Optimized plasma-assisted bi-layer photoresist fabrication protocol for high resolution microfabrication of thin-film metal electrodes on porous polymer membranes

Graphical abstract


Specification
Transfer the plasma treated glass substrates to a spin coater and spread PVA using a transfer pipette or syringe and then spin at 800 rpm for 30 s. 9 After spin coating of the PVA release layer, pre-cut PET membrane piece (slightly larger than the carrier substrate) are carefully place it on the carrier substrate.
Note: Higher molecular weight PVA is more viscous and might hinder the membrane release after fabrication.
Note: Try to avoid wrinkles (it helps to bend the membrane a little)once the membrane has been in contact with PVA it shouldn't be moved anymore! Note: Place the membrane onto the carrier while the PVA is still wet! 10 In order to dry the PVA on the substrates with the membrane attached, place it on a hotplate and ramp temperature to 150 C (the LOR3A resist needs to be baked at this temperature).
Note: If no hotplate with a ramping function is available, or the ramping is too time consumingthe samples can be also baked gradually using hotplate set to 70 C, 100 C, 120 and 150 C for 180 s each.
Note: If the samples are baked too fast the evaporating water will cause wrinkles on the membrane.
11 After dehydration cool the samples to room temperature and cut membrane pieces that are overlapping the carrier substrate using a scalpel Note: The temperature should be ramped (or as mentioned above baked gradually).
13 Once the LOR3A has been soft baked, AZ5214E resist (or a simple negative resist) is spin coated at 3000 rpm for 30 s and then soft baked at 100 C for 30 s. 14 Using a photo mask, transfer the desired electrode geometry to the sample by UV light (365 nm) exposure with a dose of 40 mJ/cm 2 . 15 After exposure, bake the sample at 120 C for 70 s and then flood expose (without photo mask) with a dose of 240 mJ/cm 2 . 16 Develop sample in AZ726MIF (TMAH based developer) for 120 s and rinse with diH 2 O.
Note: usually AZ5214E needs to be developed for 60 sbut TMAH dissolves LOR3A and allows for an undercut of the actual photo resist 17 Dry the samples (e.g. with Nitrogen spray gun, overnight in a desiccator). 18 Before depositing the metal layer, subject samples to an Argon plasma (50 W RF; 10 sccm Ar; 60 s), thereby modifying the parts of the membrane not covered by photoresist. (this can be done with a plasma asher or a Reactive Ion Etching System (RIE), we used a RIE because the power can be adjusted more precisely) 19 Deposit a gold layer of approximately 80 nm by sputtering (25 W, 2 Â 60 s sputter duration, base pressure: 2*10 -5 mbar, working pressure 8*10 -3 mbar) or evaporation.
Note: The sputter power should not exceed 25 W, otherwise the membrane might overheat, or the metal might crack or spall during lift-off because of strain/tensile forces.
Note: Deposition can be also done with an evaporation system Note: In case different metals are to be used, plasma treatment with a different gas species might improve adhesion (we found that O 2 plasma improved adhesion of Chromium) Note: Protocols for the fabrication of membrane integrated electrodes have been reported before [1], these protocols didn't include a plasma treatment prior to metal depositionwe found that this step improves adhesion of Chromium and Gold layers to the cell culture treated PET membranes supplied by it4ip, Belgium.
Note: In case other PET membranes from other suppliers are used, the plasma parameters might change.
20 After sputtering, soak the samples in N-methyl pyrrolidone or N-Ethyl pyrrolidone for 10 min and sonicate at low power to remove the photo resist and non-patterned gold. 21 Release the membrane by soaking the sample in diH 2 O for 1 min and then carefully pull it off with tweezers. 22 Using the process, gold electrodes can be deposited on porous membranes achieving a resolution down to 2.5 mm.
Note: This process can also be used to structure other metals (e.g.: copper, chromium, titanium), or combinations thereof.
23 When depositing metal combinations using sputtering, only a low sputtering power should be used to avoid spalling or cracking of the metals. 24 The PVA release layer allows rapid detachment of the membrane from its carrier and aids further processing.

