Ethical Implications of Overclocking

My PC recently crashed at an inopportune moment. In fact the perfection of the inopportunity led me to question whether it did this on purpose.

The subject of semiconductor ethics led me to only one name:  At first reading (or listening) you may find Deepak Chopra’s theories confusing.  This is the core of his theory.  As many of us do not have the time for an in depth study of his some four dozen books and various audio and video materials, I will try as best I can to summarize his argument.

Quantum physics is far too complicated for him to understand therefore no one can. Hence there must be a higher intelligence who does. The important consequence of this for our discussion is that this higher intelligence may work within Heisenberg uncertainty and quantum probability to affect the outcome of events. Importantly, Chopra argues that our minds are capable of doing the same thing, giving us free will. We can extend this to its natural conclusion and say anything that operates in a capacity which can be affected by this principle has the capacity for free will.

Thus we now have a criterion from which we may begin to investigate my question. If the crash was due to an effect governed by quantum probabilistic theory, then the computer had the capacity to change the outcome. It did not, ergo it lost my data intentionally. From here it is a simple matter of examining the operation of the semiconductors in my computer to see if this could have been the cause.

As we all know, most of the information handling work done by our computers happens on silicon based chips. The reason for this is that silicon has some interesting electrical properties. To understand them we are going to need to back up to about 11th grade physics and draw an analogy with celestial mechanics. The strength of the earth’s gravity pulls on objects with a force proportional to the inverse square of the distance and some constant. For electrical charges the constant is different but the math is the same.

To understand what is going on inside a silicon atom you will need the concept of potential energy. If you are standing on the earth and you throw a ball up, it will slow down, stop and then speed up. What is happening here is the ball is trading kinetic energy for gravitational potential energy. The further from the earth’s core the ball gets the higher its potential energy. The same is true of electrons around a nucleus.

OK – so now we all know an electron is attracted to the nucleus with a force proportional to the inverse square of its distance. The closer it is the stronger the force. For reasons I won’t go into detail on, the electron can only get so close. When an electron is at this minimum distance we will say it is in its base state. If we want to move the electron further away, the force between the proton and electron will resist it. To move it a finite distance we need to do ‘work’ to get it there. In so doing we add potential energy to the electron. Again drawing an analogy to gravity, this would be like carrying a water balloon to the top of a building. This energy can be released by dropping said balloon off the side. As the balloon is released is will trade potential energy for kinetic energy (it accelerates).

Because the force between the electron and proton is proportional to the inverse square of the distance, the force decreases rapidly with distance. This means that if we calculate the amount of work it takes to drag the electron an infinite distance from its rest state, we will find it takes a finite amount of energy. This amount of energy is the electron’s ‘ionization energy’. If we give the electron more than this amount of energy, then it becomes free to move anywhere.

As a thought experiment, let’s take a hydrogen atom, one proton and one electron. Now let’s place another proton nearby. Now we have a situation where we can drag the electron toward the second proton. When we do this the force we are working against is the force of the original proton less the force of the second, as it is also dragging the electron toward it. Now we only need to do enough work to get the electron from its rest state to the point where the two forces balance and then let the second proton take over and finish the job of dragging the proton the rest of the way. In this case what we have done is change the ionization energy. We don’t need to drag it to infinity, just to the midpoint between the protons.

Supposing we did this – something else interesting would happen. Once we gave the electron enough energy to move into orbit around the second electron, it wouldn’t loose any total energy after we let go (though it would trade some potential energy for kinetic energy). This means the electron is free to move back and forth between the protons (though not out to infinity). When this happens we say that the energy of the electron is above the potential barrier between the nuclei. If we were to now assemble a line of protons, each the same distance form the last (ie the same potential barrier between nuclei), we could move all the way along the line without any additional work. We can extend this analogy to three dimensions but that would be needlessly complicated for this discussion.

Hydrogen is simple to understand but it does not make good computers. If we want to understand other materials I will need to introduce the Pauli exclusion principle. This states (for our purposes) that no two electrons can have be in the same state. Put more simply, no two electrons can have the same amount of energy inside given potential well. Even more simply, no two electrons can orbit a proton at exactly the same distance. In fact they cant even orbit at close to the same distance. An electron bound to a nucleus can only exist at discreet distances and energy levels. This is not entirely true. Quantum mechanics is much more complicated and subtle than this. But this model will suffice for us here.

The important part to take away from this is that each additional electron you want to add to an atom must be further from the core than the previous one. Per Quantum mechanics, for reasons needlessly complicated to describe, the electron’s energy can only take finite values. This is where things start to get interesting. In solids atoms are naturally close together. These materials have a naturally low potential barrier between nuclei. In conductors the natural energy level of the outermost electrons is at or above the natural potential barrier. This lets these electrons move around freely inside the material. The outermost electrons in an insulating material are well below this barrier.

Pure silicon is an interesting case. When it is in a crystalline lattice the potential barrier falls between the highest energy level normally occupied and the the next level an electron would occupy. It falls firmly on the insulator side of this continuum but it does not take much energy to to kick an electron up to an ionized state. Once this electron is freed, you can then move it around using electric and magnetic fields. If you can do this then you can drive a current through what would normally be an insulator. The more free electrons you have the more current you can drive with a given electric field. Or as electrical engineers describe it, the more free electrons you have the lower the electrical resistance is.

In fact this energy can be supplied by the thermal energy of the material. And here is where quantum theory really starts to play a role. Average thermal energy is a continuous function but the energy of a given electron must be one level or the other without any intermediate step. Temperature tells you the probability that a given electron will be at one level or the other. I would like to avoid as many of the details of this equation as possible. What is important is that the probability is an exponential function that increases proportional to temperature and inversely proportional to the ionization energy of the electron. For conductors you can expect free electrons at all temperatures. Insulators on the other hand will melt before a significant number of electrons are freed.

