Talk:Integrated circuit

Are integrated circuits reconfigurable or not?

According to the article Programmable logic device, integrated circuits "consist of logic gates and have a fixed function." However, according to the article Field-programmable gate array, an FPGA is an "integrated circuit designed to be configured by a customer or a designer after manufacturing," which seems like a contradiction to the former. So, can an integrated circuit be configured (and reconfigured) after it has been manufactured or not? —Kri (talk) 07:18, 7 May 2019 (UTC)

I think the lead of Programmable logic device could be better written. It says, in part, "Unlike integrated circuits (IC) which consist of logic gates and have a fixed function a PLD has an undefined function at the time of manufacture."

I think what this passage really means is "Unlike other integrated circuits (IC) which consist of logic gates and have a fixed function, PLDs are a type of IC have an undefined function at the time of manufacture."

There is no doubt that some ICs are reconfigurable. Jc3s5h (talk) 11:49, 7 May 2019 (UTC)

Invention of IC

I am going significantly rewrite this section as there is several mistakes and exaggerations. The claim that "The monolithic integrated circuit chip was enabled by Mohamed M. Atalla's surface passivation process, which electrically stabilized silicon surfaces via thermal oxidation, making it possible to fabricate monolithic integrated circuit chips using silicon. This was the basis for the planar process, developed by Jean Hoerni at Fairchild Semiconductor in early 1959, which was critical to the invention of the monolithic integrated circuit chip", is exaggeration. Arjun Saxena in his book Invention on Integrated circuit, say that surface passivation was one of several factors that contributed to Hoerni's invention of planar process(page 95-102), but he did not consider it critical at all. Same with Bo Lojek's History of Semiconductor engineering, Atalla is briefly mentioned in his book, most of the information is on Hoerni. The claim appers to be based on one sentence remark by Sah. I am going to rewrite it according to Saxena.

The claim " Atalla's surface passivation process isolated individual diodes and transistors,[11] which was extended to independent transistors on a single piece of silicon by Kurt Lehovec at Sprague Electric in 1959" appeared be OR. I cant find anything about influence of passivation process on Lehovec.

"Atalla first proposed the concept of the MOS integrated circuit (MOS IC) chip in 1960, noting that the MOSFET's ease of fabrication made it useful for integrated circuits"- again this is wrong. Moskowitz says that Atallah, after proposing MOS transistor noted that it will be useful in IC as it is easier to manufacture, that's not the same as proposing "concept" of MOS IC. Ross Basset in his Book To The Digital Age,eplain this in details(page 28). Here is the quaote : "Except for a few special applications, Atalla and Kahng’s device would be useable only within a subset of the design space covered by the silicon bipolar device. Its main advantage, ease of fabrication, had little relevance to the industry at the time. To call Atalla and Kahng’s device an invention was almost a contradiction in terms, for it was inferior by every relevant standard.39 The one area in which Kahng and Atalla recognized their device might be advantageous was of no interest to Bell Labs. Kahng mentioned that the device would be suitable for integrated circuits". DMKR2005 (talk) 22:43, 3 May 2021 (UTC)

Made practical by MOS

The article lead claims "Integrated circuits were made practical by technological advancements in metal–oxide–silicon (MOS) semiconductor device fabrication." This is pretty much false. TTL and other technologies were commercially available as ICs before CMOS ICs. TTL continued in wide use even when CMOS became available because of its speed advantage. CMOS was preferred at first only in low speed applications where it had other advantages such as low power consumption and high fan-in/fan-out. Perhaps what is meant is that CMOS made practical single chip processors and other VLSI applications that could fit in a small package without getting hot enough to fry bacon on. SpinningSpark 15:39, 12 February 2022 (UTC)

I agree, bipolar transistor integrated circuits were both practical and important for at least a decade before MOS became important. Jc3s5h (talk) 17:37, 12 February 2022 (UTC)

But by that argument, Bit slicing would have been much more popular, which didn't happen. --Ancheta Wis   (talk | contribs) 01:08, 14 February 2022 (UTC)

That's a non-sequiter. You'll need to explain what you mean by that. It doesn't follow from anything I said. The only argument I have made is that it is false that MOS made ICs practical. How can it be concluded from that that bit slicing should have been more popular? SpinningSpark 12:39, 14 February 2022 (UTC)

There would have been more applications of devices like Texas Instruments SBP0400, 74181 ... but they were superseded. Even minicomputer companies stopped using them. --Ancheta Wis   (talk | contribs) 12:49, 14 February 2022 (UTC)

It was't Fairchild's emitter coupled logic family, it was RCA's COS/MOS that won out. CMOS only consumes power on the transitions between states (except for any leakage current). --Ancheta Wis   (talk | contribs) 13:41, 14 February 2022 (UTC)

I'm still not getting your point. Being superseded does not mean that there were no practical ICs before MOS. It is not in any way a justification for keeping an untrue statement in the article. SpinningSpark 13:43, 14 February 2022 (UTC)

Microprocessors were the battleground; by upping the clock, ECL consumed more power, where COS/MOS consumes power only during the rise/fall times of the clock pulses (in the times between transitions of voltage levels). So microprocessors could just expand the number and scope of their applications, because they could exploit Moore's law by upping the clock, and by decreasing their feature size, and by increasing wafer size. Otherwise, ICs would only have remained part of the glue logic of circuits; higher levels of integration (VLSI) would not have been warranted by remaining implemented as bipolar transistor circuits alone. But semiconductor memory chips would have just have consumed more power at higher clock rates, if memory chips had remained implemented as bipolar transistors only. Intel understood this, and planned to make their money on MOS memory chips, before microprocessors were invented. --Ancheta Wis   (talk | contribs) 14:25, 14 February 2022 (UTC)

You continue to argue against a point I haven't made. I don't disagree with what you say. You are beating up a strawman. The sentence I highlighted still remains untrue. This article is about ICs, not specifically uprocessors or VLSI. I already said in my opening comment CMOS made VLSI practical. But there were practical ICs prior to CMOS and prior to VLSI. SpinningSpark 14:37, 14 February 2022 (UTC)

So how about "Integrated circuits were made practical ubiquitous..." --Ancheta Wis   (talk | contribs) 14:42, 14 February 2022 (UTC)

Better to say that "Large-scale integration was made practical..." Ubiquitous is a bit vague, and of course there remained many practical applications in 1960s electronics where discrete transistors were more appropriate. SpinningSpark 15:01, 14 February 2022 (UTC)

Spinningspark clearly very large scale integration was made practical by CMOS. But up through the late 1980s and first few years of the 1990s bipolar integrated circuits were built by IBM for their System/390 systems with over 30,000 transistors; the threshold of LSI is typically considered to be about 10,000 transistors. Jc3s5h (talk) 20:10, 14 February 2022 (UTC)