ALD2023 Session AM-MoP: ALD for Manufacturing Poster Session

In order to develop a sustainable ALD processing cluster tool for 300 mm wafers, it is necessary to establish a manufacturing technology for a high-productivity, high-efficiency ALD process chamber that reduces the intrinsic excessive consumption of energy and materials.¹ In the present study, as the part of countermeasure to the excessive consumption, a micro-gap ALD process chamber is considered for the optimized design with the process space volume decreased. The changes in the flow field of nitrogen in the process space of the process chamber with the gap sizes of 1 mm and 10 mm respectively are observed at 200 ^oC, utilizing computational fluid CFD numerical analysis. For the present nitrogen flow field with a base pressure of 1 Torr and a temperature of 200 ^oC, the Knudsen number Kn<0.1 and Reynolds number Re<<2300 are evaluated, and consequently the continuity and momentum equations of a steady-state compressible laminar flow field are considered.^2,3

Acknowledgment

This work was supported by the Korean Ministry of SMEs and Startups, under Award no. S2960951.

References

1. C.Y. Yuan and D.A. Dornfeld, 2010, J. of Manufacturing Science and Engineering132, 030918 (2010).

2. M. R. Shaeri, T.-C. Jen, C. Y. Yuan and M. Behnia, International Journal of Heat and Mass Transfer89, 468 (2015).

Processing atomic layer deposition (ALD) coatings on porous high-surface-area particles is of increasing interest related to applications as heterogeneous catalysts. ALD on particles can be made in various reactor configurations such as fixed powder bed, fluidized bed, and rotating drum. Also reactors meant for thin film processing are used for porous particles, by placing the particles on a tray and allowing the gasses to flow over the bed and diffuse into the bed.

Although the fundamental ALD mechanisms are the same irrespective of the geometry of the substrate material, specialized particle coating reactors differ significantly from mainstream thin-film ALD reactors. Most important is to take into account the much larger reactant doses needed to saturate the surface with the adsorbed species. One gram of a high-surface-area often contains ~100 to 1000 m², compared to ~0.1 m² of a typical silicon wafer. The required amount of reactant scales directly with the surface area to be coated. Furthermore, porous high-surface-area materials can be estimated to have extremely high aspect ratios (HAR): mesopores with 5 nm diameter and particle size of 1 mm give an aspect ratio of ~100 000. Coating HAR substrates requires longer reactant exposures, allowing the reactant to diffuse into the structure. Of the various specialized particle coating reactors, the fixed bed reactors are the most simple to construct, and oldest in use: both historical development branches of ALD - atomic layer epitaxy (ALE) and molecular layering (ML) - employed such reactors.

In this work, we present a new ALD reactor design for coating porous high-surface-area particles in a fixed bed. The reactor reported in this work is aimed for fundamental laboratory-scale studies, allowing the coating of a few grams of porous high-surface-area material at a time. Loading and unloading of the sample inertly is possible, enabling the processing of air-sensitive substrates and the investigation of the adsorbed species without air contact. Pre-treatment can be made at temperatures up to 800°C under controlled gas flow. ALD reactants are loaded in sources which can be heated up to about 200°C. Gaseous sources are included. The reactor is equipped with an afterburner and condenser for treating the unreacted reactant. Port for gas analysis e.g. via mass spectroscopy is foreseen.

