Issue



Remote clean improves Applied HDP-CVD


04/01/1997







Remote clean improvesApplied HDP-CVD

Applied Materials has produced a new remote plasma chamber as part of the company`s new production worthy Ultima HDP-CVD Centura system. The remote chamber is reportedly 100% efficient in dissociating NF3 into atomic fluorine gas for dry chamber cleans. Conventional chamber cleans create the plasma in the deposition chamber, which is less efficient and degrades chamber surfaces through exposure to harshly energetic ions and electrons.

Developed over a period of one year to improve electrostatic chuck (ESC) degradation during chamber cleans, the remote chamber uses a standard microwave source at 1-2 torr to generate its "soft" plasma. It is located <10 cm upstream of the process chamber (see figure).

ESC degradation resulting in deposition uniformity and particle generation problems has been a well-known problem for all HDP-CVD systems ("HDP-CVD: Trying to lasso lightning," SST, April 1996, p. 63). Like most system manufacturers, Applied uses an aluminum containing ceramic for the critical top surface of the chuck; during standard in situ plasma cleans, the top surface is degraded by the formation of aluminum-fluoride due to fluorine ion exposure. Since the ions and electrons are contained within the remote chamber, the neutral free fluorine does not have enough energy to react with the ceramic and no cover wafer is needed to protect the center of the chuck during cleans.

The chemical potential of free fluorine is still sufficient to etch away all glass deposited on edges of the ESC and any traces on other chamber surfaces. System literature claims 10,000 mean wafers between manual chamber cleans.

Minimizing the clean time seems to be the biggest remaining challenge with HDP-CVD chamber cleans, since clean time reduces system availability and directly increases cost of ownership. The most optimal set-up would allow for a complete dry-clean in the short time that the handler is shuffling wafers between chambers, but this is currently impossible. The main reason seems to be the difference between <5 mtorr deposition conditions and clean pressures. Higher pressures are needed for cleaning since the 1000-fold greater concentration of reactive gases allows cleaning to occur quickly. Dana Tribula, Applied Materials HDP-CVD Product Manager, said, "What`s tricky is the pressure ramps. The challenge is managing them for cleaning throughput." - E.K.