Axial-Stressed Piezoresistive Nanobeam for Ultrahigh

Aug 30, 2012 - With the scheme of axial stress, there is no longer the neutral-plane .... doping depth and, theoretically, can be thinned unlimited, w...
0 downloads 0 Views 2MB Size
Download PDF
Article pubs.acs.org/ac

Axial-Stressed Piezoresistive Nanobeam for Ultrahigh Chemomechanical Sensitivity to Molecular Adsorption Ying Chen, Pengcheng Xu, and Xinxin Li* State Key Lab of Transducer Technology, and Science and Technology on Microsystem Lab, Shanghai Institute of Microsystem and Information Technology, Chinese Academy of Sciences, 865 Changning Road, Shanghai 200050, China S Supporting Information *

ABSTRACT: This study proposes nanothickness piezoresistive double-clamped beams that are used in a double-side adsorbing mode. The axially stressed clamped beam exhibits continually increasing sensitivity as it is thinned down to nanoscale, and the thinning is theoretically without limitation. Sensing experiments to part-per-million levels of trimethylamine vapor well verify the proposal. A 93 nm thick beam sensor exhibits higher than 1 order of magnitude sensitivity compared to typical piezoresistive cantilever sensors, and its chemomechanical sensing resolution is comparable with that obtained by the off-cantilever optical detection method. With the nanobeam, a surprisingly ultrahigh sensitivity to surface molecular self-assembly induced surface stress is also obtained that is about 150 times higher than that obtained from a conventional cantilever. With additional advantages of elimination of single-sided adsorption induced bimetallic effect noise, tinier size, and easier fabrication, the ultrasensitive nanothick beam sensors show promise to replace the state-of-the-art piezoresistive cantilevers for bio/chemical nanomechanical detection.

M

to further improve sensitivity and detection of piezoresistive cantilevers, the cantilevers are expected to be made even thinner to increase the surface-stress induced bending. Unfortunately, a technical obstacle blocks this cantilever thinning approach. Limited by the impurity doping techniques, the doping depth is hard to be made very shallow and the cantilever is hard to be thinner than a couple of hundred nanometers.31−34 By now, Harley and Kenny of Stanford University have reported the thinnest piezoresistive cantilever of 100 nm that was fabricated with a doped-silicon epitaxial process.34 With the doping process limited shallowest piezoresistor doping depth, d, as a fixed value, following theoretic analysis will conclude that sensitivity improvement of piezoresistive cantilevers cannot be directly gained by shrinking the cantilever thickness into nanometric scale. Figure 1a shows the cross-sectional schematic of a piezoresistive cantilever sensor. Surface adsorption induced surface stress, σs, causes bending stress along the cantilever longitudinal direction, which can be read out by the integrated self-sensing piezoresistor. Assuming the cantilever thickness, h, is k times of the shallowest piezoresistive layer, i.e., h = kd and k ≥ 1, the piezoresistive sensitivity, represented by σl/σs, can be deduced as35

olecular self-assembly or specific binding at solid surface induces a surface-stress sensing signal.1−3 This ultrasensitive phenomenon has been used for nanomechanical bio/ chemical sensing by detecting bending deflection of a cantilever, where the sensing layer is coated only at one side of the cantilever and the cantilever undergoes bending due to the surface stress generated at the single side.4−7 For the cantilever sensors used at the early stage, the specific adsorption induced cantilever bending signals were read out with reflective optical beam detection techniques.8−15 This optical readout method features high resolution but makes difficulty in on-thespot rapid detection, as such kind of optical beam deflection detection setups like AFMs (atomic force microscopies) are generally expensive and bulky. Besides, the light-reflecting detection is difficult to be implemented for detection in a lowtransparent liquid environment. To further improve the cantilever chemomechanical sensor being capable of on-thespot rapid and low-cost detection, recently on-cantilever integrated elements have been widely used to build self-sensing cantilever sensors, where piezoresistive, piezoelectric, capacitive, or field-effect transistor (FET) sensing elements are integrated for in situ and real-time electric signal readout.3,16−19 Integrated with impurity-doped piezoresistors for bending stress detection, silicon piezoresistive cantilever sensors have become quite popular in such kinds of applications.4,20−30 Compared with optical detection cantilevers, self-sensing piezoresistive cantilevers are advantageous in low-cost and onsite portable detection. However, piezoresistive cantilevers generally feature relatively lower detection sensitivity. In order © 2012 American Chemical Society

