John, without going into the physics, I could simply say it is a fact that isotopically pure single crystal have better thermal conductivity than non isotopically pure one. The physics is really very simple, the main factor determining thermal conductivity is the "electron/phonon" scattering (it also affects very strongly the electrical conductivity, at least in metals and thus good electrical metallic conductors are also good thermal conductors). In an isotopically pure crystal, there are fewer vibrational modes (or the phonon "spectral distribution" is narrower, so to speak), because the masses on all the lattice sites are exactly the same and create a relatively perfect "harmony". When there are fewer of these modes, there are fewer scattering events between electron and photons and thus better thermal conductivity.
Now, you are sorry you asked, right? (VBG). The second part of your statement, is that the impact of dopping (and other impurities) might be larger than those of the native "perfection" of the host crystal, and you are right, but since in both cases one would have the same amount of these impurities, there will still be a residual improvement of the isotopically pure silicon over the non pure.
The major question is , are these advantages practical? I doubt that you can get thermal conductivity (at room temperature, at very low temperatures the story is very different) difference of much better than about 10% to 20% in the dopped epi (relative to about 60% improvement in non dopped materials).
Finally, I think that if there is value in isotopically pure epi layers, these are more related to increased charge carrier mobility than in thermal conductivity. But remembering the lessons of GaAs (the semiconductor material of the future for the last 20 years or so and still the material of the future), you must have twice the performance at half the cost if you are going to break through "fortress silicon" with a new material in any big way.
Zeev |
