 |
 |
 |
 |
|
Comments? Suggestions? Send us your feedback.
MicroMagazine.com |
PROCESS EQUIPMENT—ETCHControlling corrosion, particles in metal etch-and-strip cluster tool systemsShams Tabrez, Changhun Lee, David Yang, and Yun-Yen "Jack" Yang, Lam Research Many benefits can be derived from integrating two metal etch and two strip process modules on a single platform. For example, an increase in throughput per square foot of floor space is an important advantage because of the spiraling costs of new fab construction. Clustering serial processes reduces cycle time and the number of stations that operators have to supply with wafers. Cluster tools also provide ways to reduce particulate and molecular contamination and lower defectivity at both the chamber and system levels. Finally, on metal etch systems, clusters lessen the risk of corrosion from residual chlorine, since wafers are moved between etch and strip processes under continuous vacuum without exposure to moisture in ambient air. To obtain all the benefits of integrated serial processing, additional burdens are placed on the strip module design. For the etch module to operate continuously, the combined strip cycle and wafer transport time must be shorter than the time it takes to etch a wafer. This means the stripper must be fast and efficient so the highest possible level of system throughput can be achieved.
This article examines design considerations that permit an integrated serial etch-and-strip system to maintain high throughput and productivity while still guaranteeing a particle- and corrosion-free product. Particle and corrosion control findings will be presented that apply to both the process chamber and the integrated system. For reference, a serial metal etch-and-strip system, the TCP 9600PTX 2 x 2 with two metal etch process modules and two strip modules on an Alliance cluster platform (all made by Lam Research, Fremont, CA), is shown in Figure 1. Particle Contamination Cluster systems implement particle control at both the process module and system levels. Generally, clusters help reduce particles because two processes are served by one set of loadlocks, which lessens the number of times wafers are exposed to atmospheric contamination. System-level precautions can be implemented inside the loadlocks to further control contamination. Pressure differentials can be established among the transport module, loadlocks, and process modules via a continuous nitrogen purge. The pressure differences between different parts of the system can be used to sweep particles into the nearest turbomolecular pump. In addition, a difference in pressure between the transport and processing modules can help confine corrosive gases to the process chamber. Inside the chamber, pressure and temperature changes that loosen particles most often occur when the chamber is opened. Carefully designed hardware and controlled operating procedures can prevent contamination during this critical stage of the etch step. The following process module precautions also help minimize or eliminate particle contamination:
Table I: Comparison of particle performance using a pendulum valve and butterfly valve. Corrosion Contamination at the Wafer Level After metal etch, chloride usually exists on the wafer in the form of aluminum chloride (AlCl3), the principal etch by-product. Corrosion occurs when chloride compounds, which are absorbed onto the wafer surface during metal etch, hydrolyze when they react with ambient moisture. Hydrochloric acid (HCl) is then generated, which attacks the aluminum sidewalls. The following equation describes the corrosion reaction:
The return arrow in the equation identifies the iterative reaction between AlCl3 and moisture that generates HCl and aluminum oxide. The aluminum oxide stays on the surface as a stable reaction product, but the AlCl3 starts the corrosion reaction over again. As shown in Figure 2, the resulting aluminum oxide that grows out of the film can short between traces.
Removing Chlorine during Photoresist Strip The only way to prevent corrosion after metal etch is to remove the residual chlorine from the wafer. After an initial preheat step, photoresist stripping typically consists of a two-step passivation and strip process. The preheat step consists of elevating the strip chamber to a uniform temperature that volatizes AlCl3 so it can be vented through the chamber exhaust. A passivation step then exposes the wafer to deionized water vapor that removes residual chlorine trapped in the photoresist. An onboard system delivers water vapor at the volume, pressure, and temperature specified by the recipe during this step. Passivation is followed by the active strip step in which atomic oxygen from the microwave plasma is used to burn off the organic-based photoresist at elevated temperatures. The strip process also reduces the remaining AlCl3 to Cl2, which again is exhausted. The typical steps required for photoresist strip and the range of typically used process parameters are listed in Table II.
Table II: Typical steps required for photoresist strip and range of process parameters. The contribution that each step shown in the table makes to chlorine removal is shown by the following first-order reactions, where Ln([Cl]0/[Cl]) = kt:
Figure 3 shows variations in residual chlorine levels following the different postetch treatment steps. Approximately 2800 ng/cm2 of chlorine remains after the metal etch process. After preheating the wafer at 250°C in the downstream microwave stripper for 15 seconds, the trace chlorine level was reduced to 1300 ng/cm2. Further postetch treatment with H2O and O2/N2 plasma—based ashing subsequently reduced the residual chlorine to levels below the corrosion threshold.
