I remember there is an old Czech SciFi movie about an utopian society where men are obselete and women procastrinate by cloning. Anyone remember then name?
Ubuntu 7.1 finally made me realize that it is actually possible to use Linux as a Desktop OS. For at least 10 years Linux had been the ugly OS I only accessed via SSH on my server.
Nope, we are talking about thin film transistors. Main problem is getting the semiconductor off current under control (purity), improve mobility (crystallinity, purity, doping) and getting a good gate insulator that is not attacked by the deposition process. With certain materials it is literally possible to build transistors in your kitchen. (eg. CdS)
Look at the early work in thin films transistors. Schockleys attempts at building FETs suffered from poor silicon deposition. It took until the 70ies until people found out how to work with a-Si:H.
No quantum mechanic is required to describe field effect transistors in accumulation mode. And that is exactly what Lilienfeld proposed. The only theory that is required is that of space charge limited current, a field Lilienfeld has several publications in.
Lilienfeld did in fact invent the working principle of the transistor. Whether he built one is not known. However he did all the groundlaying work on electrolytic capacitors as they are still used today. Therefore he knew very well how to create extremely thin insulating Al2O3 film that were a necessity for the type of transitor he described in his patents. It does therefore not appear entirely unlikely that he built some of the devices.
The stuff about surface states is mainly important for bulk silicon transistors. Even without solving this problem it is possible to demonstrate amplification as numerous publications on II-VI thin film transistors show.
The field effect transistor, the device that is relevant today, was invented and patented in 1926 by Julius Edgar Lilienfeld. Due to his patents many claims by Bell Labs were thrown out.
The device that was invented by Bell Labs in 1947 was a point contact transistor. An inherently fragile device not fit for mass production. The same device was invented in parallel in France by two german Scientists: Welker and Matere see here.
Schockley himself did however invent the bipolar junction transistor a couple of years later. This invention was truly a streak of genius as it is the most complex of all devices.
So, thanks to american corporate giants history was rewritten again.
Ok, the article is talking about science fiction solutions that have been demonstrated for single bits at universities, but nobody has any idea how to mass produce it. Meanwhile flash memory in production is approaching feature sizes of 30 nanometers with 2 or even 4 bits stored per cell. Also stacking of several memory layers on the same die has been demonstrated.
Scaling is not driven by lithography anymore. It is driven by material science advances. Intel introduced Hafnium based gate dielectrics in 45nm, which is an incredibly impressive feat given that they are 1-2 years beyond all other companies.
Scaling by lithography was in the 80ies and early 90ies.
The rediculous hyping of research results from universities makes me barf. They make the industry appear modest and honest in comparison.
How is that a breakthrough? 60nm line patterning is nothing. Flash memory that is currently in production uses a smaller pitch - and that is not even just lines, but also more complex structures. Also it is done with orders of magnitudate less defects.
This "breakthrough" may have some nieche applications, but it is still far away from a "revolution".
The problem is that you do not have the slightest clue about semiconductor technology. You should not comment on something you dont understand.
Lets start here: How about scaling in nature? Why are small animals like insects "built" with an exosceleton while large mammals have bones? Is it possible that the concepts do not scale arbitrarily? Very similar in the semiconductor world. To make things smaller you actually have to change how they are made, both in terms of processing and materials.
CZO (Actually Zirconium-Yttrium-Oxide) is only similar in appearance to diamond. In all other respects it is completely different.
The most important difference is that diamond does burn in oxygen while CZO is an excellent oxygen conductor (yes, a crystal that conducts ions by a hole transport mechanism). This is also the effect that is used in fuel cells. There is really no relation to diamonds. Pure popular BS-science.
Actually, logic, analog, and memory are all CMOS (or CMOS-ish in the case of Flash) and can all be made on the same silicon production lines
Can be made on the same production line, yes, that is at least partially true. A big difference is however that the process flows will differ. Each of the different circuit types use different transistors (they are all mosfets but with variations in geometry, gox, implants..). Combining all these different transistor types is simple not economical since process complexity will increase significantly.
