Analyzing available alternatives for point-of-use abatement
07/01/2003
Overview
The continually growing array of materials used in deposition and other wafer processing applications requires continuous evaluation of exhaust abatement methods. In addition, newer point-of-use methods of abatement may provide opportunities for cost savings. In any case, however, performance characterization of an abatement unit is an important activity for both existing and emerging processes to ensure there is no negative impact to abatement performance.
Point-of-use (POU) abatement is often used to minimize the overall environmental, health and safety impacts of various semiconductor processes, including deposition and etch. Unreacted feed gases as well as by-products can be removed or destroyed prior to entering the fab exhaust stream, thus preventing release to the environment. The potential of exposure to these compounds during exhaust duct maintenance activities can also be reduced if POU abatement is employed.
Analysis of abatement unit performance is important as new materials and processes are introduced into existing toolsets to ensure that the type of abatement technology is adequate to prevent impacts on the factory infrastructure. Opportunities for cost savings through operating temperature (energy) or water flow reduction may also be discovered with existing equipment or with newer technologies; however, performance must not be compromised.
As semiconductor technology advances, new materials and processes are continually being introduced. Transition metal oxides and silicates are high dielectric constant materials that are potential candidates to replace silicon dioxide in gate applications. Metals, such as titanium, tantalum, zirconium, hafnium, strontium, yttrium, and others, have been identified as possible components in high-k gate dielectrics. Metals such as titanium, tungsten, tantalum, ruthenium, and iridium may be used for complementary metal electrodes in advanced gate stacks as well as in barrier and conducting film applications.
Many of these processes will use metal organic, metal halide, metal nitrate and other precursors for MOCVD or atomic layer deposition (ALD). Emerging low-k CVD processes use silicon-based precursors with carbon and, in some cases, oxygen in various Si-C and Si-O bond arrangements. New silicon precursors are also being used for lower temperature depositions of traditional films. All of these new precursors for advanced CVD applications present challenges in determining EHS properties as well as characterizing and treating process emissions.
Emissions characterization
To determine POU abatement performance for existing and emerging CVD processes, various emissions characterization methodologies can be used. This characterization of advanced CVD processes serves several purposes. The most important function is to identify and, where possible, quantify the by-products of the CVD reaction. Emissions typically contain unreacted precursor, breakdown products of the precursor, and other by-products formed from reaction with co-flow materials. Emission characterization data can also be used to determine the level of hazard during chamber maintenance.
This data may also be useful in determining whether POU abatement is recommended and, if so, the type that is most appropriate. In some cases, abatement may already be required based on the precursor itself (highly toxic or flammable) or a co-flow material (e.g., ammonia). For cases where POU abatement is required, emissions data collected upstream and downstream of the abatement unit can be used to determine abatement efficiency. Emissions data can also be used for process diagnostics; in some cases, by-product emissions correlate with deposition parameters. The same characterization can be used for existing processes to verify abatement performance, particularly if operating parameters are modified for cost reduction purposes.
Using extractive FTIR, we have conducted several case studies of abatement performance for a single wafer CVD chamber with POU abatement. The analytical arrangement could also be used for two-wafer twin chamber or multiple wafer LP-CVD furnace applications. FTIR is useful for quantification of species, particularly if the species are known and calibration curves have been generated.
POU abatement technologies
Selection of a POU device for a new process is mainly based on the need to remove or abate the precursor and by-products in the exhaust, such as organic ligand or acid gases, or entrain the metal compounds. Three types of POU abatement technologies are available for various existing and emerging processes:
- a hot-bed dry reactor (HBR) with air addition,
- a resistively heated thermal abatement unit (RTAU) with water scrubbing, and
- a fuel-fired thermal abatement unit (FTAU) with water scrubbing.
The HBR (Fig. 1) consists of two stainless steel cartridges in parallel. Each cartridge has two stages: a reactive bed for heat transfer and thermal decomposition and calcium oxide for removal of acid gases and metal halides. This system can be equipped with air injection capability to ensure oxidative destruction of organic ligands present in both unreacted precursor and CVD process by-products. The reactive bed material is capable of removing halogens, such as chlorine and bromine, to prevent emissions of acid gases from the process. The bed material is also capable of removing the metal portion of unreacted precursor to prevent deposition further down in the exhaust stream.
null
The RTAU (Fig. 2) consists of a resistively heated column with a corrosion-resistant liner operating around 850°C. The process gases enter through the top of the unit and are mixed with air and N2 prior to the heater. After passing through the heater, the gas stream is quenched by a stream of water and particles are removed at the drain.
 Figure 2. Resistively heated thermal abatement unit. |
null
The FTAU (Fig. 3) uses hydrogen and oxygen or natural gas. The unit has an integrated wet scrubber that absorbs acid gases and removes particles. The scrubber can be configured with a recirculating alkaline liquid (e.g., KOH). The liquid is pumped from a tank and sprayed above the combustion chamber and the scrubber, allowing cooling of hot gases from the combustion area. Scrubbed gases exit the system via a de-mister in the main exhaust line. All components of the recirculation system are fabricated from corrosion-resistant materials. The automated recirculation system controls the flow and pH of the scrubber liquid. This type of abatement unit can treat emissions from multiple chambers, which can result in less water and energy used over multiple units handling only one chamber each. This unit can also enter into an idle mode when not actively abating, further saving energy.
