Anodic Bonding Setup at U of R

ByGreg Madejski

Previous Posts Anodic Bonding Theory, Anodic Bonding at RIT

I’ve put together an anodic bonder setup here at the U of R and will be moving it inside the cleanroom shortly.

660px-B-a-schemeanodicbondprocess

 

IMG_2684
Voltage controlled Voltage supply (Left + Center) and Hotplate with bonding plate (Right). Red + white wires (0-12 V <1.2A) Green and Black Wires (0-700 V, <10 mA)
EMCO Voltage Curve
EMCO Voltage Curve

 

I tried a couple of known combinations to see if the bonder was working properly. The general principle is to liberate sodium ions from glass (move to anode), and allow oxygen to move to the interface (move to cathode), reacting at the interface with the other substrate’s silicon material to make silicon oxide, forming a bond. Ions in the non-glass substrate must be prevented from leaching out into the glass; the high electric field must be maintained at the interface to promote the oxidation.

Si-Coverslip, single interface (400 C, 30 minutes, ~500 V)

IMG_2679
~15 minutes, glass to silicon has a bond.
IMG_2680
~20 min. The bond front expands across the interface, squeezing out interstitial air.
IMG_2683
~25 min. The bond front has worked around the channels.
IMG_2685
The backside silicon was bonded to the glass, but the pnc-si did not as efficiently. I believe this to be a matter of good contact rather than inability of bonding. The etched channel material was fused to the glass, even though the bulk chip was not fused.

 

Silicon Nitride – Coverslip (400 C, 30 minutes, ~650V)

IMG_2687
10 minutes.
IMG_2688
30 minutes. The bonding is proceeding a lot slower than the previous case.

 

Everything seems to be working appropriately. Now we will have to see if there is enough impurity to make the bond work for the flowchip/oxide channel combination.

Similar Posts