Making a Chip One Atom at a Time
Everyone knows that the transistors in a modern microprocessor are on the small side. So small, in fact, that it’s hard to get a grasp on the concept. Some of the critical film layers in the transistor are only a few atoms thick and well over a million transistors would fit inside the period at the end of this sentence.
So how do we actually make these infinitesimal structures? One technique that is becoming increasingly common is atomic layer deposition, or ALD. The ALD process builds up material directly on the surface of the chip, a fraction of a monolayer at a time, to produce the thinnest, most uniform films possible.
In the video above, David Thompson, Ph.D., who heads up ALD chemistry research at Applied Materials, explains how ALD works and explores the specific challenges of applying ALD to high-volume chipmaking.




Comments
Cool Presentation! However, a Question...
It looks like you're making layers that are covalently bonded. Can you make layers joined by ionic or metallic bonds?
Great question!
It's a great question. Ionic solids, covalent solids, and metals are all accessible by ALD. While the covalent solids and ionic solids generally involve reactions that incorporate atoms from both the precursor (the Hf atom from HfCl4 in the video above) and the reactant (the O atom from water), in pure metal ALD we typically deposit atoms only during the precursor part of the cycle and use the reactant to remove a passivating group.
At the risk of diving too deeply into the technical details it's also worth noting that HfO2 might be considered by many as a polar covalent or even ionic solid. The video shows an initial covalent reaction of a tetrahedral HfCl4 molecule with surface -OH groups to generate a tetrahedral Hf atom at the SiO2 interface. However, using 'ionic accouting' in the bulk HfO2 solid each Hf(4+) 'cation' has 8 O(2-) 'anions' as its nearest neighbors in the bulk solid.
The chemistry of how the hafnium atoms rearrange from tetrahedral (4 nearest neighbors) to octahedral (8 nearest neighbors) during the deposition process is fairly complex, but the general ALD chemistry of alternating self limiting reactions leading to halide and hydroxide terminated surfaces still applies.
Hope that helps,
David
Could this technique lead to
Could this technique lead to molecular manufacturing? Sounds interesting!
While we're not at any real
While we're not at any real risk of realizing K. Eric Dressler's vision in Engines of Creation (http://en.wikipedia.org/wiki/Engines_of_Creation) anytime in the next 10 years we're definitely implementing a flavor of atomic and molecular manufacturing today!
There's actually an analogous technique to ALD called MLD (molecular layer deposition) where instead of growing atom by atom we can graft self-limiting molecules together to generate complex organic structures on substrates.
So whether it's atom by atom or molecule by molecule the chemistry is available to be exploited ... but 'universal assemblers' won't be in the product offering for quite some time.
Hope that helps,
David
Enjoyed your video
I very much enjoyed Dr. Thompson's video about hafium and growing layers in controlled conditions "one atom" at a time. This explanation and demonstration is suitable for everyone from high school students to physicists and chemists .. and the technique is really quite as impressive as it appears to be. Thank you.
Post new comment