AIXTRON G5/G10 Susceptors: How Coating Thickness Affects Performance

The susceptor is an unobtrusive but highly significant element in the AIXTRON G5 and G10 epitaxy systems. It keeps the wafer in position and maintains the uniformity of heat spread in the high temperature steps. One factor that is not usually taken into account is the thickness of the coating on the Susceptor plate surface. When the coating is too thin it can wear out more rapidly and may lose stability following repeated heating cycles. Too thick can cause uneven heat transfer and impact wafer quality. It's important to strike a balance and maintain the process to prevent small problems from becoming larger production issues over time.

Optimal Thickness Design for CVD SiC Coating in AIXTRON G5/G10

To understand the details of the susceptor coating on AIXTRON G5 and G10 systems, it is best to consider it as a protective skin. It lies between the hard reactor environment and the graphite base and is responsible for the extreme heat, rapid temperature changes and reactive gases in epitaxy. Thickness of CVD SiC coating is not simply a number on a spec sheet, but a factor that directly impacts on heat flow, surface stability and wafer uniformity. Even small differences in the coating if not balanced will be later revealed as uneven growth of the film or by early wearing down of the Susceptor semiconductor surface.

Trade-Off Between Thermal Conductivity and Mechanical Stability

AIXTRON engineers are often confronted with a simple but tricky trade-off when they adjust the thickness of CVD SiC coatings on their susceptors of AIXTRON G5 or G10 type: thinner coatings provide better heat transfer, while thicker coatings provide higher durability. A thinner SiC layer normally allows heat to flow faster from the graphite base to the wafer. This will ensure smooth temperature ramping and help to achieve uniformity in fast process steps. For some production runs, particularly those with tight temperature profiles, operators will find that readings of wafers are more stable with a thinner coating. However, there's a catch. Thin coatings last shorter when thermally cycling more often. With sufficient runs the surface becomes inconsistent and eventually develops small cracks. From then on, particle risk and process drift may manifest. Thicker coatings have different behavior. They play a similar protective role to a more effective barrier. The surface of the Carbon Susceptor is more durable in prolonged production runs and under severe chemical attack. This stability is often welcomed by many of the fabs that are manufacturing in large volume batches as it will help in cutting down the need for frequent refurbishment. However, thicker isn't always better. The rate of heat transfer decreases slightly and in certain instances, temperature differences can arise within the wafer. This slight mismatch can have a detrimental impact on the quality of the epitaxial layers, particularly in types of devices that are sensitive. In real production environments, one silicon carbide coating shop reported that they were able to swap from a middle range thickness to a slightly thicker build and decrease the rate of susceptor replacement by almost half. But they also needed to tweak the temperature profile to prevent the slight edge uniformity.

Coating Thickness Impact on Wafer Temperature Uniformity

One of the things that can appear to be uniform on a screen but contain some small issues under the hood is wafer temperature uniformity. The uniform application of heat across the wafer surface is dependent on the CVD SiC coating thickness on the susceptor in AIXTRON G5 and G10 systems. Heat will flow faster through the wafer if the coating is thinner. This may help the system to respond quickly to a setpoint change. It can be beneficial in certain applications for short-cycle stability. Thin coatings can increase the surface's sensitivity to small changes in gas flow or heating, however. This sensitivity is often expressed as small differences in temperature at edge to edge inside the film that will be visible in the thickness of the film. The thicker the coating, the more like a buffer layer it will act. It registers and dampens down shock changes in temperature. This can aid to steady the temperature field throughout the wafer, particularly in longer runs. In one production line that was operating epitaxy of gallium nitride, the researchers found that they could reduce the cooling effect at the edge of the wafers by increasing the thickness of the coating by a small amount when the deposition runs had been longer. This led to a more uniform thickness of film in the batch. However, there is a limit. However, the thicker the coating, the less heat will transfer. The susceptor may require a longer time to stabilize, particularly during the ramp up and ramp down steps. This can cause a temperature difference between the center and the edges of the wafer. This will be detrimental to repeatability over time, particularly in narrow process windows. Not only is coating thickness not the sole determinant of temperature uniformity, but it is not a measured value. It also takes into account gas flow design, susceptor geometry and reactor tuning. The thickness of the coating dictates the basic way heat will conduct before all those other factors are considered. In reality, many engineers will want to stay away from both extremes, and will seek a range somewhere in between. They test wafer mapping results following thermal cycling and make incremental changes, monitoring for the edge roll-off and hot spots in the center. The pattern of uniformity can change more than anticipated by small changes, typically just a few microns.

Failure Risks: Delamination vs. Premature Wear

The two most common forms of coating failure in AIXTRON G5 and G10 susceptors are delamination and premature wear. At first, they sound alike, but they have very different behaviours in the reactor and they also originate from different conditions. Delamination is when the CVD SiC coating starts to separate from the graphite base. It frequently starts at weak spots that are present on the surface during the surface preparation or during coating growth. It very rapidly starts spreading under thermal cycling. A very good tech once said that it was the surface just peeling in places after weeks of looking so good. What can be tricky is that all will seem well until the bond eventually fails. The distinction between the two normally relies on the way the coating was created and its utilization. Delamination is more associated with bonding strength and bonding surface cleanliness. For wear it is much more related to the coating density and the choice of coating thickness and intensity of the processes. A sensible approach to dealing with both is to observe early warning signs. The first signs are usually the small variations in particle concentrations, the slight shifts in wafer temperature mapping or the unusual color changes on the susceptor surface. Thermal cycling is also beneficial because it can be inspected regularly, before problems get too big. Many production lines seek more than failure avoidance, they want to ensure that the coating design meets actual use. Coatings which achieve good life in one tool may not perform as well in a different tool, due to varying cycle speeds or gas chemistry.

Customization Strategies for Different Epitaxy Recipes

Susceptors are not the same for different epitaxy recipes. The surface appearance of the AIXTRON G5 and G10 systems may be similar, but there are considerable differences between the stress of the CVD SiC coating when switching from silicon processes to GaN or SiC processes. Coating thickness customization can come in handy there. A slightly thinner coating would be desirable for recipes that are designed for fast thermal cycling. To make these processes work, heat must be able to transfer rapidly from the graphite base to the wafer. With real production, the teams that have short cycle silicon epitaxy have noticed that the ramp stability is better with thin coatings, particularly when changing between different recipes during a single shift. The downside is maintenance checks must be conducted more frequently because of the early appearance of wear. Research has shown that a thicker coating tends to be more effective for long duration high temperature processes such as GaN or SiC epitaxy. These recipes are intended for prolonging the time that the susceptor is under heat stress, and so surface protection becomes more of a priority than speed. One production line reported that once they transitioned to a thicker coating system to achieve long runs of GaN, the lifetimes of the susceptors increased significantly, and particle interruptions decreased during longer batches. The process remained stable for a longer time, but with a slight decrease in the initial rate of temperature response.