Scoping Review of AI, Metrology, and ESG in the Semiconductor Sector

Implications for Safe and Sustainable by Design (SSbD)

Karen Ang & Han-Teng Liao

2026-06-11

Scoping Review of AI, Metrology, and
ESG in the Semiconductor Sector

Implications for Safe and Sustainable by Design (SSbD)

• RS-MI-2B: AI, Generative AI & Digital Transformation, ICE 32nd Ed.

• Tuesday, 23/June/2026: 3:00pm - 4:20pm; Location: Room Miragaia

Karen Ang 0009-0008-5923-0106
Infineon Technologies

Han-Teng Liao 0000-0003-1081-5599
Independent Researcher

Welcome everyone. My name is Karen Ang, and I am a Senior Manager at Infineon Technologies based in Malaysia. Joining me is Han-Teng Liao, an independent researcher specializing in sustainable manufacturing. Today we will present our scoping review on the intersection of AI, metrology, and ESG in the semiconductor industry, with a focus on implications for Safe and Sustainable by Design, or SSbD.

🌱 Section I: The Semiconductor Twin Transition

• Precision Scaling: Rapidly advancing sub-5 nm processing nodes.
• Sustainability Mandates: Regulatory pressure from EU CBAM and CSDDD.
• Systemic Volatility: Geopolitical friction and historical supply chain disruptions.

Caution

Screenshot (not in the paper): Source

Semiconductor Business Connect 2026

Today, with the SSbD both as EU policy and emerging design concept, we look at the semiconductor industry’s Twin Transition. Manufacturers must scale advanced processing nodes using AI while simultaneously meeting strict environmental mandates like the EU’s Carbon Border Adjustment Mechanism. This tension unfolds inside a highly volatile geopolitical and supply chain landscape.

🌱 I.A Motivation: GRC Landscapes & SSbD

• GRC: Governance, Risk, and Compliance alignment strategies.
• SSbD Framework: Safe and Sustainable by Design material lifecycle safety.
• UN SDGs: Integrating Goal 9 (Innovation) and Goal 12 (Responsible Production).

JRC SSbD framework

Note. The above image is not in the paper but sourced directly from the EC JRC site.

To navigate this volatility, firms are turning to Governance, Risk, and Compliance strategies. The European Union’s Safe and Sustainable by Design framework is driving this shift. It forces companies to move from basic administrative accounting to science-based, empirical proof of sustainability throughout the product lifecycle.

🌱 I.B Objective. Current RegTech Service Ecosystem

Platform	Core Functionality	Standards Alignment	CBAM Relevance
EcoVadis	Supplier ESG ratings	GRI, ISO 26000, SBTi	Strong procurement
IntegrityNext	Compliance automation	SAP Ariba integration	Risk detection
Workiva	Audit-ready compliance	CSRD, ESRS, GRI	Disclosure potential
AuditBoard	Risk management	SOC 2, ISO, NIST	Governance audit
SAP Ariba	Procurement	ISO	Procurement readiness

Note: Data from Table I.

How can a scoping review inform the SSbD-aligned architectures integrating AI, metrology, and ESG compliance in semiconductor manufacturing?

Objectives:

• 1 – Integration
• 2 – Constraints and Gaps
• 3 - System-Level Efficiency and Assurance

As shown in Table 1, current RegTech platforms like EcoVadis and IntegrityNext automate supplier compliance. However, these tools operate as discrete silos. They are focused heavily on corporate procurement and reporting, but remain completely disconnected from actual semiconductor factory-floor metrology and manufacturing process data. This motivates our research with objectives outlined for real-time manufacturing processes.

📚 Section II: Literature Review

Ecosystem Gaps & The SoS Challenge

• Industry 3.5 to 4.0: Phased migrations over abrupt transitions.
• Green Digital Transformation: Linking natural, business, and digital assets.
• System of Systems (SoS): Integrating fragmented technical disciplines.

