How The AFM Works 
This is a (somewhat crude) drawing of the guts of an AFM.
The Wafer: The wafer is a tiny rectangular piece of silicon nitride measuring no more than a milimeter across and only a few milimeters long. The wafer holds at least one cantalever and tip, which have been etched out of the wafer using techniques similar to those used to create microchips, at each end. A small clamp holds one end of the wafer in the AFM.
The Cantalever: The cantalever is a tiny protrusion at the end of the wafer hardly visable to the naked eye. This extremely thin structure usually comes in the form of a V (as shown in the diagram), or a rectangular plank. At the end of the cantalever is the tip which actually makes contact with the sample's surface. The cantalever is somewhat flexible and reflective.
The Tip: The tip is a pyramidal protrusion at the end of the cantalever. The very tip of a good tip measures no more than a few tens of nanomteres across (around 0.000000070 meters). The tip is scanned over the sample's surface and rides over any surface variations. This causes the cantalever to bend and twist.
The Piezo Electric Scanner (PES): It's a mouthful, and it's also the most expensive component of the AFM. The sample is placed on the PES and is scanned under a stationary cantalever tip (there are AFM models in which the tip is scanned over a stationary sample). The PES is a very precise component and is able to accuratly move the sample through a scan (a back and forth raster pattern) of only a few hundred nanometers. The PES consists of three piezo electric elements which expand and contract independently depending on how much voltage is placed across the individual elements.
The Laser: A laser beam (denoted by the red line in the diagram) emanating from a laser diode is shined on the top surface of the cantalever near the tip. It is then reflecetd off the tip onto an adjustable mirror and is finally detected by a photodiode. By measuring where the laser is on the photodetector we can determine by how much the cantalever has been deflected.
The Photodiode: The photodiode is divided into two sections, say A and B. We can determine by how much the cantalever has been deflected by determining the position of the reflected laser beam on the photodiode. This is accomplished by measuring how much of the reflected laser light falls on section A of the photodiode versus how much falls on section B. Some AFM's have a C and D section (their photodiode is a square divided in fourths) to detect torsional (twisting) cantalever difflections.
The principles behind the AFM are relativly simple. Suppose you are asked to determine what the floor of a room looks like but you cannot look at or touch the floor. You might think for a little while and then come up with the idea of feeling around the floor with another object, say a long stick. Even better, you could get a fairly precise measurment of the height of objects on the floor relative to one another if you stuck a mirror to one end of the stick, shined a flashlight on the mirror, and observed the position of the light relfected off the mirror onto the adjacent wall as you scanned back and forth over the floor. Pressing down hard on the stick would cause it to bend to varying degrees as it traversed objects on the floor. This would cause the angle of the mirror relative to the wall to change which would in turn cause the reflected spot of light to move up or down.
Maybe there would be an easier method, but the one described above basically describes how an AFM works. Instead of the stick we use a cantalever and tip. The cantalever is reflective so we don't need a mirror. The tip is initially pressed firmly onto the sample's surface so it bends slightly. The raster scan (down and then diagonal up ect..) is then initiated by the PES and the cantalever is bent more or less as the tip traverses the sample's surface variations. The laser is reflected off the cantalever and onto the photodiode (our wall). The photodiode returns a voltage describing the laser's position (i.e. positive voltage = the upper portion of the detector, negative voltage = lower portion, zero voltage = center of detector). This voltage along with the XY coordinates of the scan are fed into a computer which crunches the data and produces an image depicting the heights of the sample along the scan lines. Using this method the AFM can image objects only a few tens of nanometers in dimension, far smaller than the wavelengths of visable light.
In reality the process is a little more complicated. There is a feedback loop from the computer to the PES which raises and loweres the sample during the scan so as to maintain a relativly constant pressure and bending on the cantalever. From this feedback loop we can obtain another image consisting of PES height correction versus the location along the scan line. This is called the error signal. So in the end we have two images: One is produced by measuring the difflection of the laser as the tip scans the surface. The other is produced by measuring how much correction is needed to maintian a relativly constant pressure on the cantalever.
That is basically how an AFM works. With it a person can "see" bacteria, viruses, individual lipids in a cell membrane, enzymes, DNA molecules, and in tunneling mode (which I won't go into) you can see individual atoms. It's an amazing machine!
Back to the AFM Homepage