Why Chip Fabs Can’t Keep Yield and EHS Systems Apart Much Longer
China, Japan, Taiwan, Malaysia, and others Must Prepare
DPP, CBAM, EU Regulation
For twenty years, the fab floor has run on a simple division of labor. Yield Management Systems (YMS) watch the wafers — chamber pressure, defect density, tool drift. Environmental Health and Safety (EHS) systems watch the plant — water draw, gas venting, waste streams. The two rarely talk to each other, and until recently, nobody needed them to. A scrapped lot was a yield problem. A permit filing was an EHS problem. Different dashboards, different owners, different meetings.
That division is getting expensive, and not because of an internal sustainability push. It’s because Brussels soon put a price tag on the gap between them. Fabs must be smart, safe, and sustainable.
The fix isn’t a new dashboard
The fix that’s actually being piloted — by equipment makers and by outfits like PDF Solutions and imec exploring cryptographically signed, wafer-level carbon footprints — is to fuse YMS defect telemetry with EHS resource telemetry at the point of generation.
Every time a lot is scrapped mid-process, you’ve already burned the ultra-pure water, the specialty gas, and the grid power that went into it up to that step. That’s a carbon cost with no die to show for it. A fab that tracks yield and carbon in the same data model can start optimizing for both instead of discovering the tradeoff at quarter-end.
This is also, not coincidentally, the fifth and sixth layer of a six-layer Safe and Sustainable by Design (SSbD) architecture we’ve been developing for the IEEE ICE conference (Ang and Liao 2026) — the layers where verified fab-floor data has to surface as something a customs declaration or an ESG auditor can actually use. (The trust mechanics that make that data verifiable without leaking proprietary recipes are their own separate problem, one we tackle in a companion paper on zero-trust industrial data spaces (Liao and Ang 2026).)
It plays out differently by region
A fab that can’t produce verifiable, lot-level carbon and chemical data doesn’t get a discount for good intentions. The exposure today runs through the CBAM goods it buys, not the chips it makes — steel, aluminum, and electricity — and for those inputs, no verified data means falling back to CBAM’s default emissions values, calculated on the assumption that the producer is among the dirtiest in its category.
As CBAM’s product list extends further downstream, that same logic becomes a direct risk for the fab’s own output, not just its supply chain.
Effectively, the regulation assumes the worst about the fab’s carbon footprint unless someone proves otherwise.
And this isn’t one problem with one shape:
- Taiwan carries the sharpest Scope 2 exposure — EUV lithography is power-hungry, and the grid behind it is still carbon-heavy.
- Japan’s risk sits upstream, in fluorinated photoresist and etch chemistry that PFAS rules are squeezing hardest.
- China’s legacy-node fabs face a different threat: without verified telemetry, they default to CBAM’s worst-case penalty benchmark rather than their actual footprint.
- Malaysia’s OSAT (Outsourced Semiconductor Assembly and Test) hubs are exposed through packaging inputs — copper leadframes, bonding wire — as CBAM’s raw-material scope creeps toward metals.
Four very different fabs, one shared fix: make the carbon and chemical data trustworthy enough, and specific enough to a wafer lot, that it survives contact with a customs official.
Where this is headed
We’re developing the fuller version of this argument — with the regional trade-exposure analysis worked through in more depth — as a short paper for a journal on Green Technologies or Sustainability. If you work in fab operations, EHS, or trade compliance and this split between yield and environmental systems sounds familiar, we’d like to hear how you’re bridging it, or whether you’ve concluded it isn’t worth bridging yet.