Examining the
Future of Wafer Cleaning

New applications Demand Novel Solutions

Bijan Moslehi, PhD, is chief technology officer and senior vice president, semiconductor technology research, for The Noblemen Group (www.noblemengroup. com), a boutique investment banking, strategic advisory, and business development firm. Moslehi has some 20 years' experience working in the semiconductor and semiconductor equipment industries. He can be reached at bmoslehi@noblemengroup.com.

The April installment of Reality Check examined the wet wafer-cleaning landscape. Today, virtually all wafer-cleaning steps use wet processes, with immersion and spray batch tools controlling the market. However, the emergence of high-productivity, multichamber single-wafer cleaning systems, particularly in back-end-of-line (BEOL) applications, is challenging the dominance of batch equipment.

Over the past 15 years, many predictions of the displacement and replacement of wet cleans by dry cleaning methods have not materialized. Although achieving an all-dry-clean process flow would be ideal, nonplasma dry cleaning accounts for less than 1% of the total market. These predictions have ignored several fundamental facts behind the continued success of batch and single-wafer wet cleans. One critical factor has been the unique ability of engineered liquid chemicals to perform cost-effective, nanometer-scale substrate cleaning, particularly the removal of small particles and metallic contaminants.

However, anxiety is mounting over the extendability of wet-clean technology beyond the 65-nm node for certain applications. The primary areas of concern include BEOL low-k resist strip and sidewall polymer removal, cleaning and drying of high-aspect-ratio structures, and damage-free removal of smaller particles without affecting materials and device structures. Although significant efforts are under way to extend the capabilities of batch and single-wafer wet-cleaning systems, the technological advancements are stymied by the general lack of profitability associated with the production and sale of wet-process tools.

One emerging technology—supercritical carbon dioxide (scCO₂) cleaning—promises to address many of these issues. Based on the use of CO₂ at supercritical conditions (~ 3000 to 6000 psi and ~ 35° to 100°C), scCO₂ has an operational range that makes it possible to achieve the best aspects of both the wet and dry worlds. Since scCO₂ benefits from the higher density of liquids plus the lower viscosity and near-zero surface tension of gases, it can be extremely effective in cleaning and drying nanometer-scale geometries. To achieve its full potential, scCO₂ technology requires very small amounts of several types of cosolvents and additives; a number of companies are actively developing these specialty chemicals.

Furthermore, material suppliers are working on products and strategies associated with the infrastructure needed for the successful introduction of scCO₂ technology into the fab, including a fabwide CO₂ delivery system. Recycling techniques are also being researched for CO₂ and cosolvents, which would address environmental concerns about the use of a greenhouse gas. Since CO₂ is a safe and benign solvent, supercritical technology offers strong positive environmental benefits versus today's liquid solvents and may significantly reduce DI water usage. However, any anxieties about the safety issues related to the installation of highly pressurized scCO₂ chambers in fabs must be fully addressed by tool manufacturers. Even the perception of a potential safety hazard would act as a barrier to this technology's adoption.

Supercritical CO₂ is quite compatible with nonporous and porous low-k dielectric materials; in fact, several recent investigations have demonstrated supercritical technology's capabilities in advanced copper/low-k applications. When small amounts of additives are mixed with scCO₂, it successfully conditions and repairs low-k materials damaged by plasma and other processes. For example, the addition of hexamethyldisilazane (HMDS) restores the dielectric's k-value, which increases during the etch and strip processing steps. Furthermore, the use of chelating agents effectively removes sidewall copper residues generated during the etching of the etch-stop layer in the dual-damascene process. The cleaning and drying of porous low-k dielectrics for nanostructures ≤45-nm has also been verified. As an enabling technology, the repair and drying of low-k materials could be an immediate critical application for scCO₂.

Supercritical technology also shows promise at 45 nm and beyond in such applications as BEOL postdielectric-etch resist strip and polymer residue removal, and the cleaning of high-aspect-ratio trenches and via structures. The success of supercritical-type systems could potentially displace some wet cleans and could eventually replace even some dry-ash steps. However, to be operationally successful, these tools must provide reasonably high throughputs with short cycle times.

