Silicon Chip Images

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 8:27 am

I started this thread at the request of the moderators to show some image of chips I am working on and how I have progressed. These are silicon chips we and other design.

I started out with a Nikon 105 on extension tubes, and tried some lens magnifies you place on the front of the lens. The results were just OK. Then moved to the microscope objective lens using a Nikon 70-200 as a "tube" lens @ 200mm. This improved the results considerably but I was not satisfied with the lighting, especially when shooting chips with small solder balls use for a technique call "flip-chip mounting". These solder balls are tiny spherical mirrors and reflect everything. Wirebonds are another chip mounting technique where tiny 0.001" gold wires are used to make electrical connections to and from the chip, these are also highly reflective. The chip surface is another highly reflective subject. I have now migrated with the kind help of Rik and others (many posts here on this) to a reversed Raynox lens and a complex set of adapters with the microscope objective. Optically this is the best setup so far but the mechanical stability lacks, discussed later.

My lighting started out with a couple Nikon SB800 diffused flashes, and some support lights. Then moved to a cheap light tent which helped, then to another smaller light tent inside the larger tent..this improved the lighting a little more. Then I went to strobes instead of the flashes because my stacks were getting long (>50, now they are >500!!). These strobes are in light boxes 2' square. I followed with 3 fixed light sources in 2' by 3' boxes. This setup (shown) produced a very uniform lighting as needed when shooting the chips with solder balls, but was very large and difficult to setup and move. I have now moved to a simpler light setup with 4 Ikea LED lamps and a Styrofoam cup diffuser. This doesn't produce a very uniform light source, so I don't expect good results with a chip with solder balls.

Along the way the optical setup changed with the lenses and mounts. At first I used a good tripod, but it was somewhat too flexible and slow to damp vibrations (low natural frequency for you scientists and engineers). Then moved to a modified copy stand which help a little but still not stable enough as I progressed to 10X shooting attempts. Plus I was getting into long stacking sessions (>200) and had to wait up to 15 seconds between shots, this took a long time to capture a stacking session. I finally went to a sturdy Thor aluminum base which improved things more.

I still have the mechanical stability of a long set of adapters (9") with the Raynox reversed and then another couple inches to the objective. This bounces like a trampoline!! I am looking into a lens support like Steve used which should really improve the lens setup considerably.

All of this is a work in progress!!

Cheers,

GemBro · Post by **GemBro** » Wed Dec 17, 2014 9:27 am

Oh wow ... that last shot ... the colours are brilliant ... nice and sharp too ...

So Mike ... which setup was used to get the last shot? ... the tent? ... or the multiple reflectors? ...

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 9:32 am

Gem,

This was done with an early lens setup but with the dual light tents and strobes with diffusers.

I am trying to post more images but something is blocking it. I have better examples with later techniques and some with solder balls. I even tried creating another thread "More chip images" but couldn't upload any images.

Sorry I can't show anymore, maybe the moderator can help??

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 9:40 am

Gem,

As you can see the wirebonds are reflective, as is the chip surface and sides. The QFN package base is also highly reflective, you can see the wirebond shadows in the base. With a more direct light source these shadows become very pronounced and don't look nice. You can remove them post processing but it takes time. With the more unirform light from the tents and strobes the shadows look better IMO.

Cheers,

ChrisR · Post by **ChrisR** » Wed Dec 17, 2014 10:03 am

There are no blocks that I know of. Forum host hiccup? Try again?

Do you get an error message? [Edit] there were 4 images a moment ago - still having trouble?

15 seconds twixt exposures seems a lot. Is most of that for recycling the flash? Most folk use the heads/guns much closer, with fractional power, then the movement-settling periods are all that's really needed.

No of steps - yeah, frightening isn't it

. If you use a 10x NA 0.45 objective on a subject 4mm deep, you need 2000 frames..

GemBro · Post by **GemBro** » Wed Dec 17, 2014 10:06 am

The way I get my images up is this way ...

