Silicon Chip Images
Moderators: rjlittlefield, ChrisR, Chris S., Pau
- Charles Krebs
- Posts: 5865
- Joined: Tue Aug 01, 2006 8:02 pm
- Location: Issaquah, WA USA
- Contact:
Chris,
The pads are on 100 micron pitch I believe (at home now and can't verify by asking someone). We have some that were on 75 micron pitch, but don't think these are images of those chips.
The pads are on 100 micron pitch I believe (at home now and can't verify by asking someone). We have some that were on 75 micron pitch, but don't think these are images of those chips.
Last edited by mawyatt on Fri Dec 19, 2014 8:27 am, edited 1 time in total.
One of my colleagues saw this thread and suggested I provide more details about what these images represent for folks that may not be familiar with chip design and processing.
First off, these are of silicon chips with minimum feature sizes of 130, 90 and 65nm (nm means nanometer). The 65nm chips have the bright colors. Some are Complementary Metal Oxide Semiconductors (CMOS), others are Silicon Germanium Bipolar CMOS, or BiCMOS. The transistors are so tiny and fast that millions can fit on a single chip less than 10mm on a side, and can switch states in a few picoseconds!
There are multiple levels of interconnect metal, up to 11 different layers in some chips. If you look closely you can see down into the chip surface, below the first layers of metal and dielectric insulators. These insulator layers are so thin you can see through them!!
The design of these chips is very involved and can take a team of engineers years to complete utilizing very specialized and expensive software.
Chip fabrication involves some of the most sophisticated and complex processes on the planet. Making the masks that are used in creating the multiple layers and controlling the dopants is a science all in itself. Some of these chips require 38 individual mask sets. Mskmaking requires some of the most powerful supercomputers in the world working 24/7/365 computing the patterns for the masks, they are that complicated.
Anyway, that's at little insight into these chips & their images.
Cheers,
Mike
First off, these are of silicon chips with minimum feature sizes of 130, 90 and 65nm (nm means nanometer). The 65nm chips have the bright colors. Some are Complementary Metal Oxide Semiconductors (CMOS), others are Silicon Germanium Bipolar CMOS, or BiCMOS. The transistors are so tiny and fast that millions can fit on a single chip less than 10mm on a side, and can switch states in a few picoseconds!
There are multiple levels of interconnect metal, up to 11 different layers in some chips. If you look closely you can see down into the chip surface, below the first layers of metal and dielectric insulators. These insulator layers are so thin you can see through them!!
The design of these chips is very involved and can take a team of engineers years to complete utilizing very specialized and expensive software.
Chip fabrication involves some of the most sophisticated and complex processes on the planet. Making the masks that are used in creating the multiple layers and controlling the dopants is a science all in itself. Some of these chips require 38 individual mask sets. Mskmaking requires some of the most powerful supercomputers in the world working 24/7/365 computing the patterns for the masks, they are that complicated.
Anyway, that's at little insight into these chips & their images.
Cheers,
Mike
-
- Posts: 215
- Joined: Sun Jul 22, 2012 2:30 pm
- Location: Bedford UK
- rjlittlefield
- Site Admin
- Posts: 23626
- Joined: Tue Aug 01, 2006 8:34 am
- Location: Richland, Washington State, USA
- Contact:
If I recall correctly, masks are now computed as a sort of "inverse problem" that figures out what the mask has to be, so that when the mask is run through the imaging process to make a resist layer, and then the resist is used for etching, the final structure ends up being what's actually desired. It is a very computationally intense process that involves a lot of trial configurations, each followed by forward simulation to predict what would be produced.Mskmaking requires some of the most powerful supercomputers in the world working 24/7/365 computing the patterns for the masks, they are that complicated.
--Rik
Rik,
Exactly. The mask set may require 2 or 3 individual masks to get the desired result. This is working the the problem backwards, you must manipulate the problem to produce the correct answer...the pattern in this case. We do this routinely in communications, it's called pre-distortion.
The masks are producing features far below the diffraction limits of traditional optics. They even must account for neighbor effects as well.
We will be working at 14nm soon!!
Exactly. The mask set may require 2 or 3 individual masks to get the desired result. This is working the the problem backwards, you must manipulate the problem to produce the correct answer...the pattern in this case. We do this routinely in communications, it's called pre-distortion.
The masks are producing features far below the diffraction limits of traditional optics. They even must account for neighbor effects as well.
We will be working at 14nm soon!!
-
- Posts: 188
- Joined: Sat Nov 08, 2014 11:12 pm
- Location: Portland, OR
It would be fascinating to see 3d synthetic stereo pairs you can generate (with no extra work for you. Lots for the computer) with Zerene. The layers might show up very well. The 2d versions are wonderful.mawyatt wrote:If you look closely you can see down into the chip surface, below the first layers of metal and dielectric insulators. These insulator layers are so thin you can see through them!!
If your pictures aren't good enough, you're not close enough. - Robert Capa
-
- Posts: 188
- Joined: Sat Nov 08, 2014 11:12 pm
- Location: Portland, OR
I think Rik's guide will explain much better than I would, since I'm still kind of new to it myself.
