This is pretty close, but the other part of the mechanism is that, in an
anodic bond, the final bond reaction is a chemical one - specifically
oxidation.

Basically, removal of the sodium ions has one specific result - a large
number of ionic (negatively charged) oxygen species that no longer have
a positive ion to balance them.  The next best thing - some sort of
metallic or semi-metal species that can be oxidized.  The oxygen will
latch onto this material (be it silicon, iron - as in KOVAR, gallium -
as in GaAs, or other material) and form a covalent bond.

So the two primary material ingredients for an anodic bond are:

1) A glass containing a positively charged ionic species (for example,
sodium).

2) An oxidizable material.

So if a layer of oxidizable material is sputtered onto one of the glass
pieces, this becomes possible.

Best Regards,
Chad Brubaker

EV Group       invent * innovate * implement
Technology - Tel:  480.727.9635, Fax:  480.727.9700  e-mail:
[email protected], www.EVGroup.com

-----Original Message-----
From: [email protected]
[mailto:[email protected]] On Behalf Of Joseph Grogan
Sent: Wednesday, November 08, 2006 4:52 PM
To: General MEMS discussion
Subject: Re: [mems-talk] Glass to glass bonding

You cannot anodically bond glass to glass with SiO2 in between them. The

way anodic bonding works is that the glass is heated to the point where
alkali ions (usually sodium) become mobile. When you apply a large
voltage (maybe ~1000V) the ions are pulled to the cathode (negative
side) leaving a depleted negative charge on the glass surface which then

gets attracted down to the other substrate (traditionally silicon) and
bonds. To bond glass to glass, you need to put a diffusion barrier in
between the glass, so that the top glass has something to reach down and

bond to, otherwise the sodium from the bottom glass just gets pulled up
into the top glass and no bonding occurs. SiO2 (glass) does not serve as

a diffusion barrier for sodium since the process is based on sodium
being mobile in glass at high temperature. For a good list of usable
films as diffusion barriers check this paper:
"Glass-to-glass anodic bonding with standard IC technology thin films as

intermediate layers" Berthod et al, Sensors and Actuators, 2000.

I've never tried doing it, but I don't believe it's possible to bond
quartz to quartz because there are no mobile ions to migrate around. You

might be able to bond pyrex to quartz since the quartz has no ions to
leech across the gap, however, you need to make sure that you have
matched thermal expansion coefficients otherwise the thing will bond and

then shatter when it cools.

PDMS bonding is a finiky art if you've never done it before. The key is
low power plasma for short time. I believe the reason is that long time
high power forms a glassy layer on the PDMS which prevents bonding. I
found this paper to be particularly helpful for settings and cleaning
procedures: "Three-Dimensional Micro-Channel Fabrication in
Polydimethylsiloxane(PDMS) Elastomer" Byung-Ho Jo, Journal of
Microelectromechanical Systems, Vol 9. No 1, 2000.

good luck,
Joe Grogan