Is Moore’s Law really over?

Old 1 Comment on Is Moore’s Law really over? 78

So… Moore’s law is over and done. Well, not quite, as it would seem. Indeed, we are approaching the limit of how small we can make traditional transistors… But this does not necessarily our computational power will not increase over time. Take the new D-Wave quantum processors; phosphorus atoms in a pure silicon crystal structure. This is the fundamental miniaturisation of a transistor, either the nucleus or an election functions as a transistor; usually the outermost singular electron. Let’s use the electron as our example. Electrons can be in different states; spin up or spin down. This is called quantum mechanical spin. Physicists can control the spin of an electron with magnetism, they can artificially make it flip; spin up or spin down; 1 or a 0. Much like traditional transistors they can have an ‘off’ and an ‘on’ state, but much unlike traditional transistors they can also be both at the same time. This is due to a property called ” quantum superposition”. Because of its superpositional properties the electrons we can be: 1, 0 and everywhere in between. This allows computation to simulate real world situations in a much smaller array of transistors; certain algorithms can also be solved insanely fast! Much much faster than they would be solved in traditional computation simply because of the nature of only having 0s and 1s. It would take many traditional transistor to emulate the computational ability of the quantum transistor.

However, this has a catch. The matter of compatibility is a tricky one. Ask a traditional programmer to write a program for a quantum computer and they’d probably think you’re mad. It’s like asking a square to be a cube; not really on the same dimension. Compatibility is not the only drawback. The most fundamental problem you would encounter with a quantum processor like this is cooling. It’s not that quantum processors produce heat that need to be dispersed, more that, inherently, if you’re going to base your calculations off the state of an electron… you need to make sure that the temperature of the atoms is (near) absolute zero. Just the tiniest bit of thermal energy would make the electron unstable, thus making your processor just a lump of phosphorus doped silicon.

To overcome the problem of cooling you need something called a “u-tube”. Without getting into too much detail, a U-Tube is a tube in the shape of a U that has a liquid in it. To cool a quantum processor you need a mixture of helium 3 and helium 4. Helium is a noble gas; this means it isn’t ever attracted to other elements. However, if you put two identical helium atoms together you see a very very small attraction; this is due to Van Der Waal’s force. If you put a helium 3 and a helium 4 atom together you see a slightly greater attraction than if they were identical; this is because of the different densities making the atraction stronger. As you would expect, helium 4 will sink bellow the helium 3 due to helium 4 being more dense. This, for the most, part is true; you will get a layer of helium 3 on top of the helium 4 pool. However, due to the slightly greater force of attraction between the helium 3-4 than the helium 3-3 some helium 3 atoms will spontaneously dilute themselves into the helium 4 pool. This results in helium 3 being absorbed through the helium 3-4 interface and dissolving into the helium 4; the helium 3-4 interface is the point at which the helium 3 and the helium 4 contact. Enough helium 3 is absorbed to have a consistant concentraition throughout the helium 4. The theoretical concentration of helium 3 in the helium 4 at absolute zero would be 6.4%.

To go into more detail… The helium 3 will defuse into the helium 4 and then travel from one side of the U-tube to the other to balance out the concentration. This is called an osmotic pressure. Helium 3 requires less energy than helium 4 to evaporate because helium 3 is less dense than helium 4. If you get the temperatures right, the helium 3 will evaporate out of the helium 4 side of the U-tube. Then, if you have an extractor feeding the evaporated helium 3 from the helium 4 side back into the helium 3 layer it will allow more helium 3 to pass into the helium 4. Because you are pulling helium 3 out of the helium 4 side through evaporation more helium 3 will dissolve into the helium 4 at the helium 3-4 interface due to the osmotic pressure.

Here’s the good bit, helium 3 is being absorbed into helium 4 because of osmosis. As they do so, they are also absorbing heat! This is called dilution refrigeration. When you take atoms from a pure phase and put them in a dilute phase you increase the entropy. The atoms around the interface are absorbing heat so that they can jump from an ordered phase to a disordered phase. So anything to attach to the pipe at the helium 3,4 interface will be cooled to near zero kelvin. The lowest recorded temperature recorded though this method is 1.9 millikelvin; which is extremely extremely cold. This is how they cool a quantum processor. At temperatures near 0 kelvin (absolute zero) the outermost electron of the phosphorus is stable enough to be used as a transistor. Physicists can use microwaves to determine the state of the electron, whether is spin up, spin down or a little in between.

Now, if you’ve followed everything this far you’re probably wondering “How is this at all practical?” Well… that’s it. That is the fundamental problem. First of all you are limited by the bus speeds and secondly, at it’s current state, it’s not very optimised for commercial use. The D-Wave quantum processor units are already huge and that’s because they have to fit in a big U-tube cooling system, a device to read the state of the electrons, the processor and all the equipment and control panels necessary to make the whole think function and stay as stable as possible. At this moment in time I’m not willing to say this will be something we will see in a moble device like a phone. However, I am willing to say this is a technology we will see in desktop computers; one day.

The reason people say Moore’s law is finished because we a reaching how small we can make classical transistors; there is a fundamental limit to how small we can make traditional transistors and we are approaching it rapidly. However, if you look at quantum computation we have only just scratched the surface. There’s a whole new layer of computational ability to be created. The fundamental idea of Moore’s Law is that computational ability would double every year. This is evident in D-Wave’s progress with their quantum processors. These processors are defined by the amount of “qubits” they contain. A qubit is the name given to an electron that is used as a transistor. In 2013 D-Wave released a processor called “Vesuvius” which contained 512 qubis. D-Wave’s processors are consistently doubling in qubits. Moore’s law is not done just yet.

This technology is not something we will see available to the public for a very long time. It’s expensive to produce and the amount of apparatus involved in making it stable is mind boggling. However, the smarter of you will start to see a pattern. Will things repeat themselves? Could be a Déjà vu of how traditional computation came to be? Maybe, just maybe, Moore’s Law extends to more than just traditional computing. Maybe it extends to quantum computing. Personally, I would call this revelation ‘Quick’s Law’ after myself however I believe ‘Moore’s Quantum Law” is a little more fitting.


Tyler McIntosh

You can usually find me either programming, fixing computers or developing an operating system I have been working on for a number of years now. I offer piano and computer science lessons over the internet. I have also acted as a voluntary IT consultant for a mental health charity. I am currently 15 years old however I will turn 16 on the 10th of August. My goals and aspirations are to follow both quantum mechanics and computer science to PhD then work either at the university of New South Wales in Australia developing the phosphorous electron quantum processor or at D-Wave making quantum processors ready for domestic use. I am situated in the United Kingdom.

Related Articles

1 Comment

  1. Nick Best July 23, 2014 at 10:32 am

    This is a very thorough summary of quantum computing with good information I did not know. Historically Moore’s law has fit, more or less, a lot of metrics of advancement even with different technologies. When vacuum tubes were being used for computers a theoretical limit was reached on how many tubes could be put together. This limit was obviously broken by the silicon chip. If historical computational power were graphed to present you could see that Moore’s law smoothly fits vacuum tubes all the way to today’s multi-core processors. If this trend continues that would mean we will still experience exponential growth in computation no matter what technology we use.

Leave a comment

Back to Top