What’s inside a microchip ?

Microchips are advance electrical devices that made up of a dense network of small semiconductors that formed in layers on a thin wafer of semiconducting material, commonly silicon. The surface layer of a microchip’s silicon die typically hidden from view. However there are several instances where we allowed to look inside. And what we can see when we do so is simply amazing. I’ll attempt to take you on a brief tour of a microchip’s visible internal workings. And explain as best I can the sections I’m showing you and how they work.

So, here’s an illustration of a chip.

Here’s one that’s a lot more rational. EPROM, or EROM, stands for ‘Erasable (Programmable) Read Only Memory,’ and these chips are used to give a system with operational code or firmware. It’s designed to be programmed once and deleted only by shining an ultraviolet light through this quartz glass window onto the silicon.

And as you can see, the window makes it quite handy for us to peek inside and examine the silicon’s surface layer. So let’s examine it with an optical microscope. There isn’t a lot of detail apparent at 100x magnification, but we can see the majority of the essential elements.

The memory blocks are two enormous. Seemingly empty spaces that can contain billions of tiny transistors, each of which stores a bit of data. This particular chip is a ST Microelectronics 27C512, which can store 512Kbits of data. Therefore there will be at least 512,000 of these transistors, each of which will store a 1 or a 0 by flipping on or off.

If you increase the magnification, you might be able to see a regular pattern of transistors. This is about 800x magnification, so you can see how tiny these are. To see any acceptable detail in the individual transistors, I’d need far more than this. I could acquire it by using this third objective lens, but because the silicon is so deep inside the package. Focusing on it would require this lens to be so near that it would block all of the incoming light and we wouldn’t be able to see anything. Nothing is flawless, as usual.

The supporting circuitry found around the memory blocks. The only 28 pins on this package utilized to interface between the thousands of transistors in the memory blocks and the external circuitry distant from the chip.

Read More: How computer memory works

This circuitry does something convoluted to connect all these transistors in such a way that they can all accessed through these few pins. It would highly impractical to have one pin dedicated to each transistor. That would require over half a million pins on the package, which for obvious reasons can not done. So this circuitry does something convoluted to connect all these transistors in such a way that they can all accessed through these few pins. I’m not sure how this one accomplishes it, but I’m guessing it uses some sort of addressing system that connects a small number of the transistors storing the relevant data to the output as and when they needed, dependent on some sort of address input. ‘M752’ appears to be some sort of identification code.

I have no idea what that’s for, and I can’t find any information about it. Unfortunately, that’s the closest we get to any kind of chip art on this one, which isn’t very creative. The output buffer amplifiers, which interface the high output impedance of the silicon circuitry with the low input impedance of the external circuitry, found closer to the silicon’s extremes. This prevents extraneous circuitry from loading and interfering with the silicon circuits’ operation. In concept, this is similar to a voltage follower using a unity gain op-amp.

We can see the bonding wires connecting the silicon to the pins on the package off to the sides here. There are 30 of them, presumably a couple paired up for current carrying capability. There’s a pair and then there’s a third.

Those are most likely the vcc and ground. These wires are all made of silver, which is the best conductor of electricity. They appear to have been pressed onto the pads on the silicon die; I’m not sure if they resistance welded or not, but they affixed to establish an electrical connection in some way. There’s nothing else in this package save the die in the centre and the bonding wires that connect to the pins, so there’s a lot of wasted space. It’s astonishing to believe we can accomplish this with just a little ROM chip. When compared to current microprocessors, this is nothing.

And this is just the visible layer; there’s a lot more going on beneath the surface that we can’t see.

It’s quite broad, and it’s fascinating to see how much has evolved in the field of electronics. Electronics with integrated circuits (ICs) now have virtually little personality. It’s all tucked away and thrown away, referred to as ‘integrated.’

Also practical and reliable, but it lacks something that doing everything with individual components does. A certain dexterity. I’ll still use a pair of transistors in a circuit if I only require a single logic gate, just because I believe it’s better than putting in an anonymous DIP to handle everything for you. It might be simpler to just use a 4000 series gate, but my electronics work is extremely important to me, and I sometimes prefer to do things myself rather than rely on an off-the-shelf item to do everything for me.

That’s not to say that ICs are bad. They do a great job, and I’ll use them when I need convenience. But it’s wonderful to have some control over some things.