ALD/ALE 2025 Session PS-MoM: Plenary Session

For decades, Dynamic Random Access Memory (DRAM) has been the cornerstone of main memory in computing, meeting the demands of an ever-expanding workload. The need for greater capacity, higher bandwidth, and lower power consumption has continually driven advancements in DRAM technology. The recent surge in artificial intelligence (AI) applications has further accelerated the demand for high-speed memory, catalyzing the rise of High Bandwidth Memory (HBM). While the success of HBM underscores the importance of packaging innovations, traditional scaling continues to play a critical role in improving DRAM core die performance, such as power efficiency, area, and cost, ensuring its competitiveness.

DRAM is primarily composed of three key components: the cell transistor, storage node (SN) capacitor, and peripheral transistor. Each component presents unique scaling challenges due to differences in structure and functionality, yet they share similar material requirements. These include the ability to accommodate nanometer-scale dimensions, achieve high aspect ratios, and meet specific electrical property demands. Atomic Layer Deposition (ALD) has been instrumental in enabling advanced DRAM technology by providing atomically precise control over thickness, composition, and conformality.

As 2D DRAM approaches its scaling limits at sub-10nm nodes, next-generation architectures such as 4F² vertical gate DRAM (VG-DRAM) and lateral 3D DRAM are emerging as potential solutions. VG-DRAM offers significant benefits in scalability and performance, though its implementation has posed considerable challenges. However, recent advancements in process technologies have revitalized its potential as a strong contender for next-generation DRAM. Meanwhile, lateral 3D DRAM presents unique advantages for future extensions by circumventing the limitations of areal scaling.

Future opportunities for ALD materials are vast, particularly in addressing the demands of these new architectures. For example, lateral 3D DRAM relies heavily on lateral processes with exceptional conformality and uniformity across hundreds of layers, which is challenging even for ALD technique. Furthermore, new ALD materials may enable more disruptive capacitor-less DRAM based on 2T gain cells or ferroelectric FET.

This presentation will provide an overview of DRAM applications and key components, highlighting the primary challenges in scaling and the critical role of ALD in overcoming these obstacles. Finally, the discussion will explore the future evolution of DRAM architectures and the new opportunities that lie ahead.

Metal (boro)phosphate coatings are of interest for a wide range of applications, in particular for lithium ion batteries, electrocatalytic water splitting, and as biocompatible and protective coatings. In view of the need for thickness control and conformality in these applications, there has been a growing interest in developing ALD-based approaches for the deposition of such coatings.

Identifying a suitable phosphate precursor has proven a major challenge. In theory, phosphoric acid would be an ideal candidate, but its use in vapor deposition is prohibited by its low vapor pressure at room temperature and its tendency to decompose rather than evaporate upon heating. Trimethyl phosphate (TMP) is an attractive alternative, offering a vapor pressure of 15 mbar at 70°C. Thermal ALD processes using TMP have been demonstrated for a variety of metal sources, but it has proven difficult to incorporate a sufficiently high atomic percentage of phosphorus in the growing film, requiring an unpractically large number of phosphate sub-cycles for each metal pulse.

Dobbelaere et al. (Chem. Mater. 26, 6863 (2014)) explored the use of a TMP plasma, where the TMP precursor vapor is introduced into an inductively coupled plasma (ICP) discharge. At low substrate temperatures, exposure to a TMP plasma results in CVD-type coating in a process akin to plasma polymerisation. However, at sufficiently high substrate temperatures (300°C), saturated growth could be achieved in a true ALD regime, resulting in ALD growth of Al phosphate. Similar TMP-plasma based approaches proved successful for ALD of Ti, V, Zn, Sn, Fe, Ni, Co phosphates, as summarised in the review by Henderick et al. (J. Appl. Phys. Rev. 9, 011310 (2022)).

Inspired by the TMP-plasma based approach, Dhara et al. explored plasma activation of trimethyl borate (TMB) towards ALD of Al borate. Using TMB alone in the plasma form does not yield self-limiting ALD growth, as polymerized species continuously accumulate on the substrate. By co-dosing H₂O in the TMB plasma, saturated growth of Al borate films could be achieved at and above 250°C, with a high growth per cycle (~3.5 Å).