(255b) Development and Testing of a Robust Method for Destruction of Perfluorinated Compounds in Semiconductor Manufacturing

Perfluorinated compounds (PFCs) are commonly used in many manufacturing endeavors. Semiconductor manufacturing commonly employs a wide array of PFCs in wafer fabrication, and the industry's PFC discharges are a significant source of global warming gas discharge world-wide. The high global warming potential of these gases makes them a serious environmental concern. These gases are very difficult to abate in that they do not readily absorb, burn, or adsorb. Since the semiconductor industry has found it very difficult to find substitutes for these gases, there is a strong need to find a method to abate semiconductor process discharge in a fashion that captures or destroys the global warming PFCs. In order to continue to meet world-wide demand for semiconductor products it is critical that methods be developed to enable manufacturing of these products in a fashion that complies with global treaties on reductions in global warming gases.

Today the vast majority of global warming gas releases are in the form of carbon dioxide. PFCs comprise a small percentage of total global warming gases releases; however most PFCs have extremely long lifetimes and therefore have extremely high global warming impact. A number of approaches have been tried to reduce the PFC discharge from semiconductor factories. However, most of the currently employed technology is lacking in one or more areas. One method employed is the use of high temperature methane flame. This method has several drawbacks. First, the extremely high temperatures required demand a large flow of methane burnt in the presence of pure oxygen. This is expensive and results in the release of large quantities of carbon dioxide. Hence one global warming gas is destroyed at the expense of the generation of large quantities of another global warming gas. In addition, destruction of some PFCs such as carbon tetrafluoride is very difficult with methane and the efficiency at removing this commonly used PFC is frequently quite poor.

Another currently used solution for the abatement of PFCs is to route the reactor discharge through a plasma device. The problem with this methodology is that the plasmas require sophisticated maintenance technicians and the devices fail in environments where flow vacillates or in discharge streams where there is heavy fouling.

A third method to reduce semiconductor perfluorinated compound emissions is through the use of catalytic destruction. There are a number of challenges in developing a robust and sustainable catalytic abatement system for semiconductor process discharges. The first challenge is that the catalyst must be effective in promoting the decomposition of perfluorinated gases that are very stable and typically not very reactive. Second, if a catalyst is successful at abating PFCs it must be incorporated into a system that protects the catalyst from poisoning and fouling from the acid, halogen, and particulate laden semiconductor process tool discharge. Finally, the catalyst substrate combination must be capable of withstanding very high temperatures required to activate the reaction and must resist decomposition by the acid gases formed in the destruction of the PFC.

The benefits of using a catalytic reactor to abate PFCs are many. First, the system is relatively simple and is easy to operate. Once set in place it is easy to maintain the device's operating parameters so that the abatement tool functions at peak efficiency. Catalytic destruction is cost effective and results in minimal generation of other global warming gases.

The authors of this paper have engaged in research to determine an effective and robust process to destroy those PFCs commonly used in semiconductor manufacturing. We have conducted extensive laboratory testing, pilot reactor testing, and factory testing of abatement systems designed to destroy perfluorinated compounds by reacting the perfluorinated compound with steam over a variety of catalysts. Our testing has revealed that a select group of metal catalysts can help achieve high destruction efficiency for many common PFCs. We have also researched and developed suitable catalyst substrate composition and manufacturing procedures to ensure that the catalyst is economically feasible to operate in the harsh reactor conditions encountered during PFC destruction.

In prior testing we had determined that a reaction occurs between many PFCs and water at very high temperatures (> 1400 Celsius) that results in effective decomposition of the PFC into carbon dioxide and hydrogen fluoride. However, the energy required for this reaction is very high and results in a process that is undesirable due to the cost of energy, environmental impact, and reactor cost. We set forth to develop a catalyst to promote this reaction and reduce the energy and temperatures required to react the PFC gases with water.

Laboratory testing results

Preliminary testing was conducted across a broad array of catalysts. We placed all of the potential catalysts on an aluminum oxide substrate as prior work had convinced us that alumina had very desirable lifetime in the harsh conditions (high temperature / high fluoride) in which the catalyst would be required to operate. During bench scale testing we tested several noble metal catalysts including platinum and palladium. We also tested zirconium oxide, gallium oxide, gallium sulfate, aluminum phosphate, and aluminum fluoride. Later we tested catalysts created using combinations of those ingredients including a Pt/ZrO catalyst and a Ga sulfate catalyst. We conducted pilot testing at temperatures ranging from 20 C to 1000 C. During this testing we concentrated on determining the ability of each catalyst to promote the hydrolysis of three particular PFCs with water. Those PFCs were carbon tetrafluoride, hexafluoroethane, and octafluoropropane. The reason we tested these three gases is that these are the PFCs most commonly employed in the semiconductor industry. Our testing demonstrated that the gallium sulfate - aluminum oxide catalyst promoted the hydrolysis of carbon tetrafluoride and hexafluoroethane best.

Field testing results

After determining that the gallium sulfate on aluminum oxide catalyst had the highest efficiency at promoting the hydrolysis of the perfluorinated compounds we manufactured a suitable quantity of these materials for testing in an actual semiconductor factory. The catalyst was tested in an abatement tool that processed discharge from a silicon nitride deposition tool which employed hexafluoroethane as a chamber clean gas. The deposition reaction used silane, ammonia, and nitrous oxide. During the chamber clean 2500 sccm of hexafluoroethane is reacted with oxygen in plasma. The reactor dissociates 40 ? 45 percent of the hexafluoroethane into carbon monoxide, carbonyl fluoride, hydrogen fluoride, fluorine, and carbon dioxide. These gases react with silicon in the reactor chamber to produce silicon tetrafluoride gas. The un-reacted C2F6 is also part of the discharge stream. This discharge is blended into a nitrogen stream that is supplied at a rate of 120 liters per minute. This mixture is then sent to the scrubber for abatement.

In order to ensure that the catalyst had an acceptable lifetime we incorporated the catalytic reactor in a system of abatement tools where several pretreatment steps were accomplished on the semiconductor discharge prior to catalytic treatment. First, the discharge stream was reacted with air at high temperature to burn off any flammable gases that might generate particulate. This is important because the reaction of silane in the catalyst bed itself would result in rapid fouling of the catalyst. We then employed a particulate scrubber to remove solids from the process stream.

The next pretreatment pre-scrubbing module removed silicon tetrafluoride, fluorine, hydrogen fluoride, and carbonyl fluoride. This pre-treatment phase is critical because the non-PFC fluorinated compounds would quickly destroy the catalyst substrate and bind active sites on the catalyst surface. We developed light-off curves for the catalyst in this process which demonstrate that at temperatures in excess of 900 Celsius the hexafluoroethane is destroyed with efficiencies in excess of 99%.

One notable fact regarding the factory testing is that the flow and concentration of hexafluoroethane was significantly higher than the pilot scale flow rate or concentration. Despite the higher flows and concentrations destruction efficiencies remained high.

Conclusions

Our tests demonstrated that perfluorinated compounds could be effectively destroyed in a reactor by the reaction of the PFC with water over a gallium sulfate ? aluminum oxide catalyst. Destruction efficiencies of 95 % could be achieved on an industrial scale at temperatures of 850 C and destruction efficiencies in excess of 99% percent could be achieved on an industrial scale at temperatures of 900 C. This has major ramifications for firms seeking to limit their discharge of global warming gases. This system has demonstrated high reliability abating semiconductor discharge in an actual manufacturing environment for over 18 months. The mean time between service for this system is a minimum of twice the preventive maintenance interval for the process reactor which makes the abatement system transparent to facility operation.