Well, I’m on my way to Blizzard Technology in San Diego, leaving Vermont just ahead of an actual blizzard in the Northeast.  But I have a few more things to report from my time at IBM.  On Friday we visited the failure analysis (FA) lab. 

This is Kevin, the first person I met in the FA lab.  When I asked Kevin what was the best part of his job he said “We have the best toys.”  Kevin has a degree in Electrical Engineering from Pennsylvania State University.  He started working in Failure Analysis at IBM right out of college and never left because he does something different everyday.  

Well, I took pictures of some really cool “toys” to show you.  But first I need to tell what the FA lab does.  IBM is continually testing their chips to see how long they will run before something breaks.  When one breaks it goes to the FA lab to figure out what went wrong.  So the “toys” in the FA lab are really scientific tools used to figure out why a computer chip stopped working. 

This toy is an OBIRCH system.  OBIRCH stands for Optical Beam Induced Resistance CHange.  It’s inside a big metal box so that when you close the door it will be very dark.  The optical beam needs to be the only source of light to do its work. 

Here I am, inside the box looking at the machine.  See the blue thing right next to my foot.  That’s over 100 individual wires connected to the chip being tested and to the electrical tester.  There are more wires coming in from the other side.  Unlike the characterization lab which tests one transistor at a time, here they test the whole chip to see how the transistors work together.  (Well, actually these are chips that are broke so we are looking for how they fail to work together.) In the middle is the chip being tested. 

Right in front of me is a microscope.  The microscope is used to focus a laser beam onto a very small part of the chip.  If that part of the chip is working well the laser will not have any effect.  But if that part of the chip is not working correctly the laser beam will cause a small change in the chip and the tester will detect a difference in the signals coming out of the chip.  Now we know what part of the chip has the problem and we can examine it to see what caused the fail.

The microscope can also be used to look for very faint light given off by the chip.  When a transistor is leaking it can give off light.  Often a failing chip has a transistor at the fail site which is leaking more than the transistors on the rest of the chip that are working correctly.  So by finding a spot that is giving off more light than the rest of the chip we can find the failing location

This is Greg and I am standing on the tester.  Here you can see the other end of the blue wires we saw inside the OBIRCH system.  This is a “logic tester” It tests logic chips, also called digital chips.  Digital chips are chips that recognize inputs of ones and zeros (A one is a high voltage and a zero is a low voltage.) and send ones and zeros as output.  The logic tester can send the chip millions of signals every second and record the signals that come out. 

Here we have chips which are the brains of three different video game system.  On my left is the chip from a Nintendo Wii ©.  In the middle is the X-box ©   chip and on my right is the playstation chip ©.

Here they are in a bigger view.  The green and pale yellow part is a chip carrier.  It contains wires to connect chip to the rest of the game system.  The chip is under the shiny black part in the middle.  (The X-box board has two chips a big one and a little  one.) The shiny black is epoxy put over the chip to protect it.   On the smallest chip the epoxy has been removed so the chip can be examined with the laser.

Once you know approximately where the defect is, the FA lab has lots of other toys to help you see what the defect is.  Here I am about to load a chip into a Scanning Electron Microscope (SEM).  This is a microscope that doesn’t use any light at all.  Instead it uses a beam of electrons to make an image.  With electrons you can “see” things much smaller than you can see with light.  But electrons don’t travel very well through air so we have to put the chip in here, push it in so the door seal and pump out the air.

Here Jim is showing me how to control the microscope.  You can zoom in and out and move the chip around so you can see every part of it.  As soon as I tried it the picture when blank and a message said “filament broken”.  But Jim said I didn’t really break it.  The filament is just light a light bulb and after you use it for a while it burns out.  Jim came to IBM with a 2yr degree from Connecticut State Technical College. Then he got a 4 year engineering degree while working at IBM.  He has worked at IBM for 28 years because everyday is different and because of the great toys.

Here is a sample of what you can see with a scanning electron microscope.  To get a picture like this first you have to slice the chip in two and then polish the edge of one of the pieces.  The large ball in this picture is a bit of metal used to connect the chip to the chip carrier.  It is really about 60 micro meters in diameter.  That is a little bigger than a spec of dust floating in the air.  The top part is lighter than the bottom because it is made of a different kind of metal.  Below the solder ball, the light colored spots are where we have cut through wires that connect different transistors on the chip to each other.  The smallest ones are only a fraction of a micrometer across.  That’s about the size of a bacteria.

If you need to look at really small things you go down into the basement and talk to Phil.  This is Phil and he’s showing me an atomic force microscope.  Phil is from Syracuse, NY and got is BSEE at the university of Vermont.  He likes what he does because with this tool you can collect information that isn’t available any other way.

Here is a closer view.  For an atomic force microscope you do not need a vacuum because an atomic force microscope “sees” by dragging a very sharp probe across the surface of the sample.  It records the height of the surface at every point on along the path.

Phil showed me this picture to explain how atom force microscope can measure such small difference in height.  When the tip moves a little bit the cantilever bends a little bit.  When the cantilever bends the laser beam reflects off in a slightly different direction.  Because the photo detector is a long way from the probe tip spot where the laser beam hits moves a much longer distance than the distance the tip moved.

This is Leon showing me a plot from a problem he was working on when I came down.  The surface is very flat except for one place.  I enlarged the graph to show you.  See the one peak marked by a blue triangle.  That part of the surface sticks up about 1 nano meter higher than the rest of the surface.  One nano meter is about the size of 3 or 4 atoms.  Noel grew up in Rochester, NY and went to community college there.  He likes working at IBM because he is always learning something new.

To make a 2 dimensional map of a surface the tip traces another path right next to the first one.  After many, many passes it creates a map of the surface that looks like this.

This is what the shiny side of aluminum foil looks like with an atomic force microscope.  Although aluminum foils looks smooth to your eye and feels smooth to your fingers, at the scale of molecules it is very rough.

This is the surface of one of the wafers IBM builds chips on, before anything has been added.  This explains why some of the engineers I met have wafers hanging in their offices to use as mirrors.

The people in the Failure Analysis lab like working there because it’s like being IBMs Crime Scene Investigators.  When something goes wrong with a chip they work like detectives to figure out how it happened so they can make sure it doesn’t happen again.