Routine Femtogram-Level Chemical Analyses Using Vibrational Spectroscopy and Self-Cleaning Scanning Probe Microscopy Tips

Anal. Chem. 2008, 80, 3221-3228 Routine Femtogram-Level Chemical Analyses Using Vibrational Spectroscopy and Self-Cleaning Scanning Probe Microscopy Tips Keunhan Park,, Jungchul Lee, Rohit Bhargava,*,, and William P. King*,, Beckman Institute for Advanced Science and Technology, Department of Bioengineering, and Department of Mechanical Science and Engineering, University of Illinois at UrbanasChampaign, Illinois 61801 Simultaneous structural and chemical characterization of materials at the nanoscale is both an immediate need and an ongoing challenge. This article reports a route to address this need, which can be rapidly adopted by practitioners, by combining the benefits of widely available scanning probe microscopy and vibrational microspectrometry. In an atomic force microscope (AFM), the probe tip can provide a nanoscale topographic image. Here, we use a temperature-controlled probe tip to selectively acquire an analyte from a specified location and determine its mass in a thermogravimetric manner. The tip is then analyzed via complementary Raman and Fourier transform infrared microspectrometers, providing a molecular characterization of samples down to the femtogram level in minutes. The probe can be self-cleaned and employed for repeated use by rapidly heating it to vaporize the analyte. By combining the established analytical modalities of AFM and vibrational spectrometry, a complete physical and molecular characterization of nanoscale domains is possible: mass determination is facile, thermal analyses can be integrated on the probe, and the obtained spectral data can be related to existing knowledge bases. The atomic force microscope (AFM) 1 has provided significant opportunities to probe structures and manipulate materials at the nanometer scale. A functional AFM tip may probe nanometerscale mechanical 2 or electrical 3 properties. Consequently, various schemes have been developed for coupled materials analysis, including micromechanical thermogravimetry, 4 nanocalorimetery, 5 and micro- 6 /nano- 7,8 thermal analysis. Though these techniques * To whom correspondence should be addressed. E-mail: rxb@uiuc.edu; wpk@uiuc.edu. Beckman Institute for Advanced Science and Technology. Department of Bioengineering. Department of Mechanical Science and Engineering. (1) Binnig, G.; Quate, C. F. Phys. Rev. Lett. 1986, 56, 930. (2) Yu, M.-F.; Files, B. S.; Arepalli, S.; Ruoff, R. S. Phys. Rev. Lett. 2000, 84, 5552. (3) Martin, Y.; Abraham, D. W.; Wickramasinghe, H. K. Appl. Phys. Lett. 1988, 52, 1103. (4) Berger, R.; Lang, H. P.; Gerber, C.; Gimzewski, J. K.; Fabian, J. H.; Scandella, L.; Meyer, E.; Guntherodt, H. J. Chem. Phys. Lett. 1998, 294, 363. (5) Efremov, M. Y.; Olson, E. A.; Zhang, M.; Schiettekatte, F.; Zhang, Z. S.; Allen, L. H. Rev. Sci. Instrum. 2004, 75, 179. (6) Pollock, H. M.; Hammiche, A. J. Phys. D: Appl. Phys. 2001, 34, R23. provide physical characterization, their chemical analysis capabilities are limited and can be easily exceeded by spectroscopic measurements. Spectral measurements by AFM, however, have not proven straightforward, and chemical characterization of very small quantities of materials remains a theme at the forefront in both microscopy and spectroscopy. 9,10 Similarly, many research groups have tried to combine nanometer-scale structural measurements with molecular vibrational spectroscopy, for example, using tip-enhanced Raman spectroscopy (TERS) 11-14 and near-field IR spectroscopy. 15-18 The best achievable spatial resolution with these methods is 50 nm for TERS 12,13 and 100 nm for near-field IR spectroscopy 15,18 with a mass resolution that has been claimed to be 10 fg. 11 Unfortunately, these spectrometers are not readily accessible to most laboratories due to complexities in the system setup and sample preparation. Moreover, their low signal-to-noise ratio (SNR) necessitates long data acquisitions and optical coupling complicates interpretation, making the integration of a scanning tip with a spectrometer quite challenging and interpreting the observed spectra rather suboptimal. 19 In a different approach, the AFM picks up material that is then delivered elsewhere for chemical analysis. There are a few examples of using this approach: the sampled material has been analyzed by mass spectrometry, 20,21 surface-enhanced Raman spectroscopy (SERS), 11 or photothermal infrared (PTIR) spec- (7) King, W. P.; Saxena, S.; Nelson, B. A.; Pitchimani, R.; Weeks, B. L. Nano Lett. 2006, 6, 2145. (8) Nelson, B. A.; King, W. P. Rev. Sci. Instrum. 