ALD/ALE 2025 Session AS-TuP: Area Selective ALD Poster Session

Area-Selective Deposition (ASD) is an attractive process for semiconductor manufacturing[1]. It utilizes the differences of the surface chemical reaction between a precursor molecule and different substrates. Surface reaction simulations are useful tools for screening appropriate ASD precursors[2]. In this study, we investigated the validity of adsorption energy calculation using density functional theory (DFT) by comparing the calculation with the process outcomes.

We calculated the adsorption energy of Trimethylaluminium (TMA) molecule on H-terminated Si, OH-terminated SiO₂, and NH-, OH- and F-terminated SiN substrates (Fig1(a)). Here, TMA is a well-known precursor that is often used for Al₂O₃ deposition, and the choices of these surface terminations were based on the usual dilute HF (DHF) treatment prior to precursor introduction to remove native oxides[3]. It is widely known that H (OH) terminated surface was formed for Si (SiO₂) surface after DHF. For SiN surface, it was found from X-ray photoelectron spectroscopy (XPS) measurement (Fig1(b)) that fluorine was clearly detected on the surface, suggesting that not only OH and NH but also non-negligible amount of F was terminated on the surface, which is consistent with the previous study [3]. The calculated adsorption energy (Fig1(a)) was lower in order of the SiO₂(-OH), SiN(-OH), SiN(-NH), Si(-H), and SiN(-F) cases. This suggested that TMA is most likely to adsorb on the SiO₂(-OH) surface, whereas most unlikely to adsorb on SiN(-F) surface.

To compare the simulation results with experiments, we actually introduced TMA to Si, SiO₂, and SiN substrates in a chamber after DHF. Fig1(c) shows the XPS results (Al 2p) of these substrates after TMA flow. The largest amount of Al (resulting from TMA adsorption) was observed on the SiO₂ substrate, which agrees with the speculation from the simulation results. For SiN substrate, we also performed H₂O dipping treatment for 12.5 h at maximum after DHF, and consecutively, TMA flow experiment. As shown in Fig2(a) (XPS results), amount of fluorine decreased for 12.5-h H₂O dipping, suggesting that -F is replaced by -OH. Corresponding to this, it was found from XPS measurement (Fig2(b)) that amount of Al increased. This experimental result is consistent with the adsorption energy calculation (Fig1(a)) where TMA is more likely to adsorb on SiN(-OH) than on SiN(-F). These results demonstrate the usefulness of the DFT calculations to estimate selectivity in developing ASD.

[1]G. N. Parsons and R. D. Clark, Chem. Mater. 2020, 32, 4920−4953.

[2]P. P. Wellmann et al., Chem. Mater. 2024, 36, 7343−7361.

[3]L-H. Liu et al., J. Phys.: Condens. Matter 28 (2016) 094014.

Area-selective deposition (ASD), a “chemical patterning” method that can reduce the fabrication cost and achieve novel structures beyond lithography, is gathering increasing global interests from researchers in semiconducting industry.^1–3 One of the key items to realize ASD on two specific types of areas is the selection of reactants and/or inhibitors from over thousands of candidate molecules. Compared to experiments, the first-principles simulation is a more accessible approach with lower cost and higher speed. Yet, a thorough simulation for an adsorption event requires searching for several most energetically favorable saddle points (i.e. transition states) in the potential energy surface, which makes it demanding in computation resources for high-throughput exploration.

In this work, we developed a simple surrogate automatic simulation workflow for evaluation of a molecule’s adsorption behavior on surfaces by using density functional theory (DFT) software VASP and automation packages fireworks and atomate⁴. First, we performed manual simulations of the adsorption behavior for N-(Trimethylsilyl)dimethylamine (TMSDMA) C5H15NSi on 7 types of surfaces by using the NEB method. It is found that the activation energy E_a is almost in proportion to the adsorption energy after molecule dissociation, which is denoted as E_{d_ads}. (Fig. 1) Such empirical relationship is similar to the well-known Bell-Evans-Polanyi principle⁵, so we believe E_{d_ads} can serve as a reasonable surrogate metric for E_a.

