Strain Plethysmography Using a Hermetically Sealed MEMS Strain Sensor

Figures

• 

  Figure A1. Die level FEM parametric simulation describing effects of the device layer and the handle layer thickness on strain sensor sensitivity: (a) The strain sensor sensitivity improves slightly with a thinner device layer (5–60 µm). (b) The strain sensor sensitivity can be improved substantially by reducing the thickness of the handle layer (100–600 µm).

• 

  Figure A2. Simulation result of strain sensitivity with different cantilever thicknesses. The error is due to the loading of the strain sensor being less than 10% when the thickness of the cantilever beam is 4 mm.

• 

  Figure 1. Capacitive MEMS strain sensor: (a) SEM image of the MEMS strain sensor. (b) Cross-section of the MEMS sensor showing the differential capacitive sensing pair. C+ and C− share a common terminal connected to the ground. The inset shows the wafer-level-packaged MEMS strain sensor die placed on a US quarter.

• 

  Figure 2. Fabrication process flow for the capacitive MEMS strain sensor. (a) The SOI device layer is patterned and filled with TEOS. (b) Pattern TEOS. (c) Deposit and pattern poly-Si. (d) Oxide etching to release the suspended plate. (e) Prepare the cap wafer. (f) Wafer bonding and stealth dicing.

• 

  Figure 3. Analytical model for strain sensors. The strain sensor is treated as a prismatic beam with a central symmetric plane.

• 

  Figure 4. Strain sensing system: (a) The strain sensing system includes the MEMS die, an application-specific integrated circuit, a thin printed circuit board, and a metal cap. The complete strain sensing system features a form factor of 15 mm × 17 mm × 3 mm. (b) An image showing the small form sensor placed on the fingertip.

• 

  Figure 5. (a) COMSOL simulation of strain distribution in the strain sensor. (b) Die-level FEM simulation result shows a strain sensitivity of 2.34 fF/µε with a linear range of 800 µε. (c) Strain sensitivity characterization test setup. The strain sensor is placed near the fixed end of a long cantilever, and the input strain is controlled by the displacement δy at the free end of the cantilever using a micrometer. The cross-section AA’ is shown on the right [18]. (d) The characterization result is compared with FEM simulation result using the cantilever setup specifications. The strain sensor system shows a sensitivity of 316 aF/µε, a linear range of 700 µε, and a linearity of 99.85%.

• 

  Figure 6. The strain sensing system responds to dynamic input strain. (a) Test setup. The strain input to the strain sensor is controlled using a rate table. (b) The sensor output closely follows the dynamic input strain with ~300 aF/µε sensitivity and high repeatability.

• 

  Figure 7. Frequency response of the MEMS strain sensor to strain input simulated using FEM analysis. The sensor exhibits a high bandwidth of 65 kHz. The inset shows the first eigenmode of the sensor at 46.3 kHz.

• 

  Figure 8. Time series noise output of the strain sensing system. The sensing system exhibits a noise of ~400 aF. This corresponds to a strain sensing resolution of 1.26 µε.

• 

  Figure 9. Experiment setup for cardiovascular parameter estimation. The SCG signal (blue) and SPG signal (red) collected from the sensors are sampled by the signal processing hub and displayed on the computer monitor.

• 

  Figure 10. SPG signals are collected from the fingertip pulse location for experiments 1 and 4. (a) Illustration of sensor placement at the digital artery of the fingertip. Medical tape is used to attach the sensor to the fingertip. (b) SPG signal collected from the fingertip possesses the key features of the PPG signal.