Solving Particle Issues in Epitaxy: SiC vs TaC Coating Guide

In epitaxy, it's the small particles that can cause a big problem. They can form on wafers, serve as nuclei to stop crystal growth, or create defects that would not allow yield. When dealing with high temperature systems and coated components, many engineers encounter this problem. In this area, SiC and Tac coating are two generally used materials. Both designed to protect the graphite parts in the reactors, yet with varying properties when subjected to extended production cycles, heating and gassing. In the case of particle problems, the selection of coating may be an important consideration. Understanding of the behaviour of SiC and TaC under real production conditions enables a reduction in the number of defects and a stabilisation of the production processes.

Root Causes of Particle Issues in Epitaxy: From Coating to Gas Flow

Particles generally have many origins in epitaxy. Generally depends on coating condition, gas behaviour in the coating chamber and gas flow. In case of small flakes or dust appearance on the wafers, engineers tend to verify the coated parts first, especially SiC or TaC coated graphite parts. Wear of the coating is common. Sic coating parts may develop small cracks over time under prolonged exposure to high temperatures. The cracks in these can be hard to notice at first, but will allow small particles to enter the process chamber over time. TaC coatings tend to be more temperature tolerant, but have some issues. If there is a low degree of bonding between TaC and graphite, small areas will start to lift off, after a few heating cycles. Thermal cycling is a large consideration, too. The tools in an actual line are not kept at a constant temperature. Many times throughout the day they heat up, cool down and repeat. Stresses are experienced in each cycle, both in the coating and inside of the base material. Higher wafer throughput settings in a factory are likely to have higher particle events due to the fact that such equipment is not fully stabilized. Flow of gas within the chamber is also involved. Particles may be picked up and taken directly to the wafer if the flow is not even; these particles can be formed on the chamber wall and the parts surface. One fab team found that there was a small misalignment of a gas inlet and this caused turbulence near the edge of the wafer. The particle count decreased when the flow direction was changed but no hardware changes were implemented. Another hidden cause is during maintenance, when the water can be contaminated. If not controlled by clean-up tools and/or practices, a thin layer of small debris may remain on coated surfaces. These bits can later come loose when they are run at a high heat. Overall, taking into account the coating condition, thermal behavior, gas flow and clean-up, will reduce the particle issues. A single area of development is unlikely to give the desired result.

CVD SiC Coating vs. CVD TaC Coating: Surface Stability and Particle Generation

When people compare the CVD SiC coating and the CVD TaC coating in epitaxy tools, they will generally end up considering the stability of the surface during long periods of use, and the amount of particles produced over time. Although both coatings can be used to protect graphite parts, they do not perform the same function when subjected to repeated heat cycles and to reactive gases. The coating method is widely used due to the production of a hard smooth surface as with CVD Sic epitaxy coating. Initially, it can work quite well and maintain low levels of particles. The problem appears after using it long time. Slowly, under high temperature and fast cycling, SiC can develop micro-cracks. Initially these cracks are small, but can expand with stress. In such cases, very small pieces might chip off and fall into the chamber. Some of the fabs observe this more when they have a number of tools that are not cooled down sufficiently during continuous high volume production. Coating is typically applied to achieve higher temperature stability, and is usually CVD TaC. It is more durable on its surface in extreme heat and can withstand chemical attack from some process gases. Under many practical conditions, the number of particles originating from cracks is less for TaC-coated parts than for SiC. Smoothness of the surface is also important. A smoother coating will minimize areas for particle formation or adhesion. SiC tends to start softer, but can get rougher as it gets older. TaC may remain stable for longer but if defects are found from the beginning, they are likely to remain undetected until stress is applied. Engineers usually observe the trend of the particles, rather than the type of coating, in the production process. If defects are introduced in a line using SiC, the line could operate without any defects for several months and then a gradual increase in defects may be observed. A TaC system may be more stable, but respond strongly to the presence of a coating defect. Ultimately, no coating will completely eliminate particle risk. The actual difference lies in the performance of each during aging under heat, exposure to gas and repeated use.