But just to give it a quick run:
In Zerene Options/Preferences under Stereo/Rocking, check the box to "Generate Stereo Pair or Rocking Sequence"
The defaults should be OK just to give it a try.
Then also under Preferences "Preprocessing" it's a good idea to reduce the size with "Image Presizing."
How much depends on your camera and screen. You want two images side by side to fit on your screen (maximum). Check the Image pre-sizing box and enter a pct. I go for 18% with my setup.
On that same preferences window, you probably want to check "Stack every Nth Frame" and choose "2". That will speed things up and and the reduced size, your focus is less critical, so skipping frames is fine. With those options selected, stacking will run much faster.
Then just hit the APPLY button and processes your stacks as usual. I generally run PMAX because touching up stereo is difficult and PMAX typically has fewer artifacts and the grain is less of an issue at smaller size.
Once the stacks are done, you'll have 2 new output images. Select them and choose "Tools>Stereo Preview"
That should put the two images on the screen side by side and you should be able to cross your eyes to see the 3d image form in the middle. Sometimes it takes practice. There are some great ones linked to by member Charles Krebs in this thread:
http://www.photomacrography.net/forum/v ... highlight=
If you can't see the 3d in Stereo preview, your images may be flipped left/right (that depends on the order Zerene thinks is best to run the stack, so it often is flipped the wrong way).
While still in preview, click your cursor on the left image and you should see it show you in your list of output images which one is being shown on that side. In that list, choose the other output image to switch which is shown on the left. Follow the same procedure on the right. Hopefully you can now see the 3d.
Finally... under Tools >stereo choose "stop preview" then Tools>stereo "make stereo pair"
That will create an output image in your output list with the stereo pair and you can export it like any other output image.
Phew! That took a bit longer to explain than I thought it would! But really it's very straightforward. Once you've got it going you can tweak things, make animated gifs (aka "rocking pairs"). I'm really looking forward to seeing how your images come out. They're already pretty terrific in 2d.
If I can make one comment though -- they're so perfect, if I saw them in a magazine, I'd assume they were computer generated. Somehow (shadows? Highlights?) for the sake of your own hard work, people should know they're looking at something real!
[/url]
But just to give it a quick run:
In Zerene Options/Preferences under Stereo/Rocking, check the box to "Generate Stereo Pair or Rocking Sequence"
The defaults should be OK just to give it a try.
Then also under Preferences "Preprocessing" it's a good idea to reduce the size with "Image Presizing."
How much depends on your camera and screen. You want two images side by side to fit on your screen (maximum). Check the Image pre-sizing box and enter a pct. I go for 18% with my setup.
On that same preferences window, you probably want to check "Stack every Nth Frame" and choose "2". That will speed things up and and the reduced size, your focus is less critical, so skipping frames is fine. With those options selected, stacking will run much faster.
Then just hit the APPLY button and processes your stacks as usual. I generally run PMAX because touching up stereo is difficult and PMAX typically has fewer artifacts and the grain is less of an issue at smaller size.
Once the stacks are done, you'll have 2 new output images. Select them and choose "Tools>Stereo Preview"
That should put the two images on the screen side by side and you should be able to cross your eyes to see the 3d image form in the middle. Sometimes it takes practice. There are some great ones linked to by member Charles Krebs in this thread:
http://www.photomacrography.net/forum/v ... highlight=
If you can't see the 3d in Stereo preview, your images may be flipped left/right (that depends on the order Zerene thinks is best to run the stack, so it often is flipped the wrong way).
While still in preview, click your cursor on the left image and you should see it show you in your list of output images which one is being shown on that side. In that list, choose the other output image to switch which is shown on the left. Follow the same procedure on the right. Hopefully you can now see the 3d.
Finally... under Tools >stereo choose "stop preview" then Tools>stereo "make stereo pair"
That will create an output image in your output list with the stereo pair and you can export it like any other output image.
Phew! That took a bit longer to explain than I thought it would! But really it's very straightforward. Once you've got it going you can tweak things, make animated gifs (aka "rocking pairs"). I'm really looking forward to seeing how your images come out. They're already pretty terrific in 2d.
If I can make one comment though -- they're so perfect, if I saw them in a magazine, I'd assume they were computer generated. Somehow (shadows? Highlights?) for the sake of your own hard work, people should know they're looking at something real!
[/url]
If your pictures aren't good enough, you're not close enough. - Robert Capa
-
- Posts: 215
- Joined: Sun Jul 22, 2012 2:30 pm
- Location: Bedford UK
My thoughts entirely Max. A lot of images are so good on this forum, they look almost unreal. I have spent the last day experimenting with layer masks and gradients to try an give an image a more natural look, not that I would say it's that great an image. But I'm looking at it thinking it just needs that little bit extra.MaxRockbin wrote:
If I can make one comment though -- they're so perfect, if I saw them in a magazine, I'd assume they were computer generated. Somehow (shadows? Highlights?) for the sake of your own hard work, people should know they're looking at something real!