented by Eggers et al. [
1], was utilized in the biomedical measurement. Huang et al. [
2] fabricated a piezoelectric pressure sensor to monitor the instant pressure variation of the molding injection. Berns et al. [
3] developed a micro pressure sensor array to measure the wall pressure in turbulent flows. A silica fiber optic pressure sensor, proposed by Pulliam et al. [
4], was applied in high-temperature propulsive environments.
Recently, microelectromechanical system (MEMS) technology has been applied to fabricate various micro devices. The advantages of micro pressure sensors manufactured by MEMS technology are small size, high performance and easy mass-production. Several micro pressure sensors [
5-
10] have been fabricated using MEMS technology. For instance, Lin et al. [
5] employed a surface micromachining process to manufacture a piezoresistive pressure sensor, in which the polysilicon diaphragms were deposited by LPCVD. The pressure sensor had a sensitivity of 0.15 mV/psi. A piezoresistive pressure sensor, proposed by Wu et al. [
6], was made using a bulk micromachining process. The piezoresistors were the single-crystalline 3C-SiC film grown by APCVD (atmospheric pressure chemical vapor deposition), and the diaphragm was released using an anisotropic etching of KOH solution. The sensitivity of the pressure sensor was 177.6 μV/psi at room temperature. Dai et al. [
7] presented a capacitive pressure sensor fabricated by the CMOS process, in which the membranes were released using an anisotropic dry etching and a wet etching, and the etch holes were sealed by PECVD (plasma enhanced chemical vapor deposition) nitride. Sippola et al. [
8] reported a ceramic capacitive pressure sensor produced by a thick film screen-printing technique. The pressure sensor comprised a top electrode deposited on a ceramic diaphragm and a bottom electrode deposited on an alumina substrate, and the ceramic cavity and diaphragm were created using a thick film sacrificial layer. The pressure sensor had a sensitivity of 9.2 fF/psi. An FET pressure sensor, reported by Svensson et al. [
9], was fabricated by a surface micromachining process. The pressure sensor was an MOS transistor where the membrane was the device gate, and the two silicon diffusions under the membrane were the source and the drain regions. The sensitivity of the FET pressure sensor was 0.1 mA/bar. Hynes et al. [
10] manufactured an FET pressure sensor using a surface micromachining process, in which a sacrificial oxide layer and a polysilicon diaphragm were deposited on the pressure sensing area, and then HF was used to etch the sacrificial oxide from beneath the polysilicon diaphragm. The cavity was sealed with LPCVD oxide. The FET pressure sensor had a sensitivity of 1.3 μA/psi.
Micro pressure sensors, proposed by Lin et al. [
5], Wu et al. [
6], Sippola et al. [
8], Svensson et al. [
9] and Hynes et al. [
10] without integration with circuits on a chip, had the disadvantages of high packaging cost and large chip area. The aim of this work is to manufacture an FET pressure sensor with circuits on a chip. The advantages of the integrated FET pressure sensor are low packaging cost, small chip area, and high sensitivity. The CMOS-MEMS [
11-
13] technique is the use of commercial CMOS process to fabricate MEMS devices. The benefit of micro pressure sensors manufactured by the CMOS-MEMS technique is the capability of integration with readout circuits as a system on chip (SOC). In this work, we employ the CMOS-MEMS technique to fabricate an FET pressure sensor integrated with readout circuits on a chip. The fabrication of the pressure sensor in this work is easier than that of Svensson et al. [
9] and Hynes et al. [
10]. The FET pressure sensor is constructed by 16 sensing cells in parallel, and each sensing cell consists of a suspended membrane and an NMOS. When applying a pressure to the sensing cells, the pressure sensor generates a change in current. The readout circuits are employed to convert the current variation of the pressure sensor into the voltage output. In order to release the membranes, the pressure sensor needs a post-CMOS process. The post-process is the use of wet etching to remove the sacrificial layers, and to obtain the suspended membrane. A LPCVD parylene is utilized to seal the etch holes in the pressure sensor. Experimental results reveal that the sensitivity of the FET pressure sensor is 0.032 mV/kPa in the pressure range of 0-500 kPa.
2. Structure of Pressure Sensor
Figure 1 shows the layout of the FET pressure sensor integrated with readout circuits on a chip. The pressure sensor consists of 16 sensing cells in parallel. All sensing cells that are circular shape have the same dimensions.