I've created a Photoshop template (PSD) for all of my image posts ... I've set the template to 1200 x 800 pixels (mostly for Flickr and other places) ... I then save them as PNG, then upload them here, as they are ... so their original size is 1200x800 and about 1.5mb each ... the forum then resizes them accordingly ... not the best practice but saves me having to re-size them again for this forum ...

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 10:18 am

Chris,

Probably something with our work server at my end, don't know. I tried a few times with no luck but will try again soon.

The 15 seconds was basically to let everything die down, the heavy lens combo and focus rail on the tripod took forever to damp down. plus the Nikon D800 must flip mirror up, then wait, then front curtain, then wait and trigger strobe on rear curtain. You can't (at least I couldn't) get the D800 to leave the mirror up, it resets after every exposure which is major source of vibration with my setup. I used the larger setup to get the more uniform lighting, this was required with the flip-chips with solder balls. I am sure others could have condensed the setup, but this worked so I kept it for those images that required the uniform lighting.

My latest setup (Raynox tube lens with Mit 10X on multiple adapters) still needs about 6 seconds to dampen between shots, mostly because of the long lenses arrangement that I need to secure better.

Yes I am doing over 500 steps now showing partial chip images and some smaller chips. The DoF is so small at 5 and 10X

steveminchington · Post by **steveminchington** » Wed Dec 17, 2014 11:22 am

That's a superb image of the chip... must have taken a long time to run that stack!

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 11:41 am

Thanks Steve. I was doing the stacking & PP on my rMBP until I got a Mac Pro, still takes awhile though. At 500+ images, all from a 36MP D800E at 110MB TIFF, it take some time & disk space!!

Hey think I got this image to load up!!

rjlittlefield · Post by **rjlittlefield** » Wed Dec 17, 2014 11:46 am

mawyatt wrote:Hey think I got this image to load up!!

Yep, fully visible and looking good.

So good, in fact, that now I'd like to see a crop showing those usually troublesome areas around the chip where the wires tend to obscure the stuff behind them.

Any chance?

--Rik

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 11:53 am

Here's one with the single tent lighting, note the harsh wirebond reflections. I didn't try and edit these out, but did some editing in the previous image.

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 11:58 am

Rik,

Like this?

rjlittlefield · Post by **rjlittlefield** » Wed Dec 17, 2014 12:10 pm

Yes, excellent.

This is extraordinarily clean.

Can you share some insights about how you avoid "transparent foreground" and various sorts of halos around those elevated wires?

--Rik

mawyatt · Post by **mawyatt** » Wed Dec 17, 2014 12:34 pm

Rik,

Since this is based upon your superb Zerene Stacker, you are asking me

Seriously though, I surmised that Zerene's algorithms like continuous flow of in focus thin ribbons across the chip surface. With that in mind I try and place the chip image angle such that Zerene has something to "grab" onto and most everything else is grossly OOF...and the step size is so small that these thin ribbons line up across the chip and connect well for a smooth in focus transition. This usually dictates step sizes below the DoF because of possible focus ripples between steps (you can see in some of these images at full resolution when the D800 misses a trigger and skips a step..another topic of discussion).

When the stepping starts out everything behind the image front is grossly OOF, but the very top of the wirebonds will be in focus. This is how I select the starting point. As the image capture slides across the chip and past the chip back edge the rear wirebonds become the only thing in focus and Zerene follows the focus trail towards the last image capture point which is the tip of the rear wireboonds.

Zerene does an amazing job of stacking these images together under these conditions, so my hats off to you sir for creating such an amazing piece of software.

Hope this is what you were looking for.

BTW if you want to know about the D800 issues I'll be happy to post this information.

Cheers

rjlittlefield · Post by **rjlittlefield** » Wed Dec 17, 2014 1:45 pm

Trust me, I was not fishing for compliments. Zerene Stacker is good, but it's usually not good enough to get stuff like this completely correct without human intervention in the form of retouching. Thanks for the additional information. Now I need to go away and think for a while.

--Rik