Trace explosive detection using photothermal ...

Trace explosive detection using photothermal deflection spectroscopy Adam R. Krause, Charles Van Neste, Larry Senesac, Thomas Thundat, and Eric Finot Citation: J. Appl. Phys. 103, 094906 (2008); doi: 10.1063/1.2908181 View online: http://dx.doi.org/10.1063/1.2908181 View Table of Contents: http://jap.aip.org/resource/1/JAPIAU/v103/i9 Published by the American Institute of Physics.

Related Articles Thermomicrocapillaries as temperature biosensors in single cells Appl. Phys. Lett. 102, 103505 (2013) A flash-lamp based device for fluorescence detection and identification of individual pollen grains Rev. Sci. Instrum. 84, 033302 (2013) Spectroscopic investigations of ammonia gas sensing mechanism in polypyrrole nanotubes/nanorods J. Appl. Phys. 113, 094504 (2013) Metamaterial optical refractive index sensor detected by the naked eye Appl. Phys. Lett. 102, 091902 (2013) Laser direct writing of silicon field effect transistor sensors Appl. Phys. Lett. 102, 093504 (2013)

Additional information on J. Appl. Phys. Journal Homepage: http://jap.aip.org/ Journal Information: http://jap.aip.org/about/about_the_journal Top downloads: http://jap.aip.org/features/most_downloaded Information for Authors: http://jap.aip.org/authors

Downloaded 15 Mar 2013 to 193.52.246.36. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

JOURNAL OF APPLIED PHYSICS 103, 094906 共2008兲

Trace explosive detection using photothermal deflection spectroscopy Adam R. Krause,1 Charles Van Neste,1 Larry Senesac,1 Thomas Thundat,1,a兲 and Eric Finot2 1

Oak Ridge National Laboratory, Oak Ridge, Tennessee 37831, USA and University of Tennessee, Knoxville, Tennessee 37996, USA 2 Institut CARNOT de Bourgogne, Nanosciences-Optique Submicronique, BP 47870, F-21078 Dijon, France

共Received 12 December 2007; accepted 19 February 2008; published online 2 May 2008兲 Satisfying the conditions of high sensitivity and high selectivity using portable sensors that are also reversible is a challenge. Miniature sensors such as microcantilevers offer high sensitivity but suffer from poor selectivity due to the lack of sufficiently selective receptors. Although many of the mass deployable spectroscopic techniques provide high selectivity, they do not have high sensitivity. Here, we show that this challenge can be overcome by combining photothermal spectroscopy on a bimaterial microcantilever with the mass induced change in the cantilever’s resonance frequency. Detection using adsorption-induced resonant frequency shift together with photothermal deflection spectroscopy shows extremely high selectivity with a subnanogram limit of detection for vapor phase adsorbed explosives, such as pentaerythritol tetranitrate 共PETN兲, cyclotrimethylene trinitramine 共RDX兲, and trinitrotoluene 共TNT兲. © 2008 American Institute of Physics. 关DOI: 10.1063/1.2908181兴 I. INTRODUCTION

The detection of trace amounts of explosives with high selectivity using handheld devices is of great interest for its applications in homeland defense, forensics, humanitarian demining, and military use. Since explosives have very low vapor pressures, extremely high sensitivity is essential for explosive trace detection.1 Optical techniques such as Raman spectroscopy offer high selectivity but do not have sufficient sensitivity for trace detection.2,3 Most widely used techniques such as ion mobility spectroscopy have high sensitivity and selectivity but are bulky and expensive.4 Also, ion mobility based sensors are not quantitative and can also give false alarms for some nonexplosive species.2 Microfabricated sensors such as microcantilevers have been considered as a sensor platform for next generation explosive detection due to their advantages such as miniature size, array-based detection capability, high sensitivity, real-time operation, and low power consumption.5,6 The adsorption of molecules on the cantilever results in changes in its resonance frequency. In addition, the bending of the cantilever also varies due to adsorption-induced stress.7 Both of these signals can be simultaneously detected for cantilevers that are around 100– 400 ␮m in length with a spring constant of around 0.1– 1 N / m. However, despite its demonstrated high sensitivity in detection, the microcantilever-based sensor platform does not offer any intrinsic chemical selectivity. Such lack of selectivity is common to other sensor platforms such as surface acoustic wave devices, quartz crystal microbalance, and surface plasmon resonance sensors unless coupled to another device, such as a gas chromatograph or functionalized with receptors for molecular recognition.8–13 Since sensor reversibility is an attractive feature, the receptors are often based a兲

Corresponding author. Electronic mail: [email protected].