 Figure 3. Fuel-fired thermal abatement system with a water scrubber. |
null
To demonstrate the viability of these POU abatement technologies, evaluations were conducted with various precursors, along with other gases used in deposition processes. To conduct these tests, two FTIR systems were used to measure emission species upstream and downstream of each unit. In some cases, unreacted precursor was first flowed through the chamber to obtain reference spectra for FTIR measurement, or to calculate dilution due to pump purge and abatement unit purge. We compared inlet and outlet data to determine the removal efficiency of metals, halides, and organics.
POU with CVD
Novel low-k CVD processes probably pose a bigger challenge to POU abatement technologies compared to high-k films due to the quantities of material deposited and their tendency to form polymeric by-products. A small percentage of break-through by these precursor and by-product materials can potentially have an impact on fab infrastructure by forming siloxane deposits, ultimately blocking exhaust lines.
Consider the exhaust from a CVD process using trimethylsilane where measurements are made both up- and downstream of a RTAU (Fig. 4). Correcting for the effect of dilution of the abatement unit as a contribution to destruction-removal efficiency (DRE), the CVD precursor is abated to only 90%. While this level of DRE may be adequate for some applications and materials, in this case it may be problematic due to potential reactions of the residual precursor with air, leading to downstream deposits. The abatement unit residence time and reaction stoichiometry may be inadequate for complete destruction of the process input gases. Methane and CO combustion and oxidation should occur efficiently. Possible improvements in performance might occur by tuning the device's operational parameters, using additional heat or oxidant, or improving process efficiency by reducing emissions at the chamber level.
 |
null
 |
null
We have characterized silicon nitride deposition using SiH4 and N2O on a dual chamber plasma-enhanced CVD platform equipped with an FTAU. Each chamber on the tool was equipped with a remote plasma chamber clean device using NF3, allowing abatement evaluation of both the deposition step and the subsequent chamber clean. Inlet taps were installed on the chamber exhaust forelines downstream of the pumps to accommodate FTIR testing. Two 2-in. ports downstream of the burner allowed fab air to be drawn into the exhaust to prevent condensation of moisture from the scrubber. These were temporarily 90% plugged to permit FTIR measurement downstream without excessive dilution.
As a test, NF3 was introduced to the device with the burner off and the FTIR was able to detect an intense signal using only 50sccm NF3. This sensitivity test confirmed that the device is abating to below detection limits for: SiF4, SiH4, HF, NF3, N2O (some N2O and CO are produced by the burner itself). For all concentrations of the respective deposition and chamber cleans by-products (Table 1), we observed an outlet concentration below detection limits, indicating >99% destruction-removal efficiency.
Overall, we have tested a variety of processes on the three abatement unit types. Figures 5 and 6 illustrate how the inlet and outlet streams of the abatement units, HBR (Fig. 5) and RTAU (Fig. 6) are characterized and compared to determine if the abatement unit is effectively removing the components of interest. Table 2 presents a summary of the abatement unit performance (three types) for the precursors tested [1, 2].
 Figure 4. Resistively heated thermal abatement unit performance with trimethylsilane. |
null
 Figure 5. Hotbed reactor performance for Hf-t-butoxide. |
null
 Figure 6. Resistively heated thermal abatement unit performance for bis (tert-butylamino)silane (BTBAS). |
null
POU abatement with LPCVD
POU abatement can be used in LPCVD furnaces depositing silicon nitride films using dichlorosilane and ammonia. Since the utility expense associated with operating POU abatement devices can be significant, we conducted a study to determine whether cost reductions were feasible.
As operating parameters were varied, we measured emissions from a thermal abatement unit with integrated water scrubbing. Some effort was made to measure the contribution of each abatement factor: dilution, heat, and water scrubbing. Both standard silicon and silicon-rich recipes were tested, with downstream dichlorosilane concentrations measured as high as 4% for the silicon-rich process.
 |
The RTAU device was found to be very effective under normal operating parameters, with a significant decline in performance when operated at room temperature (Table 3). Therefore, there was not an opportunity to reduce energy consumption by lowering the abatement unit operating temperature [3].
Conclusion
Performance characterization of POU abatement units is an essential analytical endeavor for both existing and emerging processes. As new processes and materials are introduced, their associated abatement unit performance requires characterization. Opportunities for cost savings with existing units can be evaluated to ensure there is no negative impact to abatement performance. In some cases, there may be opportunities to reduce the overall cost of ownership of POU abatement by moving to newer technologies that can treat emissions from multiple chambers, thus using less water and energy for the emissions load treated. It is also important to evaluate POU abatement to eliminate impacts on the factory infrastructure.
Brian Goolsby, Victor Vartanian, Laura Mendicino, Motorola DigitalDNA Laboratories, Austin, Texas
Acknowledgments
Digital DNA is a trademark of Motorola Inc.
References
1. V. Vartanian, et al., Proceedings of the 201st Meeting of the Electrochemical Society, PV 2002-15, p. 1.
2. L. Mendicino, et al., Proceeding SEMI Technology Symposium 2002, p. 31.
3. B. Goolsby, et al., Proceedings of the 201st Meeting of the Electrochemical Society, PV 2002-15, p. 77.
Brian Goolsby received his PhD in chemistry from the U. of Texas, Austin. He is a process scientist at Motorola, Semiconductor Products Sector, DigitalDNA Laboratories, 3501 Ed Bluestein Blvd, Austin, TX 78721; ph 512/933-7192, fax 512/933-6962, e-mail [email protected].
Victor Vartanian received his PhD in chemistry from the U. of Texas, Austin. He is a principal staff chemist at Motorola DigitalDNA Laboratories.
Laura Mendicino received her BS in chemical engineering from Ohio State U. She is the EHS manager at one of Motorola's 200mm fabs.