Role of Epitaxial Parts Geometry in Particle Accumulation

Although the coating quality may be excellent, it is possible to still have a particle problem from the shape of the epitaxial part. Geometry is important in its own way because it determines the flow of heat, gas and small debris through the chamber. The small details of the design can mean the difference between particles remaining in the design and particles getting on the wafer. Particles can also be trapped in deep grooves or narrow channels. In the process of processing, small deposits can settle in these areas. These can be released later if the gas flow is great enough, in bursts. This is frequently encountered in components such as carrier structures, or susceptor grooves. They can quickly enter the wafer zone after leaving in the open. Large, flat surfaces are not always safe. If the surface is wider and is not sufficiently supported under it, it may expand and contract inconsistently in the thermal cycling. This slow moving action can cause coating fatigue. Engineers have observed that fine dust occurs in the middle of large planar components after several cycles in some cases where the component is coated with SiC. This is something that is better managed by TaC coatings, but if the design was out of balance, it can still manifest itself as geometry stress. One actual application is a production line that experienced sporadic particle spikes despite changing the coatings. The problem was identified following inspection as being an issue with the carrier design and tight inner corners. The deposits in those corners were being pushed by gas, and began to split during heating. The problem was solved by a slight redesign and smoother inner curves, but without changing the coating type. In reality, materials are not the only concern for good particle control. Each part's shape determines the build-up of the stress and the site where particle formation is likely to occur.

Process Optimization in Veeco EPIK and LPE (ASM) Systems

Optimization in epitaxy systems such as Veeco EPIK and LPE (ASM) epitaxy tools is frequently a matter of maintaining a stable and predictable process. These platforms are designed to be very precise, but any adjustments to the set up or running practice can cause particle problems or uneven film quality. Temperature control is one important aspect. Typically, heaters are not ideal throughout the chamber's entire surface area in real production. If one zone is hotter than another, the wafer and coated parts will expand at different rates. This difference may cause a strain of SiC or TaC coated parts over time. Many engineers discover that particle counts can slowly increase without any apparent signs of drift, even for a relatively small temperature drift. Gas flow tuning is another important step. Gas is introduced into the chamber by designed inlets in both systems and the actual pattern of the gas flow may change depending on the wear of parts or slight alignment movements. When flow is not even, it can lead to the formation of dead zones where byproducts can accumulate. Later, such deposits may detach from the body and touch the wafers. Frequently reducing flow balance or cleaning inlet paths will help to reduce particle levels once more. Timing is more important than many realize for maintenance. To conserve downtime, some fabs operate tools for too long between clean cycles. This can backfire. When a thin layer of deposition has been formed on chamber walls or on the surface of parts, it begins the process of flaking at a certain thickness. A regular cleaning routine (not necessarily as frequent as described above), generally will keep particle spikes in check. One of the other practicalities is part matching. Mixing of worn parts and new parts with coated parts in a system such as Veeco EPIK or ASM LPE can result in unpredictable performance. The newer TaC-coated susceptor and the older SiC-coated carrier can have different heat cycle responses. This difference may cause slight changes in gas flow rates that compound to cause a change over time.

Cleaning, Refurbishment & Lifecycle Control of Coated Components

Particle control in epitaxy tools is de facto made easier by cleaning and refurbishment of coated components. Even when coated with such materials as SiC or TaC, parts do not remain clean forever. As time goes on, thin layers of process byproducts accumulate on surfaces and cause a gradual modification in the behavior of the part under heat. In most fabs, the first step involves routine cleaning. This is typically performed to eliminate loose deposits prior to them ending up being hard layers. Deposit build-up can become very difficult to remove if it is not cleaned too soon. When it is that point, more aggressive cleaning may be required and will introduce more risk of damage to the surface. Regular maintenance can be more effective than lengthy periods of no cleaning followed by intensive cleaning. Refurbishment is fitted when cleaning doesn't cut it! Micro-cracks or surface roughness may occur after extended use for SiC coated parts. If the underlying graphite is still good, it can be re-worked or re-coated. When talking about TaC-coated parts, refurbishment might be mainly on the bond area to identify and eliminate any spot that the coating has curled or weakened. One actual example from a sample line with a high temperature epitaxy system indicated that the number of particles increased after about 6 months of stable operation. Initially, the team thought there may be a gas flow problem. Upon inspection, they discovered the susceptor surface had an uneven and thin deposition, which wasn't completely removed during cleaning cycles. After the part was suitably reconditioned and refurbished, the particle count returned to its normal level, no changes were required to the process recipe. Another important aspect is lifecycle control. For any coated part there is a limit, regardless of how good it looks. To track usage time, thermal cycles and cleaning history to determine when a part is likely to begin producing particles. Note that in some fabs the logs are simple and in others full tracking systems are used. Either way, the objective is to either replace or refurbish parts before they become a contaminator. Good lifecycle management is not just the use of parts as long as possible. It involves replacing them in time before wafer defects begin to develop on the surface of the wafer.