Our team of 3 built our first alloy transistor, at MSU, as part of a class project, and it had a gain of 4. We were third best, with another group getting a gain of 8, and number two getting a gain of 6. The rest of the class all got gains below 1. We got an A-grade, while numbers one and two got an A+plus. I complained to our TA, and he told me to suck it up, since the rest of the class got C's. Academic Life was harsh back then. So later I became an IC designers, with several ASICs under my belt.
The FET, and especially the MOSFET, was a close equivalent to the way vacuum tubes work. This fact is often skipped in the history of transistors. In electronic design, the vacuum tube was an easily understood device. Put a voltage on the grid, stop current. Remove the grid voltage, current goes. On/off.
Yes, the MOSFET closely resembles the way the old vacuum tube works. In fact, tubes are more like JFETs than MOSFETs because to actually shut off a vacuum tube, you need to apply a negative voltage to the grid, just as you need to apply a negative voltage to a JFET's gate to shut it off.
You're talking about depletion mode vs enhancement mode. Depletion mode devices are on with no gate voltage and switch off when the gate is pulled 'negative'. Enhancement mode devices are off with no gate voltage and switch on when the gate is pulled 'positive'. As far as I know, JFETs are always depletion mode devices. Most MOSFETs are enhancement mode devices but depletion mode MOSFETs exist. Valves were very similar to depletion mode FETs. They definitely were not just on/off devices. Like transistors, they could be used in linear mode or as switches.
@@nicholasvinen I appreciate you explaining the errors in the original commenter's statements. I picked up on that also. I endorse your response in the context of this forum.
another correction might be that tubes were not used as switches in majority of their applications.... there was a lot of stuff happening right between on and off...ie most of it... essentially, tubes are modulation devices....
One late night around 1975, my dad and I had a conversation. He designed satellites. In the conversation, he mentioned how everyone went with bipolar TTL but he was using RCA’s CMOS. It worked much better in the environment of space. I was in high school at the time. I built a stopwatch using RCA CMOS SSI chips. I still have the RCA books that he showed me that night.
@@ccshello1 I believe I was the first engineer to design-in CD4000 CMOS devices into a product in New Zealand - early 1970. It was so EASY to use this logic. But one detail of my schematic for the impulse counter boards was overlooked by the draftsperson, and 20 PCB's were made and put into service. Oh my! About 3 months after commissioning the power-load-monitor for a large paper mill, the first card failed. Reason? They had all been operating perfectly well without a Vdd supply. And functioned just fine because of the input protection diodes on each chip pumping the supply rails from input signals only - until one card did not. Quickly fixed of course. Very red face over that, but it taught me to henceforth go deep on the physical performance of any design and thoroughly inspect the actual physical details around anything I subsequently designed. I was not quite 20 years old at the time.
I used to design with CMOS. Loved the technology. When run in the static regime it would draw next to no power. Made an alarm for my garden shed. It ran for about ten years on the same PP3 battery!
I'd like to add that one of my instructors, Robert W. Bower, made CMOS cost effective with his Self-Aligned Gate MOSFET, U.S. Patent No. 3,472,712 . This resulted in the MASS production of MOSFETS.
He has been using this block diagram for quite some time that shows the Source and Drain in different colors, which imply they are different materials or are doped differently, when in fact the Source and Drain regions are, effectively, identical in construction, and what makes one a “source” and the other a “drain” really only has to do with how the transistors are hooked up and connected to the rest of the circuit. I think it would behoove him to update his cartoon model to show the source and drain in the same color, indicating they are identical in construction and composition. The colors should be used to indicate whether the regions are n-type or p-type, and therefore the colors can be used for both “wells” (silicon substrate under the transistor) and “active areas” (source, drain)
Yup, to add, NMOS is named after the activated/inverted channel, the chemical doping is the opposite of the 'ON' channel. When gate is turned on, the p-doped Si inverts to form n-channel. A separate FET device is the TFT with metal S/D where n-type channel gets accumulation and becomes even more conductive, no need for inversion.
Source/Drain doping determines the transistor type. If the doping is n-type, it's NMOS. If the doping is p-type, it's PMOS. There could be only one well in a more straightforward process or two wells in a more advanced process.
it's incredible hearing stories about people from all over the world beating the odds and moving to america to try their best in the highest areas of competitiveness. we should celebrate that kind of thing more
Wizards inscribe secret runes upon thin sheets of metal and then harness the power of lightning through it to force the metal to think, and you still think magic isn't real?
Look at it through the eyes of the Scientific Method …. Electrons flow through traces of a electro conducting metal embedded in a di-electric material …..
"Magic is any technology so advanced that it's working principles are completely incomprehensible to the observer". Most things are magic to most people. There was a time when producing fire from something not already on fire was magic.
In the description op MOSfets, you made some mistakes. Firstly, an nMOS transitor is in a p subtrate or well and a pMOS is in an n well or substrate. The n refers to the carriers after the channel is inverted (ie, when the fet starts to conduct). Also, when applying a positive voltage to the gate, electrons are attracted into the channel and with negative voltage holes are. Opposite charges attract and equal charges repel.
Federico Faggin while at Fairchild worked on the self-aligned Si gate MOS technology. He then went on to design the 4004 CPU at Intel using the same technology. Cf his autobiography.
I worked at a Philips chip factory in the Netherlands in the late 80's, and we did have Ion cannons there, but most of the processes were still done with ovens that would get 1300ºc hot and the wafers would be in there for about 13 hours so we never would handle the wafers we had put ourselves in the oven. We were creating the one chip TV from Phiips. The good old times when Philips was still a big big player in the world market.