Received: May 22, 2012 Accepted: August 30, 2012 Published: August 30, 2012 8184

dx.doi.org/10.1021/ac301388k | Anal. Chem. 2012, 84, 8184−8189

Analytical Chemistry

Article

increases with thinning the cantilever. Unfortunately, when further thinning the cantilever from k = 2, the sensitivity will rapidly decrease. The sensitivity will finally decrease to zero when the cantilever is thinned to k = 1. This principle really hinders the piezoresistive cantilever scaled down to nanothickness. The sensitivity decreasing versus cantilever thinning reflects the piezoresistance neutral-plane restriction rule. When the bottom of the piezoresistor layer steps over the middle plane (i.e., the neutral plane) of the silicon cantilever, the bending stress will change its sign (e.g., changed from tensile stress to compressive stress). The oppositely signed stresses will counteract with each other, which leads to the sensitivity decrease. In the worst case, when the cantilever is as thin as the piezoresistive layer, i.e., k = 1, the sensitivity will decrease to null. Due to both the piezoresistor neutral-plane restriction rule and the limited shallowest impurity doping depth, it is not preferred to further thin the piezoresistive cantilever into nanoscale for increased sensitivity. In order to break through the above-mentioned obstacle standing in the way toward nanometric sensing, this study, for the first time, proposes nanothickness double-clamped beams to replace the bending cantilevers for ultrasensitive piezoresistive sensing to surface stress. As is schematically shown in Figure 1c, the sensing layer is coated on double sides of the clamped−clamped beam, instead of the single-sided coating for cantilever sensors. Due to the specific molecule adsorption induced symmetric surface stress at the double surfaces of the double-clamped beam, bending will not be induced. Instead of that, axial stress along the beam longitudinal direction will be built in the silicon, while the stress along the transverse direction of the beam can be freely released. For nanothick (or very thin) beams, the double-sided generated surface stress can be approximately equivalent to an effect of axial force that is perpendicular to the cross section of the beam. With the whole silicon beam as a piezoresistor, the axial-stress signal will be read out by piezoresistive output. Obviously, with the beam thinned continually, the equivalent axial stress acting on the beam will continually increase, and the piezoresistive output will increase proportionally. If the average axial stress is also expressed as σl, sensitivity of the beam piezoresistor can be calculated (see the Supporting Information) as

σl /σs = 2/h

The sensitivity versus beam thickness is also plotted in Figure 1b for comparison with the bending cantilever. The intersection point of the two curves corresponds to the critical thickness of h = 3d. When the cantilever is thinner than 3d, the clamped beam of identical thickness will exhibit superior piezoresistive sensitivity. Obviously, the thinner the thickness is, the higher the sensitivity of the clamped-beam sensor is. With the scheme of axial stress, there is no longer the neutral-plane piezoresistive restriction and the shallowest doping limitation effect on the beam. Before the appearance of thinning-induced nanoscale size effects, e.g., surface effects, eq 2 indicates that a thinner beam will gain higher piezoresistive sensitivity. However, the minimum beam thickness is also limited by fabrication techniques. Using the fabrication technique developed in this study, currently we can achieve high-yield fabrication of the beams with the thickness as thin as about 90−100 nm. Along with technical development for fabricating even thinner piezoresistive beams, significant sensitivity improvement can be expected by replacing the conventional bending cantilevers with the nanothickness piezoresistive clamped beams.

Figure 1. (a) Schematic of a bending cantilever with its single side coated for sensing induced surface stress. The dimensional parameters are denoted for analysis. (b) Nominally defined as (σl/σs)d, the piezoresistive sensitivity is calculated as a function of the ratio in thickness between the whole cantilever and the doped piezoresistor layer. In comparison, the piezoresistive sensitivity of the herein proposed clamped beam is plotted together, where the piezoresistor is doped throughout the whole beam thickness of h. (c) Schematic of the piezoresistive clamped beam with its double surfaces coated for axialstress sensing.