Now consider what happens if there is an impurity added in place of a silicon atom. Before any device fabrication begins, the chips inside your PC start as high purity mono-crystalline silicon. This means it is a single crystal which has something on the order of 10 defects per cubic centimeter. Only a fraction of these will be from impurities – the rest will be deviations from the perfect crystal lattice. To put this in perspective, there are 10 000 000 000 000 000 000 000 silicon atoms in that same space. For any sensible purpose then we can say that the crystal is perfect. Suppose we were to replace a single atom of silicon with a phosphorous atom. Phosphorous as you may recall from high school chemistry is just like silicon with one extra proton.

With this proton comes one extra electron. We know that if you want to add an extra electron to a silicon lattice it must immediately become ionized. This is almost what happens. The extra proton keeps the electron held in place but only just. It takes far less energy to ionize this electron than any of the silicon electrons.

If you were to intentionally add say 10 000 000 000 000 000 phosphorous atoms to a cubic centimeter of silicon something very interesting would happen. Nearly all of these atoms would ionize at a very low temperature as compared to the silicon. Because the relation between temperature and the probability of ionization is exponential, you end up with a range of temperatures where nearly all these impurities are ionized but almost no silicon atoms have been ionized. This phenomenon allows engineers to ‘tune’ the electrical properties of the material. From this it is possible to create the diodes, transistors, and sundry components that the logic systems on your computer are made from.

Lucky for us this temperature range covers room temperature. An interesting side note for LN2 overclockers is that below 100k there begins to be a sizable departure from complete ionization of the impurities. This is called freeze out, there is simply not enough thermal energy in the material to ionize all the impurities at once. This is what causes the so called ‘cold bug’ components simply will not conduct electricity effectively.

Much more common, certainly to anyone reading this, is what happens at the other end of the temperature spectrum. If the silicon gets too hot then the number of ionized silicon atoms begins to become comparable to the number of ionized impurity atoms. At any temperature the electrical properties of the material will change exponentially with an increase in heat. At low temperatures this works in your favor keeping the electrical properties linear. Once the concentration of excess electrons becomes non-negligible, the electrical properties will change with temperature in a highly nonlinear fashion. In more simple terms, at high temperatures semiconductors become conductors. Transistors cease to be transistors and become wires. This caused my computer to crash.

This brings me back to my original problem. Based on the knowledge that free will is a result of the existence of quantum probabilistic effects, and that free will implies responsibility, I can conclude that my computer was responsible for crashing. Thermal ionization of electrons is a probabilistic effect. According to Deepak Chopra, the existence of the indeterminacy of any given electron becoming ionized allows the computer a means of acting to effect the outcome of all consequent actions. Thus my PC is personally responsible for its failure to act to stop this crash. Thus the BSOD I received after the crash was a personal insult.

Now some of you may argue that I brought this on myself from ‘cooling’ the CPU with the Buffalo heatpipe CPU cooler (I reviewed the box here several months ago HERE.). Had I chosen a better heatsink I could have avoided a thermal crisis. This is true but it would have meant denying my PC free will. While this would have avoided the problem it would not have solved it. I may have given it the power of self determination but I was not the one who caused the crash. Thus my PC is solely responsible for its actions. If anything it is Evercool’s fault for making a cooler that gave my PC enough free will to rebel. Even in this case it was still the CPU itself that made the decision to crash.

Some of you may also object to my characterization of quantum physics. I never intended to give a complete picture of the theory. If you want a better understanding of what is going on you should start by Reading QED by Richad Feynman (ISBN: 0691125756 ) and then move on to a basic university text (Griffiths’ is highly regarded ISBN: 0131118928 ). I gave the simplest explanation of the smallest subsection of quantum physics I possibly could write still conveying enough information to understand the issue.

I would even go so far as to say that everything I described here is oversimplified in one respect or another. I did this so as not to obfuscate the underlying concepts with unimportant details. For example, the intrinsic ionization energy of an atom does not actually change when you put it next to another. Only the effective ionization energy is changed and then only in relation to the second nucleus. This detail is true but it does not add anything except confusion to the discussion.

Some, including many from the previous group, would argue that my PC would have needed to allow more than just one electron to jump to a higher energy level – it would need to allow several quadrillion to ionize. They argue that this is a problem because quantum indeterminacy only predicts probabilistic behavior for individual wavicles (a ‘wavicle’ is an object that acts both as a wave and particle). And that the laws of quantum mechanics make hard and fast predictions for large numbers of electrons. Thus even if it could effect a single quantum event, the CPU could not possibly have prevented the crash. To this I respond simply that the PC could have acted on whichever the threshold electron was. In other words it really only needed to prevent one electron from ionizing. This one was whichever was the proverbial straw that broke the camel’s back.

To this one may respond that electrons are indistinguishable. You could replace any individual with any other and have absolutely no way to tell the difference. Thus there was no single threshold electron. All electrons, not just the excess electrons, were collectively responsible for the crash. Unless the PC could control all electrons it had no capacity to prevent the event.

Further, if it could control all the electrons then it would be controlling the very things that give it the control in the first place. At this point I should remind this group of people that an appeal to authority renders all contradictory empirical evidence null and void. Deepak Chopra’s theories clearly state that the CPU qua mind could have acted on each electron individually regardless of the number, collective improbability, or inherent paradox. Thus my PC is responsible for the crash.

Ian Anderson