Our literature review highlights a massive architectural gap. While models for phased industrial migrations exist, process-level data does not propagate down to regulatory platforms. We argue that metrology, AI analytics, and ESG compliance must be treated as interdependent services within a unified System of Systems framework.

🧩 Section III: Methodology

🧩 III-A: Research Questions

• RQ1: Data Interoperability: How can semiconductor analytics and measurement systems integrate heterogeneous data and generalize across customers and process nodes while remaining interoperable with ESG frameworks, including through federated data space architecture integration?
• RQ2: Scarcity Adaptation: How can analytics and measurement platforms be designed and tuned for scarce or imbalanced data while maintaining reliability for critical defect classes, leveraging soft-sensing, VM, and outsourced pipelines for better resource allocation, risk management, and lifecycle assurance?
• RQ3: Eco-Efficiency: How can industrial analytics improve yield and throughput while preserving performance for underrepresented classes and aligning with AI safety and sustainability mandates (e.g., Scope 3, CBAM) – framing optimization as a SoS challenge rather than a purely algorithmic one?

This SoS approach underscores the necessity for integrated data architectures to achieve safe, climate-neutral, and circular innovation.

🧩 III-A.B.C: Query Design & Roadmapping

Domain	Rationale	Records (WoS/Scopus)
1. Supply Chain & Emissions	Scope 3, neutrality, resilience	121 / 237
2. Policy & RegTech Platforms	ITU, IEEE, ISO, CBAM, SaaS	52 / 260
3. Federated Governance	Data spaces, interoperability	45 / 285
4. Metrology & AI	Machine learning, in-situ sensors	99 / 265
5. Foundation Models	Multimodal, safe intelligence	35 / 14
6. Circular Economy	LCA, zero-waste engineering	17 / 47

Note: Data from Table II.

To find a solution, we executed a scoping review in March 2026 across Web of Science and Scopus. As detailed in Table 2, we mapped six distinct domains spanning technical metrology, federated data spaces, and ESG platforms to understand how these systems currently interact or fail to interact.

🧩 III-D.E: Dataset & Metrics

Category / Metric	Web of Science	Scopus
Total Documents (\(N\))	363	1102
Source Publications	202	649
Annual Growth Rate	4.22%	6.28%
Document Average Age	4.86 years	8.19 years
Conference Papers	6	540
Articles	318	452

Note: Data from Table III.

1,465 total documents analyzed. - Web of Science: academic sample with strong keyword depth - Scopus: active, industrial knowledge cycles contributing a substantial volume of 540 conference papers highlighting real-world, corporate R&D activity

Table 3 outlines our dataset. We analyzed 1,465 total documents. Web of Science provided a highly curated academic sample with strong keyword depth. Scopus captured active, industrial knowledge cycles, contributing a substantial volume of 540 conference papers highlighting real-world, corporate R&D activity.

👩‍🏫 Section IV: Findings

👩‍🏫 IV.A: Institutional Affiliation

Table IV. Top Affiliations: WoS vs Scopus

Our institutional mapping in Table 4 exposed a severe structural hole. Academic networks in Web of Science focus heavily on fundamental modeling. Meanwhile, Scopus is dominated by corporate heavyweights like Infineon, IMEC, and ASML. There is an absence of formal mechanisms translating university research into industrial practice.

👩‍🏫 IV-B: Global Collaboration Networks

Fig 1: International Co-authorship Networks

• Geopolitical Interdependence: Dense trans-Pacific and Euro-Asian edges.
• Hub Dominance: US, China, South Korea, and Taiwan lead networks.
• Regulatory Friction: High research collaboration vs. fragmented ESG compliance.

When looking at national collaboration in Figure 1, the networks reveal dense trans-Pacific and Euro-Asian connections. Functional decoupling has not materialized in research. However, this high level of technical co-authorship stands in sharp contrast to fragmented global ESG reporting standards.

