Damage During Cleans Evaluated by AFM

An increasing problem with no known solution is the damage of small structures during wafer cleaning. The damage is seen as pattern collapse or simply as missing structures, blown off the surface of the wafer by the force of the clean. It's already a well-known result of aggressive megasonic cleans, but can also result from aerosol jet cleaning and newer laser-induced plasma shockwave cleans. "There is not solution at this point," said Ahmed Busnaina, the director of the Center for Microcontamination Control at Northeastern University (Boston). "People are looking at ways to adjust and be selective, but I don't think there is a solution on the horizon yet."

Busnaina and fellow researchers from Hanyang University (Ansan, Korea) and IMEC (Leuven, Belgium) have recently conducted research to quantitatively measure the amount of force that results in damage to various types of structures, including photoresist and gate stacks as a function of linewidth.

The researchers used an atomic force microscope (AFM) to simulate and measure wafer cleaning damage. They were also able to get an idea of how the damage occurred. They found that SiON/polySi/SiO₂ hard mask gate stacks were damaged from bending and delamination; SiO₂/polysilicon lines were broken and photoresist lines were torn and deformed. The researchers also observed a trench on the underneath pattern structure when the structure was collapsed. Photoresist patterns were weaker than SiON/polySi/SiO₂ hard mask gate stack and SiO₂/polySi stack. The collapse force of a SiON pattern, which was a 50 nm linewidth, was 23 μN and linearly increased as a function of the linewidth of patterns.

"We're closer to a solution because people are starting to study and try to understand damage," Busnaina said. "One of the solutions people are looking at is using more chemistry and less physical force. If we can increase the chemical part of the cleaning to maybe 90% and the physical part to 10% instead of 90%, then we can prevent damage."

Researchers are trying to optimize the amount of force used during wafer cleaning to effectively remove particles but avoid damage. Recent research shows removal force of two-days-aged PSL particles on hydrophilic silicon was 180 µN — two orders magnitude lower than the pattern collapse force.

Part of the challenge lies in providing enough force to overcome adhesion forces to remove particles, but not so much force as to damage the structure. Typically, the damage force is 100-1000× more than the adhesion force for the particles, Busnaina said, but notes that smaller structures are more easily damaged. "A small particle gets a very tiny fraction of the force applied, but the structure has a large area so receives a lot of force. It's very simple but that doesn't make it easy," he said.

The amount of force required to remove a particle is a function of adhesion forces that are, in turn, a function of the particle's surface area as well as Van der Waals forces. Busnaina said that Van der Waals forces were relatively tiny and decreased linearly with the size of the particle. "There's a misconception that nanoparticles are very difficult [to remove] because the adhesion force is so high. That's not true at all," Busnaina said. "The adhesion force of a 1 μm particle is 100× larger than for a 10 nm particle." In the AFM work to be presented next week, the researchers showed that a removal force of two-days-aged polystyrene latex (PSL) particles on hydrophilic silicon was 180 μN and it was two orders magnitude lower than pattern collapse force. PSL particles are commonly used in such studies as a standard that mimics how real particles behave.

The real reason smaller particles are more difficult to remove, he said, is that it's simply more difficult to apply force to them. "Even though the adhesion force is 100× smaller for a 10 nm vs. a 1 μm particle, the removal force has to be about 10,000× larger," he said. "You need to change the removal force by a large amount so that part of it will catch the particle. If you can get to the particle and apply the force directly, then you don't need such a large force, but the problem is we don't do that. We have to apply it through a fluid or a gas."

Busnaina believes the solution to the damage problem lies in better understanding of the fundamentals. "For the last 10 years, the damage process was not understood at all because nobody actually looked at the fundamentals. Now there's really a big push to look at the fundamentals and try to understand exactly how and why it happens."

Find more information on wafer processing.