Specific Process Knowledge/Etch/DRIE-Pegasus/Pegasus-4/SiO2 Etch/SiO2 etch with resist mask

From LabAdviser

Feedback to this page: click here

Design of experiment "DOE03" - SiO2 etch with DUV resist pattern

Done by Berit Herstrøm @DTU Nanolab Q1 2021

The design of the experiments "DOE03"

The design factors were:

  • Coil power: 2500-4000 W
  • Platen power: 200-300 W
  • Total flow: 100-500 sccm
  • He/C4F8 ratio: 5-30
  • C4H8/H2 ratio: 1-3

Fixed settings were:

  • Spacers used: 30 mm
  • APC valve: 100% open (to keep pressure as low as possible) - Pressure will be very correlated to the total flow rate.

DOE03 run plan:
This has been designed in the software JMP PRO. Click on the image to make it larger.

Samples:

  • One piece 2x2cm DUV patterened bond to a Si carrier wafer
  • ~2µm SiO2 - ~90nm Barc, ~900nm Neg DUV resist - mask: Pegreticle
  • Barc etch before the SiO2 etch: BarcO2 for 30 s

The responses I have evaluated are (See decriptions below):

  • Etch rate of SiO2
  • Etch rate selectivity to resist
  • Etch profile angle in low aspect ratio areas
  • Etch profile angle in higher aspect ratio areas
  • Change in critical dimensions
  • Aspect ratio dependent etch rate

Results from the run of DOE03 - SiO2 etch with DUV resist pattern

The DOE analysis of "DOE03"

Model of all responses together. Se explaination on the responses below.

  • Here the responses are fittet one at a time (This gives better models for the individual responses but the overview is less good):
    • Interactive HTM file: COPY THE HTM PATH INTO YOUR BROWSER, this will only work if you have acess to the Nanolab cleanroom drive - Try out the prediction profilers:

file:///O:/Intern/Equipment%20groups/STS/Udvikling/Pegasus%204%20oxide%20etch/DOE3%20inkl.%20H2%20and%20He/Fit%20Model%20all%20responses%20individually%20fit%202.htm

  • Here they are fittet together (This gives a good overview but goes not give the best model each of the responses):
    • Interactive HTM file: copy the link into your browser, this will only work if you have acess to the Nanolab cleanroom drive) - go to the bottom of the HTM page and try out the prediction profiler:

file:///O:/CleanroomDrive/_Equipment/Pegasus%204/DOE3/Fit%20Model%20all%20responses%202.htm

About the .htm interactive files: Copy the htm link into your browser and the document will open if you have access to the Cleanroom drive. The prediction profiler is interactive! Be aware that the resulting response from the prediction profiler might have large statistical uncertainty. Unfortunately the document does not include the coinfidence interval so you should only use it as guide lines. If you look at the below .pdf file for one selected set of parameters in the prediection profiler you can see the coenfidence interval for that set of factors.

Explanation of the responses that are included in the models:

Response Etch rate SiO2 in low aspect ratio: Etch rate in low aspect ratio areas
Here I have modelled the etch rate in the low aspect ration etched areas (4 µm pitch and 1 µm lines).

Response Selectivity: Etch rate selectivity to the resist
This is defined as the SiO2 etch rate in the low aspect ratio etched areas (4 µm pitch and 1 µm lines) devided by the resist etch rate also on the 1 µm lines in these areas.

Response SiO2 profile angle LAR: Etch profile angle in low aspect ratio etching
Modelling the etch profile angle is difficult because it depends on the aspect ratio of the etched structures. If the mask opening around the structure is small compared to the depth of the etch the aspect ratio is high. If the opening is large compared to the etch depth the aspect ratio is low. Here I have looked at low aspect ratio - more precisly I have looked at the profile of lines with 4 µm pitch and 1µm line width. The etch depth was been target to 1 µm. A different etch depth or a different starting thickness of the resist pattern will most likely give another profile angle.

Response SiO2 profile angle 1:1: Etch profile angle in 1:1 aspect ratio etching
Modelling the etch profile angle is difficult because it depends on the aspect ratio of the etched structures. If the mask opening around the structure is small compared to the depth of the etch the aspect ratio is high. If the opening is large compared to the etch depth the aspect ratio is low. Here I have looked at 1:1 aspect ratio - more precisly I have looked at the profile of lines with 2 µm pitch and 1 µm line width. The etch depth was been target to 1 µm. A different starting thickness of the resist pattern will most likely give another profile angle. I have not modelled higher aspect ratio structures since for the higher spect ratio structures in this study some of the samples has not been etched at all in the small opening and then the profile angle response cannot be modelled.

Respose ER 1:05/ER4:1: Aspect ratio dependent etch rate Different mask opening sizes etches with different rates - especially when the aspect ratio goes above 1 meaning the etch is deeper that the size of the opening. Here I have modelled the spect ratio dependant etch rate by modelling the ratio of the etch rate in the lines 4 µm pitch and 1 µm line over the etch rate in 1 µm pitch and 0.5µm lines. The etch depth in the large opening is about 1 µm.

Respose CD top/Bottomdifference: Change of critical dimensions (CD change)
Modelling the change in critical dimensions - that is the difference between the mask opening before etch and after etch. This has been measured as an average of the different size seen on the SEM images. It has been measured both in the top and the bottom. This modelling also includes the difference in CD change between top and bottom - this is actually another measure for the profile angle.

SEM images