If I were to ask you what the most important advancement humankind has ever made. What would you say? Many would make the case for fire, our species was in an evolutionary until we harnessed it’s power. It gave us the warmth to survive harsh winters, it’s light extended our day, it’s destructive powers gave us a weapon against predators and it allowed us to cook our food which allowed our brains to grow. Our discovery of fire transformed not only our technologies, but our culture and way of life.
But, I would argue that we are currently living in a time of equal or even greater change. We are living in an time of incredible growth, one that has already started to transform the way we live our lives. The age of information. With 40% of our world’s population currently connected to the internet, the human race is more intertwined than ever before.
https://ipfs.pics/ipfs/QmaYQN6saXb28KGiiDZXY3f73tNqP8uqdKo2mtHZpNxqtK
But what led to this amazing tool. What single invention gave rise to our smartphone equipped generation? The transistor. The transistor is information itself, even this post is just a series of ones and zeros beaming across the planet to be interpreted by the processors in your computers. Without the transistor, I wouldn’t have access to the wealth of information online to do my research, and I most certainly wouldn’t have been able to share it here for the world to see.
The transistor is so simple, but it is the foundation for all our modern computers. To understand it’s impact we need to understand the history and science behind it. Before the transistor existed, we used vacuum tubes. Which were these bulky evacuated glass bulbs. The triode vacuum tube consisted of three parts, the cathode, grid and anode.
A current is passed through the cathode and it begins to heat up, causing it to release electrons, as the gases have been removed from the tube the electrons have very little resistance to their movement and they are attracted to the positively charged anode. This completes the circuit and a current flows.
But we can manipulate this flow of electrons in many useful ways with the grid. For example we can use it as a switch, if we place a light bulb it will only light up when there is a positive voltage across the grid.
If we apply a negative voltage the negative charge will repel electrons trying to pass through.
This is the foundation for binary coding, which is the 1s and 0s that gave birth to the age of information. Here 1 is a positive voltage and 0 is a negative voltage. 1 turns the light on, 0 turns it Off.
The world’s first general purpose electronic computer, The ENIAC, used 18 thousand vacuum tubes to perform calculations. Designed by John Mauchly & J. Presper Ekert, it was completed in 1945. It was purpose built to calculate trajectories for artillery during world war 2. A calculation that would take a human a day to calculate, took ENIAC 30 minutes. But this thing weighed 30 tonnes and took up an entire room. It was incredibly power hungry, as the vacuum tubes cathodes needed to be heated to work, which also meant that the vacuum tubes burnt out regularly and needed to be replaced.
https://ipfs.pics/ipfs/QmUusjzQLcnYUfdhJWnzSFovWLuhX5wBhbTnb24svyDc3H
https://ipfs.pics/ipfs/QmYpV6hCUgX8SmfsgdSTCnmYyuwfAqYp7PcoLFFEFo6TTH
All this to perform a function that your phone basically does within Angry Birds.
Today it’s computing power could be contained on a silicon chip no larger than a grain of sand and that is thanks to the transistor. A modern phone has around 2 billion transistors, which perform the exact same job as the vacuum tube, but on a nano-scale. Let’s look at how it works.
Many of you will recognize the transistor as one of in the image,
https://ipfs.pics/ipfs/QmPicjbvNaiea7FeUXnBvQNnQ8qk7RC7U5rD7E69hfTZuE
but this is a through hole transistor that you can buy from hobby electronics stores for your DIY projects. The transistors in your CPU are microscopic and are manufactured
https://ipfs.pics/ipfs/QmWxNVB5uFWsCjHWgc27DgodPS6LiX6k1zwwkUWeW5ifW2
with incredible precision with machines on thin wafers of silicon crystal that are sliced off silicon ingots, like in the image.
https://ipfs.pics/ipfs/QmcSKco24mWKfnyLxfMx7BP4Hca2JJ8fXf538KCGt2DXEH
So what makes silicon so special that an entire section of the San Francisco Bay area has been nicknamed after the material. Silicon is a semiconductor, which means that its conducting properties can be tailored by introducing impurities to the crystal structure. Silicon has 4 electrons in it’s valence shell
https://ipfs.pics/ipfs/QmUeXhCys3zsMqw7WchkwoNcLdcSC3ffYYox5Mpu4A8Rhx
this is the outermost orbit for electrons and it determines many of the chemical properties of the atom. Atoms want 8 electrons in that shell, as this makes them very stable. So silicon readily forms covalent bonds to 4 neighbouring silicon atoms to gain those extra Electrons.
https://ipfs.pics/ipfs/QmdgMH3Hyp5wcqYcn1rtQVsPBYLZxY6y1xMKpN35fQQLsD
Now if we introduce those impurities to this pure silicon crystal, we can change how it conducts a current. If we add phosphorus, which has 5 electrons in it’s valence shell, the extra electron is left free to roam the crystal structure. This extra electron makes the N-type negatively charged, which where the name comes from. The P-type is positively charge because it is doped with Boron, which has three electrons in it’s valence shell.