Accordingly, an automatic E_{d_ads} simulation workflow that can break molecules into fragments and search for the most stable adsorption configuration automatically has been implemented (Fig. 2) and tested for 4 types of molecules on Ru and hydroxyl-terminated SiO2 (SiO2-OH) surface. For each molecule-substrate pair, multiple adsorption configurations have been investigated, from which the most stable one is chosen to represent the final adsorption energy of the molecule on the specific surface (Fig. 3). Within the 4 molecules we investigated, we find that the difference of alkynes’ adsorption energies on Ru and SiO2-OH is the largest, outperforming TMSDMA, carboxylic acids, and pyridine. Alkynes could be promising inhibitors blocking Ru surface compared to SiO2-OH.

References

¹ G.N. Parsons, and R.D. Clark, Chem. Mater. 32(12), 4920–4953 (2020).

² S. Balasubramanyam, et al., ACS Mater. Lett. 2(5), 511–518 (2020).

³ C.T. Nguyen,et al., Chem. Mater. 35(14), 5331–5340 (2023).

⁴ G. Kresse, and J. Furthmüller, Phys. Rev. B 54(16), 11169–11186 (1996).

⁵ M.G. Evans, and M. Polanyi, Trans. Faraday Soc. 31, 875 (1935).

Area-selective atomic layer deposition (AS-ALD) is one of the next-generation technologies that enable bottom-up fabrication, in contrast to the traditional top-down fabrication used in semiconductor manufacturing. AS-ALD has the potential to reduce the number of lithographic patterning steps considerably. Furthermore, AS-ALD enables angstrom-level accuracy in specific regions of substrates with diverse chemical properties when fabricating thin film devices. In this study, we applied small molecular inhibitors (SMIs) in AS-ALD to address the issues of low volatility and uniformity in gas-phase processes of conventional self-assembled monolayers (SAMs). We compared two SMIs, trimethoxyphenylsilane (TMPS) and tri(dimethylamino)phenylsilane (TDMAPS), containing methoxides and dimethylamines as the head groups that can interact with the substrates.

The synthesized SMIs were characterized using nuclear magnetic resonance spectroscopy. SMIs were deposited using ALD, and water contact angle measurements were performed on four substrates (SiO2, SiN, TiN, and Cu) to evaluate the coverage and passivation effectiveness. After depositing a Ru film on the SMIs-coated substrate using the Ru precursor [Ru(II)(η⁵-C₇H₇O)(η⁵-C₇H₉)], the film thickness was measured via X-ray fluorescence analysis. X-ray photoelectron spectroscopy was performed to conduct a precise elemental analysis of Ru. TDMAPS deposited on SiO2 and SiN substrates showed 100% selectivity for 150 cycles of Ru deposition, while deposition occurred on Cu and TiN substrates after a slight nucleation delay, confirming the substrate-specific chemo-selectivity. In particular, the TDMAPS-coated SiO2 and SiN showed higher selectivity of Ru precursors compared to those of TMPS, which is attributed to the of higher reaction rates amino groups with the substrate. This study demonstrates the importance of the head groups of SMIs to enhance the selectivity in AS-ALD and provides guidelines for the design of SMIs.

As semiconductor devices continue to scale down while demanding higher performance, precise process control and a fundamental understanding of atomic-scale surface reactions become essential. Area-Selective Atomic Layer Deposition (ASALD) has emerged as a promising technique, enabling material growth in designated regions by leveraging surface chemical properties. Unlike conventional ALD, ASALD eliminates the need for additional patterning steps, inherently achieving selective deposition at the nanoscale. This makes ASALD particularly relevant for next-generation semiconductor devices, where precise material placement is critical, such as in 3D-integrated circuits and functional thin-film applications.

