Years ago I answered a "for sale" newspaper ad for used office desks and some test equipment. I showed up at the address and it was a small dingy old building and most of the employees seemed to be 55+ year-old women. When I asked what they did and was told they made diodes, you could have knocked me over with a feather. Like he did, I assumed that all electronics parts manufacturing was high-tech, cleanroom work, etc. But here were little old ladies putting pieces together.
I had a similar epiphany years later when a friend told me he was buying a thermocouple manufacturing business for $20,000!!! The business consisted of two senior citizens who wanted to retire. No high-speed pick&place robots working in inert argon atmospheres, just two old people putting wire in jigs and spot-welding them together (probably in an inert atmosphere, though).
The same thing repeats itself over and over: a huge amount of what we think of as sophisticated technology is being done by hand, or by ancient machinery, in dirty, poorly lit corners of the US.
I ised to work in electronic subcontract engineering. They had some automation (little pick and place machine with oven and screen printer; a flow oven for populated PCBs) but there was a lot of work done by hand.
It was a terrible time of my life and it informs a lot of my opinion about poor management and inefficient working and about ISO900x quality assurance systems.
Passing these management craps are not completely meaningless. Sometimes, people do realise what was the problem during the process of producing these management craps.
Clean rooms are very old technology, they are made at least since photography exists. And, of course, you only need them if the features you want to press are very small (and a better one the smaller feature size), big components can go without.
I wouldn't make leds at home, but that's because of the chemistry, not because it can't be done...
This is why we need more posts like these and more of those How things are made blogs. Too often people don't even begin getting into (high) tech because they think it's too complicated or expensive.
The title here is slightly misleading- this isn't how LEDs are made, it is how LED dies are packaged. The more difficult part of all of this is creating the dies themselves, which is requires a clean room and all the automated tools I'm sure most people here were expecting to see.
There are so many more parts to making an LED, too. First you need the wafer that it's grown on - either silicon carbide or alumina. The latter of which requires cones to be etched down onto the growing surface which is a significant manufacturing process all on its own. The former of which requires heating SiC until it metls at 2700C and slowly pulling out a single perfect crystal cylindrical ingot.
Then you have the actual crystal growth portion where you shove all those wafers into an evacuated chamber, heat them to a thousand degrees, spin them around, and spray them toxic pyrophoric gases. This process takes multiple hours to grow just a few microns of light producing crystals.
Then you have to grind the wafers down to the appropriate die thickness, go through multiple stages of etching, add a phosphor for white LEDs, deposit gold and ITO for electrical contacts, and use lasers to cut them into individual die. Then, finally, you can ship them to the people in the linked article. Honestly, this article is the most boring part of the whole process.
Source: I used to work in R&D for an LED manufacturer.
I'm surprised by how manual the whole process is. I think labour is so expensive in the west that any manufacturing of small low value items is incredibly highly automated (see youtube for lots of videos of how huge automated production lines work). There's no way you could employ someone to manually bond 80 LEDs a minute in the west and still have a competitive product.
I'm not so sure. Let's assume you're paying someone $15/hour and she actually costs you $25/hour. Assuming a 10 minute break every hour (that's gotta be tiring work and you want to hold onto your employees), then that's 2500/(80 * 50) = 0.625 cents of labor cost per LED. I don't know what all the other costs are, but that one operation seems surprisingly cheap.
Just looking on Digikey, the unit price for a 5mm LED is somewhere between 3 and 4 cents. In that price there's distributor markup, transport, supply chain markup, the cost of the components and the cost of labour. I'm not too familiar with the economics of electronic components, but I'd be very surprised if the manufacturer sees more than 30% of the final price.
That would mean the manufacturer would be spending half the cost of the finished product on a single step of the manufacturing process. I'd be surprised if that's economically viable.
I used to work as a lab tech at one of the leading US LED manufacturers, and automated tools most definitely exist for all the steps pictured here (and they were typically done in a cleanroom, as well). Can't speak to what was done in China, though.
I've been watching "how it's made" tv series with my kids.
That has been an eye opener on where there are lots of people doing things I assumed (ignorantly) would be automated
Helps me be less of a tight ass too knowing how much manual labour is in some things!
It's at least a semi-automated bonder, so once you get the step and repeat tuned in, it's usually a matter of just pushing the button and visually verifying. Still though, looking into a microscope 12 hours/day is tedious.