I don't come close to understanding all the details in even this video, which is a tiny, tiny piece of the whole understanding of microelectronics technology. What I do appreciate is the astounding scale of the big picture. What a long, strange trip it's been, from those early cat whiskers to an IC with feature sizes of a few nanometers, and many trillions of devices fabricated on one wafer. If this doesn't blow your mind, you must not have a mind to be blown. And we're not finished yet. Far from it.
Not tired of this diagram yet. Used all of this for so long in industry. In hobby. And as a child, marveled at using the same. Happy to watch this and to watch this again. Important history for so many reasons. So many layers. Pun intended
"since we are in the caveman days before the British commercialised ion implantation..." a video on the Rise and fall of Lintott (later Applied Material Implant Division) would be cool. I worked there in the late 90s...
When I started in IC design, the first chip had a metal gate process. I sure don't miss that mess! :D Neither do I miss processes with only one metal layer.
I was around when dual-layer metal was being attempted. It was not uncommon back then to get zero yields due to "pin holes" in the insulating SiO2 insulator between the two layers that shorted the two metal layers. Thankfully, they figured it out over time!
@@demef758 I suppose the only thing to miss about single metal is how easdy it was to reverse engineer the chips. For a while, I had a side hustle doing that for some Intel chips and the NEC floppy disk controller.
Amazing how long the planar transistor technology has dominated the industry. I wonder which technological break throughs enabled the transition to vertical gates and the current FinFET transistors. Thank you for this well researched piece which was delivered by you in the usual concise manner. I really like your presentation style. Thank you.
Actually vacuum tubes or valves as the British call them are field effect devices made well before MOSFET devices. Tubes were the first devices to use field effect.
To the Digital Age is an awesome book. And to think in not too soon a time we're going to have precision tunable analog "transistors" working together with digital that can do things digital transistors cannot do efficiently. Mostly in the fields of research for the time being but I'm sure people will come up with multitudes of uses for such nearly invariant analog signals.
I love your deep dive into the early days videos... unfortunately for viewers man I'm going to assume those are also the ones that take the most time to research.
In the (more common) case of an enhancement type MOSFET yes, but a depletion type MOSFET also exists and takes a negative voltage to turn off. Both existed early on, though the enhancement type version was the first one to be made
I find your videos a very good background and deeply reserched, into the subject being viewed. Most of them were being developed whilst I was at school and into the electronics industry I joined. Namely , English Electric, Marconi Insruments, and I feel the best Ferranti Defence Systems and the semiconductor system. They were developing a silicone on saphire F200L version of their F100L whilst I worked for them, and well missed in 😊these times, we could do with an̈ updated version of the Ferranti Bloodhound, to protect us in the UK. You glanced over Telydynes Fetron, high voltage fet's swap into valve replacement for several common US and European valves. It made it brief inroads mainly in the Britidh Post Office, long-distance telephone links, but was quickly overtaken by conventional semiconductor products. Thanks for the video's.
I think your technology series are great, but could you draw the source and drain in the same color? They're interchangeable on a mosfet, just both n or p regions.
Just being pedantic but one does not "suffer" an aortic aneurysm - they may suffer an aortic aneurysm rupture but many have aneurysms without being aware.
They really improved mosfet transistors in the last decade or so in every possible way. They make nitride and carbide mosfets that can operate at hundreds of watts of RF power and at temperatures that would cook silicon. On the other extreme they have the ones on microprocessor chips that have nearly zero leakage when off and can switch on at insane speed.
ck in the 60's I remember having 8 transisters on a board that gave us 8 inverters. Double th transisters and we had an amplifier. Put these in a main frame with "AND", "ORs" & "Flip Flops" with other items and you had a "DATA PROCESSOR". Those were the days. New inventions every day.
6:50 Dr. I. M. Ross. In between 1962 and 1973, per US Government request, few hand-picked Bell Labs researchers and engineers were transferred out to a newly formed company: Bellcomm Inc. in Washington DC. The firm's task is to serve System Engineering role and coordinate with many NASA contractors to make Lunar landing possible. Yes, that's Apollo program and later Skylab. After they completed the tour of duty, they returned back to BTL. Many of them soon got promoted to senior leadership positions: Ian (Bell Labs president), Day (BL Chief Architecture SVP), Martersteck (BL Switching Systems SVP), and many others. ====== BTW, 3N99 is a depletion mode MOSFET, a not-so-common semiconductor component.
During the brief hey-day of NMOS, depletion mode transistors were VERY common thanks to ion implantation, which allowed manufacturers to place ions under the gate, "pre-inverting" the surface. Aside from that, you are correct that finding a discrete N-channel depletion-mode device was like trying to find Big Foot.
Another aspect is VLSI, how do you then fabricate a chip with millions of these transistors. The birth of layout techniques gave rise to VLSI. We all had to study the book by Carver aMead and Lynn Conway.
And It was Federico Faggin that has invented the silicon gate technology back where he used to work at Fairchild and insisted in using it for the Intel 4004 later.
As a Certified Geek, I took an electronics voc-tec class in the early 1960's. I still perfectly understand how vacuum tubes work. But solid state electronics still baffles me. What? How does a "hole" travel? Anyway, the MOSFET works just like an old triode vacuum tube. Meaning that a charge in the grid or gate controls the flow of the electrons trying to move from filament to plate or source to drain.
