There were two main objectives of the NE 471L laboratory:
i) fabricate the MEMS devices designed in Layout Editor in the Fall 2010 term,
ii) examine and test the devices, analyzing their performance relative to the theoretical performance as
modeled in COMSOL during the Fall 2010 term.
Recalling our work done in Fall 2010, 2 distinct MEMS devices (and many variations of each device) were
designed. One design was a hot-arm cold-arm thermal actuator and the other was a scratch drive
actuator. Both were first modeled in COMSOL to determine design parameters (ex. How thin can our
arms be without melting the device?) and input/output parameters (ex. Amount of deflection given a
certain applied voltage). After modeling, work was done in Layout Editor to create a 4-layer mask for
fabricating the two devices.
The 4 masks were used in this Winter 2011 term to execute fabrication. The type of fabrication executed
in this lab is surface micromachining, including the use of structural (aluminum, chromium, nitride) and
sacrificial materials (photoresist). The first lab comprises the Cr layer formation, nitride layer formation
by PECVD, photolithography and dry etch of the nitride, and the sacrificial layer formation. The second
lab comprises the aluminum layer deposition via sputtering, and the imaging and release of the
sacrificial photoresist layer.
The results of I/O testing on a four-point probe station are included in this report and the comparison to
the theoretical performance. Comments on success and failures are included as well as
recommendations for future designs.
Results and Discussion
For a detailed overview of the fabrication process, please refer to the following two sources:
- Ivanova, Rossi, University of Waterloo. Nanotechnology Engineering Advanced Laboratory
2 NE 455L/471 Nanoelectronics Winter 2011 Lab Manual, January 2011.
- Ivanova, Rossi. University of Waterloo. DC 3701 Cleanroom Processing Procedures,
No major changes to the procedure were made. The entire process may be replicated by
following the procedures outlined in those manuals.
Testing of the Film Properties
The following voltage measurements for sheet resistance of aluminum were made in laboratory:
Average: 3.417 mV The measurements were made with I = 0.4532 A. We know Rs = 4.532*V/I, therefore using the
average value for V, the sheet resistance for the chromium layer was
Rs = 4.532*3.417mV/0.4532A = 34.17 m. This is exactly what would be expected of the
aluminum resistance; it is very close to the theoretical value .
The following measurements were made for the sheet resistance of chromium were made:
V = 7.673 mV
I = 4.5320 mA
The sheet resistance is calculated via Rs = 4.532 *V/I = 7.673 . This is extremely high, as the
theoretical value for the sheet resistance of chromium is 0.43 . However, since we are not
utilizing the chromium layer for any current flow, this should not be a major problem for the
devices described here.
Examining the thickness of the chromium and nitride layers, the Dektak machine was utilized.
The figure below shows the profile of the chromium and silicon nitride layers, not corrected for
DekTak Data: Chromium and SiN Thickness
0.00 200.00 400.00 600.00 800.00 1000.00
Height (a.u.) SiN
Each increment on the y-axis represents 500 angstroms. Thus, the thickness of chromium is on
average 1782.5 angstroms and the thickness of silicon nitride is on average estimated to be
1500.25 angstroms. The thickness of both the chromium and nitride layers were supposed to be
closer to 3000 angstroms, thus could have indicated some difficulties in the fabrication process.
This discrepancy in the process has the potential to explain some difficulties encountered in
testing the devices. This is discussed further in the sections below.The change in thickness could also mean the COMSOL models were inaccurate. For example, the
predicted deflection of a plate caused by the capactive force would no longer be accurate,
because separation between the two plates is smaller than what was originally modeled.
First Simulated Device: Thermal Actuator
Our device is designed to be like a woodpecker. Joule heating will cause the arm to extend the
asymmetric design of the device allows one arm to expand more than the other, causing the end of the
actuator to hit the aluminum block. When they come in contact, the actuator short circuits and halts
Joule heating. As a result, the aluminum contracts and brings the tip out of contact with the aluminum
block. Joule heating can thus begin again and the cycle continues.
The COMSOL simulations were run also to ensure the aluminum did not exceed the melting point of
933K. Using a parametric solver in COMSOL ranging from 0V to 0.2V with a 0.01V step size, it is the
temperature of the MEMS structure at the tip is 900K when a v