Safe AIXTRON Susceptor Cleaning Without Damaging CVD SiC Coating

The CVD SiC coated susceptor surface must remain clean, but is fragile and susceptible to damage due to poor handling procedures. In many fabs small mistakes in the cleaning procedures create a localized roughness or micro-scratches that affect wafer quality downstream. One maintenance engineer actually reported that switching from the current scrub pad he was using to a soft compliant wipe cured ongoing problems with coating wear. Not only is the residue being removed, but the SiC layer is kept intact so that it will continue to fulfill its function of thermal isolation of the Sic coated graphite susceptor .

Best AIXTRON G5/G10 Susceptor Cleaning Methods for Long Service Life

The best way of cleaning the AIXTRON G5 and G10 susceptors is not "stronger cleaning", but cleaner is cleaning and incrementally removing buildup so as not to damage the CVD SiC coating. Many production lines will have a longer life of the Epitaxial susceptor if cleaning is part of the whole process rather than an aggressive step.One typical method used for cleaning is in-situ gas cleaning inside of the reactor. It means cleaning away the deposit layers of SiC or carbon with the help of the halogen based gases at the controlled temperature. The key to success is the controlled temperature, so as to protect the coating from attack. A lot of engineers like this way as it does not have contact (which causes micro-scratches).Low impact wet cleaning when undergoing maintenance shutdowns is also widely used. It contains some solvent cleaning in order to clear up loose materials, then careful immersion of chemical dips so as to eliminate contamination from metals and oxides. A fundamental rule to remember is 'no scrubbing tools, no rough wiping'. Even a minor scratch on the surface can act as a particle generator during a high temperature process.Another approach being adopted by some fabs is the use of a cleaning rotation cycle. Some light cleaning after each run is carried out and deep chemical cleaning is performed only during scheduled maintenance shutdowns, rather than during every process. This reduces the strain on the SiC coating. One SiC production line claims reduced coating failures since it shifted from weekly aggressive cleaning to brief daily purging processes and a monthly clean based on visual inspection results.Many people do not recognize the importance of controlling the cooling process. The gradual cooling down of the susceptor before cleaning is good at reducing the thermal shock. Temperature changes may be a cause of the coating cracking, though no immediate effect may be observable.In fact, normally a much longer life span of the susceptor is achieved from combination of the three procedures; controlled in-situ cleaning, low impact wet cleaning and good handling discipline, so as to avoid any early failure of the susceptors (G5/G10) and ensure a longer lifetime and better yield.

Avoiding Particle Issues in Epitaxy During Chemical Cleaning Processes

Particles are probably one of the most important variables in a stable epitaxy process when cleaning with chemical steps. All remaining loose film or dirt however minute will end up on the wafer surface and cause hard-to-trace defects. Yet, in many cases, it is the cleaning step itself that provides the problem rather than solving it.An obvious problem is insufficient rinsing after chemical baths. A chemical not fully rinsed away from the part, will during subsequent heating cycle dry on the susceptor or on the chamber and become small particles. Drip of high purity water and allowing it to dry normally without use of forced air, which can spread it further can minimize this.Another issue is post-cleaning. In many fabs, the parts are clean and left for a long time in the open air. Dust can settle on the surface without being observed by anyone - simply dust from the cleanroom. A tech in a fab was reported to have immediately eliminated repeating particles problems by just using a cover for the part-transferring tray, using exactly the same cleaning chemistry as before.The compatibility of the cleaning with parts is also important. The chemistry might react differently with older Sic coating or metal chucks and create a thin layer which can flake off afterwards. This is the reason that many groups are checking the cleaning chemistry on spares.Control of the drying phase is another aspect which can be overlooked. The spots may occur due to water left within the chuck itself and they can leave mineral spots. Generally, a slow drying phase hence, longer cycle time will provide cleaner finish.The summary is that there is not one specific significant mistake contributing to the particle issues. It is the combination of rinsing, handling and drying. Epitaxy can become stable and repeatable by performing the three above procedures properly and gently.

How High-Purity Graphite Reacts to Acid and Plasma Cleaning Cycles?

It is advantageous to use high purity graphite for epitaxy systems because of the very high operating temperatures and extended usage time. However, the surface of the graphite changes in subtle ways over time in the acid/plasma cleaning processes that may not be readily visible:Graphites are not melted or quickly dissolved during the acid cleaning phase, but reaction may occur on the surface. Strong acids are able to 'attack' the surface and create pits on an originally smooth surface. Such rough surfaces may become a point of contamination for particles or cleaning residues. It is known in many fabs that, in time, older graphite parts start to accumulate more particles, while the cleaning method remains the same. Usually, this is caused by long-term acid exposure.Plasma cleaning has a similar but different damaging effect. In contrast to acid cleaning, removing carbon layers by oxygen and/or fluorine plasma is effective, but the surface itself can be etched away layer-by-layer. Slow etching of graphite surfaces leads to texture changes, and it can cause the parts to become brittle at certain places. In such cases, flakes can fall off at high temperature cycles and act as particle sources in the reactor.This problem was identified on a production line. After several months, a set of brand new parts that was over-cleaned (multiple plasma cycles) started to produce random wafer defects. The plasma was etched away, but also worked perfectly for contamination removal. Changing the time for plasma cleaning, increasing the number of visual inspections, has reduced the defect rate.To avoid these phenomena, it is desirable to manage the lifetime of the graphite surface and the cleaning power. Acid cleaning should not last too long and should only remove visible contamination. It has been observed that short plasma cleaning pulses work more effectively than longer pulses. Furthermore, alternating the graphite parts used for the cleaning process reduces wear on the graphite to some extent.Crucially, high-purity graphite is intended to be reused, not modified for all time, and as it is used on a daily basis, every cleaning cycle removes the particle and modifies the surface.

