Semiconductor industry as a whole is facing many challenges, from rising demand to developing products faster to handling a much higher than usual product mix. As a result, the industry is looking towards digital tools such as MES to comprehensively deal with these challenges.
For semiconductor R&D facilities, the challenges faced by the industry at large are amplified: the need to be more efficient across the board, the need to handle a much larger product mix with shared tools and shorter timelines, to have accurate master data and reusable process flows/recipes/instances, to deliver a contamination-free production environment with complex flows and urgencies, eliminate or re-configure old legacy IT tools with new ones, and above all sell the vision of digitalization at large to scientists and executives who may be used to legacy, point solutions for their facilities.
For most R&D labs, it holds true that process technology is rarely aligned with the latest IT systems which actually need to be driving the process and its execution. This misalignment creates inefficiencies, delays and losses, both in terms of processed wafers and time spent in processing. Semiconductor labs, even the ones with the most sophisticated tools and process technologies, tend to rely on paper or spreadsheets and manual labor when it comes to material movement and task execution.
As process technology upgrades and becomes more complex, compounded with pressures exerted from customers for a larger product mix with shorter lead times, the disparity between IT and OT begins to take its toll. It is not uncommon for R&D facilities to have very little to no insight into their master data, and most do not use IT systems such as MES for equipment, WIP, recipe and material management.
However, just like the manufacturing industry at large, semiconductor manufacturers and R&D facilities are starting to realize the benefits a modern MES Platform has to offer. With industry specific, modular and highly configurable functionalities, these systems can completely transform any facility and bring in Industry 4.0 in a short period from the date of roll-out.
Pre-MES scenario and establishing the need
In order to transition to a modern MES application, understanding where things are with the current systems and the interplay between existing process technology and IT applications is the obvious first step. Most R&D facilities are plagued with the rampant use of paper and spreadsheets, which likely means master data is neither collected, nor stored properly. When master data is not being analyzed, process improvements are compromised.
If this is the scenario in your lab, it can serve as a good starting point to build a case for a new MES. Going paperless and replacing legacy tools and spreadsheet based applications can be a major driver for getting a modern MES; not only to better record master data, but to ensure every iteration, transaction and metric are captured, documented and versioned. Legacy applications have increasingly become hard to maintain and costly to upgrade. Another drawback with legacy applications is that they are dependent on an individual, and once the person who designed the application moves on or retires, making changes may become virtually impossible.
Another challenge faced by any R&D lab is the massive amount of disparate, individual material sets moving through the shared tools at a given point of time, all with different and individual flows. Scientists need a planning tool which helps them control, document and plan each individual flow separately and accurately. A modern MES not only allows the planning of each flow, it allows re-use of recipe steps and partial flows, while maintaining the integrity of the master data. It reduces the time needed by process participants to copy a process flow, and perhaps make modifications each time a recipe executes, with slight changes in measurement or execution parameters.
Another aspect to consider is experiments management; it’s not normally well defined in labs. As scientists change the process steps and sub-recipes based on their experience and creative thought-process, these changes aren’t necessarily documented and may not even end up on a spreadsheet. In such a scenario where iterations provide the ideal results and yields, not having the exact steps and recipe data might lead to re-work until the specifications are finally documented. Ideally, the MES supports iterative approaches often used by scientists in the bid to create something new, while using the same industry tools, parameters, recipes and equipment.
Market drivers which require a broader product mix to be delivered in shorter lead times can be an obvious reason to go for a new MES. However, it would be worthwhile for the team or leader initiating this transition to pinpoint the manufacturing flow. How do multiple lots moving through shared tool-sets compete for processing? How do you execute without a proper scheduling tool? It becomes increasingly difficult to get the maximum asset utilization and process efficiency. There are also quality and contamination related challenges, which may decline when handled through the MES, as every process step is defined and checks exist to ensure issues pertaining to cross-contamination and other quality incidents are kept under check through automated oversight.