A hole, event though is has "mobility" and a speed, does not actually travel, since it is a "vacancy" or essentially a "missing" part of the bulk. What travels is the energy wave around the hole, which translates as an evanescent wave through the structure, where the outer electrons are essentially a protective wall, around the vacancy. This electron bubble continues to move toward some potential, until it is absorbed in some Fermi band, or hits some kind of interference, such as a defect within the structure, where that energy shell is disrupted in random directions, and then the hole collapses into the thermal zone. The hole "per se" contains no energy, but the shell does. A crude analog might be a tornado, or an eddy current. The "vacancy" inside those also can contain no significant mass or other obvious energy, and it can be a vacuum, but the shell can suck up whole houses.
your diagram is the other way around at 12:13, NMOS has p-type bulk. the structure of the device becomes NPN. if you have p bulk (NMOS), you need to apply positive voltage NOT negative voltage to create electron channel. on the other hand PMOS has n-type bulk. the structure of the device becomes PNP. if you have n-type bulk, you need to apply negative voltage to create channel. the diagram shows NMOS (NPN) with the operation voltage of a PMOS which does not make sense negative voltage does not attract electron which has negative charge, and push holes which has positive charges, positive voltage does.
Interesting that back then all these different variations were more seen like competitors. These days they just all have their pros and cons and just coexist.
In the USSR, the first research work on a semiconductor triode (transistor) was carried out at NII-160 (now NPP Istok) by MCTI graduate Susanna Gukasovna Madoyan. A laboratory prototype of a (point-point) transistor started working in February 1949. Serial production of point-point transistors (TS1 - TS7) began in 1953, planar transistors (P1) - in 1955.
Just a comment about MOSFETs verses bipolar. MOSFETS are basically a capacitor. Its not rocket science to figure out that a capacitor is going be slower than a transistor that is not based on a capacitor. Moving stored charge around costs time. However, simple math would tell you that as ICs scale, the capacitor would become less and less of an issue, until a crossover point is reached. In the 1980s, I programmed an IC simulator that modeled transistors as an RC (resistor-capacitor) with a switch. IE, the function of the waveform out is that of a resistor discharging a capacitor. The idea comes from Bell Labs, and it allows analog simulation of transistor networks with a dramatically simpler model than SPICE, and thus much faster math than SPICE, which must carry out a relaxation because of reactive effects.
re: "In the 1980s, I programmed an IC simulator that modeled transistors as an RC (resistor-capacitor) with a switch" So, a simulator with less dependence on the physics-based model (which becomes computationally complex), but rather an effective model that incorporated the overall 'gross' or total known effects of the physical transistors for the purpose of modelling an IC as a device?
@@uploadJ Sure. The first simulator I did was a standard switch level simulator. These are good, but don't consider many physical attributes such as the resistance of the switch other than being able to produce "weak" states, so that transistors with greater power can take over a node. The RC simulator gives a simulation between switch level and SPICE, and gives actual switching curves, a real picture of strong vs. weak switches, and yet gives reasonable speed.
@@scottfranco1962 Interesting, Thanks. My background is RF, and we use different tools many of which are S-Parameter based and quite different from SPICE. I have used SPICE before to model RF filters, to get an idea of node voltages and currents, for the rating of components for part selection.
@@uploadJ Yea, the view of an IC as a series of capacitors discharging through resistors by switches is obviously limited. However, its nice to see curves coming out of such a simple simulation. I had hopes for implementing the algorithm in hardware, and still do. Like getting a large net that executes reasonably fast, but is actually analog based. The impetus for all that was that our regular switch based simulator at Zilog could not handle resets, so that was done by node sets. I was interested in seeing if I could make a simulator that didn't need that kludge, mainly because I had just solved a serious reset bug in one of our chips! I was able to do that, mainly by introducing weak states into the simulation (this was the digital only simulator). I think that is standard now. They just killed my old chip, the Z80. Ah well. Crappy instruction set really. I made a compiler for it, but I can't believe I did that now.
@@scottfranco1962 No kidding - the Z80? I developed a digital controller for a trunked receiver (old Motorola analog format) using the Z80 and its associated chip set (CTC, PIO, SIO) ... wrote the software to phase track the incoming GMSK datastream looking for Zero crossing which triggered a PIO interrupt wherein I would inspect the CTC count in a timer ... adjustment would follow ... also did timer reloading P/Q for M/N cycles since 4 MHz clock was not a direct integer of 3600 baud GMSK datastream rate. A rate 1/2 Convolutional ECC algorithm followed coupled with a 10 bit parity check to insure the ECC worked its magic.
I will recommend that you remind people to check the subscribe status. For some reason, TH-cam unsubscribes people from channels. I see every single video, often the same day they are released, so it's not due to lack of activity.
I suppose it would be a bit of a tangent, but I'm very interested to know why some solid-state transistors - like FETs in general? - are able to simulate effects of vacuum tubes, in technologies where that might be desirable (e.g. audio amplification), while others don't appear to have that ability. It's a quality which is probably not very helpful in some applications, but quite desirable in others.
It's mainly to do with the transfer curve. Valves and FETs are both transconductance devices, ie, use voltage to control a current. But they have different curves of input voltage vs output current. They also saturate differently. It is possible, but not easy, to make a JFET have a curve very much like a signal valve. It requires careful biasing and usually adjustment of every device built.
Your history is missing an important practical development, that happened in the mid 20-30 of the 20th century, an that is the invention and construction of the fist semiconductor electronic device by Dr. Thomas Henry Moray. Dr. Moray was showing and explaining his invention to Dr. Harvey Fletcher, in his several trips to Utah, who late became the head of the Bell Labs department where the "invention" of the transistor was pronounced 20 years after Dr. Moray' invention.
To your point about patentability, a patent only has to be non-obvious and novel. It does not have to be correct. It's hilarious to go from the rigors of academic writing to writing patent applications.
@@TheLegendaryHacker for the same reason, early access tier on Patreon! I’m part of a tier that gets some videos early but not all. My Patreon bill is also so high lol. The scanning microscope one got released eventually.
Not really. Manufacturing transistors (and other solid state devices, including CCD's) requires a significant support structure to exist. Note the first transistor section later in the vid.
