CRYSTALSIM

initializing lattice

§ /for/manufacturers

Process Compatibility for Crystal-EM Transistors

Where the Crystal-EM thesis intersects existing CMOS, GaN, and BEOL flows — what's drop-in, what needs new tooling, and which insertion points are realistic in a 3-year horizon.

§ compatibility matrix
ComponentSi CMOSGaN-on-SiBEOLSpecial tooling
GaN channelpartialyesnoMOCVD
PZT gatethermal budget issuepossibleyesSol-gel / PLD
EM coupleryesyesyesStandard metal
HfO₂ bufferyesyesyesALD
TiN gateyesyesyesPVD / ALD
§ tooling cost vs EUV
TechnologyCapExReference
EUV lithography (3 nm node)$180–250M / scannerASML NXE:3600D
MOCVD GaN-on-Si$8–15M / reactorVeeco / Aixtron
PZT sol-gel + RTA$1–3M / lineExisting thin-film tools
Standard BEOL metalamortizedAlready on every fab floor

Crystal-EM tooling reuses existing GaN/thin-film lines. No EUV-class capital required for the analog/RF insertion path.

§ realistic insertion roadmap
Year 1
Low — uses existing GaN-on-Si lines
Analog / RF discrete (GaN HEMT + EM coupler)
Year 2–3
Medium — thermal budget integration
Mixed-signal (PZT-gated SoC blocks)
Year 4–6
High — yield + density open
Digital logic insertion (selected blocks)
Year 7+
Speculative — requires Phase-3 prototype data
Full CMOS replacement in target nodes
§ open questions for foundry partners
  • 1. What is the maximum thermal budget your line tolerates after M1 metallisation? (PZT crystallisation needs 600–700 °C RTA.)
  • 2. Is your GaN-on-Si line qualified for < 200 nm gate length, or limited to power-electronics geometries?
  • 3. Do you have BEOL ferroelectric experience (HfZrO, PZT, or similar)? At what wafer size?
  • 4. Acceptable defect density target for an analog/RF prototype run (parts/cm²)?
  • 5. Is there interest in a multi-project wafer (MPW) shuttle for a 4–8-mask test chip?
§ foundry inquiry

Talk to the project lead

NDA-friendly. Replies within 5 business days. The form below opens your mail client with a pre-filled template addressed to research@crystalsim.io.

Or open mail client directly