2007, 78, 023702. (9) Levin, I. W.; Bhargava, R. Annu. Rev. Phys. Chem. 2005, 56, 429. (10) Novotny, L.; Stranick, S. J. Annu. Rev. Phys. Chem. 2006, 57, 303. (11) Anderson, M. S. Appl. Phys. Lett. 2000, 76, 3130. (12) Stockle, R. M.; Suh, Y. D.; Deckert, V.; Zenobi, R. Chem. Phys. Lett. 2000, 318, 131. (13) Anderson, M. S.; Pike, W. T. Rev. Sci. Instrum. 2002, 73, 1198. (14) Pettinger, B.; Ren, B.; Picardi, G.; Schuster, R.; Ertl, G. Phys. Rev. Lett. 2004, 92, 096101. (15) Knoll, B.; Keilmann, F. Nature 1999, 399, 134. (16) Reading, M.; Grandy, D.; Hammiche, A.; Bozec, L.; Pollock, H. M. Vib. Spectrosc. 2002, 29, 257. (17) Hammiche, A.; Bozec, L.; German, M. J.; Chalmers, J. M.; Everall, N. J.; Poulter, G.; Reading, M.; Grandy, D. B.; Martin, F. L.; Pollock, H. M. Spectroscopy 2004, 19, 20. (18) Dazzi, A.; Prazeres, R.; Glotin, E.; Ortega, J. M. Opt. Lett. 2005, 30, 2388. (19) Novotny, L.; Hecht, B. Principles of Nano-Optics; Cambridge University Press: Cambridge, 2006. (20) Lee, D.; Wetzel, A.; Bennewitz, R.; Meyer, E.; Despont, M.; Vettiger, P.; Gerber, C. Appl. Phys. Lett. 2004, 84, 1558. (21) Wetzel, A.; Socoliuc, A.; Meyer, E.; Bennewitz, R.; Gnecco, E.; Gerber, C. Rev. Sci. Instrum. 2005, 76, 103701. 10.1021/ac702423c CCC: $40.75 2008 American Chemical Society Analytical Chemistry, Vol. 80, No. 9, May 1, 2008 3221 Published on Web 03/27/2008

Figure 1. (a) Schematics of the spectroscopic analysis of thermally sampled paraffin on the probe tip (not scaled). Paraffin is locally melted and sampled on the tip while a heated probe scans over a paraffin specimen in an AFM platform. Raman and FT-IR spectroscopy are then employed for the chemical analysis of the sampled paraffin. After the analysis, paraffin is removed by heating the probe high above its melting point, allowing the reuse of the probe. (b) The SEM image of a heatable cantilever probe used in this study. troscopy. 16 While the utility of nondestructive vibrational spectroscopy has been demonstrated to be useful in these studies, SERS and PTIR are not routine sampling techniques in most laboratories. Furthermore, the spectral results are not straightforward to interpret and may not be compatible with existing, comprehensive databases of IR and Raman spectra. Hence, the utilization of available spectrometers in common sampling modes (i.e., transmission or reflection) would make the approach widely useful. In principle, the combination of commercially available AFMs and vibrational spectrometers is a strategy that might be easily implemented in most laboratories to enable practical molecular recognition in very small samples. While AFM and spectroscopy are often performed on the same sample, to our knowledge, the combination of AFM techniques with vibrational microspectroscopy has not been reported. Here, we report a systematic approach to femtogram-scale spectroscopic analysis, including sampling, spectroscopic analysis and facile probe cleaning. EXPERIMENTAL SECTION Sampling. The experimental procedure consists of three steps, shown in Figure 1a. The first step is to topographically image and extract a defined sample region using the functional tip, the second is to spectroscopically sense the extracted sample, and the third is to clean the tip for reuse. Our experiments use a silicon cantilever probe with an integrated heater-thermometer, as shown in Figure 1b, in an AFM (Asylum MFP-3D). The cantilever tip temperature can be precisely controlled over the temperature range 25-1000 C, using a resistive heater integrated into the cantilever. This type of heated cantilever is made from doped single-crystal silicon, the fabrication and characterization of which are described elsewhere. 22,23 Similar heated cantilevers have been adopted for data storage, 24 manufacturing, 25,26 and materials analysis. 7,8 The resistive heater element also serves as a resistive thermometer, and so the probe temperature may be calibrated, (22) Lee, J.; Beechem, T.; Wright, T. L.; Nelson, B. A.; Graham, S.; King, W. P. J. Microelectromech. Syst. 2006, 15, 1644. (23) Park, K.; Lee, J.; Zhang, Z. M.; King, W. P. J. Microelectromech. Syst. 2007, 16, 213. (24) King, W. P.; Kenny, T. W.; Goodson, K. E.; Cross, G. L. W.; Despont, M.; Dürig, U. T.; Rothuizen, H.; Binnig, G.; Vettiger, P. J. Microelectromech. Syst. 2002, 11, 765. (25) Nelson, B. A.; King, W. P.; Laracuente, A. R.; Sheehan, P. E.; Whitman, L. J. Appl. Phys. Lett. 2006, 88, 033104. (26) Szoszkiewicz, R.; Okada, T.; Jones, S. C.; Li, T. D.; King, W. P.; Marder, S. R.; Riedo, E. Nano Lett. 2007, 7, 1064. 3222 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008