After years of migration toward Marangoni-style drying techniques, the industry appears to be headed toward supercritical-based drying methods, particularly for high-aspect-ratio structures. An important area of interest is the drying of BEOL trenches and vias in the copper/low-k dual-damascene process. MEMS manufacturers already use the scCO₂ drying capability to prevent stiction in certain microelectromechanical structures. However, one much-anticipated scCO₂ application has not yet materialized—postdevelop drying to prevent lithography image collapse. The image collapse issue, which once seemed to represent a major opportunity for supercritical technology to enter mainstream chip manufacturing, has recently been ameliorated by adding suitable surfactants to the rinsewater.

The question on many people's minds is: Can conventional aqueous-based solvent cleans and existing drying methods be successfully extended to meet the integration requirements of porous low-k materials and high-aspect-ratio structures at the 65- and 45-nm nodes? The answer will determine the insertion node for the introduction of supercritical fluids technology and the point at which the investors in scCO₂ start to get some return on their R&D investments. Whether the industry will value supercritical systems more highly than they have valued wet-clean technology, and whether successful business models can be developed for the production, sale, and support of this type of tool, remains to be seen.

To summarize the current wafer-cleaning landscape, wet cleaning has become a major enabling segment of wafer-processing technologies. Wafer-cleaning steps constitute at least 20% of deep-submicron process flows. The wet-clean market is large and growing, and the batch segment still seems to have a significant life span remaining. However, the productivity and utilization of 300-mm batch tools must improve. Batch suppliers face mounting economic pressures caused by the difficulties of running profitable businesses in the cleaning space.

Single-wafer wet-cleaning tools (spin/spray processors, and scrubbers) have fostered new markets for unique side-selective cleaning applications, such as wafer thinning and backside particle removal and cleans. In the near term, batch tools will continue to dominate the critical front-end-of-line (FEOL) cleaning applications, while single-wafer wet cleans have the potential for continued BEOL growth. The development and especially the market adoption of single-wafer tools with suitable, solid process capabilities and competitive productivity for critical FEOL cleans have been very difficult. However, certain cleans, particularly for some BEOL applications, should continue transitioning into single-wafer spin/ spray processors, fueling the growth of the single-wafer market. Consequently, batch tools may experience future erosion in their overall BEOL market share. If successful, the integration of single-wafer wet cleans with dry-process tools would gradually undermine the position of stand-alone single-wafer cleaning tools in certain applications, resulting in permanent market share loss to captive OEM clusters.

The cleaning of high-aspect-ratio porous low-k materials and other BEOL applications may require the introduction of new approaches such as supercritical technology. A mix-and-match strategy for cleaning tools will be essential and could include immersion and spray batch, stand-alone single-wafer spin/spray processors, scrubbers, and supercritical tools, as well as integrated cleans. As always, the industry will avoid changes that it perceives as unnecessary, too costly, or too risky. In order to succeed, any new tool or process must provide enabling technology solutions that meet critical industry requirements that cannot otherwise be met with conventional technologies or their extensions.

Other new market opportunities include the need to cultivate suitable or alternative damage-free cleans for new materials and processes, such as high-k gate dielectrics and metal gate stacks. The development of better surface and interface control and more-suitable postetch cleans are also required. Environmentally friendly manufacturing considerations have also generated interest, such as continued reduction of chemical and water usage, the development of more-benign chemistries, and, where possible, the elimination of liquid solvents and chemicals. Interestingly, another emerging application—immersion lithography—has seen the leveraging of wet-clean technologies, such as liquid degasification and drying.

Although the wet-clean business continues to be highly fragmented, the inevitable market consolidation has begun. It will accelerate as 300-mm technology is increasingly adopted, because only a few suppliers have a significant presence in the new fab lines. In addition, smaller suppliers do not offer the broad product portfolio needed to address the wide range of wafer-cleaning applications. As the market matures, the big will continue to get bigger at the expense of the weaker players.