EE245 Homework 3

Due 9/25/01


First Name: 
Last Name: 
SID #: 

E-mail Address for confirmation: 

  1. Go to the LAVA web page (or try the backup) and under "Applications" run the "deposition and metalization" simulation.  You should see a yellow "substrate" with a 1 micron trench etched into it, and some straight lines above it representing line-of-sight deposition.
    1. Don't change any defaults, just press "Submit to Sample2D".  You should see that the trench is filled in, but that the corners didn't fill in well.  Using one of the red dots on the trench, or the sliders on the side, change the top aperture to 1.1 microns and run the simulation again.  Assuming that this is a metal deposition onto an insulator, will there be electrical connectivity between the metal on top and the metal in the hole?  Change the top aperture to 0.9 microns and re-run the simulation.  Will there be electrical continuity now?
    2. Reset the simulation so that the trench is 1 micron square again, and change the source type to "Hemispherical", and run the simulation.  Using the mouse, determine the deposition rate in the field, and in the center of the bottom of the trench.  Change the deposition time to 400s and graph the two deposition rates on the same plot.  Put the URL for this plot here:  What can you say qualitatively about the deposition rates in the field, at the bottom of the trench, and on the sidewalls?
    3. Reset the simulation to a 1 micron square trench, use hemispherical deposition with cosine flux, and set the positive and negative angles to +-90.  How long of a deposition is necessary before the top of the keyhole is less than 0.1 microns wide (use the mouse to measure)? What film thickness does this correspond to?
    4. Reset the simulation to a unidirectional source, and select a "line" profile.  Change the base of the line to a width of 0.75 microns, leaving the top at 1micron.  Run the simulation twice, once with a 200second deposition, and once with 100 seconds.  If the "line" were made of photoresist, and the film deposited were gold, which of these two thicknesses would likely work if the goal were to pattern the metal by liftoff?
    5. Reset the simulation to hemispherical source, uniform flux, and +-90 angles, a line profile, and 100s deposition.  What fraction of "blue sky" can the top and side of the beam see (in degrees):    What deposition rate would you predict for the top and sidewalls of the beam? What are the deposition rates on the top of the beam, on the side of the beam, and in the lower/inner corners of the beam?   Does the simulated deposition rate fit with what you predicted? Try the same simulation with longer depositions, and think about how reflow would affect an oxide shaped like this.
  2.  Figure 3.19 (from Senturia, Microsystem Design) is a clear conceptual illustration of a surface micromachining process, but it is somewhat lacking in practicality.  Lets assume that the structural layer is polysilicon, the sacrificial is LTO, and the substrate is bare SCS.  Assume that:
    3. Re-draw the third and fourth cross-sections of Figure 3.19 with these assumptions, using the following hints: Now assume that we modify the process by adding a uniform (unpatterned) 0.5 micron silicon nitride layer before any other processing.  Re-draw all of the cross-sections from Figure 3.19 with this nitride layer and the assumptions and your calculations above.  URL for cross-sections:
  3. Calculate the actual selectivities in the oxide and polysilicon etches that would be typical in the Berkeley microlab, assuming the Lam etch for poly above, and the lower-power Lam 590 oxide etch from William's paper for the LTO etch.


or 

(K. Pister)