Braunschweig (Germany) - Electron mobility researchers at the Physikalisch-Technische Bundesanstalt (PTB) have developed a high vacuum semiconductor crystal generation process which produces 5x more purity than traditional epitaxy systems. The new silicon promises a significant furthering of research into the manufacturing of semiconductor-based single-electron pumps as well as an investigation into quantum Hall resistance metrology, which could ultimately lead to a fundamental redefinition of the electrical current as a function of frequency and electron charge.

The crystal generation process involves an ultrahigh vacuum that is 15 orders of magnitude lower than standard atmospheric pressure, or about 0.00000000000003 psi. To accomplish this, the epitaxy facility employs enormous multi-stage vacuum pumps: Only after a partial vacuum has been achieved using lesser pumps, the final stage can be engaged to reach the ultrahigh vacuum levels.


Big vacuum is key

It is the ultrahigh vacuum which is key to this process's success, but it is also extremely difficult to achieve in a production environment as the law of diminishing returns comes into play.  As more and more air is removed,  there is less air left to draw out with each stroke. The final stage of this process is so powerful that it could not even operate (due to the air's drag resistance) if a significant partial vacuum weren't already achieved before it was switched on.


The process

Once the big vacuum is achieved, the process inside is relatively straight-forward. It's called "molecular-beam epitaxy" and basically allows the formation of extremely pure individual atomic layers of crystal structure in a highly controlled environment. The controlled layering ultimately produces a type of "semiconductor-based racetrack" allowing electrons to move at extremely high speeds due to the lack of impurities which would otherwise slow them down.

A heterostructure of GaAs (gallium-arsenide) or AlGaAs (aluminum-gallium-arsenide) are used as the base. Gallium and arsenic are then evaporated within the vacuum and deposited one atomic layer at a time. Specially created cooling panels are also used to "absorb" impurities before they can be introduced and incorporated into the crystal.

The final product is an extremely pure semiconductor, which can be used to push high-end research for future production-based manufacturing processes and technologies which will ultimately enable Moore's Law to continue almost indefinitely.

At the most recent IDF, Intel commented that right now it is known Moore's Law will continue through at least all of the next decade (up to 2020, and probably beyond). Research like this at PTB will continue to push it far beyond.


Engineering hurdles in a vacuum

There are enormous engineering hurdles which must be overcome with a high vacuum. Materials that are completely stable at atmospheric pressure, or even at low vacuum levels, begin to out-gas as the pressure drops. The out-gassing effect has probably been seen by most people in high-school chemistry experiments whereby a beaker of water is placed in a vacuum and then, at room temperature, it begins to boil as a small vacuum is created.

It boils because the pressure is so low that it can out-gas without resistance.

Big problems arise for machine designers, however, as the same out-gassing effect holds true for other more common products used in electrical components, such as rubber and plastics. While these components are needed to control stepper motors and servo mechanisms to move things around within the chamber, buying off-the-shelf hardware doesn't typically work at high vacuum.

Devices must often be disassembled and recreated manually using materials that will not out-gas. Sometimes canisters are employed which maintain an internal air pressure upon the hardware so it can do its job with only non out-gassing wires going in and coming out. In the alternative, complete redesigns are sometimes needed which do not use any rubber or plastic insulation at all on crucial parts. And all of this becomes expensive very fast and it truly is quite a feat of achievement to accomplish what the PTB team has done here in such an ultra-high vacuum.


Sidenote

In 1998, I had the opportunity to work with engineers at General Electric’s Gas Turbine division developing a computer controlled part positioning system (a type of robot), which operated within a similar high vacuum chamber in front of a stationary electron beam welder. Our chamber was basically a 6 foot cube. The walls of the cube were 1” thick steel with several large reinforcing plates welded to the sides. Our vacuum chamber was pulled down to roughly 8 orders of magnitude below atmospheric pressure (compared to PTB's 15 orders of magnitude).

The GE team used an old inline 6-cylinder engine (no kidding) which had been converted into a vacuum pump. That vacuum pump was part of a multi-stage sequence we used and could only be switched on once the first stage pumps had pulled down a sufficient level. A gauge on the outside of the chamber indicated the vacuum level. When the big pumps were turned on, the vacuum level dropped down powerfully at first, and then quite slower after just a few seconds. All told, it took close to 20 minutes to create the full vacuum.  Once created, it took about 3 minutes to weld each part.