Why are Chips Square, but Wafers Round?

In popular perception, a wafer is a thin, round piece of high-purity silicon crystal. Various electronic circuit elements can be fabricated on this high-purity silicon crystal, turning it into an integrated circuit product with specific electrical functions. Numerous components are neatly arranged on a single crystal silicon material, all square and orderly. Hence, in practical applications, wafers need to be cut into square shapes.

Why are silicon wafers round? Why “wafers” and not “squares”? Why do some wafers have no chips on the outer edges, while others have incomplete chips around the perimeter due to the circuitry covering the entire silicon piece?

Those familiar with semiconductor manufacturing know that all manufacturing processes, prior to chip cutting and packaging, are performed on the wafer. However, the chips we see are square, and manufacturing them on a round wafer inevitably leads to some unused areas. So, why not use square wafers to increase utilization?

The answer to this question is quite straightforward. Wafers (initially silicon slices) are cut from cylindrical silicon rods, so their cross-section can only be circular.

Let’s explore the curiosity about why wafers are more suitable than “silicon squares” for making chips by examining the manufacturing process of silicon slices and wafers.

The production of silicon wafers can be summarized in three basic steps: silicon refining and purification, single crystal silicon growth, and wafer formation.

The process begins with the purification and melting of silicon sand. This stage primarily involves dissolving, purifying, and distilling to obtain polycrystalline silicon.

Next comes the growth of single-crystal silicon. High-purity polycrystalline silicon is placed in a quartz crucible and melted at high temperatures in a protective atmosphere. A small seed crystal is slowly lifted from the rotating melt, allowing the vertical pulling of a large diameter single crystal silicon ingot.

The final step is wafer formation. Single crystal silicon ingots are typically cylindrical, with diameters ranging from 3 inches to several inches. After slicing and polishing the silicon ingot, single crystal silicon slices, or wafers, are obtained.

In essence, it’s not wrong to say that chips are fundamentally a bunch of sand. The foundation of chip manufacturing, single-crystal silicon, is derived from quartz sand.

The key to transforming sand into chips lies in the development of silicon purification and single crystal silicon fabrication technologies. In 1916, the Polish chemist Jan Czochralski experimentally verified that metal wires are composed of metal single crystals and that the diameter of these single crystals could reach the millimeter scale. Subsequently, this method, through continuous iterations by chemists, eventually led to the production of single crystal silicon. This technique is known as the Czochralski method or the pulling method.

The pulling method involves first heating high-purity silicon in a crucible to a molten state, then placing a seed crystal at the end of a rod with a precisely oriented direction, and immersing the end into the molten silicon. The rod is then slowly pulled upward and rotated. Through precise control of the pulling rate, rotation rate, and temperature, a large cylindrical single crystal silicon rod can be formed at the end of the rod. Subsequently, the silicon rod is ground, polished, and cut into usable circular silicon slices. Therefore, the round shape of the wafer is due to the “round” silicon rod.

Currently, the pulling method is the most commonly used technique for growing wafers. Besides the pulling method, another commonly used method is the float-zone method, also known as the Fz method.

The concept of “zone leveling” was first conceived in 1939 by W.G. Pfann at Bell Laboratories. With the assistance of Henry Theuer and Dan Dosi, they grew high-purity germanium and silicon single crystals, eventually earning a patent.

This method involves using thermal energy to create a molten zone at one end of a semiconductor polycrystalline rod, allowing it to recrystallize into a single crystal. The molten zone is slowly moved along a defined direction towards the other end of the rod. This process transforms the entire polycrystalline rod into a single crystal rod. The float-zone method also requires a seed crystal, and the crystal orientation of the resulting cylindrical single crystal ingot matches that of the seed crystal.

The float-zone method is divided into two types: horizontal float-zone and vertical floating zone. The former is mainly used for the purification and single crystal growth of materials like germanium and GaAs, while the latter is primarily used for silicon. The main difference between the float-zone method and the pulling method is that the former generally does not use a crucible, introducing fewer impurities, resulting in a material with lower impurity content.

In summary, since the single crystal silicon rod is cylindrical, the silicon wafers produced using this method are naturally circular.

Indeed, a silicon rod could theoretically be cut into a rectangular shape before slicing, which would yield square “wafers.” However, silicon rods used for chip production aren’t processed this way due to several reasons:

Firstly, the circular shape is more conducive to photolithography and coating processes.

Additionally, due to the presence of edge stress, circular wafers have higher structural strength compared to square ones. Silicon slices undergo multiple processes like photolithography, etching, and chemical mechanical polishing before becoming wafers. These processes tend to accumulate stress along the wafer’s outer edges. Therefore, the sharp corners of a square shape would concentrate edge stress, making them more prone to damage during production, thus affecting the overall yield rate.