0021-8979/2008/103共9兲/094906/6/$23.00

on weak interactions that can be broken at ambient temperature. This reversibility requirement limits the number of chemical interactions that can serve as a basis for receptor design, and receptors based on weak interactions are not specific enough to produce unique responses. Therefore, reversible receptor-based approaches for chemical selectivity fail when complex mixtures are present due to the lack of orthogonality of chemical interaction responses. Increasing the number of sensor units does not generally lead to higher selectivity because of the limited chemical interactions that can be used as a basis for receptor design. Resorting to higher energy chemical interactions may offer higher selectivity at the expense of sensor reversibility. Therefore, despite all the advantages of a micromechanical sensor platform, such as miniature size, low-power consumption, and the remarkably high sensitivity, its use as a practical sensor for vapor phase small molecule detection is questionable unless novel methods are utilized by which chemical selectivity can be achieved. Here, we show that detection using adsorption-induced resonance frequency shift together with photothermal deflection spectroscopy 共PDS兲 can satisfy the conditions of high selectivity, sensitivity, and sensor reversibility. We demonstrate highly selective detection of subnanogram quantities of adsorbed explosives, such as RDX, PETN, and TNT using PDS. The sensitivity of the technology presented here can be improved further by optimizing the bimaterial effect. This method has the potential for detection of submonolayer amounts of molecules with high selectivity and sensitivity. Microcantilever-based photothermal spectroscopy was first demonstrated by Barnes et al.13 Since then, it has been used for detection of a range of chemicals including DNA, explosives, and chemical warfare agents.14–19 Cantilever based photothermal spectroscopy utilizes the extremely high thermal sensitivity of a bimaterial cantilever. This sensitivity is high enough to detect femtojoule thermal changes.14 A

103, 094906-1

© 2008 American Institute of Physics

Downloaded 15 Mar 2013 to 193.52.246.36. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

094906-2

J. Appl. Phys. 103, 094906 共2008兲

Krause et al.

bimaterial cantilever with adsorbed molecules shows bending when exposed to infrared 共IR兲 radiation. The adsorbed molecules absorb the IR energy and heat the bimaterial cantilever causing it to bend. The deflection of the cantilever tip due to the bimaterial effect can be found from the following equation:20,21 3 t1 + t2 l3 P, z = − 共 ␣ 1 − ␣ 2兲 2 4 t2K 共␭1t1 + ␭2t2兲w

共1兲

where z is the cantilever deflection at the tip, ␣2 and ␣2 are the coefficients of thermal expansion for the two layers, l is the length of the cantilever, t1 and t2 are the layer thicknesses, ␭1 and ␭2 are the thermal conductivities, w is the width of the cantilever, and P is the total power absorbed by the cantilever. The subscripts 1 and 2 represent gold and silicon substrate, respectively. The parameter K stands for the expression K=4+6

冉冊 冉冊 冉冊 冉冊 t1 t1 +4 t2 t2

2

+

E1 t1 E2 t2

3

+

E2 t2 . E1 t1

共2兲

In the PDS technique, the cantilever with adsorbed chemical species is scanned using successive pulses of monochromatic IR radiation. The cantilever deflection as a function of IR wavelength resembles the IR absorption spectrum of the adsorbate molecules. Cantilever-based photothermal spectroscopy, therefore, combines the extremely high sensitivity of a cantilever beam and the selectivity of molecular vibration spectroscopy. The limit of detection of this technique is limited by the thermal sensitivity of the cantilever which can be optimized by properly designing the bimaterial aspect of the cantilever. The amount of explosives adsorbed on the cantilever surface can be determined from measuring the variation in the resonance frequency of the cantilever. Assuming the change in the cantilever spring constant due to adsorption was negligible; the mass of explosives ␦m on the cantilever can be calculated from the equation

␦m = m

冉 冊

␯21 −1 . ␯22

共3兲

Here, ␯1 and ␯2 are the initial and final resonance frequencies of the cantilever during mass adsorption and m is the cantilever mass. For cantilever sensors, the bending and the resonance frequencies can be simultaneously measured.

