Technology Index
Memristors
Nanotubes
Phase Change
Ferroelectric
In regards to breaking the laws of optics, check out silicon chip fabrication. Current CPUs from Intel use
32 nanometer features. Note that they are using Argon Florine lasers which emit light whose
wavelength is 193 nanometers. That is a breaking of the laws of optics, isn't it?
Since air begins to absorb significantly around the 193 nm wavelength, moving to sub-193 nm
wavelengths would, the argument once went, require installing vacuum pump and purge equipment on
the lithography tools which would be a significant challenge. In addition, insulating materials such as
silicon dioxide (SiO2), when exposed to photons with energy greater than the band gap, release free
electrons and holes which you don't want in your semiconductors unless you specifically put them there.
Remember when the Japanese were going to leapfrog ahead of Intel (and the US) by using X-ray
lithography? It didn't happen because of problems with the X-rays and the "tricks" that they do to get
finer resolution with the excimer laser lithography. IBM made more than one of the breakthroughs.
Non-linear photomasks are one of the tricks. At one point they substituted distilled water for for the air
which allowed them to have numerical apertures over 1.3. Then they had to worry about heat "waves" in
the water at one point. They then went to circulating the water. I think that they went to other fluids after
that, but I am not up on the latest tricks. If you want more details, read the book Advanced Processes for
193-nm Immersion Lithography. They have done down to 10 nm on the lab bench and some people are
talking about going down to 2 nm. Don't know how they are going to do that, but I don't put it past them
anymore.
Note that we are still talking about the fact that modern semiconductor facilities use 193 nm lasers to
produce 32 nm lines on your currently manufactured computer chips. They fully plan to keep doing this
down to the 10 nanometer features and the theoretical limit is 2 nanometers, believe it or not.
After that, it is on to the nanotubes and Memristors. I am excited that HP is going to have memristors in
production by 2013. Memristors act in a manner which is similar to neurons.
Please note that we have been producing integrated circuits whose density of function exceeds that of
the human brain since more than a decade ago. Read Eric Chaisson's book Cosmic Evolution for further
information on this. (Eric is a Harvard PhD and knows thermodynamics thoroughly.)
The problem is that we are presently producing commercial chips which are limited to two dimensions
and small chip sizes. Note also that dynamic RAM (DRAM) uses one transistor and one capacitor per bit
which makes it very dense, but also means that it must be constantly refreshed. The smaller the
transistor and capacitor, the more frequently it has to be refreshed. The more you refresh the more
power you burn. With the introduction of some of the new technology, there will be lower power
requirements and integrated circuits will be stacked in the third dimension. This calls out for devices
such as memristors which hold their state without any power.
The other thing that takes up so much power is driving wires with electrical signals. The wires have
capacitance to overcome and the longer the wire, the more capacitance. They are starting to move to
fiber optics. Consider one fiber which is carrying ten or more frequencies and each frequency is being
modulated at 50 Gigahertz. (http://www.intel.com/pressroom/archive/releases/20100727comp_sm.htm)
And no cross talk with the next fiber like there is with two parallel wires. This is working in the labs at
Intel right now. Just needs to go into production next year.
By 2030, we should be capable of producing a device with the dimensions of 14 cc or about one cubic
inch whose information processing capabilities compare with that of the 1400 cc human brain.
don
don
Computer memory types
Volatile
DRAM (e.g., DDR SDRAM)
SRAM
In development
T-RAM
Z-RAM
TTRAM
Historical
Delay line memory
Selectron tube
Williams tube
Non-volatile
ROM
PROM
EPROM
EEPROM
Flash memory
FeRAM
MRAM
PRAM
In development
CBRAM
SONOS
RRAM
Racetrack memory
NRAM
Millipede
Historical
Drum memory
Magnetic core memory
Plated wire memory
Bubble memory
Twistor memory DARPA
Computing Speed Progress due to software improvements
Memristors
Nanotubes
Phase Change
Ferroelectric
In regards to breaking the laws of optics, check out silicon chip fabrication. Current CPUs from Intel use
32 nanometer features. Note that they are using Argon Florine lasers which emit light whose
wavelength is 193 nanometers. That is a breaking of the laws of optics, isn't it?
