Macro Imaging Setup: Homage to "Powers of 10" Movie

[Andy Resnick]

Ok...? I don't have scale bars on the images, either.

Edit- I don't mean to sound ungracious: I am very happy and pleased that you spent time and effort thinking about my images! I probably should acquire a calibration image for the 100x lens...

I suppose I should also offer the following disclaimer: "The images presented here and on the lab blog are intended for educational and outreach purposes only; no federal research funds were used to obtain these images and they are not considered a product of scholarly activity."

 

[dlgoff]

On my screen, the penny in the .1X photo is about 8 times the size of an actual penny. The chip in the 1X photo is about 15 times the size of the chip in the .1X photo. The size of the text 'GU7868' in the 10X photo is less than 2 times the size of the text in the 1X photo. The size of the text '800A' in the unlabeled photo between 10X and 100X is about 2 times the size of the text in the 10X photo. The radius of the circular object in the 100X photo is 3 times the radius in the unlabeled photo and the the radius in the 1000X photo is 3 times the radius in the 100X photo.

What? You haven't calibrated your monitor screen?

 

[Andy Resnick]

There appears to be (at least) two major classes of chip fabrication technologies, in addition to some hybrids (radio-frequency circuits, mostly). The images I've posted so far are of one fabrication method- the wires conform to an uneven substrate and are joined with 1-micron diameter 'spot welds'.

This next series is of the other major type, where the traces are planar and there does not appear to be any obvious joining. The overall feature sizes are much smaller as well, resulting in much more colorful images:

http://imageshack.us/a/img694/7672/dsc00041cs.jpg

http://imageshack.us/a/img43/8510/dsc00084ut.jpg

http://imageshack.us/a/img834/8384/dsc00078wy.jpg

http://imageshack.us/a/img846/7118/dsc00079jy.jpg

http://imageshack.us/a/img84/9937/dsc00098qe.jpg

http://imageshack.us/a/img210/5783/dsc000981o.jpg

Even at moderate numerical apertures (0.75 and 0.9, corresponding to f/0.66 and f/0.55), the large color blocks appear uniform, as the wires are too small to be resolved. Only the 100/1.47 lens is capable of showing the traces.

It's clear that I'm pushing a few optical limits- aside from diffraction, the Petzval sum is greatly magnified, as are the chromatic aberrations. I've also slightly blurred the images to reduce speckle. I'm thinking about ways to compensate- a field flattener (a.k.a. a plano-concave lens located near the sensor) may not be realistic, but monochromatic illumination may be interesting and useful...

 

[Andy Resnick]

Now that I've done this for a bit I've gotten some additional ideas. The physical sizes of these things varies considerably- here's a test arrangement of 'powers of 10' in terms of area and mass for integrated circuits and simple MOSFETs:

http://imageshack.us/a/img845/2104/dsc02233q.jpg

http://imageshack.us/a/img197/6589/dsc02230copy.jpg

http://imageshack.us/a/img692/6172/dsc02229copy.jpg

The largest is an nVidia chip- the largest IC I've unearthed so far. The smallest device is a LED. The data:
Area [mm^2] mass [mg]
Chips
1 188 293.25
2 25.8 14.35
3 5.2 2.94
4 0.48 0.14
5 0.20 0.09
MOSFETs
1 16.8 4.41
2 1.13 0.40
3 0.09 <0.01
4 0.063 <0.01
5 0.022 <0.01

Roughly 5 orders of magnitude in area and 6 in mass.

There is also a huge variation in aspect ratio- the ratio of thickness to (say) width. The nVidia chip is massive, memory chips are incredibly thin: this is somewhat obvious, but some tiny ICs are also really thick- nearly as thick as the nVidia (that's why one of the chips is out of focus). I'm not sure how to to represent this (yet).

Yet another possibility are series showing changes in number of transistors (or circuit density) and storage on RAM chips. I'm not sure how to make this visually compelling, tho.