5
\$\begingroup\$

I've watched several programs which explain the fabrication of microchips essentially similar to developing film to the layperson. There is a silicon wafer, a special designed mask, and then light is used to "burn" the design into the silicon wafer in a multi-step process. One of the most modern ways to do this is called Extreme Ultraviolet Lithography.

Frankly, I understand the concept quite well. What I don't understand is how the burned-in result on the wafer includes a bunch of tiny electrical components which actually do something when potential differences are applied. If I search the Web for "transistor" images, I get a quite familiar electrical component with leads that could be plugged into a breadboard, for example.

My question is, what makes a "standard" transistor like one you may find in an electronics store in a bag, different from these etchings and how do we go from burning a pocket into a wafer to a functioning electronics component like a transistor? Additionally, what types of electrical differences may we see in a transistor used in a microchip vs a traditional, visible-to-the-eye transistor? If the answer to this question is too complicated to summarize here, some references would be appreciated.

\$\endgroup\$
  • \$\begingroup\$ "which actually do something " : In a chip they are connected to form complete circuits. \$\endgroup\$ – Oldfart Oct 2 '19 at 5:58
4
\$\begingroup\$

what makes a "standard" transistor like one you may find in an electronics store in a bag

You mean discrete semiconductors like transistors and diodes.

Most modern discrete semiconductors are indeed made in the same way, or at least a very similar way, as integrated circuits are made.

Both are made on silicon wafers using photolithography. Most structures aren't "burned-in" but etched or implanted.

The main difference between discrete semiconductors and integrated circuits is the process that is used. A process is like a recipe, it determines the thickness of layers for example. Also discrete semiconductors generally use more "coarse" structures compared to most IC processes. In general "coarse" processes are much lower cost than "fine" processes (like the ones using E-UV).

Also, in an IC manufacturing process we want to connect many components together to form a circuit, this usually requires much more complex wiring than a single transistor would need. For this, IC processes usually have many metal layers available (2 for very simple ICs, more than 8 for complex ones). For discrete components, this generally isn't needed so many of the processes used for this purpose only have one metal layer available. This also reduces cost.

\$\endgroup\$
  • 3
    \$\begingroup\$ One thought arises from a long discussion with an IC designer at Burr Brown. In an IC design, they know the load and source details intimately. This allows for precision sizing (and positioning.) Pin driver sections had no idea what they'd be driving or attached to, so those designs and aluminizing had to be conservative. Discrete devices would be 100% unknown, so their designs would be a balanced design of many compromises. \$\endgroup\$ – jonk Oct 2 '19 at 8:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.