PROCESS, PROCESS, PROCESS ….
What I like most on processes is that even if they have a common concept, it can be made in so many ways that the results can be quite unique…
A process is made of several steps.
Let me show that if we choose a simple process like the initial cleaning of the wafer with the well-known, RCA process, everybody knows what will be done … but we have many steps to reach our objective.
Even obvious I will put that in bullets to be clear about the steps:
- Define the correct time and temperature for the cleaning process,
- Prepare the recipient or the equipment where the RCA process will be executed,
- Inspect optically the wafers,
- Put the chemicals in a certain mixture,
- Raise the solution temperature,
- Verify the temperature,
- Prepare how to introduce the wafers into the solutions,
- Introduce the wafers in the solution and let them stay there for a certain time,
- Neutralize the wafers using DI water,
- Control the purity of the DI water
- Inspect optically the wafers.
We have several small steps to be able to reach the major process line.
We need to remember that every single step can change completely the results of the major process and can change the final characteristic of the device or the yield of the production, or the production cost.
Processes has lots of possibilities, there are better options, but not unique one, and the influence of our choices can change completely the yield, cost, reproducibility, and characteristics of each device
State-of-the-art devices demand new processes, and creating the best new process is not a simple task. We need to embrace the idea that there is always possibility to improve it.
Even in old technologies we can do better like in rectifier diodes as an example, we can have a forward voltage drop 30-40% higher or lower depending on the process we choose.
Those that are working with packaging limitations of SiC devices, I´m sure that the lower heat dissipation will give new possibilities to this device. A lot of times, solving one issue will reflect in a major problem.
“Every day, every process can be rethought, restudied, redeveloped to get better results.
Process is correlated with the final device characteristics, yield, costs, and reproducibility, that´s a fact.”
When we automate a process, we are looking for cost´s and scrap´s reduction, improving reproducibility, increasing production speed, but we need to take care not to lose our ability to rethink the processes, to accept things as unique. We will always have the possibility to do it better, and processes are fundamental to it!
I saw an interesting article about reducing the energy consumption in the semiconductor area through using of ML – machine learning. The discussion was about using or not ML to reduce it. (https://semiengineering.com/rethinking-machine-learning-for-power/)
I think we all should read it, but we could add an idea that different processes are able to produce better devices … it would be more interesting yet to read it thinking that way, it gives us a very different perspective to solve a lot of issues.
Semiconductor Experience Sharing 