Thermal Stress and NPN Tent Transfer

ByGreg Madejski

From my last two posts (Anodic Bonding, Ozone Bonding), it is clear that we have some sort of bonding going on between the freestanding membrane and the posts intended to support the membrane. However, we do not understand the nature of the transfer. To gain more insight into this process, I took more SEMs of freestanding membranes that have transferred over to the oxide spacer layer. Previously, I used ozone bonding to get bring the membranes into contact, where the highly charged surfaces will draw together into contact. The following SEMs are of my attempts using the Anodic Bonding setup (1 hr, 500 C hotplate, ~220 C surface Temp, 700 V). I previously observed that the freestanding regions transfered over, but there was no bulk bonding between the two surfaces. These are all 45 degree views

Anodic_bonded_00
Macroscopic View, showing a region where the NPN has bonded inside the channel (oxide spacer substrate, 100nm thick)
Anodic_bonded_02
The NPN membrane transfer has produced uneven bonding. The ‘capillaries’ formed by the wrinkled NPN form a network. Could heating the chips so much cause the contraction of a broken membrane over the posts?
Anodic_bonded_04
Some of the tubes between the pillars are pinched off, while others maintain a large branching network
Anodic_bonded_06
Some of the wrinkles/tubes make it to the edge
Anodic_bonded_10
The large nanopore sensor region was destroyed on this chip.

 

To test whether it was the heating or the voltages producing these wrinkled structures, I heated some chips on a hot plate (400 C, 1 hr, 7-17 Pa contact pressure). Membranes were transferred over much like in the anodic bonding and ozone bonding treatments. These images are all flat.

Heat_bonded_14
Large amount of wrinkling observed. This is just with heat and contact pressure.
Heat_bonded_24
If the tented areas can be wetted, we have an intact nitride membrane that can be electro-drilled!
Heat_bonded_20
Nanoporous nitride still has open pores

Heat_bonded_18

So it appears the heat alone can cause the membrane to shrink wrap. I was curious to see if a larger tented scaffold could transfer and keep the tent open. I used my old BBB-chip and anodically bonded it to a coverslip. This chip has a 50 nm nitride hex layer on top of a pnc-si membrane. I put the nitride grid in contact with the coverglass (1000 V, 500 C hotplate, 40 minutes), and watched the membrane delaminate from one side of the chip to the other (unzipped). Here are the SEMs of the transfer:

 

BBB_COVERSLIP_00
Large freestanding membrane pattern (3 mm length) transferred to coverglass. Charging is an issue for this film stack, so lower voltages were used to get images.
BBB_COVERSLIP_02
Upon further inspection, the hexes are nice and flat, but the nanomembrane has broken away. In a previous post (MgF2 fabrication on pnc-si), I observed that film stress will cause the membranes to buckle into crescent moon shapes. Here it appears that the chips are more solidly bonded. I wonder if the cooling causes membrane to contract from the hex grid equally, explaining some of the rips.
BBB_COVERSLIP_05
The pnc-si is draped over the nitride hex grid (like a post). Which comes into contact and bonds first, the hex grid or the pnc-si?

 

Based on these images, we should consider lower temperature bonding, with some mild contact pressure, to minimize the heating and shrinking of these layers. I believe if we can make the film stack with higher density post spacing, we should be able to suspend the nanomembranes over any smooth surface. The major question is whether or not the space inside the tent can be wetted and have the membrane remain intact.

 

Similar Posts