Zinc oxide (ZnO), a transparent conductive oxide, has garnered significant interest due to its excellent electrical and optical properties. It is widely used in next-generation semiconductor devices, transparent electrodes, optoelectronic components, and gas sensors. While previous studies on ZnO ASALD have primarily focused on inhibitor-based selective deposition, the influence of inherent substrate properties on ZnO film growth and incubation cycles remains largely unexplored.

In this study, Density Functional Theory (DFT) and molecular dynamics (MD) simulations were conducted to examine how different substrate compositions (SiO₂, Ti, Cu, and Ru) influence ZnO ALD precursor adsorption. To better capture realistic surface configurations, a machine learning force field (MLFF) was implemented, enabling MD simulations that more accurately reflect atomic interactions. Additionally, extensive DFT calculations were performed to quantify the intricate interactions occurring at different surfaces. The analysis revealed substantial variations in precursor-surface interactions, including significant differences in adsorption energies during physisorption. Furthermore, discrepancies in ligand exchange reaction energies were observed, leading to notable shifts in ZnO growth kinetics depending on the substrate. These differences directly impacted incubation cycles, providing a theoretical foundation for substrate-driven selectivity in ZnO ASALD.

By leveraging these insights, we establish a theoretical framework for understanding how chemical reactants interact with different surfaces and how substrate characteristics can be intentionally modified to induce inherent selectivity. This work highlights the importance of a strong theoretical foundation in optimizing atomic layer processing for unprecedented selectivity, offering a pathway to precise pattern control in advanced semiconductor manufacturing.

With the increasing demand for higher integration and enhanced performance in semiconductor devices, area-selectiveatomic layer deposition (AS-ALD) is playing a crucial role. AS-ALD is a technology that can precisely depositnanometer-level thin films in desired areas, effectively addressing issues that may arise during unnecessary pattern formation or processing. Particularly, selective deposition of metal materials in metallization, and further advanced packaging, where multiple layers of metal need to be deposited sequentially, has the advantage of implementing more complicated structures. Molybdenum carbide (MoC_x) is gaining attention as a promising material for the next generation semiconductor manufacturing due to its excellent mechanical and electrical properties as well as high thermal and chemical stability, there are few reports on Mo-based materialsfor AS-ALD.

In this study, we developed molybdenum carbide (MoC_x) thermalALD using a novel combination ofMo precursor and co-reactant, and conducted analysis of the film properties. Notably, itwas observed that the film growth behavior varied depending on the substrates by analyzing Mo areal density and film thickness. MoC_x films growth was effectively suppressed on Al₂O₃ substrate, whereas significant film deposition was observed on Cu substrate. Despite being a dielectric substrate, the SiO₂ substrate showed similar growth behavior to Cu, suggesting these processes preferred to suppress MoC_x deposition on Al₂O₃. The MoCₓ process achieved ~10.4 nm selective deposition on Cu over Al₂O₃ (94.2 % selectivity) and ~13.9 nm on SiO₂ over Al₂O₃ (92.8 % selectivity).These findings demonstrate the feasibility of selective deposition on different substrates, particularly with Al₂O₃ as non-growth area (NGA). This study aims to present an effective approach to overcome technical limitations in next-generation device manufacturing by proposing a method for ASD not only in metal-dielectricapplications but also between dielectrics through theinherentAS-ALD process of molybdenum carbide.

Acknowledgements

This work was supported by Samsung Electronics, and supported by the Technology Innovation Program(RS-2024-00509266, Development of Next-generation dielectric and electrode process equipment for logic 1nm or less and memory xnm level) funded By the Ministry of Trade Industry & Energy(MOTIE, Korea)

The scaling of semiconductor devices has significantly increased transistor density, which has improved the performance of various electrical devices ranging from computers to smartphones. Through such an evolutionary miniaturization, a semiconductor manufacturing technology has required more fine deposition and patterning techniques. However, in the existing nano- fabrication processes based on lithography, it is highly challenging to form an accurate pattern alignment and limited to handle interfacial phenomena inside device. Accordingly, the semiconductor industry has intensively focused on ‘area selective deposition (ASD)’. It provides a powerful technique to yield area-selective interfaces of metal/low-k material by using the self-aligned molecules as an inhibitor. In this article, we discussed the application of ASD into logic products. In particular, we introduced the evaluation results for the reverse selective barrier metal (RSBM) and fully aligned via (FAV) that are being applicable to the backend of line (BEOL).

