Source: arXiv

Paper

Scoping Review of AI, Metrology, and ESG in the Semiconductor Sector: Implications for Safe and Sustainable by Design (SSbD) — Ang, Liao (2026; IEEE ICE/ITMC proceedings framing)
ID: arXiv:2607.23082

What it claims

Chipmakers are being pulled in two directions at once: use AI to run manufacturing tighter, and prove sustainability under rules such as the EU Carbon Border Adjustment Mechanism (CBAM). The authors scope 1,465 Web of Science/Scopus documents across AI-integrated metrology, supply-chain ESG, and federated industrial data spaces. Network analysis, they argue, shows a fragmented core–periphery map with a structural hole between AI process optimization and downstream sustainability governance.

Their answer is a six-layer Safe and Sustainable by Design (SSbD) architecture in a System-of-Systems frame, with “grid-to-core” and “standards-through-supply-chain” pathways. Virtual metrology (predicting wafer quality from sensor streams without extra destructive tests), localized federated learning (training models across fabs without centralizing sensitive lot data), and defensive RegTech are offered as ways to build provenance-aware data fabrics so compliance can drive innovation rather than only paperwork.

The breakdown

This is not a Korea case study, but it is a Korea-touching industrial map: memory and foundry supply chains that Korea anchors sit exactly where CBAM-style carbon accounting and AI fab tooling collide. The paper’s useful split is metrology/AI on the line versus ESG data that buyers and regulators can trust downstream. If those literatures barely connect, exporters and customers will keep buying “AI efficiency” stories that do not travel into customs or Scope-3 questionnaires — a gap that matters when Seoul-linked fabs sell into EU buyers already asking for carbon footprints at the border.

Why readers outside the lab should care

If you buy Korean HBM/DRAM, audit a supplier for EU market access, or design RegTech for chip logistics, the structural-hole claim is the overseas hook. Korea’s fabs can be technically excellent while still failing a buyer’s CBAM or ESG evidence pack if provenance data never leaves the metrology silo.

Procurement teams that only ask for “AI-driven yield” slides will miss whether carbon and substance data can travel with the lot. The paper’s value is naming that missing bridge before a customs or customer questionnaire does.

What travelers and expats should watch

• Do ask suppliers how virtual metrology or AI process data links to exportable ESG evidence — not only yield slides.
• Do treat CBAM/SSbD language in RFPs as a data-architecture problem, not a CSR brochure ask.
• Don’t assume a Korea plant tour that shows AI dashboards answers carbon-border documentation.
• Expect the paper’s six-layer model to be a proposal, not an adopted Korean standard — use it as a checklist, not law.

Context

Use the six-layer sketch as a question list for supplier calls: where does metrology data become exportable evidence?

Read this as a map of a missing bridge between fab AI and trade compliance, not as proof any named Korean firm is non-compliant. Korelay frame: for Korea-touching chip buyers, the behavior change is to demand provenance paths that survive the factory gate.

Source

arXiv:2607.23082 — abstract framing cited; open the OA PDF for network figures and full architecture. Do not republish the PDF.