σl /σs =

3⎛1 1⎞ ⎜ − 2⎟ d⎝k k ⎠

(2)

(1)

where σl is the average bending stress across the whole piezoresistor thickness. The derivation of eq 1 is detailed in the Supporting Information. According to eq 1, the analytic results are shown in Figure 1b. When k is larger than 2, the sensitivity 8185

dx.doi.org/10.1021/ac301388k | Anal. Chem. 2012, 84, 8184−8189

Analytical Chemistry

■

Article

EXPERIMENTAL SECTION Fabrication of the Nanothickness Beam. The silicon nanobeams are fabricated by using micro/nanomachining technologies that are shown in Figure 2a and described as

follows. (1) The process starts from bonding SOI (silicon on insulator) wafers that are supplied by the Sichuan Microelectronic Institute of China Electronics Technology Group. The original n-type top layers are 200 and 800 nm in thickness, respectively, and the buried silicon dioxide layer is 1 μm thick. Dry oxidation is used to thin the top silicon layer to the desired thickness of the beams. (2) With patterned photoresist as a mask, deep reactive ion etch (DRIE) is used to shape the beams. Then, boron ions are implanted and diffused to dope through the whole silicon beam into a piezoresistor. (3) Contact holes for the piezoresistors are opened, and a Ti/Au/ Cr multilayer metal film is sputtered, patterned, and sintered for interconnection. With a patterned photoresist as mask, buffered HF is used to release the nanobeam into free-standing. The etching starts from the double sides of the beam and, by lateral under-etching of wet HF to the SiO2 beneath the silicon beam, the double-clamped beam can be finally freed. By a plasmaenhanced ashing process, a very thin layer (thinner than 10 nm) of SiO2 is low-temperature grown on the double surfaces of the beam for both insulation and following surface selfassembly of the siloxane molecule sensing layer. The fabricated nanothickness beam and the close-up view of a 93 nm thick beam segment are shown in the scanning electron microscopy (SEM) images of Figure 2, parts b and c.

■

RESULTS AND DISCUSSION Nanobeam Sensors with Varied Thicknesses for Specific Sensing to TMA Vapor. To verify the sensitivity improvement effect of the proposed axial-stressed nanobeams, specific sensing to ultralow concentration gas of trimethylamine (TMA) is conducted. Detection of part-per-million levels of

Figure 2. (a) Main process steps for fabrication of the nanothickness piezoresistive clamped beam. (b) SEM image of a fabricated beam for axial-stress sensing. (c) Close-up view showing the sidewall of a 93 nm thick beam. The beam in the image is inclined with a 20° angle from the horizontal plane.

Figure 3. Schematic sequence shows the sensing group immobilization and specific adsorption of TMA molecules on the double-side surfaces of the beam. 8186

dx.doi.org/10.1021/ac301388k | Anal. Chem. 2012, 84, 8184−8189

Analytical Chemistry

Article

Figure 4. (a) Comparison in sensitivity to 5 ppm TMA between a conventional piezoresistive cantilever sensor and the proposed piezoresistive nanobeam sensors with varied beam thicknesses. (b) Curvature of the nanobeam sensing signal in terms of beam thickness that well fits the relationship obtained from analysis. (c1) Ultrahigh sensing signal to self-assembly induced surface stress is recorded online by using a nanobeam sensor that is compared with the much lowered signal (c2) where the conventional piezoresistive cantilever sensor is used. Due to the difference in cleaning−treatment extent to the individual beam surface and the process tolerance of the sensing layer self-assembly, nonuniformity in specific adsorption properties is inevitable among different devices. These are the main factors to induce the difference in time response of different beams (see panels a and c1−c2). When the gas is introduced to the sensor, the instantaneous instability signal shown in panel c1 comes from the disturbance of handling.

