New memory/storage technology could be a massive game changer.

[autorouter]

I think the significance of the link in the OP is that they have cracked the problem of making III-V structures on Silicon substrates.

Silicon is not the ideal material for electronic devices, but it has a number of properties that make it very easy to print patterns on, and make devices on a massive scale. Other substances are better, including compounds made from Group 3 and Group 5 of the periodic table, called III-V compounds.

GaAs for example can make very fast devices, which are used where money is not a problem in military radar signal processing etc. III-Vs are common in discrete devices like LEDs, but are difficult to integrate into large circuits like silicon chips.

The III-V layers are built up on top of the silicon wafer base (substrate) by 'molecular beam epitaxy' (MBE), which basically means building up a crystal one atomic layer at a time by beams of vapourised atoms in a vacuum. MBE is mainstream technology in silicon fabrication plants, what's new is the different buffer layer between the silicon and the III-V devices, which allows these devices to be built on a bigger scale.

 

[GMLS]

Hope they will do a version that I can implant

 

[Gromett]

Jeez you went deep there. It has been 30(ish) years since I studied this stuff.

I think the significance of the link in the OP is that they have cracked the problem of making III-V structures on Silicon substrates.

Erm, we have been making III-V structures on Silicon substrates for a long, long time from memory using Boron (III) and Phosphorous (IV).
The 4 valency electrons of Silicon is what makes this whole thing possible.

Silicon is not the ideal material for electronic devices, but it has a number of properties that make it very easy to print patterns on, and make devices on a massive scale. Other substances are better, including compounds made from Group 3 and Group 5 of the periodic table, called III-V compounds.

Silicon is the ideal substance for semiconductors or as near to it as possible. Group 3 elements such as Boron have been used for years, Group 5 such as Phosphorous and Arsensic also.

GaAs for example can make very fast devices, which are used where money is not a problem in military radar signal processing etc. III-Vs are common in discrete devices like LEDs, but are difficult to integrate into large circuits like silicon chips.

I remember looking at GaAs with envy when I was a radio amateur. Not looked at them since to be honest.

The III-V layers are built up on top of the silicon wafer base (substrate) by 'molecular beam epitaxy' (MBE), which basically means building up a crystal one atomic layer at a time by beams of vapourised atoms in a vacuum. MBE is mainstream technology in silicon fabrication plants, what's new is the different buffer layer between the silicon and the III-V devices, which allows these devices to be built on a bigger scale.

From memory Molecular beam epitaxy is usually used for non silicon substrates. For silicon based semiconductors they tend to use one of a couple of deposition methods alongside lithography and resist layers.

This all pretty much from memory so forgive me if my memory has led me astray.

 

[autorouter]

Erm, we have been making III-V structures on Silicon substrates for a long, long time from memory using Boron (III) and Phosphorous (IV).
The 4 valency electrons of Silicon is what makes this whole thing possible.

I think you're remembering the III and V elements that are put into silicon as impurities (less than 1 part per million) to make the p-n junctions inside the silicon wafer. This new development is concerned with III-V compounds like GaAs, InSb etc, with no silicon in their structure at all. They are built on top of the silicon wafer, with a buffer layer.

Silicon is a fair-to-middling semiconductor, with a mediocre band gap and poor carrier mobility. It's the remarkable properties like the oxide layer that makes it so brilliant for large scale integrated circuits. The ability to build GaAs devices directly on top of the silicon and connect with other silicon is what will make this a key technology.

 

[Gromett]

I think you're remembering the III and V elements that are put into silicon as impurities (less than 1 part per million) to make the p-n junctions inside the silicon wafer. This new development is concerned with III-V compounds like GaAs, InSb etc, with no silicon in their structure at all. They are built on top of the silicon wafer, with a buffer layer.

The silicon impurities as you call is the doping process which is what I was referring to when I was talking about deposition methods. You can dope with any of the group 3 or 5 compounds including GaAs and InSb. Using these traditional deposition methods you can build up 3D structures without having to resort to 'molecular beam epitaxy'. Although now I re-read the article I can see they are using that method (thanks for pointing that out). That actually makes it less exciting to me sadly as it is a slower process and complicates things quite significantly .

Silicon is a fair-to-middling semiconductor, with a mediocre band gap and poor carrier mobility. It's the remarkable properties like the oxide layer that makes it so brilliant for large scale integrated circuits. The ability to build GaAs devices directly on top of the silicon and connect with other silicon is what will make this a key technology.

I am confused by this a little bit. The band gap and carrier mobility are what I thought made it an excellent semi conductor. These two factors make it a lousy conductor and a not so perfect insulator, but an excellent substrate for semi conductors. Higher carrier mobility would result in higher leakage currents from memory. You want a band gap that is small enough to be able to be forced into conduction (sorry terminology may be incorrect) but not so small that it will conduct freely. The fact that silicon can be doped in such a way to create the p-n junction is what makes it all work?
A large band gap would result in it being a complete insulator as the energy required to move electrons from the valence band to the conduction band would be too high. Too small a band gap and you end up with a conductor where the electrons can pop free of the valence band into the conduction band at the slightest provocation.

Although if you are saying it is a fair to middling semiconductor in comparison to others based on those parameters you may be correct, I honestly don't know enough about the other semiconductors. The only one I ever played with other than silicon was Germanium which had a higher reverse leakage current from memory. But the properties of a material are not the only thing to consider in what makes something good or not. The ease of manufacture not only of the substrate but the ease of doping or building the structures, in this case I do believe silicon has massive advantages.

I am getting a little out of my depth here from memory, and I don't have time to refresh my memory of what I learned all those years go. But thanks for making me re-read it again and pick up details I had missed first couple of times through. Also thanks for refreshing my memory on the stuff I learned all those years ago.

^{_{Subscribers  do not see these advertisements}}

 

[Gromett]

autorouter If my memory has failed me or I am misunderstanding something, please don't hesitate to educate me please. I love learning new stuff or even being reminded of stuff I had long forgotten

 

[autorouter]

The fact that silicon can be doped in such a way to create the p-n junction is what makes it all work?

If you look at the diagram in the link, you will see that there is a silicon layer at the base of the whole structure, but above that there is no silicon at all. All the structures above are only III-V compounds, like GaSb, InAs and AlSb. Fabricated by MBE.

It is notoriously difficult to fabricate III-Vs onto silicon, with many problems including thermal expansion for starters. This team has engineered a method including a buffer layer that allows possible large scale integration of III-Vs onto silicon, and if it functions as described it could be a game-changer.