Improving Thermal Cleaning Processes

Chamber cleaning is a required periodic maintenance operation for all CVD systems. In general a cleaning process using gases is much preferred to a chemical clean approach. Chemical cleaning usually requires opening the chamber to atmosphere, dismantling it for off-site cleaning with aggressive chemicals, re-installing a clean chamber kit, and then drying, pumping down and requalifying the chamber. Such a process suffers from many drawbacks, such as the need for maintenance personnel, the uncertainly of reassembly of all parts (leaks, parts management etc), the management of a second chamber kit (quartz, boats), and the exposure risk associated with highly doped processes.

In-situ chamber cleaning alleviates many of these drawbacks by removing deposits within the process chamber without affecting the integrity of the tool. Depending on the tool type (batch vs. single wafer for instance) and construction material, thermal cleaning (i.e. heat is the main activation factor of the cleaning gas) or plasma cleaning have been successfully applied throughout the industry for many CVD processes and for almost all type of materials (poly, oxide, nitride, etc.). A variety of chemical sources are being used to generate F atoms, which is the critical intermediate in all cleaning processes, with the notable exception of HCl, which is commonly used to clean Si in Epi reactors. Thermal cleaning processes are essentially based on NF3, F2, F2+HF, and ClF3, while NF3 and more thermally stable PFCs such as C2F6, CF4, C3F8… are being used on plasma enhanced cleaning recipes.

Thermal cleaning processes are especially desired for CVD processes that are not plasma enhanced to avoid the investment in the plasma generator and for large batch furnaces in which the plasma source is difficult to integrate with the quartz assembly. Also, cleaning uniformity is more difficult to achieve in such large furnaces, even with remote-plasma activation of the cleaning gas, due to the difficulty of evenly distributing the active species throughout the chamber.

When the chambers are not compatible with high temperature, an in-situ capacitively coupled plasma source is commonly used, while a remote plasma clean system is effective in increasing the dissociation of NF3 and hence the etch rate of the residues.

As a result, batch systems today using gas cleaning rely upon thermal energy to provide the F atom for the cleaning process. Figure 1 shows a picture of a 200mm quartz boat used in production. This boat had received 50 consecutive NF3 cleans and did not show signs of roughening. At the time of the picture, the system had operated for 14 months of production without one single wet clean while maintaining a particle level <30adders at 0.08µm with an NF3 clean every 6 runs.

While thermal cleaning is indeed possible and implemented with NF3, the process can be improved as the thermal dissociation of NF3 in typical cleaning conditions (pressure and temperature) is rather low. Increasing the thermal clean temperature and decreasing the pressure are means of increasing the thermal cleaning rate by thermodynamically favouring the dissociation of NF3. However these conditions need to be balanced with the loss of etch rate due to the density decrease at lower pressure, and the temperature ramp-up / ramp-down time. Also, harsh cleaning conditions tend to yield lower etch selectivity versus the quartz, yielding roughening of the chamber surfaces that may lead to particle excursions and film contamination due to out gassing of fluorinated species.

ClF3 is a thermal cleaning gas that will disassociate more readily than NF3 at the conditions that are used, however in many locations the requirements for the implementation of ClF3 are very stringent and costly. Pure F2 can also be used as a thermal cleaning gas, with improved etching rates at the same condition compared to NF3.

In spite of the relatively low dissociation efficiency of the thermal cleaning process, it is however successfully applied as the tool productivity and uptime can be increased compared to wet cleans.

Air Liquide has recognized the benefits of thermal cleaning for a number of years. In order to improve the performance of the thermal cleaning process Air Liquide has developed additive gases(1) that improve the performance of this process (etch rate, selectivity, etc.), and further contribute to favouring in-situ cleaning versus wet cleaning.

Table 1 shows a comparison of results in equivalent temperature and pressure conditions, between two different cleaning compositions, the F2 + HF mix being the initial tool recipe, and the F2 + additive the results of Air Liquide’s process optimization.

In recent years we have extended this work to the thermal cleaning of new materials such as TiN and TaN. An initial success was the implementation of a gas phase cleaning process for TiN(2). The existing gas phase cleaning process was difficult to implement at the site, and wet cleaning was used. The TiN is an IR reflective layer and by implementing a gas phase cleaning process, it was feasible to run a clean after each run which improved the reproducibility of the process. Further developments by Air Liquide identified a new additive species that dramatically improved the etching rate as shown in Figure 2. Over an order of magnitude improvement in the thermal cleaning rate was achieved using a trace additive to the F2 cleaning gas.

Similar additive and recipes have been successfully developed for TaN with performance improvement results as shown in Figure 3.

The additive gases provide a drop-in performance benefit as the implementation on an existing tool is straightforward, requiring minimal hardware changes. The performance benefits rapidly pay for the small capital investment while the gas consumable costs are in general decreased due to the much more effective usage of the F atoms that are introduced into the reactor.

The higher efficiency usage is beneficial for the customer and the environment (as much less emissions are created). The dramatic process improvements are examples of how Air Liquide brings its technology to bear to improve customer’s process and benefiting the environment.

Acknowledgements
The authors would like to acknowledge Helmuth Treichel and Thierry Lazerand, both of AVIZA
TECHNOLOGY for the fruitful exchanges on the subject.

References
(1) Method of cleaning a film-forming apparatus and film-forming apparatus”, Y.Sato, N.Tamaoki,
S.Seta, R.Zils, J.Sonobe, T.Kimura, K.Momoda, US2005/0082002.
(2) ISESH 2005 – Portland, Oregon. June 19-23 - Session PFC. ClF3 ALTERNATIVES FOR INSITU
ALD TITANIUM NITRIDE CHAMBER CLEANING
Nicolas Blasco, Brian K. Raley, Kayo Momoda, Julien Gatineau, Nicolas Jourdan, Cécile Miquel