The researchers looked at silicon cantilevers that were 3&ndsah;5 µm long, 1.4–1.5 µm wide and around 30 nm thick. Measurements with a laser Doppler vibrometer revealed the resonant frequency of the beams before the team added a coating containing protein molecules and an antibody to the vaccinia virus.

After remeasuring the beam's resonant frequency, the team submerged the beam in a mixture containing the vaccinia virus (which was used in smallpox vaccine). The antibodies coating the cantilever captured the virus particles, binding them to the beam and enabling their detection.

On attachment of the protein coating the resonant frequency of the beam either decreased (as would be expected) or increased.

"The longer cantilevers showed an increase in resonant frequency, while the shorter cantilever beams showed a decrease," Amit Gupta of Purdue told "This could only be explained if the longer cantilever beams had a thicker or higher protein molecule density on their surface, as compared to the shorter cantilever beams."

Fluorescence microscopy revealed that longer cantilevers absorbed more protein molecules per unit area of surface than shorter cantilevers. The team attributed these findings to different diffusion and attachment kinetics of the molecules depending on the size of the cantilever. In addition, more protein molecules assembled at the tip of the cantilever rather than near the base.

"The finding that the protein attachment density depends on the cantilever surface area was surprising as it was counterintuitive," said Gupta. "We would have expected the smaller cantilever beam to capture more protein molecules, as it was more spherical or 3D in shape. From a technological viewpoint, this is important in designing detectors on the micro- and nanoscale. It is also important as it improves our understanding (however slightly) of protein molecules' behaviour."

The thickness of the protein layer also affected the change in resonant frequency - above a critical protein layer thickness the resonant frequency of the cantilever increased on addition of the protein layer but for protein coatings below the critical thickness the resonant frequency decreased. For a 25 nm thick cantilever 3, 4 or 5 µm long the critical protein layer thickness was 56 nm.

The researchers say their work has implications for the design of nanoscale biosensors. "Such nanocantilever-based sensors can have a wide range of applications, but may be particularly important for the detection of airborne pathogens," said Gupta. "The detection of the mass of the analyte provides a unique method for real-time, almost instantaneous, sensing of the target analyte – whether bacteria, cells, or viruses. Other sensitive methods either need to extract the DNA or to detect indirectly other signatures such as released toxins by the microorganisms."

Now the researchers reckon that the next step is to have on-chip electrical-based resonant frequency detectors, rather than using optical-based detectors. "There is also the challenge of [developing] a robust technique for attaching antibody molecules to the surface of the cantilever beams," said Gupta. "And we are currently working towards techniques that will allow us to perform detection using the cantilevers in a viscous medium such as blood."

The researchers reported their work in PNAS.