STS Pegasus 4/Processes/LNF Recipe 1
|About this Process|
Recipe 1 for 20 min, 2 µm Trench
|Equipment||STS Pegasus 4|
|Mask Materials||PR, SiO2|
|Gases Used||SF6, C4F8, O2|
|Date Created||Jan 1, 2009|
|Authored By||SPTS, Brian Vanderelzen, Kevin Owen|
LNF Recipe 1 is the base silicon etch for the pegasus tools. It has a high selectivity, moderately fast etch rate and a clean profile.
LNF recipe 1 has 2 main process (DRIE) steps. For recipes shorter than 20 minutes, set the first DRIE step to the desired length. For recipes longer than 20 minutes, set the first DRIE step to 20:00 and the second DRIE step to the remaining time.
Run LNF Recipe 1 for up to 20 minutes with the 1 minute automated clean recipe, longer etches should be run with the 3 minute automated clean recipe.
|ICP Power||2000 W||2800 W|
|Bias Power||0 W||60 W|
|Bias Pulsing||N/A||40 Hz - 80%|
|Pressure||24 mTorr||30 mTorr|
|C4F8 Flow||250 sccm||30-0 sccm (ramped)|
|SF6 Flow||0 sccm||390 sccm|
|O2 Flow||0 sccm||39 sccm|
|Time||2.0 s||2.6 s|
This process, as is true with all Bosch processes, varies depending on feature size and density, due to RIE lag. The below table contains typical results for 2 µm, 10 µm, and 100 µm trenches and the following sections expand on these results. It is strongly recommended that the process is tested with any new mask.
|Parameter||2 µm Trench||10 µm Trench||100 µm Trench|
|Etch Rate||3 µm/min||4.5 µm/min||5.6 µm/min|
|Undercut||160 nm||200 nm||200 nm|
|Scallop height||370 nm||390 nm||370 nm|
|Scallop depth||70 nm||50 nm|
The etch rate will vary with the size of the trench being etched. Larger trenches tend to etch faster, this trend can be seen below.
- Photoresist Etch Rate
- SPR 220: 500 Å/min
- SiO2 Etch Rate
- Thermal oxide/HTO: 166 Å/min
- LTO: 178 Å/min
- PECVD: unknown
- Silicon nitride Etch Rate
- LPCVD Nitride: 356 Å/min
- LPCVD Low Stress Nitride: 291 Å/min
Mask undercut, scalloping, and sidewall angle are shown below.
Running recipe 1 for 20 minutes with 3 µm of SPR 220 on a silicon wafer with our standard DRIE characterization mask can produce a profile like this.
LNF Recipe 1 is known to work under a range of conditions. Outside of these conditions, etch performance is not guaranteed. Etch rates will vary depending on feature size, etch length, open area, etc. We strongly recommend testing the process prior to running device wafers.
One common problem seen when running outside of these conditions is grassing. This occurs when the C4F8/SF6 ratio is unbalanced causing micromasking. This creates tiny pillars of un-etched Si and teflon on the surface of the feature. An example of grassing can be seen below.
- Generally 30 sec - 20 minutes is known to work well.
- Recipe 1 is known to work well up to 20 min of etching. The process is tested for 20 min monthly to verify performance. Longer etches, particularly >30 min will likely cause undesirable results.
- Etches longer than 20 minutes should use the second DRIE step for any time past 20 minutes. This removes the C4F8 from the etch step to help reduce grassing.
- Tested on features from 1 μm - 100 μm.
- The standard characterization pattern used to qualify Recipe 1 has open lines, trenches, and grating patterns from 1 μm - 100 μm. Larger and smaller features have not been tested and may not provide the expected results.
- Aspect ratio limitations vary depending on feature size. For instance, 2 μm features have an aspect ratio limit of around 50:1. For larger features, aspect ratio is typically limited to etch length (see above).
- Very large features and masks with total open area >20% are known to degrade etch performance and should be tested prior to running device wafers. For masks with more than 20% open area Recipe 3 may provide better results.
- Multi-step processing (i.e. running two or more etches with overlapping features) with the Bosch process is known to cause undesirable results. Please contact the tool engineers if you wish to run such a process so we can help reduce or eliminate issues.
Every third Friday Recipe 1 is run for 20 minutes on a Si wafer with our standard DRIE characterization mask, with 3 µm of SPR 220. This is performed after the weekly clean, and a 100 µm trench is measured in 5 locations with the Zygo NewView 5000. This etch rate is reported in the chart below.