II. EXPERIMENTAL

The experiments were carried out using commercially available silicon microcantilevers with dimensions of 350 ␮m length, 35 ␮m width, and 1 ␮m thickness 共MikroMasch, Oregon兲. The microcantilevers were prepared by depositing a 600 nm layer of gold with 10 nm of chromium as an adhesion layer. Prior to gold evaporation using an e-beam evaporator, the cantilevers were thoroughly cleaned in acetone and ethanol. The cantilever deflection was monitored using an optical beam deflection arrangement. In the optical beam deflection method a focused beam from laser diode is

FIG. 1. Schematic of the experimental setup used for photothermal deflection spectroscopy 共PDS兲. The IR from the monochromator is focused on the cantilever and chopped at 80 Hz. The wavelength is scanned between 2.5 and 14.5 ␮m. The resulting deflection data are collected with the lock-in amplifier. The spectrum analyzer is used to measure the resonance frequency to calculate the mass of explosives adsorbed on the cantilever.

reflected off the cantilever into a position sensitive detector 共PSD兲. The output voltage from the PSD is directly proportional to the cantilever deflection. The bimaterial cantilevers were then used as a substrate for explosive vapor deposition. To deposit the explosives on the cantilever surface we used a custom-built vapor generator 共Idaho National Laboratory, INL兲 that can deliver precise concentrations of explosive vapors in these experiments. The explosive vapors were produced by flowing nitrogen through a chamber containing the explosives dissolved in an acetonitrile solution. This solution is deposited onto thickly packed glass wool kept at a constant temperature. Separate vapor generators were used for generating PETN, RDX, and TNT vapor streams to avoid cross contamination. Two thermoelectric elements kept the reservoir at constant temperature generating a saturation vapor pressure within the reservoir that is proportional to the temperature of the thermoelectric elements. For the TNT experiments, the generator was heated to 60 ° C. Due to the lower vapor pressures of RDX and PETN, their vapor generators were heated to 75 ° C. The nitrogen flow was kept at 200 SCCM 共SCCM denotes cubic centimeter per minute at STP兲 for all of the experiments. Photothermal spectra of the adsorbate on the cantilever were obtained by illuminating the cantilever beam with monochromatic infrared radiation. We have used the IR source and the interference filter wheel of a Foxboro Miran 1A-CVF spectrophotometer for these studies. The experimental setup is shown in Fig. 1. The IR energy from the source is focused onto the cantilever using a concave mirror. The spectrometer has the capability of varying the wavelength from 2.5 to 14.5 ␮m by using an interference filter wheel. The IR source was used for illuminating the cantilever beam as shown in the experimental setup in Fig. 1. The bending of the cantilever beam was monitored using an optical beam deflection arrangement and a PSD. The PSD signal was fed to an in-house fabricated amplifier circuit. The circuit output was calibrated using a standard atomic force

Downloaded 15 Mar 2013 to 193.52.246.36. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

094906-3

Krause et al.

FIG. 2. Photothermal deflection spectrum of vapor phase adsorbed PETN on the cantilever. The peak near 6 ␮m is caused by the asymmetric stretching of the O – NO2 bonds. The peak near 8 ␮m is from the symmetric stretching of the same bonds.

microscope to determine the signal’s correlation to the cantilever bending. A PSD signal of 1 V was found to correspond to 90 nm of cantilever deflection. Prior to exposure to the explosive vapor, before each experiment, a baseline IR spectrum was taken using the Stanford Research Systems SR850 lock-in amplifier with the chopper set at 80 Hz. The resonance frequency was then measured using a Stanford Research Systems SR760 spectrum analyzer. The cantilevers were then exposed to explosive vapors by placing the cantilever 共silicon side facing the flow, gold side facing away兲 directly in front of the vapor stream in open air prior to exposure to IR light. The cantilever, therefore, is exposed to other molecules in the ambient air including relative humidity just as it would be under field application conditions. As the explosive vapors condense on the cantilever, the cantilever bends and the resonance frequency varies. The resonance frequency of the cantilever was then measured after exposure to the explosive vapors to determine mass loading by explosive molecules. After the resonance frequency measurement, the monochromator, with its IR light focused on the cantilever, scans between 2.5 and 14.5 ␮m wavelengths, while the lock-in amplifier records the cantilever bending and generates an IR profile of the explosive-cantilever combination.

