ALD/ALE 2026 Session AS2-WeM: ASD Process II

Area selective deposition (ASD) continues to emerge as a pivotal technique for enabling next–generation semiconductor manufacturing, particularly as device geometries shrink and integration demands intensify. However, the long–standing goal of achieving perfectly selective growth often drives processes toward excessive complexity, high precursor consumption, and limited long–term sustainability. In this work, we explore the practical balance between achieving robust selectivity and maintaining overall process efficiency, emphasizing the potential benefits of reducing chemical usage without compromising integration–relevant performance. By adopting a more pragmatic viewpoint, we demonstrate that high–quality selectivity does not necessarily require aggressive surface treatments, extreme exposure times, or high precursor flows. Instead, a carefully tuned, eco–friendly process regime can deliver excellent material discrimination while significantly lowering chemical burden and overall environmental impact.

To evaluate performance under these more sustainable conditions, we employ top–down scanning electron microscopy (TDSEM) as a straightforward yet powerful metrology approach. TDSEM enables rapid assessment of nucleation behavior, feature–scale uniformity, and pattern fidelity, providing a clear window into the relationship between precursor flow, growth onset, and ultimate selectivity. Our results show that excellent selectivity can still be reliably achieved even when precursor usage is dramatically reduced, illustrating the feasibility of simplifying ASD processes for manufacturing–scale deployment. Electrical resistance–capacitance (RC) measurements and integrated wafer–level data further confirm that processes exhibiting modest deviations from perfect selectivity can nonetheless support high–performance interconnect scaling and patterning fidelity. These findings challenge the traditional assumption that optimal integration requires flawless selectivity and instead suggest that controlled, minimal nucleation on non–growth surfaces may be tolerable - especially when weighed against the benefits of lower cost, reduced chemical waste, and substantially improved throughput.

Taken together, this work highlights the value of shifting from a perfection–driven mindset to a more holistic framework that prioritizes precursor efficiency, environmental responsibility, and integration robustness. By demonstrating that sustainable, lower–chemistry ASD regimes can still meet stringent device requirements, we outline a more realistic and scalable pathway for adopting ASD in high–volume semiconductor manufacturing. This approach not only reduces overall process strain but also strengthens the connection between small–scale laboratory studies and real–world wafer–level performance, ultimately enabling faster development cycles and broader implementation of ASD–driven patterning strategies.

Area-selective deposition via atomic layer deposition (AS-ALD) offers a bottom-up approach to fabricate complex and functional nanostructures, being robust for the scaled architectures even with any 3D structures compared to the conventional top-down approach using multi-patterning processes. Inherent-type AS-ALD is the most simplified process, exploiting intrinsic adsorption properties of the precursors depending on the substrates. Compared to inhibitor-assisted area-selective processes, the inherent AS-ALD also highlights practical advantages such as reducing fabrication steps or eliminating concerns of residual inhibitors, which is attractive for the application of interconnect metallization.

Although the inherent AS-ALD has typically suffered from a lower selectivity compared to that of inhibitor-assisted one, a higher inherent selectivity against Si-based dielectrics such as SiO₂, low-k or SiN is essential for the intended purposes (e.g., void-free/seamless bottom-up metallization). Therefore, we demonstrated high selectivity against SiO₂ via Ru thermal AS-ALD using a novel Ru precursor, [Ru(TMM)(p-cymene)], and O₂ as a reactant without any inhibitor or other area-selective activation methods. This precursor has two different ligands, trimethylenemethane (TMM) and isopropylmethylbenzene (p-cymene), and exhibits high thermal stability (> 400 ^oC). A previous report on this precursor revealed that a surprisingly long incubation period of >1000 cycles was observed on SiO₂ while only ~8 cycles were required on TiN at 300 ^oC [1], overcoming the challenges of high selectivity which previous Ru AS-ALDs have suffered from. Computational simulations in the same report also discovered that dissociative adsorption proceeds on metallic substrates such as Ru, but is energetically unfavorable on SiO₂, providing the theoretical support for the high inherent selectivity against SiO₂.

In this study, we expand our understanding of the inherent AS-ALD Ru on technologically important metallic substrates including Ru, TiN, Mo, W against dielectric substrates such as SiO₂, low-k, SiN, Al₂O₃. The results demonstrate that the high selectivity against dielectrics was confirmed both on blanket wafers and patterned substrates. We also discuss how process conditions affect the selectivity and the experimental factors required to emerge the selectivity with mechanistic insights.