[labnetwork] PlasmaTherm 720 and a selective SiO2 etch ?

[Jörg Hübner]

Hi,
We do not have a PlasmaTherm 720 RIE but have been into etching of SiO2 for some time using our STS machines. The principles are the same regardless the machine and the selectivity of 10 is though (and will most likely be dirty) so here are a few thoughts put together mostly by Prof. Henri Jansen:

As usual, recipes are very rare to be transferable between etch tools. Even if the tools are identical (I mean having the same name) tool history and actual use (contamination) is a source of "recipe drift". Of course, different brands will make the direct exchange of a recipe even more tricky.

At Danchip we use the advanced oxide etcher from S(P)TS with a dual source arrangement (ICP to create plasma species and CCP to extract the ions). As far as I remember, the PlasmaTherm 720 has a Reinberg reactor design with only a CCP source, so very different.

Having said the above, a way to go is not to copy a recipe, but to follow a procedure to find one, because this is independent of the tool. In general, the CF4+H2 plasma produces CF2 radicals that passivates surfaces and CFx+ ions that removes material. The ions are forced towards the substrate in an almost straight line by the dc self-bias as created by the CCP (also called RIE source) and as such the profile will be directional.

To enable sufficient selectivity between silicon and silicon dioxide, one should put the recipe exactly there where the fluorocarbon deposit (from the CF2 radicals) will protect the silicon while the SiO2 is still etching. The oxide will etch quicker than the silicon because it's oxygen content will "burn" the FC deposit and consequently the SiO2 surface will be free from the passivating film.

Typically, the difference in etch rate between Si and SiO2 is rather small and a ratio of 10 is tough and this will put the system into a position where a lot of FC deposit will start to "grow" on all surfaces, including the reactor walls. This contamination can be minimized using a dual source arrangement, such as Danchip's STS tool, because these tools can create more easily an ion-rich and radical-poor environment.

As far as my opinion counts, I would start with a certain setting where the plasma is "happy", i.e., a stable CF4 plasma. Then you change the hydrogen content until the silicon almost stops etching. If correct, at that particular setting the SiO2 etch will still be reasonable, even though it won't be very fast (several 10th of nm per minute).

As a final "advice", maybe you could contact PlasmaTherm as they will have sufficient field experience with respect to this tool. The 720 is a very common tool around the globe, so maybe they have a very reasonable starting point even though I believe that a selectivity above 10 cannot be provided directly as many people do not like the frequent cleaning procedure that goes with the fast FC growth.

Hope this story is helpful...

  Regards
                             Henri and Jørg


From: labnetwork-bounces at mtl.mit.edu <labnetwork-bounces at mtl.mit.edu> On Behalf Of James C. Sturm
Sent: 14. august 2018 22:43
To: labnetwork at mtl.mit.edu
Cc: Eric N. Mills <enmills at Princeton.EDU>
Subject: [labnetwork] PlasmaTherm 720 and a selective SiO2 etch ?

Hi, we have a PlasmaTherm 720 RIE etcher plumbed with CF4 and H2.

Does anyone have a recipe for selective etching of thermal SiO2 (stopping on silicon with selectivity > 10).

Right now the chamber has an Al feature plate (outside of the wafer) which we'd prefer to leave in if possible.

Thanks, Jim Sturm and Eric Mills, Princeton University
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mtl.mit.edu/pipermail/labnetwork/attachments/20180816/2af56de8/attachment.html>