Case Study: DATA-DRIVEN PRECURSOR DISCOVERY FOR NEXT-GENERATION LOW-k DIELECTRICS

Customer: Global Semiconductor Manufacturing Equipment Company 

Outcome: materialsIN rapidly generates scientifically-grounded recommendations and shortlists of precursor candidates for experimental evaluation in advanced process development.

THE CHALLENGE

Advanced microelectronic devices require continuous innovation in materials and deposition processes to sustain performance scaling. Predicting synthesizability remains the primary bottleneck to accelerate the transition from theoretical materials discovery to practical fabrication. Unlike the discovery phase, there is no unifying theoretical or experimental framework that reliably predicts whether a material can be synthesized or provides a viable fabrication recipe. As a result, identifying synthesis pathways and processing conditions still relies heavily on empirical iteration, forming a fundamental barrier to rapid commercialization.

A global leader in semiconductor manufacturing equipment needed to explore next-generation low-dielectric-constant (low-k) thin-film materials that rely on complex multi-component coordination networks deposited by vapor-phase techniques such as atomic and molecular layer deposition (ALD/MLD). 

A fundamental barrier to experimental progress was the identification of suitable vapor-phase precursor molecules that satisfy stringent, often competing, constraints: sufficient volatility, thermal stability, controlled surface reactivity, and compatibility with self-limiting ALD/MLD reaction cycles. Traditional trial-and-error precursor discovery is slow, expensive, and high-risk, particularly when exploring unfamiliar multi-metal or ligand-mediated reaction spaces.

The company, led by an R&D team, engaged materialsIN to accelerate and de-risk early-stage process development by creating a scientifically grounded, data-driven method to identify and prioritize viable precursor candidates for experimental evaluation.

THE APPROACH

materialsIN applied its proprietary materials informatics and machine-learning methodology to develop a systematic precursor screening and ranking framework grounded in the foundational principles of organic chemistry. Its key elements include:

• Screen and select potential chemical precursors for synthesis from millions of candidate compounds;
• Identify processing and synthesis pathways with high probability of successful fabrication through a data-fusion system linking databases, experiments, and relevant materials modeling tools; 
• Predict complex information spaces using materialsIn’s uncertainty-informed decision-making tools that harness large models to explore alternative reasoning paths for materials synthesis and process optimization; and 
• Develop a science-informed AI platform using a Chemical Reasoning System that enables users to interact with AI analytics to refine the interpretation of results, balancing speed with robust decision-making.

THE RESULTS

materialsIN delivered a scientifically defensible, vendor-accessible precursor shortlist, along with a transparent selection logic linking molecular features to process feasibility.

Key outcomes included:

• Ranked library of candidate vapor-phase precursor molecules meeting defined volatility and stability thresholds;
• Chemically diverse screening subset suitable for immediate experimental testing;
• Mechanistic reaction-pathway map guiding experimental sequence design; and
• Closed-loop informatics framework enabling continuous refinement as experimental results become available

This approach allowed the client to transition from conceptual materials targets to experimentally actionable precursor screening while substantially reducing early-stage discovery time and risk.

THE IMPACT

The engagement demonstrates how materialsIN’s materials-informatics platform accelerates precursor discovery for advanced semiconductor process development in that it:

• Compresses precursor identification timelines;
• Reduces reliance on trial-and-error experimentation provides significant cost savings;
• Provides chemically interpretable, data-driven selection logic; and
• Enables data-driven guidance in linking modeling and experimentation.

By delivering ranked precursor shortlists and a reusable discovery framework, materialsIN equips semiconductor manufacturers to explore emerging deposition chemistries with higher confidence, lower cost, and faster time-to-experimental validation, and accelerates the introduction of products into the market

Download this case study as a PDF