"Atomic force microscopy was initially developed as an instrument that shows atomic resolution for nonconductive surfaces," Hans Peter Lang of IBM's Zurich Research Laboratory told nanotechweb.org. "In the wake of developing a chemical AFM, we realized that we could use the surface of the cantilever to detect chemical and biochemical processes."

Lang, his co-worker at IBM Zurich Christoph Gerber - one of the co-inventors of the AFM - and colleagues from the University of Basel, Switzerland, gave an overview of the applications of cantilevers in nanosensors in a tutorial paper in the journal Nanotechnology.

"With cantilevers we can build miniaturized versions of calorimeters, thermogravimetry devices, chemical sensors, mass detectors and biosensors," added Lang. "The level of sensitivity surpasses that of conventional instruments by orders of magnitude, and miniaturized devices can be batch-fabricated at low cost, they show quick responses and might also be disposable."

The beginnings were humble. In 1993, scientists at IBM Zurich Research Laboratory first used an aluminium-coated cantilever to look at the reaction between hydrogen and oxygen. The heat given off as the two gases combined to create water caused the beam to bend because the aluminium coating expanded more than the silicon underneath.

Similarly, the researchers examined phase transitions in hydrocarbon waxes by placing a few nanogrammes of the wax on the tip of a metal-coated cantilever. Heating the sample made the beam bend, but when the wax changed phase it absorbed energy, temporarily stopping the beam from flexing further.

Smelling success

Subsequently, the scientists found that even an event as small as the adsorption of molecules onto the surface of a cantilever can change the surface stress and cause bending. What's more, adding a functionalized layer to the cantilever alters its adsorption properties, a feature which led the researchers to develop an "electronic nose" based on cantilever sensors.

The team made an array of eight silicon cantilevers, coating each one with different polymer layers. The gases under analysis diffuse through the different layers of polymer at different rates, causing the polymers to swell and bending the cantilevers. Examining the bending pattern of the eight cantilevers (using a technique such as neural network or principal component analysis (PCA)) provides a fingerprint for the gas.

The technique has applications in process and quality control, environmental monitoring, characterizing complex odours and vapours, fragrance design, oenology (the study of wine) and in medicine, analyzing patients' breath.

"To give an example of a real-world application we studied the bending patterns of an array of polymer-coated cantilevers when we exposed them to the vapour of various perfume oils," explained Lang. "It is possible to train a neural network to recognize these vapours. We did the same to distinguish several types of soft drinks and various whiskys."

Perhaps more importantly, the electronic nose can also detect the presence of small quantities of acetone in human breath, which may be an early sign of diabetes mellitus. So far, the researchers have simulated the breath of a person with diabetes by mixing acetone vapour with exhaled air, but now they are in collaboration with a hospital in Basel, Switzerland to obtain breath samples from real patients. The technique may be more sensitive than conventional methods and help doctors to detect diabetes earlier.

Lang says that the cantilever array electronic nose is much smaller than conventional electronic noses. This means that it can be portable, give faster response times and requires smaller volumes of vapour for the analysis. The device also gives a response pattern for each gas under examination, which is easier to analyse than the output of a gas chromatograph.

Fluid focus

What's more, Lang and colleagues have adapted the nose for use with liquids by incorporating a measurement cell. They've employed the device to detect processes like DNA hybridization, protein adsorption and antibody-antigen reactions.

"The special thing about our technique is that we can detect such biochemical processes without labelling the molecules - we don't need any fluorescent or radioactive tags," said Lang. "The detection proceeds in a purely mechanical way."

To detect DNA hybridization, for example, the researchers coated the cantilevers with gold. Then they applied short single-stranded oligonucleotide DNA sequences functionalized on one end with a thiol group. The thiol groups bound the oligonucleotides onto the surface of the cantilever. When an oligonucleotide that complemented one of the sequences attached to the cantilevers was present in the sample, it hybridized with that sequence. The resulting double stranded DNA took up more space on the cantilever surface, causing it to bend because of steric repulsion.

In a cunning twist, the researchers coated some of the cantilevers in the array with probes unspecific to the target molecules. "This gives us reference cantilevers for the reactions that we want to observe," explained Lang. "We think it is very important to use reference cantilevers as it can cancel out reactions that take place unspecifically on any cantilever, as well as thermal drifts."

A few aspects of cantilever sensors require further development, however. "We need to find a way to mount the cantilever array into the chamber without having to align it optically," said Lang. "We have to develop a reliable mechanism to coat or functionalize cantilever array chips on a wafer scale. It would also be useful to have a cantilever response database for a large number of analyte vapours and sensor coatings."

So what's next? Lang says that both Veeco Digital Instruments, US, and Concentris, a spin-off from the University of Basel in Switzerland, plan to bring commercial cantilever sensor devices to market in a few months. It looks like the technology is here to stay.