/sci/ - Science & Math

>>11119877
Good luck making rectangular monocrystals

because you touch yourself at night

>>11119887
how hard can it be? just use a cast or something

https://en.wikipedia.org/wiki/Czochralski_process

They are cut from pure silicon ingots which happens to be a cost effective way to produce pure enough silicon for semiconductor substrates.

>>11119905
>invented the method in 1915
seriously? in 100 years we couldn't find a better method for a such simple thing?

Maybe it's just easier to do circles, for the same reason it's easier to make a circular pancake. You ever try making a square pancake that are lab grade?

>>11119914
Well the process is more refined now than 100 years ago but the basic principles are the same. Any excess silicon from the cutting process is just reused to grow another large crystal.

gravity

>>11119892
For what purpose? The sharp points might act as another core of crystalization.

>>11119914
much like the wheel and oldie but a goodie

>>11119914
transistors were invented over half a century ago
>durr but we couldnt think of something better???
the process isnt identical to how it was back then dumbfuck

>>11119892
/sci/ in a nutshell.

>>11119892
Then it will crystallize at different places and you will have grains instead of monocrystal

>>11119887
[de-ices ur path]

>>11119877
the big dick silicon cilinder is produced inside a (very) high pressure furnace and grown into an approximately cylindrical shape
in order to manufacture the square monocristalline solar cells the sides of the cylinder are cut off and wasted
in chip fabs they don't waste this material

>>11119887
Someone enlighten my why they need to be single crystals in the first place.

>>11120449
So that you have an uniform structure for the wafers

>>11120479
Right but does that matter for the performance of the chips? Or could they'd be made polycrystals?

>>11120486
It's absolutely critical, if you want to make any sort of functional logical circuits on nanoscale. Building it on polycrystals would be like building an astronomical observatory in middle of Las Vegas.

>>11120495
Not sure if I understand the analogy. I mean, do polycrystals conduct current differently?

>>11120510
It's kinda bad analogy, tbeh. Imagine printing patterns on dirty paper vs. clean paper, when you want the end results to be as uniform as possible.

>I mean, do polycrystals conduct current differently?
They do, but that doesn't matter. Poly are a fucking mess. You need uniform ground, because you are building a ridicolously complex, ridicolously small network for signal processing. Is there a grain boundary where you want a feature to be? Chip won't function.

>>11119914
That's like shiting on computers because we used to have abaci.

>>11120510
Which do you think is a better conductor?
They use monocrystals because you can alter the conductance of the wafer by doping it, if you were already using something which is "pre-doped", you would have to work around those structural defects.

>>11120591
Conduction doesn't matter. You aren't using it as a bulk conductor.

>>11120603
It's more about the aspect of "tabula rasa" I was trying to get across.
You want something with as little structural defects to start, such as the grain boundries, so you have more control other the end product through doping, rather than hoping.

>>11119877
They grow the silicon crystal by dipping a wick into liquid silicon, like the way the make candles, and then they slice the wafers from the candle which is called a "bool."

>>11120603
hahaha, oh wow
i see now, the problem you know literally fuck all about how these devices are built

the way transistors are made in silicon is regions of the material are doped and then connected by metal or polysilicon tracks laid on top.
so, the material is already there, you "stencil" the transistors into the silicon using chemicals, masks, etc.

now, go back and look at the image you replied to.
which transistors will work more predictably, allowing you to develop and replicate the same circuit multiple times throughout the wafer?
transistors created in the "single crystal" image? or those formed in "polycrystal" or "amorphous solid" with random holes and cracks everywhere?

as an analogy, imagine drawing a picture on a new, flat, clean sheet of paper.
then, imagine trying to draw the same picture on paper that had been crumpled up hard and then smoothed out.

>>11120603
>Conduction doesn't matter.
also, transistors conduct current, so yes it does
even more so when the transistor width is measured in nanometers

>>11119877

They are round because that is the shape they are made.

As the wafers get larger the wastage goes down.

>>11119923
>a square pancake
You mean a waffle?

>>11119877
made from eggos

>>11119914
We still use water in our power plants for energy production and wheels for our transport constructions. This spieces is DOOMED!

What's more interesting is why the wafer has flat edge on one side and that's to orient the the crystal structure for fabrication. At least I found that more interesting.

>>11119914
And we still use steam for energy production

>>11121610
>they are round because they are round

There are so many silicon wafers in the world (because of the tech revolution) that it becomes very heavy to transport them. They are circular so they are easier to transport.

>>11122242
what in the fuck are you talking about

>>11122295
We've proved optimal circle packing, not optimal rectangle packing you idiot.

>>11119888
trips confirm

>>11121592
You aren't using it as a bulk conductor. Uniformity of crystal structure is the big issue issue.

>>11122430
High quality b8

>>11119892
>>11119914
>>11119887
>>11119877
https://en.wikipedia.org/wiki/Shaping_processes_in_crystal_growth
problem is, heterogenity in thermal profile, convection currents and issues with thermal conductivity mean you can only make relatively thin rods, ribbons or tubes
if you tried going thicker than about a millimeter, the inside would be liquid while the outside solidified, which would crack as it cools down
but on top of that, there are more irregularities in crystal lattice using methods like these, rather than good'ol czochralski
it has it's uses though, for example, monocrystaline capillary from pure synthetic sapphire is pretty much the hardest, chemically and thermally resistant tubing you can have
>>11119888
and also this
>>11119923
anon, I think you are confusing waffles with wafers