Using Video Microscopy to Characterize Micromechanics of Biological and Man-Made Micromachines (invited)

Dennis M. Freeman and C. Quentin Davis

Presented at the Solid-State Sensor and Actuator Workshop

Hilton Head Island, SC, June 1996.

Part IV: Application to MEMS

still picture (130K gif)
I began today's talk by comparing the dimensions of MEMS structures on this 1 cm die to the dimensions of hair cells. Now let me zoom in and use computer microvision to analyze the motion of a cantilever beam system.
still picture (220K gif)
This is an image of the comb drive portion of a cantilever beam system as taken with our video system. The associated video illustrates the motions that result when the combs are driven differentially with a 2 volt AC sine wave superimposed on a 62 volt DC bias. The frequency is 7.7 kHz.
still picture (220K gif)
Just as we did for the biological targets, we can indicate a region of interest for the comb drive. Here and in the associated video I've enclosed the central shuttle region.
still picture (11K gif)
We've applied our motion estimation algorithms images inside this region of interest. The motion in the associated video is 1.18 micrometers peak-to-peak and the phase of the displacement lags that of the electrical stimulus by 60 degrees. We can repeat these measurements, varying frequency and holding the voltage constant, to obtain a frequency response.
still picture (10K gif)
The magnitude and phase of the displacement of the shuttle are plotted here for 53 different frequencies of excitation. The results are well fit by a second order system with a Q of 13 and a best frequency near 8 kHz.
still picture (15K gif)
We repeated this experiment 5 times to estimate the accuracy and repeatability of the measurement. The results show standard deviations on the order of 3 nm. That's more than 40 dB below the signal at low frequencies and more than 60 dB below the signal at resonance. The signal to noise ratio is smaller above 10 kHz, where the amplitude of the signal is falling with increasing frequency. The standard deviation of the phase measurements are about 1 degree except at the highest frequencies, where the magnitude is small.
still picture (21K gif)
The motion measurement system can also be used to estimate three dimensional motions. This slide and the associated video shows a 3D image that contains two teeth of one of the comb drives. One tooth is moving and the other tooth is stationary. Here, the comb is being excited differentially with 60 volts AC added to a 62 volt DC bias. The frequency is 20 kHz.
still picture (30K gif)
We can use computer microvision to estimate all three components of the displacements of both teeth. The top panels and the associated video show quantitative estimates of the horizontal displacements. This experiment was repeated 10 times to estimate the standard deviation of the measurement. Results showed that the moving tooth moved about 500 nm peak-to-peak with a standard deviation of about 1 nm. Measurements of motions of the stationary tooth provide a different type of error assessment. These measurements also indicate that our noise floor is on the order of nanometers.

Measurements of the orthogonal component of in-plane motion (center panels) are small: about 100 times smaller than the horizontal motions in the top panels. The orthogonal components of in-plane motions of both the moving and stationary tooth are near our noise floor.

Out-of-plane motions (bottom panels) for the moving tooth are clearly resolved. These motions are only 10 times smaller than the horizontal motions in the top panels and are significantly larger than either the standard deviations or the out-of-plane motions of the stationary tooth.

still picture (68K gif)
Now one might expect that out-of-plane motions could depend on in-plane position. To analyze this possibility we examined motions in three regions: one on the left, which contained a moving tooth; one near the center of the shuttle; and one on the right, which contained a moving tooth. Results in this slide and in the associated video show that motions in the left and right regions of interest had comparable magnitudes but were nearly out of phase. Motions of the shuttle were nearly an order of magnitude smaller than those on either side. These measurements show that the out-of-plane motions correspond to a rocking mode of motion.
still picture (134K gif)
To summarize (video): we have developed a computer microvision system that combines video microscopy and computer vision. We have applied the system to study sound-induced motions of inner ear structures, and the results have provided important new insights into cochlear mechanics. We have also demonstrated the use of computer microvision in MEMS.

Computer microvision has several properties that are unique among measurement systems. Unlike most motion measurement systems, computer microvision provides estimates of all 3 components of 3D motion. Although the images are acquired with a light microscope, motions can be resolved with nanometer precision. Lastly, computer microvision allows one to simultaneously measure motions of all visible structures in the field of view, with no additional built-in sensors.

Thank you for your attention.
Are there any questions?

Back to the talk's title page.