Progress in NEMS

Sensor Review

ISSN: 0260-2288

Article publication date: 7 September 2012

Citation

(2012), "Progress in NEMS", Sensor Review, Vol. 32 No. 4. https://doi.org/10.1108/sr.2012.08732daa.006

Publisher

:

Emerald Group Publishing Limited

Copyright © 2012, Emerald Group Publishing Limited


Progress in NEMS

Article Type: Nanosensor update From: Sensor Review, Volume 32, Issue 4

In this journal’s first Nanosensor update (Vol. 31, No. 1, 2011) a single-walled CNT-based NEMS resonator with a mass resolution of 1.4 zg (1.4×10−21 g) was reported which had been developed by workers from the Spanish Centre Investigacions Nanociencia Nanotecnologia (CSIC-ICN) in Barcelona in 2008. Theoretical work published by Cornell University in 2010 suggested that even greater resolutions, down to yg (10−24 g) levels, would be achieved in the near future and this has now been accomplished by the same Spanish group. The device consists of a suspended CNT, 150 nm long and 1.7 nm in diameter which acts as a beam (Figures 1 and 2) and caused to resonate at about 2 GHz by an FM mixing technique. Adding mass to the resonator reduces its resonant frequency. Experiments conducted under ultra-high vacuum conditions (∼3×10−11 mbar) to reduce interference from ambient molecules and at a temperature of just 4 K to minimise thermal effects showed that the resonator had a mass resolution of 1.7 yg. This is the greatest resolution yet achieved by such a device and to place it into perspective, this is roughly equal to the mass of a single proton (1.67 yg). To optimise the sensitivity of the device, an electrical annealing technique was applied whereby a current of ∼8 μA was passed through the CNT to remove any absorbed atoms. This work is clearly of great scientific interest, as nanoresonators operating at ultra-low temperatures might aid fundamental studies in quantum physics but the development also has practical potential. The sensor has been used to detect single naphthalene molecules and measure the binding energy of a xenon atom on the nanotube surface and the group suggests that it could find applications in future mass spectrometry systems.

Producing NEMS devices on an industrial scale has traditionally posed all manner of technological problems. Now, following several years’ collaboration with the French CEA/LETI-MINATEC research institute in Grenoble through the “Alliance for Nanosystems VLSI”, workers from the Kavli Nanoscience Institute and the Department of Physics at Caltech have reported the fabrication of VLSI NEMS arrays. These are produced from CMOS-compatible materials using microelectronic lithography and etching techniques, combined with nanoscale alignment. The arrays were fabricated from a 200 mm SOI wafer with a 160 nm-thick silicon layer and a 400 nm buried oxide layer. The metal film, a proprietary alloy which is fully compatible with CMOS, was deposited by a sputtering technique at 175°C and its thickness varied between 45 and 70 nm, depending on the design. Optical deep ultra-violet (DUV) lithography was then used to pattern the thin-film metal features: the wire bonding pads, lead-frame and the NEMS array itself. With this process, the first 200 mm wafers containing more than 3.5 million silicon NEMS devices were produced, thus achieving a density of approximately 60,000 resonators/mm2, with a functional device yield of ∼95 per cent. Figure 3 shows details of the array. While this work has used CMOS/MEMS technology to create nanoscale resonators, several groups are combining MEMS with nanomaterials. An example is an air flow sensor, reported in 2012 by a group from the National University of Singapore. This uses a silicon microcantilever with dimensions 90×20×3 μm, produced by a CMOS-compatible process, with 90 nm-wide silicon nanowires embedded close to its fixed end. When the sensor is placed in an air stream, the cantilever deflects and stresses the nanowires which are piezoresistive and change their resistance in proportion to the flow rate. Experiments showed a high flow sensitivity – 198 Ω/m/s and a sensing range of up to 65 m/s.

 Figure 1 Schematic of the CNT-based nanoresonator

Figure 1 Schematic of the CNT-based nanoresonator

 Figure 2 SEM of the NEMS device. The faint horizontal line is the CNT

Figure 2 SEM of the NEMS device. The faint horizontal line is the CNT

 Figure 3 A 200 mm SOI wafer containing more than 3 million NEMS devices

Figure 3 A 200 mm SOI wafer containing more than 3 million NEMS devices

As with their macroscale and microscale counterparts, NEMS resonators have the potential to act as platforms for high sensitivity gas and vapour sensors, whereby adsorbed molecules add mass and cause a reduction in the resonant frequency. Several groups have investigated this technology, although most have used cantilevers rather than beams as the resonating elements. In research reported in 2012 by the Caltech workers, a chemically modified version of the NEMS array was used as a high performance chemical vapour sensor. This was shown to detect a chemical warfare simulant at a part-per-billion concentration with an exposure time of just 2 s. In late 2011, a company, Analytical Pixels, which arose from the Caltech/CEA-LETI alliance, was set-up to commercialise the VLSI NEMS technology. The company will initially focus on multi-gas sensing systems based on a novel generation of NEMS-enabled chromatographs. Anticipated applications include air quality monitoring, uses in the petrochemicals sector and breath analysis systems for the early detection of biomarkers for medical diagnosis.

 Figure 4 Schematic of the AFM technique

Figure 4 Schematic of the AFM technique

 Figure 5 SEM of a nanomechanical transducer with an integrated optical
readout for high-sensitivity force and displacement measurements

Figure 5 SEM of a nanomechanical transducer with an integrated optical readout for high-sensitivity force and displacement measurements

Nanocantilevers are also poised to play a role in atomic force microscopy (AFM) which is an important tool for nanoscale surface metrology. Typical AFMs map local tip-surface interactions by scanning a flexible microcantilever probe over a surface (Figure 4). They rely on bulky optical sensing instrumentation to measure the motion of the probe, which limits sensitivity, stability and accuracy and precludes the use of probes much smaller than the wavelength of light. Recent research by the Center for Nanoscale Science and Technology (CNST) at NIST (the US National Institute of Standards and Technology) is expected to overcome these limitations. The CNST workers have fabricated a novel integrated sensor combining a nanomechanical cantilever probe with a high sensitivity nanophotonic interferometer on a single silicon chip. While probe stiffness was kept comparable to conventional microcantilevers in order to maintain high mechanical gain, the probe size was reduced to 25 μm in length, 260 nm in thickness and 65 nm in width. Because these smaller structures have an effective mass of less than a picogram (10−12 g), the detection bandwidth is dramatically increased, reducing the system response time to a few hundred nanoseconds. Readout is based on “cavity optomechanics”, with the probe fabricated adjacent to a microdisk optical cavity at a gap of less than 100 nm (Figure 5). Due to this close separation, light circulating within the cavity is strongly influenced by the motion of the probe tip. The cavity has a high optical quality factor (Q), meaning that the light makes tens of thousands of round-trips inside the cavity, thereby accumulating information about the probe’s position before leaking out. The combination of small probe-cavity separation and high Q (∼105) imparts sensitivity to probe motion of less than 1 fm/√Hz, while the cavity is able to sense changes in probe position with a bandwidth of over 1 GHz. The entire device is nanofabricated as a single, monolithic unit on a silicon wafer and is therefore compact, self-aligned and stable. Fibre optic waveguides couple light to and from the sensor, allowing it to be readily interfaced with standard optical sources and detectors. Finally, through simple changes to the probe geometry, the mechanics of the probe tip can be varied greatly, allowing for the different combinations of mechanical gain and bandwidth needed for a variety of AFM applications. A displacement sensitivity of ∼4.4×10−16 m/√Hz and a dynamic range of >106 is estimated for a 1 s measurement.