by Lin Yangchen
It’s even more primitive than a magnifying glass. This microscope can’t actually see anything. All it does is poke at the specimen with a pin.
But that’s where the invisible voodoo happens. The pin is so sensitive that it can feel and touch individual atoms.
Apparently you can assemble an atomic force microscope yourself, although it won’t be quite as sharp as those at MIT. There’s one hidden away in a far corner of a lab on a dark upper floor of the Science Centre. The whole contraption is suspended from hot pink bicycle hooks, as it is highly temperamental. The tiniest vibration can drive it crazy.

The pin protrudes from the end of a fine cantilever carved from a single crystal of silicon. The diameter of the tip of the pin is on the order of 10 nanometers. Once you’ve slotted the sample into the microscope, you gingerly lower the pin closer and closer to the sample while the pin gently vibrates and feels the space below it. If you’re not careful, it smashes into the surface like a meteorite and is destroyed. The only way to know if it's still there is to put it under a scanning electron microscope.
This particular model of AFM uses intermittent contact mode, in which the cantilever oscillates up and down at approximately its resonant frequency, about 150,000 times a second. As it scans across the sample surface with the help of a piezoelectric motor, it rapidly touches and leaves the surface instead of maintaining continuous contact. This reduces the risk of the pin being sheared off by a “cliff”.
A laser beam aimed at the reverse side of the cantilever gets reflected onto a photodiode segmented into four quadrants. As the probe follows the topography of the sample, the corresponding movements of the reflected laser are measured through fluctuations in the relative excitation intensities of the photodiode quadrants. The measurements are calibrated using a silicon chip etched with three-dimensional patterns of known height.



Another problem is that the surface relief of the stamp is on the order of tens of micrometers. This is too deep for the probe which is only about 15 μm long. The image contains dark featureless voids where the tip traversed empty space.
A way around these problems is to extend the tip with a carbon nanotube. But that will be for another day.
Acknowledgements
I am grateful to Wulf Hofbauer for valuable discussions and assistance.
References