Method validation
Several process parameters such as plasma power, type, duration and etc. were constantly evaluated, resulting in the optimized protocol presented above. Overall process parameters and results are summarized in Table 1. The gold microstructures fabricated on porous membranes achieved a resolution down to 2.5 mm. Fig. 2 shows a comparison of four parameter sets, Fig. 2A and B show thin-film microelectrodes fabricated using single-layer resist lift-off. The PET substrate in Fig. 2A was subjected to an O 2 plasma before Au deposition, while the substrate in Fig. 2B was subjected to an Ar plasma both with a power of 50 W and an oxygen flow rate of 10 sccm. Fig. 2C and D shows microelectrodes fabricated using a bi-layer resist fabrication approach, both substrates were subjected to Ar plasma with a power of 50 W. In case of the sample in Fig. 2C the Ar flow rate was 20 sccm while the flow rate in Fig. 2D was 10 sccm, resulting in better adhesion of Au to the substrate. Also, O 2 plasma at 10 sccm and 50 W could achieve results comparable to the best Ar treated samples down to a resolution of 2.5 mm only in the presence of a 5 nm chromium adhesion layer. As shown also in Table 1, any fabrication approaches that used only a single photoresist were inferior in terms of resolution to the optimized bi-layer photoresist fabrication methods. Also evident is the influence of high plasma and metal deposition powers resulting in structural and functional artifacts including thin-film delamination or cracking/ spalling of the metal layer. The improved performance of the presented optimized bi-layer lift-off fabrication protocol is not caused by physical etching known to increase surface area. For instance, Fig. 3 shows that plasma-treated PET membranes display similar surface roughness, thus area, in comparison to pristine non-modified samples. Additional phase images however revealed an increase in hydrophilicity due to incorporation of more hydrophilic surface groups, which significantly improved metal film adhesion.
To validate the electrode quality, frequency sweeps were recorded for Dulbecco's Minimal Essential Medium (DMEM) on 80 nm gold thin-film electrodes either fabricated on glass or PET track-etched membranes. The electrodes were connected to a VMP-3 Multichannel potentiostat (Bio-Logic Science Instruments, France) using spring contacts. Impedimetric measurements were performed with an excitation voltage of 50 mV and a frequency between 1 Hz and 500 kHz. Fig. 4A confirms that the frequency behavior on PET membranes is similar to microelectrodes  Fig. 4B shows that the presented membranebound high-resolution thin-film electrodes can be used in a state-of-the-art tetrapolar measurement setup frequently used for electric resistance evaluation of cell-based barrier models (transepithelial/-endothelial resistance, TEER) to eliminate the artificial resistance of the porous membrane. This can improve the overall sensor performance because industrial track-etched membranes are known to have high batch-to-batch variations due to mass production of track-etched membranes. As a final application, trans-epithelial resistance of Bewo placental epithelial cells at a seeding density of 100k cells/ cm 2 in DMEM supplemented with 10 % fetal calf serum and 1 % antibiotics mix was monitored using a tetrapolar TEER setup.
Thin-film gold electrodes on 3 mm porous PET membranes were compared to conventional Transwell 1 3 mm inserts tested with an EVOM2 Vohm meter equipped with an STX3 Ag/AgCl electrode. [3] Fig. 4C shows that membrane-bound thin-film electrodes can be readily used in any TEER or impedance-based measurement setup and can monitor cell barrier dynamics over several days. In summary, we have optimized a metal deposition method for the fabrication of high-resolution microstructures down to 2.5 mm on porous PET membranes. This improved resolution in combination of using ultra-thin porous membranes enables the development of novel microfluidic devices in addition to other applications, which require conformal electrodes with excellent resolution and high surface adhesion.