Changing the recipe and learning to do things better

The project is starting to stall. One of the teams is having a hell of a time getting the device working. They changed too many things too quickly and the boss wasn't expecting so many problems. But what else do you expect?

I worked on a different project and just finding a replacement material and then actually using it took about a year to get things to where they are now. The team wants to change the entire design and keep all the previous good characteristics of the old design and have the whole thing done in half a year. Unfortunately that didn't work out. The lead time to make something and then have things tested is much longer then that.

The problem with semiconductor processing is that it is a long drawn out process to make something. When things do work, however, it can be quite amazing. I think there are possibilities for big changes in the industry if the R&D aspects of developing new devices can be
significantly sped up.

The problem with semiconductor device R&D is that a lot of expensive, manual and slow machines are used. Washing and making sample surfaces clean is a 30 minute process and even then, sometimes we have to redo it if some large dust particle falls upon the surface. The patterning
process takes at least 45 mins to 1 hour, depositing metal contact takes about 2 hours. Removing unwanted metal takes probably about another 30 minutes...

Each processing step takes about 1 ~ 3 hours depending on the step and to create a single device, there are at least 40~50 steps. While processing the semiconductor material, it is also important to check the wafers as they go through the machines to make sure nothing goes
wrong, and if strange does happen, it is important to figure out the causes. With all the checking and debugging steps going on, it easily takes 3~4 weeks to finish a batch of devices and then send them into evaluation. After evaluation, the results are fed back into the process team and the cycle restarts. It takes about 2 months to feed back data and have new results. The catch is also that for the devices that I am making, the process is relatively short.

A crucial part to commercializing new technology is to have a R&D department that can rapidly carry out new experiments and get results with blazing fast speed, because in research it is a race to see who is the first able to develop something new for commercialization. The problem about working in R&D is that people exist in a bubble. In order to prevent leaks, there isn't a whole lot of cross talking going on unless you have friends in different departments that know what is going on. When it comes to knowing what is going on in other companies, you are going to have no clue because of strict secrecy.

The interesting thing is that there are big surprises to be had when it comes to corporations competing against each other in research. Basically the ones that are the most creative, the best equipped (and by this, I do not mean by having the most expensive equipment) and the fastest are the ones that will win when it comes to coming to market with new technology. The thing is that both party will have no idea how each group goes about doing their work.

The good thing about working in the R&D department of a large corporation is that they have a lot of money to throw around (as an aside, materials for use in R&D is generally 2~3x more expensive, the mark up via suppliers must be huge) and that they can spent plenty of cash for new equipment. The problem is that most of the equipment is pretty standard stuff and there is a lot of room for improvement to make the machines we use more efficient at getting things done.

I believe that my biggest gripe about the R&D process that I know so far is that companies are willing to invest in equipment to get things done, but there isn't a lot of people in the field that consider how to do things better. If there are companies out there that can make tools to do semiconductor R&D better then there is good money to be made here.

The problem however, is knowing how to get your foot into the door. As an outsider, most companies are too secreteive about their research projects to let anyone else know what is going on to develop things that might be useful, but at the same time, not knowledgeable enough to
develop equipment to do research better. It's a catch-22 problem and one of these conditions have to give before faster advancements in semiconductor technology can be achieved.

The only solution I can think of now is for a company to hold their own R&D semiconductor processing line and invest actively in improving how it is used internally for research, because without more openness, it is going to be hard to innovate in this field. The only company actively doing this that I can think of is Intel.

There is always the academic route (universities), the problem however, is that universities are usually underfunded when it comes to wanting to innovate in this area.