If you want REAL drama, study the invention of the transistor at Bell Labs under Shockley. Prior to this endeavor, Shockley was reported to be a real good guy to work under. But after it, he lost his marbles, jealous as hell of Bardeen and Brattain who found that Shockley's original idea would not work, and then those two had the temerity to go off and create the device that did work (i.e., "demonstrated gain"). Shockley was never the same after that, eventually quit the Labs, and headed to the West Coast. This move eventually led to the creation of Fairchild Semiconductor where the planar transistor and integrated circuit (the REAL integrated circuit, not TI's Kilby transistor that was NEVER used to manufacture a single IC) were perfected.
Vacuum tubes need a lot of power, those filaments need to be heated to emit electrons. Also not as reliable if you have thousands of them. Consider that the filament will slowly evaporate unless you drive at very low emission setting. Similar failure mode for light bulbs, which also have a filament but optimized for light emission not electron emission.
Integrated circuitry amazes me. I wonder, are the occurence of IC's just an inventive exploit stumbled upon by random chance, or a fundamental phenomenon hidden in our universe only waiting to be discovered and utilized? Anyone can look at a simple IC architectural diagram and wonder how anything like it could have been concieved, much less crafted and brought functionally into existence. Thanks for making this video!
HELLO this are excellent in-depth videos that I just found and am very impressed ! Well done! I would love you to do a video on China's progress on digital memory. i.e DRAM & NAND. I know CXMT is the leader in DRAM, but there is v little info on how it's faring under export controls, as DRAM needs DUV and EUV lithography also. Can you do a video, or if not pls post links to this message where I can track CXMT's progress. thank you !
No current flows through the gate. A bipolar amd threadripper would require a nuclear reactor to power. FETs require very little comparatively. My understanding is this revolutionized old hot mainframes with bjt chips.
But if you only use it to construct sine and cosine then you only need a bipolar zeener junction transistor to explain two simultaneous GREATER | LESS -> SWITCH CASE, so a gate array is a high level operation trained at *_fairchild budy pal_* for snake game on a 32MB PLA slice.
Yeah... about that... For historical reasons, that nobody bothered correcting, your intuitive idea is correct, but the established nomenclature is the opposite of what actually happens. Sorry. It bothers me too.
@@andersjjensen yeah they guessed the electron charge wrong and by the time they actually found out what it really was it was too late to change everything to how things really are. So here we all are today with negatively charged electrons. Most of the time it doesn't really matter. But when it matters then it does.
Our team of 3 built our first alloy transistor, at MSU, as part of a class project, and it had a gain of 4. We were third best, with another group getting a gain of 8, and number two getting a gain of 6. The rest of the class all got gains below 1. We got an A-grade, while numbers one and two got an A+plus. I complained to our TA, and he told me to suck it up, since the rest of the class got C's. Academic Life was harsh back then. So later I became an IC designers, with several ASICs under my belt.
What, no safe space to retreat to?
@@demef758 back in those days, we called that the bench, and people there were called bench warmers.
Given the spread, A grade was very sensible. I wonder why you complained and if your academic life was harsh back then is sarcasm or not.
Meritocracy in practice!
Bravo
The FET, and especially the MOSFET, was a close equivalent to the way vacuum tubes work. This fact is often skipped in the history of transistors. In electronic design, the vacuum tube was an easily understood device. Put a voltage on the grid, stop current. Remove the grid voltage, current goes. On/off.
Yes, the MOSFET closely resembles the way the old vacuum tube works. In fact, tubes are more like JFETs than MOSFETs because to actually shut off a vacuum tube, you need to apply a negative voltage to the grid, just as you need to apply a negative voltage to a JFET's gate to shut it off.
You're talking about depletion mode vs enhancement mode. Depletion mode devices are on with no gate voltage and switch off when the gate is pulled 'negative'. Enhancement mode devices are off with no gate voltage and switch on when the gate is pulled 'positive'.
As far as I know, JFETs are always depletion mode devices. Most MOSFETs are enhancement mode devices but depletion mode MOSFETs exist.
Valves were very similar to depletion mode FETs. They definitely were not just on/off devices. Like transistors, they could be used in linear mode or as switches.
@@nicholasvinen I appreciate you explaining the errors in the original commenter's statements. I picked up on that also. I endorse your response in the context of this forum.
another correction might be that tubes were not used as switches in majority of their applications....
there was a lot of stuff happening right between on and off...ie most of it...
essentially, tubes are modulation devices....
What about a p-channel MOSFET. Put a voltage on the grid; there is no hole current. Remove the grid voltage, still no hole current?
One late night around 1975, my dad and I had a conversation. He designed satellites. In the conversation, he mentioned how everyone went with bipolar TTL but he was using RCA’s CMOS. It worked much better in the environment of space. I was in high school at the time. I built a stopwatch using RCA CMOS SSI chips. I still have the RCA books that he showed me that night.
CD4000 series, THE Classic!
@@ccshello1 I believe I was the first engineer to design-in CD4000 CMOS devices into a product in New Zealand - early 1970. It was so EASY to use this logic.
But one detail of my schematic for the impulse counter boards was overlooked by the draftsperson, and 20 PCB's were made and put into service. Oh my! About 3 months after commissioning the power-load-monitor for a large paper mill, the first card failed.
Reason?
They had all been operating perfectly well without a Vdd supply. And functioned just fine because of the input protection diodes on each chip pumping the supply rails from input signals only - until one card did not. Quickly fixed of course. Very red face over that, but it taught me to henceforth go deep on the physical performance of any design and thoroughly inspect the actual physical details around anything I subsequently designed. I was not quite 20 years old at the time.
Me Too. I have all the Motorola IC Design Books, heavily marked up.
I used to design with CMOS. Loved the technology. When run in the static regime it would draw next to no power. Made an alarm for my garden shed. It ran for about ten years on the same PP3 battery!
