Reactive Ion Etching of Si:AZ1512 using the Bosch Process

Motivation

The Bosch process is a deep reactive ion etching process that is commonly used in the fabrication of microstructures such as micro electro mechanical sensors (MEMS). The Bosch process can etch at rates of 3 microns per minute with selectivity to photoresist or oxide greater than 50 to 1. This process can produce vertical sidewalls with scallops that are a few hundred nanometers deep.

Objective

Characterize the etch rate of Si and photoresist (AZ1512) for the Bosch process. Optimize the process parameters to achieve the highest selectivity.

Design of experiment

The Bosch process is a deep reactive ion etching process that etches deep structures with straight sidewalls. The Bosch process can etch silicon with rates of a few microns per second and has selectivity of up to 100 with either photoresist or silicon dioxide. The objective of this work is to run the Bosch process recipe that came with the Oxford RIE180 and characterize the etch rate and selectivity to AZ1512. The experiment is described below:

  1. Start with a silicon wafer
  2. Spin on AZ1512
    • Spin at 2500 rpm, 1250 rpm/s, 60 s
  3. Bake on a hotplate at 100 °C for 1 minute
  4. Expose a pattern at 80 mJ/cm2
    • The pattern used was a 1 mm wide cross
  5. Develop in AZ300 for 1 minute
  6. Measure the height of the pattern with a profilometer
  7. Place sample on a 4 inch Si carrier wafer
    • Smear PTFE oil on the back of the sample to improve thermal contact with carrier wafer
  8. Etch patterns in a reactive ion etcher
    • Oxford DRIE 180
    • The deposition and etch step is repeated 15 times
    • Common process parameters:
    • RF Power [W] 25
      ICP Power [W] 700
      Pressure [mTorr] 30
      Temperature [°C] 15

    • Deposition step parameters:
    • C4F8 [sccm] 100
      SF6 [sccm] 1
      DC Bias [V] 156
      Duration [s] 5
    • Etch step parameters:
    • C4F8 [sccm] 1
      SF6 [sccm] 100
      DC Bias [V] 151
      Duration [s] 6 to 8
  9. Wipe off the PTFE oil with acetone
  10. Measure the height of the pattern with a profilometer
  11. Strip the remaining photoresist patterns by immersing the sample in acetone for 5 minutes
  12. Measure the height of the pattern with a profilometer
  13. Image the sample with a scanning electron microscope

Results

Figure 1 is a graph of the etch rates for Si and AZ1512 for etch times between 6 and 8 seconds. The etch rate for Si and AZ1512 increases monotonically with etch time, as expected. If we normalize the etch rate by the etch time, we learn that the normalized etch rate at 6 seconds is 30% slower than the normalized etch rate at 7 and 8 seconds.

Figure 1: The etch rate increases monotonically with etch time and the photoresist (AZ1512) etches significantly slower than Si.

Figure 2 is a graph of the etch selectivity of Si and AZ1512. The selectivity is approximately the same for etch times of 6 and 7 seconds, but falls significantly at an etch time of 8 seconds. From this narrow dataset, the maximum selectivity of Si to AZ1512 (or photoresist) is approximately 34 to 1.

Figure 2: The etch selectivity of Si and AZ1512

Figure 3 is is an image of the scallops that form on the sidewalls due the the deposition and etch cycle of the Bosch process. The 15 scallops is consistent with the 15 repeats of the deposition and etch cycle. The inset shows a FIB cross-section of the scallops, revealing the width of the scallops to be approximately 250 nm.

Figure 3: An image of the scallops formed on the sidewalls of the etched pattern.

Figure 3: The 15 scallops that form on the sidewalls of etched pattern is caused by the Bosch process.

According to the Bosch process recipe sheet provided by Oxford, the maximum selectivity is 75:1 for a deposition power of 10 W and a DC bias of 70 V. This same recipe should also achieve a selectivity of up to 120 using an oxide mask instead of a photoresist. When we reduced the RF power of the deposition step to 10 W, the DC bias reduced to 93 V. The overall etch rate reduced (Figure 4) but the selectivity remains approximately the same (Figure 5).

Figure 4: The etch rate of Si reduces slightly with lower deposition power.

Figure 5: The etch selectivity between Si and AZ1512 does not change significantly with deposition power.

The rate at which silicon etches is dependent on the amount of exposed area. As a general rule, the etch rate begins to reduces as the exposed silicon is larger than 20%. The samples used for these experiments are approximately 3 cm x 3 cm squares sitting on top of a 4 inch silicon wafer, so roughly 90% of the of the area is exposed silicon. When the same recipe (deposition with 25W) was used on a wafer where less than 10% of the area is being etched, the silicon etch rate is 4100 nm/min with a selectivity of 94. Due to the complexity of reactive ion etching, it is important to calibrate the etch rate and selectivity for each sample pattern.