supplied to the Wheatstone bridge. A series of nanobeam sensors with variable thicknesses is sequentially tested, where the beam thickness varies from 640 down to 93 nm. The beam thickness is measured by both using a step profilometer to scan a designed testing structure for the beam and using SEM direct imaging. For comparison, a well-developed high-sensitivity piezoresistive cantilever (named as Proxi-Lever, 1 μm in thickness) sensor in our lab is also used to detect the same vapor.25−30 All the recorded sensing signals are plotted together in Figure 4a. Obviously, the clamped beams exhibit much superior sensitivity to the cantilever, and the nanobeam sensitivity increases along with thinning the beams. When the beam is thinned from 640 to 93 nm, about 15 times sensitivity improvement is gained. Compared to the cantilever, the 93 nm beam is with the sensitivity improved by about 35 times. With the output voltage for 5 ppm TMA vapor as representative, the relationship between sensitivity and thickness of the nanobeams is plotted in Figure 4b. On the basis of the experimentally obtained sensing signal and the built sensitivity modeling of the cantilever,25 the specific adsorption induced surface-stress value (σs) can be extracted. According to eq 2 and piezoresistive sensing principles (detailed in the Supporting Information), the modeled nanobeam sensitivity varying trend in terms of the beam thickness is calculated and also plotted in Figure 4b, which generally fits the experimental results. From the dramatic increasing trend in sensitivity with thinning the beam, very thin nanobeam sensors working in the proposed double-side sensing

TMA vapor has been identified as a mark to evaluate freshness of fish and seafood.36 Immobilization of the sensing molecule layer onto the beam surface and the surface specific adsorption scheme of TMA molecules are sequentially illustrated in Figure 3. The sulfonic acid group of −SO3H is self-assembled on the double-side surfaces of the beam to specifically capture TMA molecules. Self-assembly of the sensing groups is based on covalent-bond formation between the siloxane head of the molecule of 2-(4-chlorosulfonylphenyl)ethyltrimethoxysilane (purchased from ABCR) and the −OH terminals existing on the double SiO2 surfaces of the silicon beam. The self-assembly process is implemented for 12 h under dry toluene at 70 °C, and then, the sensing group of −SO3H is obtained by further hydrolysis of the self-assembled molecules at room temperature for 12 h. The specific adsorption mechanism of TMA molecules is based on hydrogen-bond-like interaction between the central N atom of TMA and the H in the −SO3H sensing group. Purchased liquid TMA is diluted by syringe injection into a volume-fixed chamber to form 5 ppm TMA vapor. Then the diluted vapor is switched on and exposed to the as-modified beam sensor for detection. The whole beam body serves as the self-sensing piezoresistor to detect the specific TMA adsorption induced surface stress. The sensing piezoresistor is electrically interconnected with three value-fixed resistors (integrated at the same silicon chip simultaneously by the doping process) to form a Wheatstone bridge configuration to output a differential voltage signal for data recording. The dc 200 mV is power8187

dx.doi.org/10.1021/ac301388k | Anal. Chem. 2012, 84, 8184−8189

Analytical Chemistry

Article

Compared with the conventional piezoresistive cantilevers, the proposed piezoresistive nanobeam sensors exhibit not only much higher sensitivity to specific adsorption induced surface stress but also other advantages in the following aspects. (1) To form absorbing induced bending, a cantilever sensor is normally coated with gold at its front-side surface. With the gold precoating, either a thiol sensing layer can be self-assembled on the gold surface or a siloxane sensing layer can be immobilized on the SiO2 back side. The thermal mismatch between the gold layer and the cantilever material inevitably causes a significant bimetallic effect that is an interfering noise to the surface-stress sensing signal. For the clamped-beam sensor, it is not necessary to precoat the gold layer, and an equivalent axial stress can be formed from the mutual balance between the double-side surface adsorption (where a siloxane sensing layer can be double-side modified on SiO2 surface). Hence, the beam sensor is immune to the bimetallic effect induced noise. Of course, environmental temperature change induced thermal-mismatch stress may cause influence to the axial beam that needs additional compensation for the sensing signal. (2) The beam body itself is used as the piezoresistor, and there is no need to make any piezoresistor pattern on the beam. In contrast, the Ushaped cantilever body or piezoresistor layout on the cantilever has to be patterned normally to locate the two electric terminals of the piezoresistor at the cantilever root for interconnection. With the two piezoresistor terminals at the double clamping ends, miniaturization of the beam sensor is easier. (3) During the micro/nano fabrication process, especially the structure release step, single-clamped cantilever structures are more mechanically fragile than the double-clamped beam counterpart, which is probably why the cantilever sensors have been seldom made as thin as the beams in this study. According to our fabrication results, the nanobeam sensors are indeed suitable for high-yield and low-cost batch fabrication. By replacing the conventional cantilever sensors, the ultrasensitive piezoresistive nanobeam sensors are expected to be widely used in on-the-spot detection of trace-concentration bio/chemical molecules.

