MAMEWorld Forums - EmuChat - Re: An interesting prototype has turned up

> > Yes those are the extreme circumstances I was talking about. Regardless no matter
> the
> > specs, there is fixed a voltage between the entry and exit of the semiconductors
> > which will determine the amount of energy it will sustain at any time. That amount
> of
> > energy is way, way inferior to the energy contained in the chemical bonds, so the
> > chemicals should not react. But there is always an epsilon risk of them reacting
> and
> > the bigger the total amount of energy it sustains over time the higher that epsilon
> > is of course.
>
> Voltage across the semiconductor junctions doesn't have much to do with it. The very
> thin oxide layers in modern CMOS chips easily degrade from heat alone. The
> first-generation Motorola 7400 CPUs (in FCPGA packages before they went BGA) are
> well-known for this happening even when run within specified temperature limits. Then
> you lose the insulation between the gate and the channel, killing (part of) your
> chip. The IBM POWER chips had far thicker oxide layers, so they used far more power
> at the same clock speed, but they could run hotter and didn't have problems with
> oxide degradation.
>
> It's similar with Schottky barrier diodes - the chemical bonds within the metal used
> for the anode are not particularly strong, and the metal atoms will readily form
> bonds with the metalloid semiconductors in the cathode. This means atoms from the
> metal anode will slowly diffuse into the cathode and poison it. Higher temperatures
> will speed it up, but it will happen at room temperature eventually.
>
> The fact is, console ROMs have survived very well because of extreme circumstances.
> Sitting in a box at room temperature is not really how semiconductor products are
> designed to be used.

heat alone? but isn't heat equal to energy?

the semiconductor has a certain resistance r to electrons movements depending on its structure, the semiconductor will sustain an electrical intensity I = U/r, and an energy equal to UI*t, U being the voltage between the entry and exit of the semiconductor, t being the time during which the voltage is active, and unless you throw the chip into the lava, the heat is almost entirely the result of the friction of electrons which itself depends directly on the voltage, so the voltage has everything to do with everything

and as long as this voltage is appropriate, there should be no way the chip will receive enough energy to break the chemical bonds and have them reacting which would end in corrupting the chip

I really wish I could say this in a way everyone could understand but I'm not very good at vulgarizing science