Next-generation abrasive particles for CMP

12/01/2004

As IC device structures become more complex and more CMP steps are added to manufacturing these devices, CMP process yield comes under greater scrutiny. Abrasive particles are critical to achieving controlled material removal without sacrificing planarity and can offer both mechanical and chemical functionality to the polishing slurry. Along with their benefits, however, abrasive particles are a contributor to yield-affecting defects. Though nanoscale particles have long been utilized as CMP abrasives, a new generation of nanoparticle abrasives are being explored and developed to meet the demands of more advanced technology nodes.

Limitations of traditional processes

Traditional CMP slurries (nanoscale abrasive particles in aqueous formulations) have been largely based on fumed metal oxides such as silica and alumina. The fuming process is a gas-phase decomposition of metal chlorides in the presence of an oxygen-hydrogen flame. The metal atom reacts with the free oxygen to form the metal oxide solid particle, and the chloride reacts with hydrogen to form gaseous HCl, which is separated from the solid particle.

These fumed metal oxides typically form nanometer-sized primary particles that chain together to form a larger secondary aggregate particle with a typical mean size from 80-150nm. The structured morphology of the fumed aggregate is a unique, distinguishing characteristic of these nanoparticles. These secondary particles agglomerate into loose powders that are readily dispersed in aqueous solutions, forming highly stable colloidal dispersions. Appropriate chemistries can be added to these dispersions to tune the performance of the abrasive particles in the CMP process.

In addition to the fumed metal oxides, the CMP process has utilized colloidal or precipitated silicas that traditionally have been used in primary silicon wafer polishing. These colloidal silicas come in a variety of mean sizes (5-300nm), distribution widths, and morphologies. The particles are commonly formed from the controlled hydrolysis and condensation of silicate ions derived from soluble salts such as ammonium, sodium, or potassium silicates. By controlling the seed concentration (or nucleation event) and condensation growth rate, a desired particle size distribution can be achieved.

An inherent problem with many of these high-volume particle manufacturing processes is the risk of solids contamination by impurities. Impurities come in a variety of forms, ranging from elements that are present in solid or soluble form to large particles having the same composition as the abrasive, but substantially larger than the bulk of the main distribution. Even at the parts-per-billion level, these contaminants can affect CMP performance and create microscratches that impact device yield.

A key to suppressing this impurity contribution is to have an intimate knowledge of how these impurities are created within the process used to manufacture the particles. In some instances, these lessons are learned through deviations in CMP polishing performance and are not easily predicted upfront.

Impact of nanotechnology research

A new breed of nanoparticle generation processes has been explored and developed over the last decade and a half due largely to the advent of nanotechnology research and applications [1, 2]. A variety of wet and vapor-phase methods have come into existence, with a select few reaching application at a commercial scale. The advantage of these processes over the conventional nanoparticle processes is the claim of a high degree of control over particle size, size distribution, composition, and degree of aggregation and agglomeration.

By definition, these processes are targeting the synthesis of particles <100nm, but due to the high associated costs of these materials, they are not easily substituted into existing slurries. In fact, a unique level of performance has to be achieved with these state-of-the-art materials for them to be considered for CMP application, which generally translates into getting more with less. Typical percent solids levels of commercial CMP slurries range from 2-25% by weight across many suppliers and applications. This range is moving to lower levels with next-generation products, however, largely due to cost savings and improvements in defect reduction.

Next-generation nanoparticle technology

Most CMP slurry suppliers at this point in time have made an investment in nanoparticle technology for the next generation of slurry products. These investments have come through partnerships, joint ventures, or acquisitions of particle technologies, many of which are undergoing commercial development.

Slurry designers always look for a broad range of chemical and physical properties, which are usually obtained in the small molecule and polymer additives that are formulated into the slurries. Being able to achieve chemical and mechanical functionality from the abrasive particle itself is an attractive proposition, compared to the current capabilities of fumed and colloidal particles.

One of the early entries from the nanoparticle platforms that is receiving attention and interest for CMP slurries is nanoparticle ceria (CeO2). Ceria has been effectively used for polishing glass materials for many years due to its chemical/mechanical interaction with hydrated silica films that form during polishing.

Conventional glass-polishing ceria slurries are far too aggressive for application in CMP because of the coarse particle sizes that cause extensive scratch defects. When generated in the nanoparticle form, however, ceria can be effectively formulated into CMP slurries with defect levels at or below those achieved with commercial silica-abrasive slurries. This performance is achieved with ceria at a fraction of the amount used in silica-based slurries. This improved performance enhancement combined with low solids is the type of response that will enable using nanoparticles as CMP abrasives.

Many success stories are in the making at CMP slurry suppliers using nanoparticle technology. For implementation in CMP slurries, the nanoparticle supply must be robust and well controlled, and the property enhancements must be realized with very low concentrations of these materials. Some of these new nanoparticle processes are being designed for CMP, in contrast to the historical particle processes that are adapted to CMP. Designer abrasives will continue to gain traction, and, at some point, will be required to meet the needs for next-generation semiconductor technology nodes.

References

1. A.S. Edelstein, R.C. Cammarata, Nanomaterials: Synthesis, Properties, and Applications, Institute of Physics Publishing, Philadelphia, 1996.
2. K.J. Klabunde, Nanoscale Materials in Chemistry, J. Wiley, New York, 2001.

John Parker is a senior scientist at Cabot Microelectronics Corp., 870 N. Commons Dr., Aurora, IL 60504; ph 630/375-5534, e-mail [email protected].