Figure 2. (a) AFM image of a paraffin specimen after the heated probe scoops paraffin on its tip. The elliptic mark shown in the figure designates the topographic change of the specimen after sampling. The inset shows the change of depth profiles before and after sampling, which are respectively drawn in dashed and solid curves. (b) The AFM image demonstrating that sub-10 fg of paraffin can be acquired simply by modifying the operation condition, such as heater temperature and contact force. (c) The SEM image of the probe tip before the paraffin sampling. (d) The SEM image of the probe after the paraffin sampling, which clearly shows paraffin around the tip. Both scale bars in (c) and (d) indicate 3 µm. monitored, and controlled. During the sampling process, the heater temperature is ramped to slightly above the melting point of the analyte. The melting of the analyte, which can be monitored calorimetrically, is confined to the region near the tip and wets the tip surface in a highly localized manner. The analyte remains adhered to the probe due to the oleophilic nature of silicon after cooling and disengaging the probe from the substrate. Hence, experimental parameters such as heating rate, surface functionalization, tip geometry and load conditions represent an opportunity to control the amount of material that is extracted. Physical and Spectroscopic Characterization. The second step in our procedure is to analyze the extracted material. Two measurements were performed for characterization: spectroscopic imaging and gravimetric analysis using the fundamental resonance frequency change of the probe. A Renishaw InVia Raman microscope was employed for Raman spectral measurements. A 488-nm Ar + laser was used through a 50 /0.75 objective, providing a focal spot of 1 µm in diameter. The output laser power was 50 µw, which is low enough to prevent laser heating of the probe. 22 The integration time was set to 300 s, a reasonable time for analytical measurements. A Varian 7000 FT-IR/600 UMA microspectrometer with a 32 32 mercury cadmium telluride focal plane array was used without any modifications. The spectral resolution was set to 4 cm -1 in a rapid scanning mode with a maximum Fourier frequency of 5 khz. To achieve a satisfactory SNR, 256-1024 coadded scans were averaged. Though an undersampling ratio of 2 was employed to provide interferograms, data spanning the 4000-1000-cm -1 bandpass were saved due to the lower detector cutoff. Details of data processing can be accessed elsewhere. 27 In the present study, paraffin was chosen as a demonstrative analyte due to its active vibrational modes for both Raman and IR spectroscopy. However, it should be noted that other materials can also be sampled and analyzed using the same method: see Supporting Information. A paraffin sample was prepared by spincoating on a glass substrate at 3500 rpm for 20 s to provide a 1-µm-thick paraffin film that partially covers the glass surface. Panels a and b in Figure 2 show topographic images of the paraffin film after the sampling. Our objective is to obtain local sampling in a controlled manner in terms of a sampling position as well as a sampling amount. Similarly, one could generate an AFM image of two phases of a composite material and sample one of the phases selectively. In Figure 2a, we have collected paraffin in a relatively large area (i.e., 15 µm in length and 1 µm in width) as a demonstrative example. However, as will be further discussed, the mass of mounted paraffin was measured to be 117 ( 5 fg. (27) Bhargava, R.; Wang, S. Q.; Koenig, J. L. Appl. Spectrosc. 2000, 54, 1690. Analytical Chemistry, Vol. 80, No. 9, May 1, 2008 3223

Self-Cleaning. A critical consideration is the possibility of repeated use of the same tip for measurements. Hence, the third step in our process is to clean the probe tip. We hypothesized that the sample could be removed by heating the tip to well above the decomposition temperature of the analyte. Since the tip can be heated up to 1000 C, most organic materials can be vaporized and removed. To test the approach, we repeatedly measured the amount of sample on the tip as a function of heating time and temperature. Using the internal resistive heater, the probe was heated for 1 min at indicated temperatures and subject to measurements. Figure 4a is a series of optical microscope images taken after each heating step. Steps 0 and 1 denote before and after the sampling, respectively, and steps 2-5 correspond to heating steps at different heater temperatures for 1 min. A small spot in step 0 is an oxide residue deposited during the fabrication and has no effect on the experimental results. When compared to step 1, the paraffin at step 2 occupies a slightly larger area with an interference fringe, indicating that paraffin melts and spreads, which is expected due to the oleophilic nature of silicon. The heater temperature at step 2 is 77 C, which is slightly higher than the paraffin melting point of 65 C. At each of the following steps, the paraffin shrinks, indicating thermal vaporization. 