You might wonder why some wafers have no chips at their outer edges while others have incomplete chips around the perimeter due to circuitry covering the entire silicon piece. This variation occurs because the layout and design of the chips on the wafer are optimized for space utilization and production efficiency. In some designs, it’s more efficient to use the entire wafer surface, including the edges, even if it results in incomplete chips. These incomplete chips may still contain functional parts or be used for testing and calibration. In other designs, the outer edge is left empty to avoid the complexities and lower yields associated with edge stress and the irregular shapes of incomplete chips. The choice depends on the specific requirements of the chip design and the manufacturing process.

The issue with the edges of the wafers and the presence of incomplete chips is indeed related to the size of the photomask (or Mask) used in photolithography. The photomask usually covers the entire area of the wafer, leading to the appearance of incomplete small squares at the edges. Additionally, during the chip manufacturing process, the wafer thickness increases, particularly in the latter stages involving metalization and via creation, which utilize multiple Chemical Mechanical Polishing (CMP) processes. If there are no patterns at the wafer’s edge, it can lead to a slower grinding rate at the edge, resulting in a height difference between the edge and the center. This discrepancy can then affect the adjacent complete chips in subsequent grinding processes.

However, chips at the outer edge of the wafer are generally not used. As mentioned earlier, due to the production process, there will inevitably be some stress at the periphery of the wafer. Chips produced in these areas will also retain internal stress, making them more prone to damage during subsequent cutting, packaging, and transportation processes. This is why some wafers have chips at the edges while others do not.

Overall, circular wafers are more conducive to chip manufacturing and have a higher yield rate. Given that wafers used for chip fabrication are not conveniently made square, why can’t chips be circular?

The answer lies in the practicality and efficiency of design and usage. Square or rectangular chips make better use of space on a printed circuit board (PCB) and allow for more efficient routing of electrical connections. In contrast, circular chips would waste PCB space and complicate the layout and wiring of the board. Additionally, square chips can be tightly packed without wasting space, making them more cost-effective and efficient for mass production. Thus, despite the initial production of circular wafers, the final chip design is square or rectangular to optimize functionality and manufacturing efficiency.

Indeed, manufacturing circular chips is more challenging than square ones.

Silicon wafers undergo processes like coating, photolithography, etching, and ion implantation to create individual chips. However, these chips are still attached to the wafer and need to be cut out to become separate units.

Square chips can be quickly and efficiently sliced from the wafer with just a few cuts. In contrast, cutting out circular chips would require significantly more time and effort. Most importantly, circular chips wouldn’t solve the problem of silicon area wastage.

Maximizing wafer area utilization has always been a critical concern. The more chips that can be produced from a single wafer, the higher the production efficiency and the lower the cost per chip. Currently, the best way to improve production efficiency is by increasing the wafer’s size, a concept well-understood in the field of microelectronics.

Besides chip fabrication, silicon wafers are also crucial in the photovoltaic (PV) industry.

The early stages of manufacturing single-crystal silicon for photovoltaic cells are similar to those for chip fabrication. The choice of square shapes for PV cells is also straightforward. If solar cells were circular, gaps would appear when multiple cells are arranged into a solar panel, reducing the overall conversion efficiency.

Compared to chip manufacturing, the purity requirements for silicon used in solar panels are slightly lower. The purity standard for PV silicon is 99.9999%, which is less than the 99.999999999% required for chip fabrication. This difference reflects the varying demands of electronic functionality versus solar energy conversion.


To answer the questions posed in the title:

Why are chips square?

Square chips are easier to cut and handle in subsequent packaging stages. Most importantly, square chips address the issue of wafer area wastage more efficiently than circular chips. The shape allows for better utilization of the wafer surface, reducing waste and lowering production costs.

Why are wafers round?

Round wafers are more convenient to produce due to mechanical factors and yield higher production efficiency. The natural shape of silicon rods, from which wafers are cut, is cylindrical, leading to the round shape of the wafers. The round form also aids in processes like photolithography and reduces stress concentration, which is more pronounced in corners.

However, in the field of photovoltaics, square silicon slices are preferred because they do not waste space when assembled into solar panels. Photovoltaic silicon slices are square to maximize the area coverage of solar cells on panels, enhancing the overall conversion efficiency of sunlight to electricity. The purity requirements for silicon in photovoltaics are slightly lower than those for chip fabrication, reflecting the different demands of these applications.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top