Lasers Could Perform On-line Contaminant Identification

In semiconductor manufacturing, problems on a production line can potentially send millions of dollars down the drain. Particle contamination is just one possible problem, but certainly not an insignificant one. Particularly with the introduction of 300 mm wafers — each yielding perhaps a couple hundred pricey chips — damaged wafers, and/or any downtime associated with producing the wafers, could translate into an extremely costly mistake.

Engineers at Purdue University (West Lafayette, Ind.) have developed laser-based methodologies to save millions of dollars worth of downtime every year. Lasers are already used by semiconductor manufacturers to detect dust particles on silicon wafers. But detecting those particles is just the first step; chipmakers must shut down production lines to determine what's causing the contamination so that they can put a stop to it. What the researchers at Purdue propose is a method that uses lasers for the next level as well — identifying the contaminant and its source during the in-line laser inspection process. The same lasers used for detection could be used for identification, tracing the contamination source within seconds.

E. Dan Hirleman, a professor and head of Purdue's School of Mechanical Engineering, and doctoral student Haiping Zhang have developed mathematical models that reveal a particle's identity by how it scatters laser light. Each type of material has its own "fingerprint." Such scattering techniques — in which laser light is monitored as it bounces off a given particle — are not particularly new, having previously been used in a variety of environmental and atmospheric applications. But the researchers in this case have developed and modified modeling techniques for use specifically with semiconductor wafers.

These graphs illustrate differential scattering cross sections for the ring region for a 0.305 µm PSL sphere above a silicon substrate with 0.25 µm SiO2 film. Incident beam is at 70° incidence and l=632.8 nm. The top graph is s-polarized, and the bottom graph p-polarized. (Source: Purdue University)

To accelerate the computation of the system linear equations in DDA, the researchers applied a 3-D fast Fourier transform (FFT) method. Although the 3-D FFT method is typically used for free-space measurements — a 2-D FFT method is typically applied for particles on a surface — the researchers found that the 3-D method worked well for surfaces, with or without film layers.

From all this, they have developed a code called DDFILM — based on the DDSURF code developed by Schmehl and Nebeker, modifying the surface interaction terms, and applying a 3-D FFT routine.

The Purdue researchers compared their DDFILM modeling results with experimental results obtained at Arizona State University (Tempe, Ariz.) with a scatterometer. The scatterometer uses a photodetector consisting of 32 rings and 32 wedges that detect angle-resolved scattering patterns from the sample. The Figures compare the DDFILM model and the scatterometer measurements, showing angle-resolved differential scattering cross sections. The results shown are for the scattered field from a 0.305 µm polystyrene latex (PSL) sphere on a silicon surface coating with a 0.25-µm-thick SiO₂ film.

The Figures show that the two results are in good agreement, validating the researchers' model. They also compared their modeling results with results calculated by a code called POLAR. In this case, the two numerical models agreed perfectly.

The researchers presented their results at the Conference on Design, Process Integration, and Characterization for Microelectronics. Co-located with this year's SPIE Microlithography in Santa Clara, Calif., the conference was the first of its kind. Such a conference was needed because of the growing need for measurement and sensor techniques in nanotechnology, Hirleman noted.

At the 70 nm node, for example, circuits could be ruined by dust particles as small as 30 nm. Purdue researchers also are working on a deep ultraviolet (DUV) scatterometer that could detect smaller particles.

For additional information on emerging technologies, go to www.semiconductor.net/emerging