Saturday, October 12, 2013

The MEMS post

This summer I was fortunate enough to be selected to participate in the Lutchen Distinguished Fellowship Program. Essentially a more rigorous UROP, it allowed me to continue my MEMS resonator research with Professor McDaniel through the summer.



Rewinding a bit, I started working with MEMS in late 2012, applying for UROP for spring and fall 2013. The Lutchen Fellowship is offered to 10 students each year who have partnered with a university professor to work on a research project. The program ends with a presentation by the students to the other Fellows, their faculty mentors, and Dean Lutchen.

My first semester of research involved learning MEMS CAD software, called L-Edit, which is a delightful program that only runs in 256 color. To further show how much I love this program, I will also mention that the student version lacks the ability to include text on the design. Unfortunately, this is the software that MEMSCAP, the company that manufactures the dies, wants us to use. After slaving for hours, the final designs from L-Edit are submitted four times a year to MEMSCAP and the products are shipped back to you with just enough (not nearly enough) time for you to release, test, and redesign devices for the next production run.

All ~20 of my devices fit on one of those little squares, which are 2.5mm x 2.5mm
A few SEM images for your enjoyment:

Wide angle shot of whole die
A pair of double camped beams
coupled double-clamped beams
Close up of capacitive gap
I began my summer research by a frequency divider for lab use. Although I didn't end up using this extensively, I gained valuable experience in circuit and PCB design. I used CircuitLab to simulate and Eagle to design the board, which was cut by my friend Steve. Thanks, Steve!

Thanks, CircuitLab!

Green in, Yellow out.

Next came learning how to use the Network Analyzer. The HP 3577a Network Analyzer is a charming machine that looks, feels, and smells like the 1980s. Gradually learning the quirks of this particular machine, I wrote a delightful instructional manual which can be downloaded here. About twice as old as I was at the time, the NA had seen its fair share of wear and tear (talk about rhyming). Several of the buttons were sticky, the GPIB/LabView interface was essentially unusable, and the knob that moves the cursor on screen was... wait for it... totally broken. The silver lining of this situation is that I learned to perform the measurements I needed the hard way, which led me to find menus I didn't know existed, giving me a better understanding of what the machine could do.

After a few weeks of trial and error with the measurement methods and vacuum chamber, I eventually found a circuit that worked and found several resonances. The Qs of these resonators were not as high as we had expected, but we realized we were limited by design parameters. The main issue, we speculate, is that the minimum gap between the resonating beam and the driving pad is .75 microns, which forces us to have a large DC bias and drive voltage. This means that we are driving the beam really hard, which causes changes in the center frequency and reduction in quality.


My vacuum chamber
Wire Bonding (cell phone microscope picture)

The first resonance I could find in my pictures
What we established:

Resonant frequency is proportional to AC drive and inversely proportional to DC bias
Quality is inversely proportional to both AC drive and DC bias

Take a look at the hastily-constructed charts below for more details.



This research will continue through (at least) the fall semester and hopefully after that as well. Stay tuned for more updates and beautiful charts.