and axially stressing piezoresistive mode are promising in ultrasensitive bio/chemical molecular detection. It is worth pointing out that the experimentally obtained relationship between the sensitivity increase and the beam-thickness decrease seems a little bit earlier than that expected by theoretical calculation. Several factors may influence to this result, e.g., signal nonuniformity from measurement to measurement and some size effects (such as surface effects) that may become to play important roles in the nanobeam behavior. Control experiments for cross-sensing response of this sensor to interfering gases have been implemented. According to the results (shown in Figure S4 of the Supporting Information), the sensor demonstrates negligible cross signal. Sensing Resolution of the Piezoresistive Nanobeam Sensor. On the basis of the experimentally obtained sensing signal from the cantilever sensor and the sensitivity model of the signal transformation from surface stress to piezoresistance,25 the TMA molecule specific adsorption in our experiment should generate 117 mN/m surface stress (detailed in the Supporting Information). If referred to the noise floor of a couple of microvolts and the 168 μV sensing output of the 93 nm thick beam sensor, about 1.4 mN/m surface-stress sensing resolution is obtainable (detailed in the Supporting Information). According to the about 4.8 μV piezoresistive sensing signal from the cantilever used in our experiment and about 1 μV noise-floor limited resolution (detailed in the Supporting Information), the surface-stress detecting resolution of the piezoresistive cantilever sensor is about 24.4 mN/m. Equivalently, the sensing resolution to the cantilever mechanical deflection is about 3.6 nm.25 In comparison with the cantilever sensor, the 93 nm thick beam sensor has exhibited about 18 times improvement in detecting resolution to surface stress. The 18 times improvement in surface-stress sensing resolution of the nanobeam can be considered being equivalent to about 2 Å deflection resolution of the cantilever sensor (if the cantilever is still used). Obviously, such angstrom-level deflection resolution has to be obtained by using a cantilever with an AFM-like optical detection method, rather than the self-sensing piezoresistive cantilever. Ultrasensitivity to Self-Assembly Induced Surface Stress. Even higher sensitivity has been obtained in our molecular self-assembly experiment, where we use phenethyltrichlorosilane (PETCS, purchased from Sigma-Aldrich) as the material to self-assemble the monolayer on the double-sided SiO2 surfaces of a 260 nm thick beam. The surface was pretreated with hot Piranha solution [7/3 (v/v) of 98% H2SO4/30% H2O2] for 3 min to densify the silanol groups. After being rinsed with deionized water for several times and dried with purified nitrogen gas, the nanobeam is immersed into 20 mL of ethanol (electrically not conductive) solution for sensor output baseline recording. Then, 20 μL of PETCS is smoothly injected for self-assembly onto the double sides of the nanobeam through the reaction between Si−OH and Si− Cl. Still power supplied by dc 200 mV, the recorded piezoresistive sensing output is as high as 14.9 mV, which is shown in Figure 4c1. For comparison, a traditional cantilever is also used in the control experiment. Herein the PETCS molecules are self-assembled only on the SiO2 surface of the cantilever back side, rather than modified on the metal surface of the cantilever front side. As is shown in Figure 4c2, the output signal is lower than 0.1 mV, which is 2 orders of magnitudes lower than that obtained by using the 260 nm thick beam.

