Translator: mariana vergnano Reviewer: Laura Díaz Aguirre [Pedro Julián - A trip inside a microchip] There is much more room at the back. "There is much more room at the back" is the title of a talk by Richard Feynman, a famous Nobel Prize winner in Physics in the second half of the last century, and obviously he was not talking about a room or a place that was at the back, he was talking about matter. He was talking about atoms and the room that was there.
And in that talk he said that in principle there was no obstacle to somebody taking atoms one by one, and move them to store information to build machines. Since that time, we've seen enough evidence of this, particularly in electronics. And to demonstrate that, we're going to shrink ourselves and go inside a microchip.
We are going to get really small. Several times over. But don't worry, we will do this together.
We are going to do it in six stages. And after that I will restore you to your normal size. Let's start with any microcontroller board.
You can see small black squares with legs. This is usually called a chip. Actually, that is a plastic casing.
The chip is inside, but to go there, we have to shrink a thousand times. When we shrink a thousand times, we see a square of a few millimeters in size. This is silicon.
Silicon is a crystal that is so pure that electrons can move inside it with almost no difficulty. If we get 300 times smaller, we will be 30 microns tall, which is more or less three times thinner than a human hair. At least on those of us who have finer hair.
And at that size, we can go inside through that fine wire, from the casing into the microchip. If we get 500 times smaller, we will get to see the wires in the microchip. Now, it's like we're standing on the roof of a building.
Looking down, we see wires. In a modern chip, they may have up to 10 different floors where these wires carry signals from inside the microchip. And on a chip of 1 cm by 1 cm, you would find up to 10 km of wire.
Let's get down along those floors. We will find that on any floor, they can connect any place you want with some place else. At any point each one can go up or go down.
Suddenly, we feel like in a maze. We continue going down and we get to the first 2 floors. When we get to the first 2 floors, we have to get 2 or 3 times smaller, because the wires get smaller.
And if we get to the ground floor, the ground is crystalline silicon. And there, we will find the transistors. What are transistors?
Basically, a transistor is like a tube where electrons pass from one side to the other. And over it I have a switch, a material that is like a switch, with which I can open or close the tube to allow the electrons to pass from one side to the other. Now, if we want to get through that tunnel, we'll have to get 10 times smaller, to 3 nanometers.
3 nanometers is the size of a transistor. And if we go with some electrons there, while we are passing, the only thing that we will see is 6 silicon atoms. Remember what Feynman said.
And what are transistors used for? Well, in principle, we could see them as switches that allow electrons or current to go through or not, and have some memory, because if you flip the switch, it remains flipped, somehow it remembers it is open, and if flips it back, somehow it remembers that it is closed. What can we use that for?
Well, it will depend very much on the number of switches that we are able to set up. If we use a single switch, we can turn a light on and off. If we use a hundred, we can control all the wiring of a house.
If we use a thousand, then we will be able to control something simple, like the electronic panel of a microwave oven. If we put in more than a thousand, we can make computers. And people always wanted to make computers to calculate.
In the year 1946, ENIAC was created as the first digital computer. It had 17 000 valves or switches. The valves were lamps of this size.
There were 90 000 other components, and it required 5 million welds made by hand. It weighed the equivalent of 5 adult male elephants, occupied the size of an entire house, and consumed the equivalent of 150 irons simultaneously plugged in. With this machine, the switches could be turned on or off 5000 times per second.
And this achieved a record for the time, namely, it did 300 multiplications per second. Obviously, with all these valves, after a long time, you got a valve failure every two days. So the valve had to be located and replaced, and that took an average of 15 minutes for a trained person to do.
In the year 1958, the concept of integrated circuits was invented. The idea was to put all these things together on the same silicon chip. In 1961, the first commercial microchip was manufactured.
It had 4 transistors. At that time, each transistor cost 10 dollars, and 10% of transistors failed. So nobody had much faith in this technology, except for the ones who were involved in making it.
The first portable calculator was made in 1964. The advertisement boasted about a chip that had 120 transistors, and the device was promoted as the first calculator made with space-age microcircuitry. That was the advertising strategy they used.
One year later, Gordon Moore, the co-founder of Intel, formulated his famous "Moore's Law. " He predicted that the number of transistors on a chip will double every 1. 5 years.
He even said that in a not-too-distant future, you would be able to buy computers in a supermarket, together with perfumes and other things. Indeed, in 1971, the first microprocessor was made for a personal computer, and it had 2300 transistors. And they could turn on and off 700 thousand times per second.
From there, progress was rapid. In the '80s, chips had hundreds of thousands of transistors. In the '90s, they had millions of transistors.
In the 2000s, they had tens of millions. Today, in the year 2013, the largest existing CPU has 7. 1 billion transistors, and those switches can turn on and off 3 billion times per second.
This is a bit difficult to imagine, so let's use an object from your everyday life. Imagine that in 1971, together with the first microprocessor, one got the most popular car of that year, which was the Fiat 127. If both had evolved at the same pace, what would that car look like today?
