Tag Archives: Extreme UltraViolet

Chip Wars by Chris Miller — Part III — why smaller is better

The following quote from part II says it all — “As Silicon valley crammed more transistors onto silicon chips, building them became steadily harder.    Russia stole the equipment to make them, but they had no way to get spare parts.  The Russian military didn’t trust the chips produced in country, so they minimized the use of electronics and computers in military systems.    The math they put into  their guidance computers was simpler to minimize the strain on the onboard computer.”

The more transistors you put on a chip of a given size, the more computing it can do in a given time, particularly when time is of the essence with missiles, and artillery.  So making transistors smaller and smaller makes them able to do more things and faster as well.

The non-techies can skip the rest, but it’s too fascinating (to me at least) to see how it’s done.   Please note that most of this material is based on Miller’s book.

As of 9/22 “The smallest transistor size that has been used in commercial central processing units (CPUs) or graphics processing units (GPUs) is currently 5 nanometers (nm). Several semiconductor companies, such as Intel, AMD, and TSMC, have released or are in the process of releasing CPUs (central processing units) and GPUs (graphics processing units) with 5nm transistors.”

I’m a chemist and chemists think in Angstroms, because the smallest atom (Hydrogen) has a diameter of 1 Angstrom.  A nanoMeter is 10^-9 meters (a billionth of a meter), and 1 nanoMeter is 10 Angstroms.

The nearest neighbor distance between silicon atoms in crystalline silicon is 2.35 Angstroms (which has the diamond structure of carbon which is a tetrahedral structure — the angle between bonds in a tetrahedron is 109 degrees, so the distance between any two silicon atoms linked by a common silicon atom is 2 times sin 54.5 (.814) times 2.35 or 3.8 Angstroms), so the actual number of atoms along a distance of  500Angstroms  (5 nanoMeters) in a silicon crystal is only 132 !  That’s how small lithography at this distance is chopping up Silicon.  Get much smaller than this and quantum mechanical effects come in to play (if they aren’t there already)

The smallest wavelength of visible light is around 4,000 Angstroms.  Waves will only be reflected by something of the order of their wavelength. Surfers ride waves in to shore, but they don’t change the speed or direction of the waves they ride.  Essentially waves can’t ‘see’ the surfers riding on them.

Similarly to ‘see’ and carve objects as small as 500 Angstroms, you need light of much shorter wavelength — called extreme ultraviolet light (EUV).  Producing such light isn’t easy — here’s how it’s done currently. To be honest the book calls EUV 135 nanoMeters, and doesn’t explain how this could make features nearly 3 times smaller (50 nanoMeters)

Producing  EUV requires pulverizing a small ball of tin with a laser.  A 30 micron ball of tin moving at 200 miles/hour was shot twice with a laser: the first pulse to warm it up, the second to vaporize it into a plasma with a temperature of 500 kiloKelvin.  The process is repeated 50,000 times each second to produce enough EUV to fabricate a chip.  The lasers produced to do this contain 457,329 parts.   Cymer, a company founded by two laser experts in the USA does this.

Focusing EUV to carve patterns on silicon requires extraordinarily precise optics done by Zeiss. The mirrors to reflect the EUV are the smoothest objects ever made.

The EUV lithography machine has hundreds of thousands of components that took 10s of billions of dollars and several decades of research.   The machines cost 100 million dollars each.

Zeiss is in Germany, ASML ,the company that makes the lithography machines is in the Netherlands, Cymer is in the USA, so it is impossible for a single country to duplicate the supply chain for EUV.  The book contains an estimate that it would cost China 1 trillion dollars to do this for computer chip production, and there is no guarantee that politics wouldn’t get in the way (as it already has in China and Russia).

So, as I said, Chip Wars is really about manufacturing (of which I and probably most of the readership were blissfully unaware).

Next up:  Chip Wars by Chris Miller — part IV beating China with silicon will be much harder than beating Russia