I like the use of a picture of a MOSBURGER store.
There's humor everywhere. The 1960s oven, for instance, with a vacuum fluorescent display and a matching microwave.
I'd like to add that one of my instructors, Robert W. Bower, made CMOS cost effective with his Self-Aligned Gate MOSFET,
U.S. Patent No. 3,472,712 . This resulted in the MASS production of MOSFETS.
Oh... but you see - he's no ASIAN, so it doesn't count.
Quick correction: it is the opposite: nMOS are n-channel MOSFETs, so p-doped wells, while pMOS have an n-well. Thank you for your great videos!
I presume you are referring to 11:15
He has been using this block diagram for quite some time that shows the Source and Drain in different colors, which imply they are different materials or are doped differently, when in fact the Source and Drain regions are, effectively, identical in construction, and what makes one a “source” and the other a “drain” really only has to do with how the transistors are hooked up and connected to the rest of the circuit.
I think it would behoove him to update his cartoon model to show the source and drain in the same color, indicating they are identical in construction and composition. The colors should be used to indicate whether the regions are n-type or p-type, and therefore the colors can be used for both “wells” (silicon substrate under the transistor) and “active areas” (source, drain)
Yup, to add, NMOS is named after the activated/inverted channel, the chemical doping is the opposite of the 'ON' channel. When gate is turned on, the p-doped Si inverts to form n-channel.
A separate FET device is the TFT with metal S/D where n-type channel gets accumulation and becomes even more conductive, no need for inversion.
Source/Drain doping determines the transistor type. If the doping is n-type, it's NMOS. If the doping is p-type, it's PMOS.
There could be only one well in a more straightforward process or two wells in a more advanced process.
When your content dropped, it was the perfect end to my father's day. Thanks lad.
it's incredible hearing stories about people from all over the world beating the odds and moving to america to try their best in the highest areas of competitiveness. we should celebrate that kind of thing more
It’s worth noting that William Shockley was an avowed racist and eugenist.
Wizards inscribe secret runes upon thin sheets of metal and then harness the power of lightning through it to force the metal to think, and you still think magic isn't real?
Look at it through the eyes of the Scientific Method …. Electrons flow through traces of a electro conducting metal embedded in a di-electric material …..
"Magic is any technology so advanced that it's working principles are completely incomprehensible to the observer". Most things are magic to most people. There was a time when producing fire from something not already on fire was magic.
Oh that's not right at all. Silicon is considered a metalloid, not a metal. The part about wizards and secret runes, though? Spot on.
TWO NOBEL PRIZES? Well ok then, that's a very short list of people (5)
In the description op MOSfets, you made some mistakes. Firstly, an nMOS transitor is in a p subtrate or well and a pMOS is in an n well or substrate. The n refers to the carriers after the channel is inverted (ie, when the fet starts to conduct). Also, when applying a positive voltage to the gate, electrons are attracted into the channel and with negative voltage holes are. Opposite charges attract and equal charges repel.
Federico Faggin while at Fairchild worked on the self-aligned Si gate MOS technology. He then went on to design the 4004 CPU at Intel using the same technology.
Cf his autobiography.
this is only 25 mins but seemed that Faggin was noticeably absent
I worked at a Philips chip factory in the Netherlands in the late 80's, and we did have Ion cannons there, but most of the processes were still done with ovens that would get 1300ºc hot and the wafers would be in there for about 13 hours so we never would handle the wafers we had put ourselves in the oven.
We were creating the one chip TV from Phiips. The good old times when Philips was still a big big player in the world market.
It's not that bad. We have 2 electric toothbrushes and one airfryer from Philips. I bet those toothbrushes are good cash cows.
Applause for the day I never missed your video 😁😁 ,thanks for your detailed information once again 👏
I don't come close to understanding all the details in even this video, which is a tiny, tiny piece of the whole understanding of microelectronics technology. What I do appreciate is the astounding scale of the big picture. What a long, strange trip it's been, from those early cat whiskers to an IC with feature sizes of a few nanometers, and many trillions of devices fabricated on one wafer. If this doesn't blow your mind, you must not have a mind to be blown.
And we're not finished yet. Far from it.
The comments under your videos are just as amazing as the videos itself!
My Sunday evenings consist of coding javascript, and learning about the history of transistors; thank you again.
Sorry about the JavaScript 😅
@@fensoxx It's a blast, actually; like ATARI player-missile graphics on crack
There are support groups for people struggling with JavaScript related cognitive impairment. And remember: Anything is better than suicide :P
@@fensoxx I'm currently 'trying' to learn javascript, what do you mean?
Not tired of this diagram yet. Used all of this for so long in industry. In hobby. And as a child, marveled at using the same. Happy to watch this and to watch this again. Important history for so many reasons. So many layers. Pun intended
11:49 I'm glad you had as much trouble explaining exactly how transistors work as I did trying to understand them.
"since we are in the caveman days before the British commercialised ion implantation..." a video on the Rise and fall of Lintott (later Applied Material Implant Division) would be cool. I worked there in the late 90s...
When I started in IC design, the first chip had a metal gate process. I sure don't miss that mess! :D Neither do I miss processes with only one metal layer.
I was around when dual-layer metal was being attempted. It was not uncommon back then to get zero yields due to "pin holes" in the insulating SiO2 insulator between the two layers that shorted the two metal layers. Thankfully, they figured it out over time!
@@demef758 I suppose the only thing to miss about single metal is how easdy it was to reverse engineer the chips. For a while, I had a side hustle doing that for some Intel chips and the NEC floppy disk controller.
Lots of good chips were designed with a single layer of aluminum.
@@allanflippin2453 You cannot tell anyone, but our company also used to de-cap chips, and reverse engineer them.
