Gallery: How Teeny, Tiny Transistors Are Born in a Near-Total Vacuum
01image-10
Nanotransistors just got a lot more nano. A new chip construction process cooked up by Applied Materials in Santa Clara creates transistors so small they can be measured in smatterings of atoms. The company can now coax a few dozen of the little guys to assemble themselves into a base layer that helps control the flow of electricity on computer chips. The biggest development is the manufacturing process: Applied Materials devised a way to keep several interconnected manufacturing machines in a near-total vacuum—at this level, a single stray nanoparticle can ruin everything. The other part of the breakthrough is making this base from hafnium (used also in nuclear control rods) instead of the standard silicon oxynitride, which is terrible at holding back electrons on a supersmall scale. (Gordon Moore himself has called this technique the biggest advancement in the field in 40 years—and it is likely to keep processors advancing on pace with his eponymous law for the foreseeable future.) Applied Materials' system means transistors can be about 22 nanometers wide, as opposed to the current standard of about 45 nanometers, resulting in smaller, cheaper computing devices. Here we explain how the shrinking happens.
Caren Alpert02prep-the-wafer
Prep the wafer -------------- A silicon wafer is first bombarded with high-energy particles to define the area that will form the foundation of each transistor. The wafer weighs 128 grams and costs about $100. After as many as 300 billion transistors are added, the weight goes up by only about a gram, but the finished slab can be carved into chips worth up to $100,000. *Photo: Caren Alpert*
Caren Alpert03load-it
Load it ------- An extremely precise robotic arm pulls the wafer into a machine, where a filtration system removes dust particles from the surrounding atmosphere and a spray of negatively charged ions neutralize static electricity. A second robotic arm then moves the wafer to the next chamber. *Photo: Caren Alpert*
Caren Alpert04seal-the-chamber
Seal the Chamber ---------------- To make the manufacturing environment even cleaner, vacuum pumps suck nearly all the air out of the next chamber until the pressure is 1/10,000th of atmospheric pressure. This chamber is carefully designed to prevent condensation and turbulence, which could contaminate the wafer. *Photo: Caren Alpert*
Caren Alpert05lay-down-the-base
Lay down the base ----------------- The wafer is heated to as high as 1,800 degrees Fahrenheit and exposed to single oxygen atoms, which are extremely reactive. The oxygen reacts with silicon in the wafer, forming a layer of silicon oxide. This three-atom-thick layer creates a sticky base for the rest of the transistor to adhere to — like applying primer before a finish coat of paint. *Photo: Caren Alpert*
Caren Alpert06deposit-the-hafnium
Deposit the hafnium ------------------- Now the machine deposits a single layer of hafnium oxide (HfO~2~) molecules to the sticky base on the wafer’s surface. The process is repeated until there are about 12 layers of atoms (about 3 nanometers high). This thin layer of HfO~2~ — not the standard silicon oxynitride — will form the base of the gate. (The gate is what turns the transistor on and off.) *Photo: Caren Alpert*
Caren Alpert07form-nitrogen-compounds
Form nitrogen compounds ----------------------- A combination of HfO~2~ and nitrogen is better at holding a charge than HfO~2~ alone. The two are combined in the next chamber. Several nitrogen-containing gases are zapped to form a plasma, and voltage is applied to the wafer. This causes the nitrogen atoms to shoot down into the wafer and deposit themselves among the HfO~2~ molecules. *Photo: Caren Alpert*
Caren Alpert08complete-the-transistor
Complete the Transistor ----------------------- The wafer is moved to a rapid thermal processing chamber and heated again to between 1,300 and 1,800 degrees to distribute the nitrogen atoms and bake them into place. Once this step is complete, several other machines add connections, wiring, and components to finish the chips. *Photo: Caren Alpert*
