Scanning Tunnelling Microscopy is all very well, and allows you to look at surfaces of conducting (metal or semiconductor) samples with atomic resolution. But what about non-conducting samples? One solution is to cover the sample with a thin layer of metal so you can at least see the general shape of the sample. This is not ideal, and so to get round this problem a new version of scanning microscopy was invented in 1987*, called Atomic Force Microscopy (AFM)

The concept of AFM is extremely simple, although the execution is often very sophisticated. A small* cantilever is held approximately parallel to the sample surface. The free end of the cantilever is attached to a probe that interacts with the sample. The probe is normally in the form of a sharp tip. Any force acting on the tip causes the cantilever to bend, and the resultant deflection is measured by some means. If the probe is brought into contact with a surface and then scanned over that surface, the profile of the surface can thus be measured with the resolution determined by the sharpness of the probe (often less than 10 nm).

There are several methods that can be used to detect the cantilever deflection in AFM. There are three main requirements:

1. the accuracy must be high enough to detect cantilever deflections in the angstrom or even sub-angstrom range.
2. probes can be changed without requiring time-consuming realignment/calibration of the system
3. the probe is not perturbed significantly by the measurement system

The main methods in use are discussed below.

Cantilever detection methods

• Tunneling sensor - i.e. coat the back of the cantilever with metal, stick another electrode very close above it, create a voltage difference between the two and measure the tunneling current. Very accurate, but only works over a very short range (a few nm). You can get round this using more complicated schemes.
• Capacitance sensor. Again, coat the back of a flat(ish) cantilever with metal, and fix another flat metal electrode some distance above it. This gives you a capacitor, the capacitance of which will change with the distance between the plates. The problem with this method is that measuring the capacitance involves putting a voltage difference between the plates, which results in an attractive force between the plates.
• Strain gauges. It is possible to manafacture small strain gauges (usually of the piezo electric type) on the cantilever itself. This is not actually as expensive as you might think.
• Optical interferometry. Use the back of the cantilever as a mirror at one end of an interferometer.
• Optical beam deflection. Again using the back of the cantilever as a mirror, bounce a laser beam off it at an angle. As the cantilever bends, the position of the reflected beam changes, and this position can be measured by a simple photodiode system. One interesting feature of this system is that if the cantilever twists, the reflected light will move orthogonal to the first direction. This information is not avaliable using the other detection methods.

Types of AFM probes

Wire-based probes

Wire-based probes were a popular type of probe extensively used in early AFM studies, due to the fact that they can be made fairly simply. Typically the wire used is several tens of microns in diameter and made of Tungsten or Platinum-Iridium alloy; the main criteria being that the material is stable and resistant to oxidation. To obtain a sharp point the end of the wire is subjected to electrochemical etching, giving sharp and well-defined points. After etching, the sharp end of the wire is bent downwards, forming the AFM probe. This process can yield good tips with high aspect ratios* and small tip radii (typically 30 nm or less). The process of etching and bending is performed manually, and thus consistency from one tip to another is usually poor. An alternative method of manufacturing wire probes is to stretch a portion of the wire to breaking point. Probe consistency is also variable using this method. Currently wire probes are used mainly for STM studies.

Silicon Nitride probes

SiN probes are produced by first creating a mould by conventional lithographic techniques* on a single-crystal silicon wafer. To define the tip, a pit is etched into the silicon through a lithographically defined square window using an anisotropic etchant such as potassium hydroxide. If the silicon wafer has been cut along the correct crystal axis, this process gives you a pit in the shape of a square-based pyramid. A layer of Silicon Nitride is then deposited into this mould and patterned using further lithographic processes to form the cantilever and tip. After a glass block (about 1x1x2 mm in size) is bonded to the rear end of the cantilever for handling purposes, the remaining Si is etched away, releasing the cantilever.

The tips produced by this method are square-based pyramids with sides that make an angle of 45° to the tip axis. The tips produced by this method normally have apex radii of 20 to 50 nanometers, which is useful for lower resolution work.

Monolithic silicon probes

Using similar processes to those used in mould manufacture for SiN probes, complete probes can be fabricated directly from silicon. The probes can take a variety of shapes depending on the application and the details of the manufacturing process. The tip radius is often 10 nm or less (when uncoated), easily small enough for all but the most demanding work.

Modes of operation

So we have a scanning system, and we have a sharp tip/cantilever combination to scan with. How is this actually done? The simplest method is simply to bring the tip in contact with the sample* and then start scanning. Measure the cantilever deflection as the tip scrapes over the surface, and display the result as some kind of image. Simple.

And also not terribly clever. There are several problems with this approach. In particular, if the sample is quite rough, when the probe passes over the high points the cantilever will be bent up, and so will press down harder onto the sample. In other words, the force exerted on the sample by the probe will change with the sample height. It is easy to see that if you are trying to image something squishy (like a cell, for example) this is not desirable. To counteract this problem, a more advanced and complex imaging mode is introduced - the constant force mode

Constant force, feedback and all that jazz

Tapping (intermittent, AC etc.) mode

Introduction to other applications, MFM, EFM, etc