👩‍🏫 IV-C: Regional Funding Distributions

Region / Funder Hub	Key Contributing Agencies	Shares (%)
China	NSFC, National Key R&D, CAS	21.77%
Non-China East Asia	NRF (Korea), NSTC (Taiwan), JSPS	20.97%
European Union	Horizon Europe, ECSEL JU, ERC	11.29%
United States	NSF, NIST, DoD, DoE, DARPA	9.68%

Note: Data from Table V.

Funding agency distributions in Table 5 show that East Asian agencies back over 42% of the funded projects. Yet, despite driving the volume of innovation, these regional R&D ecosystems remain isolated from Western market access rules, representing another critical structural hole.

👩‍🏫 IV-D: Conceptual Structure - Factorial

Fig 2: Relational Factorial MCA Map

• Cluster 1 (Red): AI-Driven Smart Manufacturing (VM, Deep Learning).
• Cluster 4 (Purple): Sustainability and Device Performance.
• The Structural Hole: Isolation between environmental metrics and the AI-core.

Figure 2 shows our Factorial Multiple Correspondence Analysis. It maps a clear division. Axis 1 separates operational factory execution from strategic governance. The most alarming structural finding is the complete isolation of Cluster 4, sustainability, from the high-density AI manufacturing core in Cluster 1.

👩‍🏫 IV-E: Conceptual Structure - Keyword Co-occurrence Topology

Fig 3: Keyword Co-occurrence Network

• Core Core-Periphery Structure: Dense core around fabrication and VM.
• Disconnected Peripheries: Standards and Power Electronics are isolated.
• Required Fix: Establishing explicit “grid-to-core” data pathways.

The network topology in Figure 3 confirms this fragmentation. Mainstream semiconductor fabrication terms form a tight, central hub, while standards and energy efficiency sit at the absolute periphery. To fix this, we must build a grid-to-core pathway that feeds real-time power metrics directly into compliance loops.

🗪 Section V: Discussion

• RQ1: Data Interoperability: How can semiconductor analytics and measurement systems integrate heterogeneous data and generalize across customers and process nodes while remaining interoperable with ESG frameworks, including through federated data space architecture integration?
• RQ2: Scarcity Adaptation: How can analytics and measurement platforms be designed and tuned for scarce or imbalanced data while maintaining reliability for critical defect classes, leveraging soft-sensing, VM, and outsourced pipelines for better resource allocation, risk management, and lifecycle assurance?
• RQ3: Eco-Efficiency: How can industrial analytics improve yield and throughput while preserving performance for underrepresented classes and aligning with AI safety and sustainability mandates (e.g., Scope 3, CBAM) – framing optimization as a SoS challenge rather than a purely algorithmic one? This SoS approach underscores the necessity for integrated data architectures to achieve safe, climate-neutral, and circular innovation.

Integrating the Research Questions

We can now synthesize our three core research questions. Data interoperability, scarcity adaptation, and eco-efficiency cannot be treated as separate issues. They represent an interdependent System of Systems challenge that requires structural integration across the entire manufacturing lifecycle.

🗪 V.A: Research Questions Revisited

• RQ1: Data Interoperability: Solved via federated data spaces.
• RQ2: Scarcity Adaptation: Addressed through virtual metrology soft-sensors.
• RQ3: Eco-Efficiency: Balancing local yields with global Scope 3 mandates.

Fig 4: Safe and Sustainable by Design architecture

Note: Conceptual outline of Fig. 4.

Answering all research questions, we propose this six-layer Safe and Sustainable by Design architecture. This framework acts as a conceptual scaffold to integrate factory floor metrology directly with high-level regulatory compliance, mapping data flows from molecule to market.

🗪 V: Base-Substitution & RegTech Layers

Fig 4: Safe and Sustainable by Design architecture

• SSbD Substitution: Targeting PFAS, PFCs, and scarce mineral abatement.
• RegTech Layer: Deploying digital product passports for clear provenance.
• Defensive Prompting: Preventing compliance gaming in automated reporting.