Amazing how long the planar transistor technology has dominated the industry. I wonder which technological break throughs enabled the transition to vertical gates and the current FinFET transistors.
Thank you for this well researched piece which was delivered by you in the usual concise manner. I really like your presentation style. Thank you.
Actually vacuum tubes or valves as the British call them are field effect devices made well before MOSFET devices. Tubes were the first devices to use field effect.
To the Digital Age is an awesome book. And to think in not too soon a time we're going to have precision tunable analog "transistors" working together with digital that can do things digital transistors cannot do efficiently. Mostly in the fields of research for the time being but I'm sure people will come up with multitudes of uses for such nearly invariant analog signals.
I love your deep dive into the early days videos... unfortunately for viewers man I'm going to assume those are also the ones that take the most time to research.
2:30 that is why he is called Shockley
And don't forget Schottky. His diodes were better.
Ahh, positive gate voltage turns on NMOS device…
In the (more common) case of an enhancement type MOSFET yes, but a depletion type MOSFET also exists and takes a negative voltage to turn off. Both existed early on, though the enhancement type version was the first one to be made
I find your videos a very good background and deeply reserched, into the subject being viewed.
Most of them were being developed whilst I was at school and into the electronics industry I joined.
Namely , English Electric, Marconi Insruments, and I feel the best Ferranti Defence Systems and the semiconductor system.
They were developing a silicone on saphire F200L version of their F100L whilst I worked for them, and well missed in 😊these times, we could do with an̈ updated version of the Ferranti Bloodhound, to protect us in the UK.
You glanced over Telydynes Fetron, high voltage fet's swap into valve replacement for several common US and European valves. It made it brief inroads mainly in the Britidh Post Office, long-distance telephone links, but was quickly overtaken by conventional semiconductor products.
Thanks for the video's.
that image at 24:14 alone justifies all your videos !
I think your technology series are great, but could you draw the source and drain in the same color? They're interchangeable on a mosfet, just both n or p regions.
The MOSFET was indeed a game-changer, revolutionizing the semiconductor industry. Stellar video.
It's now 4 in the morning in southern Europe.
Me: 👀🍿 new asiometry video.
Just being pedantic but one does not "suffer" an aortic aneurysm - they may suffer an aortic aneurysm rupture but many have aneurysms without being aware.
They really improved mosfet transistors in the last decade or so in every possible way. They make nitride and carbide mosfets that can operate at hundreds of watts of RF power and at temperatures that would cook silicon. On the other extreme they have the ones on microprocessor chips that have nearly zero leakage when off and can switch on at insane speed.
Several misspelled names occur in your video.
ck in the 60's I remember having 8 transisters on a board that gave us 8 inverters. Double th transisters and we had an amplifier. Put these in a main frame with "AND", "ORs" & "Flip Flops" with other items and you had a "DATA PROCESSOR". Those were the days. New inventions every day.
6:50 Dr. I. M. Ross. In between 1962 and 1973, per US Government request, few hand-picked Bell Labs researchers and engineers were transferred out to a newly formed company: Bellcomm Inc. in Washington DC. The firm's task is to serve System Engineering role and coordinate with many NASA contractors to make Lunar landing possible. Yes, that's Apollo program and later Skylab. After they completed the tour of duty, they returned back to BTL. Many of them soon got promoted to senior leadership positions: Ian (Bell Labs president), Day (BL Chief Architecture SVP), Martersteck (BL Switching Systems SVP), and many others.
======
BTW, 3N99 is a depletion mode MOSFET, a not-so-common semiconductor component.
During the brief hey-day of NMOS, depletion mode transistors were VERY common thanks to ion implantation, which allowed manufacturers to place ions under the gate, "pre-inverting" the surface. Aside from that, you are correct that finding a discrete N-channel depletion-mode device was like trying to find Big Foot.
I hate cliffhangers, can't wait to see what happens in 2011
Go through his back-catalogue and find the episode on FinFETs. I don't recall the title, but I know it's there.
Another aspect is VLSI, how do you then fabricate a chip with millions of these transistors. The birth of layout techniques gave rise to VLSI. We all had to study the book by Carver aMead and Lynn Conway.
Morgan Sparks and Gordon Teal. Image in the video is correct, but you mish mashed the names in the audio.
Wow incredible content, never heard about passivation before, never thought the oxide layer was that important!
And It was Federico Faggin that has invented the silicon gate technology back where he used to work at Fairchild and insisted in using it for the Intel 4004 later.
He was Faggin
As a Certified Geek, I took an electronics voc-tec class in the early 1960's. I still perfectly understand how vacuum tubes work. But solid state electronics still baffles me. What? How does a "hole" travel?
Anyway, the MOSFET works just like an old triode vacuum tube. Meaning that a charge in the grid or gate controls the flow of the electrons trying to move from filament to plate or source to drain.
A hole, event though is has "mobility" and a speed, does not actually travel, since it is a "vacancy" or essentially a "missing" part of the bulk. What travels is the energy wave around the hole, which translates as an evanescent wave through the structure, where the outer electrons are essentially a protective wall, around the vacancy. This electron bubble continues to move toward some potential, until it is absorbed in some Fermi band, or hits some kind of interference, such as a defect within the structure, where that energy shell is disrupted in random directions, and then the hole collapses into the thermal zone. The hole "per se" contains no energy, but the shell does. A crude analog might be a tornado, or an eddy current. The "vacancy" inside those also can contain no significant mass or other obvious energy, and it can be a vacuum, but the shell can suck up whole houses.
RIP Mohamed Atalla and Dawon Kahng. It's a shame they never won a Nobel Prize for inventing the MOS transistor.