Have you used an industrial quality binocular microscope? I do - though not continuously - and they're not at all unpleasant. The eyepieces are individually adjustable, as is the interocular spacing.
We should make a raspberry powered microscope that can project onto a 22" HD display. It would be awesome if Lytro had a realtime USB version of their camera.
Video microscopes work OK for inspection but for working under the microscope, the video delay and the lack of stereo vision are pretty intolerable. What would probably work better is a Vision Engineering eyepieceless stereo scope.
Sounds good for a second, and then you realize that it's going to induce a conflict between your visual and proprioceptive senses that'll make it impossible to work at best, and probably make you toss up your lunch besides.
Needs to be steroscopic or else you'd be nailing bonds down in the wrong location. No substitute for a good stereo microscope for microelectronics work.
Maybe, but then you've got the lag and pixilation. Seems to be a solution in search of a problem. Still nothing nicer than human eyes looking through good glass of a stereo inspection microscope. Really not even the eyes, but the same posture for hours on end hurting the neck muscles. I suppose you can get used to it, but with a wire bonder there is not much leeway in adjusting the microscope position. Then again, for a guy manually wire bonding LEDs in China, it beats toiling in a rice patty.
Even the USB microscope cameras are garbage. Much better pictures just holding a digital pocket camera up to the eyepiece.
Very surprising how manual this is, unless this is just for the purpose of showing how the LEDs are made. I would have expected everything in this process to be automated since electronic component manufacturing is very high volume.
Also amazing how small the components (LED dies) are.
Everyone else has moved on to 0402-sized surface mount LEDs a long time ago. Could this manual process just be because nobody is building the old through-hole LEDs in massive quantities?
You're probably right. However there is a big market for quality through hole LEDs in non-mass produced consumer crap (read industrial electronics and test gear for example) so some of the big guys still produce. However when someone picks out a BOM for a production run from Farnell and there's a 0.05 no brand LED and a 0.25 Kingbright through hole, the no brand wins every time.
(says me who just bought 200 for a small prod run). So that's £10 or £50 out of the budget. Thanks to cheap LED guys, I can take my family out for that extra £40 :)
Ha ha, try hand placing 01005. Literally specs of black pepper. Must use ceramic tweezers due to residual magnetism in even the highest quality, demagnetized stainless. I was placing them between BGA balls to solve a noise issue.
The lady I have hand building my boards can place 0201 all day. Pretty amazing.
The company I ised to work for would sometimes buy components from RS. They'd supply them in strips of 10, 50, or 100.
The pick and place machine needed blank strip of about ten components (just for lead-in on the cassete).
Transfering components from one tape to another was really annoying. And one of the bosses just did not understand that the fractions of a penny saved were obliterated by time wasted moving components over.
And until you've had to do something like that you can't really understand the benefit of $60 Swedish tweezers.
If you look closely, you'll see it's actually in the shape of a reflector - usually a cone. Also not shown is that some higher-power leds use a blob of silicone around the chip to protect the bond wire during thermal expansion -- the cone helps hold the silicone while curing.
This is clearly an old process - when was the last time you saw a 7-segment display, or those big red round thru-hole LEDs, outside of hobby electronics?
I'd be very curious to see how this process differs from the high-power lighting LED manufacturing process (e.g. CREE), which is a more modern technology by several decades. I would guess there is much more of the automation people are expecting...
"when was the last time you saw a 7-segment display, or those big red round thru-hole LEDs"
Every HP common slot server PSU has one of those round through hole LEDs on it. I have an HP server sitting here (DL580) that has two 7 segment displays on its motherboard. These things are definitely still used in modern products.
>This is clearly an old process - when was the last time you saw a 7-segment display, or those big red round thru-hole LEDs, outside of hobby electronics?
Remember folks, there is a world outside of Silicon Valley.
"when was the last time you saw a 7-segment display, or those big red round thru-hole LEDs"
At work yesterday, at a microwave communications equipment manufacturer. Through-hole LEDs are the easiest kind to mount on the front panel of rack mount equipment. And there are 7-segment displays on two of our adjustable power supplies.
And so this makes an interesting point, LED shapes are controlled by the mold makers, but when do we get just a flat thin LED and you can 3D print the shape you want on top? :-)
Make a trip to your local Michael's or other art/craft store and you can do it today. Silicon sealant to make a mold + liquid molding acrylic and you're off.
