# Part II: Controlling electricityFrom doped silicon to the transistor: electricity controlling electricity, and the supply chain that builds it at scale. ## Semiconductors A semiconductor is a material whose conductivity lies between a conductor and an insulator. Its conductivity can be controlled by altering its composition and also by applying electric fields. This controllability (not merely having conductivity somewhere between a conductor and an insulator) is what makes semiconductors useful. Silicon is the most widely used semiconductor material thanks to its abundance, stability, and compatibility with large-scale manufacturing. Small amounts of other elements can be added through a process called doping. When silicon is doped with elements like boron or phosphorus, its conductivity becomes highly controllable. Depending on the element used for doping, this either introduces extra electrons or creates "holes" that electrons can move into. The motion of these electrons and holes is influenced by electric fields, based on electromagnetism, which means their behavior can be controlled without any mechanical movement. By arranging differently doped regions and controlling them with electric fields, engineers can construct devices in which one electrical signal controls the movement of another. ## Transistors Transistors are the basic active component of all modern digital electronics. It is the smallest element in the stack where control appears. ### What a transistor is and how it works A transistor is a structure made from semiconductor material, carefully shaped and doped so that electrical current can be allowed or prevented from flowing depending on an applied voltage. What makes a transistor special is that this control is not made mechanically, like a physical switch, but electrically. In other words, electricity is used to control electricity. Under normal conditions, the transistor sits in a "closed" state: electric current does not flow through it. When a specific voltage is applied, it changes the internal electrical environment of the semiconductor material which reorganizes the charges inside the transistor and opens a path for the electric current to flow freely. If the voltage is removed or reduced below a certain point, the internal conditions return to their original state and the path disappears, which causes the current to stop flowing. In conclusion, a transistor can reliably switch between two states: allowing current to flow (on) or blocking it (off). By assigning meaning to those states (on as 1, off as 0) electrical signals become information. When many transistors are connected, the state of one can influence others, enforcing logic and rules like "if this is on and that is off, then allow current here." This is the basis for computation. ### How transistors are made The earliest transistors were experimental devices developed in the late 1940s and 1950s. They were physically separate devices made from chunks of a semiconductor material called germanium, assembled by hand and wired together one by one. The next phase emerged with the introduction of doping, forming the so called PN junctions. Fabrication remained largely manual, and each transistor was still produced as an individual component. A major transition occurred with the adoption of silicon. This made it possible to use the planar process for transistor manufacturing, which consists in building them layer by layer on a flat piece of silicon. The starting material is a silicon wafer, which is a very thin, shiny and smooth disk cut from a large crystal of pure silicon. To control where the surface is modified, photolithography is used: The wafer is coated with a light-sensitive material and exposed to light through a mask, which is a plate with a specific pattern that blocks light in some areas and allows it in others. This creates precise patterns on the surface. Specific regions are then chemically modified by adding small amounts of other elements to change their electrical properties. {{< figure src="images/tech/from-atoms-to-pixels/wafer-processing-eng.webp" alt="Diagram of semiconductor wafer processing: photolithography, patterning, and layer-by-layer modification of the silicon surface" caption="Wafer processing builds transistors layer by layer on a silicon disk through patterning and chemical modification." >}} By repeating these steps, complete electronic devices are built on the surface of the wafer. This turned the transistor from an individual laboratory device into a repeatable building block that could be manufactured by the thousands, millions, and eventually billions on a single chip. In the last decades, the basic principles of the planar process have stayed the same, but its scale has improved enormously. Precision is down to the nanometer scale, fabrication is done in ultra-clean environments, where even a tiny particle can ruin the process. At this scale, modern chips such as those found in smartphones can integrate tens of billions of transistors (~20,000,000,000), each switching billions of times per second. The modern transistor supply chain is highly fragmented and specialized: 1. **Raw materials and chemicals:** The base of the supply chain, with companies providing ultra-pure materials. Major examples are silicon wafers ([Shin-Etsu Chemical](https://www.shinetsu.co.jp/en/), [SUMCO](https://www.sumcosi.com/english/)) and specialty chemicals ([Air Liquide](https://www.airliquide.com/), [Linde](https://www.linde.com/), [JSR](https://www.jsr.co.jp/jsr_e/)). 2. **Manufacturing equipment:** This layer is the most technologically concentrated. Advanced lithography is effectively monopolized by [ASML](https://www.asml.com/) (whose systems critically depend on ultra-precision mirrors supplied exclusively by [Carl Zeiss SMT](https://www.zeiss.com/semiconductor-manufacturing-technology/home.html)). Etching and deposition tools are supplied by companies such as [Applied Materials](https://www.appliedmaterials.com/), [Lam Research](https://www.lamresearch.com/), and [Tokyo Electron](https://www.tel.com/). 3. **Fabrication plant (fabs):** Leading-edge production is dominated by [TSMC](https://www.tsmc.com/), with additional capacity from [Samsung Electronics](https://semiconductor.samsung.com/) and [Intel](https://www.intel.com/). These companies operate fabrication plants costing tens of billions of dollars each. 4. **Chip design (fabless companies):** Many firms design chips but do not manufacture them. Examples include NVIDIA, AMD, Qualcomm, and Apple. They rely entirely on external manufacturers for production. 5. **Packaging and testing:** After fabrication, chips are cut, packaged, and tested by specialized firms such as [ASE Technology](https://www.aseglobal.com/), [Amkor](https://www.amkor.com/), and [JCET](https://www.jcetglobal.com/). This extreme specialization leaves key materials and equipment dependent on very few suppliers, making disruptions very hard to absorb. High capital costs, long fab build times, rising manufacturing complexity and geopolitical tensions (i.e. Taiwan) increase system fragility and elevate overall supply-chain risk. ### The layers above and below The semiconductor layer beneath the transistor provides the physical mechanisms which the transistor organizes into reusable elements of control. The layers above assume that these rules can be chained together, that the output of one decision can serve as the input to another, and that these chains will behave consistently over time. The next part examines how physical arrangements of transistors become gates, memory, and the first true layer of computation.