■

CONCLUSIONS

Theoretical analysis indicates that the single-side-coated bending piezoresistive cantilever is difficult to be thinned into nanoscale for improving the sensitivity of specific adsorption induced surface stress, due to the technical limitation in minimum doping depth and the middle-plane restriction rule for piezoresistors. In contrast, when the proposed piezoresistive clamped beam is shrunk into nanothickness, it gains upgraded piezoresistive sensitivity from the double-side adsorption induced equivalent axial stress. The axial-stressed nanobeam is no longer restricted by the piezoresistor doping depth and, theoretically, can be thinned unlimited, where the doping can be simply throughout the whole beam thickness. Experiments of chemical adsorption and sensing have verified that, surpassing the widely used piezoresistive cantilevers, the sensitivity of the clamped-beam sensors continually increases along with thinning of the beams into nanometers. The sensing resolution of a 93 nm thick beam sensor even approaches that normally obtained by using a laser diode photonic detection method (e.g., atomic force microscopy). 8188

dx.doi.org/10.1021/ac301388k | Anal. Chem. 2012, 84, 8184−8189

Analytical Chemistry

■

Article

(21) Yoshikawa, G.; Akiyama, T.; Gautsch, S.; Vettiger, P.; Rohrer, H. Nano Lett. 2011, 11, 1044−1048. (22) Boisen, A.; Thaysen, J.; Jensenius, H.; Hansen, O. Ultramicroscopy 2000, 82, 11−16. (23) Kooser, A.; Gunter, R. L.; Delinger, W. D.; Porter, T. L.; Eastman, M. P. Sens. Actuators, B 2004, 99, 474−479. (24) Goericke, F. T.; King, W. P. IEEE Sens. J. 2008, 8, 1404−1410. (25) Li, P.; Li, X. X.; Zuo, G. M.; Liu, J.; Wang, Y. L.; Liu, M.; Jin, D. Z. Appl. Phys. Lett. 2006, 89, 074104. (26) Zuo, G. M.; Li, X. X.; Zhang, Z. X.; Yang, T. T.; Wang, Y. L.; Cheng, Z. X.; Feng, S. L. Nanotechnology 2007, 18, 25501. (27) Zuo, G. M.; Li, X. X.; Li, P.; Yang, T. T.; Wang, Y. L.; Cheng, Z. X.; Feng, S. L. Anal. Chim. Acta 2006, 580, 123−127. (28) Chen, Y.; Xu, P. C.; Li, X. X. Nanotechnology 2010, 21, 265501. (29) Yang, Y. L.; Chen, Y.; Xu, P. C.; Li, X. X. Microelectron. Eng. 2010, 87, 2317−2322. (30) Chen, Y.; Xu, P. C.; Liu, M.; Li, X. X. Microelectron. Eng. 2010, 87, 2468−2474. (31) Gel, M.; Shimoyama, I. J. Micromech. Microeng. 2004, 14, 423− 428. (32) Gel, M.; Shimoyama, I. In Proceedings of the IEEE Sixteenth Annual International Conference on Micro Electro Mechanical Systems, Cho, Y., Tabata, O., Eds.; IEEE, Robotics and Automation Society: Kyoto, Japan, Jan 19−23, 2003; pp 494−497, ISBN: 0-7803-7744-3. (33) Barlian, A. A.; Park, W. T.; Mallon, J. R.; Rastegar, A. J.; Pruitt, B. L. Proc. IEEE 2009, 97, 513−552. (34) Harley, J. A.; Kenny, T. W. Appl. Phys. Lett. 1999, 75, 289−291. (35) Bao, M. H. Analysis and Design Principles of MEMS Devices, 1st ed.; Elsevier Science: Amsterdam, The Netherlands, 2005; pp 44−47. (36) Xu, P. C.; Yu, H. T.; Li, X. X. Anal. Chem. 2011, 83, 3448−3454.

ASSOCIATED CONTENT

S Supporting Information *

Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org.

■

AUTHOR INFORMATION

Corresponding Author

*Phone: +86-21-62131794. Fax: +86-21-62513510. E-mail: [email protected]. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The research is supported by the Chinese 973 program (2011CB309503) and NSFC project (91023046, 61021064, 61161120322). Xinxin Li also thanks the Korean WCU project (R32-2009-000-20087-0).