There are ways to integrate DRAM or flash into a logic process. However, usually the memory arrays are much less efficient than in a dedicated process.
TI and Analog Devices have been able to incorporate MEMS with CMOS bits quite successfully in their respective MEMS devices (though they are the only two to do so commercially, to my knowledge).
TIs DLP devices use the upmost metallization layers only, with a process flow that is similar to a normal CMOS backend. Things are a bit more difficult with accelerometers which are based on polysilicon. Analog devices basically uses a dedicated MEMS process with tacked on analog part. Only very simple circuits are possible in that process. ("simple" as compared to todays average logic circuits)
Ignore all other replies, they are only half truth.
The truth is: Different manufacturing processes are required depending on the function of the chips. There are many different types of integrated circuits in a cellphone: Logic (processor), analog parts (Silicon and exotic III-V semiconductors), Memory (NAND flash, NOR flash, DRAM), Sensors (think MEMS). Each of these require a different process flow. Combining those is often extremely expensive to impossible.
The way it is usually done is to use different circuit techniques to achieve the same functionality in a silicon logic process. However in many situations this is not possible or economical, yet.
That is all there is. The propulsion principle has been known for at least a hundred years. The only 'new' thing is to use a diode to generate a DC field from externally applied AC. But actually that does not really solve any practical problem.
Slashdot Mirror
I remember there is an old Czech SciFi movie about an utopian society where men are obselete and women procastrinate by cloning. Anyone remember then name?
AMD divested of their memory business years ago. You should look more often.
Actually memory and logic are completely different worlds.
Ubuntu 7.1 finally made me realize that it is actually possible to use Linux as a Desktop OS. For at least 10 years Linux had been the ugly OS I only accessed via SSH on my server.
Nope, we are talking about thin film transistors. Main problem is getting the semiconductor off current under control (purity), improve mobility (crystallinity, purity, doping) and getting a good gate insulator that is not attacked by the deposition process. With certain materials it is literally possible to build transistors in your kitchen. (eg. CdS)
Look at the early work in thin films transistors. Schockleys attempts at building FETs suffered from poor silicon deposition. It took until the 70ies until people found out how to work with a-Si:H.
No quantum mechanic is required to describe field effect transistors in accumulation mode. And that is exactly what Lilienfeld proposed. The only theory that is required is that of space charge limited current, a field Lilienfeld has several publications in.
Mod parent down for being plain stupid.
Lilienfeld did in fact invent the working principle of the transistor. Whether he built one is not known. However he did all the groundlaying work on electrolytic capacitors as they are still used today. Therefore he knew very well how to create extremely thin insulating Al2O3 film that were a necessity for the type of transitor he described in his patents. It does therefore not appear entirely unlikely that he built some of the devices.
The stuff about surface states is mainly important for bulk silicon transistors. Even without solving this problem it is possible to demonstrate amplification as numerous publications on II-VI thin film transistors show.
Me me me! Long time ago.
In between I worked on organic transistors, normal silicon transistors, high-k devices.. you name it.
You mean without Teal? But that is for Silicon devices.
ITT Intermetall did manage to mass produce transistors without any license or technology transfer from Bell labs in the late 40ies...
The field effect transistor, the device that is relevant today, was invented and patented in 1926 by
Julius Edgar Lilienfeld. Due to his patents many claims by Bell Labs were thrown out.
The device that was invented by Bell Labs in 1947 was a point contact transistor. An inherently fragile device not fit for mass production. The same device was invented in parallel in France by two german Scientists: Welker and Matere see here.
Schockley himself did however invent the bipolar junction transistor a couple of years later. This invention was truly a streak of genius as it is the most complex of all devices.
So, thanks to american corporate giants history was rewritten again.
