Home News Publications Members Links Contact

Resonant Mass Sensors Containing Nanofluidics

Robert A. Barton, Rob Ilic, Scott S. Verbridge, and Harold G. Craighead


The ability of nanomechanical resonators to sense mass in a liquid environment is compromised by a large dissipation of energy to the surrounding liquid. In order to overcome this problem, we have designed and fabricated nanoscale resonators that contain fluidic channels. The channel allows delivery of analytes to the resonator in liquid while the resonator is operated in vacuum to avoid viscous damping. We estimate that the channels will be sensitive to masses as small as an attogram based on the mechanical resonance properties of the devices presented here.

Research Summary

Figure 1: Sequence of steps employed to fabricate resonant nanochannels.

Nanomechanical resonators can be extremely sensitive mass sensors and have been used to weigh objects as light as a single gold atom [1]. Unfortunately, maximizing the sensitivity of these devices generally requires that they be operated in a vacuum, where the lightest, most sensitive resonators will experience minimal damping from their surroundings. This restriction precludes many applications of nanomechanical resonators to biology, where it is useful to weigh analytes from a liquid solution [2]. One way to measure mass from solution while avoiding the strong viscous damping associated with operation in liquid is to package the analyte solution inside of a hollow nanomechanical resonator and leave the outer surface of the resonator in contact with vacuum [3]. Previously, this technique has been exploited to measure the mass of nanomolar concentrations of proteins and single live cells in solution. Weighing single viruses, single proteins, and DNA from solution, however, will require hollow resonators dramatically smaller than anything demonstrated previously.

Here we demonstrate the fabrication of resonant nanochannels with f Q > 10^11 and masses on the order of picograms. The details of the fabrication are summarized in Figure 1. Reactive ion etching (RIE) is used to define trenches in a silicon surface. We use chemical vapor deposition to fill the trenches with oxide and chemical mechanical polishing to planarize the oxide/silicon surface. Next, low pressure chemical vapor deposition (LPCVD) is used to deposit a layer of polysilicon (Figure 1a). We use photolithography to pattern the polysilicon into doubly clamped beams of various lengths and widths over the trenches, and then we suspend the polysilicon beams by etching the underlying oxide with hydrofluoric acid (Figure 1b). Subsequently, we use LPCVD to conformally coat the suspended beam and the surrounding patterned polysilicon with silicon nitride, which defines the nanochannel and the surrounding reservoirs that will feed it with liquid. We pattern the nitride with irrigation holes that provide access to the sacrificial polysilicon (Figure 1c) and then etch out the polysilicon using a XeF2 etch. The remaining nitride comprises a hollow channel with two reservoirs on either side that allow us to feed in fluid. We used CVD and lithography to fill in the irrigation holes, which prepares the device to hold fluid (Figure 1d).

Figure 2: (a) Scanning electron microscope image of a hollow suspended nanochannel. The twenty-micron-long channel has been hollowed out by a gas etch that enters through circular holes on either side of the channel. These holes have been filled with oxide to eliminate leaking, leaving visible impressions. Scale bar, five microns. (b) Resonance of the channel in vacuum, demonstrating a quality factor of 4700.

One of the devices we have fabricated is shown in Figure 2a. The twenty-micron-long device consists of an inner hollow region one micron wide and several hundred nanometers tall surrounded by a conformal coat of high stress nitride approximately 200 nm thick. The circular etch holes (visible impressions on either side of the channel) evenly distribute the stress of the high stress nitride. We have also filled the holes with oxide to eliminate leaking.

Resonant mechanical motion was excited and detected using optical techniques described in the literature [4]. The response for this channel when filled with vacuum is shown in Figure 2b. The fundamental resonant mode of the device is at 22.17 MHz, and it has quality factor 4700. These resonant properties are comparable to what has been demonstrated by Manalis and coworkers, but with roughly one thousand times smaller mass and one hundred times greater frequency [3]. Thus, we estimate that these devices will be able to measure the mass of particles in solution with sub-attogram precision, which will for the first time enable measurement of the masses of single viruses and proteins in solution.


  1. K. Jensen, Kwanpyo Kim, and A. Zettl, Nature Nanotechnology, 20 July 2008 (10.1038/nnano.2008.200).
  2. R. Lucklum and P. Hauptmann, Anal. Bioanal. Chem. 384, 667–682 (2006).
  3. T. P. Burg, M. Godin, S. M. Knudsen, W. Shen, G. Carlson, J. S. Foster, K. Babcock and S. R. Manalis, Nature 446 (7139), 1066-1069 (2007).
  4. Ilic, B.; Krylov, S.; Aubin, K.; Reichenbach, R.; Craighead, H. G., Applied Physics Letters 2005, 86 (19). Abstract