■

REFERENCES

(1) Berger, R.; Delamarche, E.; Lang, H. P.; Gerber, C.; Gimzewski, J. K.; Meyer, E.; Guntherodt, H. J. Science 1997, 276, 2021−2024. (2) Fritz, J.; Baller, M. K.; Lang, H. P.; Rothuizen, H.; Vettiger, P.; Meyer, E.; Guntherodt, H. J.; Gerber, C.; Gimzewski, J. K. Science 2000, 288, 316−318. (3) Wu, G. H.; Ji, H. F.; Hansen, K.; Thundat, T.; Datar, R.; Cote, R.; Hagan, M. F.; Chakraborty, A. K.; Majumdar, A. Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 1560−1564. (4) Boisen, A.; Dohn, S.; Keller, S. S.; Schmid, S.; Tenje, M. Rep. Prog. Phys. 2011, 74, 036101. (5) Buchapudi, K. R.; Huang, X.; Yang, X.; Ji, H. F.; Thundat, T. Analyst 2011, 136, 1539−1556. (6) Senesac, L.; Thundat, T. G. Mater. Today 2008, 11, 28−36. (7) Waggoner, P. S.; Craighead, H. G. Lab Chip 2007, 7, 1238−1255. (8) Hansen, K. M.; Ji, H. F.; Wu, G. H.; Datar, R.; Cote, R.; Majumdar, A.; Thundat, T. Anal. Chem. 2001, 73, 1567−1571. (9) Lavrik, N. V.; Sepaniak, M. J.; Datskos, P. G. Rev. Sci. Instrum. 2004, 75, 2229−2253. (10) Thundat, T.; Warmack, R. J.; Chen, G. Y.; Allison, D. P. Appl. Phys. Lett. 1994, 64, 2894−2896. (11) Medalia, O.; Englander, J.; Guckenberger, R.; Sperling, J. Ultramicroscopy 2002, 90, 103−112. (12) Baselt, D. R.; Lee, G. U.; Hansen, K. M.; Chrisey, L. A.; Colton, R. J. Proc. IEEE 1997, 85, 672−680. (13) Baller, M. K.; Lang, H. P.; Fritz, J.; Gerber, C.; Gimzewski, J. K.; Drechsler, U.; Rothuizen, H.; Despont, M.; Vettiger, P.; Battiston, F. M.; Ramseyer, J. P.; Fornaro, P.; Meyer, E.; Guntherodt, H. J. Ultramicroscopy 2000, 82, 1−9. (14) Lang, H. P.; Berger, R.; Andreoli, C.; Brugger, J.; Despont, M.; Vettiger, P.; Gerber, C.; Gimzewski, J. K.; Ramseyer, J. P.; Meyer, E.; Guntherodt, H. J. Appl. Phys. Lett. 1998, 72, 383−385. (15) Lang, H. P.; Berger, R.; Battiston, F.; Ramseyer, J. P.; Meyer, E.; Andreoli, C.; Brugger, J.; Vettiger, P.; Despont, M.; Mezzacasa, T.; Scandella, L.; Guntherodt, H. J.; Gerber, C.; Gimzewski, J. K. Appl. Phys. A: Mater. Sci. Process. 1998, 66, S61−S64. (16) Mukhopadhyay, R.; Sumbayev, V. V.; Lorentzen, M.; Kjems, J.; Andreasen, P. A.; Besenbacher, F. Nano Lett. 2005, 5, 2385−2388. (17) Vancura, C.; Li, Y.; Lichtenberg, J.; Kirstein, K. U.; Hierlemann, A.; Josse, F. Anal. Chem. 2007, 79, 1646−1654. (18) Adams, J. D.; Parrott, G.; Bauer, C.; Sant, T.; Manning, L.; Jones, M.; Rogers, B.; McCorkle, D.; Ferrell, T. L. Appl. Phys. Lett. 2003, 83, 3428−3430. (19) Shekhawat, G.; Tark, S. H.; Dravid, V. P. Science 2006, 311, 1592−1595. (20) Arlett, J. L.; Maloney, J. R.; Gudlewski, B.; Muluneh, M.; Roukes, M. L. Nano Lett. 2006, 6, 1000−1006. 8189

dx.doi.org/10.1021/ac301388k | Anal. Chem. 2012, 84, 8184−8189