How much do defects cost for manufacturing companies ?
I am preparing the presentation for my phd thesis and I would like to add some really concrete motivations to the introduction slides.
My thesis focuses on the 3D-growth of III-V semiconductors monolithically grown on silicon. In the first part of my talk I should introduce the defects due to the monolithic growth of III-V on silicon and the different strategies used to filter these defects.
Therefore, I was thinking of starting with a slide of main motivation which will let the audience understand the following problem: defects affect the cost in semiconductors devices, therefore manufacturing companies are wasting money due to this. If we solve this problem we could potentially save this amount of money.
However, I want to be more clear than that and give them some actual numbers. How much do defects cost for manufacturing companies ?
Of course, it would be amazing if you could point out some market analysis (qualitative and quantitative analysis), survey papers, etc.
Thanks for your help.
You touche a very important point of the effects of material defects on the final cost of the devices or integrated circuits. What you ask for is the yield of the manufacturing process which is measured by the ratio of the accepted chips to the total number of chips on a wafer. The yield is a resultant of material defects and processing faults because of not fully controlled manufacturing parameters.
There is an important concept that there is specifications of the material before going to the next step of the process.
In your case you have to fully assess your epitaxial layer such that you have to be sure that they will result in good operating devices. The characterization includes the thickness, the homogeneity, the doping concentration if any, the crystallographic defects. Crystallographic defects play a very important role in the characteristics of the devices made from such material. Ideally, one has to avoid completely such defect formation during growth process. This is not practically impossible and so there is the accepted level of defects which means the defects must be kept below certain threshold to avoid affecting the device performance.
The defects are geometrical defects and chemical defects. The chemical defects are the deviation from Stoichiometry , and the inclusion of impurities. The geometrical defects are point, line , area and volume defects. Point defects as vacancy Line defects such as dislocations , area defects such as stacking faults and volume defects such as tripyramides.
One has to control his process to reduce these defects to acceptable limit for the device production.
It is so there will be additional defects that will be produced during processing so one has always to monitor the process during the fabrication.
Geetak's document addresses the so called manufacturing yield due to wafers' size, die size, and defective dies distribution function (that is manufacturing parameters initiated defect distribution model on a wafer). It is not exactly what you looking for, as you want to show a single die yield improvement by such process steps that reduce the density of defects on a die by changing process and / or A3B5 components parameters. In other words, wafer statistics are not your key interest at first, until you understand what are the key mechanisms that reduce defects on a single die. If you know the answer, all you need to do is to assess the cost increase or decrease that is related to the defects reduction and then to see, if the overall manufactured dies cost sustainable or not with the positive effect of each particular device functionality improvement (as far as say reliability, parameters of operation improvement, etc.). The next more sophisticated step could be integrating your one die improvements model into the overall manufacturing yield. However, for the PhD theses probably you can postpone it.