Talk:Transistor

Attributed to Lincoln

"If you call a tail a leg, how many legs does a dog have? Four. A tail is not a leg." — Preceding unsigned comment added by Wtshymanski (talk • contribs)

I presume that you are talking about discrete transistors on die. I searched several semiconductor websites (not an exhaustive search by any measure) and did not find any. All the discretes were in packages. The only thing I found on die were integrated circuits. When I think of a discrete transistor, I mean a transistor with a collector, base and emitter and nothing else. A single transistor on a die would have, I think, a substrate. I'm leaning to saying that is not discrete. Constant314 (talk) 23:07, 21 May 2018 (UTC)

No he is talking about discrete transistor dies. He is confusing a die diced from a wafer having one transistor with an monolithic integrated circut die diced from a wafer and having many transistors. A die as a single transistor has the exposed discrete three connections you expect (emitter, base & collector for bipolar or source, drain & gate for MOS). Such die are typically connected by solder bumps. Whereas none of the many transistors in a monolithic IC ever exposes the three discrete connectors you expect, they are inconnected as part of the deposition process and are never discrete from a connection perspective. We don't build much these days with such discrete transistors but consider IBM SLT as one example - note the usage of the term "discrete transistors" in Steps in manufacturing Solid Logic Technology hybrid wafers. The reason yr search didn't turn up any references is probably because there are no standards for transistor dies but they were used heavily in the 1960s and are still are available, see e.g. & wafer. When a wafer is diced into individual transistors, they are discrete, it doesn't matter how they are subsequently packaged. Tom94022 (talk) 00:31, 22 May 2018 (UTC)

Of course I'm so stupid that I can't tell the difference between a single unpackaged transistor on a die and an integrated circuit with multiple transistors on a die. This is why I come to Wikipedia, to learn. In any case, so long as the System 360 isn't on the list of discrete transistor computers, I'm content. --Wtshymanski (talk) 02:10, 24 May 2018 (UTC)

Singular and plural die and dice

Unfortunately, both terms are sometimes used both as singular and plural, though dice as singular is just wrong except when referring to a game; a pair of dice is two, while each one is a die. In books, "cut into die" is not a thing, while "cut into dice" is common. In reference to wafers, see this search, and this search (sometimes it's "wafer into die chips" or "into die sized pieces" just because die is so awkward as a plural).

An anon has been reverting with the claim that dice is singular and die is plural. That's just backwards. A better argument for reverting dice to die would be that die is preferred in this industry; but that's questionable. Dicklyon (talk) 14:02, 23 May 2018 (UTC)

In my experience, "dice" can also be a verb: "dice the wafer". See this patent, within the browser, search on "DICE AND PICK". The phrase refers to the operation where, after the wafer was tested, the wafer is sawn into dice (or diced) while attached to a plastic adhesive-coated sheet. Then a machine consults the wafer map, which shows which chips passed, and picks the good chips off the plastic sheet for further processing. The remaining chips are discarded. Jc3s5h (talk) 14:16, 23 May 2018 (UTC)

IEEE Std. 100 (7th ed), owing to some sloppy proofreading, is not our friend here. The entry for "Die" says in part "...Note: this is the plural form of die. ... See dice". The entry for "dice" syas in part "...Note:this is the plural form of die." Perfectly ambiguous. And authoritative, it says so right on the frontispiece. --Wtshymanski (talk) 02:06, 24 May 2018 (UTC)

As far as the integrated circuit usage is concerned, Googling "Integrated Circuit Dice" turns up only three hits (ignoring Wikipedia) where all three words are present, and they are all links to the company 'Danube Integrated Circuit Engineering' (DICE). However: Googling "Integrated circuit die" turns up pages of hits for the subject of this discussion. On this basis there does not seem to be any on-line evidence supporting the notion that integrated circuits are constructed on dice rather than die. 86.149.136.154 (talk) 15:28, 25 May 2018 (UTC)

Counterexample.] Jc3s5h (talk) 15:31, 25 May 2018 (UTC)

Well found. I would make the observation that a patent is written by an individual (usually a patent clerk) and is not therefore an authoritative source on correct word usage in any context. In view that the industry generally uses 'die', I would suggest that WP:COMMONNAME suggests that we stick with it. 86.149.136.154 (talk) 17:30, 25 May 2018 (UTC)

I feel judging by the number of Google hits can be very misleading. When a series of words are put in quotes, they must all occur in the order given, which greatly restricts the number of hits. I hypothesize that publications that delve deeply enough into integrated circuits to use the word "die" or "dice" would often not use the phrase "integrated circuit", but instead would use shorter words like wafer, chip, device, IC, etc. If instead you google for the words silicon dice you get lots of hits; several on the first two pages support dice as the plural of die. Jc3s5h (talk) 17:47, 25 May 2018 (UTC)

Sorry to have gotten us into this semantics issue but as it turns out the IEEE dictionary definition partially quoted about is actually is quite helpful:

• "die (1) A single piece of silicon that contains one or more circuits and is or will be packaged as a unit. ..."
• "dice Multiple pieces of silicon, each of which contains one or more circuits and is or will be packaged as a unit. ..."

This supports a die/dice usage. Die as singular is also the SIA definition. Personally I have always used dies as the plural but can live with die/dice. Tom94022 (talk) 18:13, 25 May 2018 (UTC)

Evidently, only if it's made of silicon. Jimbo help us describing dies made of GaAs, SiC, germanium, or other materials. It's a very poorly edited dictionary entry. ==Wtshymanski (talk) 19:55, 25 May 2018 (UTC)

Transistor aging compared to a vacuum tube

Is it really appropriate to include aging as a limitation in a section comparing them to tubes? In isolation it would be a valid concern but compared against a tube with almost infintiely worse long term reliability it becomes an advantage rather than a limitation. I was tempted to simply snip that on sight but I figure I'll raise it here first, I know people can be hesitant to remove sourced material even if as here the source does not directly support the assertion made - the source does not even mention tubes. 3142 (talk) 13:55, 7 January 2019 (UTC)

I used to work for a semiconductor manufacturer. One of my duties was to waive failures in the accelerated aging tests, provided a reason for the premature failure could be found that would not occur in real products. I agree the bullet about aging of transistors should be removed. Jc3s5h (talk) 15:28, 7 January 2019 (UTC)