Iridium (Ir) has a low figure of merit (ρ₀×λ) and a high melting temperature, so it has recently been spotlighted as a very important copper (Cu) alternative interconnect material in next-generation semiconductor devices. In particular, atomic layer deposition (ALD) enables the deposition of ultra-thin, conformal, and uniform films with excellent step coverage, even in highly complex or narrow trench structures of several-nm dimensions, due to its inherent self-limiting characteristic. In this regard, ALD-Ir is considered one of the most suitable metallization processes for advanced semiconductor interconnect applications.

To develop a reliable ALD-Ir process with improved film quality, considerable efforts have been made. Recently, an ALD-Ir process exhibiting low electrical resistivity and negligible oxygen impurities was reported using Tricarbonyl (1,2,3-η)-1,2,3-tri(tert-butyl)-cyclopropenyl iridium (C₁₈H₂₇IrO₃ or TICP) precursor and oxygen [1]. However, despite these advantages, this TICP precursor exhibits a long incubation period and high nucleation delay, making it difficult to deposit extremely thin and continuous Ir films on the hydroxyl-terminated oxide surface.

Therefore, in this study, a method for depositing highly uniform and continuous ALD-Ir thin films with low resistivity even on oxide materials was explored by significantly reducing the incubation period and promoting nucleation through impurity-free accelerators under simple process conditions. This additional step does not change the existing device structure while enabling the formation of high-quality Ir thin films with strong resistance to external impurities. Furthermore, we systematically compared and analyzed the nucleation and growth behavior, as well as film properties, of ALD-Ir on the oxide surface under process conditions with and without the accelerator. As a result, we successfully obtained ultrathin ALD-Ir films with superior uniformity, low surface roughness, and low resistivity, even on oxide surfaces.

References

1. Park, Na-Yeon, et al. Chemistry of Materials 34.4 (2022): 1533-1543.

Photoelectrochemical CO₂ reduction (PEC-CO₂RR) is a promising approach for producing sustainable fuels and chemicals. While many semiconductors have been investigated, silicon-based electrodes modified with metal nanoparticles (NPs) remain underutilized, despite their potential for scalable and cost-effective applications. Among metal catalysts, copper (Cu) NPs are widely studied for their ability to generate hydrocarbons and oxygenates. However, their instability under reaction conditions—including aggregation, detachment, and dissolution—leads to loss of activity and altered selectivity. These issues are exacerbated under PEC-CO₂RR conditions due to combined light-induced and electrochemical stress, as well as the low catalyst loadings required for efficient light absorption.

Alloying Cu with other metals like Ag or Pd can improve stability, though the effect depends on alloy composition, synthesis, and operating conditions. We recently demonstrated that Metal-Assisted Chemical Etching (MACE) is an effective and low-cost method for depositing bimetallic NPs (e.g., AgCu, PdCu) on p-Si supports, yielding promising performance in terms of current density and selectivity. However, Cu-rich systems still degrade rapidly, highlighting the need for additional stabilization strategies such as protective coatings or nanostructuring.

In this work, we investigate a novel stabilization method using Area-Selective Atomic Layer Deposition (AS-ALD) to coat silicon-based photocathodes with functional oxide layers. AS-ALD leverages the chemical reactivity contrast between SiO₂ and metal surfaces, enabling selective deposition of Al₂O₃ onto the silicon substrate while leaving metal NP surfaces largely uncovered. By tuning precursor chemistry, deposition temperature, and cycle number, we developed nanometer-thick Al₂O₃ coatings that protect the substrate without blocking catalytic sites.

