Pgmea

In order to generate water-in-oil droplets, a soft photolithographic process was employed to fabricate the microfluidic devices used. The height and width of the device used to produce the tubular microgels were 50 and $100\mu \text{m}$ respectively. In brief, a $50\mu \text{m}$ thick photoresist layer (SU-8 3050, MicroChem) was spin-coated onto a silicon wafer and soft-baked for 15 min at $95{}^{\circ }\text{C}$ . The photo-mask was placed onto the wafer, which in turn was exposed to UV light. This was postbaked for 5 min at $95{}^{\circ }\text{C}$ . In order to remove excess photoresist, the master was developed in propylene glycol methyl ether acetate (PGMEA, Sigma-Aldrich). For the double emulsion experiments, a two-step photolithographic process was used^{21 (link)}.
In order to fabricate microfluidic devices, a 10:1 elastomer PDMS to curing agent (SYLGARD 184, Dow Corning, Midland, MI) mixture was used. This was cured for 3 h at $65{}^o\text{C}$ . The hardened PDMS was cut, peeled off the master and holes of 0.75 mm were punched into the PDMS. This was then bonded onto a glass slide by treating with a plasma bonder (Diener Electronic, Ebhausen, Germany).

The precursor solutions for producing MoS₂ were synthesized using our recently developed approach^{10 (link)}. A 1 M sulfur solution was prepared by dissolving S (Alfa Aesar, Fisher Scientific) in carbon disulfide (CS₂, Yakuri Pure Chemicals Co., Ltd). The precursor solution was obtained by dissolving ammonium tetrathiomolybdate ((NH₄)₂MoS₄, 99.97%, Sigma Aldrich) in 4 parts of ethanolamine (Sigma Aldrich) and 4 parts of butylamine (Sigma Aldrich) with the S solution. Then, 2 parts of n,n dimethylformamide (DMF, Sigma Aldrich) were added to the solution to form the precursor solution. In our CVD-free method, S was added during solution preparation to yield an S-rich precursor instead of adding S powder separately, as proposed in other CVD methods.
The silver paste was formulated by mixing 100 parts of original Ag paste (4000 cps, AD-V7-108) with 1 part of Silveray (solvent) and 3 parts of propylene glycol methyl ether acetate (PGMEA, Sigma Aldrich). This was done to make an even paste and prevent clogging at the tip capillarity during the printing process according to our recent publication^{13 (link)}. We modified it in order to be relevant to this research.

PDMS moulds for hot embossing and PDMS chips were fabricated similarly by casting PDMS on SU-8 (2005 and 2050, MicroChem Corp., USA) master moulds as previously described [32 (link)]. Briefly, the master moulds were obtained by spin-coating thin layers of SU-8 (5–100 µm) on silicon wafers using a spincoater (Laurell Technologies, USA) according to manufacturer's instructions. Microstructures were then generated using high-resolution printed masks (JD phototools, UK) and 365 nm UV exposure (UV KUB 2, Kloe, France) followed by PGMEA (Sigma-Aldrich) development. The master moulds were then treated with trichloromethyl-silane (Acros Organics, Belgium) by vapour deposition (room temperature overnight under vacuum). PDMS moulds for hot embossing were produced by casting a 5 : 1 PDMS mixture on these SU-8 master moulds and curing at 75°C for 30 min followed by an additional baking step (200°C for 1 h). These moulds were used for direct hot embossing of COC at high temperatures (up to 200°C), accommodating pressures up to 1 tonne without significant deformation of the structures. PDMS chips were fabricated using a similar protocol, but with a 10 : 1 PDMS mixture and a single-step curing at 75°C for 1 h. For the mounting of closed PDMS devices, both chips and coverslips were plasma-treated for 1 min (Harrick Plasma), allowing direct and reliable bonding.

The multi-level microchannel was designed in AutoCAD (AutoDesk, San Rafael, CA, USA) and exported to GDS format for maskless lithography using KLayout. The design consists of three layers: the first alignment mask layer, the flow channel layer, and the growth channel layer (Figure 1). The flow channel layer (for nutrient delivery and bacterial removal) was 15 $\mathsf{\mu }$ m in depth and the growth channel layer (for bacterial entrapment, outgrowth, and long-term imaging) was 1 $\mathsf{\mu }$ m in depth.
To create the microstructure mold, we developed a lithography approach using a single negative photoresist (mr-DWL-5, Micro Resist Technology GmbH, Berlin, Germany) and a maskless writer (DL-1000, Nanosystem Solutions, Okinawa, Japan). The hard baking temperature and exposure dosage protocol was based on the manufacture instructions. Briefly, after lithographic exposure, the wafer was hard baked and developed in propylene glycol methyl ether acetate (PGMEA, Sigma-Aldrich, Burlington, MA, USA), washed thoroughly with isopropanol, and dried with nitrogen air. The true depth of the microstructures was confirmed with a stylus profilometer (DektakXT, Bruker, Billerica, MA, USA).