J. Appl. Phys. 103, 094906 共2008兲

FIG. 3. Photothermal deflection spectrum of RDX adsorbed on the cantilever from vapor phase. The peaks seen at 6.4 and 7.6 ␮m are caused by the asymmetric and symmetric stretchings of the N – NO2 bonds, respectively.

vacuum oven or 共2兲 exposure to UV-ozone followed by a short etch in piranha solution. The collected spectra were identical for the subsequent PDS runs. We have also carried out the PDS experiments on different cantilevers with slightly different resonance frequencies. The spectra were identical and the peaks agreed very well with known peaks of PETN, RDX, and TNT. Photothermal signals depend on the thermodynamic and energy transfer properties of the sample and the cantilever beam. Temperature changes resulting from absorption of infrared energy are directly related to the vibration modes of the adsorbates as well as the heat capacity and thermal conductivity of the cantilever beam. The observed photothermal spectra correspond to various vibrational modes of the explosives. The adsorbed molecules are excited into vibration modes by absorption of IR photons and the nonradiative deexcitation of the molecules results in thermal energy. The thermal energy is then transferred to the bimaterial substrate causing cantilever bending. The extent of bending is proportional to the amount of thermal energy transferred to the cantilever. The peak at around 7.4– 8 ␮m, found in all three explosive spectra, is caused by the symmetric stretching vibration of the NO2 共nitro兲 group bond.23–26 The peak near

III. RESULTS AND DISCUSSIONS

Figure 2 shows the PDS spectrum of PETN on a silicon cantilever. The spectrum shown in Fig. 2 is obtained by dividing the IR profile by the baseline of the bare silicon cantilever taken prior to explosive adsorption. The resulting spectrum is then normalized for qualitative comparison. The observed peaks are in excellent agreement with IR absorption spectra of bulk PETN.22 Figures 3 and 4 show the photothermal deflection spectrum of adsorbed RDX and TNT, respectively. The observed peaks also agree very well with IR absorption spectra of the respective explosives.22 We have repeated the experiments on the same cantilever after desorbing the adsorbed explosives and thoroughly cleaning the cantilever using one of two cleaning methods: 共1兲 rinsing with acetone and ethanol followed by 30 min in a

FIG. 4. TNT photothermal deflection spectrum. The stretching peaks at 6.5 and 7.4 ␮m can be seen. These are caused by the asymmetric and symmetric stretching of the C – NO2 bonds, respectively.

Downloaded 15 Mar 2013 to 193.52.246.36. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

094906-4

Krause et al.