Figure 2 illustrates the schematic cross-sectional view of a sensing cell, which is composed of a suspended membrane and an NMOS. The suspended membrane is the gate of the NMOS, and the two silicon diffusions under the membrane are the source and drain regions. The diameter of each sensing cell is 100 μm. The membrane is a sandwiched structure that contains silicon dioxide, metal and silicon oxide layers, in which the metal layer is the electrode and each layer has a thickness of about 1 μm. The gap between the membrane and the substrate is about 0.65 μm. As shown in
Fig. 2, the dielectric of the NMOS consists of a 1 μm oxide layer, a 0.65 μm air gap and a 1 μm oxide layer. The structure can be taken as a series of three capacitors and the total capacitance pre unit area
Ct can be expressed as,
where
Cgap and
Cox represent the individual capacitance per unit area of the air gap and oxide, respectively. The membrane generates a deformation when applying a uniform pressure to the sensing cell, leading to the total capacitance of the NMOS changes. In the saturation region, the drain current,
Ids, of the NMOS can be expressed as [
10],
where
L represents the channel length of the NMOS,
W is the channel width of the NMOS,
Ct is the total capacitance of the NMOS,
μn is the mobility of the electrons,
Vt is the threshold voltage and
Vgs is the gate-to-source voltage. According to
Eq. (2), the drain current,
Ids, changes as the capacitance,
Ct, varies. Therefore, the FET pressure sensor produces a change in current upon applying a pressure to the membranes of the sensing cells.
The finite element method software, CoventorWare, is utilized to simulate the behaviors of the FET pressure sensor. The model of only one sensing cell is constructed because all sensing cells are the same. The triangular elements are employed to mesh the model. The materials of the membrane are aluminum (Young's modulus of 70 GPa, Poisson's ratio of 0.3 and mass density of 2679 kg/m
3) and silicon dioxide (Young's modulus of 69 GPa, Poisson's ratio of 0.17 and mass density of 2200 kg/m
3) [
10]. The boundary condition is that the membrane edge is fixed, and the load is a uniform pressure applied to the membrane. The stress and displacement of the membrane can be computed through the simulation.
Figure 3 presents the stress distribution of the membrane at a pressure of 500 kPa. The simulated result shows that the maximum stress situates at the edge of the circular membrane.
Figure 4 shows the displacement of the membrane in the sensing cell at a pressure of 500 kPa. The maximum displacement is located at the center of the membrane. The relation between the pressure and displacement at the center of the membrane is shown in
Fig. 5. The result depicts that the displacement at the center of the membrane is about 0.44 μm at a pressure of 500 kPa. The transverse displacement of a clamped circular thin plate can be expressed as [
7,
15],
where
p represents uniform distributed load,
D is the flexural rigidity of the plate and
a is the radius of the plate. Substituting P=500 kPa, a=50 μm, h=2.65 μm and r=0 into
Eq. (3), we obtain that the displacement at the center of the circular plate is about 0.41 μm, and the value approximates to the simulated result of 0.44 μm.
The professional circuit simulation software, HSPICE, is utilized to simulate the current-to-voltage characteristics of the FET pressure sensor and the output voltage of the readout circuits.
Figure 6 presents the simulated results of
Ids/
Vds output characteristics for the FET pressure sensor. As shown in
Fig. 6, the current of
Ids depended on the voltages of
Vgs and
Vds, and the current of
Ids is about 1.5 mA at
Vgs=3 V,
Vds>1.5 V.
Figure 7 illustrates the readout circuits with an operational amplifier that are used to convert the current variation of the FET pressure sensor into the voltage output. As shown in
Fig. 7, R1,
R2,
R3,
R4 and
R5 represent the resistances,
Ms is the sensing MOS of the pressure sensor,
Vdd is a voltage power supply, and
Vout is the output voltage.
Figure 8 shows the design of the operational amplifier circuit, where
Vdd represents a voltage power supply and
Vss is the ground.
Figure 9 depicts the simulated results of the frequency response for the operational amplifier. The dc open loop gain of the operational amplifier is approximately 93 dB, and the phase margin of the operational amplifier is about 80°.
Figure 10 displays the simulated output voltage of the readout circuits. In this simulation, the resistances
R1,
R2,
R3,
R4, and
R5 are set with 0.1, 1, 10, 0.1, and 0.1 kΩ, respectively. The
Vdd and
Vgs voltages are given by a voltage of 3.3 V and 3 V, respectively. The
Ids current of the pressure sensor is set to change from 1560 to 1585 μA. The simulated result shows that the output voltage changes from 1085 to 1065 mV as the current of the pressure sensor varies from 1560 to 1585 μA.
3. Fabrication of Pressure Sensor
The commercial 0.35 μm CMOS process of Taiwan Semiconductor Manufacturing Company (TSMC) is employed to fabricate the FET pressure sensor integrated with readout circuits.
Figure 11 displays the process flow of the FET pressure sensor with circuits.