Common Causes of CVD SiC Coating Cracks After Repeated Maintenance

CVD SiC coatings are designed to be strong, but maintenance should be done repeatedly and will cause stress to the coating over time, leading to cracking. Cracks can often be caused by a combination of errors. They accumulate over time with minor changes in handling, cleaning and thermal cycling.A common cause is thermal shock due to the cleaning cycle. If a susceptor or coated part is heated or cooled rapidly, the SiC layer and the base material will expand at different rates. This mismatch generates a strain. After a few cycles, small cracks begin to form, and after more cycles, a crack grows larger, until finally, nothing is left intact. An early maintenance team found that longer cooling time before parts moved to cleaning stations led to fewer cracks.Another concern is chronic exposure of the chemicals. The bond between the SiC coating and the substrate can gradually break down with strong acids or Reactive cleaning solutions. Although each type of cleaning might appear to be innocuous, repeated applications can lead to weakening of the coating in certain areas. As time goes on, those weak points evolve into crack initiation sites.Mechanical handling damage is also a big factor. Tiny surface defects can be caused by simple actions, such as resting parts on hard surfaces, operating with metal tools, or minor impacts on parts during transfer. These may not be a problem initially but when the temperature rises, they will develop into prominent cracks.Another hidden cause is plasma over-cleaning. Plasma can be used to remove residues, but can also cause uneven etching of the coating surface if long or frequent exposure. This builds up tension in the structure, causing stress points to weaken it.A team's change in the way they maintained the production line resulted in a real case where the cracking problems were reduced. They cut down on aggressive chemical dips, introduced less abrasive handling trays and decreased plasma cleaning time. No changes were made to the tool, but noticeably, the coating's life increased.The basic concept is very simple: SiC coatings do not fail due to their weakness, it's because repeated maintenance gradually forces them to exceed their stress limit. It is advisable to follow each cleaning step gently, to get a long coating life and a stable epitaxy performance.

When to Repair or Replace Epitaxial Parts in AIXTRON Reactors?

An art to repairing or replacing epitaxial parts within AIXTRON reactors is to be able to read the slight changes and recognize them before they result in part failure. You won't often find that fab parts such as liners, susceptors and graphites will suddenly cease working. Instead these parts will wear away gradually over time and at that point, the problems may begin to appear on your wafers.A definite indicator of repeat particle problems despite cleaning, is if the same problem keeps reoccurring. If your wafers continue to show the same type of random defects and despite a cleaning procedure, nothing changes, the state of the surface of your part is likely to be too poor to restore. For example one line kept increasing cleaning strength to combat the particle problems they were seeing, only for it to be a case that their susceptor under their SiC coating was micro-cracked.Visible colour change or wear on the coating is also a definite tell. Typically SiC coated parts look consistently the same. Should you begin to see dark areas that cannot be washed off, or duller/different colored textures on parts, then it is very likely the coating has been eaten away or compromised. You have waited too long to repair at this stage.Cracks and edge chipping show that replacement is going to be your best bet. High temperature cycle times tend to propagate these cracks relatively quickly even if they are initially minute. Polishing or re-cleaning these cracks will often just decrease stability of the process, rather than improve it.Changes in your process such as process drift can be a more subtle indicator that part wear is beginning to become a problem. As the parts wear out the flow of gas or distribution of heat within the reactor is altered and you may begin to see changes in temperature uniformity, deposition rate or wafer consistency without the recipe ever having been changed. Often these changes will initially be written off as process variability rather than the mechanical issue that the worn part represents.Repair is really best suited for cases where the contaminants are only skin deep, such as a little contamination, or where light deposition has been put down on parts. In these situations either a controlled chemical clean, or low power plasma may bring parts back within limits, but should there have been any structural damage, a part replacement will be more reliable in the long run.One fab was telling of how they have reduced their total downtime from when a part fails to just a simple rule. If they see the same particle problem twice in a row after cleaning the part is pulled and sent for inspection, not for use in production. It was found this reduced the time when a sudden drop in yield would occur.The most practical means for running operations day to day will be to build maintenance of replacement parts into the process control, rather than building the repairs around maintenance reacting to situations. With careful monitoring, slight part wear can be avoided for full production runs by keeping epitaxy stable and through early decision making.