Thomas Beeg, in a recent post on Wafer Fab Size, and Does Size Matter, explains how two factories running on the exact same technologies and process flows may end up having different cycle times due to the wait times for the lots being processed. The post points to the fact that processing time would be the same per tool, per process, for a specific operation, and the only real factor affecting the cycle time would then become the wait time.
To understand this phenomenon further, we must consider the numerous factors which effect the cycle time in a fab. The first and biggest factor is the Fab size itself; however, other factors from the product mix to overall equipment uptime and stability also play critical roles. For the purpose of this post, I will focus on the factory size and specifically, how having multiple tools capable of performing the same process step might be beneficial in improving the overall cycle time.
(Factory) Size Matters
Thomas explains that no equipment deployed in a Semiconductor fab can possibly have 100% uptime, which means unless there are redundant equipment for a given process step, lots would have no path to move forward and would come to stand-still if there was downtime. The graph below shows the operating curve to demonstrate the argument.
Having more than one tool massively reduces the average lot waiting time and helps avoid the ‘one of a kind’ or single point of failure situation. The difference between having 2 tools with the same capabilities and having 3 tools is not as great, but it becomes clear that having more tools is better than having one when it comes to cycle times and utilization of individual tools for processing.
Another extremely important factor to consider when considering cycle times is the number of times a lot comes back to a given tool-set for processing. The example below highlights the impact the number of passes may have. Considering the operating curve shows an X-Factor of 3 and a processing time of one hour, meaning each lot has a wait time of 2 hours. It becomes clear upon reviewing the example that the more the number of passes, the higher the overall cycle time, which indicates and at times necessitates the inclusion of 3-4 tools for processing lots instead of just 1-2 tools.
An extremely important consideration to make is to ensure that if multiple tools are deployed on the shop-floor, they should remain available for processing and not be disqualified or down when making execution plans. This comparison between actual tool availability against planned capacity should be made periodically to understand whether the deployed tools are being utilized to the full extent.
When process complexity, meaning the number of passes per tool per job are taken into consideration, is coupled with the number of tools per fab, it becomes increasingly clear how a larger number of tools will almost invariably be better than having a single or dual tool set-up. It is also understood that having more tools means spending more money, but with increased demands for better outputs and with ever decreasing lead times, fab owners aren’t left with much choice it seems if they are to survive in this highly competitive marketplace.
Thomas suggests going big – even massive if possible – as having larger number of tools for a busy fab means increased utilization, drastically lower cycle times and a possibility of reduced CAPEX. Go big or go home is the message!
The Role of MES in Improving Cycle Times
It is well known that investing into and building massive fabs can be extremely capital intensive, (estimates of building a new fab are over $1B) but as clearly indicated, having additional tool sets is extremely important in maintaining the requisite cycle time and utilizing available tools to the maximum. With increased process complexity, larger product mix and lower overall customer lead time expectations, investing in technology like a modern MES platform can be greatly beneficial while increasing tool redundancy incrementally. Let’s understand how.
The MES brings first and foremost to the process Edge Computing capabilities, which means enhanced visibility into what’s happening on the shop-floor and being able to decide what alternate courses exist for a given process step affected, in real time. MES through IoT brings clarity of material and machine utilization from the shop-floor, which translates to knowing exactly how a piece of equipment is performing and whether or not lots need to be re-routed to another ‘ready’ tool in case of an unexpected disruption or failure event.
When fabs do not have the option of going big, they must rely on technology, like the MES to ensure their existing tools are qualified, ready and able to process lots which are both planned and re-routed towards them. With the MES in place, not only do lots get processed as planned, but in case of disruptions, lots can be sent on an alternate path for processing automatically. The application is tasked to ensure they are processed following the right recipe and that the overall schedule remains unaffected to the maximum possible extent.
Further, the application tracks material movements and provides complete Material Logistics within its functionality to ensure any waiting time for lots is exactly as per the plan and not because of any negligent systemic or manual disruption. It is evident from Thomas’s post that waiting times of individual lots per process step can have a massive bearing on the overall cycle time. Add to it the complexity of multiple and repeated tool visits, the MES provides necessary oversight and control in such cases to ensure that waiting times and processing times are kept at an optimal level and the material movement which needs to happen in order to execute even highly complex recipes is done so in an automated and well-orchestrated manner.
While theoretically an X factor of 3 and a processing time of 1 hour indicate a wait time of 2 hours per lot; on the shop-floor in real operating conditions a lot can happen to increase this time. This is where the MES makes a major impact, as it brings with it the capability of not only scheduling production lots, but actually punctuating the lot movement, ensuring tool readiness and controlling material flow to ensure planned execution and actual shop-floor execution are as aligned as possible. Knowing where the material is in real-time, where it is headed and whether or not the tool supposed to process the lot is ready and qualified can help ensure higher utilization with a tight control on planned cycle times. It translates to better results even in a two, three or four tool set up.
With the right application in place, CAPEX does not need to skyrocket going from one tool set to ten; rather it can be incremental and phased, where the availability of 3-4 tools working efficiently delivers on the cycle time expectations. In such cases while the MES becomes more akin to a justified OPEX, owing to its cloud-based and modular nature, where roll-outs can be use case based, delivering value as the expenditure increases. Semiconductor companies can truly invest as they grow and be able to utilize existing and new tools in a better and more reliable manner.