30 Finally, in step 5, we observe that paraffin is completely removed. Figure 3. Raman spectra of the probe without paraffin, bulk paraffin, and paraffin-mounted probe. The presence of paraffin on the tip, despite its extremely small amount, provides several characteristic peaks of paraffin on the probe baseline spectrum. Peaks marked as p1-p7 come from paraffin while s1 and s2 come from the probe itself. Although the area modified by the probe tip is relatively large, most of the paraffin is pushed and piled up at the boundary during the sampling process, and a smaller amount is mounted on the probe. This feature can be clearly verified in the inset of Figure 2a, which compares depth profiles before and after sampling. Significantly more localized sampling should be possible with better control over sampling size, contact force, and heater temperature. Figure 2b demonstrates that with the present setup sub-10 fg of paraffin can be obtained with a 100-nm lateral resolution: the measured paraffin mass was 7 ( 4 fg. The presence of paraffin on the probe tip can be verified from Figure 2c and d, the SEM images of the probe before and after sampling. The presence of paraffin can also be verified spectroscopically. Figure 3 compares Raman spectra from the probe before sampling, from the bulk paraffin sample on a glass substrate and from the probe after sampling. The Raman spectrum of the paraffinmounted probe shows characteristic peaks of paraffin (p1-p7) along with those of the probe (s1 and s2). Peaks p1-p7 are associated with C-C stretching and CH 2 and CH 3 deformation in the straight-chain hydrocarbon structure of paraffin, 28 and peak s1 is associated with the scattering of the transverse optical phonon of Si. 29 The peak s2 is not a characteristic band arising from Si and is likely due to contamination during fabrication or storage of this particular probe. It was not detected in other probes. (28) Lin-Vien, D. The Handbook of Infrared and Raman Characteristic Frequencies of Organic Molecules; Academic Press: Boston, 1991. (29) Wang, R. P.; Zhou, G. W.; Liu, Y. L.; Pan, S. H.; Zhang, H. Z.; Yu, D. P.; Zhang, Z. Phys. Rev. B 2000, 61, 16827. RESULTS Thermogravimetric Results. The mass of paraffin and its reduction during cleaning can be measured using nanothermogravimetric analysis. Here, Figure 4b shows the fundamental resonance frequency change of the probe and associated paraffin mass at each heating step. The paraffin mass ( m) can be estimated from m ) k(f -2 2 - f1-2 )/4π2, where k and f are the spring constant and fundamental resonance frequency of the probe, respectively. 31 A noise spectrum measurement in the AFM reveals k and f of the probe before sampling as k ) 0.76 ( 0.01 N/m and f ) 69.39 ( 0.02 khz. Upon the paraffin sampling (step 1) as shown in Figure 2a, the resonance frequency decreases to 68.39 ( 0.01 khz, or 1.44%, equivalent to the mass increase of 117 ( 5 fg. When heated in steps, the cantilever resonance frequency increases at each step, eventually returning to the presampling frequency to indicate complete removal of paraffin. This measurement clearly demonstrates the self-cleaning capability of our probe. It also demonstrates that the cleaning can be rapidly accomplished, in a matter of minutes, without any adverse effects on the cantilever. Although not a focus of this study, femtogram-scale thermogravimeric analysis should be possible using the heated tip and could be used to further characterize the analyte or corroborate spectroscopic findings. Raman Spectroscopic Characterization. Figure 4c shows Raman spectra of each heating step. Characteristic peaks attributable to paraffin can be detected up to step 3, at which point the paraffin mass is 52 ( 6 fg. Peak intensities become smaller due to decreasing mass of the analyte, and the Raman spectrum at step 4 does not demonstrate peaks significantly above the noise level, indicating that the current configuration can analyze between (30) Plyushch, G. V.; Pryadka, G. A. Powder Metall. Met. Ceram. 1970, 9, 392. (31) Chen, G. Y.; Warmack, R. J.; Thundat, T.; Allison, D. P.; Huang, A. Rev. Sci. Instrum. 1994, 65, 2532. 3224 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008