6 – 6.6 ␮m is caused by the asymmetric stretching vibration of the same bond.23,26 The slight shift in these two peaks is due to the nitro group being bound to a different atom in each explosive compound. In PETN, the nitro groups are bound to oxygen 共O – NO2兲. This bond has a symmetric stretch vibration at wavenumber of 1285 cm−1 共7.78 ␮m兲 and an asymmetric stretch vibration at wavenumber of 1658 cm−1 共6.03 ␮m兲.23,25 In RDX, the nitro groups are bound to nitrogen 共N – NO2兲. This bond has a symmetric stretch vibration at wavenumber of 1310 cm−1 共7.63 ␮m兲 and an asymmetric stretch vibration at wavenumber of 1570 cm−1 共6.37 ␮m兲.23 In TNT, they are bound to carbon 共C – NO2兲. This bond has a symmetric stretch vibration at wavenumber of 1350 cm−1 共7.41 ␮m兲 and an asymmetric stretch vibration at wavenumber of 1530 cm−1 共6.54 ␮m兲.23,24 The relative intensities of the peaks were slightly different for a photothermal spectrum as compared to a conventional IR spectrum. Also some peaks that are not very prominent in conventional IR spectra appear to have higher intensity in the photothermal spectra. This may be directly related to the efficiency of nonradiative decay of these excited states. We did not observe a large shift in the peaks due to substrate effects. The extent of bending is directly proportional to the adsorbed material, the impinging power of IR radiation, the absorption mode, and the thermal sensitivity of the cantilever. The thermal sensitivity of the cantilever can be tuned by controlling the thicknesses of the cantilever and the metal layer and by judicial choice of the bimaterial couples. It is possible to increase the sensitivity of photothermal deflection spectroscopy by optimizing the bimaterial cantilever parameters as well as increasing the power of the illuminating IR source. For example, by selecting different metals and optimizing the thickness of the coating, it is possible to make a bimaterial cantilever very sensitive to thermal changes. It is also possible to fabricate cantilevers with an optimized spring constant for increased thermal bending or to pattern the cantilever surface for increased adsorption. Another way to increase sensitivity is by restricting the heat flow from cantilever into the base of the cantilever. The dynamic range of detection for this setup was determined by adsorbing the explosives on the cantilever and simultaneously monitoring the variation in PDS peak and the resonance frequency shift as a function of desorption time. The dynamic range for TNT adsorption is shown in Fig. 5. The silicon cantilever used had 600 nm of gold on one side and a mass of 170 ng. The mass of explosive adsorbed on the cantilever was calculated using Eq. 共3兲 for each spectrum taken. We observed that in open air all explosives from vapor phase adsorbs on both sides of the cantilever 共since after adsorption the laser used to measure the deflection was diffusively reflected off the gold surface as opposed to specularly reflected as was observed before adsorption兲. Therefore, the mass of TNT on either side may be approximately half of the adsorbed mass. Assuming the spring constant does not change due to TNT adsorption, this corresponds to a thickness of 40 nm on each side. It is worthwhile mentioning that the photothermal spectrum is caused by the explosives adsorbed on the silicon side only since gold is an excellent

J. Appl. Phys. 103, 094906 共2008兲

FIG. 5. Height of TNT peak near 7.4 ␮m in arbitrary units as a function of adsorbed mass. At regular time intervals, the resonance frequency was measured just before the spectrum was taken to find the peak. As the mass desorbs, the peak height increases at first, then decreases. The initial increase is due to the initial overabundance of TNT on the cantilever preventing the thermal energy from reaching the cantilever. As the TNT mass desorbs the thermal energy reaches the cantilever more efficiently causing the peak height increase. The height then decreases due to less infrared absorption by a diminished amount of adsorbed TNT. This graph shows a dynamic range from 800 pg to almost 9 ng. However, only half of the adsorbed mass is contributing to PDS spectra.

reflector for IR. The linear nature of the desorption rate supports the assumption that the spring constant remains the same during adsorption and desorption process. The dynamic range measurement for TNT in Fig. 5 shows the height of the 7.4 ␮m peak of adsorbed TNT as a function of mass desorption from an initial mass of 12 ng. Initially, as the desorption starts, the peak height increases as a function of desorption and reaches a maximum at around 8 ng. However, as the mass of the TNT decreases due to desorption, the peak height decreases after reaching a maximum value. The initial part of the curve in Fig. 5 agrees well with the fact that more heat is absorbed when more TNT is present on the surface. However, at a certain thickness the amount of heat transferred to the cantilever substrate decreases probably due to thermal insulation of the TNT. It is also possible that the thick explosive layer prevents IR from reaching the bottom most layers.27 As the explosives desorb and the layer becomes thinner, heat is more readily transferred to the cantilever and the height of the peak increases. As the layer further desorbs, there are less explosives to absorb the IR and the peak decreases. The minimum detectable photothermal signal was obtained for 800 pg of adsorbed material. However, since the bimaterial effect is caused by TNT adsorbed only on one of the surfaces of the cantilever, the limiting value is 400 pg. This value corresponds to an average of 10 ML of TNT. The dynamic range, as seen in Fig. 5, extends from 400 pg 共half of 800 pg兲 to as high as 4.25 ng 共half of 8.5 ng兲. As the adsorbed mass increases past 8.5 ng 共4.25 ng per side兲, the peak decreases in size and the detector loses sensitivity. Upon inspection of Fig. 5, we can see that within the dynamic range the peak height can be calibrated to the mass adsorbed on the cantilever. Therefore, once calibrated, the