24:19 LOL @ MOSBurgers!!!!! 😅😅😅😅😅
The amazon 1 click patent diss though 😂😂😂
your diagram is the other way around at 12:13,
NMOS has p-type bulk. the structure of the device becomes NPN. if you have p bulk (NMOS), you need to apply positive voltage NOT negative voltage to create electron channel.
on the other hand PMOS has n-type bulk. the structure of the device becomes PNP. if you have n-type bulk, you need to apply negative voltage to create channel.
the diagram shows NMOS (NPN) with the operation voltage of a PMOS which does not make sense
negative voltage does not attract electron which has negative charge, and push holes which has positive charges, positive voltage does.
I could listen to your essays on semiconductor technology for hours on end......cheers !!
Interesting that back then all these different variations were more seen like competitors.
These days they just all have their pros and cons and just coexist.
Trivia: Amazon's 1-click patent expired in 2017
Like the mosfet won the world, asianometry has won TH-cam!
Here here 👏
In the USSR, the first research work on a semiconductor triode (transistor) was carried out at NII-160 (now NPP Istok) by MCTI graduate Susanna Gukasovna Madoyan. A laboratory prototype of a (point-point) transistor started working in February 1949. Serial production of point-point transistors (TS1 - TS7) began in 1953, planar transistors (P1) - in 1955.
@@TyphoonUSSRI'm gay but that's ok
Ingratiation will you get you only so far in life, but top comment on youtube
@@Abhishek_78 yummy
Just a comment about MOSFETs verses bipolar. MOSFETS are basically a capacitor. Its not rocket science to figure out that a capacitor is going be slower than a transistor that is not based on a capacitor. Moving stored charge around costs time. However, simple math would tell you that as ICs scale, the capacitor would become less and less of an issue, until a crossover point is reached.
In the 1980s, I programmed an IC simulator that modeled transistors as an RC (resistor-capacitor) with a switch. IE, the function of the waveform out is that of a resistor discharging a capacitor. The idea comes from Bell Labs, and it allows analog simulation of transistor networks with a dramatically simpler model than SPICE, and thus much faster math than SPICE, which must carry out a relaxation because of reactive effects.
re: "In the 1980s, I programmed an IC simulator that modeled transistors as an RC (resistor-capacitor) with a switch"
So, a simulator with less dependence on the physics-based model (which becomes computationally complex), but rather an effective model that incorporated the overall 'gross' or total known effects of the physical transistors for the purpose of modelling an IC as a device?
@@uploadJ Sure. The first simulator I did was a standard switch level simulator. These are good, but don't consider many physical attributes such as the resistance of the switch other than being able to produce "weak" states, so that transistors with greater power can take over a node. The RC simulator gives a simulation between switch level and SPICE, and gives actual switching curves, a real picture of strong vs. weak switches, and yet gives reasonable speed.
@@scottfranco1962 Interesting, Thanks. My background is RF, and we use different tools many of which are S-Parameter based and quite different from SPICE. I have used SPICE before to model RF filters, to get an idea of node voltages and currents, for the rating of components for part selection.
@@uploadJ Yea, the view of an IC as a series of capacitors discharging through resistors by switches is obviously limited. However, its nice to see curves coming out of such a simple simulation. I had hopes for implementing the algorithm in hardware, and still do. Like getting a large net that executes reasonably fast, but is actually analog based.
The impetus for all that was that our regular switch based simulator at Zilog could not handle resets, so that was done by node sets. I was interested in seeing if I could make a simulator that didn't need that kludge, mainly because I had just solved a serious reset bug in one of our chips! I was able to do that, mainly by introducing weak states into the simulation (this was the digital only simulator). I think that is standard now.
They just killed my old chip, the Z80. Ah well. Crappy instruction set really. I made a compiler for it, but I can't believe I did that now.
@@scottfranco1962 No kidding - the Z80? I developed a digital controller for a trunked receiver (old Motorola analog format) using the Z80 and its associated chip set (CTC, PIO, SIO) ... wrote the software to phase track the incoming GMSK datastream looking for Zero crossing which triggered a PIO interrupt wherein I would inspect the CTC count in a timer ... adjustment would follow ... also did timer reloading P/Q for M/N cycles since 4 MHz clock was not a direct integer of 3600 baud GMSK datastream rate. A rate 1/2 Convolutional ECC algorithm followed coupled with a 10 bit parity check to insure the ECC worked its magic.
I will recommend that you remind people to check the subscribe status. For some reason, TH-cam unsubscribes people from channels. I see every single video, often the same day they are released, so it's not due to lack of activity.
I suppose it would be a bit of a tangent, but I'm very interested to know why some solid-state transistors - like FETs in general? - are able to simulate effects of vacuum tubes, in technologies where that might be desirable (e.g. audio amplification), while others don't appear to have that ability. It's a quality which is probably not very helpful in some applications, but quite desirable in others.
Vacuum tubes and FETs use the same process - electron flow control via electric fields.
It's mainly to do with the transfer curve. Valves and FETs are both transconductance devices, ie, use voltage to control a current. But they have different curves of input voltage vs output current. They also saturate differently. It is possible, but not easy, to make a JFET have a curve very much like a signal valve. It requires careful biasing and usually adjustment of every device built.
It’s so weird that RCA is now just a brand that the owner doesn’t even use but licenses to various other manufacturers.
Ah, I miss Mos Burger...
Crazy how small the circle of people who made all this happen really is.
he really just mentioned POGs. i didnt know how happy this would make me, i feel so validated lol
Hmm... a few glaring technical errors, I believe
Your history is missing an important practical development, that happened in the mid 20-30 of the 20th century, an that is the invention and construction of the fist semiconductor electronic device by Dr. Thomas Henry Moray. Dr. Moray was showing and explaining his invention to Dr. Harvey Fletcher, in his several trips to Utah, who late became the head of the Bell Labs department where the "invention" of the transistor was pronounced 20 years after Dr. Moray' invention.
