Single Wafer Ozone Clean Done Without Megasonics

With the transition to 300 mm, single-wafer cleaning has proved its merit. Conventional ozonated DI processes used in single-wafer cleaning have limitations such as high pH for particle removal and rapid ozone decay at elevated temperature applications. To counter these issues, one alternative is the OzoneJet ozone clean approach developed by NOVO Research Inc. (San Jose), which achieves high particle removal efficiency without megasonics. In fact, experimental results showed particle removal performance of >95% in 110 sec.

It was accomplished using ammonia chemistry with ozone. This OzoneJet clean concept is not limited by the pH and temperature, and could also be applied to photoresist strip. It does not use H₂SO₄, H₂O₂ or HCl. The amount of DI and HF consumption is ~2% of the traditional RCA clean — making it more cost-effective than conventional RCA cleans.

This cleaning method delivers the ozone in its gaseous form through multiple orifices in a nozzle directly to the wafer surface, while treating the wafer with chemicals. The pressure of the individual columns of ozone gas coming out from the orifices parts the surface chemical layer at multiple points of contact, making possible the direct reaction between the highly oxidative ozone species and the wafer surface. The nozzle runs along the radius as the wafer rotates, covering the whole wafer. Multiple nozzles can be used for higher efficiency. Therefore, a given point on the wafer experiences repeated exposures to chemical and ozone. Other processes use ozonated DI or diffusion of ozone through a diffusion layer, but the ozone effect is limited by ozone decay and low ozone concentrations.

The baseline process consists of consecutive applications of HF OzoneJet, DI rinse, ammonia OzoneJet, DI rinse and Rotagoni drying (total process time is 107 sec). It could be applied to FEOL, BEOL and post-CMP cleaning with appropriate chemistry. One other variation used hot DI instead of room-temperature DI, resulting in a significant increase in the photoresist strip rate. The decay issue in high pH and high temperature can be ignored because gas phase ozone is used, not dissolved ozone.

Particle removal efficiency tests were conducted on bare silicon wafers with >5000 silica particles. Particle measurements were done for particles >0.2 µm. Figure 1 shows the particle removal efficiency of HF, HF OzoneJet, HF OzoneJet plus ammonia, and HF OzoneJet plus ammonia OzoneJet. Results indicate HF or HF OzoneJet particle removal efficiency does not increase significantly with HF (0.35 wt %) time up to 50 sec. The addition of ozone to the HF process enhances particle removal efficiency. Furthermore, the addition of ammonia to the HF OzoneJet process enhances the particle removal efficiency to ~80%. Particles are removed and prevented from re-adhesion by the ammonia's zeta potential. Results show that the addition of ozone to ammonia increases the particle removal efficiency to >95%. Non-optimized HF OzoneJet and ammonia OzoneJet resulted in pitted bare silicon surfaces.

1. Particle removal efficiency of various processes. (Source: NOVO)

NOVO researchers also evaluated pattern damage (Fig. 2), following high-pressure test on an aluminum patterned wafer. The pressure used for the experiment was about 2× higher than that for the OzoneJet method. As can be seen from the SEM, the effect of high pressure of the OzoneJet on the wafer surface is non-destructive. No pattern damage was observed on the entire wafer.

2. Cross-section SEM micrograph after a high-pressure OzoneJet process. No pattern damage was observed on the entire wafer by the SEM inspection. (Source: NOVO)

"NOVO's OzoneJet process, which lowers the chemical boundary layer down to nearly zero and controls the effect of ozone and liquid chemicals independently, will impact the cleaning market significantly," said Yongbae Kim, NOVO's founder and president.

For additional information on clean processing, go to www.semiconductor.net/clean