[mems-talk] proper glass-glass bonding

[Gareth Jenkins]

Hi

I doubt companies have any special tricks which aren't already out there 
in the literature but they probably have more rigorous process 
development and better facilities to obtain better reliability. In 
particular they probably invest in commercial bonding systems rather 
than use a furnace which may suffer from poor uniformity.
Flatness of your wafers and thorough pre-treatment are very important. 
You can also use a grid channel network around your structures to 
improve air circulation and minimise the risk of voids. Minimise the 
area you use (i.e. don't expect to be able to bond a full size wafer and 
then dice multiple structures afterwards).
There are some low temperature glass bonding techniques which may be 
worth a look. One involves dilute HF to activate the surface.
Part of the trick may be getting a good "pre-bond". If you have flat 
clean wafers, they will stick together tightly forming a vacuum bond. 
Then put them in the bonding oven.
As Isaac says, there is no one "magic pill" but if you have adequate 
tools and are prepared to spend time on rigorous development then you 
should be able to achieve a reproducible protocol.
Best regards

Gareth

Mikael Evander wrote:
> Hello.
>
> I'm familiar with the "standard" glass-glass thermal fusion bonding method where you wash your glass wafers, preferable surface activate them with O2-plasma or such-like and thermo-cycle them in an oven a bit over Tg with some pressure on the wafers. As far as I know most people using this get a pretty low bonding efficiency, some voids and usually not the nice "mirror-finish" on the glass that you usually see before you bond.
>
> I'm also aware that the companies doing glass-structuring and bonding commercially must have, and have, much better bonding efficiencies and incredably good looking bonding. 
>
> I find it kind of surprising that whatever tips and tricks are being used in getting at least a slightly better bond hasn't seeped out from the industry. As far as I know at least ...
>
> Does anyone have any nice tips and tricks to share that would help me, and hopefully lot of other people, to get more efficient bonding?
>
>