Manufacturing Process for Semiconductors Today. Could There be Better Supply Results?
“Could it be a better result? By this, I mean should there really have been an extreme shortage and long lead times?”
A fellow writer was an engineer at OnSemi. As he stated once, his job was watching the crystals (future “wafer material) grow. More coming on this comment shortly. You will not be experts (I am not either) in the process. Hopefully, you will understand why the issues associated with manufacturing semiconductors are questionable. I was purchasing for automotive and electronics companies following spec-sheets.
I will offer an opinion at the end of this dialogue.
The Silicon Wafer Production Process
Silicon wafers result from the slicing of manufactured ingots which undergoes multiple processes before becoming a part of a semiconductor:
The first step in silicon wafer production is growing a nugget of silicon, also referred to as a silicon ingot. Growing a single silicon ingot can take as less as one week to up to one month. Size, quality, and the specifications of the wafer determines the time taken for the ingot growth. One of the most common methods used to grow the crystal is the Czochralski Method or the CZ Method. Processing using CZ process, consists of placing polycrystalline silicon pieces in a quartz trough.
Adding small quantities of dopants determines crystal properties. Dopants are elements from Group 3 and 4 of the periodic table. Boron, Phosphorus, Arsenic, or Antimony are Dopants. The properties of the resultant wafer material may be P (positive) type (Boron) or N (negative) type (Phosphorous, Arsenic, Antimony). Heating the material above a temperature of 1412 degrees (usually above 1500 degrees) liquefies the crystal.
A silicon crystal having the same orientation (P or N) is the intended end result of the process is placed inside as a seed to facilitate the growth of the crystal with minimum defects. Once the diameter of the crystal, the process is stopped by stabilizing the conditions.
The resulting material is referred to as electronic grade silicon to EGS. Silicon ingots are cut and molded into the shapes needed for the final silicon wafers for the semiconductor, like those discrete semiconductors produced by NXP Semiconductors.
Ingot diameters are usually a little larger than what is in finishing specifications. The ingot goes through a slicing process creating the wafer after inspection. A diamond edge saw (or thin wire) process creates a thinner and smaller wafer without causing any major damage. Builders who are in need of cast iron to use for construction should learn about the benefits of hiring a brass foundry expert.
Lapping is the next step in the process. It consists of removing the marks, abrasions, and defects from the sawing. This step also serves to thin out the wafer a bit more as well as relieve the stress that the wafer has undergone during the slicing process.
After lapping, the wafers pass through chemically etching and a cleaning process. This further helps to alleviate any cracks or surface defects that may still be there, even on a microscopic level. The bath consists of sodium hydroxide, acetic acid, or nitric acid.
After cleaning, the wafer passes through polishing to reach its final stage. Performing wafer polishing is in a clean room where particle quantity per cubic foot of air is to Class 1 to Class 10,000 specifications. Workers have to wear clean suits, work under a fan that blows away excess particles, and maintain a level of cleanliness on their workstations. High quality silicon wafers usually go through two to three stages of polishing which can be a single side or double side polishing. Most wafers are subject to polishing on the front. The polishing process itself has two stages;
- The first stage being the stock removal process removing a thin layer of silicon to make sure the surface does not have any defects and
- A second stage of a chemical mechanical polish, not removing any silicon layers creating a shiny, mirror finish.
What Happens after the Wafers are complete? “Packaging“
The silicon wafers go through polishing and pass through a series of cleaning baths. The wafers pass through inspections multiple times under high intensity lights to ensure there are no excess materials or defects on the surface. After the cleaning and inspection, the wafers meeting specifications go through packaging and sealed. Vacuum sealing the wafer in plastic packaging is important as to protect them from damage and keep moisture out. Upon the completion of packaging, the wafers go to Fabrication facilities and used as semiconductor substrates.
Better explained pictorially (click on it to make it larger)
It was but a little more than a decade ago when the US was experiencing similar semiconductor shortages. My thinking back then was we will learn from the economic shutdown and prepare. Twenty twenty-one and we were no more prepared for an economic shutdown now as we were then. Closing down the economy shut down a low labor input manufacturing process that could have still manufactured semiconductors. Moreover, if you’re planning to purchase label printing for the packaging, you might want to try these Brother compatible labels. You must use specially designed and manufactured labels.
