Feasability of SU8 3010 Negative Resist for Lift-off.

SU-8 3010 (MicroChem) was chosen for its improved adhesion strength, and reduced stress. MicroChem also provides a datasheet on the usage of this negative resist that was followed as a baseline. For Post Exposure Bake (PEB) two methods were implemented, standard procedure, and part of Li, et al process: 55C for 2 hours to reduce stress.

Before attempting an SU8 structure on a membrane, it was tested on oxide and nitride for bond strength and stress/shrinkage after lift-off. Once the lithography steps were complete, the wafer pieces were then placed in 10:1 BOE. The oxide under the SU8 was 30 nm thick, and the etchant had to travel at least 12.5 microns, at a rate of 0.1 um/min, 125 minutes is the minimum. Once the membrane could be removed from the wafer, it was remeasured to determine the shrinkage across 7 lines and spaces.

A single fragment of each PEB method was then placed in 10:1 BOE for 3 hours. After the etch, the fragment was placed in water and the film could be peeled off with tweezers. This experiment determines the effect of solvent evaporation on the total size of the freestanding support structure after lift-off. The results from measuring the lines and spaces on Si showed that both PEB methods provided the same results before and after lift-off (see Figure 4). The 7 line-space measurements on SiN showed a decrease of ~9 microns with the 95C PEB. The 55C PEB produced a reduction of ~1um after lift-off as compared to the SU-8 still on the SiN.

On Si, the SU-8 bond was strong, no lift-off or weak sections before the BOE. SiN did not share the strong adhesion. The SU-8 on SiN started to peel in sections after developing, this was easy to scratch off the SiN. In an attempt to improve the bond, the remaining structure was hard baked at 200C for 3 minutes (an additional recommendation from MicroChem). This did improve the bond strength of the SU-8 on SiN. Later examination of the results and process revealed a missed note in the MicroChem process: when exposing on nitride, the dose should be increased by 40%. This was not done and could have caused the low bond strength seen here due to partial cross-linking of the SU-8.

Ideally, the BOE would not affect the SU8 on SiN, and would eventually lift-off from the Si wafer with thermal oxide. This has been confirmed for the oxide, but nitride requires more testing. Given that the SU8 on SiN is fully cross-linked, they do not separate. This will have to be reconfirmed after the extended exposure time.
Samples from all test conditions have been placed in water and PBS at body temperature (37C) for the past two weeks to mimic the environment the membranes would be used in. None of the samples have shown any signs of delamination, which further verifies the adhesion strength of SU8 3000 on substrates of interest.