How LEDs are made (2014) | Hacker News Reader

How LEDs are made (2014)

learn.sparkfun.com

149 pointssmig021 comments

Man, I miss photo articles like this that I can read at my leisure, without sound. Nowadays this would likely be a (probably frantic) video.

A non-frantic (related) video worth the watch:

Why It Was Almost Impossible to Make the Blue LED

https://www.youtube.com/watch?v=AF8d72mA41M

I remember reading about the Blue LED when it first started to appear. It was interesting reading what they thought how it would be used at the time with being able to do true R/G/B the thing everyone was talking about. Not sure how much later before they were used for shiny round discs but it wasn't part of the article's prognostications. This is all a bit nostalgic as I read about that in a printed magazine when those were still a thing.

Back at Caltech, one of the students realized that the only thing limiting the brightness of an LED was heat dissipation. So, he dipped an LED into liquid nitrogen, and cranked up the current. It got pretty bright before it melted.

Naturally, he realized that the clear plastic blob it was inside was an insulator. How to fix - he filed it down to the bare minimum that would hold it together. This time, it would light up a whole room!

Liquid nitrogen is all one needs to make bright LEDs.

People dedome their LEDs regularly these days for a different colour tint and beam profile. I never considered what it did to the thermal profile.

>... to make bright WHITEish LEDs.

While I'll readily admit to remembering nothing from her class, I took physical chemistry [πchem] from one of the co-inventors [2] of white LEDs. This is where my own limited fleshtelligence began searching for Heisenberg's god...

[πchem] e.g: how metals behave when struck with electron[-like thing]s)

[2] ONE of two known-methods, then

----

But yes, once any process exists, it's usually only a matter of heat management to keep it working full wall-slam-ed-ly [ƒpu]

[ƒpu] which is why to run GPU fanspeeds high enough to keep <65°C – don't care about the noisiness if they'll then last forever; change your car's oil (and keep topped-up)

> Back at Caltech, one of the students realized that the only thing limiting the brightness of an LED was heat dissipation

It takes studying at Caltech to realize semiconductors output are limited by their heat generation? I thought everyone knew this.

The average person knows literally nothing about semiconductors. I would say it’s not very intuitive how diodes work, unless you’re an electrical engineer.

Having taken a course in the quantum mechanics of semiconductors, I can vouch for anything involving QM is not intuitive at all.

"I think I can safely say that nobody understands quantum mechanics" -- Richard Feynman

1. Fascinating overall

2. "they can align over 80 per minute or about 40,000 per day." - terrifying, as I assume this is a metric workers are held against :O

80 per minute is less than a second for what sounds like several movements - move the die over, align, push down, move it out. While your eye is stuck to the microscope.

For context this is a 12 year old article about an outdated factory before LED die bonders got cheaper. The humans are working as glorified pick and place machines doing very repetitive motions, not manually aligning each die through a microscope. This only works because the tolerances on the placement between the die and anode/cathode are huge and the surface tension of the adhesive does most of the work.

Neat article.

I'd love to know more about how candle-flicker LEDs are assembled, because that source of [apparent] randomness is very interesting to me. I'm not sure if it's an LFSR or true HRNG, and I'm sure there are lots of different designs out there for the simulcrum of natural candle light.

You can get a better sense of their operation if you wire up the LED to an audio circuit where they'll make a pleasingly happy beep boop sound.

(Edit: There was an article somewhere that explored the entropy and concluded that their component operated on a LFSR, as they binned all the brightness outputs into integer values and waves hands did fancy math to conclude that the brightness it was likely modulated by a LFSR. I'll see if I can find it.)

(Edit 2: https://cpldcpu.com/2013/12/08/hacking-a-candleflicker-led/ here's that article for those interested. https://news.ycombinator.com/item?id=25530895 was the original submission to HN.)

De-doming these things (as discussed in another comment) is quite a chore; I de-domed 30 LEDs (candle-flicker, of course) in order to diffuse the light and fit under the keys of a small 3x10 keyboard I was building. But the effect is neat when the backlight is on as it almost looks like a shadow is randomly typing away as the entire array flickers.

I guess I had read this article 12-13 years back. I think it was this same article.

One of the things I vaguely remember was reading somewhere that working on this LED manufacturing severely damages the workers' eyes. I don't know how much of it is true and if it is, whether that is still the case.

So every LED die is manually aligned?

Surely 10 years on that isn't true anymore??

no longer true indeed. comment a bit above mentions more modern automated alignments. this and things like blue leds coming down in price

I doubt it would even be true 10 years ago. This article is from small-mid scale led production.

China has improved a lot in commerical, high power LED. 10 years ago, they could not even touch the performance of CREE or Luxeon or Osram LEDs, now thay are on par in term of performace, and much cheaper.

Neat, I was expecting more about how the semiconductor part is made. I toured the Lumileds/Phillips fab that closed in San Jose but you can't really see much.

You could probably see more if they used their own product.

This was a lot lower-tech than I was expecting. Very cool!

I wonder why some leds have a high TDP and if even that it is efficienty and how it could be fixed...

Reply on news.ycombinator.com