It's very interesting. It is almost like a solid state vacuum tube. The electric field controls the flow of electrons.
So when did transistors become woke and now IGBT transistors are all the rage?
The fact give us the name. The name had to comes from somewhere and now we knows where.
Great video, a very interesting historical guide to the field effect transistor. Thank you.
The self references in this videos are at much higher levels.
T. Henry Moray called it a valve.
Thanks, Jon. as always a great video!
Shockley's boom "Electrons and Holes" is still the best solid state physics book I've ever read. The MOS is quickly discussed as an idea in the book.
One click buying - "Are you sure you want to purchase this item?" Double click button to continue... lol
ten bucks says you ain't old enough for pogs, my man.
20:00 Wow, that reference lol
To your point about patentability, a patent only has to be non-obvious and novel. It does not have to be correct. It's hilarious to go from the rigors of academic writing to writing patent applications.
I can’t seem to find your electron microscope link on TH-cam. Can you post it? I’d love to see it!
th-cam.com/video/rvr5w5bbbj4/w-d-xo.html
Two months ago??? What?!
@@TheLegendaryHacker for the same reason, early access tier on Patreon! I’m part of a tier that gets some videos early but not all. My Patreon bill is also so high lol. The scanning microscope one got released eventually.
th-cam.com/video/rvr5w5bbbj4/w-d-xo.htmlsi=3BXx1SSjtcYkarNM
1:00 whoa my victorian dudes could have invented digital cameras in 1873 o.o
Not really. Manufacturing transistors (and other solid state devices, including CCD's) requires a significant support structure to exist. Note the first transistor section later in the vid.
No, but simple light detectors used selenium cells. You could build an analog TV with one of those scanning wheels.
I didn’t get CS/EE degree so I don’t know if they already have this, but if they don’t, universities should have semiconductor history courses.
If you want REAL drama, study the invention of the transistor at Bell Labs under Shockley. Prior to this endeavor, Shockley was reported to be a real good guy to work under. But after it, he lost his marbles, jealous as hell of Bardeen and Brattain who found that Shockley's original idea would not work, and then those two had the temerity to go off and create the device that did work (i.e., "demonstrated gain"). Shockley was never the same after that, eventually quit the Labs, and headed to the West Coast. This move eventually led to the creation of Fairchild Semiconductor where the planar transistor and integrated circuit (the REAL integrated circuit, not TI's Kilby transistor that was NEVER used to manufacture a single IC) were perfected.
I know the "point contact transistor" as a "cat whisker" from old ham books. cool stuff.
The cat whisker was a crystal diode.
Wasn’t it more to do with size than reliability? Vacuum tubes are very reliable, as long as there’s not a leaky capacitor(leaking DC) burning it up.
Vacuum tubes need a lot of power, those filaments need to be heated to emit electrons. Also not as reliable if you have thousands of them. Consider that the filament will slowly evaporate unless you drive at very low emission setting. Similar failure mode for light bulbs, which also have a filament but optimized for light emission not electron emission.
12:00 mixed up positive and negative voltage here.
Integrated circuitry amazes me.
I wonder, are the occurence of IC's just an inventive exploit stumbled upon by random chance, or a fundamental phenomenon hidden in our universe only waiting to be discovered and utilized?
Anyone can look at a simple IC architectural diagram and wonder how anything like it could have been concieved, much less crafted and brought functionally into existence.
Thanks for making this video!
Dont be messing with my electric vibes baby
Amazing video as always sir. Thank you.
Left us on a cliff hanger. What will evil ibm do with this dangerous tech
HELLO this are excellent in-depth videos that I just found and am very impressed ! Well done! I would love you to do a video on China's progress on digital memory. i.e DRAM & NAND. I know CXMT is the leader in DRAM, but there is v little info on how it's faring under export controls, as DRAM needs DUV and EUV lithography also. Can you do a video, or if not pls post links to this message where I can track CXMT's progress. thank you !
lol i think you mean positive voltage on gate at around 12 minute mark, right ?
Superb in all respects, as usual!!
Transistor means life.
Please make a video on photo molecular effect! Thanks!
I love it when you get technical.
So I guess the next episode will be about finfet and 3d transistors?
thats a history class
I was hoping to hear about the development of the RCA 40673 Dual-Gate MOSFET transistor, an excellent VHF amplifier transistor in its day.
No current flows through the gate. A bipolar amd threadripper would require a nuclear reactor to power. FETs require very little comparatively. My understanding is this revolutionized old hot mainframes with bjt chips.
I would like to hear more about HSM and attala e.t.c.
It's "in passing" from the French "en passant".
Sry for the noob qs, but why does a negative voltage attract electrons? Dont same charges repel?
But if you only use it to construct sine and cosine then you only need a bipolar zeener junction transistor to explain two simultaneous GREATER | LESS -> SWITCH CASE, so a gate array is a high level operation trained at *_fairchild budy pal_* for snake game on a 32MB PLA slice.
Shockley seems to have pissed off the right people. I’m curious now.
I remember when it was THE selling point.
12:08 Somehow I always thought, that negative voltage would repel electrons, but what do I know.
Yeah... about that... For historical reasons, that nobody bothered correcting, your intuitive idea is correct, but the established nomenclature is the opposite of what actually happens. Sorry. It bothers me too.
Negative does repel electrons.
@@andersjjensen yeah they guessed the electron charge wrong and by the time they actually found out what it really was it was too late to change everything to how things really are. So here we all are today with negatively charged electrons. Most of the time it doesn't really matter. But when it matters then it does.
24:18 Do they sell FETBurger there?