4.12.2010

It's the little things that make life so interesting.......

From time to time my clients at places like Freescale, Motorola or AMD will come alive and call me with an assignment to photograph a chip "die".  The schedule is usually as short as the fuse on a bottle rocket and the part is usually a beta product with a blemish or defect.  If I remember correctly this little beauty was about 2 or 3 mm in width.  It also had some scratches.  I needed to make it look intriguing and needed to do it by end of next business day, 22 hours and counting.

I don't know if you routinely photograph anything that small but they are tiny enough that a sneeze sends them flying and a piece of dust looks like a boulder on the landscape of the die.  To get the kind of magnification you'll need you'll want to have a bellows to stick in front of your camera and you'll want to use a fairly short focal length lens that's computed for high magnification.  Maybe a 12.5mm Photar or a 25mm Zuiko lens made specially for magnifications that start at five times lifesize.

Many nasty things happen when you go above 5x lifesize.  First, the finder gets really, really dark.   The image gets really hard to focus.  Wide open on an f4 lens the depth of field becomes the width of a human hair.  If you stop down you are immediately confronted with sharpness robbing diffraction.  You must be exactly planar from the sensor plane to the lens plane to the subject plane or you will never achieve sharp focus over the whole object.  We're talking fractional degrees of angle here.

Finally, the chip die is actually a dull finished piece of silicon.  The bright colors on these samples and on samples you've doubtless seen elsewhere is the painstaking end product of moving lights around at different angles and different altitudes until your get a reflection of a layer and the refraction of the reflection creates a dominant color.  Shorter way of saying that:  You gotta move the main light all over the place till you get a blast of color.  But it's never the same color twice.

There are hundreds of ways to get sabotaged. Could be dust in the air, could be an alignment just a tiny bit out of whack.  You can spend hours tracking it down.  But what the heck, the fun is in the challenge.
In the old days the dies were giant.  Some where half an inch by half an inch.  Easy as pie.  We routinely shot that stuff with 4x5 sheet film.  But every 18 months the chip dies get smaller and smaller.   At this point it's no longer feasible to put a standard macro lens like a Nikon or Canon 50 because the lens is too long to achieve the magnification you need on the available bellows.

In a few more 18 month generations this kind of documentation will  probably need to be done on low mag industrial microscopes with oculars for camera attachment.  Right now, if you can do this reliably you'll be in demand.  Not as glamorous as shooting lingerie models but I'm going to be these kind of assignments actually pay better.

Here's two more:



Wouldn't you hate to solder the leads onto the connectors by hand?  Makes me laugh just to think about it.  I hear the micro chip industry is coming back to life.  Should keep some people a bit busier.  They sure are sexy.  Wish I could tell you which camera but I've been through so many I can't.

Next post.  Why some people (me) change cameras a lot.  Conjecture from a famous psychiatrist.

12 comments:

J said...

I honestly had no idea what happens beyond the reach of the 300mm or even 500mm macro lenses. This is exciting to read. To think of such small depth of field... how do you make sure it's perfectly perpendicular to your line of sight?
And they look really sci-fi, which I assume is a good thing?

J said...

Oh and the big sizes are definitely worth checking!

Bruce Walker said...

Thank goodness lingerie models don't keep getting smaller every 18 months. :-)

Those are very cool shots, Kirk.

ginsbu said...

Very interesting. One question: I was surprised that a short focal length is required for this sort of work, would you take a moment to explain why? Thanks!

Bold Photography said...

I'm certainly busy! I was wondering when you'd get the next call to shoot some chips...

kirk tuck said...

ginsbu, the longer lenses require much too long an extension to get them to focus close enough. The solution for micro work has always been wider angle lenses which increase magnification much more quickly as they are moved away from the film plane. 12.5, 25 and 35 are typical focal lengths for micro work.

Robert said...

maybe not every 18 mos. model are getting smaller and it is disturbing to me I don't know where it will stop. I recently heard someone say high fashion is a size 0. yuck. I prefer a size 8. but Marylin Monroe was a size 12\14. she was the hottest thing in her time, and she would be considered a plus size model today.

Wolfgang Lonien said...

Very nice and professional job, Kirk! You're truly a pro.
And thanks to Bruce for a good laugh ;-)

Nick said...

You could do all that work, if you want your photos to look nice or something, or you could just shove an old Canon G5 up against one eyepiece of a binocular microscope and shoot away. Not that I've ever published the result of something so kludge-y...

Alan Fairley said...

Yeah, way to go on this is a binoclular microscope - if you have one with a camera attachment tube, not a kludge, but the right tool for the job. And you can pick up a binoclular scope for less than what you would shell out for bellows, short FL macro lenses, ect. Just sayin'

ginsbu said...

Kirk, thanks for the explanation!

kirk tuck said...

Alan, The disconnect with the microscope for me, up to this point is that there are few available low magnification optics that I'm aware of. Seems that most microscope lenses start at 25X and up.
Let me know if you know otherwise....