Figure 4. Change of paraffin mass and Raman spectrum during the self-cleaning process. In the cleaning process, the probe was jouleheated for 1 min at various probe heater temperatures. (a) Optical microscopic images of the heater region show that paraffin is removed by heating the probe. (b) Paraffin mass change at each step, obtained by monitoring the fundamental resonance frequency change of the probe. (c) Raman spectra at each step. Without any special sample preparation, conventional Raman spectroscopy can perform the chemical analysis on 50 fg of paraffin. 10 and 50 fg. From the linear extrapolation of the peak intensity at 1133 cm -1 (peak p2), the minimum analyte amount is roughly estimated to be 20 fg for these conditions. The limit of detection could be further reduced by changing experimental parameters such as the integration time and laser power: a simple increase in recording time and moderate increases in power could provide femtogram-level sensitivity in a matter of hours. As with conventional spectroscopic studies, the tradeoff between sensitivity and speed of analysis is left to the operator. FT-IR Spectroscopic Characterization. Although the lateral resolution is diffraction-limited to a few micrometers, FT-IR spectroscopic imaging is preferred for obtaining spatially resolved spectroscopic data from a large area in a relatively short time. 9 Hence, we used FT-IR imaging to image the paraffin-mounted probe with a 5.5 µm 5.5 µm pixel size. Since this pixel size is not small enough to resolve the width of the v-shaped heater (i.e., 6.6 µm) in Figure 1b, we instead used a different type of heatable probe having a relatively large triangular-shaped heater. Once the analyte was sampled, the probe was fixed on an IR-transparent BaF 2 plate for the transmission measurement. Figure 5a shows the absorbance image of the probe at 2916 cm -1, and Figure 5b shows the absorbance spectra at two pixels representing the heater region (A) and leg region (B). For clarity, the probe boundary from integrated optical microscopy is overlaid on Figure 5a. The IR spectrum at pixel A has three peaks between 2800 and 3000 cm -1 due to CH 2 and CH 3 stretching vibrational modes, 28 confirming the presence of paraffin in the heater region. As shown in Figure 5a, paraffin cannot be simply distinguished from the probe in the raw absorbance image due to high reflection and scattering of the leg at 2916 cm -1. The IR spectrum of the leg becomes more reflective at the mid-ir region, due to its heavy doping level and resultant increase in free carriers. The fringed patterns in absorption spectra of the heater and leg are attributed to the interference of multireflected light inside the probe. 32 As common in IR spectral interpretation, the recorded spectrum is baseline-corrected with a two-point linear procedure across each peak to remove optical effects of the probe. Figure 6a provides the absorbance spectrum from the heater region when the probe is heated at different temperatures. The corresponding absorbance peak height images at 2916 cm -1 are shown in Figure 6b. The paraffin is clearly distinguished from the background, and its removal can be monitored during the stepwise heating process. In both panels, as expected, the paraffin fingerprint is reduced to finally disappear as the probe is heated to higher temperatures. Interestingly, a dim spot at the upper leg region in Figure 6b can be seen from paraffin residue accidentally acquired during the sampling process. This paraffin residue was (32) Born, M.; Wolf, E. Principles of Optics: Electromagnetic Theory of Propagation, Interference and Diffraction of Light; 7th ed.; Cambridge University Press: Cambridge, 1999. Analytical Chemistry, Vol. 80, No. 9, May 1, 2008 3225

Figure 5. FT-IR microspectroscopic image of the paraffin-mounted probe. (a) The absorbance image of the probe at the wavenumber of 2916 cm -1. (b) The absorbance spectra of the heater and leg regions, which are taken from the pixel A and B in (a), respectively. Three peaks in the absorbance spectrum of the heater region are signature peaks of paraffin, and the fringe of the spectrum is due to the interference effect. not removed due to the low temperature at the leg region. Only the initial mass of the paraffin was measured using the resonance method as the probe position was fixed for pixel-to-pixel comparisons between images. The initial paraffin mass was measured to be 74 ( 7 fg. The