Letting microbots feel their way

Much of the effort going into microrobotics these days is rather macroscopic in nature. The precision robots used for micro-assembly are not really robots in themselves, but elements in robotic systems: most of the sensing and control functions are carried out centrally, while the "robot" itself just follows orders like a glorified actuator. There are good practical reasons for this, not least that the sort of equipment generally used for micro- and nano-scale imaging is generally very big. However, this may be about to change. Researchers at the Swiss Federal Institute of Technology (ETH) in Zurich are currently working on applying recent research in microscopy to machine vision. The result may be a more autonomous microbot.

Andreas Stemmer and his colleagues at the ETH are working with a technique based on atomic force microscopy (AFM): a technique for measuring the topology of a scanned surface by monitoring the external force applied to a tiny cantilever tip. This is well-known technology and far from new. But more recently, the technique was extended using the Kelvin method, which allows simultaneous measurement of the topology and surface potential of a sample.¹ This is done using an applied AC voltage operating at the cantilever's resonance frequency (see figure 1). The interaction between this voltage fluctuation and the potential of the sample make the cantilever oscillate, and this oscillation feeds back into the system as a DC offset. When the cantilever stops moving, its DC offset is equal to the surface potential of the sample.

Figure 1. The Kelvin method for measuring surface potential, as applied to the atomic force microscope.

Figure 2. (a) Topography and (b) surface potential of the cleaved surface of a AlGaAs/AlAs vertical cavity laser. Variations in the topology (from white to black represents a height difference of 36 nm) are caused by oxide on the AlAs layers. Variations in potential are caused by the chemical difference in the layers themselves and the way they have been doped (n-doping on the left, p- on the right).

An example of what the Swiss researchers have achieved with this technique is shown in figure 2, where the sample used was the cleaved surface of an AlGaAs/AlAs vertical cavity laser. The alternating material layers are clearly seen in the topographic map because of an oxide layer on top of the AlAs. In the potential map, these layers are also visible, and the demarcation between the n- doped and p-doped regions of the device also become clear. The high resolution of the ETH version of this potential imaging technique, here around 50 nm, comes not from simply looking at the absolute DC-offsets, but in small offset variations. This is because the potential at any particular point is affected by all the points around it. Nevertheless, theoretical models developed by ETH researchers suggest that they should be able to get higher resolutions still.

A major advantage of this technique over, for instance, scanning electron microscopy, is that the sample need not be kept in a vacuum. This makes it particularly suitable for biologists. Combined with inverted optical microscopy (where the optical microscope images from below and the AFM scans the sample from above²) has already been applied to biological samples-living proteins-under conditions of high humidity. An obvious next step is to image in liquid, thus allowing the scanning of cells, a development that is in the pipeline.

But the ETH approach is to do more than just perfect another kind of imaging. Atomic force microscopy has the advantage that it uses a local probe: it does not measure from afar. A cantilever attached to a tiny robot, for instance, could allow the machine to locate itself on a sample with nanometer precision. One example of how this might be useful would be in examining a semiconductor gate. By sensing just the sample's surface potential and topology, the robot would be able to recognize the gate potential when it found it and-for instance-scan its topology by following lines of equal potential. A more futuristic application might be to have nanobots fixing broken "wires" on chips by following their potential signatures as tracks. In the nearer term, such probes may simply help to guide microscopes to specific areas of interest, thus simplifying the job of the associated machine vision system or human operator.

1. Andreas Stemmer, Heiko Jacobs, Helmut F. Knapp, "Approaching the nanoworld," SPIE Proc. 2906, 1997.

2. A. Stemmer, "A hybrid scanning force and light microscope for surface imaging and three dimensional optical sectioning in differential interference contrast," Journal of Microscopy, Volume 178, Part 1, pp. 28-36, April 1995.