Youngjun Kweon / Cleanroom
Cleanroom · UCSD Nano3

Cleanroom Fabrication

The GFET process flow I run in the Qualcomm Institute Nano3 cleanroom, written up step by step with the tools, recipes, and the physics and chemistry behind each process.

Nano3 cleanroom facility ↗GFET device physics →

Process traveler

Each step links to a full write-up with parameters, recipe, mechanism, and tool anatomy.

# Process Tool Write-up
01 Substrate clean (SC-1 / RCA-1) Wet bench Substrate preparation
02 Resist coat and develop AZ 1529, LOR5B / AZ 1518, AZ MIF 300 Photoresist and surface profile
03 Lithography Heidelberg MLA Photolithography
04 Cr/Au contacts and lift-off Temescal BJD-1800 e-beam E-beam evaporation and lift-off
05 SiO₂ passivation Oxford PlasmaLab PECVD PECVD oxide passivation
06 SiO₂ etch Trion ICP-RIE Reactive ion etching

After these steps come graphene transfer (APS Cu etch with PMMA support), wire bonding to a custom PCB, and electrical testing on a probe station.

Failure modes and solutions

Two fabrication problems I diagnosed and fixed while building back-gated GFETs.

Failure mode 1: Resist edge bead

ProblemAn edge bead raised the wafer edge, so the maskless aligner couldn't find focus and patterns came out misaligned.

(a) Aligner error: high edge detected.
(a) Aligner error: high edge detected.
(b) Visible edge bead on the wafer.
(b) Visible edge bead on the wafer.
(c) Misalignment after lithography and development.
(c) Misalignment after lithography and development.
(d) Alignment after edge-bead removal.
(d) Alignment after edge-bead removal.

FixA two-step spin coat (slow spread, then fast spin to target thickness), followed by edge-resist removal with acetone, then IPA.

ResultThe aligner focused normally and patterns were well aligned after development (d).

Failure mode 2: Resist residue after graphene transfer

ProblemThick single-layer AZ 1529 couldn't be stripped cleanly from the graphene channel, and sonication wasn't an option because it damages graphene.

(a) Resist residue after a removal attempt (AZ 1529).
(a) Resist residue after a removal attempt (AZ 1529).
(b) Clean graphene channel (LOR5B / AZ 1518).
(b) Clean graphene channel (LOR5B / AZ 1518).

FixSwitched to an LOR5B / AZ 1518 bilayer. The LOR undercut lets the stack lift away in PG Remover without sonication.

ResultComplete resist removal after graphene patterning, with no visible residue in the channel (b).

Equipment I operate

  • Photoresist spin coating (AZ positive photoresist series)
  • Maskless lithography (Heidelberg MLA)
  • Photoresist development (AZ MIF 300)
  • Reactive ion etching (Trion ICP-RIE)
  • E-beam evaporation (Temescal)
  • Plasma-enhanced CVD (Oxford PlasmaLab)
  • Atomic layer deposition (Beneq ALD)
  • Surface profilometry (Bruker Dektak XT surface profiler)
  • Wet etching (Cu foil etching in ammonium persulfate)