At the foundation, the substitution layer drives chemical safety and mineral abatement based on the IRDS roadmap. Directly above it, the RegTech layer uses digital product passports to automate compliance. Crucially, it includes defensive prompt structures to stop vendors from gaming automated screening tools.

🗪 V: Core-Federated Spaces & Metrology

Fig 4: Safe and Sustainable by Design architecture

• Federated Data Fabrics: Preserving cross-border data sovereignty.
• Metrology Optimization: Real-time closed-loop Statistical Process Control.
• Grid-to-Core Pathway: Connecting sensor power metrics to CBAM logs.

The middle layers solve the data sharing dilemma. Federated data fabrics allow secure information exchange across different countries without losing data sovereignty. This enables inline sensors and virtual metrology models to pipe verified energy and emissions data directly into compliance applications.

🗪 V: Twins & Transitions

• Digital Twins: Simulating sustainability trade-offs before physical scaling.
• Industry 5.0 Vision: Prioritizing human-machine synergy and resilience.
• Socio-Technical Ecosystem: Multipolar diversification and automated compliance.

The top layers complete the ecosystem. Digital twins simulate manufacturing changes to verify energy and yield trade-offs before physical wafers are ever processed. This turns regulatory demands into an automated, self-correcting driver of actual industrial innovation.

🎁 Section VI: Conclusion

Toward Intelligent, Sustainable, and Scalable manufacturing

• Silo Elimination: Merging factory analytics with ESG reporting.
• System of Systems Paradigm: Local process optimization must serve global limits.
• Interdependence: Global research links remain intact despite policy friction.

Caution

Screenshot (not in the paper): Source

Semiconductor Business Connect 2026

In conclusion, our scoping review shows that we must eliminate the silos separating factory floor analytics from ESG compliance. Moving forward, the semiconductor industry must adopt a System of Systems paradigm where local manufacturing optimizations actively support global ecological constraints.

🎁 VI: Industrial Relevance & Use Cases

• IMEC.netzero: Virtual fab modeling of chemical and energy footprints.
• ESMC Dresden Fab: Embedding EU sustainability mandates into production.
• Infineon Smart Power Fab: Designing circularity directly into operations.

European Semiconductor Manufacturing Company (ESMC)

This architecture is grounded in pioneering 2026 initiatives. For instance, the IMEC netzero platform models process footprints, while the new ESMC fab in Dresden and Infineon’s Smart Power Fab are embedding these exact circular design and compliance data flows directly into advanced-node fabs.

🎁 VI: Standardized Knowledge Transfer

• Historical Precedent: Leveraging “Copy Exactly!” and SEMI frameworks.
• Modern Expansion: Extending technical standards to cover sustainability.
• Global Value Chain: Expanding IRDS and ISRS multi-stakeholder workshops.

IRDS/ISRDS Workshop in Granada

The semiconductor sector has always thrived on rigorous, standardized knowledge transfer models like Intel’s “Copy Exactly.” We must now extend this historical capability to sustainability by expanding multi-stakeholder workshops to build an open, standardized net-zero value chain.

🎁 VI: Future Research Directions

• Industrial Pilots: Testing federated data spaces in cross-border fabs.
• Metrology Validation: Substantiating CBAM disclosures with sensor data.
• Ecosystem Co-Development: Refining the architecture via industry feedback.

Thank you for your time. Moving forward, we call for industrial pilots to test these federated data spaces in active cross-border fabs. This will ensure that process-level metrology truly validates regulatory reporting, driving a secure, climate-neutral transition.

Thank You for Your Attention!

Scoping Review of AI, Metrology, and
ESG in the Semiconductor Sector

Implications for Safe and Sustainable by Design (SSbD)

• RS-MI-2B: AI, Generative AI & Digital Transformation, ICE 32nd Ed.

• Tuesday, 23/June/2026: 3:00pm - 4:20pm; Location: Room Miragaia

Karen Ang 0009-0008-5923-0106
Infineon Technologies

Han-Teng Liao 0000-0003-1081-5599
Independent Researcher