UV instruments on NASA space missions such as Hubble Space Telescope (HST), the Far Ultraviolet Spectroscopic Explorer (FUSE), and the Galaxy Evolution Explorer (GALEX) have made groundbreaking astrophysical discoveries in areas as diverse as galaxy evolution, star formation, and molecular cloud chemistry. These spectrometers have all benefited from the use of Al mirrors that are highly reflective in the ultraviolet (UV). Aluminum is the only reflective metal that offers broad ultraviolet/visible/near-infrared response, making it highly relevant for use in all far ultraviolet (FUV, 90-200 nm) instruments. However, aluminum is very reactive and susceptible to oxidation, which can limit its reflective performance in the FUV. Ultra-thin coatings of metal-fluorides such as AlF₃, MgF₂, and LiF can be used to protect the Al surface while preserving its high reflectance in the FUV. Atomic layer deposition (ALD) provides unparalleled uniformity and thickness control, making it the ideal process for coating these curved mirrors and shaped optics. However, the Al mirrors are typically fabricated via a separate physical vapor deposition (PVD) process, requiring the mirror to move between one vacuum system to another and exposing it to air, which results in the immediate formation of the native oxide on the Al surface. Therefore, it is necessary to coat the Al surface with the metal-fluoride before exposing it to air.

Spatial Atomic Layer Deposition (sALD) offers a unique opportunity for localized deposition due to its physical separation and isolation of precursor and co-reagent dosing.^[1] While simple in theory, due to well-developed examples of sALD, in practice miniaturization of sALD requires substantial effort into the creation of suitable micro-nozzles.^[1] Uniquely, ATLANT 3D has developed proprietary sALD micronozzles, called microreactor Direct Atomic Layer Processing - µDALP^TM.

The µDALP^TM process undergoes the same cyclic ALD process but is only done in a spatially localized area.^[2] The microreactor or micronozzle confines the flows of gases used for ALD within a defined µm-scale centric area on the substrate, to deposit the desired material. Similarly, to spatial ALD, the creation of this monolayer then hinges on the movement of the substrate.^[1,2]

Since sALD and the µDALP^TM process are based on physical separation, it is theoretically compatible with any ALD material process however requires development as ALD processes are highly tool dependent.^[3] As such, the material capabilities can match traditional ALD and exceed other patterning techniques, such as lithography, which can be costly and time-consuming, especially for rapid prototyping required for innovation.^[4,5]

sALD using the µDALP^TM technology also vastly increases the efficiency and innovation potential of material and precursor development. Using a small amount of precursor (due to low flow rates required) multiple film thicknesses can be deposited onto a single wafer used to calculate a processes growth rate within only a few hours, compared to days for a traditional ALD process (Fig 1). Multiple depositions can also be performed at varying temperatures for the calculation of temperature dependent growth rate (for “ALD window”), and film characteristics all within a few hours on a single sample. The µDALP^TM process has also been used to demonstrate the selective deposition of different materials on the same substrate without the need for masking shown in Fig 2. By facilitating the more efficient development of ALD processes, µDALP^TM sALD can help to enable continued and more efficient growth of the ALD industry and the development of new and innovative technologies. Multi-material sALD also enables unseen potential for versatile patterning and complex geometry formation, applicable to efficient, iterative, and low-cost device and sensor development.

In this work, we describe how Hitachi Kokusai (HiKE) vertical furnaces developed the novel thin film formation method, to provide the second tool type of FinFET hard mask silicon nitride (HM SiN) process that widely used in tsmc very large scale integrated (VLSI) devices fabrication line, The background is the first tool type Tokyo Electron (TEL) not able to provide formula furnace with end of maintenance (EOM).

We develop an innovation HiKE low-temperature 500C process with good film property control of silicon nitride (SiN) formed by low-pressure atom layer deposition (LPALD), and it has been developed by using hexachlorodisilane (HCD, Si2Cl6). The potential scope of application to lead the advanced production line extend from the conventional low-pressure vapor deposition (LPCVD) technique performed at 600C and produced by TEL formula vertical furnaces. The film formed by HiKE furnace shows excellent uniformity with the same refractive index 2.28 and silicon-nitride ratio (Si:N ratio), to lead the same performances of film properties and is advantageous for thermal budget and cost reduction.

In this study, HCD-SiN deposition characteristics, temperature dependence of the film composition and film properties under VLSI fabrication processes are reported, and the differences with the conventional LPCVD HCD-SiN are discussed.