Send-Ahead Processing: Your Canary in the Coal Mine

In general, all defined manufacturing process steps are safely assumed to be stable – that is…the result can be predicted and repeated with a high degree of certainty and accuracy over an extended period. Without these stalwart, predictable steps in a process, it would be nearly impossible to build reliable, uniform, consistent parts, components, and products. The unknowns coming from inconsistencies, variations, and unreliability would be costly and antithetical to a successful business venture. Even with the expected reliability in today’s tech-infused manufacturing processes, some steps can still be tenuous and difficult to construct to ensure a consistent result is achieved every time. Without tests and validations at specific points in the process, the product can be at risk of failure. If 999 of the 1,000 steps in a product’s creation are performed perfectly and a single value-added step is performed incorrectly or the process deviates in some way at this one step, the result could be millions of dollars in lost revenue, unrecoverable raw material expenses, and wasted labor.

Parent & Child Relationship

One of the ways to prevent this type of unfortunate event is to gate a sensitive, unpredictable process with what is commonly referred to as a send-ahead process. Send-ahead processing also goes by the term piloting. Send-ahead processing takes many forms depending on the industry, the MES being utilized, and the data needed ahead of time to validate the results of the current process conditions. In general, a send-ahead process utilizes a finite number of materials in a carrier, container, or front-opening unified pod (FOUP) to create a child lot from the parent lot. As the parent lot waits for the child lot results, the child lot processes through the unpredictable step(s) to the point where data can be measured and evaluated to ensure the process is acceptable for the parent lot. If the child lot results are acceptable, then the parent material can be released to continue processing using the same parameters as used on the child material. The child lot waits for the parent lot to “catch up”. Once the parent lot is at the same step as the child lot, and assuming the child and parent lots measures acceptably – then the child lot is (or can be) merged into the parent lot and the parent lot can continue on its way (with all of the child wafers). If, however, the child material fails to measure nominally, then engineering personnel can address the problem, create another child lot out of the parent, and send the second child material through the process. Again, the second child lot is measured and evaluated. This cycle can continue until the correct send-ahead results are obtained which allows the parent lot to be processed.

Send-Ahead Processing vs. Run-To-Run Processing

Send-ahead processing differs from traditional run-to-run scenarios. Run-to-run processing is more dynamic than send-ahead process which is more static in nature. Run-to-run processing uses the latest downstream measurements from the last material measured to influence processing parameters on an upstream process. Whereas, send-ahead processing uses material in a single lot or container to test the process for a known good result before committing the entire lot to the process. Secondarily, send-ahead processing validates a given process on a single lot, whereas run-to-run processing uses current measurements as the catalyst for recipe parameter adjustments for future lots. In the end, send-ahead processing fulfills a specific role in validating a tenuous process step.

Pros & Cons of Send-Ahead Processing

An expected outcome in this type of processing is to – if necessary – sacrifice one (or more) of the materials in the lot to ensure the material obtains the correct results. The hope, of course, is the child material is successful the first time through the target process and can be merged into the parent material later for continued processing; however, when the process is not stable and results in deviations, then only a small number of materials are affected rather than the entire batch or lot or container of materials. This is the classic “canary in a coal mine” scenario – if coal miners sent a canary into a coal mine and could continue to hear the canary “chirp”, this meant the level of toxic gas was low enough that it was safe for humans to remain. But when the canary would stop chirping (because it died from the gases), then the miners would leave because conditions were unsafe.

Here are a few questions send-ahead processing might help answer:

• Is the ambient barometric pressure in the fab such that my atmospheric deposition tool will process material with uniform results on target?
• Has the replacement of a raw material supply affected a given process?
• Does a raw material supplied by a new vendor produce the same tensile strength as the existing raw material supplier?
• How does a change in energy dosage on a process in Photolithography impact end-point etch times in Etch?
• Is the equipment capable of processing with minimal elevated particle generation now that the preventive maintenance has been delayed?

Send-ahead processing offers the following benefits:

• The process on the parent material should mirror the process on the child material.
• A sampling (every 10th material, for example) can ensure the process is stable, thus, reducing scrap potential and saving financial and human resources by avoiding unnecessary measurements and equipment maintenance.
• Scrap counts can be reduced by keeping sacrificial material at a minimum.
• The send-ahead processing concept can be implemented in almost every MES as well as in environments lacking MES applications.

Send-ahead processing is not without its concerns, which may include the following:

• A stubbornly unstable process could require repeat attempts, thus, consuming more than one of the parent materials.
• There is no guarantee good child material results will translate to good parent material results, especially as the processing time between the child material and the parent material increases.
• The further along in the process, the more opportunity costs exist with send-ahead material. In other words, doing a send-ahead process at the beginning of a 1000-step process flow is far less costly than a send-ahead process at step 900 (of course, the costs are realized only when the child lot is deviant and, thus, needs to be scrapped).

As you can see, send-ahead processing is yet another tool in a Process Engineer’s toolbelt to ensure the stability and accuracy of a process, develop baseline data for future comparison, and increase quality and profitability by reducing the occurrences of detrimental processing results.

With a deep respect for the intricacies of manufacturing, SYSTEMA’s team of seasoned experts is dedicated to crafting and integrating bespoke automation solutions that address your most pressing challenges while optimizing your production processes. Our rich experience in the realm of manufacturing, coupled with a nuanced understanding of automation, positions us to significantly elevate the reliability, efficiency, and profitability of your operations. For more information, please feel free to contact us.