Microwave Strippers and Corrosion Control In the strip chamber, some embedded chlorine is driven off by heat, some by passivation with vaporized DI water, with the remainder expelled as a gaseous by-product when the photoresist is ashed with an oxygen-rich plasma. A microwave source is characterized by plasma density of 1011 cm—3 at the source region. Downstream from the source, microwave plasma is dominated by atomic oxygen, a neutral species that ashes the organic photoresist. The neutrals also produce high strip rates on the order of 5 to 8 µm/min on blanket resist. The recombination of the chemistries in the downstream region prevents high-energy ions from reaching the wafer surface, thus eliminating the potential for device damage to thin gate oxides (<60 Å). In addition, a baffle located between the plasma and wafer blocks UV radiation from the microwave source to prevent device damage, and it diffuses species evenly over the wafer surface. The reactants' composition is well established by the emission spectra of the discharge and the measurement of ion density at the wafer surface. At the wafer surface, the high ratio of neutral species accomplishes the basic reaction required in a stripper, which is expressed as:
Figure 4 shows an open strip chamber with its diffusion baffle above and a wafer in a transport arm below.
At two production sites where the stripper had been installed, throughput was the priority strip requirement. The newly designed microwave, a small-volume chamber that can be pumped down quickly, met throughput requirements in terms of strip rate and process cycle time. The etcher, not the stripper, became the critical path in the four-module system. When used as a replacement for the previous stripper on stand-alone modules, the microwave stripper lowered strip process time by 30—50 seconds, which increased throughput by 2.5 to better than 3 wafers/hr. This increase limited the integrated serial process etch rate, rather than the strip rate. Equally important, the wafers passed 24-hour wet-box tests without corroding, indicating that the passivation cycle had effectively eliminated the chloride residuals. To accomplish this, the factors that contribute to corrosion were carefully studied, and the hardware and process were designed to remove chloride residuals as quickly as possible at each process step. Residual Chlorine and the Threshold of Corrosion In order to understand the strip process and to design the most efficient strip chamber in terms of corrosion control, the concentration of chlorine left on the wafer after each of the three strip steps was studied. A logarithmic scale ratio of initial to residual chlorine after treatment was plotted against each process step. The results showed a straight line for both the H2O passivation and O2/N2 processes, which indicates that chlorine removal reactions are governed by first-order kinetics. All corrosion control steps can be treated as first-order reactions. This means a chlorine concentration threshold exists below which the chlorine by-product will not combine with moisture in the air quickly enough to cause corrosion problems. Once the threshold is identified, process steps can be developed to reach it within the shortened process time allowed for maximum system productivity. To determine the threshold level of corrosion-inducing residual chlorine, wafer samples were prepared and analyzed using small-extract ion chromatography (IC). The small-area IC extraction analysis indicated that the wafer edges have a higher chlorine concentration than the center. From these tests, the critical chlorine level needed for corrosion to occur could be determined from a wafer sample that showed corrosion at the edge, but was corrosion free in the center. If the chlorine level could be lowered to between 5.9 and 6.9 ng/cm2, no corrosion occurred, even after the standard 24-hour wet-box corrosion test. Further investigation revealed a range of 6 to 18 ng/cm2, which coincides with other investigations that found a range of 6.8 to 15.3 ng/cm2. Experimental data could then be extracted after heating, passivation, and plasma stripping to determine which combination of hardware and process was the most effective in lowering chlorine below the threshold levels and in preventing corrosion on the finished wafer. The chamber and processes could then be designed to implement the most efficient strip and corrosion-preventive chamber. Integrating Processes One fairly obvious way to eliminate corrosion is to transport wafers under vacuum on a cluster platform. There are more subtle, less obvious relations between the etch and strip processes also worth considering. During metal etch, polymers form on the etched metal sidewalls. The polymers prevent the etch from becoming isotropic and protect the underlying metal film from chlorine attack. If the protective polymer layer is too thin or becomes damaged by energetic ions, chlorine can attack a sidewall locally and form voids in the aluminum. In the scanning electron microscope (SEM) image in Figure 5, three voids appear as dark spots on the sidewalls, in the upper- and lower-right-hand corners, and below the left center.