Since air begins to absorb significantly around the 193 nm wavelength, moving to sub-193 nm
wavelengths would, the argument once went, require installing vacuum pump and purge equipment on
the lithography tools which would be a significant challenge. In addition, insulating materials such as
silicon dioxide (SiO2), when exposed to photons with energy greater than the band gap, release free
electrons and holes which you don't want in your semiconductors unless you specifically put them there.
Remember when the Japanese were going to leapfrog ahead of Intel (and the US) by using X-ray
lithography? It didn't happen because of problems with the X-rays and the "tricks" that they do to get
finer resolution with the excimer laser lithography. IBM made more than one of the breakthroughs.
Non-linear photomasks are one of the tricks. At one point they substituted distilled water for for the air
which allowed them to have numerical apertures over 1.3. Then they had to worry about heat "waves" in
the water at one point. They then went to circulating the water. I think that they went to other fluids after
that, but I am not up on the latest tricks. If you want more details, read the book Advanced Processes for
193-nm Immersion Lithography. They have done down to 10 nm on the lab bench and some people are
talking about going down to 2 nm. Don't know how they are going to do that, but I don't put it past them
anymore.
Note that we are still talking about the fact that modern semiconductor facilities use 193 nm lasers to
produce 32 nm lines on your currently manufactured computer chips. They fully plan to keep doing this
down to the 10 nanometer features and the theoretical limit is 2 nanometers, believe it or not.
After that, it is on to the nanotubes and Memristors. I am excited that HP is going to have memristors in
production by 2013. Memristors act in a manner which is similar to neurons.
Please note that we have been producing integrated circuits whose density of function exceeds that of
the human brain since more than a decade ago. Read Eric Chaisson's book Cosmic Evolution for further
information on this. (Eric is a Harvard PhD and knows thermodynamics thoroughly.)
The problem is that we are presently producing commercial chips which are limited to two dimensions
and small chip sizes. Note also that dynamic RAM (DRAM) uses one transistor and one capacitor per bit
which makes it very dense, but also means that it must be constantly refreshed. The smaller the
transistor and capacitor, the more frequently it has to be refreshed. The more you refresh the more
power you burn. With the introduction of some of the new technology, there will be lower power
requirements and integrated circuits will be stacked in the third dimension. This calls out for devices
such as memristors which hold their state without any power.
The other thing that takes up so much power is driving wires with electrical signals. The wires have
capacitance to overcome and the longer the wire, the more capacitance. They are starting to move to
fiber optics. Consider one fiber which is carrying ten or more frequencies and each frequency is being
modulated at 50 Gigahertz. (http://www.intel.com/pressroom/archive/releases/20100727comp_sm.htm)
And no cross talk with the next fiber like there is with two parallel wires. This is working in the labs at
Intel right now. Just needs to go into production next year.
By 2030, we should be capable of producing a device with the dimensions of 14 cc or about one cubic
inch whose information processing capabilities compare with that of the 1400 cc human brain.
don
don
Computer memory types
Volatile
DRAM (e.g., DDR SDRAM)
SRAM
In development
T-RAM
Z-RAM
TTRAM
Historical
Delay line memory
Selectron tube
Williams tube
Non-volatile
ROM
PROM
EPROM
EEPROM
Flash memory
FeRAM
MRAM
PRAM
In development
CBRAM
SONOS
RRAM
Racetrack memory
NRAM
Millipede
Historical
Drum memory
Magnetic core memory
Plated wire memory
Bubble memory
Twistor memory DARPA
Computing Speed Progress due to software improvements
AskTheBot's Site for Discussion of Robotic Issues
 | |  | |  | |  |