Figure 11(a) shows the schematic cross-section of the pressure sensor after the CMOS process. In the pressure sensor, the membranes are composed of the oxide, metal and oxide layers. The etch holes are filled with metal and via layers, which the materials of the metal and via layers are aluminum (Al) and tungsten (W), respectively. The metal layer under the membrane and the Al and W layers in the etch holes are the sacrificial layers. In order to obtain the suspended membranes, the FET pressure sensor needs a post-CMOS process to remove the sacrificial layer. The post-process used wet etching to etch the sacrificial layers, and to release the suspended membranes.
Figure 11(b) displays that the pressure sensor is immersed in two etchants: one is an Al etchant with phosphoric acid, nitric acid, acetic acid and DI water in the ratio 14:1:2:3 and the other is a W etchant with sulfuric acid and hydrogen peroxide in the ratio 2:1. The sacrificial layers are removed and the suspended membranes are released, which an air gap between the membrane and substrate is formed.
Figure 12 shows the photograph of the FET pressure sensor with readout circuits on a chip after the wet etching process.
Figure 13 depicts a scanning electron microscopy (SEM) image of the FET pressure sensor after the wet etching process. A white light interferometer (Zoomsurf 3D from Fogale Nanotech Co.) is utilized to measure the flatness of the membranes. The result shows that the maximum deflection of the membranes is about 0.06 μm. The etch holes in the pressure sensor have to be sealed.
Figure 11(c) shows that the etch holes are sealed using a LPCVD parylene, and the paraylene film is patterned by a dry etching. The thickness of the parylene film is about 1.6 μm. The FET pressure sensor that the cavities are nearly at vacuum is an absolute pressure sensor due to the LPCVD parylene is processed in a high vacuum chamber.
4. Results and Discussion
The FET pressure sensor was mounted in a pressure chamber. The nitrogen pressure source was supplied to the pressure chamber, and the nitrogen pressure in the chamber could be tuned through the gas valves. A calibrated pressure sensor was utilized to monitor the gas pressure in the pressure chamber. The power supply provided the
Vgs and
Vds (drain-to-source voltage) voltages of the FET pressure sensor. The output current of the FET pressure sensor without the readout circuits was detected using the digital multimeter.
Figure 14 reveals the measured results of
Ids/
Vds output characteristics for the FET pressure sensor. In the measurement, the FET pressure sensor was tested at different
Vgs (from 0 to 5 V) and
Vds (from 0 to 3.3 V) voltages without pressure in the pressure chamber. The experimental results of
Ids/
Vds characteristics for the FET pressure sensor were in agreement with the simulated results in
Fig. 6.
The FET pressure sensor without the readout circuits was tested at different pressures. The power supply provided the
Vds voltage of 3.3 V and the
Vgs voltage of 3 V to the FET pressure sensor, and the output current of the pressure sensor was measured using the digital multimeter.
Figure 15 displays the relation between the pressure and output current for the pressure sensor. The experimental results showed that the output current of the pressure sensor changed from 1560 to 1571 μA as the pressure varied from zero to 500 kPa, and the curve was a linear with a slope of 0.022 μA/kPa.
The FET pressure sensor integrated with the readout circuits, in which the circuits were applied to convert the output current of the pressure sensor into the output voltage. The power supply provided the
Vds voltage of 3.3 V and the
Vgs voltage of 3 V to the FET pressure sensor, and the output voltage of the FET pressure sensor with the readout circuits was measured at different pressures using an oscilloscope.
Figure 16 shows the relation between the pressure and output voltage for the pressure sensor. The measured results revealed that the output voltage of the pressure sensor varied from 1083.6 to 1067.6 mV as the pressure changed from zero to 500 kPa, and the curve was a linear with a slope of 0.032. Thereby, the pressure sensor had a sensitivity of 0.032 mV/kPa at
Vds=3.3 V and
Vgs=5 V under the pressure of 500 kPa.
Svensson et al. [
9] and Hynes et al. [
10] employed surface micromachining process to fabricate the FET pressure sensors without readout circuits. Comparing with Svensson et al. [
9] and Hynes et al. [
10], this work used the CMOS-MEMS process to manufacture the FET pressure sensor integrated with readout circuits on a chip, in which the output current of the FET pressure sensor was converted into the output voltage by the readout circuits. On the other hand, the pressure sensor reported by Lin et al. [
5] had a sensitivity of 0.15mV/psi. Wu et al. [
6] presented the pressure sensor that had a sensitivity of 177.6 μV/psi. A comparison with Lin et al. [
5] and Wu et al. [
6], the sensitivity of the pressure sensor in this work exceeded that of Lin et al. [
5] and Wu et al. [
6]. The capacitive pressure sensor proposed by Dai et al. [
7] had a chip area of 2×2 mm
2. The area of the FET pressure sensor in this work, which was about 0.6×0.8 mm
2, was less than Dai et al. [
7].