Downloaded 15 Mar 2013 to 193.52.246.36. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

094906-5

J. Appl. Phys. 103, 094906 共2008兲

Krause et al.

sensitivity. We calculated that a subnanogram limit of detection can be obtained for vapor phase adsorbed explosives, such as TNT, RDX, and PETN. The observed deflection peaks of explosives are in excellent agreement with spectra obtained with conventional techniques. It is possible to improve the limit of detection of this detection method by optimizing cantilever properties. In addition to finding the spectra, we also investigated the desorption characteristics of TNT using the photothermal deflection spectroscopy technique along with resonance frequency measurements. From this study, we have calculated a 14 kJ mol−1 heat of adsorption. ACKNOWLEDGMENTS FIG. 6. Mass of TNT desorbed from a microcantilever as a function of time. The mass of TNT on the cantilever is calculated from the resonance frequency shift. A linear regression can also be seen. This regression shows that the TNT desorbs at a rate of 1.1 pg/ s.

peak height would be sufficient for calculating the adsorbed mass. The frequency shift would no longer be needed for this purpose. Another sensor characteristic that is important in many practical applications is its ability to regenerate. TNT has a vapor pressure of 1.1⫻ 10−6 Torr at 20 ° C. RDX and PETN have vapor pressures of 4.1⫻ 10−9 and 3.8⫻ 10−10 Torr, respectively. The adsorbed TNT, therefore, desorbs at a faster rate than RDX or PETN. The amount of TNT lost due to evaporation during any single experiment is negligible to cause any change in spectral intensities. Figure 6 shows the desorption of adsorbed TNT as a function of time at room temperature. From the linear regression, we find a desorption rate of 1.1 pg/ s. This corresponds to a molecular desorption rate of approximately 2.9⫻ 109 s−1. It is assumed that the explosive desorption follows an Arrhenius-type law,28,29

冋 册

kd = A exp

− Ea . NAkT

共4兲

where kd is the molecular desorption rate, k is Boltzmann’s constant, NA is Avogadro’s number, Ea is the heat of adsorption and T is the temperature, and A is the pre-exponent factor which is the molecule-surface vibrational frequency. The value of A is generally taken to be 1012 Hz.28 From the molecular desorption rate found from Fig. 6 and Eq. 共4兲, we calculate an average value for heat of adsorption of TNT as 14 kJ mol−1. This is in good agreement with the reported value of 12 kJ mol−1.29 The strong dependence of desorption on temperature can be used for regenerating the sensor. Since the thermal mass of the cantilever is very small, it can be heated to desorb the adsorbed explosive by increasing the temperature, for example, focusing the total IR power 共broad band spectrum without filtering兲 on the cantilever. IV. CONCLUSION

In conclusion, we have demonstrated that photothermal deflection spectroscopy, combined with resonance frequency shift of bimaterial microcantilevers, offers very high selectivity for trace detection of explosives while maintaining

