Tone Reversal


Nanoimprint can easily produce cavities in resist, but it is very challenging to produce protrusions. One way to achieve protruding patterns in the resist is to transform the cavity pattern into a protrusion pattern via a process called tone reversal. Developing a tone reversal process will enable nanoimprint technology to produce the full range of patterns that can be produced using the more established techniques such as photolithography and ebeam lithography.


Develop a tone reversal process for nanoimprint.

Design of experiment

We have explored a couple of approaches to achieve tone reversal without much success. Below is a brief description of our failed ideas:

  1. Spin coat resist on mold
    • Failed because mold is coated with anti-sticking layer so resist does not coat the mold
  2. Spin coat HSQ on mr-I pattern
    • Failed because mr-I resist dissolves in HSQ solvent (MIBK)
These ideas failed very quickly and the failure mechanism is quite clear. Our next idea is to use HSQ and PMMA since we know that PMMA has been used for nanoimprint and that PMMA can be etched with high selectivity to HSQ.

The experiment is described below:

  1. Start with oxide wafer
  2. Coat with PMMA
    • PMMA 950 A4.5
    • Spin at 3000 rpm at 1250 rpm/s for 60s
    • Bake at 180 °C for 2 minutes
  3. Coat with HSQ
    • Dow Corning XR-1541-4 (HSQ 4%)
    • Spin at 3000 rpm at 1250 rpm/s for 60s
    • Bake at 90 °C for 1 minutes
  4. Measure thickness of film
    1. Dip half the chip in TMAH 25% to remove HSQ
    2. Scratch both half of the film
    3. Use AFM to measure thickness of PMMA, HSQ and PMMA+HSQ
  5. Etch patterns in a reactive ion etcher using the HSQ etch
    • Etch for 30s
    • Oxford DRIE 180
    • Parameter Units Value
      RF Power [W] 70
      ICP Power [W] 200
      Pressure [mTorr] 6
      C4F8 [sccm] 45
      O2 [sccm] 5
      DC Bias [V] 280
      Temperature [C] 20
      Helium [Torr] 10
  6. Measure the thickness of the film at the same 3 spots
  7. Strip the grating patterns by immersing the sample in acetone for 5 minutes
  8. Measure the height the 400 nm grating with an atomic force microscope


Figure 1 is a graph of the etch rates for SiO2 and HSQ at various ICP power. After realizing that the etch rates for both SiO2 and HSQ is the same at an ICP power of 200 W, we decided to run the experiment again to confirm our results. From the second set, we confirmed that the data results at an ICP power of 200 watts is repeatable and therefore real. It will be interesting to perform another experiment in the future to determine what the etch rates at ICP power between 0 and 350 watts.

Figure 1: A graph showing the etch rate versus ICP power

Figure 1: The etch rate increases monotonically with ICP power and the two sets of measurements are in agreement.

Figure 2 is a graph of the etch selectivity of SiO2 and HSQ. At 350W and above the selectivity of SiO2 to HSQ is approximately 0.6 to 1. An anomaly occurs at an ICP power of 200 W, where the selectivity is almost 1 to 1.

Figure 2: A graph showing the etch selectivity of SiO2 and HSQ.

Figure 2: The etch selectivity of SiO2 and HSQ is approximately 0.6 for all ICP power values except 200 W.

Figure 3 is a graph of the DC bias that is observed at each etch process. Normally, the DC bias is deliberately decreased to increase selectivity. In this case, we have the highest selectivity at the highest DC bias. What is going on?

Figure 3: A graph showing the DC bias at various ICP Power.

Figure 3: The DC bias decreases monotonically as ICP power increases.