Same as regular chips: in a grid on a wafer, then cut and pick&placed.
Don't forget: yield goes up when the die shrinks because the defects are typically small spots. The smaller the die the better you pixelate the defects.
The wafers are only 2 inch diameter to avoid yield loss due to edge effects: at the edge of the wafer you have lowest quality components (optical ring effects)
I doubt that they 2-inch wafers to avoid yield loss from edge effects. The ratio of area to perimeter rises as you increase the wafer size. Perhaps the reasons they use 2-inch wafers are for better uniformity, better flexibility, or lower capital investment.
Your reasoning for the 2" wafer surprises me because most semiconductor work scales well, and the cost to buy and process larger wafers is only slightly more expensive while the surface area is much larger. A 6" wafer is 9 times as big as a 2" one.
A 6" wafer will also have lots of dies on the edge that don't pass automatic testing. Don't forget that most devices are located near the edge.
Secondly, upgrading your Fab to switch your process to larger wafer size has a 10-figure $ price tag. It's not trivial for digital, let alone opto electronics. LEDs are special in that they deviate severely from the standard mos process. When Monsanto first produced them they almost dumped the idea for LEDs because of all the issues to deal with the exotic materials.
> A 6" wafer will also have lots of dies on the edge that don't pass automatic testing. Don't forget that most devices are located near the edge.
I don't understand this. Given a fixed size die, the number of dies near the edge would go up linearly as the wafer size goes up. The number of interior dies would go up quadratically.
Yeah, what he/she is saying doesn't really make sense. Perhaps he/she means that there are problems with uniformity and the dies on the edge suffer most, not because they are next to an edge, but just because they are further from the center. In that case 2-inch wafers may make more sense than 6-inch wafers.
These are not the type of LEDs used in modern light LED light bulbs. They produce a tiny amount of light (maybe 5 lumens each, with a 60W equivalent requiring 800 lumens), don't work well with a heat sink, and aren't particularly efficient. Separately, a lot more goes into an LED light bulb than just the LED.
Home Depot has been selling those lamps for $4.97ea for quite a few months now. They're a very nice replacement for the traditional A19 60W incandescents.
My only complaint, and it's a tiny one, is that they only dim to about 10% before cutting off entirely. There's also a small dark spot on the spot of the bulb opposite the socket, but in practice that's been a non-issue for all my use cases.
In the US, the price varies state by state, and reflects a variety of subsidies used to encourage the switch over from incandescents. Thus for certain bulbs (like this one) the production cost is often higher than the selling price. Some details for California here: http://www.designingwithleds.com/rebated-cree-philips-led-bu...
Led light bulbs are considerably more intense. It has to contain a power supply, heat sink, and more LEDs than just the one that gets put into the single led, and all that needs to be interconnected on a chip.
The book referenced as the first source on the "die" wikipedia link claims:
> The plural of die is dice. However, it has become standard practice in the industry to use "die" as the plural.
As a native speaker of English, I'd find any of "die", "dies", or "dice" to be pretty reasonable as long as it's clear what's referred to from context. (I'd prefer "dies" over "dice" if it's not immediately clear that you're not talking about playing dice, for example.)
Interestingly enough, organic LED (OLED) technology uses dyes instead of semiconductor materials. OLED display panels can be 'printed' like ink-jet printers. http://en.wikipedia.org/wiki/OLED
Years ago I answered a "for sale" newspaper ad for used office desks and some test equipment. I showed up at the address and it was a small dingy old building and most of the employees seemed to be 55+ year-old women. When I asked what they did and was told they made diodes, you could have knocked me over with a feather. Like he did, I assumed that all electronics parts manufacturing was high-tech, cleanroom work, etc. But here were little old ladies putting pieces together.
I had a similar epiphany years later when a friend told me he was buying a thermocouple manufacturing business for $20,000!!! The business consisted of two senior citizens who wanted to retire. No high-speed pick&place robots working in inert argon atmospheres, just two old people putting wire in jigs and spot-welding them together (probably in an inert atmosphere, though).
The same thing repeats itself over and over: a huge amount of what we think of as sophisticated technology is being done by hand, or by ancient machinery, in dirty, poorly lit corners of the US.