From: Michael Huth (enigma_at_mail.lipsia.de)
Date: 2008-05-17 12:17:21
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Wolfgang Moser wrote: | Hello Michael | | Michael Huth schrieb: |> you know I don't write often here, but this might be of interest. |> I got a defect VIC2 R3 from Stephan Lesch today to map the IC. |> At first I thought this can be quite complicated and may require to put |> it into the ESEM (Environmental Scanning Electron Microscope), |> but after a brief check in an optical microscope the ICs structure was |> good visible. With a good adjustment of the image plane there are |> also the layers distinguishable. At least upto the third I guess. This |> was done at about 5000x. |> The mapping images were taken at 1000x. The width of a 'conduction line' |> is about 6 micrometer. |> I took an image every 0.5 mm whereas each image shows a wier area so |> there is an overlap to put the images together. | | I saw your message to forum-64 already. This is | really great work, the images are so damn sharp. Thanks. The sharpness is not perfect though. The microscope has a well defined image plane, that is the distance at which the image is sharpest. This means that the IC-plane has to be parallel to the image plane, whereas the VIC2 is more or less bend a bit. It is in the center 'deeper', maybe it is some artefact of the packaging. I had put the whole opened package under the microscope, thats why the chip itself is still bonded in the images. Another point is that I can move the sharpness between the layers, at least quite well for the topmost bright metal layer and the more brownish second layer. The green third layer is overall hard to distinguish. In the recorded images I tried to focus somewhere between the top and second (brownish) layer, to get good sharpness in both. Everything else would have meant to take at least two shots of the same location with different focusing. Because of the bending I had to adjust the focus length for each picture, this is why it varies a bit. This is best noticeable if you look at the second layer, where the connection turn to a filled brown/orange shape if focused. This focus length is even more defined with a higher magnification, f.e. 5000x but hey this would have meant to take about 25 times more images. | | |> To give you a brief idea how this looks I uploaded two images at: |> http://mail.lipsia.de/~enigma/vic2/10_5.png |> http://mail.lipsia.de/~enigma/vic2/10_6.png |> |> 10_6 is in the grid to the top of 10_5. | | I also played around with these two in Photoshop. | There was a slight gap between these two instead | of the mentioned overlap as it looks to me | (judging from the structures to the right side). Yes I noticed that too. As I wrote in forum-64 consider this as a first try. The overlap was checked in x-direction, but the camera has a lower resolution in y-direction, so the seen object wide is smaller. Switching to the normal (human eye) optics the visible area is circular symmetric. This is exactly the point where a putting together the images can tell if something crucial is missing. It took me about 2 hours of work, so I could take images of some specific areas again, if the microscope is unused. | | Btw. I recommend using a selective smoothener to | eliminate the noise a bit (more) and get better | compression ratios. Might be considerable for a final release. Ciao... ...Micha -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.2.2 (MingW32) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFILrCxjH+kqnb44nMRAth3AKCuJBnZSCD+W5eMG5cE8Ws61zLibwCfRgrM gOWcVhJtoun46dd5253uHjw= =j9Pu -----END PGP SIGNATURE----- Message was sent through the cbm-hackers mailing list
Archive generated by hypermail pre-2.1.8.