Nicolas,

Some nice explanation of RIE process you have there. But sorry just want
to clarify some points you made there. Actually reactive ion etching is
not a technically correct term to describe this etching process because
ions in capacitively coupled 13.56MHz rf plasma constitute typically less
than 0.1% of total plasma species in the chamber. So etching should not be
significantly caused by "reactive ions." If I remember correctly, the term
was coined before researchers understood what was really going on in an
RIE process, thus leaving such incorrect name. Ion bombardment "cleans"
the surface of passivating species, like SxFy and CxFy, as well as
amorphizing the silicon network to enhance chemisorption of reactive
neutral species, such as F, onto the surface for chemical reactions.
Surface silicon atoms seldomly left dangling for a long period of time.
Thus without this physical ion bombardment, F adsoption rate on Si surface
is very low, even though the raw Si + F reaction is very exothermic.

Back to that question, increasing rf power and cycling frequency will help
but more reactive species are also generated if rf power, not ICP, is the
means to create and sustain the plasma. I read a paper, Journal of Vacuum
Science and Technology B vo.17 no.6 pp.2768, detailing a bosch-type
process using SF6/Ar and CF3/Ar gas mixtures. Ar provides the main source
of ion bombardment. It is definitely worth a read.

Regards,

Isaac


On Mon, 11 Sep 2006, Nicolas Duarte wrote:

> The sidewall scallops come from the chemical nature of the Bosch
> process.  Specifically a Bosch process involves anisotropic etching
> (SF6), followed by a placement of a passivation layer on all surfaces
> (from C4F8), and repeat.  Some choose to add an oxygen plasma step to
> help removing the polymer that deposited on the etch surface.  Note
> that there is no such thing as an RIE that is just chemical or just
> physical...every RIE process is a combination of both, hence the name
> Reactive Ion Etching.  If it is just physical it is called Ion
> Milling and if it is just chemical then your wasting power creating a
> plasma.
>
> The reason why the Bosch process is used is that no process is 100%
> anisotropic.  For various reasons (ion bowing as an example) the etch
> part of the Bosch process (or any plasma etch left long enough for
> that matter) will cause a sort of balloon etch.  Once the passivation
> layer is placed on the walls, this balloon hole will not open up any
> more, instead the bottom of the hole will go deeper during the next
> etch phase and create a the balloon feature below it.  It is hard to
> explain without drawings but most any MEMS textbook will explain it
> in greater detail.  Also a quick search on google and I found this
> page which has some good drawings to explain scalloping and notching
> http://cmi.epfl.ch/etch/Talk_Cyrille_CMI2004.pdf.
>
> So as to why the solutions I suggested:
>
> First off higher substrate power will pull the ions down more
> vertically, giving them better directionality and less of the balloon
> effect.  Too much however and you start getting the balloon back
> since the ions will just start milling and shooting the byproducts
> back out at random directions.  With higher power also comes more
> speed, more roughness at the bottom of the pit, and it is harder to
> control etch depth.
>
> As for the shorter cycles, it will make smaller balloons before
> moving deeper.  The smaller the balloon, the smaller the scallops.
> You will also get a much slower etch and probably use up more of your
> reactant gas.