Choosing and justifying the MES
Irrespective of the initial reasons driving the transition to a modern MES, it will need to be justified to the management and the stakeholders at large, and more so to the users whose daily work-life is affected due to the implementation of the new software. Oftentimes, the resistance against a new MES might come from end users- scientists for example- who may feel that the new application restricts their freedom of ad hoc scribbling something on paper, testing it and moving on to the next iteration.
First things first. When it comes to choosing the right application, the basic ability of the vendor to understand the status-quo and showcase how their application transforms it to an Industry 4.0-level operation is a must. The MES should be scalable; for most deployments, a subset of the overall functionality might be used either across a single lab or multiple labs, depending on the project team’s decision. Tools like the Gartner Magic Quadrant can help you shortlist the top industry-specific vendors to help you choose the right application which suits your budget and needs.
Once the most suitable application is chosen and deployed, project leaders need to work on two important aspects simultaneously: one, the justification of the cost incurred, and two, managing the change and resistance associated with the launch and use of the new application.
The best way to manage both justification and change management is to have a simple vision and a pragmatic implementation team. Going paperless, for example in maintenance, sounds like a simple enough vision, with a clear deliverable and with it the attached value proposition of better tracking, automated maintenance scheduling and data capture. In realizing this or any other vision, there is an inherent amount of effort. Project leaders and MES vendors need to be clear with the team; leaders of the implementation team need to point out to the other stakeholders improvements as they start to deliver value, which unless pointed out might go unnoticed. Educating the users on how the status-quo erodes value and how the new MES helps create it might be the most important job carried out by the implementation team.
Current and future use cases
For any R&D lab, the priorities might differ, but the main focus should be on those use cases which deliver the fastest and most tangible value. The approach which has been highly successful in R&D labs making the transition to a new MES is to move in a phased manner. Focus first on getting the basics right and then move on to more advanced capabilities.
Going paperless or tracking equipment status and operation or creating a virtual twin which replicates the actual shop floor, may be the first use case/s chosen.
The goal for most labs initially is to get a better handle on operational flows and maintenance activities, along with the proper storage and use of the master data to allow re-use of certain data sets while maintaining data integrity. WIP management benefits the most from master data management, where the simple activity of creating master data sets leads to the benefit of process rationalization, with each process being analyzed for redundancy and with the spirit of having unique and clean master data sets. MES brings that clarity and eliminates redundant processes, while consolidating and creating a master database, replacing silos of data from point solutions.
With a phased approach towards MES implementation, labs should plan to fully utilize the entire set of semi-specific MES functionality: experiments management, scheduling, recipe, and reticle management. It is important to note that with existing benefits and implementation efforts, the future initiatives gain momentum. As with current use cases, future ones too have specific benefits. Experiments management for example can help process owners better understand the interaction between different processes being executed in the same chamber. Recipe management can deliver through equipment integration the right recipe to the right tool, and this is considered an important deliverable. Reticle management will focus on delivering error-free work even if the loading of photomasks themselves remains manual.
Irrespective of the MES functionality chosen to implement, what remains important is to have people on board and make consistent gains, while ensuring that every new implementation adds value to the process along with improved functionality.
The modern modular MES application has multiple complex functionalities which further support complex manufacturing flows and the advanced Industry 4.0 capabilities that are needed to be more efficient and agile in the current market scenario.
However, what becomes evident is that priorities need to be set when it comes to use cases. Even the seemingly less ambitious use case of eliminating paper, or the ability to make well-defined changes to recipes and flows, require infrastructure. The need to have uniform master data and the use of the same application across the board for comprehensive knowledge are the first steps that can lead to massive gains for the R&D lab.
Priorities may vary from lab to lab and from fab to fab, but the need to graduate from old, cumbersome and outdated systems is the same. Whether you are a semiconductor R&D facility or a high volume manufacturer, choose an MES that offers system scalability, data visibility, and advanced functionality.
Image credits: Fraunhofer Institute for Microelectronic Circuits and Systems IMS