comparison between step 4 (T H ) 134 C) in Figure 3 and step 3 (T H ) 126 C) in Figure 6 reveals that the minimum paraffin analyzable with FT-IR spectroscopy is smaller than that of Raman spectroscopy, being close to 10 fg. This lower limit of FT-IR spectroscopy arises directly from its higher SNR as a consequence of the linear absorption process being stronger than Raman scattering. Last, the total scanning time for data acquisition was less than 4 min. A tradeoff 33,34 between spectral (33) Bhargava, R.; Levin, I. W. Anal. Chem. 2001, 73, 5157. resolution, coaddition, undersampling, and acquisition time or use of a noise reduction algorithm 35,36 may be conducted by the practitioner to reduce this limit to below the 1-fg level. The obtained experimental results also reveal several important issues that need to be considered for routine use of femtogramlevel sampling and chemical analysis using combined AFM and spectroscopy. The first consideration is the experimental parameters used in material extraction and analysis. While a heated tip retracts from the locally melted sample surface, analyte can be mounted only when the viscous force at the analyte/probe (34) Bhargava, R.; Levin, I. W. Anal. Chem. 2002, 74, 1429. (35) Bhargava, R.; Ribar, T.; Koenig, J. L. Appl. Spectrosc. 1999, 53, 1313. (36) Bhargava, R.; Wang, S. Q.; Koenig, J. L. Appl. Spectrosc. 2000, 54, 486. 3226 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008

Figure 6. FT-IR spectrum changes during the self-cleaning process. (a) The absorbance peak changes at each heating step. As paraffin is evaporated due to high temperature, all the absorbance peaks finally disappear. (b) From the peak height images at 2916 cm -1, the location of paraffin and its amount can be determined. interface exceeds the surface tension of the liquid bridge. Thus, the feasibility of sampling depends on material properties, including the surface tension and viscosity, as well as experimental parameters, including heating rate, temperature, and load conditions. A systematic approach should be studied to determine the theoretical framework and feasibility of the tip-based sampling for a wider range of materials. The second consideration is the probe geometry. For example, interference fringes observed in Figure 5b are likely to result from the fine dimensional tolerances on microscopic probes. To ease routine spectral analyses, these can be removed by a rougher or reflective coating on the nonsampling region of the probe. Coatings may also be used on the thermal probe itself to adjust the surface wetting capabilities and, hence, sampling capability for various materials. With improved probe design and sampling strategy, the methodology demonstrated here can be further optimized for the physical and molecular characterization of nanoscale domains. CONCLUSIONS We have demonstrated that imaging and highly localized femtogram-level extraction of a specified region can be achieved by using a heated cantilever probe in an AFM platform. The sample mass was determined via cantilever resonance. Chemical analysis on the extracted samples using commercially available Raman and FT-IR microspectrometers, without any special treatment of the sample or change in metrology, was facile and required only minutes. The minimum detectable amount of the analyte is in the 10-100-fg range and can be reduced to the femtogram range using the usual tradeoffs in spectroscopy. Spectral data are straightforward to compare with the bulk case and interpret as, using the probe tip in this manner, near-field coupling does not dominate the data. Last, the same probe is demonstrated to be useful for repeated measurements by a simple self-cleaning process. We anticipate that this approach will help bridge the gap between nanoscale structural analysis and molec- Analytical Chemistry, Vol. 80, No. 9, May 1, 2008 3227

ular spectroscopy in the immediate future and in a manner that is useful to analytical laboratories. ACKNOWLEDGMENT Funding for this work was provided in part by University of Illinois at UrbanasChampaign through a Critical Initiatives in Research and Scholarship (CIRS) grant, and by the National Science Foundation (DMR0114103, DMI03-28162, CBET0731930). The authors thank Dr. Z. Dai, Dr. G. Srinivasan, and Ms. N. Keith for assistance. SUPPORTING INFORMATION AVAILABLE Additional information as noted in text. This material is available free of charge via the Internet at http://pubs.acs.org. Received for review November 26, 2007. Accepted February 21, 2008. AC702423C 3228 Analytical Chemistry, Vol. 80, No. 9, May 1, 2008