Why don’t wafers be square? Aren’t square silicon wafers more productive?

      Have you ever been pained by the thought of chips being cut off around a wafer?
      When making A CPU, GPU, or memory or SOLID-state disk chip, the corners of each crystal circle are cut off a lot (figure below) :

      Considering Intel’s currently tight 14nm capacity (and TSMC’s 7nm capacity is not enough), these Dies are an inhumane waste of every wafer. Why not make the wafers square? Square wafers are the perfect space usage rule for IC industry.

      All this, however, starts with the birth of the “wafer”.

      One of the reasonableness of wafer: InGOT generation rule

      The first step on the journey from sand to CPU chips is to turn the high-purity polysilicon material into silicon based chips suitable for “printing”. Here, we thank one of the biggest names in technology — J.Chralski:

      J. Chralski invented the pull method, which can produce a variety of artificial colorless sapphire, ruby and other gemstone crystals at low cost

      For the chip industry, the allure of artificial gems is certainly not enough

      Engineers have perfected the mound carats of pulling method, puts polysilicon material (seed crystal) seed crystal rod, heating in the quartz crucible (about 1000 degrees Celsius), still need to inject the appropriate proportion in the heating process of inert gas (air) rather than to simply, so that the polysilicon melting and chemical reaction with the impurities in the air created. On the way, the seed crystal rod is pulled and rotated according to the melting state of the seed crystal surface, so that the melt is in the supercooled state and crystallizes on the seed crystal (that is, the growth process of silicon crystal). Finally, the silicon crystal column ingot with cross section shape is pulled from the melt of the crucible:

      If the process of making silicon roadblocks by the above method is a little complicated, you will be able to see it in this GIF:

      In short, the method of making a cylindrical silicon anchorage is similar to the method of heating and drawing glass for thousands of years

      In the process of rotating the seed rod and stretching the seed melt, the cylindrical silicon anchorages are produced:

      Therefore, the basic shape of the silicon wafer will not change as long as the rules for making silicon anchorages remain the same.

      Wafer reasonability part ii: The round shape is more suitable for slicing, polishing and polishing of silicon substrate

      After the silicon is made, it is cut into thin wafers and pre-processed, such as polishing, before being allowed to enter the later lithography process:

      First: the diamond sand line slicing process of silicon anchorage is more suitable for circular substrate

      The above is a screenshot of Intel’s birth of Core video, made a decade ago, in which a silicon carbide cylinder is cut with a saw blade for animation purposes only. The silicon carbide is actually cut with a diamond wire:

      It can be seen that the cylindrical silicon roadbed is more suitable for uniform, rapid and complete cutting by diamond sand line slicer. If you square the wafer. I am afraid that the attrition rate of diamond sand lines and the stress dilapidations of the square silicon battleship will surpass the horizon

      Second: a wafer can withstand more grinding and polishing than a wafer

      After the diamond sand line is sliced, the surface of the silicon wafer (wafer) must be very rough. Therefore, to enter the grinding, polishing process. Obviously, a thin square of “glass” placed in a high speed rotating, vibrating polishing environment will result in a high probability of corner damage

      Third: wafers are easier to store and transport

      The finished wafer material needs to be stored and transported so that it can be easily manufactured for different IC companies. At this point, the wafer’s advantage over the “wafer” is once again apparent.

      Of course, these points do not represent the full scope of wafer rationality, only to the general reader’s convenience to understand the perspective of a brief explanation. In fact, there is a very complicated process involved in the photoligraphy of a wafer, and the shape of the wafer can be changed accordingly.

      In addition, there are other cutting shapes

      These detailed changes are generally made before the wafer goes through the final photolithography process. The main purpose is to facilitate wafer positioning, testing and so on during photolithography production.

      There is, of course, a square silicon substrate, but it is still cut from a cylindrical silicon cocoon and cannot be made directly from a furnace into a perfect silicon cube.