This work was supported in part by the Office of Naval Research 共ONR兲 and the U.S. Department of Energy’s Office of Nonproliferation Research and Development in the National Nuclear Security Administration. Oak Ridge National Laboratory is managed by UT-Battelle, LLC, for the U.S. Department of Energy under Contract No. DE-AC0500OR22725. Trace Chemical Sensing of Explosives, edited by R. L. Woodfin 共Wiley, New York, 2007兲. 2 D. S. Moore, Rev. Sci. Instrum. 75, 2499 共2004兲. 3 J. I. Steinfeld and J. Wormhoudt, Annu. Rev. Phys. Chem. 49, 203 共1998兲. 4 M. Nambayah and T. I. Quickenden, Talanta 63, 461 共2004兲. 5 L. A. Pinnaduwage, V. Boiadjiev, J. E. Hawk, and T. Thundat, Appl. Phys. Lett. 83, 1471 共2003兲. 6 L. A. Pinnaduwage, A. Gehl, D. L. Hedden, G. Muralidharan, T. Thundat, R. T. Lareau, T. Sulchek, L. Manning, B. Rogers, M. Jones, and J. D. Adams, Nature 共London兲 425, 474 共2003兲. 7 G. Y. Chen, T. Thundat, E. A. Wachter, and R. J. Warmack, J. Appl. Phys. 77, 3618 共1995兲. 8 R. A. McGill, T. E. Mlsna, R. Chung, V. K. Nguyen, and J. Stepnowski, Sens. Actuators B 65, 5 共2000兲. 9 E. J. Houser, T. E. Mlsna, V. K. Nguyen, R. Chung, R. L. Mowery, and R. A. McGill, Talanta 54, 469 共2001兲. 10 J. Bowen, L. J. Noe, B. P. Sullivan, K. Morris, V. Martin, and G. Donnelly, Appl. Spectrosc. 57, 906 共2003兲. 11 C. Hartmann-Thomas, J. Hu, S. N. Kaganove, S. E. Keinath, D. L. Keeley, and P. R. Dvornic, Chem. Mater. 16, 5357 共2004兲. 12 G. Bunte, J. Hürttlen, H. Pontius, K. Hartlieb, and H. Krause, Anal. Chim. Acta 591, 49 共2007兲. 13 J. R. Barnes, R. J. Stephenson, M. E. Welland, Ch. Gerber, and J. K. Gimzewski, Nature 共London兲 372, 79 共1994兲. 14 G. Li, L. W. Burggraf, and W. P. Baker, Appl. Phys. Lett. 76, 1122 共2000兲. 15 P. G. Datskos, S. Rajic, M. J. Sepaniak, N. Lavrik, C. A. Tipple, L. R. Senesac, and I. Datskou, J. Vac. Sci. Technol. B 19, 1173 共2001兲. 16 P. G. Datskos, M. J. Sepaniak, C. A. Tipple, and N. Lavrik, Sens. Actuators B 76, 393 共2001兲. 17 E. T. Arakawa, N. V. Lavrik, S. Rajic, and P. G. Datskos, Ultramicroscopy 97, 459 共2003兲. 18 E. T. Arakawa, N. V. Lavrik, and P. G. Datskos, Appl. Opt. 42, 1757 共2003兲. 19 A. Wig, E. T. Arakawa, A. Passian, T. L. Ferrell, and T. Thundat, Sens. Actuators B 114, 206 共2006兲. 20 J. R. Barnes, R. J. Stephenson, C. N. Woodburn, S. J. O’Shea, M. E. Welland, T. Rayment, J. K. Gimzewski, and Ch. Gerber, Rev. Sci. Instrum. 65, 3793 共1994兲; 66, 3083 共1995兲. 21 J. M. Antonietti, J. Gong, V. Habibpour, M. A. Rottgen, S. Abbet, C. J. Harding, M. Arenz, U. Heiz, and C. Gerber, Rev. Sci. Instrum. 78, 054101 共2007兲. 22 F. Pristera, M. Halik, A. Castelli, and W. Fredericks, Anal. Chem. 32, 495 共1960兲. 23 I. R. Lewis, N. W. Daniel, Jr., and P. R. Griffiths, Appl. Spectrosc. 51, 1854 共1997兲. 24 P. S. Makashir and E. M. Kurian, Transp. Res. Part E 55, 173 共1999兲. 1

Downloaded 15 Mar 2013 to 193.52.246.36. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions

094906-6 25

J. Appl. Phys. 103, 094906 共2008兲

Krause et al.

P. S. Makashir and E. M. Kurian, Propellants, Explos., Pyrotech. 24, 260 共1999兲. 26 R. W. Beal and T. B. Brill, Appl. Spectrosc. 59, 1194 共2005兲. 27 A. Wig, A. Passian, E. Arakawa, T. L. Ferrell, and T. Thundat, J. Appl.

Phys. 95, 1162 共2004兲. P. Atkins, Physical Chemistry, 6th ed. 共Freeman, New York, 1999兲. 29 L. A. Pinnaduwage, T. Thundat, A. Gehl, S. D. Wilson, D. L. Hedden, and R. T. Lareau, Ultramicroscopy 100, 211 共2004兲. 28

Downloaded 15 Mar 2013 to 193.52.246.36. Redistribution subject to AIP license or copyright; see http://jap.aip.org/about/rights_and_permissions