When many voids appear, a high corrosion rate at the void sites can usually be found. On some test wafers, all voids were accompanied by corrosion. In addition, electrically connected metal structures, such as those attached to an arsenic-doped negative well, tend to be more susceptible to corrosion than floating structures. This is probably because of the attraction of the aluminum-corroding electrochemical processes to negative potentials. Even after a final wet strip, corrosion can happen at a void site. It is not clear why this occurs, perhaps entrapped gas in the void prevents the wet chemicals from reaching the aluminum surface. However, a number of precautions can be taken during the etch step to ensure that the process does not pass voids onto the stripper. Lower chuck temperature reduces sidewall voiding, probably because of the increased polymer formation at lower temperatures. Lowering the chuck temperature does require a trade-off with critical dimension etch bias and microloading, both of which decrease at higher temperatures. Increased microloading may require extending the overetch cycle to clear the loaded areas. Lower chamber pressure also prevents voiding. Lower pressure provides a longer mean free path for the ions, decreasing the chance of the ions hitting the sidewalls and removing the protective polymer. This approach also involves a trade-off. Since chemistry, not ion energy, drives metal etch, the lower pressure tends to reduce the etch rate. Although other controls can be implemented during metal etch to enhance the polymer coating, temperature and pressure have the greatest influence. Keeping those factors under control during the etch process eliminates the creation of voids and simplifies the sequential strip process. The etch and serial strip processes are thus closely related through void formation as well as the necessity of protecting wafers from ambient air between the processes.
Final Results The SEM image in Figure 6 shows corrosion-free 0.5% Cu/Al metal-1 lines after 30 seconds of H2O passivation and 60 seconds of O2/N2 chemistry strip. In later tests, the passivation time was kept at 30 seconds, but the strip process time was reduced to 30 seconds. With either recipe, the wafer remained corrosion free after a solvent wet clean and a 72-hour wet-box test. The shorter process time translates directly into higher throughput at the system level without the risk of corrosion. These findings are also significant since most process engineers would agree that a short plasma exposure is desirable, even though microwave is known as a damage-free plasma source. At the same time, the 30—60-second plasma exposure window provides a wide latitude for recipes that may be needed in difficult cases. Conclusion Because of the corrosive nature of metal etch chemistry, immediate stripping is critical to a successful metal etch process. A serial etch system that mounts two metal etch and strip modules on a single cluster platform and transports wafers between processes under continuous vacuum successfully prevents contamination. Since corrosion cannot be tolerated in metal etch processing, the 2 x 2 configuration satisfies this important criterion. System design requirements must then address processing cycle times to attain high productivity and throughput. Acceptable cost of ownership figures depend on attaining the highest possible throughput, which means the system must be designed so that the primary etch process, not the stripper, imposes production limitations. The final design step involves integrating both processes on the platform in terms of timing cycles, chemistry, and recipe development to guarantee process excellence as well as the highest possible throughput. Shams Tabrez is part of the corporate marketing group at Lam Research (Fremont, CA). With eight years of experience in the semiconductor and semiconductor equipment industries, he worked previously at National Semiconductor (Santa Clara, CA) in marketing and general management within the semiconductor packaging and interconnect technologies divisions. Tabrez has a BS in chemistry from the University of Manchester (UK), a masters in chemical engineering and economics from Imperial College, University of London, and an executive MBA from the University of Chicago Graduate School of Business. (Tabrez can be reached at shams.tabrez@lamrc.com.) Changhun Lee, PhD, is a senior process engineer in conductor etch R&D at Lam Research. He has been involved in the development of microwave-based dry strip processes since 1995. Before joining Lam, he was a senior researcher at the optodevice lab at Samsung Advanced Institute of Technology (Kihueng, South Korea), where he was involved in R&D for optical devices for telecommunications. He holds a BS in chemistry from Korea University and a PhD in physical chemistry from the University of Iowa. David Yang, PhD, is also a senior process engineer in conductor etch R&D at Lam Research. He has been involved in postetch process development for TCP 9600SE and 9600PTX since 1996. He received a BS in physics and an MS in material science from Chinghua University (China) and a PhD in materials science from UCLA. Yun-Yen "Jack" Yang, PhD, is a process engineering manager in conductor etch R&D at Lam Research. He has managed process development of postetch treatment using RF- and microwave-based downstream dry asher integrated with TCP 9600SE and 9600PTX since 1993. Yang received a BS in chemistry from National Taiwan University and a PhD in physical chemistry from UCLA.
MicroHome |
Search | Current Issue | MicroArchives Questions/comments about MICRO Magazine? E-mail us at cheynman@gmail.com. © 2007 Tom Cheyney |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Figure 1: Serial metal etch-and-strip system on a cluster platform.
Figure 2: SEM image showing corrosion on aluminum oxide traces.
275°C
Figure 4: A downstream strip chamber with diffusion baffle and transport arm holding wafer.
Figure 5: SEM image showing sidewall voids appearing as black spots.
Figure 6: SEM image showing corrosion-free metal 1 lines on a patterned wafer.