Initial PEC-CO₂RR tests of protected electrodes (Ag, Cu, and AgCu NPs on p-Si) show enhanced NP stability and reduced hydrogen evolution, likely due to decreased exposure of bare Si to the electrolyte. Characterization via XPS, SEM, and PEC measurements confirms partial preservation of the catalyst structure. Some degradation still occurs, likely due to Al₂O₃ porosity. Ongoing work focuses on improving the robustness of the ALD coatings to further enhance long-term durability.

Non-lithographic patterning of polymer thin films enables facile fabrication of nanostructures and devices for applications spanning from soft electronics to biomedical engineering. Developing area-selective deposition (ASD) for polymers could achieve self-aligned polymer growth, fostering the generation of complex structures and device configurations with enhanced precision and versatility. However, due to the rapid radical propagation in chain reactions, achieving area-selective radical polymerization in all-dry processes still represents a huge challenge. For conventional initiated chemical vapor deposition (iCVD) and photo-initiated chemical vapor deposition (PiCVD), free radicals are primarily generated in the vapor phase, which leads to uniform deposition with negligible selectivity on different surfaces.

Here, we report a photocatalytic strategy to achieve surface-specific radical generation for polymer ASD. Because radicals are generated by the reaction between photogenerated holes and monomers, photocatalytic surface-initiated chemical vapor deposition (PS-iCVD) not only eliminates the need for initiator, but also can localize radical generation on the photoactive surfaces, which enables ASD of polymer thin films. Moreover, for semiconductors with narrow bandgaps, such as native oxide of Fe, the PS-iCVD can also enable polymer ASD under visible light for fabricating self-aligned patterns with high area-selectivity (97.4% at deposition thicknesses more than 40 nm). The PS-iCVD offers placement control over chain-growth polymers, providing insights for guiding the design of eco-friendly deposition methods and bottom-up non-lithographic nanofabrication. These polymer patterns shown in this work are expected to serve as inhibitors and templates for achieving area-selective atomic layer deposition.

Area-selective atomic layer deposition (AS-ALD) has emerged as a crucial technique for fabricating semiconductor devices, addressing the limitations of conventional top-down patterning methods based on photolithography. In particular, AS-ALD of SiO₂ on SiO₂ surfaces over SiO₂/Si₃N₄ patterns is considered essential for manufacturing charge trap layers (CTLs) in highly scaled 3D NAND devices with high aspect ratios and reduced ONO tier dimensions. Among various approaches to SiO₂ AS-ALD, the use of small-molecule inhibitors (SMIs) to passivate Si₃N₄ surfaces shows significant promise due to its effectiveness and minimal side effects. For successful implementation of SMI-based AS-ALD, a thorough understanding of the physical and chemical processes involved during film deposition is critical.

In this study, we theoretically investigate SiO₂ AS-ALD on a SiO₂/Si₃N₄ pattern using benzaldehyde (BAD) as a SMI, using density functional theory (DFT) calculations. We first examine the adsorption behavior of BAD on both Si₃N₄ and SiO₂ surfaces to confirm its preferential binding to Si₃N₄. Next, we analyze the reaction between BAD adsorbed on Si₃N₄ and ozone (O₃), a commonly used oxidant in ALD processes, to understand how BAD loses its blocking functionality under oxidative conditions. We also study the potential for BAD re-adsorption on Si₃N₄ following oxidation, evaluating whether repeated inhibitor dosing restores its blocking capability during AS-ALD cycles.

Our DFT results show that BAD selectively adsorbs onto Si₃N₄ surfaces, effectively inhibiting precursor deposition. However, exposure to ozone degrades BAD, resulting in a loss of its blocking performance. Notably, the blocking property is recoverable through repeated BAD dosing, provided the Si₃N₄ surface is only partially oxidized. Once the surface becomes fully oxidized and terminated with –OH groups, re-adsorption becomes significantly hindered.

Overall, our findings demonstrate that BAD serves as a promising inhibitor for SiO₂ AS-ALD due to its selective adsorption on Si₃N₄. Although ozone exposure degrades BAD, its blocking function is restorable through timely re-dosing. These results highlight the importance of controlling re-exposure timing in cyclic AS-ALD processes to maintain selective deposition.