The structure wants to gain it’s final electron and will steal electrons from its neighbouring atoms, this creates a mobile positive charge, called a hole. The conductivity of material has thus been increased as we have increased the number of mobile charges.
When we arrange n-type and p-type semiconductors
https://ipfs.pics/ipfs/QmcxYVC5ZbHEuFad4jpmuwu5QtPZTJ4KNukyBwCyvYibeR
like in the image and attach terminals to each, we create the world’s most prevalent transistor. The NPN transistor. The transistor works due to the interaction of those free electrons and holes at the n-type and p-type junction. Free electrons in the n-type will migrate over to fill those holes in the p-type. This creates a boundary layer called the depletion layer which prevents more electrons passing through, due to the negative charges repelling each other.
https://ipfs.pics/ipfs/QmUNm1WHKimmaELkfSXUjhmyk29h5noFr3sV6881uiFRAE
But, when a positive voltage is applied to the base it negates that depletion layer and allows current to flow through, completing the circuit. As you can see this is very similar to the function of the vacuum Tube.
So how exactly does this allow computers to perform all these complex functions that we see today.
Let’s look at a very basic example. Let’s add two numbers together. First we need to learn how numbers are represented in binary, that’s the 1’s and 0’s that are used to store data.
https://ipfs.pics/ipfs/QmZDXCiDpyDNLWZ414ssCxSdonS5nA128mLC2vhC7PqyVR
This is the number 15, which is the largest number you can represent with 4 bits.. The first bit represents 1, the next represents 2, then 4 and finally 8, added up that equals 15.
https://ipfs.pics/ipfs/QmbXv5CPDm8Vknv2buSoJRiET8hK6G3DpqRvnNmKz4e3jN
This pattern continues with each successive bit representing double the previous, so if we add an additional bit we can count up to 31. Let’s add 5 and 6 together. To do this we want a circuit that will hold a 1 in this position when either are 1 and carry the 1 forward when both are 1, as you can see this will give us the number 11.
https://ipfs.pics/ipfs/QmRfteBnTtrrdgb5QgEDYhdgTC915DZermmjbAp3LPj71V
The simplest circuit that can do this is a half adder. Which contains two types of logic gates, these are devices that can modify the binary code, they are built using transistors. The first is the XOR logic gate, which gives a 1 only when one of the inputs is one, if both are 0 or 1 it gives a 0.
https://ipfs.pics/ipfs/QmezNfZxy1p4xNa9pUMc7oxTBSZqpaPVyv6GQDTbV2vg6m
The second Logic gate is an and gate, which gives a 0 for everything except when both inputs are 1. If we wire these logic gates we create a half adder, which gives two outputs our Sum and our Carry. Which allows us to add our binary number one bit at a time.
https://ipfs.pics/ipfs/QmcNjcxGuoWjS6Ajty5YZj4r7D69wA98hptkFYvQnih2fn
A more complicated circuit is needed to perform the calculation in one step. Modern computers can perform millions of these calculations per second and they are still getting faster. The Co-founder of Intel Gordon E. Moore noticed a trend in 1965 that the density of transistors on integrated circuits doubles every two years, that trend has held until very recently, but it is starting to slow down.
https://ipfs.pics/ipfs/QmdE11gTNV7A9ebNKCsuiQTeek9rbFbBwurMV9AYtXe5eQ
One of the reasons for this is the less well known of Moore’s predictions, Moore’s second law or Rocks law, which states that the cost of manufacturing these devices will double every 4 years. Intel made an announced last year, that the rate of advancement was slowing for these reasons, it’s getting more and more difficult for chip manufacturers to shrink their product while maintaining Profit.
Another problem that transistors are facing is quantum tunneling. As these transistors get smaller, so do the barriers between different sections.
https://ipfs.pics/ipfs/QmT96BAqx2ZdXB2wY2TUSEhFkq8e4HAyXDnPxNWCumTVFc
The barriers between each section of the transistor are getting so thin, that electrons can pass right through them.
https://ipfs.pics/ipfs/QmbGNbTFqA42fAjafDUqwKfTdobpGTdwnQu9X4yqwppddL
With no definitive successor to the silicon transistor lined up this incredible period of growth over the last 50 years could plateau in the near future.
Some want to harness quantum mechanics to perform calculations faster than any transistor ever could, others want to decentralize computing power with the so called internet of things. Intel have said themselves that they plan to shift their focus from increases in speed to decreases in power consumption. One thing is for sure, the computer industry will have to redefine itself in the near future.
Great article. Very thorough. Parallel processors, such as GPUs have begun to interest me lately. They aren't suited for all applications, but for those which they are well suited, parallel processing shows a lot of promise.