We are experiencing the same issues as what were experienced after 2008. We learned nothing from the last experience of maintaining orders and lead times. The planning of future orders stopped and orders canceled. No reserves were kept to cover a future startup. Today’s shortages did not have to be. There is a strong possibility the production process could have continued with little risk to people.
All sounds vaguely familiar.
Ken and Others:
The wafer process is not Labor intensive and neither is the Packaging side. It is like SMT lines which you set them up and let them run.
Automotive did not keep the orders going to the Manufacturers for whatever reasons. These things do not decay when in inventory. There are only a few reasons I can think of for not keeping the orders flowing:
– They are stupid
– They did not want the inventory is stock.
– Or they believed they could rent take which they are doing.
I worked in automotive for a number of years. In one company I came to after they did it. They got rid of 3 million semiconductors when the economy was down after 2008 which they needed in 2009 when I came on board. German company.
I was kind of there from the beginning; did controls for drawing the ingots, the lapping (Siltec, I think), …; even earlier. I remember meeting with Fairchild, National, …; The discovery of the groundwater contamination; the birth defects among the women workers (the lines were almost always staffed with women); the ‘boats’ before automation; readying, packaging, the machines with my controls on them for shipment to Indonesia, … .
This was all in the mid to late 1970s.
I got my BA with a minor in Math in 74 after 2.5 years. I was in a hurry having married this gorgeous woman from NYC who could make great spaghetti sauce like my mom. She was working as a paralegal making more money and I. She did not care and she wanted kids. I obliged and the rest is history. Fifty years later, we are boarding the white ship to Arizona for now. Our first new home.
I digress . . . I worked for Motorola for a while and they were talking about these phenomenal little squares they were wiring to boards with wire like spider web threads at Motorola. It was slow and tghey were on the cutting edge. At Marquardt I was arguing with suppliers about the chips we needed even after the fools dumped their inventory before I got there and to no avail. I could jump and down and they did not give a shit. We got our allocation and they made money.
When you have a steady demand, you gamble, and you keep your demand going. The fools did not for the 2nd time in my life time. Either by choice or just plain stupid. The Germans did not cross me. I made it a point of knowing more than my bosses and the sup[ply base knew it.
The labor in making chips and especially the wafers is minimal.
One thing to note is that when we enter a supply shortage (we referred to it as “in allocation”) we lose track of what demand really is. People who can’t get their product will double their orders hoping that suppliers will increase their allotment. Staying in allocation too long is a nightmare of increasing fictional orders. When things start to stabilize, those extra orders will all be cancelled.
I suggest (without additional evidence) that when real demand for things consumers wanted at home that were different from what they would have wanted from work, experienced suppliers may have discounted those orders.
Thank you for your comment. I do know what you are talking about as far as allocations. Maintain lead times is vital. At capacity will be be offset by proven demand as what occurs in automotive and history secondarily.
From a wafer to an ink jet chip takes about a month. If you send the people running the machines home, you have scrap.
I see how I was not clear. When we did not get our lines running fast enough demand for inkjet could outstrip supply. Our distributors were in allocation as we had to decide who was going to get what we could produce. Demand was increasing for over a decade, so proven demand was a sketchy concept. We were still building new lines for 5 year old products even as we were bringing up new lines for new products. Although management convinced itself to always be ahead of demand we had failures. Leaving money on the table really hurt them.
That was over 20 years ago. When customers started to realize they did not need to print so much proven demand dropped more rapidly than expected. That hurt in a different way.
In some places I was the planner or P&IC Manager. In others, I was the Commodity Manager or Purchasing Manager. I scheduled, planned (MRP), implemented MRP, did the logistics, purchased, etc. I was that end of the business and was active which many did not like. I usually found ways to increase capacity for “needed” parts. I made sure I knew the process and the capacities in each step. On tooling I suggested the cavities for plastic components, etc. If they did not follow my suggestion, they would be reminded. I cost modeled the parts and talked with the suppliers on parts. Outside supply base and I had a working relationship.
I was known for being respectful to the supply base and doing what I said I would.
So yeah, I understand what you are saying. That stuff happens and you can only be as good as the tools they give you to work with.
This post is a reckoning as we have been in this situation before. They knew what would happen if they completely shut down. Automotive knew they would need the parts, the semiconductors. Labor input today for this product is minimal. I believe they could have kept running. These were clean rooms and they were gowned. Yet, here we are “AGAIN” with everyone making money from the result of industries negligence based upon supply for the same old demand. It is contrived.