Well, that car would go up to 885 kilometers per hour. This means that it would reach the Moon in a second and a half. So going to the supermarket would be a bit complicated!
Especially the braking part. (Laughter) The car would weigh 60 milligrams, which is more or less the weight of 20 mosquitoes, on a diet! (Laughter) In the trunk you could fit 20 million bags, and the car would cost 1 cent.
Obviously, to put 7100 million transistors in a thing like that, a major technological effort is required. This is important. Drwing a transistor of 3 nanometers on crystalline silicon is roughly equivalent to getting an astronaut to an altitude of 6000 kilometers, and asking him to pretend to draw a coin on the surface of the Earth with exact detail and accuracy.
Obviously, that requires significant cost. Not because of the pencil length needed, but because of the factory, right? Nowadays, an integrated circuit factory costs 10 billion dollars.
And, as you know, every year and a half the number doubles, so that factory will last 2 or 3 years, because you have to build a new one, capable of making smaller chips. At this point you may think that making integrated circuits is not only difficult but also expensive. In fact, not so much.
Why? Because, for several decades now, design and manufacture have been separated processes. The design is done by an engineer with a computer, where they draw the transistors and metal connections.
And manufacturing is done elsewhere, in the physical factory that goes through all the process of making these chips. Then, for an engineer who is designing, making a transistor is something as simple as "use green paint to draw the tube where the electrons will go through, red paint to draw the switch which will open or close, blue paint for the paths we want the electrons going through, and white paint for the boxes electrons will use to go to another floor. And so, we can draw up to 10 floors.
Obviously with a transistor we do not do much, so once we have designed a transistor, we can combine it with 2 or 3 more transistors, to make a relatively simple circuit. After we have done that, we can combine it with 3 or 4 simple circuits to make a not-so-simple circuit. Once we made the not-so-simple circuit, we can combine it with 3 or 4 not-so-simple circuits to create something a little more complicated.
When we have the more complicated circuit, we can connect it to other blocks, circuits which are not so complicated, to get something that begins to be quite complicated. Which can do functions such as adding, calculating, dividing or saving in memory. And once we have that block, which is quite complex, we have more room to connect it to another complex block.
Thus, we make something much more complex that may have other functions and we can then connect it to some other blocks that can do even more complex things. And then we can connect it to more blocks to end up making a giant microchip which has 7. 1 billion transistors and these tens of kilometers of wires.
Which I find amazing. Being faced with that task is really like being a designer who has a vast city in their hands and can work on things from the smallest to the greatest detail. That's what a designer, and even my family thinks when they see these these cute images, they say they are not beautiful, and that I should devote myself to designing clothes with this, which is a frequent discussion at home.
Now, you'll say, all this just to make computers? And, actually, it's not. Computers are one thing that has been the engine of the evolution of this, but due to some advances, having such complex chips enables us to do many other things.
Let's look at several examples. The human ear. The human ear is an exquisite device.
The outer ear receives acoustic pressure waves, which are transformed in the middle ear, through tiny bones, into mechanical vibrations, which in turn transforms them in the inner ear into vibrations of a liquid within the cochlea, which has some very tiny hairs with cells that turn this movement, these vibrations of the liquid into electrical stimulation, which in turn goes via the acoustic nerve to the brain, where the sound image is formed. If any of this is damaged, the patient loses some or all hearing. And here is where the electronics comes in, and we can put a microphone on the outer side, and we can also put a microchip inside the head which reads the signals from the microphone and generates all these channels and injects them directly into the auditory nervous system, to help the person, the brain, interpret the sound.
Recently, researchers made a chip of 3 mm by 3 mm with 1500 photosensitive cells. They receive receive light stimuli. And it was implanted in patients with untreatable forms of hereditary reticular degeneration.
This chip was implanted in the retina, in the fundus. Immediately after the procedure, 3 patients could distinguish and read letters, 5 patients said it improved their daily life significantly. And at the Max Planck Institute of Biochemistry, they discovered a potential way to connect brain cells with electrical transistors.
These researchers put nerve cells on the surface of a microchip, brain cells of a certain type of snail. These cells connected to each other, as cells like to do, with synapses, and spread out. And at certain moment, with one of the transistors, they injected a signal to one of the brain cells.
This brain cell communicated with another one, and that brain cell communicated with the transistor, and they were able read the signal. This system, transistor-brain cell, brain cell-transistor, is a neuro-electronic system which shows that brain cells can be combined with electronics. And could say there were merely four components.
This is a proof of concept, but remember that in 1961, the first commercial microchip also only had 4 components. Imagine what we can achieve, not within 60 years, but in twenty. Now, let me remind you that I transformed you all into 3-nanometer tall people.
And if I left you like this, each of the chairs you're sitting in could fit 3 million times the world's total population. So it would be quite difficult to get out of here, and I do not think the organizers would be too happy, so, I will bring you back to your natural size. Let's get away from the microchip, and return to your normal height, and let me thank you for having accompanied me on this trip inside a microchip.