Ok, the article is talking about science fiction solutions that have been demonstrated for single bits at universities, but nobody has any idea how to mass produce it.
Meanwhile flash memory in production is approaching feature sizes of 30 nanometers with 2 or even 4 bits stored per cell. Also stacking of several memory layers on the same die has been demonstrated.
Scaling is not driven by lithography anymore. It is driven by material science advances. Intel introduced Hafnium based gate dielectrics in 45nm, which is an incredibly impressive feat given that they are 1-2 years beyond all other companies.
Scaling by lithography was in the 80ies and early 90ies.
There are a lot of wafer fabs in China. SMIC being the biggest company. The number is likely to increase.
"The agriculture ministry is not in charge of Gundam," said a ministry official
I lolled.. hard!
The rediculous hyping of research results from universities makes me barf. They make the industry appear modest and honest in comparison.
How is that a breakthrough? 60nm line patterning is nothing. Flash memory that is currently in production uses a smaller pitch - and that is not even just lines, but also more complex structures. Also it is done with orders of magnitudate less defects.
This "breakthrough" may have some nieche applications, but it is still far away from a "revolution".
The problem is that you do not have the slightest clue about semiconductor technology. You should not comment on something you dont understand.
Lets start here: How about scaling in nature? Why are small animals like insects "built" with an exosceleton while large mammals have bones? Is it possible that the concepts do not scale arbitrarily? Very similar in the semiconductor world. To make things smaller you actually have to change how they are made, both in terms of processing and materials.
CZO (Actually Zirconium-Yttrium-Oxide) is only similar in appearance to diamond. In all other respects it is completely different.
The most important difference is that diamond does burn in oxygen while CZO is an excellent oxygen conductor (yes, a crystal that conducts ions by a hole transport mechanism). This is also the effect that is used in fuel cells. There is really no relation to diamonds. Pure popular BS-science.
Actually, logic, analog, and memory are all CMOS (or CMOS-ish in the case of Flash) and can all be made on the same silicon production lines
Can be made on the same production line, yes, that is at least partially true. A big difference is however that the process flows will differ. Each of the different circuit types use different transistors (they are all mosfets but with variations in geometry, gox, implants..). Combining all these different transistor types is simple not economical since process complexity will increase significantly.
There are ways to integrate DRAM or flash into a logic process. However, usually the memory arrays are much less efficient than in a dedicated process.
TI and Analog Devices have been able to incorporate MEMS with CMOS bits quite successfully in their respective MEMS devices (though they are the only two to do so commercially, to my knowledge).
TIs DLP devices use the upmost metallization layers only, with a process flow that is similar to a normal CMOS backend. Things are a bit more difficult with accelerometers which are based on polysilicon. Analog devices basically uses a dedicated MEMS process with tacked on analog part. Only very simple circuits are possible in that process. ("simple" as compared to todays average logic circuits)
Ignore all other replies, they are only half truth.
The truth is: Different manufacturing processes are required depending on the function of the chips. There are many different types of integrated circuits in a cellphone: Logic (processor), analog parts (Silicon and exotic III-V semiconductors), Memory (NAND flash, NOR flash, DRAM), Sensors (think MEMS). Each of these require a different process flow. Combining those is often extremely expensive to impossible.
The way it is usually done is to use different circuit techniques to achieve the same functionality in a silicon logic process. However in many situations this is not possible or economical, yet.
Computer geek vs. science geek battle alert!
I will take the science side any time! Web technology fads come and go, science will stay.
Yes, but the paper companies only plant single species fast growing trees. Those can not replace the complex ecosystem in the rain forests.
Cute...
Please google Onsager, Faraday, Nernst and all the other pioneers of electrochemistry.
Bull.. - please review your electronics 101 notes.
That is all there is. The propulsion principle has been known for at least a hundred years. The only 'new' thing is to use a diode to generate a DC field from externally applied AC. But actually that does not really solve any practical problem.